United States
Environmental Protection
Agency
Office of Research and
Development
Washington DC 20460
EPA/600/R-00/076
October 2000
vvEPA
US EPA Often ol Research and Development
Proceedings
Sixth US/Germany
Workshop on Ozone/Fine
Particle Science
-------
-------
EPA/600/R-00/076
October 2000
Proceedings
Sixth US/Germany Workshop
on
Ozone/Fine Particle Science
(Combined with University of California/Riverside Workshop on
Environmental Chamber Design)
Mission Inn
Riverside, California
October 4-6, 1999
US-Germany Environmental Agreement Project Managers
Basil Dimitriades
U.S. Environmental Protection Agency (EPA), USA
Dieter Jost
Umweltbundesarn.pt, Germany
William Carter
U. California/Riverside
Workshop Co-Chairmen
Karl Becker
U. Wuppertal, Germany
Basil Dimitriades
U.S. EPA
Editor
Basil Dimitriades
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711 USA
Printed on Recycled Paper
-------
DISCLAIMER
The Workshop Proceedings was developed by assembling presentation manuscripts or
presentation abstracts submitted in advance of the Workshop. Presented papers were not
subjected to peer review, and, therefore, reflect only the viewpoints of the presenters and not
necessarily the view of the U.S. Environmental Protection Agency, and no official endorsement
should be inferred.
ACKNOWLEDGMENTS
The Workshop was instigated by the US-Germany Environmental Agreement Managers
Drs. Basil Dimitriades (USA) and Dieter lost (Germany), and consisted of two parts: a one-
Session part on Ozone/Particular Matter (PM) Regulatory Policy Issues and Developments
conducted at Research Triangle Park (RTP), NC, on October 4, 1999, and a part on ozone/PM
science conducted at Riverside, California, on October 4-6, 1999. The RTP Session was
organized by Drs. William Harriett (USEPA/OAQPS) and Gary Foley (USEPA/ORD/NERL).
The Riverside part was organized by Drs. William Carter and Joe Norbeck, and Ms. Phyllis
Crabtree, U. California/Riverside, and was conducted with a U. California/Riverside Workshop on
Environmental Chamber Design. The Workshop agenda was developed by Drs. Carter and Karl
Becker, U. Wuppertal. Workshop costs were covered by the U. California/Riverside. Some
financial assistance was provided by the German Government (Ministry of Science and
Technology) and the U.S. Environmental Protection Agency to selected invited participants. The
contribution of these individuals and organizations are acknowledged with gratitude by the
Workshop organizers.
-n-
-------
PREFACE
The Workshop was held within the framework of cooperation under the US-Germany
Environmental Agreement (Agreement Number: Oil, Project Number: 002). It contributes to
high priority activities identified by Minister Toepfer and Administrator Thomas during their
December 1987 meeting. As of to date, such Workshops have been held as follows:
First Workshop:
Second Workshop:
Third Workshop:
Fourth Workshop:
Fifth Workshop:
Cologne, Germany, May 4-6, 1988
Chapel Hill, NC, USA, June 5 - 8, 1990
(Held in coordination with the EC)
Lindau/Lake Constance, Germany, June 30 - July 3, 1992
(Held in coordination with BMBF and EC)
Charleston, SC, USA, June 13 - 17, 1994
(Held in coordination with BMBF and EC)
Berlin, Germany, September 24 - 27, 1996
(Held in coordination with BMBF and UBA)
-in-
-------
CONTENTS
DISCLAIMER ii
ACKNOWLEDGMENTS ii
PREFACE.
.111
WORKSHOP ANNOUNCEMENT .. .." vi
WORKSHOP AGENDA viii
SESSION I: Gas Phase Chemistry and Modeling
Overview of Atmospheric Chemistry Studies in Europe 2
Atmospheric Chemistry of VOCs and Nox 7
Mechanisms for Air Quality Modeling: Development and Applications of the Regional Atmospheric
Chemistry Mechanism 15
Development and Evaluation of the SAPRC-99 19
Modelling Ozone Formation with a Master Chemical Mechanism 42
Simulations of EUPHORE and Field Experiments Using a Master Chemical Mechanism 43
Influence of Biogenic VOCs on Photo-oxidant Formation: Simulation Experiments in EUPHORE and
Comparison with Model Calculations 52
Atmospheric Fate of Alkoxy and Carbonyl Radicals 58
Poster Session:
Atmospheric Chemistry of Selected Hydroxy-Carbonyls 69
Products and Mechanisms of the Reactions of 1,3-Butadiene with Chlorine Atoms in Air 74
Analysis of Gas Phase Halogen Compounds Using Atmospheric-Pressure-Ionization Mass
Spectrometry 79
New Product and Aerosol Studies on the Oxidation of DMS 88
Total Non-Methane Organic Carbon: Measurements of Total and Speciated Hydrocarbons at
Azusa, California 93
Use of ab initio Quantum Mechanics to Estimate Rate Constants 103
A Measurement Method for Hydroxyacetone 105
Fast Gas Chromatography with Luminol Detection for Measurement of Nitrogen Dioxide and
PANs 110
Simulation Chamber Study of Night-time Chemistry of Aldehydes and PANs 118
EPA PM Chemistry Chamber Studies 123
Development of the Master Chemical Mechanism (MCMv2.0) Web Site and Recent Applications
of Its Use in Tropospheric Chemistry Models 125
SESSION I: Gas Phase Chemistry and Modeling (continued)
Aromatic Hydrocarbon Research in the European Photoreactor (EUPHORE) 132
Studies on Oxygenated Fuel Additives: Ethers and Acetals 145
Development of Degradation Mechanisms of Oxygenated Hydrocarbons 150
Atmospheric Oxidation of Ethers Under High and Low NOx Conditions 160
Atmospheric Oxidation Mechanism of Unsaturated Oxygenated VOCs 168
-iv-
-------
Degradation Mechanisms for the Troposphere Oxidation of Chlorocarbons 173
SESSION II: Heterogeneous Chemistry and Modeling
The UNC Chamber Auxiliary Model (Wall Model) '. 175
Measurement and Modeling of NOx Off-gasing From FEP Teflon Reactors 178
Experimental Techniques for Studying Surface Chemistry in Smog Chambers 192
Heterogeneous Reaction of N2O5 on Anorganic Aerosols: The Nitrate Effect 198
Tropospheric Aqueous Phase Chemistry Laboratory and Modelling Studies 199
SESSION III: Measurement Methods
In-Situ Radical Measurements in EUPHORE 201
Measurement of Peroxy Radicals in the European Photoreactor EUPHORE 202
Measurement of NOy and Potential Artifacts 209
Instrumentation for State-of-the-Art Aerosol Measurements in a Smog Chamber 218
Use of a Thermal Desorption Particle Beam Mass Spectrometer for Studies of Secondary Organic
Aerosol Formation 221
SESSION IV: Environmental Chamber Studies
Aerosol Formation From the Reaction of-Pinene and Ozone Using a Gas Phase Kinetics-Particle
Partitioning Model 230
Secondary Aerosol Formation From Biogenic Precursors 240
Smog Chamber Studies of Particle Formation from the Oxidation of Terpenes 241
Determination of Photolysis Frequencies and Quantum Yields for Small Carbonyl Compounds using the
EUPHORE Chamber 246
Investigation of Real Car Exhaust in the EUPHORE Chamber 254
Outdoor Smog Chamber Experiments in Mexico 263
EPA Gas Phase Chemistry "Chamber Studies 269
Atmospheric Photochemical Degradation of 1,4-Unsaturated Dicarbonyls 272
A Modelling System for Mechanism Evaluation Using Chamber Data 303
Chamber Evaluation of Process Diagnostics and Photochemical Indicators 307
Ozone Formation in Coastal Urban Atmospheres: The Role of Anthropogenic Sources of Chlorine . . 321
SESSION V: Reactivity Studies
U.S. EPA Models-3 - Status and Applications 328
Analysis of Chamber-Derived Incremental Reactivity Estimates for n-Butyl Acetate and 2-Butoxy
Ethanol 333
Atmospheric Availability as a Component of the Tropospheric Ozone-Forming Potential of Volatile
Organic Compounds 358
California's Reactivity Program ' 368
Programme of Control Concepts and Measures for Ozone 371
SESSION VI: New Chamber Projects
Design of SAPHIR (Simulation of Atmospheric Photochemistry in a Large Reaction Chamber) 374
A Research Plan for a New Environmental Chamber Facility 375
Research Plan for "Next Generation Environmental Chamber Facility for Chemical Mechanism and VOC
Reactivity Evaluation" . 378
-v-
-------
WORKSHOP ANNOUNCEMENT
SIXTH US/GERMAN WORKSHOP ON
THE PHOTOCHEMICAL OZONE PROBLEM AND ITS CONTROL.
REGULATORY ASPECTS AND INTERACTIONS WITH THE
FINE PARTICLES PROBLEM
[To be held within the framework of cooperation under the United States - German
Environmental Agreement. It contributes to high priority activities identified by German Minister
Toepfer and U.S. Environmental Protection Agency (EPA) Administrator Thomas during their
December 1987 meeting. It is supported by the USEPA and the German Government (BMU and
BMBF)].
Locations: Research Triangle Park, NC, USA
Riverside, CA, USA
October 4, 1999
October 4-6, 1999
This is to announce the organization of the Sixth US/German Workshop on the
Photochemical Ozone problem, scheduled for October 4-6, 1998, in USA. The Workshop is
divided in two parts: A Session on Ozone and Fine Particles (FP) regulatory policy developments
and issues will be held on October 4 in Research Triangle Park (RTP), NC, and two Sessions, on
Ozone/FP science issues, will be held on October 4-6, in Riverside, CA. Drs, Basil Dimitriades,
USEPA, and Dieter Jost, BF, Germany, are the Environmental Agreement Project Managers.
Drs. William Carter, U. California/Riverside, Karl Becker, U. Wuppertal, Germany, and Basil
Dimitriades, are the Workshop Co-Chairs. As with the last five workshops, the scope of this
Sixth Workshop is expanded to include participation of other countries within and outside the
European Union (EU).
Regulatory policies and programs in the photochemical ozone and fine particle problem
areas have been in effect for several years in the United States and other developed countries.
Experiences to date, however, have not always conveyed a clear picture on the effectiveness of
such policies. Questions regarding the relative effectiveness of VOC and NOx emission controls
for urban ozone reduction still persist, as there are questions regarding the effectiveness of
policies that deal with the different photochemical pollution problems separately rather than
collectively. In reaction to this, the Workshop organizers are strongly interested in presentations
and discussions on regulatory policies currently in effect, with specific focus on scientific (or
other) justifications and data-supported impressions to date regarding the effectiveness of such
policies.
The Science Sessions of the Sixth Workshop will deal, as usual, with scientific issues
related to the atmospheric chemistry and physics, ambient observations, and modeling aspects of
the photochemical ozone and FP problems.
Uncertainties in the atmospheric chemistry and other scientific aspects of the ambient
ozone and FP problems still persist. These uncertainties make it difficult to formulate sound air
quality management strategies with predictable^ffectiveness. Reacting to the need for reliable
-VI-
-------
scientific data, the Workshop organizers wish to invite researchers to discuss the latest results
from basic and smog chamber studies of atmospheric photodegradation of aromatic and biogenic
VOCS, ozone-olefin reaction mechanisms, radical reactions, atmospheric chemistry of heretofore
unstudied organic air pollutants, the role of heterogenous chemistry in ambient ozone formation,
and effects of controllable and uncontrollable factors on fine particle formation in the atmosphere.
Information is also solicited on the latest developments in the air quality modeling area, especially
on development of multi-pollutant models, and development and application of VOC-reactivity
models and of observations-based models.
Presenters are requested to bring with them to the meeting extended abstracts (1,500
words at minimum) of their presentations.
-vn-
-------
AGENDA
Monday, October 4
7:30 AM Registration and Continental Breakfast - Ramona Court/Glenwood Tavern
8:00 AM Welcome and Opening
Bill Carter/Joe
Norbeck
8:05 AM
Welcome to Riverside
8:15 AM Other announcements, etc.
Gas Phase Chemistry and Modeling - Session Chair: Karl Becker
8:20 AM Overview of Atmospheric Chemistry Studies in Europe
Atmospheric Chemistry of VOCs and NOx
8:45 AM
9:10 AM
9:35 AM
10:00 AM
10:25 AM
10:45 AM
11:10 AM
11:35 AM
12:00 PM
Mayor Ronald
Loveridge
Bill Carter
Karl Becker
Roger Atkinson
Mechanisms for Air Quality Modeling: Development and Applications Bill Stockwell
of the Regional Atmospheric Chemistry Mechanism
Development and Evaluation of the SAPRC-99 Chemical Mechanism Bill Carter
Modelling Ozone Formation with a Master Chemical Mechanism Dick Derwent
Break
Simulations of EUPHORE and Field Experiments Using a Master
Chemical Mechanism
Nicola Carslaw
Influence of Biogenic VOCs on Photo-oxidant Formation: Simulation Lars Ruppert
Experiments in EUPHORE and Comparison with Model Calculations
Atmospheric Fate of Alkoxy and Carbonyl Radicals
Lunch - Spanish Art Gallery
Friedhelm Zabel
Summaries of Poster Presentations - Session Chair: Roger Atkinson
1:00 PM Atmospheric Chemistry of Selected Hydroxy-Carbonyls
1:05 PM Products and Mechanisms of the Reactions of 1,3-Butadiene with
Chlorine Atoms in Air
1:10 PM Analysis of Gas Phase Halogen Compounds Using
Atmospheric-Pressure-Ionization Mass Spectrometry
1:15 PM New Product and Aerosol Studies on the Oxidation of DMS
Roger Atkinson
Weihong Wang
Krishna Foster
Ian Barnes
-via-
-------
1:20 PM OH Formation from Ozone-Akene Reactions
1:25 PM Use of ab initio Quantum Mechanics to Estimate Rate Constants
1:30 PM A Measurement Method for Hydroxyacetone
1:35 PM Fast Gas Chromatography with Luminol Detection for Measurement of
Nitrogen Dioxide and PANs
1:40 PM Measurements of Atmospherically Relevant Gas - Species at the Ultra Trace
Level Using Laser lonization Mass Spectrometry
1:45 PM Simulation Chamber Study of Night-time Chemistry of Aldehydes and PANs
1:50 PM EPA PM Chemistry Chamber Studies
1:55 PM Development of the Master Chemical Mechanism (MCMv2.0) web site and
recent applications of its use in tropospheric chemistry models
Suzanne Paulson
David Golden
Yin-Nan Lee
Jeff Gaffney
Thorsten Benter
Francois Doussin
Erick Swartz
Nicola Carslaw
3:00 PM
3:25 PM
3:50 PM
4:15 PM
4:40 PM
5:05 PM
5:30 PM
Aromatic Hydrocarbon Research in the European Photoreactor (EUPHORE)
Studies on Oxygenated Fuel Additives: Ethers and Acetals
Development of Degradation Mechanisms of Oxygenated Hydrocarbons
Chamber Studies on the Atmospheric Oxidation of Ethers under High and
Low NOx Conditions
Atmospheric Oxidation Mechanism of Unsaturated Oxygenated VOCs
Degradation Mechanisms for the Tropospheric Oxidation of Chlorocarbons
Adjourn
Bjorn Klotz
Ian Barnes
Harald Geiger
John Wenger
Georges Le Bras
Howard Sidebottom
Tuesday, October 5
7:30 AM Continental Breakfast - Ramona Court/Glenwood Tavern
Heterogeneous Chemistry and Modeling - Session Chair: Georges Le Bras
8:00 AM The UNC Chamber Auxiliary Model (Wall Model) Harvey Jeffries
Measurements and Modeling of NOx Offgasing from FEP Teflon Reactors Bill Carter
Experimental Techniques for Studying Surface Chemistry in Smog Chambers Laura Iraci
Heterogeneous Reaction of N2O5 on Inorganic Aerosols: The Nitrate Effect Andreas Wahner
Tropospheric Aqueous Phase Chemistry Laboratory and Modelling Studies Hartmut Herrman
-ix-
8:25 AM
8:35 AM
9:00 AM
9:25 AM
-------
9:50 AM Break
Measurement Methods - Session Chair: Ulrich Platt
10:20 AM In-Situ Radical Measurements in EUPHORE
Measurement of Peroxy Radicals in the European Photoreactor EUPHORE
NOy Measurement Methods and Artifacts
10:40 AM
11:00 AM
11:20 AM
11:40PM
Instrumentation for State of the Art Aerosol Formation Measurements in a
Smog Chamber
Use of a Thermal Desorption Particle Beam Mass Spectrometer for Studies
of Secondary Organic Aerosol Formation
Ulrich Platt
Martin Heitlinger
Dennis Fitz
David Cocker
Paul Ziemann
12:00 PM Lunch - Galleria
1:00 PM Poster Viewing
Environmental Chamber Studies - Session Chair: Rich Kamens
2:00 PM Predictions of Biogenic Aerosol Formation Using a Gas Phase
Kinetic/Gas-Particle Partitioning Model and Outdoor Chamber Data
2:25 PM Measurement of Secondary Organic Aerosol Yields and Molecular Speciation
2:50 PM Smog Chamber Studies of Particle Formation from the Oxidation of Terpenes
3:15PM Break
3:35 PM Determination of Photolysis Frequencies and Quantum Yields for Small
Carbonyl Compounds using the EUPHORE Chamber
4:00 PM Investigation of Real Car Exhaust in the EUPHORE Chamber
4:25 PM Outdoor Smog Chamber Experiments in Mexico
4:50 PM Adjourn
6:30 PM Poolside Barbeque at the Victoria Club
Wednesday, October 6
7:30 AM Continental Breakfast - Ramona Court/Glenwood Tavern
Environmental Chamber Studies (continued) - Session Chair: Harvey Jeffries
8:00 AM EPA Gas Phase Chemistry Chamber Studies
8:25 AM Recent UNC Chamber Studies
Rich Kamens
David Cocker
Jens Hjorth
Klaus Wirtz
Peter Wiesen
Julio Sandoval
Deborah Luecken
Ken Sexton
-x-
-------
8:50 AM A Modeling System for Mechanism Evaluation Using Chamber Data
9:15 AM Measurement Needs for Assessing Process Analysis and Photochemical
Indicators
9:40 AM Ozone Formation in Coastal Urban Atmospheres: The Role of Anthropogenic
Sources of Chlorine
10:05 AM Break
Reactivity Studies - Session Chair: Dick Derwent
10:25 AM U.S. EPA Models-3 - Status and Applications
10:50 AM Analysis of Chamber-Derived Incremental Reactivity Estimates for N-Butyl
Acetate and 2-Butoxy Ethanol
11:15 AM Atmospheric Availability as a Component of the Ozone Formation Potential
ofaVOC
11:40 AM Presentation on California Air Resources Board Reactivity-Related Research
11:50 AM Presentation on Research Plan for the Reactivity Research Working Group
12:00 PM Lunch - Spanish Art Gallery
1:00 PM German Environmental Protection Agency Research Programs
New Chamber Projects - Session Chair: Bill Carter
1:20 PM Design of SAPHIR (Simulation of Atmospheric Photochemistry in a Large
Reaction Chamber)
1:45 PM A Research Plan for a New Environmental Chamber Facility
2:10 PM Research Plan for "Next Generation Environmental Chamber Facility for
Chemical Mechanism and VOC Reactivity Evaluation"
2:40 PM Caltech Participation in Aerosol Studies in EPA/UCR Chamber
3:05 PM Break
3:20 PM Panel Discussion on Research Needs for Environmental Chamber Studies
4:50 PM Concluding Remarks
5:00 PM Conclusion of meeting
6:00 PM CE-CERT Tour & Buffet Dinner
Paul Makar
Gail Tonnesen
Paul Tanaka
Ken Schere
Jana Milford
Jonathan Kurland
Eileen McCauley
Don Fox
Claudia Maeder
Andreas Wahner
Rafael Villasenor
Bill Carter
John Seinfeld
Karl Becker / Claudia
Maeder
Joe Norbeck / Mitch
Boretz
-XI-
-------
-------
Session I
Gas Phase Chemistry and Modeling
Session Chair
Karl Becker
-i-
-------
Overview of Atmospheric Chemistry Studies in Europe
Karl H. Becker
Physical Chemistry, University Wuppertal, Germany
email: becker^piiyschcm.uni-wuppcrtal.de
web page: http://www.physchem.uni-wuppertal.de
EU-Research Western Europe:
At the workshop you will hear about the European photoreactor EUPHORE which has been built
within an EU project before 1995. The following overview is related to programmes after 1995.
Projects within the 4th Framework Programme of the EU Tropospheric Research:
1995 - 1999: 50 Projects with 35 Mill. EURO (approx. 35 Mill Dollars)
The projects were grouped in the following clusters, each cluster was guided by two rapporteurs.
Cluster reports are available from the European Commission. A cluster consists of several
projects which are listed below. Each project is carried out by 5 - 10 European laboratories
consisting of a reasonable cross-section of European research. The project is guided by a project
coordinator. Upon request I can give the names of the co-ordinators.
Cluster 1: Oxidising Capacity of the Atmosphere and Transport of Photooxidants
HALOTROP (The contribution of reactive halogen species to the oxidation capacity of
the troposphere)
FORMONA (Formation and occurrence of nitrous acid in the atmosphere)
CO-OH-EUROPE (Development and application of 14CO methodology for validating OH
distribution incorporating a stable isotope of Europe's CO budget)
VOTALP (Vertical ozone transports in the Alps)
- HALOTROP-SALT (The contribution of reactive halogen species: sea salt aerosols:
laboratory investigations of heterogeneous halogen activation in the troposphere)
TACIA (Testing atmospheric chemistry in anticyclones)
RIFTOZ (Regional differences in tropospheric ozone in Europe - an analysis of its
controlling phenomena)
MARATHON (Marine atmosphere oxidation capacity experiment)
CATOME (Carbonyls in tropospheric oxidation mechanisms)
VOTALP II (Vertical Ozone transport in the Alps)
FIRETRACC / 100 (Firn record of trace gases relevant to atmospheric chemical change
over 100 years)
- MAXOX (Maximum oxidation rates in the free troposphere)
- HAMLET (Halogens in the marine environment: Laboratory investigation of
heterogeneous chemistry)
Cluster 2: Aerosols and Clouds in the Troposphere
- HILLCLOUD (Use of a hill cap cloud to study cloud aerosol interactions in ACE-2)
-2-
-------
- FREETROPE (Free tropospheric aerosols and their mixing with the marine boundary
layer in ACE-2)
- LAGRANGIAN (North Atlantic regional aerosol characterisation experiment (ACE-2):
continental air mass evolution in the marine boundary layer)
- SOAP (Speciation of the organic fraction of atmospheric aerosol particles related to cloud
formation)
- CIME (Cloud ice mountain experiment)
- MOD AC (Model development for tropospheric aerosol and cloud chemistry)
- HECONOS (Heterogeneous conversion of nitrogen on aerosol surfaces)
- PARFORCE (New particle formation and fate in the coastal environment)
- INTACC (An investigation into the interaction of aerosols and cold clouds)
- NUCVOC (Nucleation processes from oxidation of biogenic volatile organic compounds)
Cluster 3: VOC, NOX and Greenhouse Gases contributions from Different Sources
- BIPHOREP (Biogenic VOC emissions and photochemistry in the boreal regions of
Europe)
- MEDFLUX (Quantification of pollutant dry deposition fluxes over Mediterranean
ecosystems)
- BEMA (Biogenic emissions in the Mediterranean area)
- GEFOS (Greenhouse gas emissions from farmed organic soils)
- FOREXNOX (European Forest, as a source of atmospheric nitrogen oxides)
- 14 C- VOC (Biogenic and anthropogenic contribution to ambient volatile organic
compounds)
- ECO VOC (Parametrisation of environmental and physiological controls of VOC
emissions from Europe forests)
- VOCAMOD (Biogenic VOC emission modelling for European forest canopies)
- RICEOTOPES (Methane from rice paddies: isotopic signals, microbial pathways and
fluxes)
- BIOFOR (Biogenic aerosol formation in the boreal forest)
- EULINOX (European lightning nitrogen oxides project)
Cluster 4: Chemical Processes and Mechanisms
- INFORMATEX (Influence of fuel formulation on atmospheric reactivity of exhaust
gases)
- BIO VOC (Degradation mechanisms of biogenic VOCs)
- SARBVOC (Structure-Activity relationships for reactions in the degradation of biogenic
volatile organic compounds)
- EURO VOC (Control strategies for European air quality based on the tropospheric
oxidation characteristics of volatile organic compounds)
- RINOXA 2 (Removal and interconversions of oxidants in the atmospheric aqueous phase)
- AEROBIC (Aerosol formation from biogenic organic carbon)
- RADICAL (Evaluation of radical sources in atmospheric chemistry through chamber and
aboratory studies)
- DOMAC (DMS: Oxidation mechanisms in relation to aerosol and climate)
- DIFUSO (Diesel fuel and soot: fuel formulation and its atmospheric implications)
-3-
-------
- HALOBUD (An investigation of the tropospheric budget of halogenated compounds)
- UNARO (Uptake and nitration of aromatics in the tropospheric aqueous phase)
- AFCAR (Atmospheric fate of carbonyl radicals)
- EUROSOLV (Reduction of tropospheric ozone formation in Europe by the employment
of alternative industrial solvents)
Cluster 4 essentially represents the laboratory studies. The outcome of the work is regulary
published:
"Chemical Mechanisms of Atmospheric Processes"
(edit.: K. H. Becker and G. Angeletti)
Air Pollution Research Report 67, European Commission, EUR 18765, Brussels 1999
(Proceedings of the Workshop 1998, Copenhagen)
" Chemical Processes and Mechanisms (EU) and Chemical Mechanism Development"
EC/EUROTRAC 2 Joint Workshop
Sept. 20-22,1999, Ford Research Center Aachen, Germany
(Proceedings will be published)
Cluster 5: Measurement Techniques
- EUPHORAM (In situ EUPHORE radical measurement)
- MEDUSE (Monitoring and prediction of the atmospheric transport and deposition of
desert dust in the Mediterranean Region)
- PRICE II (Peroxy radical intercomparison Exercise II)
- DCHOR (Development of a compact transportable instrument for the measurement of
tropospheric OH and HOa on remote and airborne platforms)
- HRDLCDEM (High resolution diode laser carbon dioxide environmental monitor)
- PRIME (Peroxy radical initiative for measurements in the troposphere)
- SAMPLER (Sampling device for the measurement of peroxy radicals in atmospheric
systems)
Future EU-Research:
New European Research Programme 5th Framework Programme, Part for the Troposphere
First Phase (2000-2003)
Key Action "Global Change, Climate and Biodiversity"
• Atmospheric Composition
(12 Projects: 10 Mill. EURO)
Key Action "City of Tomorrow and Cultural Heritage"
• Local Air Pollution / Air Quality and Instrumental Developments (approx. 10 Mill. EURO)
New EU Clusters which are problem orientated (atmospheric composition)
Cluster 1: Ozone Budget
-4-
-------
- POET (Precursors of ozone and their effects in the troposphere)
- FUTURE-VOC (BVOV emissions of European forests under future CCh levels: influence
on compound composition and source strength)
- TROTREP (Tropospheric ozone and precursors - trends, budget and policy)
- SUB-AERO (Subgrid scale investigations of factors determining the occurrence of ozone
and fine particles)
- EXACT (Effects of the oxidation of aromatic compounds in the troposphere)
Cluster 2: Aerosols
- MINATROC (Mineral dust aerosol and tropospheric chemistry)
- OSOA (Origin and formation of secondary organic aerosol)
- NITROCAT (Nitrous acid and ist influence on the oxidation capacity of the atmosphere)
- NICE (The nitrogen cycle and effects on the oxidation of atmospheric trace species at
high) latitudes
Cluster 3: Climatic Impact Gases
- CUT-ICE (Chemistry of the upper troposphere: Laboratory studies of heterogeneous
processes on ice)
- IAFAEE (Impact of alternative fluorinated alcohols and ethers on the environment. A
laborators and modelling study)
- EL CID (Evaluation of the climatic impact of dimethyl sulphide)
Cluster 4: European Component of Global Observing Systems
- ERA-40 (A forty-year European re-analysis of the global atmosphere)
- MAGPROX (Screening and monitoring of anthropogenic pollution over Central Europe
by using magnetic proxies
Support for Research Infrastructures
- (EARLINET (A European aerosol research LIDAR network to establish an aerosol
climatology)
The cluster "Chemical Processes and Mechanisms" will not continue. Laboratory based projects
will receive less support in the future!
Germany:
German Research Programme (BMBF) on-going:
• Tropospheric Research (mid 1996-mid 2000: 40 Mill. DM/23 Mill. $)
• Aerosol Research (1997 - 2000: 20 Mill. DM/11.5 Mill. $)
-5-
-------
• Stratospheric Research (1989 - 1999: 100 Mill. DM/58 Mill. $)
The structure of the research programmes were differently organised, between single projects,
weakly co-ordinated (no financial support for co-ordination) and co-ordinated with significant
financial support.
The Tropospheric Research Programme (TFS) is well co-ordinated in three parts: with nearly
equal financial support:
• CT Modelling and Model Developments
• Emission Studies
• Process Analysis (Laboratory studies and the field campaign BERLIOZ-Berlin Ozone)
Future Atmospheric Research in Germany (BMBF):
Atmospheric Research AFO 2000 (2000-2005): 100 Mill. DM (approx. 58 Mill. $)
• Improvement of the Understanding of the Atmospheric System
(mainly world-wide field campaigns)
• Development and Provision of Tools for Environmental Policy
(meanly model evaluation and developments, studies of biogenic emissions, studies and
vertical transport processes)
• Generous Support of Young Scientists in the Field of Atmospheric Research
Laboratory based projects will not be supported within the new BMBF-Programme; laboratory
work should be supported by the German National Science Foundation (DFG).
In the USA a similar discussion took place as reported hi Nature, Vol. 400, 492 (1999).
Conclusions:
During the last 10 years atmospheric science has been quite well supported by the EU as well as
by Germany (BMBF). In particular, kinetic and mechanistic studies in the laboratory, during the
first stage mainly in the gas phase but in the second stage also in the aqueous phase and
heterogeneous phase (multi-phase) received significant support. During the coming years the
support for atmospheric science seems to be reasonable, however, other areas like soil and water
pollution and biodiversity is also attracting a larger proportion of the available resources.
Laboratory studies of atmospheric chemical systems will certainly receive less support in the
future by the EU as well as by Germany (BMBF). At present it is still an open question how
much support can be provided by the National Science Foundation (DFG) in Germany. In France
laboratory studies are regularly supported by CNRS, however, on a low level. In the UK a new
programme has been submitted to NERC, the outcome remains open. European Chemical
Industry represented by CEFIC does not see it as a priority to support atmospheric science;
ecotoxicological studies have higher priorities.
At present integrated research projects combining laboratory work, field studies and modelling
might attract further support. The knowledge on reaction kinetics and product yields of
atmospheric processes seems to be sufficiently developed, at least as viewed outside from of
chemistry circles.
-6-
-------
ATMOSPHERIC CHEMISTRY OF VOCs AND NOX
Roger Atkinson
Air Pollution Research Center,
Department of Environmental Sciences,
and
Department of Chemistry
University of California
Riverside, CA 92521
Introduction
Large quantities of volatile organic compounds (VOCs), are emitted into the troposphere from biogenic
and anthropogenic sources [1,2]. In the troposphere, VOCs (and their reaction products) are removed by the
physical processes of wet and dry deposition [3] (not discussed here), and are degraded by the chemical processes
of photolysis, reaction with hydroxyl (OH) radicals, reaction with nitrate (NO3) radicals and reaction with 03
[4,5]. Additionally, reactions of Cl atoms with alkanes and other VOCs have been observed in the lower Arctic
troposphere during springtime [6], and it is possible that similar occurrences occur in other localities [7].
The atmospheric chemistry of the various classes of VOCs have been discussed in recent reviews and
evaluations [4,5,8,9]. Photolysis and the initial reactions of many VOCs with OH radicals and NO3 radicals lead
to the formation of alkyl or substituted alkyl (R) radicals, and the reactions of O3 with alkenes and other VOCs
containing >C=C< bonds lead to the formation of organic peroxy (RO2) radicals. The tropospheric degradation
reactions of VOCs can be schematically represented (at least in part) by Reaction Scheme 1, with the important
intermediate radicals being alkyl or substituted alkyl radicals (R), alkyl peroxy or substituted alkyl peroxy radicals
(RO2), and alkoxy or substituted alkoxy radicals (RO).
VOC
ROOH
carbonyl
alcohol
RON0
products
Reaction Scheme 1
-7-
-------
However, the formation of biradicals from the O3 reactions with alkenes, and the formation of
hydroxycyclohexadienyl-type radicals from the OH radical reactions with aromatic hydrocarbons (see below) lead
to intermediate species whose subsequent reactions differ from those of, RO2 and RO radicals formed from
alkanes, alkenes and other VOCs. As may be expected from Reaction Scheme 1, the tropospheric reactions of
VOCs share many reaction sequences in common, and certain areas of uncertainty which affect the products
formed are also common to almost all VOCs. These areas of uncertainty include those involving the reactions of
RO2 and RO radicals.
Reactions of Organic Peroxy Radicals
Organic peroxy radicals react with NO, NO2, HO2 radicals, organic peroxy radicals and NO3 radicals.
M
ROo-l-NO
RONO2
RO + NO2
(la)
(Ib)
RO2 + NO2
ROONO2
RO2 + HO2
ROOH + O2
(3)
RCH(OO)R' + RCH(OO)R'
RCH(0)R' + RCH(0)R' + O2
RCH(OH)R' +RC(O) +O2
(4a)
(4b)
ROo+NOi
RO + NO2 + O2 (plus other products)
(5)
Alkyl peroxynitrates thermally decompose rapidly back to reactants at around room temperature, and hence the
R02 + NO2 reaction is unimportant in the lower troposphere for R = alkyl or substituted alkyl (but not for R =
acyl). In the troposphere, important daytime reactions of RO2 radicals are with NO and HO2 radicals, and the
competition between these reactions determines whether net O3 formation or net O3 destruction occurs. Kinetic
and product data for RO2 radical reactions with NO and with the HO2 radical are available mainly for alkyl
peroxy radicals formed from alkanes [8,9], and kinetic and product data for the reactions of a wide variety of
organic peroxy radicals with NO and HO2 radicals are needed. When NO3 radicals are present at significant
concentrations, NO concentrations will be low, and hence nitrooxyalkyl peroxy radicals are expected to react
primarily with NO2, to form thermally unstable peroxy nitrates, and with HO2 and NO3 radicals, with self-
reactions of nitrooxyalkyl peroxy radicals or combination reactions with other peroxy radicals being important in
laboratory studies.
Organic Nitrate Formation
Organic nitrates (RONO2) are formed in one channel [reaction (la)] of the reaction of organic peroxy
radicals with NO. The formation yields of organic nitrates are significantly temperature and pressure dependent,
decreasing with increasing temperature and with decreasing pressure [10]. To date, measured organic nitrate
formation yield data are available only for secondary alkyl peroxy radicals formed from the NOx-air
photooxidations of C2-Cg n-alkanes and for a number of other alkyl and -hydroxyalkyl peroxy radicals formed
-8-
-------
from alkanes and alkenes [8,10-12]. Additional yield data are needed for organic nitrates formed from the wide
variety of VOCs observed in ambient air.
Reactions ofAlkoxy Radicals
As shown in Reaction Scheme 1, alkoxy radicals are formed as intermediates during the tropospheric
degradations of VOCs. The subsequent reactions of alkoxy radicals determine the products formed (and, in the
presence of NO, the amounts of O3 formed). In the troposphere, alkoxy radicals can react with O2, unimolecularly
decompose by C-C bond scission, isomerize by a 1,5-H shift through a 6-membered transition state, or, for alkoxy
radicals of structure RC(O)OCH(O)R', undergo a rearrangement to form RC(O)OH plus R'CO [8,13,14].
The alkyl and -hydroxyalkyl radicals formed from the decomposition and isomerization pathways,
respectively, react further by reaction schemes analogous to those shown in Reaction Scheme 1. Thus, for
example, the subsequent reactions of the CH3CH(OH)CH2CH2CH2 radical formed by isomerization of the 2-
pentoxy radical in the presence of NO can lead to formation of 5-hydroxy-2-pentanone, as shown in Reaction
Scheme 2.
CH3CH(OH)CH2CH2CH2 + °2
CH3CH(OH)CH2CH2CH202
NO _
CH3CH(OH)CH2CH2CH2O
isomerization
CH3CH(OH)CH2CH2CH2OH
0^
CH3C(O)CH2CH2CH2OH + HO2
Reaction Scheme 2
Reactions of -hydroxyalkyl radicals such as CH3C(OH)CH2CH2CH2OH (which are also formed by H-atom
abstraction from the C-H bonds of >CHOH and -CH2OH groups in alcohols) with O? lead to the carbonyl plus
the H02 radical [4,9].
CH3C(OH)CH2CH2CH2OH + O2 CH3C(O)CH2CH2CH2OH + HO2 (6)
The products formed from alkoxy radicals therefore depend on the reaction pathway occurring. There is now a
semi-quantitative (or better) understanding of the tropospheric reactions of alkoxy radicals formed from alkanes
and of -hydroxyalkoxy radicals formed from the OH radical-initiated reactions of alkenes [8,13], and this
understanding appears to hold for substituted alkoxy radicals formed from certain other classes of VOCs.
However, we do not have such a quantitative understanding of the reactions of alkoxy radicals of structure
-9-
-------
>C(O)OR (R = alkyl) formed from ethers and glycol ethers, nor of the reactions of alkoxy radicals of structure -
R'C(O)OCH(O)R (R = alkyl) formed from esters. This may also be the case for alkoxy radicals containing other
structural features.
In addition to reaction with O2, decomposition by C-C bond scission and isomerization through a 6-
membered transition state, halogenated alkoxy radicals formed from haloalkanes and haloalkenes can also
decompose by elimination of a Cl or Br atom.
The above generalized reactions hold for alkanes, for the OH and NO3 radical-initiated reactions of
alkcnes and of certain VOCs containing >C=C< bonds, and for the OH radical-initiated reactions of many
oxygenated VOCs. However, as noted above, the reaction mechanisms of the reactions of 03 with alkenes and
VOCs containing >C=C< bonds and of the OH radical with aromatic hydrocarbons are significantly different
[4,5], and these reaction systems are briefly discussed below.
Reactions of Alkenes and VOCs containing >C=C< bonds with O3
O3 initially adds to the >C=C< bond to form an energy-rich primary ozonide, which rapidly decomposes
as shown in Reaction Scheme 3 to form two sets of (carbonyl + biradical), where [ ]J denotes an energy-rich
species [8]. Recent studies have shown that the relative importance of the two decomposition pathways of the
primary ozone to form the two (carbonyl plus biradical) products depends on the structure of the alkene [8].
C = C
0
Ri —
o
c
\
o
c •
I
R4
RjC(O)R2 + [R3R4COO]*
Reaction Scheme 3
R3C(O)R4
The biradicals can be collisionally stabilized or decompose by a number of pathways, as shown in Reaction
Scheme 4. The isomerization/decomposition reactions of the "energy-rich" biradicals include the "ester channel"
and the "hydroperoxide" channel, as shown in Reaction Scheme 4. O(3P) atom elimination has not been observed
for alkenes at room temperature and atmospheric pressure of air [8]. However, OH radicals are formed from the
reactions of O3 with alkenes, often in unit or close to unit yield [8]. This formation of OH radicals from the
reactions of O3 with alkenes leads to secondary reactions of the OH radical with the alkene and, unless the OH
radicals are scavenged, the O3 reaction involves OH radical reactions and hence the products observed and their
yields may not be those for the O3 reactions.
-10-
-------
[RjCH2C(R2)OO]* + M
[R]CH2C(R2)OO]*
[R!CH2(R2)OO]*
RjCH2C(R2)OO + M
RjCH2 O
X
O
[R1CH=C(OOH)R2]*
[R]CH2C(O)OR2]*
decomposition
PRODUCTS
(including R]CH3 if R2=H)
R2CHC(O)R2 + OH
Reaction Scheme 4
The formation of OH radicals is postulated to occur via the "hydroperoxide" channel, as shown in Reaction
Scheme 4. Assuming that the OH radicals observed to be formed in these reactions arise via the hydroperoxide
channel, then the organic co-product radicals (for example, the CH3C(O)CH2 radical formed together with the OH
radical from the [(CH3)2COO]' biradical) will react as described in Reaction Scheme 1, leading to a variety of
carbonyl, hydroxycarbonyl and hydroperoxycarbonyl products [8,15]. In the atmosphere, the thermalized
biradicals appear to react dominantly with water vapor, leading to formation of a carboxylic acid plus H2O (if
possible) or to a carbonyl plus H2O2. While there have been significant advances in our knowledge of the
isomerization and/or decomposition reactions of the energy-rich biradicals over the past few years, there is still a
need to quantitatively elucidate these reactions.
Aromatic Hydrocarbons
Benzene and the alkyl-substituted benzenes such as toluene, xylenes, and trimethylbenzenes react with
OH radicals and NO3 radicals [4,5], with the OH radical reactions dominating as the tropospheric loss process.
The OH radical reactions proceed by H-atom abstraction from the C-H bonds of the alkyl substituent groups (or
from the C-H bonds of the aromatic ring in the case of benzene), and by OH radical addition to the aromatic ring
to form a hydroxycyclohexadienyl-type radical (OH-aromatic adduct) [4,5]. The H-atom abstraction pathway
accounts for 10% of the overall OH radical reactions with benzene and the methyl-substituted benzenes at room
temperature and atmospheric pressure [4]. The tropospheric reactions of the benzyl and alkyl-substituted benzyl
radicals formed from the pathway involving H-atom abstraction from the alkyl-substituent groups are analogous to
those for alkyl radicals discussed above [4,5].
The major reaction pathway involves the formation of the OH-aromatic adduct(s). The OH-benzene, OH-
toluene and OH-xylene adducts do not react with NO, but do react with O2 and NO2 [4,5,16]. The kinetic data
show that in the lower troposphere, including polluted urban airmasses, the dominant reaction of OH-aromatic
adducts is with O2. The uncertainty in the tropospheric degradation mechanism of aromatic hydrocarbons
concerns the products and mechanisms of the reactions of the OH-aromatic adducts with NO2 and, especially, O2.
Reaction Scheme 5 shows postulated reactions of the OH-benzene adduct with O2, leading to phenol, an epoxide-
alkoxy radical, a bicycloalkyl radical, aperoxy radical, and benzene oxide/oxepin [17-19] (noting that formation of
the methyl-substituted benzene oxide/oxepin has been shown to be unimportant in the toluene reaction [20]).
-11-
-------
OH
\H
02
.OH
+ H02
0
H-
rOH
~-H
+ H02
Reaction Scheme 5
-12-
-------
The potential subsequent reactions of the epoxide-alkoxy radical, the bicycloalkyl radical, and the peroxy
radical species formed in Reaction Scheme 5 lead to the formation of ring-opened unsaturated carbonyls,
dicarbonyls and epoxy-carbonyls. Product studies carried out to date under simulated atmospheric conditions for
benzene, toluene, and the xylenes generally account for only 30-50% of the reaction products [4,5], which include
phenolic compounds and ring-opened dicarbonyls [4,5,21]. 3-Hexene-2,5-dione, the expected co-product to the a
-dicarbonyl glyoxal, is formed in similar yield to gloxal from/7-xylene [22], evidence for formation of the
bicycloalkyl radical intermediate. The formation of other ring-opened unsaturated carbonyls have been reported
[4,5,21,23-25], although no quantitative yield data have been published to date. The unsaturated 1,4-dicarbonyls
and di-unsaturated 1,6-dicarbonyls formed from aromatic hydrocarbons are highly reactive. Mechanistic and
product data (and especially quantitative data for the formation yields of the various potential unsaturated
dicarbonyls and unsaturated epoxycarbonyls) are clearly needed.
References
1. A. Guenther, C.N. Hewitt, D. Erickson, R. Fall, C. Geron, T. Graedel, P. Harley, L. Klinger, M. Lerdau,
W.A. McKay, T. Pierce, B. Scholes, R. Steinbrecher, R. Tallamraju, J. Taylor, P. Zimmermann, J.
Geophys. Res. 100, 8873-8892 (1995).
2. National Research Council, Rethinking the Ozone Problem in Urban and Regional Air Pollution,
National Academy Press, Washington, DC (1991).
3. T.F. Bidleman, Environ. Sci. Technol. 22, 361-367 (1988).
4. R. Atkinson, J. Phys. Chem. Ref. Data Monograph 2, 1-216(1994).
5. R. Atkinson, Atmos. Environ., in press (2000).
6. L. Barrie, U. Platt, Tellus B 49, 450-454 (1997).
7. K. Kreher, P.V. Johnston, S.W. Wood, B. Nardi, U. Platt, Geophys. Res. Lett. 24, 3021-3024 (1997).
8. R. Atkinson, J. Phys. Chem. Ref. Data 26, 215-290 (1997).
9. R. Atkinson, D.L. Baulch, R.A. Cox, R.F. Hampson, Jr., J.A. Kerr, M.J. Rossi, J. Troe, J. Phys. Chem.
Ref. Data 28, 191-393 (1999).
10. W.P.L. Carter, R. Atkinson, J. Atmos. Chem. 8, 165-173 (1989).
11. J.M. O'Brien, E. Czuba, D.R. Hastie, J.S. Francisco, P.B. Shepson, J. Phvs. Chem. A 102, 8903-8908
(1998).
12. X. Chen, D. Hulbert, P.B. Shepson, J. Geophys. Res. 103, 25563-25568 (1998).
13. R. Atkinson, Int. J. Chem. Kinet. 29, 99-111 (1997).
14. E.G. Tuazon, S.M. Aschmann, R. Atkinson, W.P.L. Carter,./. Phys. Chem. A 102, 2316-2321 (1998).
15. H. Niki, P.O. Maker, C.M. Savage, L.P. Breitenbach, M.D. Hurley,/ Phvs. Chem. 91, 941-946 (1987).
16. R. Knispel, R. Koch, M. Siese, C. Zetzsch, Ber. Bunsenges. Phys. Chem. 94, 1375-1379 (1990).
17. L.J. Bartolotti, E.G. Edney, Chem. Phys. Lett. 245, 119-122 (1995).
18. J.M. Andino, J.N. Smith, R.C. Flagan, W.A. Goddard III, J.H. Seinfeld, J. Phvs. Chem. 100, 10967-
10980(1996).
-13-
-------
19. B. Klotz, I. Barnes, K.H. Becker, B.T. Golding, J. Chem. Soc. Faraday Trans. 93, 1507-1516 (1997).
20. B. Klotz, S. S0rensen, I. Barnes, K.H. Becker, T. Etzkorn, R. Volkamer, U. Platt, K. Wirtz, M. Martin-
Reviejo.y. Phys. Chem. A 102, 10289-10299 (1998).
21. A. Bierbach, 1. Barnes, K.H. Becker, B. Klotz, E. Wiesen, Proc. 6th European Symposium on the
Physico-Chemical Behaviour of Atmospheric Pollutants, G. Angeletti, G. Restelli, Eds., European
Commission, pp. 129-136 (1994).
22. H.L. Bethel, J. Arey, R. Atkinson, J. Phys. Chem. A, submitted for publication (2000).
23. E.S.C. Kwok, S.M. Aschmann, R. Atkinson, J. Arey, J. Chem. Soc. Faraday Trans. 93, 2847-2854
(1997).
24. J. Yu, H.E. Jeffries, K.G. Sexton, Atmos. Environ. 31, 2261-2280 (1997).
25. J. Yu, H.E. Jeffries, Atmos. Environ. 31, 2281-2287 (1997).
-14-
-------
Mechanisms for Air Quality Modeling:
Development and Applications of the
Regional Atmospheric Chemistry Mechanism
William R. Stockwell
Division of Atmospheric Sciences, Desert Research Institute, 2215 Raggio Parkway,
Reno, NV 89512-1095, USA; tel. 775-674-7058; fax 775-674-7008;
E-mail wstock@dri.edu
ABSTRACT
The Regional Atmospheric Chemistry Mechanism (RACM) is a highly revised version
of the RADM2 mechanism (RACM; Stockwell et al., 1997). The changes to the
inorganic chemistry were relatively minor but there were substantial changes for many
organic compounds. These revisions included improvements to the mechanisms for the
oxidation of alkanes by HO radical, the ozonolysis of alkenes, the reaction of alkenes
with NOs radical, peroxy radical reactions, aromatic chemistry and better treatment of
the chemistry of isoprene and terpenes.
The RACM mechanism has been applied to evaluate incremental reactivities for the
production of ozone from volatile organic compounds and to assess the aerosol particle
formation reactivity of nitrogen oxide (NOX) emissions (Stockwell et al., 1999a;b). The
concept of incremental reactivity was extended to multi-day scenarios. Ethane and
acetone are regarded as unreactive compounds but for a five day scenario their reactivity
is appreciable. The maximum ozone incremental reactivities (MOIR) of ethane, acetone
and dimethyoxymethane (a proposed low reactivity replacement solvent) have
incremental reactivities that are about equal for a scenario with a duration of six days.
A similar approach was developed to assess the aerosol particle formation reactivity of
nitrogen oxide (NOX) emissions for wintertime conditions in central California
(Stockwell et al., 1999b). Our calculations found that about 33% of emitted NOX were
converted to particulate nitrate on a molar basis and about 0.6 g of ammonium nitrate is
produced for each gram of NOX emitted (the mass of NOX calculated as NO2). This
estimate is in reasonable agreement with field measurements.
INTRODUCTION
The gas-phase chemical mechanism is one of the most important components of an
atmospheric chemistry model. It is difficult to include all significant chemical reactions
in an air quality model because the organic chemistry of the polluted atmosphere is very
complicated. There are large numbers of emitted volatile organic compounds (VOC)
and each compound's degradation mechanism may include many chemical
intermediates and reactions.
The Regional Atmospheric Chemistry Mechanism (RACM; Stockwell et al., 1997) is a
compromise between chemical detail and accurate chemical predictions. The RACM is
a new version of the Regional Acid Deposition Model mechanism (RADM2; Stockwell
et al., 1990). There were relatively minor changes made to the inorganic chemistry but
-15-
-------
the mechanisms for many organic compounds were highly revised. The alkane
chemistry was revised to decrease the ratio of the yields of aldehydes to ketones. The
yield of HO from the reaction of alkenes with ozone was increased. The branching ratios
for the reaction of acetyl peroxy radicals with NO and NO2 were revised and the
reactions of organic peroxy radical + NOs reactions were added and these changes cause
predicted concentrations of peroxyacetyl nitrate (PAN) from RACM to be lower than
RADM2. The reactions of unbranched alkenes with NOs produce relatively large
amounts of nitrates during the nighttime and this is now included. The RACM
mechanism has a new condensed mechanism for aromatics, isoprene and terpenes but
further improvement will be necessary in view of recent laboratory measurements. The
RACM is becoming widely used in Europe, Asia and the United States. It has been
accepted as the baseline mechanism for the German Tropospheric Research Program.
The RADM2 and the new RACM mechanisms were tested against the same set of
representative environmental chamber experiments used previously to evaluate the
RADM2 (Stockwell et al., 1990). An agreement of ±30% in peak ozone concentrations
is within the limits imposed by uncertainties in the chamber experimental characteristics
including wall loss, wall radical sources, actinic flux and initial conditions. The peak
ozone concentrations predicted by the RACM mechanism had a mean normalized
deviation of 19% from the chamber experiments while the mean normalized deviation
was 13% for the RADM2 mechanism (Stockwell et al., 1997). The RACM and the
RADM2 mechanisms predict peak NO2 concentrations and the timing that the peak
occurs very well. The mean normalized deviations of NO2 peak concentrations from the
chamber experiments are both near 10%. The mean normalized deviations of timing of
the NO2 peak concentrations for both mechanisms are near 3%.
The concentrations of O3, H2O2, H2SO4, HNO3 and PAN predicted by the RACM and
the RADM2 mechanisms were compared for a set of urban and rural conditions
(Stockwell et al., 1997). Although the differences between the two mechanisms for O3,
H2O2, H2SO4 and HNO3 are small the differences in predicted PAN concentrations are
very significant. The average predicted noontime PAN concentration ratios of RACM to
RADM2 were 0.58.
MECHANISM APPLICATIONS AND METHODS
The RACM mechanism has been applied to the calculation of incremental reactivity and
the equivalence between NOX emissions and ammonium nitrate particle formation.
Incremental reactivity, IR, is the change in the peak ozone concentration, A[O3], divided
by an incremental change in the VOC present in the atmosphere, AfVOC], (Carter,
1994).
IR = A[O3] / A[VOC]
A base scenario of initial VOC concentrations and emissions is defined and a NOX
emission rate that gives the highest peak ozone concentration for the available VOC is
determined. This scenario is known as the maximum ozone incremental reactivity
scenario (MOIR). Another scenario with the NOX emission rate adjusted to give the
greatest possible incremental reactivity for VOC mixture is known as the maximum
-16-
-------
incremental reactivity scenario (MIR). The RACM mechanism was modified to include
with an explicit chemical mechanism for dimethoxymethane and acetone (Stockwell et
al., 1999a). Simulations were made to determine the incremental reactivity of these
compounds for single and multiple day scenarios.
The box model with the RACM mechanism was modified to investigate the equivalence
between NOX emissions and ammonium nitrate particle formation (Stockwell et al.,
1999b). The model included the heterogeneous reaction of N2O5 with water, deposition
loss processes for ozone, NO, NO2, PANs, HNO3, MONO, N2O5, organic nitrates, SO2,
SO4~, H2O2 and organic peroxide and the yield of ammonium nitrate particles from
nitric acid. The yield of nitric acid was defined to be the fraction of gaseous nitric acid
that combines with ammonia to form paniculate ammonium nitrate for the conditions of
central California. The yields were calculated with the Simulating Composition of
Atmospheric Particles at Equilibrium model, version 2 (SCAPE-2) (Kim and Seinfeld
1993a,b; 1995).
RESULTS AND CONCLUSIONS
Single day scenarios do not give a complete picture of a VOC's incremental reactivity.
Our calculations showed that the incremental reactivity of ethane, acetone and
dimethoxymethane increase as the duration of the scenario increases (Stockwell et al.,
1999a). The MOIR of dimethoxymethane was greater for scenarios between one and six
days but the three compounds reached almost the same MOIR value for the six day time
period. The MIRs of ethane, acetone and dimethoxymethane increased as the duration of
the scenario increased. Although the MIR of dimethoxymethane was greater than the
MIR of ethane for all six scenarios the MIR of acetone increased so that its MIR became
greater than the MIR of dimethoxymethane for scenarios lasting longer than 2 days. This
result suggests that even unreactive compounds may become relatively reactive over
multiple day scenarios. Ozone control strategies that intend to reduce the concentrations
of highly reactive VOC by increasing the use of "unreactive VOC' in urban regions
might even increase ozone concentrations over the regional scale.
A photochemical box model and an equilibrium ammonium nitrate particulate model
were used to investigate the wintertime equivalence between NOX emissions and
ammonium nitrate particle formation (Stockwell et al., 1999b). During the wintertime:
the total ammonia to nitrate ratio was greater than one; the relative humidity was greater
than 70%; and the temperature was less than 292 K in central California. Under these
conditions approximately 80% of the nitric acid resulting from NOX emissions was
calculated to be in the particulate nitrate phase. It was calculated that 33 ± 7 % of the
moles of emitted NOX are converted to particulate nitrate. This corresponds to a particle
equivalent of NOX emissions on the order of 0.57 ±0.13 g of ammonium nitrate for each
gram of NOX emitted (where the mass of NOX emissions was calculated as NO2). The
particle equivalent of NOX emissions was most sensitive to uncertainties in the mixing
height which affected the deposition of nitrogenous species. The calculated equivalence
between NOX emissions and ammonium nitrate particle formation values were in
reasonable agreement with field measurements for central California.
-17-
-------
ACKNOWLEDGEMENTS
Support for this research was provided by the German Bundesministerium fur Bildung,
Wissenschaft, Forschung und Technologic (BMBF), project "Tropospharenforschung
(TFS/LT3): ProzeBstudien zur Oxidatienbildung und Oxidationskapazitat" to the
Fraunhofer Institute for Atmospheric Environmental Research in Garmisch-
Partenkirchen, Germany, U.S. Generating Company and by Lambiotte & Cie S.A,
Bruxelles, Belgium.
REFERENCES
Carter W.P.L. (1994) Development of ozone reactivity scales for volatile organic
compounds. J. Air and Waste Management Assoc., 44, 881-899.
Kim, Y.P., Seinfeld, J.H., and Saxena, P. (1993b) Atmospheric gas-aerosol equilibrium
I. Thermodynamic model. Aerosol Sci. Technol. 19,157-181.
Kim, Y.P., Seinfeld, J.H., and Saxena, P. (1993a) Atmospheric gas-aerosol equilibrium
II. Analysis of common approximations and activity coefficient calculation methods.
Aerosol Sci. Technol. 19, 182-198.
Kim, Y.P. and Seinfeld, J.H. (1995) Atmospheric gas-aerosol equilibrium ID.
Thermodynamics of crustal elements Ca2+, K+, and Mg2+. Aerosol Sci. Technol. 22,
93-110.
Stockwell, W.R., F. Kirchner, M. Kuhn, and S. Seefeld (1997) A New Mechanism for
Regional Atmospheric Chemistry Modeling, /. Geophys. Res., 102, 25847-25879.
Stockwell, W.R., H. Geiger and K. H. Becker (1999a) Estimation of Incremental
Reactivities for Multiple Day Scenarios: An Application to Ethane, Acetone and
Dimethyoxymethane, submitted, Atmos. Environ.
Stockwell, W.R., J. G. Watson, N.F. Robinson, W. Steiner and W.W. Sylte (1999b) The
Ammonium Nitrate particle equivalent of NOX emissions for continental wintertime
conditionsl, submitted to Atmos. Environ.
-18-
-------
DEVELOPMENT AND EVALUATION OF THE SAPRC-99
CHEMICAL MECHANISM
William P. L. Carter
Air Pollution Research Center and College of Engineering Center for
Environmental Research and Technology, University of California, Riverside,
CA 92521
A. Introduction
Airshed models are essential for the development of effective control strategies for
reducing photochemical air pollution because they provide the only available scientific basis for
making quantitative estimates of changes in air quality resulting from changes in emissions. The
chemical mechanism is the portion of the model that represents the processes by which emitted
primary pollutants, such as volatile organic compounds (VOCs) and oxides of nitrogen (NOX),
interact in the gas phase to form secondary pollutants such as ozone (O3) and other oxidants. This
is an important component of airshed models because if the mechanism is incorrect or incomplete
in significant respects, then the model's predictions of secondary pollutant formation may also be
incorrect, and its use might result in implementation of inappropriate or even counter-productive
air pollution control strategies.
One example of a chemical mechanism for airshed models is the SAPRC-90 chemical
mechanism (Carter, 1990). Unlike most other chemical mechanisms developed at the time, which
were designed for efficient representation of complex ambient mixtures, SAPRC-90 was designed
for the purpose of assessing differences in atmospheric impacts of individual VOCs. The
SAPRC-90 mechanism had assigned or estimated mechanisms, for over 100 types of VOCs.
Although other state-of-the art mechanisms were available for airshed model applications (e.g.,
Gery et al, 1998, Stockwell et al, 1990), SAPRC-90 used for this purpose because it was the only
mechanism that that represented a large number of VOCs that was evaluated against
environmental chamber data. However, although this mechanism represented the state of the art
at the time it was developed, since then there has been continued progress in basic atmospheric
chemistry, and new information has become available concerning the reactions and O3 impacts of
many individual VOCs.
This mechanism has been updated several times to incorporate some of the new
information that has become available, with the major documented updates being the
"SAPRC-93" (Carter et al, 1993; Carter, 1995) and the "SAPRC-97" (Carter et al, 1997a)
versions. However, the reactions and rate constants for most of the inorganic species and common
organic products have not been updated, and the latest documented update (SAPRC-97) does not
incorporate important new information concerning mechanisms and reactivities of many classes
of VOCs.. This includes particularly improved estimation methods and new reactivity data on
many types of oxygenated VOCs that have not previously been studied but that are or may be
important in stationary source emissions, and updated mechanisms for components of mineral
spirits and other high molecular weight alkanes.
Because of this, an updated mechanism that represents the current state of the art has
been needed for state-of-the art VOC reactivity assessment. To address this, a completely updated
-19-
-------
version of the SAPRC mechanism, designated SAPRC-99, was developed. This updated
mechanism is comprehensively documented in a report to the California Air Resources Board
(Carter, 1999 - see also http://cert.ucr.edu/~carter/reactdat.htm), and its major features are
summarized in this paper.
B. The Base Mechanism
The base mechanism is the portion of the mechanism that represents the reactions of the
inorganics and the common organic products. SAPRC-99 incorporates the first complete update of
the base mechanism since SAPRC-90 was developed. The IUPAC (Atkinson et al, 1997, 1999) and
NASA (1997) evaluations, the various reviews by Atkinson (1989, 1991, 1994, 1997a), and other
available information were used to update all the applicable rate constants, absorption cross
sections, quantum yields, and reaction mechanisms where appropriate. Although many small
changes were made, most are not considered to have obviously important impacts on reactivity
predictions. The one possible exception is the -30% reduction in important OH + NO2 rate constant
based on the new evaluation by NASA (1997)'. However, a complete analysis of the effects of all
the changes has not been carried out, so the possibility that other changes to the base mechanism
may be important cannot be ruled out.
The base mechanism was also modified to improve somewhat the accuracy and level of
detail in the mechanism in representing no-NOx or low-NOx conditions. The methyl peroxy and
acetyl peroxy radical model species are not represented explicitly, without using "operator"
approximations or the steady-state approximation that was incorporated in previous mechanisms.
This should give somewhat more accurate predictions of radical fates and C, product formation
yields under low NOX or nighttime conditions when peroxy + peroxy reactions become
nonnegligible. The explicit treatment of methyl peroxy is based on the approach used in the
RADM-2 mechanism (Stockwell et al, 1990), which was shown to give a good approximation to
a version of the mechanism with explicit representation of all peroxy + peroxy reactions (Carter
and Lurmann, 1990). However, "operator" and steady state approximation methods are still
employed to represent the higher peroxy radicals, and the current mechanism, like the previous
versions, is still not capable of predicting how the C2+ organic products may differ under
conditions where peroxy + peroxy reactions compete with peroxy + NO reactions. But
approximations have little or no effect on predictions of O3 formation or O3 reactivities,
especially for the relatively high NOX scenarios used for calculating the MIR scale (Carter, 1994),
and significantly reduce the number of active species that need to be included in the mechanism.
The number of model species used to represent the reactive organic products was
increased somewhat in this version of the mechanism. These are listed in Table 1. With the model
species that were not in the SAPRC-90 base mechanism being indicated in bold font. The
additional species include (1) more species to represent low NOX reactions of Q radicals as
indicated above; (2) explicit representation of biacetyl and methyl glyoxal; (3) separate species to
represent the various types of uncharacterized aromatic fragmentation products (with the methyl
glyoxal model species no longer being used for this purpose); and (4) product species needed to
represent the reactions of isoprene, based on the "4-product" mechanism of Carter (1996). The
latter was added because of the importance of isoprene in the emissions in many regional models,
with the more detailed product representation used for greater accuracy and the ability to test
model predictions against ambient data for isoprene's major products.
1 The high rate constant in the current IUPAC (Atkinson et al, 1997) evaluation is probably
inappropriate (Golden, 1999).
-20-
-------
Table 1. Model species used to represent
mechanism. Model species not in
in bold font.
reactive organic products in the SAPRC-99
the SAPRC-90 base mechanism are indicated
• Formaldehyde •
• Lumped C3+ Aldehydes •
• Ketones, etc with kOH < 5x 1 0"' 2 cm3 molec" •
2 sec'1
• Methanol •
• Higher Organic Hydroperoxides
• Lumped Organic Nitrates •
• PPN And Other Higher Alkyl PAN •
Analogues
• Methacrolein •
• Other Isoprene Products •
• Glyoxal •
• Biacetyl •
• Photoreactive Aromatic Fragmentation •
Products (ec-Dicarbonyl Spectrum)
• Phenol •
• Nitrophenols •
Acetaldehyde
Acetone
Ketones, etc with kOH > 5xlO"12 cm3
molec"2 sec"1
Methyl Hydroperoxide
Peroxy Acetyl Nitrate
PAN Analogues Formed From Aromatic
Aldehydes
Methyl Vinyl Ketone
Methacrolein PAN Analogue
Methyl Glyoxal
Non-Photoreactive Aromatic
Fragmentation Products
Photoreactive Aromatic Fragmentation
Products (Acrolein Spectrum)
Cresols
Aromatic Aldehydes
Although the base mechanism for SAPRC-99 is more detailed than previous versions in
most respects, a few condensations were employed. The separate model species used to predict
formation of low-reactivity CrC3 organic nitrates in the reactions of peroxy radicals with NO was
lumped with the model species used to predict the formation of higher nitrates in these reactions
because of the low total yield of the low reactivity nitrates. The PAN analogue formed from
glyoxal, GPAN, is now lumped with the rest of the higher PAN analogues because of the
relatively low amounts of GPAN predicted to be formed in atmospheric simulations. The effects
of these approximations, which resulted in fewer species and significantly fewer reactions in the
base mechanism, was shown to be small even in simulations of VOCs where these model species
are predicted to be formed.
C. Mechanism Generation and Estimation System
Probably the most important single advance in this version of the mechanism is the use of
a new mechanism generation and estimation software system to derive fully detailed mechanisms
for the atmospheric reactions of many classes of VOCs in the presence of NOX. These are then
used as the basis for deriving an appropriate representation of the VOC in the model. The
automated procedure for generated alkane reaction mechanisms that was incorporated in
-21-
-------
SAPRC-90 (Carter, 1990) was updated based on the results of the evaluation of Atkinson (1997a)
and an independent evaluation of alkoxy and peroxy radical reactions, as discussed in this report.
More significantly, the software was completely revised and the capabilities of the system were
extended to include not only alkanes, but also alkenes (with no more than one double bond), and
many classes of oxygenates including alcohols, ethers, glycols, esters, aldehydes, ketones, glycol
ethers, carbonates, etc. The capabilities of the mechanism generation system in terms of the types of
compounds and reactions it can process are summarized on Table 2.
Table 2. Summary of compounds and reactions that can be processed using the
SAPRC-99 mechanism generation system.
-OH
=C<
Generates mechanisms for VOCs containing following groups:
CH3- -CH2- >CH- >CH< -O-
-CHO -CO- -ONO2 =CH2 =CH-
Currentlv cannot process reactions of following types of VOCs:
• VOCs with more than one double bond
• VOCs with more than one ring
• VOCs that form radicals whose AHF's cannot be estimated
Generates reactions and estimates rate constants or branching ratios for following types of
reactions:
• VOC + OH
• Alkene + O3
• Alkene + NO3
• Alkene + O3P
• Aldehyde + NO3
• Aldehyde + hv->R- +HCO-
• Ketone + hv -> R- +R'CO-
• RONO2 + hv -> RO- +N02
• R- +O2
• RO2 + NO->yNRONO2 + (l-y
• Alkoxy + O2
• Alkoxy radical decompositions (^-scission and "ester rearrangements")
• Alkoxy radical isomerizations
• Crigiee biradical reactions
+NO2
Although many of the estimated rate constants and rate constant ratios are highly
uncertain, this procedure provides a consistent basis for deriving "best estimate" mechanisms for
chemical systems which are too complex to be examined in detail in a reasonable amount of time.
The system allows for assigning or adjusting rate constants or branching ratios in cases where
data are available, or where adjustments are necessary for model simulations to fit chamber data.
Therefore, it could be used for deriving fully detailed mechanisms for VOCs that fully
-22-
-------
incorporate whatever relevant data are available, relying on various estimation methods only
when information is not otherwise available. The program also outputs documentation for the
generated mechanism, indicating the source of the estimates or assumptions or explicit
assignments that were used.
The various types of estimation methods employed in the current mechanism generation
system are summarized in the following section. See Carter (1999) for details.
1. Summary of Estimation Methods
a. VOC + OH Rate Constants and Branching Ratios
The rate constants for the reactions of OH radicals at various positions of the
molecule are estimated using the structure-reactivity estimation methods of Kwok and Atkinson
(1995) as updated by Kwok et al (1996). These include both OH abstraction from C-H bonds and
OH addition to double bonds. The total rate constant is derived from summing up the estimated
rate constants for reactions at various positions in the molecule, and the branching ratios are
estimated by the ratio of the estimated rate constant for reaction at each position to the estimated
total rate constant. If the total rate constant is known, then estimates are used for the branching
ratios only. If product data are available, these are used to derive assigned branching ratios, and
estimates are not used.
For higher molecular weight alkenes such as 1-butene, etc, reactions can occur
both by addition to the double bond and abstraction from non-vinylic C-H bonds. However, the
current system assumes that the only significant mode of reaction of OH to alkenes is by addition
to the double bond. This is because the thermochemical estimates needed to estimate mechanisms
for the unsaturated radicals formed following OH abstractions from alkenes are insufficient to
estimate the possible branching ratios for the radicals predicted to be formed. In addition, the
unsaturated radicals have possible modes of reactions for which estimates have not been
developed. Fortunately, OH addition to the double bond is estimated to be the major reaction
route for most alkenes that are considered.
b. Alkene + O3 Rate Constants and Branching Ratios
Reactions of O3 with alkenes are assumed to occur entirely by addition to the
double bond forming a primary ozonide, which then decomposes to a carbonyl compound and an
excited Crigiee biradical. Total rate constants and branching ratios (i.e., relative yields of the two
possible sets of carbonyl and Crigiee products for unsymmetrical compounds) can be estimated
by assuming that reactions at CH2=CH-, CH2=C<, -CH=CH-, -CH=C<, and >C=C< all have the
same rate constants and branching ratios. Thus averages of rate constants and branching ratios of
compounds where these are known can be used for general estimates when measured rate
constant data are not available. Note that this method performs relatively poorly in estimating
total rate constants for O3 reactions (e.g., see Atkinson and Carter, 1994), but fortunately the O3
rate constants have been measured for most VOCs of interest.
c. VOC + NO3 Rate Constants and Branching Ratios
Reactions of NO3 with VOCs by C-H abstraction forming nitric acid and the
corresponding alkyl radical are assumed to be relatively unimportant for all VOCs except for
aldehydes, and thus are neglected. (The reactions of NO3 with phenolic compounds are included
in the base mechanism, but the mechanism generation system is not used for aromatics.) For
-23-
-------
aldehydes, the abstraction rate constants are estimated based on measured rate constants for NO3
+ acetaldehyde, the correlation between kNO3 and kOH given by Atkinson (1991), and measured
or estimated OH rate constants.
Reaction of NO3 with alkenes are assumed to occur entirely by addition to the
double bond, with all the addition assumed to occur at the least substituted end for unsymmetrical
molecules. Total rate constants are estimated by assuming that reactions at CH2=CH-, CH2=C<,
-CH=CH-, -CH=C<, and >C=C< all have the same rate constants, and using averaged measured
rate constants for compounds where data are available.
d. Aldehyde Photolysis Rates and Initial Reactions
With lack of available data at the time the mechanism was developed, it is
assumed that all saturated C3+ aldehydes photolyze at the same rate. Thus the absorption cross
sections and quantum yields used in the base mechanism for propionaldehyde is used for all such
aldehydes. The products are assumed to be exclusively R- + HCO . However, the recent data of
Wirtz (1999) indicate that this is an oversimplification, and that this aspect of the estimation
method would have to be refined in future versions of the mechanism.
e. Ketone Photolysis Rates and Initial Reactions
In the previous versions of the mechanism, it has been assumed that all higher
saturated ketones photolyze with the same rate and overall quantum yields as methyl ethyl
ketone, whose overall quantum yield was adjusted to fit results of a limited number of outdoor
chamber runs with that compound. Since then, environmental chamber data obtained under more
controlled lighting conditions have become available not only for methyl ethyl ketone, but also
several higher aldehydes. These data could be used to derive overall quantum yields for the
photodecompositions of these ketones, and the resulting values are shown on The mechanism for
ketone photodecomposition is assumed to be scission of the weakest C-CO bond in the molecule,
forming the corresponding R- + R'CO radicals.
Figure 1. It can be seen that, contrary to the assumption in the earlier
mechanisms, the overall quantum yield declines with carbon number. Therefore, for general
estimation purposes, the overall quantum yields for ketones are estimated to depend on the
number of carbons in the molecule, and are estimated using the curve shown on the figure.
The mechanism for ketone photodecomposition is assumed to be scission of the
weakest C-CO bond in the molecule, forming the corresponding R- + R'CO- radicals.
-24-
-------
Figure 1.
Overall quantum yields for ketone photodecomposition used in the SAPRC-90
and SAPRC-99 mechanisms.
0.20
0.15
O Adjusted to Fit 2-Butanone Chamber Data
D Adjusted to Fit 2-Pentanone
I Adjusted to Fit 4-Methyl-2-Pentanone
A Adjusted to Fit 2-Heptanone
SAPRC-99 Estimate
- - - 'SAPRC-90 Estimate
S 0.10
o
0.05
0.00
C4 C5 C6 C7 C8 C9
f. Nitrate Yields from Peroxy + NO
The formation of alkyl nitrates from the reaction of peroxy radicals with NO can
be a significant radical and NOX termination process, and the importance of this process assumed
in the model can significantly affect reactivity predictions for higher molecular weight VOCs.
The previous mechanism used estimates given by Carter and Atkinson (1989), based on available
data on nitrate yields for normal and a limited number of branched alkanes. Recently, Atkinson
and co-workers (Atkinson, private communication, 1999) remeasured the nitrate yields from the
C5+ n-alkanes, and obtained significantly lower yields than were previously estimated. Atkinson
(private communication) considers the new data to supercede the older measurements, and thus
the nitrate yield estimates of Carter and Atkinson (1989) were re-derived to be consistent with the
new data. The new and old data and estimates of nitrate yields for reactions of secondary peroxy
radicals with NO are shown on Figure 2.
Note that the new data indicate that nitrate yields in the reactions of the initially
formed peroxy radicals in the reactions of NO with the C8+ n-alkanes are about -50% lower than
previously assumed. In the previous versions of the SAPRC mechanism it was found that the C8+
n-alkane chamber data could be fit by model simulations only if it was assumed that all the nitrate
formation came from the initially formed peroxy radicals, and that nitrate formation from the
reaction of the 5-hydroxy peroxy radicals formed following alkoxy radical isomerizations was,
negligible. This is not chemically reasonable because reactions involving formation of these
radicals are expected to be important in the Cs+ alkane photooxidation mechanisms, and nitrate
formation from the reactions of NO from the non-substituted analogues is significant. However,
with the lower nitrate yields indicated by the new data, it is no longer to make this chemically
unreasonable assumption for the model to fit the C8+ n-alkane reactivity data.
-25-
-------
Figure 2. Measured and estimated organic nitrate yields for the reactions of NO with
secondary alkyl peroxy radicals, as a function of carbon number.
SECONDARY RO2 NITRATE YIELDS
VS.CARBON NO.
• New or Corrected Data SAPRC-99 Est'd
A Old Data • SAPRC-90 Est'd
40% - ..----'
C4
C6
C8
C10
Figure 3. Plots of estimated overall nitrate yields formed in the photooxidations of various
VOCs forming substituted or non-secondary peroxy radicals against
experimentally determined values or values adjusted to fit environmental
chamber data.
» Alcohol, Ethers
A Esters, Etc.
1:1 Line
i
15% -
• Alkenes
© Non-Sec'y Alkyl
60%UncV
t 5% -,
5% 10%
ESTIMATED NITRATE YIELD
There are limited data on nitrate yields from the reactions of NO with non-
secondary alkyl peroxy radicals or with peroxy radicals with -OH, -CO-, or other substituents. In
most cases, the only "data" available comes from adjusting nitrate yields so model simulations
can fit environmental chamber reactivity data. Environmental chamber experiments are sensitive
to this parameter, but other uncertainties in the estimated or assumed VOC oxidation mechanisms
can affect the results. Therefore these data are uncertain.
-26-
-------
The limited data on measured or estimated nitrate yields from non-secondary or
substituted peroxy radicals indicate that nitrate yields from such radicals tend to be lower than
those from secondary alkyl peroxy radicals, but the dependence on structure is unclear. Because
of lack of information on structural dependences, the same estimates are used for all such
radicals. In particular, nitrate yields from the reaction of NO with substituted or non-secondary
are assumed to be those estimated for secondary alkyl peroxy radicals with the carbon number
reduced by 1.5. The 1.5 reduction was derived to minimize the least squares error between
estimated and measured or adjusted nitrate yields for compounds forming such radicals. The
performance of this estimation method for these radicals is shown on Figure 3, which shows a
plot of estimated vs. experimentally determined overall nitrate for various VOCs forming such
radicals. It can be seen that there are several VOCs where the overall nitrate yields are not well
predicted, but for most others the overall nitrate yields are predicted to within ±60%.
g. Estimation of Alkoxy Radical Rate Constants
Alkoxy radicals can react by a number of competing processes, and the variety of
reactions the higher molecular weight alkoxy radicals can undergo accounts for much of the
complexity of the VOC photooxidation mechanisms. The branching ratios for these competing
reactions can affect not only the number of NO to NO2 conversions involved in a VOCs
photooxidation mechanism, but also the nature (and thus the reactivity) of the organic products
formed.
Atkinson (1997a,b) developed estimation methods for the reactions of the alkoxy
radicals formed in the photooxidations of the alkanes and alkenes that was reasonably successful
in estimating the rate constants involved. However, these estimates were found to perform poorly
when extended to the substituted alkoxy radicals involved in the photooxidations of ethers, esters,
and other substituted organics. Therefore, although the alkoxy radical estimation methods of
Atkinson (1997a,b) served as a useful starting point, they needed to be revised and extended to be
useful for general mechanism estimation purposes.
The alkoxy radical reactions for which estimates were derived included reaction
with O2, P-scission decomposition, 1,4-H shift isomerization, and the "ester rearrangement"
reaction undergone by radicals with the structure -CH[O- J-O-C(O)- (Tuazon et al, 1998). In each
case, rate constants were estimated by making separate estimates of the Arrhenius A factors and
activation energies. The A-factors estimates adopted by Atkinson (1997a,b) and Baldwin et al
(1977) were retained for this work, with the "ester rearrangement" A factor being assumed to be
the same as for the 1,4-H shift isomerization, based on expected similarities in the structure of the
transition state. Various approaches were used to estimate the activation energy depending on the
type of reaction, as discussed below.
Reaction with O?. The few measured alkoxy + O2 rate constants were found to
give reasonably good correlation between the estimated heat of reaction and the activation energy
derived using the estimated A factor. This correlation was then used to estimate the activation
energies for the reactions of other alkoxy radicals with O2. This is considered to give more
reasonable estimates than the approach adopted by Atkinson (1997a,b), which is to use a
correlation between the rate constant and the heat of reaction. The problem with that approach is
that it estimates rate constants for some of the highly exothermic O2 reactions which are greater
than the A factors for the reactions whose rate constants have been measured.
1.4-H Shift Isomerizations. The activation energies for these reactions are
estimated based on thinking of them as an H-atom abstraction by an alkoxy radical. The
-27-
-------
correlation between C-H bond energies and activation energies for various bimolecular H-atom
abstractions by methoxy radicals are used to estimate activation energies for abstractions from
various groups. Substituent correction factors are derived by analogy with those used in the
structure-reactivity method s of Atkinson (1987) and Kwok and Atkinson (1995) for estimating
OH radical rate constants. A comparison of the activation energies derived based on the methoxy
radical reactions with the activation energies derived for the few 1,4-H shift isomerizations for
which there are data (based on the rate constant and the estimated A factors) indicate that it is
appropriate to add a ~1.6 kcal/mole "strain" term when estimating the isomerization activation
energies from those estimated for bimolecular reactions of methoxy radicals.
In order to fit the available product data for oxygenated VOCs which form
alkoxy radicals where 1,4-H shift isomerizations compete with P-scission decompositions, and to
be consistent with P-scission decomposition rate constant estimates derived for other radicals, it is
necessary to assume that isomerizations involving -CO- or -O- groups in the ring in the transition
state have higher activation energies than isomerizations of unsubstituted alkoxy radicals. The
data are best fit by an additional strain energy of 3.5 kcal/mole for isomerizations involving
intermediates with such groups.
p-Scission Decompositions. By far the most difficult problem in developing
estimation procedures for alkoxy radical reactions was deriving general estimation methods for
the various types of P-scission decompositions. Direct measurements of p-scission decomposition
rate constants are available only for a few relatively simple radicals, but estimates of rate
constants for a wide variety of other radicals could be derived by analysis of product yield data
for a variety of VOCs. In many cases, relative yields of various products can be used to estimate
rate constant ratios for decompositions relative to competing reactions of the alkoxy radicals with
Oa or by 1,4-H shift isomerization. From O2 or isomerization rate constants estimated as
discussed above, these product yield ratios can then be used to obtain estimates of P-scission
decomposition rate constants for a wide variety of compounds. These estimated rate constants can
then be used as a basis for deriving general estimation methods for alkoxy radical decomposition
reactions. Although uncertain, they provide the only available data base that can be used for this
purpose.
Based on the approach of Atkinson (1997a,b), activation energies for P-scission
decompositions, Ea, are estimated based on the assumption that
Ea^EaA + EaB- AHR
Here AHR is the heat of reaction of the decomposition reaction, EaA is a parameter that is
assumed to depend only on the nature of the radical formed in the decomposition, and EaB is
assumed to be the same for all alkoxy radical decompositions. A value of EaB ~ 0.44 was derived
from data for alkoxy radical reactions forming CH3- radicals, and is assumed to be applicable to
all P-scission decomposition reactions.
The values of EaA were derived based on available measured or estimated alkoxy
radical decomposition rate constants, or from various other estimates in cases where rate constant
data or estimates were not available. Atkinson (1997a,b) noted a correlation between the EaA
parameter and the ionization potential of the radical form, and that correlation was used for
estimating certain EaA parameters for which data were not available. However, this correlation
was not found to be useful only for estimating EaA parameters for reactions forming radicals of
the same type. This is shown on Figure 4, which shows plots of EaA parameters against IP for
reactions forming radicals for which the IP is known or can be estimated.
-28-
-------
Figure 4. Plots of EaA parameters (in kcal/mole) against ionization potential for estimating
activation energies for p-scission decomposition reactions against ionization
potentials for the radical formed in the decompositions. The filled symbols, "X"
and "+" show the parameters that were derived from measured or estimated
decomposition rate constants, and the open symbols show the parameter that
were estimated by assuming a linear relationship between IP and EaA for
reactions forming radicals of the same type.
+Alky I
AHCO.RCO
+ ROCH2
• RO.
• HOCH2,RCH(OH).
XOH
16 -
X
8 10 12
IONIZATION POTENTIAL (E.V.)
14
The decomposition activation energy and rate constant estimates are considered
to be highly uncertain in many (if not most) cases, being based in many cases on very uncertain
alkoxy + C>2 or isomerization rate constants, employing many highly uncertain and untested
assumptions, and not giving satisfactory predictions in all cases. Clearly, additional data are
needed, particularly for reactions of oxygen-containing alkoxy radicals, to test, refine, and
improve these estimates and the many assumptions they incorporate. Indeed, it may not be
possible to develop a totally satisfactory estimation method that can accurately predict rate
constants for the full variety of these reactions, without carrying out detailed theoretical
calculations for each system. Thus, rate constants or branching ratios derived from experimental
data are used whenever possible when developing reaction mechanisms for atmospheric reactivity
predictions. However, when no data are available, we have no choice but to use estimates such as
those developed in this work.
Ester Rearrangements. There are no data available concerning the rate constant
for these reactions. However, this reaction clearly is the dominant pathway in the reactions of the
CH3CH[O- ]OC(O)CH3 radical formed in the ethyl acetate system (Tuazon et al, 1998), and model
simulations of n-butyl acetate reactivity experiments can only fit the data if it is assumed that 1,4-
isomerization dominates over the ester rearrangement in the reactions of CH3CH2CH2CH(O- )O-
CO-CH3 radicals. This information, together with estimates of the rate constants for the
competing reactions of these radicals, and estimates for the A factor as discussed above, are then
used to derive an approximate estimate of ~7.4 kcal/mole for the activation energy of this
reaction. This is clearly highly uncertain.
-29-
-------
h. Crigiee Biradical Reactions
Excited Crigiee biradicals are assumed to be formed in the reactions of O3 with
alkenes and alkynes, and the subsequent reactions of these species are incorporated in the
mechanism generation system. Four types of Crigiee biradicals are considered, as discussed
below. Note that it is assumed that the biradical reactions are assumed not to depend on the
reaction forming them. The validity of this assumption is subject to question, since the reaction
forming them would be expected to affect the distribution of excitation energy in the biradical,
and the excitation energy would be expected to affect the mechanism of reaction in some if not all
cases.
CH2OO Biradicals. The recommendations of Atkinson (1994, 1997a) concerning
branching ratios for stabilization and the various decomposition routes are used without
modification.
RCHOO Biradicals. The recommendations of Atkinson (1997a) for CH3CHOO
biradicals and an analysis of OH measurements from the reactions of O3 with various 1 -alkenes in
the absence of NOX can be used to derive a general set of estimates for the various competing
reactions for these biradicals. These estimates involve predictions of OH yields of ~55%,
independent of the size of the biradical. However, results of 1-butene - NOX and 1-hexene - NOX
environmental chamber data can only be successfully simulated if it is assumed that the OH
yields from the biradical formed in these systems are only -5% and less than 1%, respectively. It
is also necessary to assume that the OH yields from the CH3CHOO biradicals are lower than
recommended by Atkinson (1997a) in order to obtain nonbiased simulations of the large body of
propene - NOX environmental chamber data. Therefore, the SAPRC-99 mechanism assumes that
OH yields from RCHOO biradicals decrease with the size of the molecule, to be consistent with
the chamber data.
The discrepancy between laboratory measurements of OH yields from 1 -alkenes
and the lower OH yields needed to fit chamber data for such compounds is clearly a concern, and
additional work is needed to resolve this discrepancy. It might be due to the interaction of these
biradicals with NOX, which is present in the environmental chamber experiments but not in the
laboratory systems where OH yields are measured. However, the SAPRC-99 Crigiee biradical
mechanisms do not include any biradical + NOX reactions; all their reactions are treated as
unimolecular processes. Note that this includes stabilization - pressure effects of Crigiee biradical
reactions are not represented in the current version of the mechanism.
RR'COO Biradicals with a Hydrogens. These biradicals are assumed to
exclusively rearrange to the unsaturated hydroperoxide intermediate, which then decomposes to
form OH radicals in 100% yield, i.e.,
>CHC(OO)- -» >C=C(OOH)- -> >C=C(0- ) + OH
OH
This is consistent with the relatively high OH yields observed in the reactions of O3 with internal
alkenes, and in this case the model assuming these high OH yields gives reasonably good
simulations of the available environmental chamber data.
Other RR'COO Biradicals. The above mechanism cannot occur for those
disubstituted Crigiee biradicals that do not have substituents with a hydrogens. It is also
considered to be unlikely if the only substituent(s) with a hydrogens are -CHO groups, since it is
expected that formation of a ketene hydroperoxide intermediate would involve a strained
-30-
-------
transition state. In the cases of biradicals with carbonyl groups, the mechanisms assumed are
based on those derived by Carter and Atkinson (1996) for reactions of O3 with isoprene products.
In the other cases, which do not occur in many cases for the VOCs currently considered in the
mechanism, we arbitrarily assume that 90% is stabilized and 10% decomposes to CO2 + 2 R- .
2. Mechanism Generation
This mechanism generation system is used as the primary means of deriving SAPRC-99
mechanistic parameters for all the classes of VOCs that it can handle, including alkanes, alkenes,
and the variety of oxygenated species as indicated above. Although the program outputs
mechanisms that can (for larger molecules) involve hundreds or even thousands of reactions or
products, various "lumping rules" are used to convert the detailed generated mechanisms and
product distributions into the lumped reactions incorporating the appropriate model species used in
the base mechanism. The use of this program has permitted estimation of detailed mechanisms for a
much larger number of compounds than otherwise would be possible without incorporating
approximations that might significantly compromise the accuracy of reactivity predictions.
Although the mechanism generation system currently cannot be used to derive
mechanisms for dialkenes and unsaturated aldehydes and ketones, the estimates in the detailed
mechanism of Carter and Atkinson (1996) for isoprene and its major products were incorporated
explicitly in the mechanism generation system, allowing full mechanisms for these species to be
generated.. The results are therefore are consistent with the detailed mechanism of Carter and
Atkinson (1996) and the condensed mechanisms of Carter (1996) for these compounds. A similar
approach was used so the system could be used to generate reactions of 1,3-butadiene acrolein,
and various alkynes.
D. Assigned or Parameterized Mechanisms
1. Aromatics
Despite progress in recent years, there are still too many uncertainties concerning the
details of the photooxidation mechanisms of aromatics and the reactive products they form to
allow for explicit mechanisms to be derived or estimated. Therefore, simplified and
parameterized mechanisms, with uncertain parameters adjusted to fit environmental chamber
data, are still employed. However, the representation of the uncharacterized aromatic ring
fragmentation products was revised somewhat based new data obtained for unsaturated
dicarbonyls (e.g., Bierback et al, 1994), and to allow for explicit representation of the oc-
dicarbonyl products formed from the methylbenzenes. As with SAPRC-97, this version of the
mechanisms appropriately represents reactivity differences among various xylene and
trimethylbenzene isomers, and is able to correctly simulate how aromatic reactivities vary with
differing light sources.
This mechanism also uses parameterized representations for naphthalenes and tetralin,
with parameters adjusted to fit the limited chamber data for naphthalene, 2,3-dimethyl
naphthalene, and tetralin. These compounds tended to have much lower mechanistic reactivities
than the alkylbenzenes, and thus their mechanisms use lower yields of the model species that
represent photoreactive ring fragmentation products. In addition, to fit the chamber data, it is
necessary for the model to assume that these compounds form radicals that react like PAN
precursors. The actual mechanistic implications of this parameterizations is unknown, and the
reliability of model predictions for these compounds is uncertain.
-31-
-------
2. Terpenes
Because the mechanism generation system cannot derive mechanisms for bicyclic
compounds, simplified mechanisms were derived for the terpenes, based on environmental chamber
data for several representative terpenes. Some parameters, such as overall organic nitrate yields and
numbers of NO to NOa conversions in the OH reaction, were adjusted based on the chamber data,
and the mechanism generation system for compounds with similar structures was employed to
derive estimated mechanisms for their reactions with ozone. The mechanism correctly predicts
observed reactivity differences among various terpene isomers, though some experiments,
particularly with [3-pinene, are not well simulated in some respects.
3. Other Compounds
Assigned mechanisms were also derived for styrene, N-methyl-2-pyrroladone, toluene
diisocyanate, and diphenylene diisocyanate, based on available kinetic and mechanistic data,
estimated or parameterized mechanisms, and results of environmental chamber experiments
employing those or related compounds.
Although C1OX or BrOx chemistries have been incorporated as extensions to the
SAPRC-97 mechanism (Carter et al, 1996, 1997c, 1997d), this is not yet incorporated in the
current version of this updated mechanism. With the exception of chloropicrin, which appears to
have relatively simple and unique chemistry (Carter et al, 1997d), the few halogenated
compounds we have studied [trichloroethylene (Carter et al, 1996) and alkyl bromides (Carter et
al, 1997c)] indicate that we cannot account for the reactivities of'those compounds with explicit
mechanisms. Therefore, the current version of the mechanisms uses a highly simplified and
parameterized "placeholder" mechanism to provide very rough estimates of the approximate
range of reactivities of halogenated compounds under MIR conditions, given their OH radical rate
constants. The predictions of these mechanisms must be considered to be highly uncertain, and
the available chamber data indicate they are almost certainly not valid under low NOX conditions.
A parameterized "placeholder" mechanism is also used to estimate the
approximate reactivity ranges of amines, given their measured or estimated OH radical rate
constants. The predictions of this mechanism for those compounds must also be considered to be
highly uncertain, especially since they have not been evaluated using environmental chamber
data. However, use of this mechanism allows at least approximate estimates to be made.
4. Mechanism Evaluation
The performance of the mechanism in simulating 03 formation, rates of NO oxidation,
and other measures of reactivity was evaluated by conducting model simulations of over 1600
environmental chamber experiments carried out the Statewide Air Pollution Research Center
(SAPRC) and the College of Engineering Center for Environmental Research and Technology
(CE-CERT) at the University of California at Riverside (UCR). These include not only
experiments in the UCR database through 1993 (Carter et al, 1995), but also experiments carried
out at CE-CERT through mid 1999 for the purpose of developing and evaluating mechanisms for
various types of VOCs . The types of experiments and number of VOCs in the chamber data base
used in the evaluation are summarized on Table 3, and the types of chambers employed are
summarized on Table 4.
2 The experiments include most of those described in the various reports on CE-CERT chamber
studies that can be downloaded from http://cert.ucr.edu/~carter/bycarter.htm.
-32-
-------
The results of the evaluation indicated that this version of the mechanism performed
slightly, though not dramatically, better than the previous versions (Carter and Lurmann, 1991;
Carter, 1995; Carter et al, 1997a) in simulating experiments with the major hydrocarbon classes
found in ambient air and complex or surrogate mixtures. The overall performance in the
mechanism in simulating the amount of O3 formed + NO oxidized for all the environmental
chamber runs modeled is shown on Figure 5. It can be seen that for most experiments the model
was able to simulate this to within ±20%, which is slightly better than the ±30% performance
obtained previously. Of course, this could in part be due to the fact that the current data set
included much fewer of the more difficult-to-characterize outdoor experiments, and includes a
much larger number of very recent experiments carried out under more reproducible conditions.
In addition, this version of the mechanism generally gave satisfactory fits to the reactivity
data for most of the experiments using the various compounds that were studied more recently,
which were either not represented or poorly represented in the previous versions. However, as
with previous evaluations of this (Carter and Lurmann, 1991; Carter, 1995; Carter et al, 1997a)
and other (Carter and Lurmann, 1990, Gery et al, 1988) mechanisms, there were cases where
satisfactory simulations were not obtained. Many of these cases of poor performance in
simulating the data can be attributed to problems with the mechanism, but this is probably not
true in all cases. The cases where less-than-satisfactory model performance may be attributable to
possible problems in the mechanism are summarized in Table 5. However, reactivities of most
VOCs were reasonably well simulated, though in many cases adjustments to uncertain portions
were made to achieve the fits. These cases are also noted in the summary of the evaluation
results.
Table 3. Environmental chamber data base used for evaluating the SAPRC-99
mechanism.
Type of experiment
Runs
VOCs
Characterization runs 76
VOC - NOX runs 484 37
Incremental reactivity runs • 447 80
Miscellaneous mixture - NOX 95
"Base case" surrogate - NOX mixtures with reactivity runs 561
-33-
-------
Table 4. Summary of major characteristics of the environmental chambers used in the
mechanism evaluation.
Walls
Lights
RH
Vol (L)
Runs
Teflon Film
Teflon Film
Teflon Film
Teflon Coated Al,
Quartz
Teflon Film
Blacklight
Blacklight
Xenon Arc
Xenon Arc
Sunlight
50%
Dry
Dry
50%
Dry
6000
3000-6000
2500-5000
6400
20,000
139
1066
323
107
42
Figure 5.
Distribution plot of model performance in simulating the amount of NO oxidized
+ NO formed, A([O3]-[NO]), for all environmental chamber runs modeled.
I
UJ
UJ
LL
o
U.
Ul
m
(CALCULATED - EXPERIMENTAL) / EXPERIMENTAL
E. Updated Reactivity Scales
The updated mechanism was used to calculate updated MIR and other ozone reactivity
scales, using the scenarios and methodology developed previously for this purpose (Carter, 1994).
Reactivity estimates were derived for a total of 557 VOC's, including many that were not in
previous tabulations, or whose estimates were based on much more uncertain or approximate
mechanisms. It is therefore recommended that these be used in any application that calls for use
of the MIR scale or any of the other scales given by Carter (1994). Although the estimates for
many of the VOCs remain highly uncertain, the present scale provides the best estimates that are
currently available.
-34-
-------
Table 5. Summary of VOC classes where model performance in simulating the
environmental chamber data suggests possible problems with the mechanism, or
where unoptimized or placeholder mechanisms were used.
Compound
Problem
Benzene
€4+ 1-Alkenes
3,4-Dimethyl Hexane
Cyclohexanone
3-Pinene
t-Butanol
Dimethyl Succinate (DBE-4)
Trichloroethylene
Alkyl Bromides.
Unsatisfactory simulations of some experiments
Could not fit data unless OH yields in O3 + alkene reactions are
lower than indicated by laboratory data.
Nitrate yields and some other mechanistic parameters were not
optimized.
Model fits some but not all reactivity experiments
O3 overpredicted in f$-pinene - NOX experiments, but incremental
reactivity experiments are reasonably well simulated.
Data are better fit if the kOH were reduced from the currently
accepted value.
Model fits some but not all reactivity experiments
"Placeholder" mechanism used for all halogenated VOCs. But
model performed surprisingly well in fitting the data for
trichloroethylene, considering its crudity.
"Placeholder" mechanism used for all halogenated VOCs. Model
did not correctly predict inhibiting effects of bromides on O3
under low NOX conditions.
The effects of the updates in the mechanism on calculated reactivities in the Maximum
Incremental Reactivity (MIR) scale are shown on Table 6. This compares incremental reactivities
of the ambient mixture and relative reactivities of selected VOCs in this version of the
mechanism and the SAPRC-97 mechanism of Carter et al (1997a). It can be seen that there is
almost no change in the ambient mixture incremental reactivity, but the relative reactivities (i.e.,
incremental reactivities of the VOCs divided by the incremental reactivity of the ambient
mixture) changed by 15-50% for many VOCs, and by larger factors for a few others. The large
change for high molecular weight alkanes is due in part to the use of a more reactive model
species to represent alkane- photooxidation products, and also to the fact that the reactivities of
these compounds, whose net reactivities are differences in large positive and negative factors, are
highly sensitive to changes in the base mechanism. The large change for acetylene is due to the
higher reactivity for its major product, glyoxal, used in the current mechanism, which was made
for the model to fit results of new experiments with acetylene (Carter et al, 1997b). The large
change for ethylene glycol is due to a new value for its OH radical rate constant. Reactivities of a
number of other individual VOCs changed due to new information on their mechanism, or the
availability of new environmental chamber data that indicate needs for modifications for their
mechanisms.
-35-
-------
Table 6.
Effects of mechanism updates on Maximum Incremental Reactivities.
VOC or Mixture
SAPRC-97 SAPRC-99 A%
Incremental Reactivities (am O3 / am VOC)
Ambient Mixture 4.06 3.98
Relative Reactivities (mass basis)
-2%
Ethane
n-Decane
1-Hexene
Isoprene
a-Pinene
Toluene
m-Xylene
p-Xylene
Acetylene
Ethanol
Ethylene Glycol
Methyl t-Butyl Ether
2-Butoxyethanol
Formaldehyde
Methyl Ethyl Ketone
Mineral Spirits "B"
(Type II-C)
0.08
0.13
1.40
2.30
0.96
1.26
3.49
0.71
0.09
0.42
0.56
0.18
0.57
1.62
0.35
0.35
0.09
0.24
1.53
2.89
1.13
1.07
2.78
1.15
0.33
0.47
0.92
0.22
0.84
2.33
0.40
0.24
12%
91%
9%
25%
19%
-15%
-20%
61%
276%
13%
63%
25%
47%
44%
14%
-29%
The California Air Resources Board (CARB)'s vehicle emissions regulations (CARB,
1993) use "Reactivity Adjustment Factors" (RAFs) in emissions standards for alternatively fueled
vehicles, where the RAFs are calculated from the ratio of MIR reactivities of the exhaust divided
by that of a standard gasoline. The current regulations uses RAFs derived from the MIR scale
calculated using the SAPRC-90 mechanism (Carter, 1994). Table 7 shows effects of the
subsequent updates to the SAPRC mechanism on the calculated RAF values for transitional low
emissions vehicle (TLEV) exhausts of interests to the CARB. It can be seen that the mechanism
updates had relatively small effects on these RAFs, with the update from SAPRC-97 to
SAPRC-99 tending to counteract the changes caused when SAPRC-90 was updated to
SAPRC-97.
The reactivity tabulations include footnotes indicating the type of mechanism or
representation employed when calculating the reactivities, the extent to which the reactivity
predictions were evaluated against experimental data, and an uncertainty ranking. Upper limit
reactivity estimates are also included. The uncertainty classification given with the scale and the
other associated footnotes can be used to indicate the qualitative level of uncertainty for any
given VOC. It is recommended that any regulatory application that employs any of the scales
given in this report appropriately take uncertainty into account for those VOCs whose reactivities
are indicated as having a high level of uncertainty.
-36-
-------
Table 7.
Effects of mechanism updates on TLEV exhaust reactivity adjustment factors
(RAFs) calculated using the MIR scale.
Exhaust
Type
RFA
M85
E85
CNG
LPG
Phase 2
SAPRC-99
RAF A%
1.00
0.38
0.67
0.19
0.52
0.99
_
14%
11%
16%
12%
1%
SAPRC-97 SAPRC-90
RAF A% RAF
1.00
0.34
0.61
0.16
0.46
0.98
_
-10%
-4%
-9%
-8%
0%
1.00
0.37
0.63
0.18
0.50
0.98
F. Additional Information Available
A draft report (Carter, 1999) giving the comprehensive documentation of the mechanism,
the results of the evaluation against the chamber data, the updated calculated reactivities in the
MIR and other scales, and the updated uncertainty classifications has been prepared and
submitted to the CARB for review. This report can be downloaded from
http://cert.ucr.edu/~carter/reactdat.htm. The CARB contracted William Stockewell, the developer
of the RADM and RACM mechanisms (Stockwell et al, 1990, 1997), to conduct a peer review of
the mechanism, and that report, and the authors response to Stockwell's recommendations, can
also be downloaded at that side. In addition, the mechanism generation system that was
developed for this mechanism was designed to be accessed by anyone from the web, and the
access to that system is also linked to the above-referenced web site.
-37-
-------
Table 8. Examples of reactivity uncertainty classifications assigned for reactivity
predictions calculated using the SAPRC-99 mechanism.
CONSIDERED TO BE RELATIVELY UNCERTAIN
• n-Butane
• Propene
• 2-Butoxyethanol
MECHANISM MAY CHANGE, BUT MIR CHANGE IS EXPECTED TO BE LESS THAN A
FACTOR OF TWO
• n-Octane
• 1-Pentene
• Toluene
• 2-Ethoxyethanol
REACTIVITY MAY CHANGE BY A FACTOR OF TWO IF COMPOUND STUDIED OR IF
BASE MECHANISM CHANGED
• n-Dodecane *
• Branched C12 Alkanes
• Trans-2-Hexene
• s-Butyl Benzene
• Ethyl t-Butyl Ether
REACTIVITY IS EXPECTED TO CHANGE IF COMPOUND IS STUDIED. (Uncertainty
adjustments should be used if estimated reactivities for these compounds are used in regulations.)
• 1-Octene
• C8 Internal Alkenes
• Methyl Acetylene
• Vinyl Acetate
SIGNIFICANT CHANCE OF MECHANISM BEING INCORRECT IN IMPORTANT
RESPECTS (Uncertainty adjustments should be used if estimated reactivities for these
compounds are used in regulations.)
• Cyclopentadiene
• Indan
MECHANISM IS PROBABLY INCORRECT OR VOC IS REPRESENTED BY A
"PLACEHOLDER" MECHANISM (Uncertainty adjustments should be used if estimated
reactivities for these compounds are used in regulations.)
• Ethyl Amine
• Vinyl Chloride
• Benzotrifluoride
-38-
-------
G. References
Atkinson, R. (1987): "A Structure-Activity Relationship for the Estimation of Rate Constants for
the Gas-Phase Reactions of OH Radicals with Organic Compounds," Int. J. Chem. Kinet,
19, 799-828.
Atkinson, R. (1989): "Kinetics and Mechanisms of the Gas-Phase Reactions of the Hydroxyl
Radical with Organic Compounds," J. Phys. Chem. Ref. Data, Monograph no 1.
Atkinson, R. (1990): "Gas-Phase Tropospheric Chemistry of Organic Compounds: A Review,"
Atmos. Environ., 24A, 1-24.
Atkinson, R. (1991): "Kinetics and Mechanisms of the Gas-Phase Reactions of the NO3 Radical
with Organic Compounds," J. Phys. Chem. Ref. Data, 20, 459-507.
Atkinson, R. (1994): "Gas-Phase Tropospheric Chemistry of Organic Compounds," J. Phys.
Chem. Ref. Data, Monograph No. 2.
Atkinson, R. (1997a): "Gas Phase Tropospheric Chemistry of Volatile Organic Compounds: 1.
Alkanes and Alkenes," J. Phys. Chem. Ref. Data, 26, 215-290.
Atkinson, R. (1997b): "Atmospheric Reactions of Alkoxy and Beta-Hydroxyalkoxy Radicals,"
Int. J. Chem. Kinet., 29, 99-111.
Atkinson, R., D. L. Baulch, R. A. Cox, R. F. Hampson, Jr., J. A. Kerr, M. J. Rossi, and J. Troe
(1997): "Evaluated Kinetic, Photochemical and Heterogeneous Data for Atmosp 11, 45.
Atkinson, R., D. L. Baulch, R. A. Cox, R. F. Hampson, Jr., J. A. Kerr, M. J. Rossi, and J. Troe
(1999): "Evaluated Kinetic, Photochemical and Heterogeneous Data for Atmospheric
Chemistry: Supplement VII, Organic Species (IUPAC)," J. Phys. Chem. Ref. Data, 28,
191-393.
Atkinson, R. and W. P. L. Carter (1984): "Kinetics and Mechanisms of the Gas-Phase Reactions
of Ozone with Organic Compounds under Atmospheric Conditions," Chem. Rev. 84,
437-470.
Baldwin, A. C., J. R. Barker, D. M. Golden and G. G. Hendry (1977): "Photochemical Smog.
Rate Parameter Estimates and Computer Simulations," J. Phys. Chem. 81, 2483.
Bierbach A., Barnes I., Becker K.H. and Wiesen E. (1994) Atmospheric chemistry of unsaturated
carbonyls: Butenedial, 4-oxo-2-pentenal, 3-hexene-2,5-dione, maleic anhydride, 3H-
furan-2-one, and 5-methyl-3H-furan-2-one. Environ. Sci. Technol., 28, 715-729.
CARS (1993): "Proposed Regulations for Low-Emission Vehicles and Clean Fuels - Staff
Report and Technical Support Document," California Air Resources Board, Sacramento,
CA, August 13, 1990. See also Appendix VIII of "California Exhaust Emission Standards
and Test Procedures for 1988 and Subsequent Model Passenger Cars, Light Duty Trucks
and Medium Duty Vehicles," as last amended September 22, 1993. Incorporated by
reference in Section 1960.
-39-
-------
Carter, W. P. L. (1990): "A Detailed Mechanism for the Gas-Phase Atmospheric Reactions of
Organic Compounds," Atmos. Environ., 24A, 481 -518.
Carter, W. P. L. (1994): "Development of Ozone Reactivity Scales for Volatile Organic
Compounds," J. Air & Waste Manage. Assoc., 44, 881-899.
Carter, W. P. L. (1995): "Computer Modeling of Environmental Chamber Measurements of
Maximum Incremental Reactivities of Volatile Organic Compounds," Atmos. Environ.,
29,2513-2517.
Carter, W. P. L. (1996): "Condensed Atmospheric Photooxidation Mechanisms for Isoprene,"
Atmos. Environ., 30, 4275-4290.
Carter, W. P. L. (1999): "Documentation of the SAPRC-99 Mechanism for VOC Reactivity
Assessment," Draft final report on California Air Resources Board Contracts No. 92-329
and 95-308. September 13.
Carter, W. P. L. and R. Atkinson (1989): "Alkyl Nitrate Formation from the Atmospheric Photo-
oxidation of Alkanes; a Revised Estimation Method," J. Atm. Chem. 8, 165-173.
Carter, W. P. L., and F. W. Lurmann (1990): "Evaluation of the RADM Gas-Phase Chemical
Mechanism," Final Report, EPA-600/3-90-001.
Carter, W. P. L. and F. W. Lurmann (1991): "Evaluation of a Detailed Gas-Phase Atmospheric
Reaction Mechanism using Environmental Chamber Data," Atm. Environ. 25A, 2771-
2806.
Carter, W. P. L., J. A. Pierce, I. L. Malkina, D. Luo and W. D. Long (1993): "Environmental
Chamber Studies of Maximum Incremental Reactivities of Volatile Organic
Compounds," Report to Coordinating Research Council, Project No. ME-9, California
Air Resources Board Contract No. A032-0692; South Coast Air Quality Management
District Contract No. C91323, United States Environmental Protection Agency
Cooperative Agreement No. CR-814396-01-0, University Corporation for Atmospheric
Research Contract No. 59166, and Dow Corning Corporation. April 1.
Carter, W. P. L., D. Luo, I. L. Malkina, and D. Fitz (1995): "The University of California,
Riverside Environmental Chamber Data Base for Evaluating Oxidant Mechanism. Indoor
Chamber Experiments through 1993," Report submitted to the U. S. Environmental
Protection Agency, EPA/AREAL, Research Triangle Park, NC., March 20..
Carter, W. P. L. and R. Atkinson (1996): "Development and Evaluation of a Detailed Mechanism
for the Atmospheric Reactions of Isoprene and NOx," Int. J. Chem. Kinet, 28, 497-530.
Carter, W. P. L., D. Luo, and I. L. Malkina (1996): "Investigation of the Atmospheric Ozone
Formation Potential of Trichloroethylene," Report to the Halogenated Solvents Industry
Alliance, August.
Carter, W. P. L., D. Luo, and I. L. Malkina (1997a): "Environmental Chamber Studies for
Development of an Updated Photochemical Mechanism for VOC Reactivity
Assessment," Final report to the California Air Resources Board, the Coordinating
Research Council, and the National Renewable Energy Laboratory, November 26.
-40-
-------
Carter, W. P. L., D. Luo, and I. L. Malkina (1997b): "Investigation of the Atmospheric Ozone
Formation Potential of Acetylene," Report to Carbind Graphite Corp., April 1.
Carter, W. P. L., D. Luo, and I. L. Malkina (1997c): "Investigation of the Atmospheric Ozone
Formation Potential of Selected Alkyl Bromides," Report to Albemarle Corporation,
November 10.
Carter, W. P. L., D. Luo and I. L. Malkina (1997d): "Investigation of that Atmospheric Reactions
of Chloropicrin," Atmos. Environ. 31, 1425-1439.; Report to the Chloropicrin
Manufacturers Task Group, May 19.
Gery, M. W., G. Z. Whitten, and J. P. Killus (1988): "Development and Testing of the CBM-IV
For Urban and Regional Modeling,", EPA-600/ 3-88-012, January.
Golden, D. M. and G. P. Smith (1999): "Reaction of OH + NO2 + M: A New View," To be
submitted to Journal of Physical Chemistry.
Kwok, E. S. C., and R. Atkinson (1995): "Estimation of Hydroxyl Radical Reaction Rate
Constants for Gas-Phase Organic Compounds Using a Structure-Reactivity Relationship:
An Update," Atmos. Environ 29, 1685-1695.
Kwok, E. S. C., S. Aschmann, and R. Atkinson (1996): "Rate Constants for the Gas-Phase
Reactions of the OH Radical with Selected Carbamates and Lactates," Environ. Sci.
Technol 30, 329-334.
NASA (1997): "Chemical Kinetics and Photochemical Data for Use in Stratospheric Modeling,
Evaluation Number 12," JPL Publication 97-4, Jet Propulsion Laboratory, Pasadena,
California, January.
Stockwell, W. R., P. Middleton, J. S. Chang, and X. Tang (1990): "The Second Generation
Regional Acid Deposition Model Chemical Mechanism for Regional Air Quality
Modeling," J. Geophys. Res. 95, 16343- 16376.
Stockwell, W.R., F. Kirchner, M. Kuhn, and S. Seefeld (1997): A new mechanism for regional
atmospheric chemistry modeling. J. Geophys. Res., 102,25847-25880.
Wirtz, K. (1999): "Determination of Photolysis Frequencies and Quantum Yields for Small
Carbonyl Compounds using the EUPHORE Chamber", presented at the Combined
US/German and Environmental Chamber Workshop, Riverside, CA, October 4-6, 1999.
Tuazon, E. C., S. M. Aschmann, R. Atkinson, and W. P. L. Carter (1998): "The reactions of
Selected Acetates with the OH radical in the Presence of NO: Novel Rearrangement of
Alkoxy Radicals of Structure RC(O)OCH(O.)R'", J. Phys. Chem A 102, 2316-2321.
-41-
-------
MODELLING OZONE FORMATION WITH A MASTER CHEMICAL
MECHANISM
R G Derwent1, M E Jenkin2, S M Saunders3 and M J Pilling3
1 Climate Research Division, Meteorological Office, Bracknell, UK.
2 National Centre for Environmental Technology, AEA Technology, Culham Laboratory,
Oxfordshire, UK.
3 School of Chemistry, The University, Leeds, UK.
A Master Chemical Mechanism containing over 2400 chemical species and over 7100 chemical
reactions is employed here to describe the atmospheric degradation of 123 organic compounds
and the associated regional scale ozone and PAN formation under conditions appropriate to the
polluted boundary layer over north-west Europe. Photochemical ozone and PAN creation
potentials (POCP and PPCP) are derived for each organic compound from their propensities to
form ozone and PAN relative to ethylene and propylene, respectively. When we have compared
the POCPs calculated under the European conditions appropriate to multi-day regional scale
photochemical episodes against the literature MIR values determined for intense USA urban
single-day conditions, the scatter plot showed distinct curvilinear behaviour. We interpreted this
behaviour as indicating that the multi-day episode allowed less reactive organic compounds to
build up in concentrations and hence to produce heightened contributions to ozone formation.
This present study confirms that this is indeed the explanation and shows that reactivity scales for
alkanes, alcohols, esters and ethers should take both single day and multi-day conditions into
account.
-42-
-------
Simulations of EUPHORE and field experiments using a master chemical
mechanism
Nicola Carslaw & Michael J. Pilling
School of Chemistry, University of Leeds, Leeds LS2 9JT, UK
Michael E. Jenkin & Garry D. Hayman
AEA Technology pic, National Environmental Technology Centre, Culham, UK
Introduction
In recent years, a master chemical mechanism (MCM) has been developed to describe the
atmospheric degradation of VOC in the atmosphere (Jenkin et al., 1997a; Saunders et al.,
1997). The MCM has been developed as part of a collaborative project between the
University of Leeds, AEA Technology pic and the UK Meteorological Office, and funded by
the UK Department of the Environment (now the UK Department of the Environment,
Transport and the Regions). The mechanism is near-explicit, and draws on the latest chemical
kinetic and mechanistic results. The MCM has been validated for ozone production against
the carbon bond mechanism (CBMIV), which in turn was validated against US smog chamber
data (Derwent et al, 1998). The latest version of the MCM treats the degradation of 123
VOC, and can be found on the internet at http://www.chem.leeds.ac.uk/Atmospheric/MCM/
mcmproj.html). In this paper, we describe the evaluation of the MCM both through smog
chamber and field experiments.
Smog chamber work
In order to test various aspects of a recently developed cc-pinene mechanism, chamber
experiments were carried out in the £C/ropean .P/ftTtochemical REactor (EUPHORE) at
Valencia. Two experimental systems were studied as followed: NOx/ethene/toluene/«-butane
and NOx/ethene/toluene/H-butane/a-pinene, the object being to investigate the effect on the
final ozone concentration of the addition of cc-pinene to the experimental system. The
systems were then modelled using the MCM, with diurnally varying photolysis rates
calculated for the latitude of Valencia, (39.5°N), and the appropriate day of the year using the
UVFLUX model (Hayman, 1997; Jenkin et al., 1997b). The initial concentrations and
conditions of the two experimental runs are summarised below in Table 1.
Table 1 - Summary of initial concentrations and conditions used in the EUPHORE chamber
Start time
End time
Average temperature (°C)
Maximum temperature (°C)
Pressure (mbar)
no a-pinene
08:40
16:29
24
26.5
1004
with a-pinene
08:15
16:05
25.1
27.3
1001
-43-
-------
Initial concentrations (ppb):
[NO]
[toluene]
[g-pinene]
177
20
254
257
73
0
178
16
246
248
68
28*
The results from the simulation are shown in Figures 1 and 2 (the smog chamber data are
preliminary and have been provided by Dr. Lars Ruppert of the University of Wuppertal).
•100 T
350 •
300 -
250 •
200 -
150 •
100 •
50 •
0 •
•
i
i i
JPJJ*»*MJS1
-
• """"^Str
,
1
ig-i • •in
7:40
7-—vu-HL'JL' ^>w X
v"i-"1-t;J7-1 - .^a^
* "^
-------
7:40
9:40
17:40
Figure 2 - results from the NOx/ethene/toluene/«-butane/a-pinene system. The measurements are indicated by
symbols and the MCM results by solid lines. (Note that the NO profiles have been omitted in order to see clearly
the a-pinene comparison)
The addition of 28 ppb of a-pinene to the system results in the production of an extra 50 ppb
of ozone under these conditions. The MCM generally reproduces the observed concentrations
of most of the components of the smog chamber systems.
Field work
Field campaigns have allowed the evaluation and testing of the MCM against field
measurements. A comparison has been made between in situ measurements of radicals (OH,
HO2 and RO2) and simulations using a constrained box model. The constraints are provided
by measurements of the stable species, such as non-methane hydrocarbons (NMHC), CO,
CH), NOX and O3. This approach depends on the short-lived nature of the radicals, which
react quickly to changes in local ambient conditions (solar flux, concentration of NOX and
NMHC), but are not directly affected by atmospheric transport. It is, therefore, sufficient to
use a zero-dimensional model to describe the fast chemical processes.
The model has been applied to field campaigns both at the Mace Head Atmospheric Research
Station (Eastern Atlantic Summer Experiment 1996, 'EASE96' and the Eastern Atlantic Spring
Experiment 1997, 'EASE97') near Galway, Ireland (53°19'34"N, 9°54'14"W) and the Cape
Grim Baseline Air Pollution Station (40°41 ' S, 144°41 ' E) in Tasmania (Southern Ocean
Atmospheric Photochemistry Experiment, 'SOAPEX'). Both sites are essentially clean air
sites, though subject to pollution in certain wind directions. During the campaigns, OH and
HO2 were determined using the PAGE technique (Creasey et al, 1997) and (HO2+SRO2) by
use of a PERCA (peroxy radical chemical amplifier) instrument (Monks et al. 1998). Other
salient measurements included the concentrations of NMHC, NO, NO2, 03, HCHO, CO,
H2O, and PAN, as well as j^D), j(NO2), aerosol and meteorological data.
-45-
-------
Measured concentrations of NMHC, CEU and CO were used to construct appropriate models
for each campaign. The product of the concentration of each hydrocarbon (and CO) measured
during each campaign and its rate coefficient for reaction with OH was calculated. Figure 3
shows the result of this calculation for several field campaigns that the Leeds group have
participated in recently. Where appropriate, campaigns have been segregated into different air
mass types. During the EASE campaigns, the air tended to be clean when from the Atlantic or
tropical regions, but polluted when it had travelled in an anti-cyclonic direction taking in
emissions from the UK or Europe. During the 1997 campaign, there was a short period where
air arrived from polar regions. During the SOAPEX campaign, there were a few periods of
baseline air when the air was very clean (NOx < 50 ppt). The AEROBIC (AEROsol formation
from jBTbgenic organic Carbon) campaign took place in a forested environment in northern
Greece and the average OH losses due to CO, CEU and NMHC are included as they are very
different to the other two locations.
As you move from left to right in figure 3, the air is becoming cleaner and cleaner, with the air
masses encountered during the AEROBIC campaign containing most NOx, and those in the
SOAPEX campaign containing least. In the clean air masses, most OH reacts with CO and
CHj (>95% in baseline air during SOAPEX cf. <15% during AEROBIC). During EASE96,
100%
Hmonoteipene I
Disobutene
|DC6alkene
iODMS
Bc-2-Butene
D Butanes
JElt-2-Butene
'01,3,5-TMB
Dl-Butene
.•Isoprene
iDPropene
- • 1,3-Butadiene
•CH4
BCO
around 95% of the OH loss due to CO and hydrocarbons could be accounted for by CO, CH*,
ethane, ethene, propane, propene, isoprene, cis- and fraws-2-butene, toluene, 1,3-butadiene,
and 1,3,5-trimethylbenzene. These species form the basis of the model, with the mechanisms
taken from the MCM. Dry deposition and heterogeneous loss terms are also added.
Figure 3 - Importance of hydrocarbons and CO for loss of OH during recent field campaigns
-46-
-------
EASE96
The model/measurement comparisons have been discussed in detail before (Carslaw et al.
1999a) and the results are summarised in Table 2. August 3rd had low concentrations of NOx
and anthropogenic hydrocarbons, whereas July 17th and 18th were both characterised by higher
NOx and NMHC, and on July 17th, high concentrations of isoprene (~ 350 ppt maximum).
Table 2 - Comparison of measured and modelled radicals from EASE96 (concentrations in molecule cm"3)
Date [OH]/106 ' [HO2]/108 £[HO2+RO2]/10*
ith
Measured Modeled
Measured
2.61
1.12
Modeled Measured Modeled
July 17th - - 2.61 2.96 2.55 3.98(3.37*)
July 18th - - 1.12 2.02(1.08f) 1.92 2.54(1.86f)
August 3rd 2.33* 3.28* - - 3.13 2.30
tva}"e obtained when isoprene peroxy radicals subtracted from model estimation 6? PERCA"observations;
value obtained when y(HO2)=l; * nonzero values only for [OH] (for more details see Carslaw et al.,
Figure 4 shows a comparison between measured and modelled peroxy radicals on July 17th.
An asymmetry about solar noon was observed in the measured peroxy radicals and model HC>2
and CBbO2 also peak after noon. Peroxy radicals formed during the initial oxidation of
isoprene by OH (isoprene peroxy radicals) peak somewhat earlier (12:30), when the isoprene
concentration is greatest.
8.00E+08 -,;
"'a 6.00E+08 -j
8 I
1 5.00E+08 -|
J, 4.00E+08 -j
e |
f 3.00E+08 -j
| 2.00E+08 -I
l.OOE+08 >
O.OOE+00
model HO2
- - model CH3O2
model 'PERCA'
-•»— isoprene peroxy radicals
• • measured HO2+RO2
—«—model 'PERCA' - isoprene peroxy radicals
4:00
6:00
8:00
10:00
12:00 14:00
time (hours)
16:00
18:00
20:00
Figure 4 - Plot showing a breakdown of peroxy radicals on J199. The model 'PERCA' predictions are shown in
black, with the actual measurements in dark grey. Note the asymmetry in the peroxy radical measurements
around solar noon (see text). Model 'PERCA' - isoprene peroxy radicals (black dashed line), model HO2
(dashed dark grey line), model CH3O2 (dashed light grey line), and isoprene peroxy radicals (solid light grey
line). The peroxy radical measurements were made by Paul Monks, University of Leicester, UK.
-47-
-------
Under clean conditions, the peroxy radical profile is symmetrical and tracks the solar flux. In
order to understand the mechanistic origins of these effects, we have looked in some detail at
the HOa production rate from various reaction channels in the isoprene scheme (Carslaw et
al, 1999c). Although the maximum production of HOa from some intermediate products in
the isoprene scheme occurs around solar noon, Figure 5 shows that there are many
interhtediate products that are responsible for HOa production much later on. The position of
these products in the isoprene oxidation scheme is shown in Figure 6. ISOPBO is an alkoxy
radical produced when peroxy radicals, formed through the initial OH oxidation of isoprene,
react with NO. Therefore, maximum HOa production occurs fairly close to the isoprene
maximum. However, some of the intermediate products such as MVK, have a much longer
lifetime than isoprene, and consequently, further breakdown products such as HMVKAO (oxy
radical) are responsible for the later generation of peroxy radicals. Indeed, some of these HOa
production routes continue to increase in magnitude throughout the afternoon (e.g. methyl-
glyoxal, MGLYOX), reflecting their position in the degradation scheme. With many such
routes producing HOa, the observed asymmetry is easily understood.
1251
10:30
11:30
12:30 13:30
time /hours
14:30
15:30
16 1
"» 14 1 (b) HMVKAO
V2
•s ion
10:30
11:30
12:30 13:30
time /hours
14:30
15:30
10:30
11:30
12:30 13:30
time /hours
14:30
15:30
Figure 5—Rate of formation of HO2 from different oxidation steps in the isoprene mechanism.
-48-
-------
ISOPBOOH
I OH
OH V^H'
HMVKBOOH >. BIACETdH
/ \
HO; / N02 \hv HOCH2CHO
/ \ if
/ *RO2 X -~
HMVKBO, —j ». HKWKBO ^— *.CH3Ca
ISOPBOj
ISOPBO -
NO2
->. MVK
(+HCHO.HOJ
/OH
•£—»•
NO
[ |hv
HMVKA02 I ""SHMVKAO -
N^
OH I
- MGLYOX
(+HCHO,HO2)
Figure 6 - Oxidation of isoprene through the MVK route (for a full explanation of all of the species, see Carslaw
etal, 1999c).
The role of formaldehyde in this context is particularly interesting. Figure 7 shows the
formation of HOa from the isoprene degradation scheme throughout the day, and separately,
the contribution from HCHO also formed through the isoprene scheme.
6.0E+06
5.0E+06
E 4.0E+06
3.0E+06 -
2 2.0E+06 -
l.OE+06
O.OE+00
10:00
-HO2 production from isoprene excluding HCHO
- HO2 production from isoprene via HCHO
11:00
12:00
13:00
time / hours
14:00
15:00
T 1.2E+06
-• l.OE+06 -r
-- 8.0E+05 |
^3
- - 6.0E+05
--4.0E+05
i.
2.0E+05 -
O.OE+00
16:00
Figure 7 - HO2 production from isoprene degradation, and shown separately, from isoprene-produced HCHO.
HCHO acts as a temporary reservoir for HC>2, locking it away until later in the afternoon.
Interpretation of the data has also shown that isoprene is responsible for 40-60% of the HCHO
formation, and 20-40% of the 2 ppb h"1 conversion of NO to NO2 (Carslaw etal., 1999c).
This analysis shows how accurate, time-resolved measurements of radical concentrations,
coupled with a detailed mechanism, can be used to understand several aspects of the
photochemical mechanism.
-49-
-------
SOAPEX
Results from this campaign are at a preliminary stage. We show a model/measurement
comparison of [OH] from February 7,1999 in Figure 8, a typical baseline day.
5.0E-HJ5 -
O.OE+00
5:00 7:00 9:00 11:00 13:00 15:00 17:00 19:00
Time (hours)
Figure 8 - Comparison of measured and modelled OH on 7* February, 1999. The PAGE measurements were
made by Dwayne Heard, David Creasey and James Lee of the University of Leeds, UK.
SOAPEX was probably the most extensive campaign to date for the Leeds group in terms of
number of observations, and we will be reporting further results in the near future.
Conclusions
The MCM has been used to model EUPHORE smog chamber data for systems including
ethene, w-butane, toluene and a-pinene. The results are very encouraging and suggest that the
MCM describes well smog chamber conditions. In addition, a methodology has been
developed, based on the MCM, for constructing box models for simulating radical
concentrations for comparison with field measurements. The model is constrained by field
measurements of stable species and used to simulate OH, HO2 and peroxy radicals. The
comparisons with field measurements are reasonably satisfactory, although modelled [OH] is
somewhat greater than measured [OH]. The detailed mechanism when coupled with
atmospheric observations can be used to understand essential aspects of the atmospheric
chemistry.
Acknowledgements
The authors wish to acknowledge S.M. Saunders (University of Leeds, UK) and R.G. Derwent
(UK Meteorological Office) for useful discussions and advice regarding this work. We also
thank other NERC ACSOE and SOAPEX participants who provided data for the model. We
would like to thank the NERC, the UK Department of the Environment, Transport and the
Regions and the EU for funding.
-50-
-------
References
Carslaw N., D.J. Creasey, D.E. Heard, A.C. Lewis, J.B. McQuaid, M.J. Pilling, P.S. Monks, B.J. Bandy and S.A.
Penkett, Modelling OH, HO2 and RO2 radicals in the marine boundary layer: 1. Model construction and
comparison with measurements /. Geophys. Res. In press, (1999a).
Carslaw N., P.J. Jacobs and MJ. Pilling, Modelling OH, HO2 and RO2 radicals in the marine boundary layer: 2.
Mechanism reduction and uncertainty analysis, submitted to J. Geophys. Res. In press, (1999b).
Carslaw N., N. Bell, A.C. Lewis, J.B. McQuaid and M.J. Pilling, A detailed study of isoprene chemistry during
the EASE96 Mace Head campaign: July 17th 1996, a case study, submitted to Atmospheric Environment,
(1999c).
Creasey D.J., P. A. Halford-Maw, D.E. Heard, M.J. Pilling and B.J. Whitaker; Measurement of OH and HO2 in
the troposphere by laser-induced fluorescence, J. Chem: Soc., Faraday Trans., 93, 2907, (1997).
Derwent, R.G., M.E. Jenkin, S.M. Saunders and M.J. Pilling, Photochemical ozone creation potentials for
organic compounds in north west Europe calculated with a master chemical mechanism, Atmos Environ 32
2419-2441, 1998.
Hayman G.D., Effects of pollution control on UV exposure. Final report (Reference,
AEA/RCEC/22522001/R002) prepared for the Department of Health on Contract 121/6377, 1997.
Jenkin M.E., S.M. Saunders, and M.J. Pilling; The tropospheric degradation of volatile organic compounds: a
protocol for mechanism development, Atmos. Environ., 31, 81, (1997a).
Jenkin M.E., G.D. Hayman, R.G. Derwent, S.M. Saunders, N. Carslaw and M.J. Pilling, Tropospheric Chemistry
Modelling: Improvements to current models and application to policy issues, First annual report (Reference,
AEA/RAMP/20150/R001 Issue 1) prepared for the Department of the Environment on Contract PECD 1/3/70
(1997b).
Monks P.S., L.J. Carpenter, S.A. Penkett, G.P. Ayers, R.W. Gillett, I.E. Galbally and C.P. Meyer, Fundamental
ozone photochemistry in the remote marine boundary layer: ,The SOAPEX experiment, measurement and theory,
Atmos. Environ., 32, 3647, (1998).
Saunders S.M., M.E. Jenkin, R.G. Derwent and M.J. Pilling (1997) Report Summary: World Wide Web site of a
Master Chemical Mechanism (MCM) for use in tropospheric chemistry models. Atmospheric Environment 31
p!249.
-51-
-------
Influence of Biogenic VOCs on Photo-Oxidant Formation:
Simulation Experiments in EUPHORE and
Comparison with Model Calculations
L. Ruppert
Fraunhofer Instilut Atmospharische Umweltforschung, IFU
Kreuzeckbahnstr. 19, D-82467 Garmisch-Partenkirchen, e-mail: ruppert@ifu.fhg.de
Introduction
The possible contribution of biogenic VOCs to tropospheric ozone formation even in highly
developed regions like central Europe or the eastern United States is strongly discussed.
Isoprene and monoterpenes account for a main fraction of the biogenic VOC emissions.
Several recent field studies have shown situations where isoprene in fact dominates the
daytime photochemistry and thereby contributes strongly to the regional ozone formation
(Staffelbach et al., 1997; Biesenthal et al, 1998; Roberts et al, 1998). Despite recent
progress, the degradation mechanisms of monoterpenes are far from being sufficiently
understood and even in the case of isoprene the quantified products account for only 60 - 70%
of the carbon balance. Due to the complexity of the compounds and their reactions, causing
exceptional analytical difficulties, this knowledge is likely to improve only gradually.
This work represents a complementary, not alternative, integrated approach to the
investigation of mechanistic details in laboratory studies.
Experimental
Several smog chamber runs have been carried out in the outdoor simulation chamber
EUPHORE in Valencia, Spain (Becker, 1996). A three-component VOC mixture (base mix)
was chosen as a reference case. The composition of the base mix (n-butane, 50%-C, ethene
and toluene 25%-C, each) was chosen (1) to represent the proportions of the main VOC
classes, alkanes, alkenes and aromatic hydrocarbons, typically measured in relatively polluted
ambient air and (2) to deal with compounds whose oxidation pathways are relatively well
known, in order to reduce the uncertainties for the modeling of the reference system. By
adding relatively small amounts of a fourth VOC of interest, e.g. a biogenic one, changes in
the NOX transformation and ozone formation relative to the base mix runs should be observed.
This experimental approach is mainly based on similar experiments by Carter et al, 1995,
who used indoor smog chambers with artificial light sources.
In this contribution we report on experiments with isoprene, a-pinene, limonene and the
potential fuel additive di-ethoxy methane (DEM) as additives to the base mix. The initial
carbon and NOX concentrations of 2 ppm-C and 200 ppb, respectively, as well as the
proportions of the base mix VOCs were kept constant in all experiments. In every experiment
the mixture was exposed to sunlight for at least 7 h, centered around midday. Besides the
reacting VOCs, ozone, NO and NO2, also the concentrations of a number of reaction products
could be measured time-resolved, e.g. formaldehyde, 2-butanone, methyl vinyl ketone,
methacrolein and PAN.
-52-
-------
Experimental Results
In Fig. 1 the concentration-time profiles of several species from two experiments, a base case
and one with added ct-pinene, are plotted for comparison. The experiments were run on two
consecutive days with clear sky conditions so that light, temperature and chamber conditions
are virtually equal during both runs.
The most obvious difference lies in the enhanced ozone formation in the a-pinene
experiment, which continues even after the a-pinene has completely reacted. The accurate
data analysis revealed that in the a-pinene experiment, compared to the base run, [OH] is
slightly suppressed in the beginning of the run, when the overall OH-reactivity is substantially
increased due to the highly reactive a-pinene, but significantly increased in the second half of
the experiment, after the a-pinene had reacted.
400-•
solid lines/signs:
base mix + a-pinene
open signs:
base mix only
7:30 8:30 9:30 10:30 11:30 12:30 13:30 14:30 15:30 16:30
GMT
Fig. 1: Comparison of concentration/time profiles from two EUPHORE photosmog runs
(April, 1998): Base mix + a-pinene (thick lines/signs) - base mix only (open signs /
thin lines)
Fig. 2 (left) shows the A[O3]-A[NO] (corrected for dilution losses) from four runs, two with
the base mix and two with a-pinene added. This quantity serves as a measure for the total
amount of NO oxidized to NO2 by peroxy radicals, the step directly linked to ozone
production. Besides the experiments from Fig. 1, two experiments were evaluated which were
characterized by partly cloudy conditions with significantly reduced radiation intensities. This
reduction leads to a prompt decrease of the oxidation rate, showing that one or several
photolysis reactions influence the radical formation. As the water vapor concentration in the
experiments is very low (ca. 0.2 mbar), the photolysis of Os, which is the important step in the
atmosphere, is not likely to be of importance in the EUPHORE chamber.
Nevertheless the increased ozone formation rate in the a-pinene runs is obvious. This is even
clearer in the case of isoprene, from the respective plot shown in Fig. 2 (right). Here five runs
-53-
-------
from July, 98 are displayed, made under very similar (clear sky) conditions. Whereas the
addition of liinonene does not lead to a significant deviation from the base-mix, the
experiment with DEM is characterized by a strong decrease in the ozone formation rate.
1 SCO-
g>400
-base mix
(cloudy)
+ a-pinene
+ a-pin (cloudy)
300-
t
-base mix
+ isoprene
+ isoprene (2)
+ DEM
+ limonene
08:15 09:15 10:15 11:15 12:15 13:15 14:15 15:15
GMT
08:45 09:45 10:45 11:45 12:45 13:45 14:45 15:45 16:45
GMT
Fig. 2: Comparison of AfOsJ-AfNO] in base mix runs with a-pinene (left) and isoprene,
limonene and di-ethoxy methane (DEM) experiments (right), respectively. The left
plot also shows the influence of variations in the sunlight intensity.
Simulation Results
A number of experiments including several base mix runs, experiments with added isoprene
and a-pinene and one experiment with a-pinene as single VOC were simulated with the
RACM mechanism from Stockwell et al., 1997. Results from the model runs, compared with
the measured concentrations are displayed in Figs. 3-5.
Experimental
data
o NO2
» n-butane
• toluene *5
• ethene
RACM
simulation
o3
fMMMMMt 002
hc3
ete
10
11 12 13
time (GMT) [h]
14
15
16
Fig. 3:
EUPHORE - photosmog run: Base mix (n-butane, ethene, toluene: 2 ppm-C +
200 ppb NOX). Comparison of experimental with simulated data (RACM).
-54-
-------
No "artificial" sources of OH ("chamber-wall source") were used in the simulations. Only
small amounts of HONO (< 1 ppb) were assumed to be present initially.
thick lines:
model results
10:00
11:00
12:00
14:00
15:00
16:00
13:00
Time (GMT)
Fig. 4: EUPHORE - photosmog run: 90 ppb cc-pinene + 280 ppb NOX. Comparison of
experimental with simulated data (RACM).
10 11 12 13 14 15 16 17
10 11 12 13 14 15 16 17
time (GMT) [h]
Fig. 5: Photosmog experiment with base-mix + isoprene. Comparison of experimental and
simulated concentration time plots. Left plot: Original RACM, right: modified
RACM code (see text and Table 1).
Whereas excellent agreement between measured and simulated concentrations (of VOCs, NOX
and ozone) was observed for experiments with the base-mix (Fig. 3), RACM clearly under-
predicts the reactivity in experiments with added isoprene (Fig. 5). Therefore parts of the
RACM dealing with isoprene have been modified and updated using results from recent
laboratory studies. An improved simulation of the measured data is obtained for all runs
investigated so far (Fig. 5, right).
-55-
-------
Only the product formation from the reaction of isoprene with OH has been changed so far,
since this is the only relevant reaction under the experimental conditions. The strongest
influence showed the decrease of the organic nitrate yield (from 0.153 to 0.044) and the
increase of the reactivity of the reaction products (see Table 1).
Table 1: Modification of the product formation from isoprene + OH in the RACM
Product
MACR
OLT
OLI
HCHO
ONIT
Product
RACMa
0.446
0.354
-
0.606
0.153
Yield
new*
0.57
-
0.43
0.57
0.044
Explanation
methacrolein
(new: C4-carbonyls)
terminal alkene
(instead of MVK)
internal alkene
formaldehyde
organic nitrates
a) Stockwell et al. 97, based on data from Paulson et al. 92; b) results from Tuazon and Atkinson, 91; Kwok et
al. 95; Chen et al. 98; Ruppert and Becker, in press
It has to be stressed that no attempts had been made to fit any of the modified variables in the
mechanism to the EUPHORE data, but that the changes introduced in the RACM are purely
based on independent laboratory studies. The modifications were introduced without
increasing the number of species or reactions in the RACM.
Summary and Conclusions
Q Addition of a-pinene / isoprene to the base mix increases the rate of ozone formation
(amount of ozone formed + NO oxidized in a certain time) (reactivity increase!). Addition
of di-ethoxy methane shows the opposite effect. If this is valid also under different
conditions (e.g. VOC / NOx-ratio) needs to be tested in future experiments.
Q a-Pinene and isoprene increase the average OH-concentration (possibly decrease at the
beginning of the experiment, due to increased initial OH-reactivity).
D Butanone was measured as product from n-butane + OH with a molar yield of 35±5 %.
Further reaction products which could be measured in the experiments by GC or FTIR
include methacrolein, methyl vinyl ketone (from isoprene), formaldehyde, acetaldehyde,
HNO3andPAN.
Simulations with RACM:
Q No chamber specific reactions, besides a constant and known dilution factor, had to be
introduced into the mechanism. Especially, no artificial OH-source was required, except a
small and realistic initial HONO concentration. Further chamber characterization studies
are required to justify this treatment.
-56-
-------
rj Base mix runs could be reasonably simulated.
Q RACM severely underestimates the ozone formation (and NO to NC>2 conversion) in
experiments with added isoprene and in a pure a-pinene experiment (Fig. 4), possibly due
to underestimation of the reactivity of the oxidation products from a-pinene.
Q Mechanism modifications according to recent laboratory results improve the mechanism
performance.
»
Acknowledgements
We thank Klaus Wirtz and his co-workers from CEAM (Valencia) for help and assistance in the
experiments in EUPHORE. This work was supported by the Bundesministerium fur Bildung,
Wissenschaft, Forschung und Technologic within the tropospheric research programme (TFS).
References
Becker, K.H. (ed), The European Photoreactor EUPHORE. EC final report EV5V-CT92-0059, Wuppertal
(1996).
Biesenthal, T.A., J.W. Bottenheim, P.B. Shepson, S.-M. Li and P.C. Brickell; The chemistry of biogenic
hydrocarbons at a rural site in eastern Canada, J. Ceophys. Res. 103 (1998) 25487-25498.
Carter, W.P.L., J.A. Pierce, D. Luo and I.L. Malkina; Environmental chamber study of maximum incremental
reactivities of volatile organic compounds, Atmos. Environ. 29 (1995) 2499-2511.
Chen, X., D. Hulbert and P.B. Shepson; Measurement of the organic nitrate yield from OH reaction with
isoprene, J. Geophys. Res. 103 (1998) 25563-25568.
Kwok, E.S.C., R. Atkinson and J. Arey; Observation of hydroxycarbonyls from the OH radical-initiated reaction
of isoprene, Environ. Sci. Technol. 29 (1995) 2467-2469.
Paulson, S.E., R.C. Flagan and J.H. Seinfeld; Atmospheric photooxidation of isoprene. Part I: The hydroxyl
radical and ground state atomic oxygen reactions, Int. J. Chem. Kinet. 24 (1992) 79-101.
Roberts, J.M., et al.; Measurements of PAN, PPN, and MPAN made during the 1994 and 1995 Nashville
intensives of the Southern Oxidant Study: implications for regional ozone production from biogenic
hydrocarbons, J. Ceophys. Res. 103 (1998) 22473-22490.
Ruppert, L. and K.H. Becker; A product study of the OH radical-initiated oxidation of isoprene: Formation of C5-
unsaturated diols, Atmos. Environ, in press.
Staffelbach, T. et al.; Photochemical oxidant formation over southern Switzerland. 1. Results from summer 1994,
J. Geophys. Res. 102 (1997) 23345-23362.
Stockwell, W.R., F. Kirchner, M. Kuhn and St. Seefeld; A new mechanism for regional atmospheric chemistry
modeling,/. Geophys. Res. 102 (1997) 25847-25879.
Tuazon, E.G. and R. Atkinson; A product study of the gas-phase reaction of isoprene with the OH radical in the
presence of NOX, Int. J. Chem. Kinet. 22 (1990) 1221-1236.
-57-
-------
Atmospheric Fate of Unsubstituted Alkoxy and Carbonyl Radicals
O. Shestakov, S. Jagiella, J. Theloke, H. G. Libuda, and F. Zabel
Institut far Physikalische Chemie, Universitat Stuttgart, Pfaffenwaldring 55, D - 70569 Stuttgart
Introduction
Alkoxy (RiC(R2)(R3)O.) and Carbonyl (RC(.)O) radicals are important intermediates in the deg-
radation chain of hydrocarbons. Both classes of radicals are subject to competing reaction chan-nels
leading to different products which exhibit different ozone formation potentials:
Alkoxy radicals -
(1) Reaction with O2 (koz, => aldehyde/ketone + HO2);
(2) thermal decomposition (kais, -> aldehyde/ketone + alkyl/H);
(3) isomerization (kto, => hydroxyaldehyde/hydroxyketone).
Carbonyl Radicals -
(4) Reaction with O2 (koz,=> RO2, + NO2 => peroxynitrate (alkyl substituted PAN));
(5) thermal decomposition (kdis, => CO + alkyl).
In the present work, the ratios kdis/kb2 were determined for a number of unsubstituted C4 and Cs
alkoxy and carbonyl radicals, linear and branched, with the main aim of measuring the effect of
branching at the a-C atom and of the chain length of R on the ratio kdis/ko2.
Experimental
Experiments are performed in a 12 L temperature controlled photoreactor from stainless steel
(Figure 1).
Fan
Ranj
Temperature Gauge
Temperature Gauge
cave Mirrors
is Lamp
Gas Inlet ' Vacuum Pump
Inlet System
Figure 1. Temperature controlled photoreactor (v = 12 L) from stainless steel
-58-
-------
Alkoxy radicals are produced by photolyzing the appropriate iodides in the presence of O2, NO, and
N2 as a buffer gas. For example, in the photolysis of 3-iodopentane the following mechanism is
effective:
C2H5C(H)(I)C2H5 + hv(254 nm) => C2H5C(H)(.)C2H5
C2H5C(H)0)C2H5 + O2 + M => C2H5C(H)(OO.)C2H5 + M
C2H5C(H)(OO.)C2H5 + NO => C2H5C(H)(O.)C2H5 + NO2
C2H5C(H)(0.)C2H5 + 02 => C2H5C(0)C2H5 + H02
C2H5C(H)(O.)C2H5 + M => C2HSCHO + C2H5 + M
The products in bold letters are analyzed by FT-IR absorption (optical pathlength = 2 m), and the
rate constant ratios kdis/koa are determined using the equation
(I)
kdis_A[C2H5CHO]x[023
kn A[C2H5C(0)C2H5]
Carbonyl radicals are formed by stationary photolysis of Br2 in the presence of the corresponding
aldehyde, O2, NO2, and N2 as a buffer gas. The important reactions taking place are:
Br2 + hv(> 420 nm) => 2 Br
RC(O)H + Br => RCO +" HBr
RCO + M => R + CO + M,kdiS
RCO + O2 + M => RC(O)O2 + M, ko2
RC(O)O2 + NO2 + M => RC(O)O2NO2 + M
The products in bold letters are analyzed by FT-IR absorption, and the rate constant ratios kdiS/kO2
are determined using the equation
(II)
kdis = A[CO]x[02]
k0, A[RC(0)02N02]
Results and Discussion
/. Alkoxv Radicals
In Figures 2 and 3, the experimental data points are shown for 3-pentoxy. For large O2 partial
pressures, the rate constant ratios ko2/kdis = A[C2H5C(O)C2H5]/(A[C2H5CHO]x[O2]) (inverse of eq.
(I)) approach Wkdis = (1.32±0.33)xlO'19 cm3 or Whto = (7.6+1.9)xl018 cm'3. This value
corresponds to the following product distribution at 298 K in synthetic air:
Reaction with O?
(40%): 3-pentanone + HO2
-59-
-------
Thermal decomposition (60 %): propanal +
1.4e-18
CM
o
1.20-18 .
l.Oe-18 .
ra
5
§• 8.0e-19 -
*-• 6.0e-19 .
••»
¥
g 4.0B-19 .
B
?• 2.0e-19 .
0.0
3-pentoxy
o
0
£ 0_ O g Q o Q Q m
400
600
1000
[mbar]
Figure 2. Determination of koz/kdis for 3-pentoxy radicals at 298 K, 1 bar (M = N2 + O2)
(open points: experimental; full points and broken line: calculated with a simple mechanism)
3 -pentoxy
200 400 600
p [mbar]
Figure 3. Determination of ko2/kdis for 3-pentoxy at 298 K, 1 bar (M = N2 + O2)
-60-
-------
The increase of ko2/kdis at low 02 partial pressures originates in an additional, 02 independent source
of 3-pentanone the nature of which is still unknown. Possible reactions which can explain this
additional formation of 3-pentanone are:
3-pentoxy + NO => 3-pentanone + HNO
3-pentoxy + M => 3-pentanone+ H + M
Depending on the nature of the additional 3-pentanone source, the C>2 channel can be larger by
20 % under atmospheric conditions (corresponding to the intercept on the vertical axis in fig. 3). A
behaviour similar to that shown in figures 2 and 3 was also observed for the other investigated
alkoxy radicals. Further studies on these and other alkoxy radicals are under way.
In table 1, data which have been determined in the present work for 3-pentoxy, 2-butoxy,
/-butoxy, and 3-methyl-2-butoxy at 298 K are summarized and compared with experimental, semi-
empirical and ab-initio values from the literature.
-61-
-------
Table 1. Data on k«jis/ko2 at 298 K for selected alkoxy radicals
2-Butoxy
kdis/ko2(298K)[cm^]
2.1xl016 "
(2.2±0.4)xi018 l)
1.9xl018
(3.6±2.1)xl018
3.7xl018
2.5xl018
(2.9±0.6)xl018
Reference
Carter etal.^
Cox et al. i}
Blitz etal.^
Heinetal. s)
Somnitz, Zellner 6;
Atkinson '*
this work
Remarks
complex reaction system , ','...
Mixture of isomers
2-butoxy from 2-butyl nitrite + hv '-
pure 2-butoxy isomer ,,- s
ab-initio calculation "'"''
semi-empirical s « • „ , ,
pure 2-butoxy isomer "
z-Butoxy
kdis^b2(298K)[cm-"]
1.3xl018
6.2xl018
(5.9±1.7)xl018
Reference
Atkinson °
Atkinson 7)
this work
Remarks - - """
semi-empirical, with IUPAC value ^ for
the heat of formation of i-C^R-?
semi-empirical, with JPL value 9? for the
heat of formation of i-CsH? '.'''.
pure i-butoxy isomer , , ;, . ,
3-Methyl-2-butoxy
kdis/ko2(298K)[cm^]
8.6xlOiy
(6.7±1.8)xlOw
Reference
Atkinson '*
this work
Remarks " -<••<•>-'*<*
semi-empirical . - •"***
pure 3-pentoxy isomer " ",- , „ "'
3-Pentoxy
kd!s/ko2(298 K) [cm^]
(3.8^-s.i.9)xl018
S.OxlO18
5.3xl018
(7.6±1.9)xl018
Reference
Atkinson etaL 10J
Somnitz, ZelltierOJ
Atkinson et al. l}
this work
Remarks _ , f/~
Mixture of isomers , '-"-..
ab-initio calculation , '
semi-empirical, " ' '•
pure 3-pentoxy isomer , - "
References and notes: !) original value extrapolated to 298 K; 2) W.P.L. Carter, A.C. Lloyd, J.L. Sprung, J.N. Pitts, Jr.,
Int. J. Chem. Kinet. 11(1979)45; 3) R.A. Cox, K.F. Patrick, S.A. Chant, Environ. Sci. Technol. 15(1981)
587;4) M.A. Blitz, M.J. Pilling and P.W. Seakins, 15th Int. Symp. on Gas Kinetics, 1998, Bilbao; ' H. Hein, H. Somnitz,
A. Hoffmann, R. Zellner, CMD- Konferenz, Karlsruhe, 1998; 6' H. Somnitz, R. Zellner, 15* Int. Symp. on Gas Kinetics,
1998, Bilbao; 7) R. Atkinson, J. Phys. Chem. Ref. Data 26(1997)215; 8) R. Atkinson et al., J. Phys. Chem. Ref. Data
26(1997)521;9) W.B. DeMore et al., JPL Publication 97-4, Pasadena, 1997;10) R. Atkinson, E.S.C. Kwok, J. Arey, S.M.
Aschmann, Faraday Discuss. 100(1995)23
-62-
-------
In table 2, all the alkoxy radicals HC(Ri)(R2)O are listed for which isomerization via a six-
membered transition state is impossible and thus intramolecular H atom migration is unlikely
(see e.g. [1,2]). During thermal decomposition, the larger (or equally large) fragment (here:
R2) generally leaves the alkoxy radical as an alkyl radical whereas the smaller (or equally
large) fragment (here: RI) ends up as the carbonyl compound RjC(O)H. Table 2 supports the
assumption underlying the semi-empirical method to estimate k
-------
For atmospheric applications, kdjS/(ko2x[C>2]) values between = 0.1 and = 1 0 (last column in
table 2) are most interesting since here the competition between channels (1) and (2) is most
effective.
2. Carbonvl Radicals
Experimental results on kdis/ko2 at 1 bar, M = Oa + Na are summarized in table 3. Since ex-
perimental data on the recombination of acetyl radicals with 62 [2] suggest that there is no
notable barrier for reaction (4), the temperature dependence of k
-------
Table 4. Estimated data on kdis, total pressure 1 bar, M = O2 + N2
RCO
ft-butyryl
«-pentanoyl
3-methylbutyryl
2-methylpropionyl (= /-butyryl)
2-methylbutyryl
pivaloyl
kdists"1], this work"
317K
< 2,300
< 2,300
3,200
40,300
52,500
880,000
298 K
< 700 2>
< 700 2)
1,0002)
12,1002)
1 5,800 2)
295,000
Notes: J> with k02 = 3.2xl(T12 cmV [3];
•' extrapolated from higher temperatures with Ea = 49.6 kJ mol"1 from Tomas et al. [4].
The present data at 298 K are higher for pivaloyl and /-butyryl than recent data of Tomas et al.
[4] and much lower than early values by Cadman et al. [5,6]. The major part of these dis-
crepancies may be due to the long range of extrapolation necessary to convert the rate con-
stants of refs. 4-6 from the high temperatures of the experiments to 298 K.
The data in tables 3 and 4 show that
(i) kdis increases by about a factor of 15 for each H atom connected to the cc-C-atom in
CH3CO which is replaced by a methyl group (corresponding to increasing branching of
R);
(ii) even for the thermally most unstable radicals of table 4, i.e. pivaloyl, only 1.8 %
decompose rather than add O2 at 298 K in dry air.
Future Work
Work on alkoxy and carbonyl radicals will be continued, using both the photoreactor shown
in fig. 1 and a new reaction chamber from quartz (figure 4) which is under construction and
close to being finished. It consists of two' concentric quartz tubes with teflon-coated end
flanges from stainless steel. The space between the quartz tubes is filled with the cooling
agent (silicon oil); photolysis lamps are placed around the outer quartz tube. The quartz tube
(v = 190 L) has several advantages as compared to the cell from stainless steel:
(i) More homogeneous distribution of temperature and photolysis light intensity;
(ii) larger sensitivity due to longer light paths (by a factor of 20 both in the IR and UV/VIS);
(iii) much better volume:surface ratio;
(iv) less reactive wall materials.
The final experimental set-up is shown in figure 5.
-65-
-------
Temerature gauge
White mirror systems (I = 50 m. UV und IR )
Photolysis lamp
Heating/Cooling
agent
•JDiode array
spectrometer
Quartz tubes
Vacuum pump
FTIR-
Spectrometer
Heating/Cooling
agent
Figure 4. Sketch of the 190 L photoreactor from quartz
DIodo array detector
Movable mirror
-- UV
Absorption cell from
stainless steel (12 L)
Photoreactor from quartz (190 L)
Figure 5. Experimental set-up for the investigation of UV spectra of gaseous
compounds and of the kinetics of chemical reactions in the gaseous phase,
using a diode array spectrometer and an FT-IR spectrometer
-66-
-------
References
[1] R. Atkinson, J. Phys. Chem. Ref. Data 26(1997)215-290
[2] H. Somnitz, R. Zellner, 15th Int. Symp. on Gas Kinetics, 1998, Bilbao
[3] R. Atkinson, D. L. Baulch, R. A. Cox, R. F. Hampson, Jr., J. A. Kerr, M. J. Rossi, J. Tree, J. Phys. Chem.
Ref. Data 28 (1999)191-393
[4] A. Tomas and R. Lesclaux, 15th Int. Symp. on Gas Kinetics, 1998, Bilbao
[5] C. D. P. Cadman, A. F. Trotman-Dickenson, and A. J. White, J. Chem. Soc. (A) (1970)3189-3193
[6] P. Cadman, C. Dodwell, A. F. Trotman-Dickenson, and A. J. White, J. Chem. Soc. (A) (1970)2371-2376
-67-
-------
Poster Session
Session Chair
Roger Atkinson
-------
ATMOSPHERIC CHEMISTRY OF SELECTED HYDROXYCARBONYLS
Sara M. Aschmann, Janet Arey and Roger Atkinson
Air Pollution Research Center
University of California
Riverside, CA 92521, U.S.A.
Introduction
Volatile organic compounds present in the atmosphere can undergo photolysis and chemical
reaction with OH radicals, NO3 radicals, and O3 [ 1 ], with the OH radical reaction being an important,
and often dominant, atmospheric loss process [1,2]. Hydroxycarbonyls are formed as atmospheric
reaction products of organic compounds; for example, a-hydroxycarbonyls are formed from the OH
radical-initiated reactions of alkanes [1,3] and a-hydroxycarbonyls can be formed from the OH
radical-initiated reactions of alkenes [1,3]. Because of difficulties in the analysis of this class of
compounds, few data are presently available concerning the atmospheric chemistry of these
compounds [1,2]. It is expected that the dominant atmospheric loss process for the
hydroxycarbonyls not containing unsaturated >C=C< bonds is by daytime reaction with the OH
radical [1], with photolysis also being possible.
In this work, we have measured the rate constants for the gas-phase reactions of the
hydroxycarbonyls l-hydroxy-2-butanone,3-hydroxy-2-butanone, l-hydroxy-3-butanone, 1-hydroxy-
2-methyl-3-butanone, 3-hydroxy-3-methyl-2-butanone and 4-hydroxy-3-hexanone with OH radicals,
NO3 radicals, and O3 at 296 ± 2 K. In addition, we have investigated the products formed from the
reactions of the OH radical with the a-hydroxycarbonyls 3-hydroxy-2-butanone and 4-hydroxy-3-
hexanone.
Experimental
Experiments were carried out in a 7900 liter Teflon chamber, equipped with two parallel
banks of Sylvania F40/350BL blacklamps for irradiation, at 296 ± 2 K and 740 Torr total pressure
of purified air at -5% relative humidity. This chamber is fitted with a Teflon-coated fan to ensure
the rapid mixing of reactants during their introduction into the chamber. Rate constants for the OH
radical and NO3 radical reactions were determined using relative rate methods in which the relative
disappearance rates of the hydroxycarbonyls and a reference compound, whose OH radical or NO3
radical reaction rate constant is reliably known, were measured in the presence of OH radicals or
NO3 radicals [4]. Providing that the hydroxycarbonyls and the reference compound reacted only
with OH radicals or NO3 radicals, then [4],
In{([hydroxycarbonyl]t0/[hydroxycarbonyl]t) - Dt} = (I)
(kj/k2)ln{([hydroxycarbonyl]t0/[hydroxycarbonyl]t) - Dt}
-69-
-------
where [hydroxycarbonyl]to and [reference compound]^ are the concentrations of the hydroxycarbonyl
and reference compound, respectively, at time t0, [hydroxycarbonyl], and [reference compound], are
the corresponding concentrations at tune t, Dt is a factor to account for any dilution due to additions
to the chamber during the reactions, and k, and k2 are the rate constants for reactions (1) and (2),
respectively.
OH
OH
} + hydroxycarbonyl - products
} + reference compound - products
(1)
(2)
OH radicals were generated by the photolysis of methyl nitrite (CH3ONO) in air at
wavelengths >300 nm, and NO was added to the reactant mixtures to suppress the formation of O3
and hence of NO3 radicals. The initial reactant concentrations were similar to those used in
analogous studies conducted in this laboratory [4]. The hydroxycarbonyls were also photolyzed in
air in the presence of cyclohexane (to scavenge any OH radicals present).
NO3 radicals were generated in the dark by the thermal decomposition of N2O5 [18], and 1-
butene or crotonaldehyde (CH3CH=CHCHO) were used as the reference compounds. Experiments
were carried out as described previously [4]. The concentrations of the hydroxycarbonyls and the
reference compounds were measured by gas chromatography with flame ionization detection (GC-
FID) during the experiments.
Rate constants, or upper limits thereof, for the reactions of the hydroxycarbonyls with O3
were determined in the dark by measuring the decay rates of the hydroxycarbonyls in the presence
of measured concentrations of O3 [4]. Cyclohexane was added to the reactant mixtures to scavenge
any OH radicals formed hi the reaction systems. Ozone concentrations were measured by ultraviolet
absorption using a Dasibi 1003-AH ozone analyzer.
Products were identified and quantified from the reactions of the OH radical with 3-hydroxy-
2-butanone and 4-hydroxy-3-hexanone, by GC-FID [4] and by combined gas chromatography-mass
spectrometry (GC-MS).
Results
Photolysis of the hydroxycarbonyls in air at the same light intensity as used in the OH radical
rate constant determinations for up to 60 min showed <2% loss of the hydroxycarbonyls. Hence
photolysis of the hydroxycarbonyls studied was of no importance during the irradiations employed
for the determination of the OH radical reaction rate constants. OH Radical Reactions
A series of CH3ONO - NO - hydroxycarbonyl - «-octane - air irradiations were carried out.
The rate constant ratios k,^ and rate constants k, obtained from least-squares analyses of the data
are given in Table 1. GC-FID analyses of these irradiated mixtures" showed "the formation of
-70-
-------
products from the 3-hydroxy-2-butanone and4-hydroxy-3-hexanone reactions (but not from the other
hydroxycarbonyls). Matching of GC retention times and mass spectra with those of authentic
standards showed that the products are 2,3-butanedione (biacetyl) from 3-hydroxy-2-butanone and
3,4-hexanedione from 4-hydroxy-3-hexanone. Least-squares analyses of the data obtained (with
corrections to the measured a-dicarbonyl concentrations to account for minor losses due to
photolysis and reaction with OH radicals) lead to formation yields of 2,3-butanedione from 3-
hydroxy-2-butanone of 79 ± 14% and of 3,4-hexanedione from 4-hydroxy-3-hexanone of 84 ± 7%,
where the indicated erors are two least-squares standard deviations combined with estimated overall
uncertainties in the GC-FID response factors for the a-hydroxycarbonyls and a-dicarbonyls of ±5%
each.
NO3 Radical Reactions
A series of reacting NO3 - N2O5 - NO2 - hydroxycarbonyl - 1 -butene (and/or crotonaldehyde) -
air mixtures were carried out. While no significant (<2%) consumption of 3-hydroxy-3-methyl-2-
butanone occurred under conditions where 78% of the initial 1 -butene had reacted, losses of the other
hydroxycarbonyls were observed. Plots of Equation (I) showed distinct curvature, with the slope of
the curves decreasing with the extent of reaction to close to zero towards the end of the reactions
(irrespective of whether 1-butene or crotonaldehyde was used as the reference compound and of the
initial NO2 concentration). While the reasons for this behavior are not presently understood, we used
the final values of {ln([hydroxycarbonyl]to/[hydroxycarbonyl]t) - DJ and {ln([reference
compound]to/[reference compound],) - Dt} to obtain upper limits to the rate constant ratios k,/k2 and
hence to the rate constants Ic, (which are given in Table 2).
O3 Rate Constants
The measured maximum losses of gas-phase hydroxycarbonyls in the presence of 3.44 x 1013
molecule cm"3 of O3 over a period of 3.8 hr were <2-3% in each case, and within the analytical
uncertainties. Assuming maximum hydroxycarbonyl losses due to reaction with O3 of 5% leads to
upper limits to the rate constants at 296 ± 2 K of k3 <1.1 x 10~19 cm3 molecule
hydroxycarbonyls.
"' s"1 for each of these
Discussion
The lack of observed reaction of the hydroxycarbonyls studied with O3, and the slow
reactions with the NO3 radical, are consistent with literature data for aliphatic alcohols and ketones
[1,5]. As is the case for the reactions of the OH radical with aliphatic alcohols and ketones, the OH
radical reactions proceed by H-atom abstraction from the various C-H bonds and (generally to a
minor extent) from the O-H bond [1,2]. The a-dicarbonyl products observed from 3-hydroxy-2-
butanone and 4-hydroxy-3-hexanone clearly arise after H-atom abstraction from the activated tertiary
H-atom of the CH(OH) group:
OH + CH3C(O)CH(OH)CH3 - H2O + CH3C(O)C(OH)CH3
CH3C(O)C(OH)CH3 + O2 - CH3C(O)C(O)CH3 + HO2
Combining our measured room temperature rate constants with a 24-hr average tropospheric
-71-
-------
OH radical concentration of 1.0 x 106 molecule cm"3 [6,7] leads to calculated lifetimes of: 1 -hydroxy-
2-butanone, 1.5day;3-hydroxy-2-butanone, 1.1 day; l-hydroxy-3-butanone, 1.4 day; l-hydroxy-2-
methyl-3-butanone, 0.7 day; 3-hydroxy-3-methyl-2-butanone, 12 days; and4-hydroxy-3-hexanone,
0.8 day.
Acknowledgements
We gratefully thank the U.S. Environmental Protection Agency, Office of Research and
Development, for supp,orting this research through Assistance Agreement R-825252-0-01. While
this research has been'supported by the U.S. Environmental Protection Agency, it has not been
subjected to Agency review and therefore does not necessarily reflect the views of the Agency and
no official endorsement should be inferred.
References
1. R. Atkinson, J. Phys. Chem. Ref. Data, Monograph 2, 1 (1994).
2. R. Atkinson, /. Phys. Chem. Ref. Data, Monograph 1, 1 (1989).
3. R. Atkinson, J. Phys. Chem. Ref. Data, 26, 215 (1997).
4. S. M. Aschmann, and R. Atkinson, Int. J. Chem. Kinet., 30, 533 (1998).
5. R. Atkinson and W. P. L. Carter, Chem. Rev., 84, 437 (1984).
6. R. Prinn, R. F. Weiss, B. R. Miller, J. Huang, F. N. Alyea, D. M. Cunnold, P. J. Fraser, D.
E. Hartley, and P. G. Simmonds, Science, 269, 187 (1995).
7. R. Hein, P. J. Crutzen, and M. Heimann, Global Biogeochem. Cycles, 11, 43 (1977).
-72-
-------
Table 1. Rate constant ratios kj/k2 and rate constants k, for the gas-phase reactions of the OH
radical with hydroxycarbonyls at 296 ± 2 K.
hydroxycarbonyl
k,/k,:
1012xk!
(cm3 molecule"1 s"1
CH3CH2C(O)CH2OH 0.893 ± 0.083
CH3C(0)CH(OH)CH3 1.19 ± 0.05
CH3C(O)CH2CH2OH 0.940 ± 0.082
CH3C(O)CH(CH3)CH2OH 1.87 ± 0.09
(CH3)2C(OH)C(0)CH3 0.108 ± 0.036
CH3CH2C(0)CH(OH)CH2CH3 1.74 ± 0.07
7.7 ±1.7
10.3 ±2.2
8.1 ±1.8
16.2 ±3.4
0.94 ± 0.37
15.1 ±3.1
"Indicated errors are two least-squares standard deviations.
bPlaced on an absolute basis by use of a rate constant of k2(«-octane) = 8.67 x 10"12 cm3
molecule'1 s"1 (±20%) at 296 K [3]. The indicated errors include the estimated overall
uncertainty in the rate constant k2.
Table 2. Rate constants k (cm3 molecule"1 s'1) for the gas-phase reactions of the
hydroxycarbonyls studied with OH and NO-, radicals and O, at 296 ± 2 K.
hydroxycarbonyl
1016xkMm
^H
CH3CH2C(0)CH2OH
CH3C(O)CH(OH)CH3
CH3C(O)CH2CH2OH
CH3C(O)CH(CH3)CH2OH
(CH3)2C(OH)C(0)CH3
CH3CH,C(0)CH(OH)CH7CH,
<3
<9
7.7 ±1.7
10.3 ±2.2
8.1 ±1.8
16.2 ±3.4
0.94 ± 0.37
15.1 ±3.1
-73-
-------
Products and Mechanisms of the Reactions of 1,3-Butadiene with Chlorine Atoms In Air
Weihong Wang and Barbara J. Finlayson-Pitts
Department of Chemistry
University of California, Irvine
Introduction
Chlorine atoms may be generated by reactions of sea salt particles transported inland with
air masses. Chlorine atom precursors have been measured at coastal sites and in the Arctic at
polar sunrise (Keene et al, 1993; Pszenny et al., 1993; Impey et al, 1997a, b; Spicer et a/.,
1998). Once chlorine atoms are generated, they can react with ozone, with a rate constant of 2.9
x 10"11 cm"3 molecule"1 s"1, or organics, with rate constants at ~10"'° cm"3 molecule"1 s"1, in air.
The Cl-organic reactions will eventually lead to ozone formation in the presence of NOX. Thus
chlorine atoms will have impacts on ozone level either way.
The reactions of chlorine atoms with organics proceed in a similar way as hydroxyl
radicals (OH). For Cl reactions with alkenes, Cl would add to the double bond and form some
unique chloro-carbonyl compounds, which would not otherwise be in the atmosphere, except for
these kinds of reactions. A potential approach to investigate chlorine atom production in the
troposphere is to identify and measure unique chlorine-containing products of the reactions of Cl
with organics, such as 1,3-butadiene, in air. 1,3-Butadiene was classified as a hazardous air
pollutant under 1990 Clean Air Act and is emitted from motor vehicles. If any unique chlorine-
containing products can be determined from the reaction of Cl with 1,3-butadiene, they could
serve as "markers" for chlorine atom chemistry in urban coastal areas where there are both
sources of Cl atoms and 1,3-butadiene.
We present here studies of mechanism and formation of 4-chlorocrotonaldehyde (CCA),
CIH2C
/
H
\
CCA
a unique chlorine-containing compound, from the reaction of atomic chlorine with 1,3-butadiene
in air at room temperature.
-74-
-------
Experimental
Product studies of the Cl-butadiene reaction were carried out by GC-MS (Fig. 1) and
FTIR (Fig. 2).
Na2CO3 Filter for C12
/ l\
Black Lamps
(300-400 nm)
Sarrple
Loop
Carrier Gas
Figurel. GC-MS experimental apparatus
-75-
-------
Impey, G.A., P.B. Shepson, D.R. Hastie, L.A. Barrie and K. Anlauf, Measurements of
photolyzable chlorine and bromine during the Polar Sunrise Experiment 1995, J.
Geophys. Res., 102,16005-16010, 1997b.
Keene, W.C., J.R. Maben, A.A.P. Pszenny, and J.N. Galloway, Measurement technique for
inorganic chlorine gases in the marine boundary layer, Environ. Sci. Technol, 27, 866-
874, 1993.
Pszenny, A.A.P., W.C. Keene, DJ. Jacob, S. Fan, J.R. Maben, M.P. Zetwo, M. Springer-Young
and J.N. Galloway, Evidence of inorganic chlorine gases other than hydrogen chloride in
marine surface air, Geophys. Res. Lett., 20, 699-702, 1993.
Spicer, C.W., E.G. Chapman, B.J. Finlayson-Pitts, R.A. Plastridge, J.M. Hubbe, J.D. Fast and
C.M. Berkowitz, Unexpectedly high concentrations of molecular chlorine in coastal air,
Nature, 394, 353-356, 1998.
-78-
-------
Analysis of Gas Phase Halogen Compounds Using Atmospheric Pressure lonization-
Mass Spectrometry
Krishna L. Foster1, Tracy E. Caldwell2, Thorsten Benter1, Sarka Langer3,
John C. Hemminger1, and Barbara J. Finlayson-Pitts1
'Department of Chemistry, University of California, Irvine, CA 92697-2025
2 NASA Johnson Space Center, Mail Code CB, Houston, TX 77058
3 Department of Chemistry, Section for Inorganic Chemistry, University of Goteborg, S-
412 96 Goteborg Sweden; also at SP Swedish National Testing and Research Institute,
Box 857, S-501 15 Boras, Sweden.
I. INTRODUCTION
Environmental chamber studies involve numerous chemical compounds with a variety of
physical properties that make it difficult to identify all of them with a single analytical
technique. Mass spectrometry is unique in its ability to detect most compounds, and is a
valuable component for environmental chamber studies because of this characteristic.
We show here that atmospheric pressure ionization-mass spectrometry (API-MS) in the
negative ion mode is a highly sensitive and selective technique ideal for measuring
halogen compounds such as HOC1, Ck, and Br2 both in laboratory systems and in air.
This presentation will focus on the quantitative analysis of HOC1 with API-MS [Foster
etal.,1999].
Field studies show elevated bromine measurements correlated with surface-level Os
depletion in the Arctic spring [Barrie et al., 1988]. In order to explain these
observations, a very large or recyclable source of gaseous bromine must be identified
\McConnell et al., 1992; Vogt et al., 1996]. Sea-salt aerosol and ice contain bromide
ions, and are a potential source of gaseous bromine compounds. Hypochlorous acid
(HOC1) is a potential intermediate in halogen activation involving deliquesced sea-salt
aerosol and sea-salt ice [Sander and Crutzen, 1996; Vogt et al., 1996; Abbott and Nowak,
1997 ;Chu and Chu, 1999]
HOC1 + Bf -> -» BrCl, Br2
(1)
Existing measurement techniques fail to distinguish HOC1 from other chlorine
compounds [Keene et al., 1993; Pszenny et al., 1993; Impey et al., 1997a; Impey et al.,
1997b]. Chlorine (Cla and HOC1) compounds other than HC1 have been identified
separate from HC1 and organic chlorine in mist chamber studies conducted at mid-
latitudes [Keene et al., 1993; Pszenny et al., 1993]. Other studies have been performed
wherein photolabile chlorine compounds in the Arctic were photolyzed to produce
chlorine radicals, which were reacted with propene to produce the measured compound
chloroacetone [Impey et al., 1997b]. This technique was used to measure up to 100 ppt
of photolyzable chlorine as CLj (200 ppt Cl radicals) in the high Arctic.
-79-
-------
Atmospheric pressure (chemical) ionization-mass spectrometry is a highly sensitive and
selective technique which has been used for specific measurements of inoganics [Spicer
et aL, 1998] and for the identification of organics [Kwok et al, 1995; Kwok et al, 1996a;
Kwok et al., 1996b] in air. For example Cla has been measured in the negative ion mode
using air as the chemical ionization reagent (CIR) gas. Here, Oz acts as the CIR ion by
accepting an electron in the corona discharge. In the ionization of Clz, electron transfer
can occur between 62" and Ob forming the Cla" ion observed at m/z = 70,72, and 74
[Spicer et al, 1998; Caldwell et al., 1999]. This technique is not only selective, but
sensitive as well. Spicer et al. have reported a detection limit of Ck using air as the CIR
gas of ~16 ppt in ambient air [Spicer et al., 1998].
Chemical ionization is a soft ionization technique wherein ion-neutral adduct formation is
possible. For example, to detect HONO in ambient air, Spicer et al. created Cl" CIR ions
by using chloroform as the CIR reagent gas. The ion-neutral adduct (HONOC1)", was
formed and used to selectively monitor HONO in air. The estimated detection limit in
this study was 0.5 ppb, illustrating that sensitive measurements of ion-neutral adducts can
be achieved with API-MS [Spicer et al., 1993].
The instruments used in these §tudies employed tandem quadrupoles which were used to
further enhance the selectivity of chemical ionization produced ions. Ql scans were used
to measure a range of m/z ratios with a single quadrupole to survey ions. Multiple
reaction monitoring (MRM) scans measured the signal intensity for parent-daughter ion
pairs. This is a highly selective technique used for quantitative analysis.
Here we present techniques for quantitative measurements of HOC1 with API-MS using
air as the CIR gas for field and laboratory studies. We also explore the effects of
alternative CIR gases, such as bromoform (CHBra), oh the selectivity and sensitivity of
HOC1 measurements with API-MS. A detailed discussion of this work will be presented
elsewhere [Foster et al., 1999]. A summary is presented below.
n. EXPERIMENTAL
A schematic of the experimental apparatus is presented in Figure 1.
Cooling
By-pass bath
API-MS >»T~
Zero-air
in flow meter
HOCI
VBubbler
-80-
-------
FIGURE 1: Schematic representation of the experimental apparatus used for the
heterogeneous titration of HOCL
Samples in 35-55 L Teflon reaction chambers were flowed into to the API-MS at an
estimated flow rate of 1 L min"1. The apparatus included a glass U-tube which was
submerged in a cooling bath and equipped with a by-pass. Samples were monitored
directly by the API-MS through the by-pass, or diverted through the U-tube.
Experiments were performed to determine the effects of H2O vapor and acids on the CIR
ions in air. Dry air samples were prepared by filling an evacuated Teflon reaction
chamber with dry zero-air. Humid air samples were prepared by bubbling the zero-air
through room temperature ultra-pure water and into a Teflon reaction chamber. The
relative humidity of the air was measured as 84% using a humidity and temperature
transmitter. Concentrated HC1 samples (~300 ppm) were prepared by flowing zero-air
through the U-tube containing 37% HC1 cooled to 273 K in a cooling bath.
HOC1 samples were synthesized by adding sodium hypochlorite (NaOCl) dropwise to
anhydrous magnesium sulfate (MgSO4) dissolved in ultra-pure water. The product was
vacuum distilled to minimize impurities. C12O and HC1 impurities were determined to be
< 1% for each species by electron-impact time-of-flight mass spectrometry \Caldwell et
al, 1999]. Molecular chlorine impurities were reduced in the experiment by bubbling the
HOC1 solution rapidly prior to HOC1 sample preparation in each experiment.
Hypochlorous acid samples were prepared by bubbling zero-air through the HOC1
solution into a 35 L Telfon reaction chamber. Further dilution was achieved by adding
additional zero-air.
To calibrate HOC1, heterogeneous titration of HOC1 with HC1 was performed to produce
gaseous C12 [Abbott and Molina, 1992; Chu et al., 1993; Hanson and Ravishankara,
1993; Donaldson et al, 1997; Caldwell et al, 1999].
HOC1 + HC1 -> C12 + H20
(2)
First, HOC1 samples were flowed directly to the API-MS through the U-tube by-pass.
Next, the HOC1 was diverted through the U-tube, which contained a thin layer of dilute
HC1 and was submerged in a 245 K cooling bath. Throughout, MRM signal intensities
for select parent/daughter ion pairs were used to monitor HOC1 and C12. The C12 signals
were calibrated, with known concentrations of pure C12 gas. Calibration of the HOC1 was
performed by plotting the HOC1 signal intensity as a function of calibrated C12 produced
in reaction 2.
-81-
-------
HI. RESULTS and DISCUSSION
A. Effects ofH2O and acids on the CIR ions in air
When air is used as the CIR gas, H2O and acids effect the CIR ions. A comparison of the
observed ions in dry air, humid air, and air with ~300 ppm of HC1 vapor is presented in
Figure 2.
Dry air
o •
; 0 4D 50 60 70 BO 90 100 110
Humid air
tTJ
vf iy.
, Jl ft L
30 40 50 BO 70 80 90 100 110
Dry air with HCI
o- I* rtT^Ss^HCMon
i* °3 I nil! adducts
[f lJuLJL
30 40 50 60 70 80 SO 100 110
FIGURE 2: Ql scans of (a) dry air, (b) humid air (84% RH) and (c) dry air with HCI
vapor (~300 ppm).
Figure 2a shows the Ql spectra of dry air. The ions labeled in Figure 2a have been
assigned in.previous work [Kotasek, 1981]. In humid air, the signal intensity of O3" rises
while the CO3" signal intensity decreases as shown in Figure 2b. Previous studies have
shown water vapor produces OH" hi chemical ionization sources [Vogt., 1969; Spinks and
Woods, 1990]. Because OH has a larger electron affinity than O3 [Lias et al, 1988],
electron transfer is expected to produce an increase hi the O3" signal intensity in humid air
where OH" concentrations are high:
OH" + O3 -» O3" + OH (3)
The ozone anion can react with CO2 to produce CO3" [Fehsenfeld and Ferguson, 1974]:
O3" + CO2 -» O2 + CO3".
(4)
-82-
-------
However, the rate of reaction 4 is slower with > 2 waters of hydration, which can explain
the observed decrease in COi in humid air despite the observed CV increase [Fehsenfeld
andFerguson, 1974].
Chemical ionization reagent ions are severely decreased by the concentrated acid vapor in
Figure 2c. The signal intensities of O2~, O3~, and CO3" are all below the detection limit in
Figure 2c. However, the NCV signal intensity remains high. The relative gas phase
acidities of ion neutrals for the ions present in the corona discharge (i.e. HO2, HOH,
HNOs) compared to gas phase acidities of acids added to the corona discharge can be
used to predict if proton transfer will occur [Dzidic et al, 1974; Yamdagni and Kebarle,
1973; Lias et al., 1988]. For example, the gaseous acid strength of HC1 > HO2 [Lias et
al., 1988], so
HCI
HO2 + cr
(5)
which can explain the observed decrease in O2~ signal intensity hi Figure 2c.
B. HOCl titration using air as the ion source
Figure 3a shows the Ql spectra of HOCl using air as the CIR gas.
HOCl through bypass
84, 86
HOCI-O2)
87, 89, 91
(HOCI-CI)'
103, 105, 107
(HOCI-OCI)'
(b)
8.
O 5-
0-
0-
HOCI over HCI
70,72,74
30 40 50 60 70 80 90 100110
m/z, amu
30 40 SO 60 70 80 90 100 110
mlz, amu
FIGURE 3: Ql scans of (a) air containing ~700 ppb HOCl and (b) products of the
HOCl titration with HCI using air as the CIR gas.
Parent peaks (m/z - 52, 54) are small in this spectra. The strongest non-background
signals are assigned to HOCl-ion adducts labeled hi Figure 3a including (HOC1»O2)~.
Unlike the HCI sample in Figure 2c, HOCl deprotonation is not found because the acid
strengths of HCI and HO2 are greater than that of HOCl [Lias et al, 1988].
Hypochlorous acid was successfully titrated by flowing HOCl over the HCI filled U-tube.
Figure 3b shows the HOCl-ion adducts have nearly vanished during titration with HCI.
These signals have been replaced by intense C12 signals which appear at m/z - 70, 72, and
74.
-83-
-------
In MRM experiments, the calibrated 35>37C12 (72/35) titration product is used to determine
the HOC1 concentration, which is monitored using the ion pair 84/35 assigned to
(HO35C1»O2)~. HOC1 concentration plotted as a function of (HO35C1»O2)~ signal intensity
at 84/35 is linear for concentrations up to 500 ppb. At higher concentrations, non-linear
results are observed. These non-linear results are attributed to the effects of acid and
water vapor described in Section HI A.
C. HOCl titration using bromoform as the CIR gas
Figure 4 shows the heterogeneous titration of HOCl with HC1 using bromoform as the
CIR gas.
(a)
35Cl2»79Br)~
(149/35) titration product is used to determine the HOCl concentration, which is
monitored using the ion pair 131/79 assigned to (HO35Cl«79Br)~. This technique yields an
estimated detection limit of 0.9 ppb, lower than the 3 ppb detection limit measured using
air as the CIR gas. Additionally, non-linear results were not observed for concentrations
up to 960 ppb monitored in this experiment, which indicate that the effects of water vapor
and acids are minimized here.
-84-
-------
IV. SUMMARY AND ATMOSPHERIC IMPLICATIONS
In summary:
• Water vapor and acids affect the CIR ions in air.
• In air, HOC1 concentrations > 500 ppb could not be measured because of the effects
of water vapor and acids on the CIR ions.
• In bromoform, HOC1 concentrations > 960 ppb were measured, and complexities due
to water and acid vapor concentrations were not observed.
• Initial detection limits were 3 ppb in air and 0.9 ppb in bromoform.
The atmospheric implications of this work are that both methods can be used to measure
HOC1 in the laboratory, however, when air is the CIR gas, care must be taken to insure
CIR ions are not severely depleted by BbO and acid vapor during quantitative
measurements. Secondly, with an order of magnitude gain in sensitivity, it will be
possible to measure ambient HOC1 concentrations. It is possible to gain an order of
magnitude sensitivity by increasing the dwell time in the MRM experiments from 0.5 s to
50 s. Similar techniques can be developed to perform selective and sensitive
measurements of other halogen compound in air as well.
ACKNOWLEDGMENTS
We thank the Department of Energy and the National Science Foundation for support of
this research. T.E.C. thanks the Camille and Henry Dreyfus Foundation for financial
support as a Dreyfus Environmental Postdoctoral Research Fellow, and S.L. thanks the
Knut and Alice Wallenberg Foundation for financial support during a visit to this
laboratory. We are grateful to C. W. Spicer and R. T. Mclver. Jr. for helpful discussions
on API-MS and ion-molecule chemistry.
REFERENCES
Abbatt, J.P.D., and MJ. Molina, The heterogeneous reaction of HOC1 + HC1 -> C12 +
H^O on ice and nitric acid trihydrate: reaction probabilities and stratospheric
implications, Geophys. Res. Lett., 19, 461-464, 1992.
Abbatt, J.P.D., and J.B. Nowak, Heterogeneous interactions of HBr and HOC1 with cold
sulfuric acid solutions: implications for Arctic boundary layer bromine chemistry,
J. Phys. Chem., 101, 2131-2137, 1997.
Barrie, L.A., J.W. Bottenheim, R.C. Schnell, P.J. Crutzen, and R.A. Rasmussen, Ozone
destruction and photochemical reactions at polar sunrise in the lower Arctic
atmosphere, Nature, 334, 138-141, 1988.
-85-
-------
Caldwell, T.E., K.L. Foster, T. Benter, S. Longer, J.C. Hemminger, and B.J. Finlayson-
Pitts, Characterization of HOC1 using atmospheric pressure ionization mass
spectrometry, J. Phys. Chem. A, 103, 8231-8238, 1999.
Chu, L., and L.T. Chu, Heterogeneous reaction HOC1 + HBr —> BrCl + H2O on ice films,
J. Phys. Chem., 103, 691-699, 1999.
Chu, L.T., M.-T. Leu, and L.F. Keyser, Heterogeneous reactions of HOC1 + HC1 -» Cl2
+ H2O and C1ONO2 + HC1 —» Cl2 + HNC>3 on ice surfaces at polar stratospheric
conditions,/. Phys. Chem., 97, 12798-12804, 1993.
Donaldson, D.J., A.R. Ravishankara, and D.R. Hanson, Detailed study of HOC1 + HC1
-» C12 + H2O in sulfuric acid, J. Phys. Chem., 101, 4717-4725, 1997.
Dzidic, I., D.I. Carroll, R.N. Stillwell, and E.G. Horning, Gas phase reactions. Ionization
by proton transfer to superoxide anions, J. Am. Chem. Soc., 96, 5258-5259, 1974.
Fehsenfeld, F.C., and E.E. Ferguson, Laboratory studies of negative ion reaction with
atmospheric trace constituents,/. Chem. Phys., 61 (8), 3181-3193, 1974.
Foster, K.L., T.E. Caldwell, T. Benter, S. Langer, J.C. Hemminger, and B.J. Fihlayson-
Pitts, Techniques for quantifying gaseous HOC1 using atmospheric-pressure-
ionization mass spectrometry, accepted for publication in Phys. Chem. Chem.
Phys., 1999.
Hanson, D.R., and A.R. Ravishankara, Reaction of C1ONO2 with HC1 on NAT, NAD,
and frozen sulfuric acid and hydrolysis of N2C>5 and C1ONO2 on frozen sulfuric
acid, J. Geophys. Res., 98 (D12), 22,931 - 22,936, 1993.
Impey, G.A., P.B. Shepson, D.R. Hastie, L.A. Barrie, and K.G. Anlauf, Measurement
technique for the determination of photolyzable chlorine and bromine in the
atmosphere,/. Geophys. Res., 102 (D13), 15999-16004, 1997a.
Impey, G.A., P.B. Shepson, D.R. Hastie, L.A. Barrie, and K.G. Anlauf, Measurements of
photolyzable chlorine and bromine during the Polar Sunrise Experiment 1995, /.
Geophys. Res., 102 (D13), 16005-16010, 1997b.
Keene, W.C., J.R. Maben, A.A.P. Pszenny, and J.N. Galloway, Measurement technique
for inorganic chlorine gases in the marine boundary layer, Environ. Sci. Technol.,
27, 866-874,1993.
Kotasek, V., Negative ion/molecule reactions in a negative corona discharge, Masters
thesis, University of Toronto, Toronto, 1981.
Kwok, E.G., R. Atkinson, and J. Arey, Observation of hydroxycarbonyls from the OH
radical-initiated reaction of isoprene, Environ. Sci. Technol., 29, 2469-2469,
1995.
Kwok, E.G., R. Atkinson, and J. Arey, Isomerization of beta-hydroxyalkoxy radicals
formed from the OH radical-initiated reactions of C4-Cg alkenes, Envoron. Sci.
Technol., 30, 1048-1052,1996a.
Kwok, E.S., J. Arey, and R. Atkinson, Alkoxy radical isomerization in the OH radical-
initiated reactions of C4-Cg n-alkanes,/. Phys. Chem., 100, 214-219, 1996b.
Lias, S.G., J.E. Bartmess, J.F. Liebman, J.L. Holmes, R.D. Levin, and W.G. Mallard,
Gas-phase ion and neutral thermochemistry, /. Phys. Chem. Ref. Data, 17
(Supplement No. 1), 652-861,1988.
-86-
-------
McConnell, J.C., G.S. Henderson, L. Barrie, J. Bottenheim, H. Niki, C.H. Langford, and
E.M.J. Templeton, Photochemical bromine production implicated in Arctic
boundary-layer ozone depletion, Nature, 355, 150-152, 1992.
Pszenny, A.A.P., W.C. Keene, DJ. Jacob, S. Fan, J.R. Maben, M.P. Zetwo, M. Springer-
Young, and J.N. Galloway, Evidence of inorganic chlorine gases other than
hydrgen chloride in marine surface air, Geophys. Res. Lett., 20 (8), 699-702,
1993.
Sanders, R., and Crutzen, P.J., Model study indicating halogen activation and ozone
destruction in polluted air masses transported to the sea. J. Geophys. Res., 101
(D4), 9121-938, 1996.
Spicer, C.W., E.G. Chapman, BJ. Finalyson-Pitts, R.A. Plastrigge, J.M. Hubbe, and J.D.
Fast, Unexpectedly high concentrations of molecular chlorine in coastal air,
Nature, 394, 353-356, 1998.
Spicer, C.W., D.V. Kenny, G.F. Ward, and I.H. Billick, Transformations, lifetimes, and
sources of NO2, HONO, and HNC>3 in indoor environments, J. Air & Waste
Manage. Assoc., 43, 1479-1485, 1993.
Spinks, J.W.T., and R.J. Woods, An Introduction to Radiation Chemistry, Wiley, New
York, 1990.
Vogt, R., P.J. Crutzen, and R. Sander, A mechanism for halogen release from sea-salt
aerosol in the remote marine boundary layer, Nature, 383, 327-330, 1996.
Vogt., D., Uber die energieabhangigkeit und den mechanismus von reaktionen bei
stossen langsamer negatver ionen auf molekule, Int. J. Mass Spectrom. Ion Phys.,
3, 81-90, 1969.
Yamdagni, R., and P. Kebarle, J. Amer. Chem. Soc., 95, 4050, 1973.
-87-
-------
New Product and Aerosol Studies On The Photo-Oxidation Of Dimethylsulfide
C. Arsene, I. Barnes and K.H. Becker
Physikalische Chemie /Fachbereich 9, Bergische Universitdt-GH Wuppertal
Gaufistrafie 20, 42097 Wuppertal, Germany
Email: barnes(a)physchem.uni-wuppertal.de
INTRODUCTION
Dimethylsulfide (DMS) is the major natural source of sulfur to the atmosphere with a source
strength of between 12-54 Tg S yr"1 (Andreae et al., 1994). The chemistry of DMS has been
postulated to play a pivotal role in regulating the Earth's radiation budget. Although the
atmospheric chemistry of DMS has been the subject of intense research an in-depth
understanding of many facets of its chemistry still remain elusive.
Laboratory experiments show, that 862, HCHO, methanesulfonic acid (MSA: GH^SOsH), aero-
sols and recently also dimethylsulfoxide (DMSO: CHsSOCHs) and methane sulfmic acid(MSIA:
CH3S(O)OH), comprise the most important end products (Patroescu et al., 1999; Serensen et al.,
1994). Formation of carbonyl sulfide (OCS) has also been observed could make a significant
contribution to the COS-budget (Barnes et al., 1996; Patroescu et al., 1999). The products of the
OH-radical initiated oxidation of DMS have been investigated for the first time as a function of
temperature and O2 partial pressure.
EXPERIMENTAL
The experiments were performed in a temperature regulated 1080 litre volume quartz glass
reactor. The 254 nm photolysis of H2C>2 was used for the production of OH radicals. Mixtures of
DMS/H2O2/(O2 + N2) were irradiated over a period of 20-30 min.All experiments were
performed at 1000 mbar (O2 + N2) total pressure at O2 partial pressures of 20, 200 and 500 mbar
and temperatures of 284, 295 and 306 K.
Reactants and products were monitored using long path in situ FTIR spectroscopy. Methane
sulfinic acid, methane sulfonic acid and sulphate were measured using ion chromatography.
Samples of 10-40 1 air were drawn through a U-tube immersed in an ethanol/liquid N2 slush bath.
RESULTS
Yield/time profiles of the products formed in the OH + DMS reaction as a function of
temperature and O2 partial pressure are shown in Figure 2: a) SO2, b) DMSO, c) DMSO2, d)
MSA, e) OCS, f) MTF. The yields have not been corrected for secondary loss processes:
reactions with OH, photolysis and wall loss.
-88-
-------
Table 1: Corrected yields for the formation of DMSO in the reaction of OH with DMS in
1000 mbar of synthetic air as a function of temperature. The yields are compared
to the fractions of the reaction occurring via the addition pathway and
abstraction pathways calculated using the results of Hynes et al. [1986] for the
specific reaction conditions.
Temperature
[K]
284
295
306
Corrected DMSO Yields
[% molar yield, ± 2a]
46.3 ± 5.0
34.8 ± 7.6
24.4 ± 2.8
Contribution of addition
pathway
52
33
17
Table 3: Corrected yields for the formation of SO2 in the reaction of OH with DMS in
1000 mbar of synthetic air as a function of temperature. The yields are compared
to the fractions of the reaction occurring via the addition pathway and abstraction
pathways calculated using the results of Hynes et al. [1986] for the specific
reaction conditions.
Temperature
[K]
284
295
306
Corrected SO2
Yields
[% molar yield, ± 2a]
84.3 ± 6.5
95.0 ± 3.8
99.0 ±6.5
Contribution of
addition pathway
I%]
52
33
17
Contribution of
abstr. pathway
I%]
48
67
83
-89-
-------
(a)S02
(b) DMSO
(c) DMSO2
i-6-
r-284K(20mbarO2)
I-295 K (20 mbar O2)
-«-306K(20mbarO2)
-A- 284 K (200 mbar O2)
-«- 295 K (200 mbar O2)
-•- 306 K (200 mbar O2)
-nfc-284 K (500 mbar O2)
-*- 295 K (500 mbar O2)
-*- 306 K (500 mbar O2)
400
800 .
1200 1600
Time(s)
(d)MSA
-*-284K(20mbarO2)
-•-295 K (20 mbar O2)
-»-306K(20mbarO2)
-*- 284 K (200 mbar O2)
-m- 295 K (200 mbar O2)
-•-306 K (200 mbar O2)
-A-284 K (500 mbar O2)
- 295 KL (500 mbar O2)
-306 K (500 mbar O2)
400 800 1200 1600 0 400 800 1200 1600
Time(s) Time(s)
(f)MTF
0 400 800 1200 1600
Time(s)
400 800 1200 1600
Time (s)
Figures 3 and 4 show plots of the corrected concentrations for SOa and DMSO against the
amount of DMS consumed. The formation yields are tabulated in Tables 1 and 2 where they are
compared against the contribution of the addition and abstraction channels.
-90-
-------
V
A284K
• Z95K
S306K
6 8 10
delta ID M S) x 10'"(mola:ijlecmJ)
!!'°J
y =0.942x - 0.395
R! = 0.993
8 10 12 14
delta [DMS]x 10"13(moleculean"3)
j|G «•
I?, »•
A- 8 •
X «:
y =0.991x-0.174
Rl = 0.987
0 2 4 6 8 10 12 14
del(a[DMSJx 10'" (moteculecm0)
DISCUSSION
The primary steps in the OH-radical initiated oxidation of DMS can be represented as follows:
OH + CH3SCH3 -»CH3SCH2 + H2O (Abstraction)
-> CH3S(OH)CH2 (Addition)
CH3SCH2 + O2 -> Products (SO2, HCHO, MSA)
CH3S(OH)CH3 + O2 -» DMSO + other products (MSIA?)
• The observed variations in SO2 and DMSO yields with experimental conditions are broadly in
line with the above mechanism.
• The high yield of DMSO suggests that reaction of the OH.DMS adduct with O2 results
mainly in formation of DMSO.
• The DMSO yields are much higher than the 0.5 yield reported by Turnipseed et al. at low
temperature and pressure.
• Earlier detection of OCS has been confirmed.
• Yields of gas phase MSA are low.
• High yields of methane sulfinic acid (MSIA) have been observed. Yields are lower limits
because of the method of collection and oxidation of MSIA to MSA.
• A large fraction of DMSO is being oxidised to MSIA in line with a recent absolute
measurement.
• The large overall formation yields of SO2 support that SO2 is formed in both the abstraction
and addition channels.
-91-
-------
• Further oxidation of MSI A must also result in formation of SCh under the present
experimental conditions.
In the absence of heterogeneous processes, the gas phase oxidation of DMS in the atmosphere
under conditions of low NO will result mainly in the production of SC>2.
x
The fate of DMSO and MSIA, other than gas phase oxidation, is likely to be a major factor in
determining the efficiency of the conversion of DMS to SO2-
LITERATURE
Andreae, T. W., Andreae, M. O. and Schebeska, G. (1994) Biogenic sulfur emissions and aerosols over the tropical
South Atlantic 1. Dimethylsulfide in seawater and in the atmospheric boundary layer. J. Geophys. Res. 99,
22,819-22,829.
Barnes, I., Becker, K. H. and Patroescu, I. (1996) FT-IR product study of the OH-initiated oxidation of dimethyl
sulphide: observation of carbonyl sulphide and dimethyl sulphoxide, Atmos. Environ. 30A, 1805-1814.
Hynes, A. J., Stoker, R. B., Pounds, A. J., McKay, T. Bradshaw, J. D., Nicovjch, J. M. and Wine P. H. (1995) A
mechanistic study of the reaction of OH with dimethyl-d6 sulfide. Direct observation of adduct formation and the
kinetics of the adduct reaction with O2. J. Phys. Chem. 99, 16,967-16,975.
Hynes, A. J., Wine, P. H. and Semmes, D. H. (1986) Kinetics and mechanism of OH reactions with organic sulfides.
J. Phys. Chem. 90,4148.
Patroescu, I. V., Barnes, I., Becker, K. H. and Mihalopoulos, N. (1999) FT-IR product study of the OH-initiated
oxidation of DMS in the presence of NOX, Atmos. Environ. 33, 25-35.
Ravishankara, A. R., Rudich, Y., Talukdar, R. and Barone, S. B. (1997) Oxidation of atmospheric reduced sulphur
compounds: Perspective from laboratories, Phil. Trans. R. Soc. Land., 332, 171-182.
Sffensen, S. Falbe-Hansen, H., Mangoni, M., Hjorth, J. and Jensen, N. R. (1996) Observations of DMSO and
CH3S(O)OH from the gas phase reaction between DMS and OH, J. Atmos. Chem. 24, 299-315.
Tumipseed, A. A., Barone, S. B. and Ravishankara, A. R. (1996) Reaction of OH with dimethyl sulfide. 2. Products
and mechanisms, J. Phys. Chem. 100,14703-14713.
Urbanski, S. P., Stickel, R. E. and Wine, P. H. (1998) Mechanistic and kinetic study of the gas-phase reaction of
hydroxyl radical with dimethyl sulfoxide, J. Phys. Chem. 102, 10,522-10,529.
Acknowledgement
Financial support for this work by the European Commission and the BMBF within the AFS
project is gratefully acknowledged.
-92-
_
-------
Total Non-Methane Organic Carbon: Measurements of Total and Speciated
Hydrocarbons at Azusa, California
Suzanne E. Paulson, Richard Meller1, and Franz Kramp2
Atmospheric Sciences Department, University of California at Los Angeles, Los Angeles, CA
90095-1565, USA
'Max-Planck-Institut foer Chemie, Abteilung Luftchemie, D-55128, Mainz, Germany
2Forschungszentrum JYlich, Institut fYr Chemie und Dynamik der GeosphSre (ICG-2), P.O.
Box 1913, 52425 JYlich, Germany
Introduction
The formation of ozone and other oxidants in urban and rural areas remains a persistent
problem that affects both the public health and economic vigor of many areas around the globe.
Oxidant formation results from photochemistry of organic compounds in the presence of
nitrogen oxides. The organic component begins mainly as non-methane hydrocarbons (NMHC,
containing only carbon and hydrogen) from biogenic and anthropogenic sources. These
compounds are progressively oxidized to CO and CO2 over periods of hours to weeks. The
variety of primary hydrocarbons and their oxidation products is large. While separate techniques
exist to measure groups of compounds (e.g., Ca-Cg hydrocarbons with some oxygenates, or
alcohols, or formaldehyde or organic nitrates, etc.) no techniques to assess the total loading of
non-methane organic carbon (TNMOC) have been widely applied in the atmosphere. The goal
of this work is to determine the relationship between the total non-methane organic compounds
and the sum of the speciated volatile organic compounds (VOC's) measured by standard
techniques. The primary scientific motivations are to define the airborne quantity of reactive
organic carbon, find how close this quantity is to the standard measurements of VOC's, and
address the question of what happens to the multifunctional products of the photo-oxidation
reactions of emitted hydrocarbons; do they stay in the gas phase or are they removed to the
aerosol phase or surfaces?
There are several reasons to expect the TNMOC measurement to result in a number that
is larger than the sum of the routinely measured VOC's. Loss of oxygenated and semi-volatile
compounds in sampling and on columns is well known [1,2], although the extent of the losses is
very difficult to quantify. Modeling results indicate that while the initial oxidation step of a
hydrocarbon may take place within hours, if our understanding of the relevant processes is
correct, complete removal of the partially oxidized fragments should take weeks [3]. Finally, in
controlled smog chamber photooxidations, a standard gas chromatograph with flame ionization
detection (GC/FID) measurement can account for much less than 100% of the reacted parent
hydrocarbon [4]
Roberts and co-workers have developed a column back-flush approach to make a
TNMOC measurement in ambient air [5]. This approach has the advantage of achieving quite
low detection limits (~1 ppbv) but may lose semi-volatile compounds to the GC column.
Roberts et al. (1998) compared their TNMOC concentrations with speciated VOC measurements
collected by other groups with instruments nearby to calculate the TNMOC/- speciated VOC's
-93-
-------
ratio. Since the calibrations, sampling frequencies and averaging times were different, a fair
degree of scatter is anticipated, and a systematic error is possible. Measuring in relatively clean
air in Boulder, Colorado, and Nova Scotia, Canada, they found differences between TNMOC
(which they refer to as "Cy") and the sum of hydrocarbons and carbonyls that were small in
absolute terms (averaging 3.8 ± 7.9 ppbC) but were significant in percentage terms. Ratios of
TNMOC/- speciated hydrocarbons + carbonyls were 1.36 ± 1, 1.16 ± 0.7. and 1.33 ± 0.4 for
samples from air with < 20, 20 2 03 2 50, and > 50 ppbv Os respectively. These values are
comparable to the ratio of TNMOC/- Speciated VOC's we measured in Azusa, California during
September and October of 1997 of 1.3 ± 0.3.
Here we describe development of a new instrument to measure the airborne total non-
methane organic carbon concentration (TNMOC), and the ratio of this value to the sum (• ) of
speciated VOC's measured by standard gas chromatography (GC-FID). The approach is to make
an in situ measurement that minimizes sample contact, cryo-trapping whole air samples with
minimal trapping of COa, CO and CELt. Samples are processed by an oxidation catalyst to
generate COa that is measured as TNMOC. Alternatively, a standard speciated VOC's
measurement is made with the same instrument. The TNMOC instrument described here has
several improvements compared to previous approaches. Samples introduced into our
instrument are exposed to only an inlet tube, valve and short length of transfer tubing, and then
immediately analyzed. Water management is not needed. Speciated VOC's are measured with
the same instrument, avoiding problems created by comparing different sample inlets and
calibrations.
Instrument Description
The technique to make the TNMOC measurement centers on a cryogenic separation of
reactive carbon from CO, CO2 and CH^. A schematic for the process is shown in Figure 1. In
step 1, the first trap (trap I) is cooled to between -55 and -80 uC. The sample passes through the
sampling tube and a heated valve before passing through trap I. The target TNMOC condenses,
while CEU, CO, and most of the CO2 pass through. Once the sampling (cryo-trapping) stage is
finished, trap I is purged with helium (40 s at 16 mL/min) to remove sample gas from the tubing
and air space within the trap. In step 2, trap I is rapidly heated; and a He/O2 mix sweeps the
desorbed TNMOC into an oxidation catalyst where the organics are converted to CO2 (200 s at
16 mL/min). An advantage of this method is that decomposition or reaction of labile compounds
during trapping and heating has no effect on the result, because the trapped organics are
immediately oxidized to CO2- The CO2 is focused in trap II, which is immersed in /N2, to
concentrate the (~50 mL) volume necessary to thoroughly desorb the contents of trap I into a
small plug for the quantification step. Finally the focusing trap is heated and the TNMOC,
which has been converted to CO2, flows into column I and is catalytically converted at the end of
column 1 to CHLj and quantified with an FID.
The TNMOC instrument is built around a dual FID GC (Hewlett Packard 5890-11). Inlet
valves (Valco) are attached to an aluminum block maintained at 50-60 uC. Transfer lines, trap
tubing and fittings are constructed from deactivated fused silica lined stainless steel (Restek) and
are maintained above ambient temperatures with heating tape. Temperature controllers (Omega)
maintain heating and cooling of trap I, the methanation and oxidation catalysts and the valve
block. Trap II is alternately cooled and heated by immersing it in liquid nitrogen and hot water.
-94-
-------
Both traps I and II are filled with silanized glass beads and are constructed of 1/4" and 1/8" outer
diameter (OD) tubing, respectively. Trap I is cooled from the center with pulses of /N2 vapor, so
that there is a significant temperature gradient along the length of the tube. As a result, the trap's
temperature, which is recorded by a thermocouple attached close to the center of the tube, is a
relative value. Each time this trap is replaced, the trapping of CO2 as a function of temperature
is determined and a new optimal operational temperature is chosen, typically falling in the range
-55 to -80 uC. The oxidation and reduction catalysts are constructed of 1/8" OD stainless tubing
packed with unsupported Pd (Aldrich), and Ru (Ann Arbor Specialty Chemicals) granules
maintained at 700 and 300 uC respectively. Mass flow controllers (Unit Instruments) provide
stable flows of He, He/O, and sample flow.
One of the most interesting aspects of the TNMOC measurement is its comparison to the
sum of the speciated VOC's as they are commonly measured. The ideal way to make this
comparison is to acquire two identical samples simultaneously, and process one sample in
TNMOC mode and the other in speciated mode, with separate GC columns. Prior to the field
measurements described below, we installed a second GC column and FID. After the field
campaign, an identical sampler in parallel was added. The columns used for both TNMOC and
Speciated measurements are 60m x 0.32mm ID x lum DB-1 (J&W). This column was chosen
because the stationary phase (100% methyl polysiloxane) is among the most widely applied in
field measurements of VOC's [1, 6-8]. Ideally, 100% of the NMOC in the air sample stays in
trap I, and CO, CO2, and CBU pass through. In practice, CO and CH4 (boiling points (Tb) of -
191.5 and -164 QC, respectively) pass through, while CO2, with an ambient concentration larger
than the TNMOC by a factor of 100-40,000 and a sublimation point of-78.5 uC, traps to a minor
degree, while the C2-C4 hydrocarbons are not trapped. The TNMOC detection limit is
determined not by the absolute amount of atmospheric CO2 that traps but by its variability. We
can achieve trapping of CO2 equivalent to about 50 ± 10 ppbv in a 500 mL sample. The amount
of CO2 trapped is traded off, however, with the trapping of C^Cs hydrocarbons. In the field, we
chose conditions that trapped 300 ± 40 ppb of CO2, thus our detection limit (~3 x the CO2 noise
level) was about 120 ppbC. The amount of trapped CO2 is measured continually, and includes
CO2 background caused by the carrier gasses.
-95-
-------
Stepl:
Whole A
Sample
Step 2:
He Can-
Cold Trap: trap reactive caibon,alo* non-reactive carbon to pass through.
Coolant Out Reactive Coolant Coolant Out 4
^^^^^^^^^
1 K , -ta ^Vl Wf'~
IAI 1
"Cool ant Gas inlet .. '
Vent
Heat Trap (resbli¥EM,desorb reactiw carbon, convert to CQ^lhen to CH* to quantify.
=^
Data Acquisition
1 1 1
Reacliw Coolant
/ -\
u
Flame lonization
J Detector Ru Catalyst
(Reduction!^
-*-He + CH^ -*-(." : -,-:. >l3f-He •
^s He +
"^ Cartjon
Pd Catalyst
(Oxidation] 1 ' QA
Figure 1. Process schematic of the approach for the TNMOC measurement.
The TNMOC instrument can be operated in three modes: 1) TNMOC: The sample is
oxidized over a Pd-catalyst to CO2, run on column 1 and analyzed with FID after conversion to
methane over Ru. 2) Speciation, column 1: No oxidation, column 1, Ru methanizer and FID.
This mode was necessary to establish the amount of CO2 captured in trap I to order to correct
TNMOC measurements. This mode is not ideal for speciated VOC's due to losses of heavy
hydrocarbons in the methanizer. 3) Speciated VOC's: No oxidation, column 2 analyzed with
FID (no methanizer). This is the standard speciated hydrocarbon analysis. TNMOC (mode 1)
and 2. Speciated VOC's (mode 3) chromatograms of ambient air acquired in Azusa, CA are
shown in Figure 2.
Field Measurements at Azusa, CA during September and October 1997
Field measurements were carried out at the SCAQMD's Azusa Station from September 4
to October 10, 1997, as part of the Southern California Oxidants Study (SCOS-97). The Azusa
site has been a PAMS site since 1994, and is located at the foot of the San Gabriel Mountains
downwind of Los Angeles near the intersection of the 210 and 605 freeways in a mixed
residential and light industrial area. The site experiences significant local sources of VOC's, and
typically experiences some of the highest 03 levels in the heavily populated portions of the Los
Angeles air basin. The summer of 1997 had, however, the cleanest air on record.
For this campaign, the sampling line was constructed from 9.7 m of 1/8" Silcosteel tubing
(Restek) with an inlet about 2.5 m above the roof. The three meters inside the heavily air-
conditioned station were wrapped with heating tape and maintained at about 50 °C to prevent
condensation of water in the sampling line. Flow through the sampling line was maintained at 1
L/m with a diaphragm pump, and the sample was drawn through a tee at 20 or 50 mL/min. The
residence time in the sampling line was about 4 s. The identification of the peaks discussed
below is based on reference chromatograms constructed using samples from an EPA PAMS
59 compound retention time standard cylinder provided by the SCAQMD. These
-96-
-------
chromatograms were taken under dry and CO2-free conditions, so the Azusa retention times were
adjusted by using the most clearly identifiable peaks as markers. Retention times were checked
against measurements made by the co-located EPA GCMS instrument, which uses the same type
of column phase. The identification of methanol, acetaldehyde, ethanol, cc-pinene, limonene,
and several other species is based on the EPA GC/MS-instrument. All identities, but particularly
the oxygenates and biogenics, should be considered tentative.
mV
90 ~
80-
70~
60~
50~
40~
30-
20-
10-
0~
10-
'c
.Wk.
H/t , " ' • "i U . ••,, uf
yj
LLJ
1.1
A
' 1 ' I ' 1 ' 1 ' 1 • 1 ' 1 ' 1 '
(246 8 10 12 14 16
Li
\^LMJL^^
i , , . , , | , , , j
18 20 22 24 26 28
Vlinutes
Figure 2. Sample chromatograms acquired 9/19/97 10:30 -11:00 AM at the Azusa station .
Upper chromatogram, TNMOC, 550 ppbC. The lower chromatogram, Speciated VOC's,
460 ppbC has been multiplied by 6.5 and shifted down with respect to the vertical axis by 10
mv. Each sample was 500 mL.
Ten calibrations with a 1.08 ppmv n-hexane standard were performed at Azusa, resulting
in uncertainty ± 2 and 3 % for the TNMOC and Speciated VOC's channels, respectively (1 a).
The variability in the trapped ambient COa was ± 40 ppbC, or ± 8 % (1 a). This variability no
doubt contributes to the scatter in TNMOC measurements, although it is a small fraction of the
observed scatter in the TNMOC measurements, the majority of which is presumably due to
airmass variability. TNMOC and speciated VOC's measurements were made sequentially
during the Azusa campaign, using two different protocols: (1) One sample/hour, alternating
TNMOC and Speciated VOC measurements, 800 mL at 20 mL/min, first 19 of 84 samples, and
(2) Two samples/hour, alternating TNMOC and Speciated VOC's, 500 mL at 50 mL/min,
remaining 65 samples. The sample frequency is limited by the 35-40 minutes needed for
chromatography of the Speciated VOC's. The TNMOC measurement (10 minute
-97-
-------
chromatogram) was made first, because this allowed the two samples to be collected as close
together as possible (15-20 minutes).
Measurements and Comparisons ofTNMOC and Speciated VOC's
Measurements of TNMOC are plotted vs. time of day in Figure 3. TNMOC averaged
530 ± 200 ppb (1 a) with a weakly decreasing trend over the course of the day, and high
variability in the morning hours. The • Speciated VOC's data has a similar trend with an average
concentration of 410 ± 150 ppbC (1 a).
Hour of Day (PDT)
Figure 3. TNMOC (ppmC) vs. Hour of day. The line indicates a linear correlation with the
equation y = -0.024x +0.8, and has an R value of 0.51. The average TNMOC concentration
was 520 ppbC.
One of the primary goals of the TNMOC instrument is to determine the amount of
organic carbon in ambient air that is not detected in conventional GC analysis. The ratio of
TNMOC/Speciated VOC's for 83 pairs of measurements made between Sept. 4 and Oct. 10 are
conventional VOC measurements using GC analysis underestimates the amount ofTNMOC.
Contribution of Light Oxygenates to the Excess TNMOC
A candidate for the source of the excess TNMOC compared to the • Speciated VOC's
observed in Azusa is the oxygenated VOC's that are detectable with GC-FID but have reduced
responses relative to hydrocarbons due to their heteroatom content [Jorgensen, 1990 #335], and
thus contribute less to the • Speciated VOC's than to TNMOC. Most oxygenates were not
quantifiable in our speciated chromatograms, but we could estimate the concentrations of
methanol, ethanol and acetaldehyde. Methanol averaged 6 ppbC or about 1.3% of the total
carbon, and acetaldehyde and ethanol each contributed about 2% to the TNMOC. The average
acetaldehyde concentration was 8.4 ppbC, in agreement with 6 ppbv measured at Azusa in 1993
[9]. Ethanol averaged 11 ppbC. Roughly half of the carbon in the methanol, ethanol and
acetaldehyde is already accounted for in the I, Speciated VOC's measurement (FID responses are
equivalent to 0.77, 1.65 and 1.02, respectively, [10]). These three compounds sum to about 5%
of the total VOC loading, 2-3% of which is already included in the • Speciated VOC's. Thus
-98-
-------
these three common oxygenates can account for 10% or less of the difference between TNMOC
and the • Speciated VOC's.
n R •
I
-•' 'I'M' ,' .
p\. ' T j>\pt}\ ,Vi I
11 ! ' "' i i r "' , ''
,,, .M.
0 4 8 12 16 20 24
Hour of Day (PDT)
Figure 4. Ratio of TNMOC to • Speciated VOC's as a function of time of day. The average
TNMOC/- Speciated VOC's was 1.29 ± 0.27
Hydrocarbon Age/Photochemical Processing
Another potential source of the 30% excess TNMOC over speciated are the multi-
functional species similar to those generated in the isoprene and m-xylene chamber experiments
described above. Azusa normally experiences quite polluted air during the "smog season" (July
to October), but 1997 was the cleanest year on record, and our data set contains 03 levels 95% of
which were at or below 70 ppbv and 60% at or below 40 ppb. To estimate the extent of
photochemical processing of hydrocarbons arriving at the Azusa station, ratios for several pairs
of aromatic hydrocarbons from the speciated VOC chromatograms were calculated [11, 12].
Ignoring the effect of dilution, the ratio of two hydrocarbons should evolve according to the
following equation:
r . -i r * ^
(kA -kB)[OH]t (E 1)
where [X] and [X]0 are the species concentrations at the time of the measurement and hi the
source, respectively, kx is the rate constant for reaction of X with OH radicals and [OH] is the
average OH concentration during the time period t. Here we focus on the m&p-
xylene/ethylbenzene ratio, since these compounds are highly correlated and the peaks could be
reliably quantified, m- And p-xylene are not separated by a DB-1 column, thus their
concentration is combined. The primary sources of these three aromatics are unburned or
evaporated fuel [13]. Rate constants for reaction ethylbenzene and m&p-xylene with OH of
0.71, 2.2 and 1.4 x 10'11 cm3 molecule-1 s-1, respectively were assumed [14, 15].
At night our m&p-xylene/ethylbenzene ratio had a value of 3.8 - 3.9 (Figure 5) in
excellent agreement with the value reported for Los Angeles gasoline hi 1975 [13] and ambient
air in 1987 [12]. The magnitude of the decrease during the day is presumably an indicator of the
level of photochemical activity. Indeed, the inverse correlation between the m&p-xylene/
ethylbenzene ratio and 03 has an R value of 0.77. From the decrease of the ratio La (m&p-xylene
-99-
-------
/ethylbenzene) from 1.35 at night to between 0.8 and 1.0 in the afternoon, an average exposure to
OH can be calculated from (El), and corresponds to 1.8 hours of exposure 0.2 ppt of OH.
With this estimate of hydrocarbon processing, a calculation of the expected degree of
increased oxygenation/nitration of the organics can be made relying on the following
assumptions: 1) the average speciated mixture of 100 hydrocarbons from the EPA for 29 cities at
6-9 AM survey [16], 2) rate constants for each hydrocarbon reacting with OH, and for alkenes
with ozone, 3) exposure to 0.2 ppt OH and 50 ppbv O3 for 1.8 hours, and 4) equation (E2):
(E2)
Under these conditions, an average of 27% of the hydrocarbons react once with either OH or
ozone. Since the organics have an average of 7 carbons [16], and can be expected to add ~1.5
functional groups (primarily alcohol, carbonyl, or nitrate) per reaction, the total mix might
increase its heteroatom content relative to the carbon by about 6%. The effect of this modest
increase on the TNMOC/- Speciated VOC's ratio cannot be calculated precisely since the effect
of oxygenation/nitration can be to either reduce the FID response or to cause the compound to be
lost or broadened in the column so as not to be quantifiable. The latter reduces the sum of
speciated VOC's more than the former. It is clear, however, that given the low level of
photochemical processing in this data set, the T/S ratio could be expected to have relatively little
dependence on the time of day (Figure 3), ozone level, or the ratio of m&p-xylene/ethylbenzene
(not shown). Indeed, no trends are observed; each relationship has a correlation coefficient of
less than 0.05.
0 8 •
0 4
"l 'I 'l h I H *M| , .,' ,,'!
l^'.Vni1'
",,(.« \V
V
0.
0 6 12 18 24
Hour of day
Figure 5. ln(m&p-xylene / ethylbenzene) versus hour of day.
Conclusions
The results from the field campaign in Azusa, CA are surprising, showing that even in air
with minimal photochemical processing, the ratio of the TNMOC to the sum of speciated VOC's
is 1.3 ± 0.3. The source of this difference, as well as its spatial and temporal variability remains
to be determined. About 10% of it is from the light oxygenates methanol, ethanol and
acetaldehyde, which are only partially measured by an FID. A contribution from directly emitted
semi-volatile heavy hydrocarbons is possible [1, 17, 18]. The difference in the Azusa data set is
probably not due to large quantities of heteroatom containing oxidation products reaching the
-100-
-------
station, since the air during most of the late summer of 1997 was unusually clean; the maximum
level of oxygenation/nitration due to photochemistry was less than 10%. The variability is due to
a combination of short term variability in both local sources and in air mass history and
highlights the requirement that TNMOC and sum of speciated VOC's to be made in parallel,
preferably by the same instrument. Clearly, these initial results are intriguing and further
investigation is warranted.
Acknowledgments
The California Air Resources Board is gratefully acknowledged for partial support of this
work. The statements and conclusions in this report are those of the University and not
necessarily those of the California Air Resources Board (ARB). The mention of commercial
products, their source, or their use in connection with material reported herein is not to be
construed as actual or implied endorsement of such products. The authors would like to thank
the many individuals who contributed to this project: at UCLA, Paul Northrop, Ping Liu and
Hugh Liu, Andy Ho, and Myeong Chung, SCAQMD staff Steve Barbosa, Phil O'Bell and Rudy
Eden, ARB contract managers Dr. Randy Pasek and Dr. Eileen McCauley, from EPA Dr. Charles
Lewis, and UCR Prof. Janet Arey and Prof. Roger Atkinson.
References
1. Zielinska, B., J.C. Sagebiel, G. Harshfield, A.W. Gertler, and W.R. Pierson, Atmos. Environ.,
1996, 30, 2269-2286.
2. Cao, X.-L. and C.N. Hewitt, /. Chrom. A, 1994, 668, 368-74.
3. Madronich, S. and J.G. Calvert. in Proc. Int'l. Conf. Global and Regional Envir. Atmos. Chem.
1989. Beijing, China:
4. Paulson, S.E., R.C. Flagan, and J.H. Seinfeld, Int'l. J. Chem. Kinet., 1992, 24, 79-102.
5. Roberts, J.M., S.B. Bertman, T. Jobson, H. Niki, and R. Tanner, J. Geophys. Res. Atmos., 1998,
103, 13581-13592.
6. Martin, R.S., H. Westberg, E. Allwine, L. Ashman, J.C. Fanner, and B. Lamb, J. Atmos. Chem.,
1991, 13, 1-32.
7. Blake, D.R., T.W. Smith, T.Y. Chen, W.J. Whipple, and F.S. Rowland, J. Geophys. Res., 1994, 99,
1699-1719.
8. Guenther, A., P. Zimmerman, L. Klinger, J. Greenberg, C. Ennis, H. Westberg, G. Allwine, and C.
Geron,/. Geophys. Res., 1996, 101, 1345-1359.
9. Grosjean, E., D. Grosjean, M.P. Fraser, and G.R.Cass, Envir. Sci. TechnoL, 1996, 30, 2687-2703.
10. Paulson, S.E., R. Meller, and M. Chung, (1999). Report No. 95-335, California Air Resources
Board
11. Roberts, J.M., F.C. Fehsenfeld, S.C. Lui, MJ. Bellinger, C. Hahn, D.L. Albritton, and R.E.
Sievers, Atmos. Environ., 1984, 18,2421 - 2432.
12. Fujita, E.M., B.E. Croes, C.L. Bennett, D.R. Lawson, F.W. Lurmann, and H.H. Main, J. Air.
Waste. Manage. Assoc., 1992,42, 264-276.
13. Nelson, P.P. and S.M. Quigley, Atmos. Environ., 1983, 17, 659-662.
-101-
-------
14. Atkinson, R., J. Phys. Chem. Ref. Data, 1997, 26, 215-290.
15. Kramp, F. and S.E. Paulson, J. Phys. Chem., 1998, 102, 2685-2690.
16. Jeffries, H.E., in Composition, Chemistry and Climate of the Atmosphere, H.B. Singh, Editor.
1995, Van Nostrom Reinhold: New York.
17. Eraser, M.P., G.R. Cass, B.R.T. Simonet, and R.A. Rasmussen, Env. Sci. Tech., 1997, 31,2356-67.
18. Fraser, M.P., G.R. Cass, B.R.T. Simonet, and R.A. Rasmussen, Env. Sci. Tech., 1998, 32, 1760-
1770.
-102-
-------
Use of ab initio Quantum Mechanics to Estimate Rate Constants
David M. Golden1, Juan Senosiain2 and Charles Musgrave2
Stanford University, Stanford, CA
1. Department of Mechanical Engineering
and Molecular Physics Laboratory, SRI International, Menlo Park, CA
2. Departments of Chemical Engineering and Material Science
Understanding complex chemical systems, such as the chemistry of the polluted urban
atmosphere, requires a mathematical model describing the physics and chemistry of the
assemblage. Despite many years of laboratory experiments that have improved our
understanding immensely, it is not possible to measure every possible reaction that
should be considered in the model. Thus, estimation techniques based on laboratory
understanding have been used extensively. Methodologies such as Benson's
"Thermochemical Kinetics" and additivity and structure activity relationships (SAR)
have been widely applied. Often however insufficient experimental measurements exist
with which to begin extrapolation to larger molecules.
Recent advances in computational quantum mechanics have made it possible to compute
potential energy surfaces for reactions that have not been measured. It would then seem
possible, to compute gas phase rate constants from judicious use of transition state theory
(TST) and/or the microcanonical version applied to unimolecular reactions and their
reverse known as RRKM theory.
We have set out to test this posit by computing the structural properties of reactants and
transition states needed to apply the above theory. We have computed these properties
for the reaction X + ethane = HX + ethyl for X = H, O, OH, NH2, CH3, and Cl. The
thermochemistry is well-known for these reactions and the rate constants have been
measured as functions of temperature for all of them. We have employed the following
quantum chemistry methods: B3LYP, MP2 and QCISD and used the gaussian basis sets:
6-31 lG(d,p) and 6-31 l++G(3df,2dp). We first examined the computed values for AH for
the reaction. We found that even the purportedly best calculation, QCISD(T) with the
larger basis set, missed the experimental values by about 3 kcal/mole in several cases (O,
OH and NHa). As has been noted by others, the structures and frequencies for the stable
molecules and the transition states were similar for all levels of calculation. Thus we
used the B3LYP structures and frequencies in conjunction with a TST code, written by
the late Alan Rodgers, that compares experimental rate constants as functions of
temperature with computed values, accepting the B3LYP structural information for
reactants and transition states, while searching for the best value of the activation barrier
(DH^o )• (Tunneling is accounted for iteratively using Eckaft corrections.) We have
found that using the ab initio structures we cannot fit the Arrhenius curvature measured
for either OH or NH2 with ethane. We have also looked at the O-atom reaction, but there
is some question about the low temperature data.
There remains little question in our minds that the approach is sound. The questions to
be resolved center around the meaning and accuracy of frequencies computed for the
-103-
-------
modes that represent internal rotation in the transition state. For instance, in the HO-
Ethane reaction the motion corresponding to HO rotating about the newly forming bond
that we may denote as HO—H—CaHs, is computed to be 41 cm"1. We can fit the
Arrhenius curvature with a value of 110cm"1. Is this within the realm of uncertainty in
these kinds of frequencies?
-104-
-------
A MEASUREMENT TECHNIQUE FOR HYDROXYACETONE
P. J. Klotz, E. S. C. Kwok, X. Zhou, J. H. Lee, and Y. -N. Lee
Environmental Chemistry Division, Department of Applied Science
Brookhaven National Laboratory, Upton, NY 11973
Introduction
Hydroxyacetone (HA) is mainly produced in the atmosphere from oxidation of hydrocarbons of the
type, CH3(R)C=CH2. Tuazon and Atkinson (1990) reported HA yield of 41% from the OH-initiated
oxidation of methacrolein in the presence of NOX. Since methacrolein is a major product of isoprene
oxidation (Carter and Atkinson, 1996), isoprene, a key biogenic hydrocarbon, is therefore expected to be an
important source for HA. Consequently, knowledge of ambient concentration of HA would provide
information needed to examine the applicability of isoprene reaction mechanisms developed in laboratory.
and to assess the contribution of isoprene to photooxidant production.
The commonly used GC-FID technique involving cryo-focusing is unsuitable for HA owing to HA's
thermal instability. When subjected to a temperature of 100 C for only a few seconds, HA was found to
disappear completely. Since HA is highly soluble in water (it's Henry's law constant being ~2 x 104 M atm"1
at 20 °C, Zhou and Lee, unpublished data), we developed a wet chemical technique similar in principle to the
one we reported earlier (Lee and Zhou, 1993), namely, based on derivatization following liquid scrubbing.
To increase the sensitivity, we adopted a fluorescence detection scheme based on o-phthaldialdehyde (OPA)
chemistry. The technique was deployed in the field during two measurement periods at a NARSTO site
located on Long Island (LI), New York. We report the principle and the operation of this technique and the
results obtained from these field studies.
Experimental Section
HA and other soluble carbonyls (e.g., formaldehyde) are incorporated into an aqueous solution using
a 28-turn glass coil scrubber; typical conditions used a sample air flow rate of 2.0 L min"1 and a liquid flow
rate of 0.30 mL min"1. The collected samples containing the soluble carbonyls are first derivatized using
NaHSOa to form the respective sulfonic acid complexes which are then analyzed using ion chromatography
(DuVal et al., 1985). The eluted sulfonic acids are detected on-line by way of a post column derivatization
technique which decomposes the acids to the respective carbonyl compounds and S(IV), and detects the
released S(IV) fluorometrically following derivatizing using OP A in the presence ofNH3. This analysis
scheme is a variation of that used for NH3 detection (e.g., Zhang et al., 1989). A schematic diagram showing
the technique in an automated configuration is given in Fig. 1. A chromatogram showing the detection of the
carbonyl compounds is shown in Fig. 2.
This technique was tested during the summers of 1997 and 1998. The measurement site, which is
located at Brookhaven National Laboratory (BNL), is exposed to a fair amount of vehicular emissions from
traffic within the laboratory. In addition, the site is within 3 miles of two major highways, subjecting it to
pollution when winds are from the south and the west. However, since LI is covered with dense vegetation,
biogenic emissions are also important.
HA was determined along with formaldehyde (FA) using a variation of the technique, namely, the
pre-column S(IV) derivatization was carried out in a batchwise fashion rather than continuously on-line. The
detailed procedure follows. Liquid samples collected in glass vials representing an integration of ca. 15 min
period were added with a pre-determined amount of 0.50 M S(IV) solution (pH 7.0) to result in a final
[S(IV)] of 10 mM. The sample was then manually injected into the 1C for analysis. The 1C system consisted
of an HPLC pump (Hitachi, model 6200A), a 6-port injection valve (Rheodyne, model 7161)
-105-
-------
CottScmbbcr
Exhaust
Into
Peristaltic
Pump
Computer
H IS
Solutions:
»1- ImM buffers
•2-10 mM HSO
»3-15mMHCl
»4-OPA-NH4Cl
95 • Boiate buffer
oltn.pHS
3
Figure 1. Schematic diagram of the
hydroxyacetone measurement
system in an automated
configuration.
Figure 2. A chromatogram showing the
detection of a mixture of aqueous
standards of the indicated carbonyl
compounds. HA and acetone were 5.0
uM and FA and methylglyoxal were 1.0
uM.
equipped with a 50 uL sample loop, an 1C column (Vydec, cat. No. 300IC405), and a fluorescence detector
(Shimadzu, model RF-551) with the excitation and emission wavelengths set at 330 nm and 380 nm,
respectively. The eluant whose flow rate was 0.60 mL min"1 contained a 2.0 mM*potassium hydrogen
phthalate buffer maintained at pH 2.73 with the presence of formic acid (40 mM). The effluent exiting the
1C column was mixed with (1) a 0.10 M borate buffer containing 4.0 mM NH4C1 maintained at pH 10.5 and
-106-
-------
(2) a 3.0 mM OPA solution containing 30% methanol. The two reagents were each mixed in at 0.31 mL
min"1 using a peristaltic pump.
Because of the substantial amount of S(TV) contained in the sample, the efficiency of the column
was found to have deteriorated after 5 or 6 injections when the column was saturated with S(IV). To remedy
this shortcoming, we limited the samples to 4 per day. At the end of the day, the column was regenerated by
flushing the column using a solution containing 20 mM H2O2 and 3 mM HC1 followed by a 50 mM HNO3
solution.
Results
The technique was found to perform reasonably well, although it is labor intensive in the manual
mode. The column performance deterioration arising from excess amount of S(FV) which limited the
number of measurements to 4 per day can be eliminated by back flushing the column in an automated
configuration. The limit of detection for HA was only marginal, i.e., ~0.4 ppbv, and was due mainly to the
post-column derivatization which resulted in dilution from the added reagents and axial mixing associated
with the increased tubing length which broadened the peaks, lowering the sensitivity. The measurement
. uncertainty is estimated to be ± 15-20%.
The concentrations of FA and HA measured during the summers of 1997 and 1998 are shown in Fig.
3; the available O^ concentrations are also included. With the limited number of measurements, FA and Os
appear to be correlated. HA exhibits a slightly stronger diurnal dependence than FA, albeit both are weak
(Fig. 4). The median concentrations of FA and HA for the two measurement periods were 2.4 ppbv and 0.9
ppbv, respectively (Fig. 4).
8/27/97 8/29/97 8/31/97
9/2/97
9/4/97
9/6/97
9/8/97
9/10/97 9/12/97
— o
j&
• FA
A HA
,.~»4lk j...
**
...A.
.'.... ..... .
8/11/98
8/13/98
8/15/98 8/17/98 8/19/98 8/21/98
Date_Time
Figure 3. Concentrations of formaldehyde and hydroxyacetone determined during summers of 1997 (upper
panel) and 1998 (lower panel) at a NARSTO site located at BNL, LI, NY. Available data on O3 are also
shown.
A correlation plot of HA and FA (Fig. 5) showed that the [HA]/[FA] ratios were confined within a
region below a line with a slope of ~ 1.2 and an x-axis intercept of ~0.7 ppbv. One may interpret this
maximum [HA] to [FA] ratio to have resulted from a common precursor which dominates the production of
these two species when present in sufficient concentration. The data points having ratios less than 1.2 reflect
the presence of other precursors of FA which do not concomitantly produce HA. A plausible candidate of
such a precursor is isoprene, which has also been shown to be an important precursor for formaldehyde (Lee
et al., 1998). Since a molar yield ratio of 1 to 6 for HA/FA is estimated from the isoprene-OH reaction
-107-
-------
/. " ,
'' . -•»
234
[FA], ppbv
Figure 4. Composite diurnal
variation of formaldehyde and
hydroxyacetone (left panel).
The curves represent weighted
averages which are similar in
magnitude to median values.
The right panel shows the box
plots of [FA] and [HA] where
the limits of the boxes
represent the mid 50
percentile. The lines in the
boxes are the median values.
Figure 5. Correlation between
hydroxyacetone and
formaldehyde. The dash line
represents an estimate of the
upper limit of the ratio of [HA]
to [FA], at a value of 1.2
(NRC, 1991), we need a ratio of disappearance rates of these compounds of ~6 to 1 favoring FA in order to
maintain this steady state concentration ratio of 1.2. For a noon time j(FA) value of 7 x 10"5 s"1 and [OH]~1
x 107 molec cm"3, the destruction rate constant ratio of FA to HA is calculated to be ~5.8 (the rate constants
of the OH reactions with FA and HA are l.Ox 10"11 and 2.0 x 10"12 cm3 molec"1 s"1, respectively), in good
agreement with the suggestion of isoprene being a candidate of this precursor. It may be noted that
anthropogenic hydrocarbons such as isobutene and 2-methyl-l-butene are also precursors to HA. However,
since their concentrations are typically small, their contributions may be unimportant in regions where
isoprene emission is significant.
Following the hypothesis given above, the observed ratio of the median concentrations of HA to FA,
namely, 0.35 (Fig. 4), can be used to estimate that isoprene contributes >l/3 of the FA, which is a key radical
precursor, at the measurement site if isoprene is the sole precursor for HA. Further, the intercept of the
straight line in Fig. 5 suggests a day time background FA concentration of ~0.7 ppbv presumably due to long
lived species such as CBL}, in close agreement with that evaluated from the study of other isoprene products,
i.e., glycolaldehyde and methylglyoxal (Lee el al., 1998).
-108-
-------
Conclusions
A technique, which involves liquid scrubbing using a glass coil, pre-column derivatizing using
S(IV), 1C separation, and post-column derivatizing using OPA followed by fluorescence detection, was
developed for measuring ambient hydroxyacetone. This technique was deployed in the field at a NARSTO
site at BNL, LI, NY during two 2-week summer periods in 1997 and 1998. The [HA] measured showed a
median concentration of 0.9 ppbv, with a maximum of ~4 ppbv, limit of detection being ~0.4 ppbv. An
argument is presented that at this measurement site isoprene is the dominant source of HA and contributes
>l/3 of the formaldehyde.
Acknowledgements
This research was performed under the auspices of the United States Department of Energy, under
Contract No. DE-AC02-76CH00016, and with Support provided by Consolidated Edison of New York and
Long Island Lighting Company. The O3 data were provided by S. R. Springston of Environmental
Chemistry Division, BNL.
References
Carter, W. P. L. and R. Atkinson, Development and evaluation of a detailed mechanism fo the atmospheric
reactions of isoprene andNOx. Int. J. Chem. Kinet, 28, 497-530, 1996.
DuVal, D. L., M. Rogers, and J. S. Fritz, Determination of aldehydes and acetone by ion chromatography.
Anal. Chem. 57, 1583-1586, 1985.
Lee, Y.-N. and X. Zhou, Method for the determination of some soluble atmospheric carbonyl compounds.
Environ. Sci. Technol. 27, 749-756, 1993.
Lee, Y. -N., Zhou, X., Kleinman, L. I. Nunnermacker, L. J., Springston, S. R., Daum, P. H., Lee, Newman,
L., Keigley, W. G., Holdren, M. W., Spicer, C., Young, V., Fu, B., Parrish, D. D., Holloway, J., Williams, J.,
Roberts, J. M., Ryerson, T. B., Fehsenfeld, F. C. Atmospheric chemistry and distribution of formaldehyde
and several multi-oxygenated carbonyl compounds during the 1995 Nashville/Middle Tennessee Ozone
Study. J. Geophys. Res., 103, 22449-22462, 1998.
National Research Council (NRC), Committee on tropospheric ozone formation and measurement,
Rethinking the ozone problem in urban and regional air pollution, Natl. Acad. Press, Washington, D.C.,
1991.
Tuazon, E. C. and R. Atkinson, A product study of the gas-phase reaction of methacrolein with the OH
radical in the presence of NOX. Int. J. Chem. Kinet., 22, 591-602, 1990.
Zhang, G. P. K. Dasgupta, and S. Dong, Measurement of atmospheric ammonia. Environ. Sci. Technol., 23,
1467-1474, 1989.
-109-
-------
FAST GAS CHROMATOGRAPHY WITH LUMINOL DETECTION FOR
MEASUREMENT OF NITROGEN DIOXIDE AND PANs
Jeffrey S. Gaffney*, Nancy A. Marley, and Paul J. Drayton
Environmental Research Division
Bldg. 203/ER
Argonne National Laboratory
Argonne, IL 60439
630-252-5178
gaffney® anl.gov
Abstract
Fast capillary gas chromatography has been coupled to a luminol-based
chemiluminescence detection system for the rapid monitoring of nitrogen dioxide and
peroxyacyl nitrates. A first-generation instrument was described recently (Gaffney et al.,
1998). This system is capable of monitoring nitrogen dioxide and peroxyacyl nitrates
(PANs; to and including the C4 species) with 1-min time resolution. This is an
improvement by a factor of five over gas chromatography methods with electron capture
detection. In addition, the luminol method is substantially less expensive than laser
fluorescent detection or mass spectroscopic methods. Applications in aircraft-based
research have been published electronically and will appear shortly in Environmental
Science and Technology (Gaffney et al., 1999a). An improved version of the instrument
that has been designed and built makes use of a Hammamatsu photon-counting system.
Detection limits of this instrumentation are at the low tens of ppt. The range of the
instrument can be adjusted by modifying sampling volumes and detection counting times.
A review of past work and of recent application of the instrumentation to field
measurements of nitrogen dioxide and PANs is presented. The data clearly indicate that
the luminol approach can determine the target species with time resolution of less than
1 min. Examples of applications for estimation of peroxyacetyl radical concentrations and
nitrate radical formation rates are also presented. This instrumentation can further be used
for evaluation of surfaces for loss of nitrogen dioxide and PANs, phenomena of possible
importance for sampling interfaces and chamber wall design. Our high-frequency field
data clearly indicate that the "real world" is not well mixed and that turbulent mixing and
plume-edge chemistries might play an important role in urban- and regional-scale
interactions. Dynamic flow systems might be required to evaluate such effects in new-
generation chamber studies.
AbriB
Schnelles kapillares Gas ist chromatography zu einem luminol chemiluminescent
Entdeckung System fur die schnelle Uberwachung von Stickstoffdioxid und peroxyacyl
nitrates verkoppelt worden hat basiert. Ein erstes Generation Werkzeug ODER Gera't ist
kiirzlich in der Literatur (Gaffney et al., 1998) vorhergehend beschrieben worden. Dieses
-110-
-------
System kann zu, Stickstoffdioxid und PFANNEN (zu und einschlieBli der C4 Spezies)
mil Imin. Zeit BeschluBfassung uberwachung. Dies ist eine Verbesserung eines Faktors
von funf iiber GC/ECD Methoden, und ist wesentlich weniger teuer als laser
fluoreszierende Entdeckung oder Masse spectroscopic Methoden. Anwendungen in
Flugzeug Forschung sind elektronisch herausgegeben worden werden und erscheinen
bald in Umweltwissenschaft & Technologic (Gafftiey et al., 1999a). Eine verbesserte
Ausfuhrung dieses Werkzeugs ODER Gera'ts ist entworfen worden gebaut ist worden
und, der Gebrauch einen Hammamatsu photon macht, der System za'hlt. Entdeckung
Grenzen dieser MeBgerate sind an den niedrigen zehn von ppt. Bereich des Werkzeugs
ODER Gera'ts kann durch Probieren von Banden und Entdeckung Zahlen Zeiten einstellt
werden. Eine Nachprufung der vergangenen Arbeit und der neuen Anwendungen dieser
MeBgerate zu Feld Messungen von Stickstoffdioxid und PFANNEN werden uberreicht.
Diese Daten anzeigen deutlich kann, daB diese Annaherung diese Spezies mil weniger als
1 min. Zeit BeschluBfassung bestimmen. Ebenso Beispiele fur Anwendungen fur
Schatzung von peroxyacetyl radikale Konzentrationen und nitrate werden radikale
Forrmmg Raten uberreicht. Diese MeBgerate konnen auch fur Abschatzung der
Oberflachen fur Verlust von Stickstoffdioxid und PFANNEN benutzt werden konnen, die
wichtig fur Probieren Schnittstellen und auch fur Kammer Wand Entwurf sein. Merken
Sie unseren fieldwork auch deutlich anzeigt wird konnen jener hohen Frequenz Daten
Schau, daB das "alltagliche" nicht gut gemischt, und jene turbulent Mischung und Feder-
Kante Chemie spielen eine wichtige Rolle in Stadtische und regionale maBstabgetreue
Wechselwirkungen. Dies kann erfordern zu, daB dynamischer Ablauf Systeme diese
Arten Wirkungen in neuem Generation Kammer Studien bewerten.
Introduction
The interaction of hydrocarbons and oxides of nitrogen in sunlight to produce
"photochemical smog" has been well studied over the years (Finlayson-Pitts and Pitts,
1986). Indeed, to assess this type of process under simulated atmospheric conditions,
many studies have used smog chambers and or other very large reaction cells and bags to
investigate the potential ozone-forming relationships between the various volatile
hydrocarbons and nitrogen oxides. These studies were very productive in establishing a
first-order knowledge of the complex chemistry involved in the formation of ozone and
other oxidants. However, problems in past work typically limited studies in smog
chambers to relatively high nitrogen oxide levels. Limiting factors included wall
adsorption and desorption of nitrogen oxide species, analytical limitations, and
difficulties in speciating the various nitrogen oxide compounds. In the past, the
workhorse for the measurement of nitrogen dioxide and nitric oxide was the
chemiluminescent reaction with ozone. This method has detection limits of
approximately 0.5 ppb for most commercial instruments, but it cannot detect nitrogen
dioxide directly, because the instrument detects NO and uses hot catalytic surfaces to
decompose all other nitrogen oxides (including NOj) to NO for detection (Finlayson-Pitts
and Pitts, 1986). The main problem with this method is inherent difficulty in the detection
of the emitting species (excited nitrogen dioxide). This species emits over a broad region
-111-
-------
beginning at approximately 660 nm, with a maximum at 1270 nm, and requires a red-
shifted photomultiplier for detection.
Use of luminol for direct detection of NO2 was demonstrated to have greater inherent
sensitivity than indirect ozone chemiluminescence detection (Wendel et al., 1983).
Detection limits of 5 ppt have been demonstrated for the luminol method. The detection
system uses a gas-liquid reaction leading to light emission with a maximum at
approximately 425 nm. This emission, at the maximum sensitivity for most
photomultiplier tubes, is responsible for the increased detection sensitivities. The biggest
problem with this method for direct measurement of NO2 has been interferences from
other soluble oxidants, particularly peroxyacyl nitrates (PANs).
Along with NC>2, PANs are important trace gas species associated with photochemical air
pollution. The PANs are a class of organic oxidants having the general chemical structure
ROOO-O-NO2. The most common PANs are peroxyacetyl nitrate (PAN; R=CH3-),
peroxypropionyl nitrate (PPN; R=CHsCH2-), and peroxybutyryl nitrate (PEN;
R=CH3CH2CH2-).
The PANs are in thermal equilibrium with the peroxyacetyl radical (RC=O-OO) and
NO2 (Gaffney et al., 1989). Because PANs are trapped peroxy radicals, they are an
important indicator species of the photochemical age of an air parcel, as well as a vehicle
for long-range transport of NO2 leading to the formation of regional ozone and other
oxidants. Typically, PANs are measured by using a gas chromatograph with electron
capture detection (BCD). Once automated, this method has been shown to be reliable and
quite sensitive, allowing levels of PANs to be measured in the troposphere at low parts
per trillion (e.g., Gaffney et al., 1993). Unfortunately, a number of other atmospheric
gases (e.g., 62, Freons, H2O) also have strong BCD signals or act as inferences and limit
the speed with which the analysis can be completed. Currently, the shortest BCD analysis
time for PAN, the simplest peroxyacyl nitrate, is approximately 5 min (e.g., Williams
etal., 1997).
Peroxyacetyl nitrate was shown to react with luminol by our group and by Stedman's
research group in the late 1980s. Both research groups demonstrated that a simple packed
column could separate the PAN signal from the NO2 signal. Indeed, a method was
demonstrated for measuring PAN and NO2 by using a gas chromatograph coupled to a
luminol detector (Burkhardt et al., 1988). This method is the basis for a commercial
instrument (Scintrex, LMA-4) that has demonstrated the potential of the approach and
has achieved detection of NO2 and PAN with 5-min time resolution. We revisited this
approach and applied fast capillary gas chromatography to accomplish the analysis for
NO2, PAN, PPN, and PEN with an analysis time of less than 1 min (Gaffney et al., 1998).
Recently, we used this approach for aircraft measurements of NO2 and PAN (Gaffney
et al., 1999a). Described here is a new design for a smaller instrument based on the
application of a Hammamatsu photon-counting system. Examples of the type of
information that can be accessed with this method are also presented and discussed.
-112-
-------
Experimental Approach
The experimental approach for the original configuration of our instrument has been
described in detail (Gaffney et al., 1998). Figure 1 is a photograph of the second-
generation instrument that we are currently developing for aircraft and ground-based
measurements.
The major improvements in the second-generation system are the replacement of the
LMA-3 (Scintrex) luminol detection system with photon counting. This gives us direct
access to the photomultiplier tube via computer control for both PMT voltage and digital
signal processing. We are currently using a simple basic program for data acquisition and
processing of the signal. An electronic sampling valve with a timer is used to collect an
air sample and inject it onto the gas chromatographic column for analysis. The sample
volume injected is typically 1-5 cm3, depending upon the sensitivity required for the
measurement. Current detection limits of this system with a 5-cm3 sample are
approximately 20 ppt for NO2 and PAN. A 5% O2/He carrier gas mixture is used at a
flow rate of 40 cnr min"1. The capillary column used, 3 m long with a DB-1 coating,
allows for sufficient separation of NO2 from PANs. Calibration standards for PANs were
prepared by using a strong acid nitration of the corresponding peracid and extraction into
n-tridecane (Gaffney et al., 1984).
Results and Discussion
Figure 2 shows raw data for three samples taken at random in a recent study in Vineland,
New Jersey, as part of the collaboration of the U.S. Department of Energy's Atmospheric
Chemistry Program in the Northeast Oxidant and Particulate Studies (NEOPS), based in
Philadelphia, Pennsylvania, in July-August of 1999. The first peak monitored, NQj, is
followed by PAN. A 2-cm3 sample loop was used for these measurements. The data
correspond to PAN levels of approximately 0.5 ppb and NO2 levels of approximately
2 ppb. The results were obtained by taking the photon-counting data at 1-sec intervals.
We are currently examining signal-to-noise levels at faster counting times as part of our
instrument development work. The second-generation system successfully acquired data
in automated operation for approximately two weeks during this study.
Our first-generation instrument, which coupled a fast gas capillary column to a
commercial LMA-3 detection system with a Hewlett Packard integrator and a laptop
computer with Peak 96 software (Hewlett Packard) for data processing was used in our
previous studies (Gaffney etal., 1998,1999a, 1999b). One advantage of the luminol
detection method is that it can collect data for bom NO2 and PAN at the same time. This
advantage can allow the estimation of the peroxyacetyl radical if pressure and
temperature data are obtained simultaneously. We have done this in previous aircraft-
based work and have measured PAN/NO2 ratios that are consistent with the anticipated
thermal properties of PAN (Gaffney et al., 1989, 1999a, 1999b). During summer 1998,
we made ground-based measurements over a period of approximately one month at a site
in the foothills east of Phoenix, Arizona. We will be reporting results from this study at
-113-
-------
the upcoming annual meeting of the American Meteorological Society (Gaffney et al., in
press).
In the Phoenix study, we noted that rapid, specific measurement of NO2 with our first-
generation instrument and concurrent measurement of ozone allowed us to determine the
production rate of nitrate radical. Data obtained in this field study clearly showed that
NOi was being produced at night by the titration of ozone with locally generated NO.
These plumes then were transported in the nighttime boundary layer winds into and past
our site at Usery Pass. Calculations clearly indicate that nitrate radical production rates
will be very high on the edges of the plumes. This is, of course, where the ozone from
background air and the high NO2 in the plume mix to form NO3. Future chamber studies
will have to address such mixing issues. Rapid measurement techniques like this method,
which we believe can be improved to reduce the analysis time for NO2 and PAN to
approximately 30 sec (see Figure 2), and other methods based on fluorescence
spectroscopy or negative ion mass spectroscopy will be useful for evaluating such effects,
because they will have the time resolution to determine the effects of spatial
inhomogeneities and turbulent mixing on the chemistry.
We have also used the rapid luminol detection method to examine losses of NO2 and
PAN on a variety of different surfaces including Teflon, aluminum, copper, and
carbonaceous soots. The method could be useful for selecting appropriate materials for
chambers used in studies of NO2 and PANs. Along this line, we have set up a simple flow
system with this detector and tested various tubing sections to determine loss of material.
Examples of results with this type of approach have been reported (Gaffney et al., 1998).
In comparisons to more conventional BCD measurement techniques in both laboratory
and field studies, the rapid luminol detection approach has been found to agree quite well,
with comparable detection sensitivities for PANs in the low tens of ppt. Calibration for
PANs can be accomplished by the thermal decomposition of PANs to form NO2 and
subsequent monitoring of the resulting increase in NO2 signal during chromatographic
analysis. Nitrogen dioxide calibration gas standards (traceable to the National Institute of
Standards and Technology) are used to calibrate the NO2 and PAN signals by comparison
of the ambient and thermally decomposed PAN standard chromatograms (Gaffney et al.,
1998). Use of a chromate converter permits use of this method for NO measurements.
If we are to expand chamber studies to investigate the influence of very low nitrogen
oxide levels on the production of oxidants and ozone, very low detection sensitivities will
be required to measure and characterize nitrogen oxide species. Fast gas chromatography
with luminol detection has real promise in this area. We are continuing detector
development, with future attention to be given to minimizing dead volumes and
optimizing signal processing aspects of the photon-counting system. An update on our
results for this system and on our work on a real-time hydrocarbon detection system will
be reported at the American Meteorological Society meeting (Drayton et al., in press).
-114-
-------
Acknowledgements
The authors gratefully acknowledge the continuing support of the U.S. Department of
Energy, Office of Science, Office of Biological and Environmental Research,
Atmospheric Chemistry Program, through contract W-31-109-Eng-38 to Argonne
National Laboratory.
References
Burkhardt, M.R., N.I. Maniga, D.H. Stedman, and R.J. Paur, 1988. "Gas
Chromatographic Method for Measuring Nitrogen Dioxide and Peroxyacetyl Nitrate in
Air without Compressed Gas Cylinders." Anal. Chem., 60, 816-819.
Drayton, P.J., C.A. Blazer, N.A. Marley, and J.S. Gaffney, In Press. "Improved
Instrumentation for Near Real-Time Measurement of Reactive Hydrocarbons, NOa, and
Peroxyacyl Nitrates (PANs)." Preprint volume, American Meteorological Society 2000
meeting, Symposium on Atmospheric Chemistry Issues in the 21st Century, Long Beach,
California, January 2000.
Finlayson-Pitts, B.J., and J.N. Pitts, Jr., 1986. Atmospheric Chemistry. John Wiley and
Sons, New York.
Gaffney, J.S., R. Fajer, and G.I. Senum, 1984. "An Improved Procedure for High Purity
Gaseous Peroxyacetyl Nitrate Production: Use of Heavy Lipid Solvents." Amos.
Environ. 18,215-218.
Gaffney, J.S., N.A. Marley, and E.W. Prestbo, 1989. "Peroxyacyl Nitrates (PANs): Their
Physical and Chemical Properties." The Handbook of Environmental Chemistry,
Volume 4/Part B (Air Pollution), edited by O. Hutzinger, 1-38, Springer-Verlag, Berlin.
Gaffney, J.S., N.A. Marley, and E.W. Prestbo, 1993. "Measurements of Peroxyacetyl
Nitrate (PAN) at a Remote Site in the Southwestern United States: Tropospheric
Implications." Environ. Sci. Technol. 27, 1905-1910.
Gaffney, J.S., R.M. Bornick, Y.-H. Chen, and N.A. Marley, 1998. "Capillary Gas
Chromatographic Analysis of Nitrogen Dioxide and PANs with Luminol
Chemiluminescent Detection." Atmos. Environ., 32, 1145-1154.
Gaffney, J.S., N.A. Marley, H.D. Steele, PJ. Drayton, and J.M. Hubbe, 1999a. "Aircraft
Measurements of Nitrogen Dioxide and Peroxyacyl Nitrates Using Luminol
Chemiluminescence with Fast Capillary Gas Chromatography." Environ. Sci. Technol.,
Internet publication 8/14, in press.
Gaffney, J.S., N.A. Marley, and PJ. Drayton, 1999b. "Regional-Scale Influences on
Urban Air Quality: A Field Study in Phoenix, Arizona." Symposium on Interdisciplinary
-115-
-------
Issues in Atmospheric Chemistry, American Meteorological Society, Dallas, Texas,
Paper 2.1, pp. 26-31, January.
Gaffhey, J.S., N.A. Marley, P.J. Drayton, M.M. Cunningham, J.C. Baird, and
J. Dintaman, In Press. "Phoenix, Arizona, Revisited: Indications of Aerosol Effects on
Os, NOa, UV-B, and NO3." Preprint volume, American Meteorological Society 2000
meeting, Symposium on Atmospheric Chemistry Issues in the 21st Century, Long Beach,
California, January 2000.
Wendel, G.J., D.H. Stedman, and C.A. Cantrell, 1983. "Luminol-Based Nitrogen Dioxide
Detector." Anal Chem., 55, 937-940.
Williams, J., J.M. Roberts, F.C. Fehsenfeld, S.B. Bertman, M.P. Buhr, P.D. Goldan,
G. Hubler, W.C. Kuster, T.B. Ryerson, M. Trainer, and V. Young, 1997. "Regional
Ozone from Biogenic Hydrocarbons Deduced from Airborne Measurements of PAN,
PPN, and MPAN." Geophys. Res. Lett. 24 1099-1102.
-116-
-------
Figure 1. Luminol instrument with luminol cell and photon-counting system
dismantled to show the various parts of the instrument, peristaltic pump, luminol
reservoirs, air sampling pump, mass flow controller, and gas chromatographic column.
An external automated sampling valve is used to sample the air pulled through a
sampling loop at 1-min intervals.
Rgure 2. Raw Data from Vineland, NJ, 7/30/99, 11:39 to 11:41 am EOT. \
30
25
20
15
10
5
21 41 61 81 101 121 141 161
Time in Seconds
-117-
-------
Simulation Chamber Study of Night-time Chemistry
of Aldehydes and PANs
J.F. Doussin and Carlier P.
Laboratoire Interuniversitaire des Systemes Atmospheriques, UMR-CNRS 7583,
61 Av. du General de Gaulle, 94010, Creteil cedex, France
doussin@lisa.univ-parisl2.fr
Introduction
Aldehydes are major secondary products of the atmospheric oxidation of most of the volatile
organic compounds (Carlier et al, 1986). Their night-time reactivity with NO3 could lead to the
production of related peroxyacylnitrate (Cantrell et al, 1986). However, the kinetic database for
the reactions NO$ + RCHO is still inconsistent and the need of absolute measurement has been
pointed out (Wayne et al, 1991). On the other hand, the night-time chemistry of the peroxyacyl
radicals could be of importance in the night-time production of OH radicals (Canosa-Mas et al,
1996). Simulation chamber experiments were conducted together with numerical simulation to
investigate the full mechanism of the degradation of formaldehyde, acetaldehyde and
propionaldehyde.
Experimental
All the experiments were performed in a 0.977 m3 pyrex simulation chamber of 6 metres length
which has been described elsewhere (Doussin et al, 1997). The reactor is equipped with IR and
UV long path mirror system of 108 and 72 m optical path lengths, respectively. Both long path in-
situ FTIR and UV-visible DOAS measurements were used in order to perform absolute studies by
measuring the concentrations of both VOCs and NO3. The concentrations were monitored by UV
spectrometry at 662 run for the nitrate radical and by FT-IR spectrometry for others NOy and
organic compounds.
was produced by mixing NO and O3 in a small flow reactor connected to the chamber.
Methods
Model calculations that include 63 reactions in the case of acetaldehyde and 93 in the case of
propionaldehyde were performed for each experiment and the unknown rate coefficients were
determined by fitting the model results to the experimental data (Doussin, 1998). The comparison
between model results and experimental data is shown in Figs 1-2.
Results and discussion
The results are relevant of three main scientific fields : reactivity of NO3 with aldehyde, reactivity
of NOa with peroxyacyl radicals and heterogeneous chemistry of peroxyacyl radicals.
a) The mechanistic studies of the reaction of aldehyde with NO3 have confirmed that these
processes proceed exclusively via overall H-atom abstraction on the aldehyde group. Under our
experimental conditions (low NO level due to ozone concentration - see Figs 1-2) the PANs yield
reach rapidly 25-35%.
The kinetic results of the first step of these processes are reported in Table 1-3.
-118-
-------
Table 1. : Rate constant of the reaction NO3 + HCHO - Comparison with previous works.
T°(K)
298+1
298
298
298
296
k(NO3 H- HCHO)
molec"1 cm3 s"1
(3.2±0.2)xlff16
(6.3+1. l)xl(T16
(5.4±l.l)xlff16
(8.0+1. 5)xlff'6
(5.2+0.9)xlff16
k(N03 + HCHO) updated *
molec"1 era3 s"1
(5.0±0.4)xlO-'6
(5.4±1.0)xlCr16
Source NO3
N205
0S+N02
O3+NO2
Oj+NOx
References
Atkinson et al, 1984
Cantrell et al, 1985
Hjorth et al, 1988
Hannachi, 1986
This work
* Data updated from new reference values in case of relative rate study.
Table 2.: Rate constant of the reaction NO3 + CH3CHO - Comparison with previous works.
T°(K)
300
298+1
299+1
298
298
k(NO3 + CH3CHO)
molec"1 cm3 s"1
(1.2+0.3)xlff15
(1.3+0.3)xlCr15
(2.1+0.4)xlff15
(2,74+Q.07)xlff'5
(2.1±0.7)xlff15
k(NO3 + CH3ono) updated *
molec"1 cm3 s"1
(2.8+0.7)xlff15*
(2.1±0.4)xlCr's"
Source NO3
NO 2 + 03
N205
N205
N205
O}+NOX
References
Morris et al, 1974
Atkinson et al, 1984
Cantrell et al, 1986
Dlugokenski et al, 1989
This work
' Data updated from new reference values in case of relative rate study.
Table 3. : Rate constant of the reaction NO3 + C2H5CHO - Comparison with previous works.
T (K)
k(NO3 + C2H5CHO)
molec"1 cm3 s"1
Source NO3
References
298
298
(5.7+0.4)xlff'-
(4.5+0.9)xl(r'
N20S
03+NOx
D'Annaetal, 1997
This work
b) The reactions between NO3 and peroxyacyl radicals have been added to the mechanism.
RC(0)02 + N03 -> RC(0)0 + N02 + O2
A third in-situ generation of nitrate radical was performed when the PANs concentrations were
high and aldehydes concentrations were low. As can be seen from Fig. 2, when the NOs
concentration increases, PAN decomposition rate increase. It is well known that the reaction
-119-
-------
Injeclton 6.22 ppm NO
ln
-------
Injection 622 ppm NO •
CH,CHO Injection
5000
10000
15000 20000
Time (s)
25000 30000
Fig. 2: Experimental (thin line) and model (bold line) results. Concentration time behaviour without and with
acetaldehyde (2 stages experiments) followed by a thud generation of NO3.
-121-
-------
of PANs with NO3 is very slow so this result is the first evidence of a reaction between
peroxyacyl radicals and nitrate radicals under simulated tropospheric conditions. This procedure
allowed us to study for the first time under atmospheric conditions these important reactions and
to determine the associated rate constants with a good sensitivity of the fit.
Table 4.: Rate constant of the reaction NO 3 + peroxyacyl - Comparison with previous works.
T* (K) k(N03 + CH3co,) molec"1 cm3 s'1
Method References
300-403 20xlCr'2 Slow flow reactor - preliminary study Biggs et al, 1994
403-443 (4±l)xlO~12 Flow reactor Canosa-Mas et al, 1996
29S±2 (3.2±1.4)xHr12 see text This work
T* (K) k(NO3 + CjHsCO:,) molec"1 cm3 s"1
298±2 (2.5±1.2)xlOr12
Method • References
see text This work
c) We took into account the wall reactions of peroxyacyl radicals (Schurath et al, 1979) in order
to simulate our experimental data. In accordance with Langer et al (Langer et al, 1992) the
following process was considered.
(Gas) CH3C03 [CH3C(0)Of -> CH3 + CO2
(Wall)
CH3CO3 -> CH3C(O)O + O
,3P
The adsorbed oxygen atom could then react with the adsorbed nitric acid to give OH and NO3.
The pseudo first order rate parameter of this process was found to be highly variable with the wall
conditioning. Considering a S/V ratio of 9 m"1 for our reactor, the value 0.1.s"1 < klst < 7.S"1.
are in good agreement with the previous results of Langer et &l(Langer et al, 1992).
References
Atkinson, R.; Plum, C. N.; Carter, W. P. L.; Winer, A. M. and Pitts, J. N. J.; Journal of Physical Chemistry , 88 ,
(1984), 1210-1215.
Biggs, P.; Canosa-Mas, C. E.; Hansen, K. J.; Owen, P. S. and Wayne, R. P.; In Eurotrac '94 ; Borrell, P. M.,
Borrell, P., Cvitas, T., kelly, K., Seiler, W., Ed.; SPB academic publishing : Garmish -Partenkirchen ; pp
139-143 ,1994.
Canosa-Mas, C. E.; King, M. D.; Lopez, R.; Percival, C. J.; Wayne, R. P.; Shallcross, D. E.; Pyle, J. A. and Daele,
V.; Journal of the chemical society - Faraday Transaction , 92 , (1996), 2211-2222.
Cantrell, C. A.; Davidson, J. A.; Busarow, K. L. and Calvert, J. G.; Journal of the geophysical research , 91,
(1986), 5347-5353.
Carlicr, P.; Hannachi, H. and Mouvier, G.; Atmospheric Environment , 20 , (1986), 2079-2099.
D'Anna, B. and Nielsen, C. J.; Journal of the chemical society, Faraday transaction , 93 , (1997), 3479-3483.
Dlugokenski, E. J. and Howard, C. J.; Journal of physical chamistry , 93 , (1989), 1091-1096.
Doussin, J. F. These de Doctoral en Sciences (Ph-D Thesis), Universite Paris 7 - Denis Diderot, 1998.
Doussin, J. F.; Rite., D.; Durand-Jolibois, R.; Monod, A. and Carlier, P.; Analusis , 2 , (1997), 236-242.
Langer, S.; Wanberg, I. and Ljungstrom, E.; Atmospheric Environment , 26A , (1992), 3089-3098.
Moms, E. D. J. and Niki, H.; Journal of physical chemistry , 78 , (1974), 1337-1338.
Schurath, U. and Wipprecht, V.; In The first European symposium on Physico-Chemical Behaviour of
Atmospheric Pollutants ; Restelli, G., Abgeletti, G., Ed.: JRC - Ispra (Italy) ; pp 157-166 , 1979.
Wayne, R. P.; Barnes, I.; Biggs, P.; Burrows, J. P.; Canosas-Mas, C. E.; Hjorth, J.; Le, B., G.; Moortgat, G. K.;
Pemer, D.; Poulet, G. and Restelli, G., Sidebottom.H.; Atmospheric Environment , 25 A , (1991), 1-206.
-122-
-------
EPA PM Chemistry Studies
E.G. Edney(1), E. Swartz(1), DJ. Driscoll(1), T.E. Kleindienst(2), W. Li(2), C.D. Mclver(2), and
T.S. Conver(2), (I) U.S. Environmental Protection Agency, Research Triangle Park, NC, 27511,
(2) ManTech Environmental Technology, Inc., Research Triangle Park, NC 27509
Extended Abstract
Although PM2.5 can be directly introduced into the atmosphere through primary emissions, the
mass concentration of PM2.5 is also affected by secondary processes such as nucleation or
condensation of nonvolatile and semivolatile compounds on pre-existing PM2.5. To address these
issues, a laboratory research program was developed at EPA to investigate the key chemical
processes that determine the contributions of secondary processes to the overall mass
concentrations of PMa.5. The program consists of experiments to (1) measure the secondary
organic aerosol (SOA) yields of atmospherically relevant hydrocarbons under ambient
concentration and relative humidities; (2) determine the chemical composition of the multi-
functional compounds in the SOA chamber samples and compare those findings with field study
results; (3) measure the partitioning coefficients of semivolatile compounds in SOA; and (4)
evaluate methods for collecting semivolatile SOA.
As part of the program a series of chamber experiments was carried out to determine to what
extent photochemical oxidation products of aromatic hydrocarbons contribute to SOA aerosol
formation through uptake on pre-existing inorganic aerosols in the absence of liquid water films
(1). The irradiation experiments were conducted with toluene, ^-xylene, and 1,3,5-
trimethylbenzene in the presence of NOX and ammonium sulfate aerosol, with propylene added
to enhance the production of radicals in the system. The mass concentration of the organic
fraction was obtained by multiplying the measured organic carbon concentration by 2.0, a
correction factor that takes into account the presence of hydrogen, nitrogen, and oxygen atoms in
the organic species. In addition, mass concentrations of ammonium, nitrate, and sulfate as well
as total gravimetric mass concentrations were measured. The reconstructed mass concentrations
were in reasonable agreement with the gravimetrically determined values. The largest secondary
organic aerosol yield of 1.59 ± 0.40% was found for toluene at an organic aerosol concentration
of 8.2 ug m"3, followed by 1.09 ± 0.27% for p-xylene at 6.4 u.g m'3, and 0.41 ± 0.10% for 1,3,5-
trimethylbenzene at 2.0 jxg m"3. In general, these results agreed with those reported by Odum et
al.,(2) and support the gas-aerosol partitioning theory developed by Pankow (3). The presence of
organics in the aerosol did not affect significantly the hygroscopic properties of the aerosol for
relative humidities between 35% and 80%.
-123-
-------
References:
-------
Development of the Master Chemical Mechanism (MCMv2.0) web site
and recent applications of its use in tropospheric chemistry models
Sandra M. Saunders, Nicola Carslaw, Stephen Pascoe, and Michael J. Pilling
School of Chemistry, University of Leeds, Leeds, UK
Michael E. Jenkin
AEA Technology pic, National Environmental Technology Centre, Culham, UK
Richard G. Derwent
Meteorological Office, Bracknell, UK
Introduction
Atmospheric research in the School of Chemistry (Leeds) involves several groups
working in closely related areas, studying a wide variety of topics of relevance to
atmospheric processes. The Tropospheric Chemistry Modelling group is involved with
all aspects of model construction and application. Part of this work has been a
collaborative project, funded by the UK Department of the Environment, Transport and
the Regions (DETR), to develop and apply predictive models to the formation of
tropospheric ozone on a range of different geographical scales (i.e. global, regional and
national). The insight gained in this manner can aid in the formulation of policy with
regards to the air quality and ambient levels of ozone in the United Kingdom. The
master chemical mechanism (MCM) underpins much of the current ozone modelling
undertaken on the behalf of the DETR.
The main intention of the web site is to provide a flexible, easily utilised platform for the
MCM that is readily accessed by the whole research community, and to promote its
collaborative development and evaluation. This paper details updates and developments
that have occurred since the launch of the MCMvl.O web site (Saunders et al. 1997).
The current MCMv2.0 consists of around 10500 reactions and 3500 species and the web
site is located at:
http://chem.leeds.ac.uk/Atmospheric/MCM/mcmproj.html
MCM chemistry updates
The degradation schemes of MCMvl.O have been revised and updated in several areas.
The inorganic chemistry section and generalized rate parameters have been completely
reviewed. Updates incorporate data available in the open literature through 1997,
including the evaluations in the IUPAC (Atkinson et al., 1997a, 1997b) and JPL
(DeMore et al., 1997) publications.
Photolysis parameter updates have been made using a newly developed UV-flux model
(photol) with an improved representation of the effect of elevated concentrations of
particles, ozone and nitrogen dioxide in the boundary layer. An important revision has
been the incorporation of recently published quantum yields for ozone photolysis.
-125-
-------
Details of the model calculations are given in an earlier project report (Jenkin et al.,
1997). Revised parameters are now available, which define the photolysis rate as a
function of solar zenith angle for 28 reactions, for conditions appropriate to the boundary
layer over NW Europe.
The schemes for aromatic VOC have been completely revised since MCMvl .0, to take
account of data available in the open literature through July 1997 (eg. Yu and Jeffries,
1997; Carter et al., 1995; Wiesen et al, 1995; Klotz et al., 1995, 1997). Full details of
the supplementary protocol implemented for the aromatic mechanism construction are
presented in Jenkin et al. (1997) and are summarised on the web site.
Recently reported kinetic and mechanistic information relevant to the degradation of
organic oxygenates has been used to update the MCMvl.0 schemes for alcohols, glycols,
ethers, glycol ethers ands esters. Details of the construction are given in Jenkin et al.
(1997) and in the open literature (Jenkin and Hayman, 1999)
New VOC schemes in MCMvZ.O
The primary VOC list of MCMvl.0 has been extended to include two of the proposed
fuel additives, dimethoxy methane and dimethyl carbonate. In addition, a degradation
scheme has been constructed for cc-pinene, in order to give a representation of a
monoterpene. The complete VOC list now totals 123 VOC species, segregated into the
classifications given in Table 1. In addition, there is a comprehensive inorganic scheme.
Table 1 - Classification of the 123 primary VOC of MCMv2.0
Classification Number of species
Alkanes
Alkenes
Dialkenes
Alkynes
Aromatics
Aldehydes
Ketones
Alcohols and Glycols
Ethers and Glycol ethers
Esters
Organic acids
Other Oxygenates
Monoterpenes
Chloro-carbons
22
15
2
1
18
6
10
17
10
8
3
2
1
MCMv2.0 web site
On entering the web site, the home page shows a menu of contents and some general
introduction. Details of the MCM project, degradation scheme construction protocol
and recent developments are located within the project description area, under the
following section headings:
-126-
-------
• Project Description
• Project Objectives
• Introduction to the Master Chemical Mechanism (MCM)
• Development of the MCMv2.0
• Construction of the MCMv2.0
• Brief Protocol Review
• Aromatic scheme development
The What's New button shows the user which changes and additions have been made
to the original MCMvl.O site. New features of the system are highlighted, including the
MCM subset assembler tool, a facility to enable the assembly of subsets of MCMv2.0.
The MCM subset assembler tool allows the user to select any number of compounds
from the primary VOC list, and extract a complete, consistent, degradation mechanism
for that subset (Figure 1). . .
To view asubmechanism enter the list of initial species, separated by commas. Alternatively view the catalogue of root
Figure 1 - The subset assembler tool extracting the degradation scheme for cc-pinene.
-127-
-------
Changes have also been made to the MCM archive. The archive download area now
includes a repository of the files available on the MCMvl.O web site, and a new file store
of facsimile codes, which incorporate the current MCMv2.0 data set.
MCM web site developments
Work is continuing to further improve the web site. One improvement will be the
enhancement of the subset assembler tool. In addition, there will be improved species
and reaction visualisation for multi-platform use throughout the site. For instance,
Figure 2 shows a visual display of the reaction within the subset assembler tool. This
new development will not require the set up of a helper application for the internet
browser in use.
Figure 2. Platform independent display of reaction structures
We also hope to make MCMv2.0 and the accompanying models on the web site,
available in Fortran code in the near future.
-128-
-------
MCM maintenance and development
The degradation schemes for aromatic hydrocarbons will be updated in line with new
kinetic and mechanistic data, as they become available. The particular features of
aromatic degradation which have the most influence on ozone formation will be
identified by performing appropriate POCP sensitivity studies.
The representation of a number of gas-phase chemical processes will be updated in line
with recently published kinetic and mechanistic data. In particular, the following
reactions:
• the reactions of OH with PAN species
• the reactions of oxy radicals formed from the degradation of esters and alkenes
• the formation of excited oxy radicals from the reactions of some peroxy radicals with
NO.
In addition, we will also be updating the photolysis rates of ozone and other inorganic
and organic species in line with the latest absorption cross-section and quantum yield
data. We will also investigate the feasibility of incorporating or improving the
description of the formation of secondary sulphate and nitrate aerosols, in order to
improve the representation of important heterogeneous processes (e.g. the reaction of
N2O5 with water).
MCM recent applications
The MCM code is being used in a wide range of tropospheric models. Recent
applications include the following work:
• An investigation of the sensitivity of POCPs for oxygenated organic compounds to
variations in kinetic and mechanistic parameters (Jenkin and Hayman, 1999)
• POCPs for 123 organic compounds under north American conditions calculated with
the MCMv2.0. (Derwent et al, 1999a)
• Comparison of modelled OH, HO2 and RO2 concentrations using the MCM with
measurements in the marine boundary layer (Carslaw et al., 1999a,b)
• Detailed study of isoprene chemistry using the MCM (Carslaw et al., 1999c)
• Hydroxyl radical concentrations estimated from measurements of trichloroethylene
during the EASE/ACSOE campaign at Mace Head, Ireland during July 1996.
(Derwent et al. 1999b)
-129-
-------
References
Atkinson R., D.L. Baulch, R.A. Cox, R.F. Hampson, J.A.Kerr, M.J. Rossi and J. Troe (1997a)
Evaluated Kinetic and Photochemical Data for Atmospheric Chemistry: Supplement V. J. Phys. Chem.
Ref.Data,26,p52l-\Ol\.
Atkinson R., D.L. Baulch, R.A. Cox, R.F. Hampson, J.A.Kerr, M.J. Rossi and J. Troe (1997b)
Evaluated Kinetic and Photochemical Data for Atmospheric Chemistry: Supplement VI. J. Phys. Chem.
Ref.Data,26, 1329-1499.
Carter W.P.L., D. Luo, I.L. Malkina and J.A. Pierce (1995) Project ME-9, Environmental
chamber studies of atmospheric reactivities of VOC. California air resources board contract A032-
0692, and south coast air quality management contract C91323.
Carslaw N., D.J. Creasey, D.E. Heard, A.C. Lewis, J.B. McQuaid, M.J. Pilling, P.S. Monks,
B.J. Bandy and S.A. Penkett (1999a) Modelling OH, HO2 and RO2 radicals in the marine boundary
layer: 1. Model construction and comparison with measurements. J. Geophys. Res. - in press.
Carslaw N., P.J. Jacobs and M.J. Pilling (1999b), Modelling OH, HO2 and RO2 radicals in the
marine boundary layer: 2. Mechanism reduction and uncertainty analysis, submitted to J. Geophys. Res.
in press.
Carslaw N., N. Bell, A.C. Lewis, J.B. McQuaid and M.J. Pilling (1999c), A detailed study of
isoprene chemistry during the EASE96 Mace Head campaign: July 17th 1996, a case study, submitted to
Atmospheric Environment.
Derwent R.G., M.E. Jenkin, S.M. Saunders and M.J. Pilling (1999a) Photochemical ozone
creation potentials for 123 organic compounds under north American conditions calculated with the
Master Chemical Mechanism v2.0 - California Air Resources Board.
Derwent R.G., N. Carslaw, P.O. Simmonds, M. Bassford, S. .O'Doherty, D. Ryall, M.J. Pilling,
A.C. Lewis and J.B. McQuaid (1999b) Hydroxyl radical concentrations estimated from measurements
of trichloroethylene during the EASE/ACSOE campaign at Mace Head, Ireland during July 1996.
J. Atmos. Chem., 34, pi85.
Jenkin M.E., G.D. Hayman (1999) Photochemical Ozone creation potentials for oxygenated
organic compounds: sensitivity to variations in kinetic and mechanistic parameters. Atmospheric
Environment, 33, 1275-1293.
Jenkin M.E., G.D. Hayman, R.G. Derwent, S.M. Saunders and M.J. Pilling (1997)
Tropospheric chemistry modelling: Improvements to current models and application to policy issues. 1st
annual report AEA/RAMP/20150/R001.
Klotz B.G., A. Bierbach, I. Barnes and K.H. Becker (1995) Kinetic and mechanistic study of
the atmospheric chemistry of muconaldehydes. Environ. Sci. and Technol, 29, p2322.
Klotz B.C., A. Bierbach, I. Barnes, K.H. Becker and B.T. Golding (1997) Atmospheric
chemistry of benzene oxide/oxepin. Chem. Soc. Faraday Trans., 93, pi 507.
Saunders S.M., M.E. Jenkin, R.G. Derwent and M.J. Pilling (1997) Report Summary: World
Wide Web site of a Master Chemical Mechanism (MCM) for use in tropospheric chemistry models.
Atmospheric Environment, 31, p!249.
Wiesen E., I. Barnes and K.H. Becker (1995) Study of the OH-initiated degradation of the
aromatic photooxidation product 3,4-dihydroxy-3-hexene-2,5-dione. Environ. Sci. and Technol., 29,
pi 830.
Yu J. and H.E. Jeffries (1997) Atmospheric photoxidation of alkylbenzenes. Atmospheric
Environment, 31, 2281-2287.
-130-
-------
Session I (continued)
Gas Phase Chemistry and Modeling
Session Chair
David Golden
-------
Aromatic Hydrocarbon Research
in the European Photoreactor (EUPHORE)
Bjorn Klotz
Bergische Universitat, Physikalische Chemie, GauBstrafie 20, D-42097 Wuppertal, Germany
Present address:
The National Institute for Environmental Studies, Atmospheric Environment Division,
16-2 Onogawa, Tsukuba, Ibaraki 305-0053, Japan, E-mail: bjoern.klotz@nies.go.jp
Introduction
Aromatic hydrocarbons are rated amongst the most important classes of compounds with regard
to the formation of photo-oxidants in the lower atmosphere. A recent theoretical study [1] using a
trajectory model approach estimated the aromatic hydrocarbons to be responsible for about 40 %
of the anthropogenic ozone formation hi north-western Europe, their contribution to the total non-
methane hydrocarbon emissions was calculated to be 26 %.
Despite their importance, the degradation mechanisms of aromatic hydrocarbons, which
determine photo-oxidant formation, are still uncertain. This is mainly due to their far greater
complexity compared to those of other compound classes like alkanes or alkenes. This
complexity and the problems associated with it will be described hi more detail below, when
different aspects of aromatic hydrocarbon photochemistry that have been studied in the
EUPHORE photochemical reaction chamber are highlighted.
Experimental
The experiments were performed hi the European Photoreactor (EUPHORE), a large volume
outdoor photochemical reaction chamber located hi Valencia/Spain. EUPHORE consists of two
half-spherical FEP-Teflon bags with a volume of ca. 200 m3 each. The chambers are equipped
with multiple analytical instruments for the detection of trace substances, i.e. DOAS and FTIR
spectrometers, on-line monitors for ozone, NOX, and several GC and HPLC systems. Light
intensities can be measured with J(NO2) and J(O !D) filter radiometers and a spectral radiometer.
The experiments presented were conducted hi chamber A, which is shown hi Figure 1.
For more detailed descriptions of EUPHORE, including instruments not used and therefore not
mentioned hi this contribution, the reader is referred to the other papers on EUPHORE present hi
this volume or the EUPHORE reports compiled Becker [2] and Barnes and Wenger [3].
-132-
-------
FEPfoii
DOAS opposite mirrors
•ventilators
PT-IR opposite
mirrors
protective
housing
main
valve
FF-IR field mirror
J(0'D)&J(NO,)
radiometers
floor panels, cooled
DOAS field mirror
Fig. 1: Chamber A of the European Photoreactor (EUPHOKE), with some of the instrumentation
located inside the chamber shown. In addition, sampling ports for on-line and off-line
instruments are situated in the central flanges. The circle shows two co-workers sitting on the
protective housing behind the chamber.
Results and Discussion
- Toluene photo-oxidation [4]
The objective of the study of the photo-oxidation of toluene/NOx/air mixtures was the in situ
determination of the yields of ring-retaining products by Differential Optical Absorption
Spectroscopy (DOAS) and the elucidation of their formation pathways. Experiments were
performed with toluene concentrations between 0.68 and 3.85 ppm and initial NOX concentrations
ranging from 3 to 300 ppb, i.e. down to the range actually observed in the lower atmosphere. The
ring-retaining product yields were found to be (5.8 ± 0.8) %, (12.0 ± 1.4) %, (2.7 ± 0.7) % and
(3.2 ± 0.6) % for benzaldehyde, o-cresol, wt-cresol and /?-cresol, respectively. Under the
experimental conditions, no dependency of the yields on the NOX concentration or the
toluene/NOx ratio could be found. The formation kinetics of the cresols are in line with a
"prompt" formation mechanism, i.e. loss of a hydrogen atom from the toluene-OH adduct. The
exact pathway could either be reaction of the adduct with molecular oxygen (toluene-OH + 02 —=>
cresol + HOj), or its unimolecular decomposition (toluene-OH —> cresol + H), as proposed by
Bjergbakke et al. [5].
-133-
-------
In addition, evidence was found that reaction with NOs radicals represents an important sink for
cresols in smog chamber studies, as predicted by Carter et al. [6]. This is well illustrated by the
concentration-time profiles of toluene, benzaldehyde and the cresol isomers shown in Figures 2a-
c. To show the effect of NOs radicals on these experiments, the concentration-time profiles were
simulated using a simplified chemical mechanism, in which the OH radical concentration was
fitted to the observed toluene decay. This mechanism included loss of benzaldehyde and the
cresols through dilution and OH radical reaction, but not by reaction with NOs radicals. From
Figures 2a-c it becomes clear that this mechanism is sufficient to describe to the behavior of
benzaldehyde, but the behavior of the cresols is only described well in the experiment shown in
Figure 2c. In the experiments shown in Figure 2b, and even more so in that shown in 2a, the
simplified scheme overestimates the observed cresol concentrations. The three experiments
shown in Figure 2a-c have an increasing toluene/NOx ratio, and the apparent NOs effect should
therefore be most significant in 2a, and least significant in 2c, which is observed. The effect of
NOs radical reactions should be insignificant for benzaldehyde, which has an NO3 radical
reaction rate constant about 4 orders of magnitude lower than that of the cresols. This too is
observed.
1.7
1.6
1,3
11.00 12:00 13:00 14:00
time of day (GMT) |h:min|
10:00 10:30 11:00 11:30 12:00 12:30
lime of day (GMT) |h:mln|
9:30 10:00 10:30 11:00 11:30 12:00
lime of day (GMT) [h:min]
Fig. 2: Concentration-time profiles of toluene (O) and its ring retaining oxidation products
benzaldehyde (+), o-cresol (•), m-cresol (T) and p-cresol (A). Estimated NOs concentrations
are shown as broken lines. Starting concentrations were (a): 730 ppb toluene 7115 ppb NO, (b):
3850 ppb toluene / 120 ppb NO, (c) 3800 ppb toluene / 28 ppb NO. The lines represent simulated
profiles, see text for details.
In addition, concentrations of NOs radicals could be estimated based on a steady state
approximation, where the reaction of NO2 with Os was the only source of NOs, and sinks were
reactions with NO and the cresols and photolysis, see Klotz et al. [4] for details. Maximum
calculated NO3 concentrations are highest in 2a (1- 108 cm"3), while 2b (2.5- 107 cm"3) and 2c
(3- 106 cm"3) show successively lower calculated NOs concentrations, in line with the lower
observed effect attributed to NOs. Moreover, in Figure 2b the formation rate of o-cresol shows an
anticorrelation to the calculated NOs concentration. When the NOs concentration rapidly drops
after 11:00 h due to limited availability of NOX, the o-cresol formation rate increases again.
-134-
-------
Similar effects are observed for m- and p-cresol, but are not as apparent as for o-cresol due to the
much lower concentrations of these isomers.
From the combined results it is inferred that reaction of the cresols with the nitrate radical is the
only viable explanation for the observed effects. Reaction with NOs radicals is concluded to be a
significant to major sink for cresols in aromatic hydrocarbon photo-oxidation experiments
conducted under conditions of NOX concentrations above the range observed in the troposphere.
It possibly also plays a role under urban tropospheric conditions, but daytime NOs radical
concentrations are insufficiently well known for a conclusion to be drawn at this time.
- Photolysis ofarene oxides [7]
A study of the photolysis of two postulated intermediates in the OH initiated photo-oxidation of
aromatic hydrocarbons, benzene oxide/oxepin and toluene-l,2-oxide/2-methyloxepin, has been
conducted under the real light conditions of EUPHORE. The main focus of the study was on the
formation of the hydroxylated aromatic hydrocarbons, i. e. phenol from benzene oxide/oxepin
and the cresol isomers from toluene-l,2-oxide/2-methyloxepin. Photolysis frequencies were
measured relative to that of NO2, which was determined with a filter radiometer. A linear
dependency was found within the experimental errors, the ratios of the photolysis frequencies of
the arene oxides to that of NO2 were found to be (4.41 ± 0.44) • 10'2 and (3.99 ± 0.48) • 10'2 for
benzene oxide/oxepin and toluene-l,2-oxide/2-methyloxepin, respectively. The yield of phenol in
the photolysis of benzene oxide/oxepin was found to be (43.2 ± 4.5) %, independent of the light
intensity. For toluene-l,2-oxide/2-methyloxepin, an o-cresol yield of (2.7 ± 2.2) % was
determined, the formation of other cresol isomers was not observed.
2.5 10'-
!l.510'J
t
«
X
il.01013
1°
3 s.o io12
no13
810"
•§•
6 10'2.—
3
410'2f
21012
o ^,
= 30
10
(b)
1200 1800 2400 3000 3600 4200 4800
reaction time [s]
46
J(NO.) I
12
Fig. 3: (a): Photolysis of benzene oxide/oxepin (•), leading to the formation of phenol (A) in
high yields, (b): Dependency of the photolysis frequency of benzene oxide/oxepin on that of
NO2.
-135-
-------
As an example, Figure 3a shows concentration-time profiles of benzene oxide/oxepin and its
photolysis product phenol in one of the experiments conducted, Figure 3b shows the observed
dependency of the benzene oxide/oxepin photolysis frequency on that of NO2.
- Photolysis ofdicarbonyls [8]
Unsaturated 1,6-dicarbonyls like 2,4-hexadienedials are ring opening products in the OH initiated
photo-oxidation of aromatic hydrocarbons. The photolysis of two isomeric hexadienedials, E,Z-
and E,E-2,4-hexadienedial, has been investigated under natural sunlight conditions in the
EUPHORE chamber. In the case of the E,Z-isomer, an extremely rapid isomerization into the
E,E-fbrm was observed. Figure 4a shows concentration-time profiles for one of the experiments
conducted. A first order decay of the E,Z-2,4-hexadienedial is observed. The photoisomerization
frequency, relative to that of NO2, was found to be J(E,Z-2,4-hexadienedial) / J(NO2> = (0.148 ±
0.012).
I2;00 12:30
13:00 13:30
9:00
9:30
10:00 10:30 11:00 11:30
time of day (GMT) [hrmln]
time of day (GMT) [h:mln]
Fig. 4: (a): Concentration-tune profiles of E,Z-2,4-hexadienedial (•) and E,E-2,4-hexadienedial
(^) in an experiment on the photolysis of E,Z-2,4-hexadienedial. (b): Concentration-time profile
of E,E-2,4-hexadienedial (4) in a photolysis experiment. Lines are simulated profiles using the
chemical mechanism shown in Figure 5. Aerosol number concentrations are also shown (V).
As evident from Figure 4b, the photolysis of E,E-2,4-hexadienedial did not show the behavior
expected from a first order process, indicating a different photolysis mechanism for this
compound. The behavior of E,E-2,4-hexadienedial could be described by the mechanism shown
in Figure 5. Here, a fast equilibrium is proposed to precede a comparably slow photolysis. For the
equilibrium reaction, relative frequencies of J(E,E-2,4-hexadienedial -» EQUI) / J(NO2) = (0.113
± 0.009) and J(EQUI -> E,E-2,4-hexadienedial) / J(NO2) = (0.192 ± 0.016) were obtained, giving
an equilibrium constant of K = (0.59 ± 0.07). For the photolysis frequencies, ratios of J(E,E-2,4-
hexadienedial -•> products) / J(NO2) = J(EQUI -> products) / J(NO2) = (1.22 ± 0.45) • 10"2 were
determined.
-136-
-------
hv
\J
E,Z-2,4-hexadienedial
EZHX
^_y
E,E-2,4-hexadienedial
EEHX
*
RE2 _
hv, fast
unknown
• compound
EQUI
slow|RE5
EQPROD
i
RE7 /hv
^ ^/slow
Fig. 5:
^^*- products
Proposed chemical mechanism of the photolysis of the 2,4-hexadienedial isomers.
Under typical atmospheric noontime light conditions for a clear sky and low aerosol
concentrations on July 1 at a latitude of 40° N, corresponding to the time and location where the
experiments were performed, an NO2 photolysis frequency of J(NO2) = (8.5 ± 0.5) • 10~3 s"' is
usually assumed [9]. Under those conditions, a photolysis lifetime of J(E,Z-2,4-hexadienedial) =
(13.2 ± 1.3) min can be expected. Lifetimes of E,E-2,4-hexadienedial are (17.4 ± 1.7) min for the
initial equilibrium reaction and (161 ± 60) min for the subsequent degradation under the above
conditions.
The results of the E,Z-2,4-hexadienedial photolysis are in line with those of an earlier study [10],
in which photolysis experiments were conducted with superactinic fluorescent lamps (320 nm <
A. < 480 nm, Xmax = 360 nm). In contrast, no photolysis of E,E-2,4-hexadienedial was observed
under the above laboratory conditions. This disagreement is most probably due to the differences
in emission characteristics between the artificial lamps and natural sunlight. Most importantly,
the superactinic lamps do not emit light in the region from 290 - 320 nm, which is of particular
importance for the hexadienedials, as their absorption spectrum shows an exponential decay
towards longer wavelengths [10]. This shows that photolysis experiments under realistic
atmospheric conditions are an indispensable tool to reliably predict the behavior of photolytically
reactive chemicals in the atmosphere.
Similar differences between photolysis frequencies measured under irradiation with superactinic
or "black" lamps can be expected for a large number of carbonyls, as the absorption cross
-137-
-------
sections of these compounds generally decrease exponentially from 290 nm towards longer
wavelengths.
A pronounced formation of particles was observed in the photolysis of both E,Z- and E,E-2,4-
hexadienedial, see Figure 4. To check for a possible influence of chamber aerosol sources, an
independent test experiment, in which the chamber was exposed to sunlight without the presence
of added reactants, was conducted. No significant particle formation was observed in this test
experiment. With the available instrumentation no size distribution measurements were possible,
so no aerosol yields can be given. However, from the available data it can be concluded that the
photolysis of 2,4-hexadienedials leads to the formation of secondary organic aerosol, which can
contribute to the aerosol formation observed in the photooxidation of aromatic hydrocarbons
[11,12].
The results indicate that the dominant fate of E,Z-2,4-hexadienedial in the atmosphere will be
photoisomerization, while for E,E-2,4-hexadienedial, both photolysis and OH radical reaction
will be important sinks. Another conclusion that can be drawn from the data is that the aerosol
formation in photolysis reactions, notably of aldehydes with longer carbon chains, deserves
increased attention from researchers.
- Chemical simulations using a box model
One aim of simulation chambers like EUPHORE is to increase the database on hydrocarbon
photo-oxidation experiments that can be used to verify the chemical mechanisms used in models
of tropospheric photo-oxidant formation. Particular uncertainties still exist regarding the
degradation channels of aromatic hydrocarbons, with several different schemes being proposed
by different authors. An overview of the currently proposed mechanisms is given in Figure 6.
The classic pathways [13,14] are shown as (a) and (b) in Figure 6. A detailed description of the
various mechanisms will not be given here, the reader is referred to existing literature for a
review [15]. Pathway (a) involves formation of a 1,3-oxygen-bridged alkyl radical, which
subsequently adds Oa and reacts with NO, followed by three |3 -scissions and reaction with Oi to
give HOa and 1,2- and unsaturated 1,4-dicarbonyls. These compounds currently represent the
largest ring-fragmentation products for which quantitative determinations exist in the literature
[15,16]. Their further oxidation is not shown for simplicity.
The same 1,2- and unsaturated 1,4-dicarbonyls can also be formed through pathway (b), by
further oxidation of an unsaturated 1,6-dicarbonyl. Such unsaturated 1,6-dicarbonyls have been
identified in the OH-initiated oxidation of aromatic hydrocarbons [17,18], though their yields and
consequently their importance remain unknown.
Unsaturated 1,6-dicarbonyls like those formed in pathway (b) can also be formed through the
further oxidation of arene oxides, as proposed in pathway (d) [19]. The unsaturated 1,6-
-138-
-------
dicarbonyl would again react on to give the 1,2- and unsaturated 1,4-dicarbonyls already
observed from pathways (a) and (b).
NO
02,NO.
ring-
scission
02,NO.
OH
NO2
H02
NO2
H02
Fig. 6: The different degradation pathways currently proposed for the addition channel in the OH
initiated oxidation of aromatic hydrocarbons. For simplicity, benzene is taken as an example.
It has to be noted at this point that though pathways (a), (b) and (d) result in the same products,
their formation is successively delayed in (b) and (d) by the occurrence of stable intermediates.
This difference will have important implication for atmospheric chemistry modeling studies, as
will be shown later.
Pathway (c) has recently been proposed based on theoretical calculations [20], and experimental
evidence for the formation of the proposed multifunctional compounds has been reported in the
literature [21]. These compounds may, however, also be formed by the OH-initiated oxidation of
the benzene oxide shown in pathway (d) [19].
It becomes clear from the above discussion that different mechanisms with different
intermediates are currently being considered feasible for the OH-initiated oxidation of aromatic
hydrocarbons. However, since these mechanisms basically all lead to the same set of ring-
fragmentation products, i.e. 1,2- and unsaturated 1,4-dicarbonyls, it has to be asked why it is
important to understand the exact chemistry.
It is well known that the OH initiated photo-oxidation of aromatic hydrocarbons results in a
significant enhancement of radical production compared to alkanes or alkenes which possess
-139-
-------
equal OH reaction rate constants [22]. These secondary radicals are produced in the photolysis of
carbonyl compounds, especially the dicarbonyls formed in the OH initiated oxidation of aromatic
hydrocarbons, see Figure 6.
Amongst the dicarbonyls shown in Figure 6, the 1,2- and unsaturated 1,4-dicarbonyls are
expected to contribute to photolytic radical formation. The unsaturated 1,6-dicarbonyls proposed
in pathways (b) and (d) of Figure 4 are thought to photolyze without significant radical formation
[8].
The photolysis of 1,2-dicarbonyls (glyoxal and its methylated derivatives) has been studied to
some extent [23,24] and can be described reasonably well. Some knowledge of the photolysis of
unsaturated 1,4-dicarbonyls is also available, but this is largely confined to photolysis
frequencies. Notably the radical yields in the photolysis of unsaturated 1,4-dicarbonyls are
currently not known.
160
140
120
~£
JjlOO
cT
-« 80
o
60
40
20
0
160
140
120
JiT
a 100
o"
* 80
60
40
20
0
30 60 90 120 150 180 210 240
reaction time IminJ
Yvrk""*X
••.•' 7* XA »
*/..
/»*v \ -0 -x
>*/ A *"*••-.. "• *"-.
tl«
-------
Fig. 7: Reactivity differences of different combinations of toluene ring-opening mechanisms, see
text for details. (A): 10 % of ring-opening proceeds through channel a in Figure 4, the rest
through channel d. (B): 30 % channel a and 70 % channel d. (C): 50 % channel a and 50 %
channel d. (D) RACM mechanism [25]. Legend: toluene (diamond, line), NO (filled circle, dotted
line), NC>2 (filled diamond, dashed line), ozone (filled triangle, long dashed line). Symbols
represent experimental data, lines show the results of the simulations.
If the ring-opening channels (a), (b) and (d) hi Figure 6 are compared, it becomes clear that the
radical producing products are formed directly, i.e. without a non-radical intermediate, only in
channel (a). In channel (b) they are formed through the further oxidation of an unsaturated 1,6-
dicarbonyl, while in channel (d) two non-radical intermediates, an arene oxide and an unsaturated
1,6-dicarbonyl, need to be oxidized to give the 1,2- and unsaturated 1,4-dicarbonyls.
When used in computer aided kinetic simulations of the atmospheric degradation of aromatic
hydrocarbons, these differences result in significant differences in reactivity. Figure 7 shows
simulations of a toluene/NOx photo-oxidation experiment using a simplified toluene oxidation
scheme. The model is based on the RACM mechanisms published by Stockwell et al. [25], in
which the toluene oxidation scheme has been updated to the current state of knowledge as shown
in Figure 6. There are two main differences between the RACM and the mechanism employed in
this study. The first is the incorporation of a direct cresol formation mechanism, shown as
channel (e) in Figure 6, utilizing the cresol formation yields recently reported [4,16,26]. The
second main difference is the replacement of the ring-opening reactions employed hi the RACM
by channels (a) and (d) of Figure 6. Channel (b) was omitted since, for the purposes of this study,
it only represents an intermediate between (a) and (d). The original RACM ring opening scheme
was based on channel (a) alone. The degradation of the unsaturated 1,4-dicarbonyls has been
updated based on the results of Bierbach et al. [27] and S0iensen and Barnes [28], no changes
were made to the degradation chemistry of the 1,2-dicarbonyls.
Simulations performed with this scheme are shown hi Figures 7a-d. Lines represent simulated
concentration/tune profiles, while symbols represent experimental data. All 4 panels of Figure 7
show the same experimental data, the starting concentrations were 700 ppb of toluene and 115
ppb of NOX (mainly in the form of NO). The corresponding experiment was performed in
EUPHORE. The different graphs in Figure 7 represent calculations using different branching
ratios between the channels (a) and (d). In graph A, the ring opening is assumed to proceed
almost exclusively through the arene oxide channel (d), a ratio of ring opening between the
channels (a) and (d) of 10:90 has been employed here. This ratio is successively increased in
graphs B (30:70) and C (50:50) of Figure 7. Graph D shows a calculation with the original
RACM scheme.
It becomes clear from Figure 7 that changing the branching ratio between channels (a) and (d) in
favor of (a) greatly increases the reactivity of the system. This reflects the much faster formation
-141-
-------
of photolytic radical producers like 1,2- and unsarurated 1,4-dicarbonyls in channel (a). Thus, the
reactivity of the system is governed by the branching ratio between the "fast" and "delayed"
formation mechanisms of 1,2- and unsarurated 1,4-dicarbonyls.
For comparison, a simulation run employing the original RACM mechanism is also included in
Figure 7. The RACM scheme leads to a significantly increased reactivity of the system. This is
expected as the RACM toluene degradation mechanism is based on channel (a) shown in
Figure 6.
As a consequence of these differences in reactivity, the aromatic hydrocarbon degradation
pathways employed in chemical computer models of atmospheric photo-oxidation systems are
expected to have an influence on the reactivity of the entire system. The RACM mechanism [25],
for example, is based on channel (a) for ring-opening and should predict a much higher reactivity
than the Master Chemical Mechanism (MCM) developed by Saunders et al. [29]. The MCM is
based on channel (d).
Since these models are primarily used to calculate/predict photo-oxidant concentrations in the
atmosphere, the aromatic hydrocarbon degradation scheme employed in a certain model will have
consequences for the outcome of such predictions.
It becomes clear that a detailed knowledge of the degradation channels actually occurring in the
OH initiated oxidation of aromatic hydrocarbons is of considerable importance for the accuracy
of models of photo-oxidant formation. Additional research is needed to identify the pathways that
actually occur in the OH initiated oxidation of aromatic hydrocarbons and their branching ratios.
In addition, our understanding of the chemistry, notably the photolysis, of unsarurated
dicarbonyls needs to be significantly unproved.
Summary
Different aspects of aromatic hydrocarbon photochemistry have been investigated in the
EUPHORE reaction chamber. These include determinations of toluene ring retaining product
yields with in situ spectroscopic methods and an array of photolysis studies. Besides providing
new data on photolysis frequencies and -products, the photolysis studies showed that the
superactinic or "black" lamps often used hi indoor experiments can lead to significantly different
results compared to real sunlight. Such data can therefore not be readily extrapolated to
tropospheric conditions. Another aspect of the photolysis experiments is the observed formation
of secondary organic aerosol, which requires further studies.
A simple modeling study showed that the different proposed degradation schemes for aromatic
hydrocarbons result hi large differences in the reactivity of aromatic hydrocarbon/NOx systems as
-142-
-------
calculated by these models. The photolysis of small 1,2- and unsaturated 1,4-dicarbonyls, and the
pathways by which these compounds are formed, apparently govern reactivity. In particular,
significant differences between the predictions of the commonly employed mechanisms RACM
[25] and MCM [29] are expected. The RACM is based on a rapid mechanism for dicarbonyl
formation and should thus lead to a significantly higher reactivity than the MCM, which is based
on a delayed formation mechanism for the dicarbonyls.
Acknowledgements
The author is indebted to a large number of people who participated in the aromatic hydrocarbon
studies in Valencia. These are notably Thomas Etzkorn and Rainer Volkamer from the group of
Ulrich Platt (Heidelberg), who set up and operated the DOAS, Klaus Wirtz (Valencia) and his co-
workers, and the colleagues from Wuppertal who were involved in the experiments: Klaus
Brockmann, Florian Graedler, S0ren S0rensen and Lars Thiiner.
Financial support of the European Commission (RADICAL project) and the "Bundesministerium
fur Bildung, Wissenschaft, Forschung und Technologic" within the TFS program for this work is
gratefully acknowledged. "
References
1. R. G. Derwent, M. E. Jenkin and S. M. Saunders, Atmos. Environ., 30 (1996) 181.
2. K.H. Becker, (ed.) The European Photoreactor EUPHORE, Final Report of the EC-Project,
Contract EV5V-CT92-0059, Bergische Universitat Wuppertal, Dept. of Physical Chemistry,
Germany, 1996.
3. I. Barnes and J. Wenger, (eds.) EUPHORE Report 1997, Bergische Universitat Wuppertal,
Dept. of Physical Chemistry, Germany, 1998.
4. B. Klotz, S. S0rensen, I. Barnes, K. H. Becker, T. Etzkorn, R. Volkamer, U. Platt, K. Wirtz
and M. Martin-Reviejo, J. Phys. Chem. A, 102 (1998) 10289
5. E. Bjergbakke, A. Sillesen and P. Pagsberg, /. Phys. Chem., 100 (1996) 5729.
6. W. P. L. Carter, A. M. Winer, J. N. Pitts, Jr., Environ. Sci. Technol, 15 (1981) 829.
7. B. Klotz, I. Barnes and K. H. Becker, Chem. Phys., 231 (1998) 289.
8. B. Klotz, I. Barnes and K. H. Becker, Int. J. Chem. KineL, 31 (1999) 689.
9. D. D. Parrish, P. C. Murphy, D. L. Albritton and F. C. Fehsenfeld, Atmos. Environ., 17
(1983) 1365.
10. B. G. Klotz, A. Bierbach, I. Barnes and K. H. Becker, Environ. Sci. Technol., 29 (1995)
2322.
11. H. J. L. Forstner, R. C. Flagan and J. H. Seinfeld, Environ. Sci. Technol., 31 (1997) 1345.
12. J. R. Odum, T. P. W. Jungkamp, R. J. Griffin, H. J. L. Forstner, R. C. Flagan and J. H.
Seinfeld, Environ. Sci. Technol., 31 (1997) 1890.
-143-
-------
13. K. R. Darnall, R. Atkinson and J. N. Pitts, Jr., J. Phys. Chem., 83 (1979) 1943.
14. M. Hoshino, H. Akimoto and M. Okuda, Bull. Chem. Soc. Jpn., 51 (1978) 718.
15. R. Atkinson, J. Phys. Chem. Ref. Data, Monograph No. 2, (1994)
16. D. F. Smith, C. D. Mclver and T. E. Kleindienst, J. Atmos. Chem., 30 (1998) 209.
17. B. E. Dumdei and R. J. O'Brien, Nature, 311 (1984) 248.
18. J. Yu, H. E. Jeffries and K. G. Sexton, Atmos. Environ., 31 (1997) 2261.
19. B. Klotz, I. Barnes, K. H. Becker and B. T. Golding, J. Chem. Soc., Faraday Trans., 93
(1997) 1507.
20. L. J. Bartolotti and E. O. Edney, Chem. Phys. Lett., 245 (1995) 119.
21. J. Yu and H. E. Jeffries, Atmos. Environ., 31 (1997) 2281.
22. W.P.L. Carter, Atmos. Environ., 29(1995)2513.
23. L. Zhu, D. Kellis and C.-F. Ding, Chem. 'Phys. Lett., 257 (1996) 487.
24. S. Koch and G. K. Moortgat, /. Phys. Chem. A, 102 (1998) 9142.
25. W. R. Stockwell, F. Kirchner, M. Kuhn and S. Seefeld, J. Geophys. Res., 102 (1997) 25847.
26. R. Atkinson and S. M. Aschmann, Int. J. Chem. Kinet., 26 (1994) 929.
27. A. Bierbach, I. Barnes, K. H. Becker and E. Wiesen, Environ. Sci. Technol, 28 (1994) 715.
28. S. S0rensen and I. Barnes, in: EUPHORE Report 1997, I. Barnes and J. Wenger (eds.),
Department of Physical Chemistry, Bergische Universitat Wuppertal, Wuppertal, Germany,
December 1998, pp. 149-150.
29. S. M. Saunders, M. E. Jenkin, R. G. Derwent and M. J. Pilling, Atmos. Environ., 31 (1997)
1249.
-144-
_
-------
Studies on Oxygenated Fuel Additives: Ethers and Acetals
I. Barnes. K.H. Becker, L. Thiiner and T. Maurer
Bergische Universitat - Gesamthochschule Wuppertal, Physikalische Chemie / Fachbereich 9,
GauBstraBe 20, D-42097 Wuppertal, Germany.
Phone: +49 202 439 2510,
Fax: +49 202 4392505,
E-mail: barnes@physchem.uni-wuppertal.de
Ever increasing political pressure for a cleaner environment has triggered large research efforts
over the past.decade within the automobile and solvent industries to develop new, oxygenated,
low atmospheric reactivity fuels and solvents, respectively. The main stimulus for the interest in
such compounds in Europe is existing or pending legislation demanding the regulation of the
reactivity of the anthropogenic VOC emissions to the atmosphere, in particular from solvents and
motorised vehicles. Obviously to assess the environmental acceptability of alternative oxygenated
solvents or fuel additives a comprehensive understanding of their gas-phase oxidation mechanism
is required, preferably prior to deployment of the compounds.
Within the EU INFORMATEX project and also the German TFS/LT3 project much effort has
been directed towards investigation of the atmospheric photo-oxidation mechanisms of organic
compounds used (or under consideration) as fuel additives. The INFORMATEX and TFS
projects concentrated on investigating the atmospheric photo-oxidation mechanisms of ether
compounds already in use for gasoline reformulation, i.e. MTBE, ETBE and TAME. Since this
material is well documented this abstract will focus on the acetals which are not so commonly
known
Acetals are a group of oxygenated organic compounds with the general diether structure,
ROCH2OR (R = an organic alkyl group)
These substances have excellent solvent properties and also good toxicological and
ecotoxicological profiles which makes them suitable as potential replacements for many of the
chlorinated and aromatic solvents currently in widespread use. Recent research has also shown
that addtion of acetals to diesel fuel greatly improves the environmental properties of the engine
exhaust. Tests on methylal (CH3OCH2OCH3) and n-butylal (GjHgOClfcOCiIfc),- for example,
have shown that:
- the engine particle emissions are greatly reduced
- the NOX emissions are substantially reduced
- butylal improves the cetane number
-145-
-------
Since the widespread of such compounds will lead to release of large amounts of these substances
to the atmosphere it is obviously desirable to assess the potential environmental implications of
such releases prior to deployment. Reported here are exploratory investigations on the
atmospheric fate of a series of acetals performed in the laboratory and also in the large outdoor
EUPHORE simulation chamber in Valencia in Spain. The data is used to assess the ozone
forming potential of this class of compound and their primary oxidation products and their likely
influence on atmospheric reactivity on local and larger scales.
Experimental
The atmospheric behavior of the acetals has been investigated in laboratory photoreactors and
also in the large outdoor simulation chamber EUPHORE in Valencia Spain.
The laboratory experiments were performed in Wuppertal in a 1080 1 quartz reaction chamber at
298 K and a total pressure of 1000 mbar using synthetic air as the bath gas. The photolysis (A.max
- 360 nm) of CH ONO/NO was used as the OH radical source. FT-IR spectroscopy was the
main analytical method used for the analysis of reactants and products.
The experiments in the large-scale (200 m3) EUPHORE photoreactor were performed at 1000
mbar total pressure and between 284 and 290 K using purified air as the matrix gas. The principle
analytical techniques employed were in situ FTIR and GC.
Atmospheric Gas Phase Reactivity
The acetals currently under investigation are listed in Table 1.
• With the exception of methylal (CH2OCH2OCH2) all the acetals investigated react relatively
quickly with OH radicals (k between 2 and 5 x 10"'l cm3 molecule"1 s"1; see Table 1).
• This corresponds to atmospheric OH-lifetimes of approx. 60 h for methylal and between 1.4
and 6 h for the other acetals (calculated with [OH] = 1 x 106 molecule cm"3).
• Reactions with O2 and NO2 are negligibly slow.
• The acetals do not absorb in the atmospheric actinic region therefore photolysis does not
occur.
Ozone Forming Potential
Tests have been performed in the EUPHORE chamber to assess the ozone formation potential of
the acetals. Two different types of "classical" smog-type experiment have been performed; one
series of experiments with the pure acetal and another with additional reactive hydrocarbons.
Examples of the ozone formation are shown in Figure 1.
• the ozone production decreases along the series ethylal-propylal-butylal
• in mixtures with ethene they reduce ozone formation on a short-term irradiation basis (5-6 h)
-146-
-------
• none of the acetals is very efficient in the production of ozone on the basis of short term
irradiation experiments acetals, which give relatively high HCHO yields show the highest
ozone production
• in mixtures with ethene they reduce the ozone formation compared to pure ethene
Products and Oxidation Mechanism
For the straight chain acetals the major products from the OH initiated oxidation are alkoxy-
methyl formates (CnH(2n+i)OCH2OCHO) and aldehydes (CnH(2n+i)CHO) in equi-molar yield. In
addition di-alkyl carbonates (CnH(2n+1)OC(O)OCnH(2n+1)) are also formed in low yields. For the
branched acetals ketones are major products. The pathways leading to the above products are
illustrated for the reaction of OH radicals are illustrated for butylal in Figure 1.
Conclusions/Possible Atmospheric Implications
• acetals will be degraded quickly in the atmosphere producing mainly aldehydes and alkoxy-
methyl formates
• the ozone formation tests suggest that acetals will not contribute to ozone formation on a local
scale and may even lead to a reduction, however, 2 to 3 day scenarios are needed to assess the
long-term impacts
• the aldehydic products will certainly contribute to ozone formation over a longer time scale:
photolysis will lead to the formation of highly reactive alkenes and thermally stable PAN-like
compounds which can transport NOX
• the formates will be oxidised only slowly, mainly to diformates, and will probably be
removed by deposition - more work is needed on the fate of these compounds
Although the addition of acetals to diesel will certainly be beneficial with respect to particle and
NOX formation and also ozone production on a local scale the results suggest that on the long
term their use may result in the spread of reactivity over wider scales. The best environmental
option with regards to the use of alternative solvents / fuel additives may eventually, in many
cases, boil down to a choice between ,,the lesser of two evils".
References
1. Wallington, T. J., M.D. Hurley, JVC. Ball, A. M. Straccia, J. Platz, L. K. Christensen, J.
Sehested, and O. J. Nielsen, /. Phys. Chem. A, 101 (1997) 5302-5308.
2. Porter E., J. Wenger, J. Treacy, H. Sidebottom, A. Mellouki, S. Teton, and G. Le Bras, /.
Phys. Chem. A 101 (1997) 5770-5775.
3. Dagaut, P., R. Liu, T. J. Wallington, and M. J. Kurylo, Int. J. Chem. Kinet. 21 (1989) 1173-
1180.
4. Becker, K. H., C. Dinis, H. Geiger, and P. Wiesen, Chem. Phys. Letters 300 (1999) 460-464.
-147-
-------
Table 1: Summary of the recommended rate coefficients for the reaction of OH radicals with
acetals obtained in this work and reported in the literature.
Compound
dimethoxy methane (DMM)
diethoxy methane (DEM)
di-n-propoxy methane (DNPM)
di-iso-propoxy methane (DIPM)
di-n-butoxy methane (DNBM)
di-wo-butoxy methane (DIBM)
di-sec-butoxy methane (DSBM)
10" koH
cm3 molecule"1 s"1
0.49 ± 0.02
0.53 ± 0.05
0.46 ±0.16
0.49 ± 0.08
0.46 ± 0.01
1.84±0.18
2.04 ±0.1 4
2.06 ± 0.01
1.68 ±0.16
2.63 ± 0.49
3.93 ± 0.48
3.47 ± 0.42
3.21 ± 0.79
3.68 ± 0.57
4.68 ± 0.05
Technique'
RR
RR
PR-UV
RR
PLP-LIF
RR
RR
PLP-LIF
FP-RF
RR
RR
RR
ELP-LIF
RR
RR
Temperature
K
300 ±1
295 ±2
346 ±3
298 ±2
298 + 2
298 ±2
298 ±2
298 ± 2
298 ±
296 ±4
295 ±4
298 ±2
298-710n)
299 + 3
299 ±4
Reference
this work
Wallington et al. [1]
Wallington et al. [1]
Porter et al. [2]
Porter etal. [2]
this work
Porter et al. [2]
Porter et al. [2]
Dagaut et al. [3]
this work
this work
this work
Becker et al. [4]
this work
this work
i) RR •» relative kinetic technique; PR-UV = pulsed radiolysis - transient UV absorption; PLP-LIF = pulsed laser
photolysis - laser laser induced fluorescence; FP-RF = flash photolysis - resonance fluorescence; ELP-LIF -
eximer laser photolysis - resonance fluorescence.
ii) Rate coefficient was reported to be independent of temperature over the temperature range investigated.
-148-
-------
O O
Di-n-butoxy Methane
OH
OH
\
H9O
CH
H 02
O O
NO
NO2
NO
NO2
cr o
scission
n-Butoxy Methyl Formate
•o
Propanal
HO2
H02
o
II
•c/V
Di-n-butyl Carbonate
Figure 1: Mechanism for the OH radical initiated oxidation of n-butylal.
-149-
-------
Developments of Degradation Mechanisms
of Oxygenated Hydrocarbons
H. Geiger
Bergische Universitat Gesamthochschule Wuppertal / FB 9 - Physikalische Chemie
Gauflstrafie 20, D-42119 Wuppertal, FRG
Introduction
The increasing use of oxygenated organic compounds like ethers, esters and alcohols as fuel
additives, alternative fuels or solvents during the last years may lead to a higher influence of
these species on tropospheric chemistry. While the rate coefficients for the OH reactions of the
relevant oxygenated VOCs are mostly well established, the knowledge of the detailed.atmo-
spheric degradation pathways for these compounds is relatively poor. The mechanistic studies
are very often confined to the detection of the primary reaction products. Reaction mechanisms,
which are derived from these data, are usually not proofed by comparison of the measured prod-
uct concentrations and the model. In other cases, where no suitable experimental results are
available, reaction mechanisms are postulated without any further verification. Both methods are
characterised of high uncertainty when the models are applied, e.g. by field modelling.
In the present work, the construction of chemical degradation schemes for single VOCs con-
sidering both experimental data and computer simulations is demonstrated. Dimethoxymethane,
dimethoxyethane, 1,4-dioxane and 1,3-dioxolane were chosen as concrete examples for oxygen-
ated VOCs, which are expected to gam importance in tropospheric photochemistry for the future.
The degradation mechanisms of these compounds under urban tropospheric conditions were
defined and checked by using recent experimental results from our laboratory. The importance of
the individual reactions has been evaluated by means of sensitivity analyses.
Experimental Data and Computer Simulation System
The experimental data used in this work for the evaluation of the degradation mechanisms
described here were obtained in our laboratory. The experimental set-up used for these experi-
ments has been described in detail elsewhere (Barnes et al., 1994). Briefly, in a 1080 dm3 quartz
reactor, mixtures of VOC, methyl nitrite and NO in air at a total pressure of 760 Torr were irra-
diated with fluorescence lamps. Time resolved concentrations of VOC, primary reaction prod-
ucts, NOX and ozone were measured by using long path FTIR spectroscopy.
All computer simulations were carried out by using the boxmodel SBOX by Seefeld (1997) and
Seefeld and Stockwell (1999). This FORTRAN program incorporating the Gear algorithm (Gear,
1971) was operated on an SGI Origin 200 workstation running under IRIX 6.5. The program
uses the public domain library VODE (Brown et al., 1989) to integrate the ordinary differential
equations.
-150-
-------
Construction of the Mechanisms
a) Photolysis reactions, inorganic chemistry and OH source
In order to describe the photolysis processes in the reaction system investigated in the pre-
sent work a set of photolysis reactions taken from the RACM mechanism by Stockwell et al.
(1997) was used. The only photolysis frequency, which was experimentally measured under the
present conditions, was that of NO2 for photodissociation into O(3P) and NO. All other photoly-
sis frequencies were not measured and had to be estimated. It can be assumed, that the photolysis
behaviour of the different species in the photoreactor is almost similar to that in the troposphere
when suitable lamps are used. Under this assumption, photolysis frequencies can be calculated
for all species relative to J(NO2) for atmospheric conditions using the algorithm of Madronich
(1987). This approximation is justifiable because the radiation strength in the photoreactor is
much weaker than in the atmosphere and the influence of the photolysis processes on the radical
budget in the reactor is low for most of the treated species.
For the description of the inorganic processes in the model, a set of 35 inorganic reactions
taken from the RACM mechanism of Stockwell et al. (1997) has been added to the chemical
model used in the present study.
In the indoor photoreactor experiments relevant to this study, OH radicals were formed by the
photolysis of methyl nitrite, CHjONO, in the presence of NO. Since it was not possible to
accurately measure the initial concentration of CEbONO, the "real" concentration of CHjONO
was obtained from a fit of the simulated VOC profiles to the experimental data by variation of
the initial CHjONO concentration. Because the CHjONO photolysis should be the only OH
source under the given conditions, this approximation can be made.
b) Dimethoxymethane (DMM)
Dimethoxymethane (DMM) is a diether which is currently used as a solvent and recently has
been considered as an alternative diesel fuel. The major reaction products under urban conditions
are methoxymethylformate (MMF, about 70 mol%), dimethylcarbonate (DMC, about 25 mol%)
and methylformate (MF, about 5 mol%). Based on the experimental results of our laboratory, the
reaction scheme was constructed as illustrated in Fig. 1. The stated branching ratios were deter-
mined by fitting the model to the experimental data. Fig. 2 illustrates comparisons of experi-
mental and simulated data for selected reactants. Both data sets are in excellent agreement.
-151-
-------
32%
00
83%
CH2OO
17%
-CH30
M/-H
methylformate (MF)
methoxymethylformate (MMF)
products
products
products
tfmelhylcarbonale (DM0)
Fig. 1: Tropospheric degradation mechanism for dimethoxymethane in the presence of NOX.
products
10 15 20 25 30
Fig. 2: Comparison of experimental (symbols) and simulated (curves) concentration-time profiles for selected
reactants. Experimental conditions: 1.36 ppm DMM, 0.40 ppm NO, 0.60 ppm methyl nitrite, 760 Torr air.
A sensitivity analysis was carried out for the DMM degradation scheme in order to rate the
importance of the single reaction steps. Relative sensitivities Sn were calculated by using time-
averaged normalised sensitivity coefficients Sai (Stockwell et al, 1995) as shown by the follow-
ing equation:
-152-
-------
Sri ~ Vo
Fig. 3 shows as an example the relative sensitivities of NO, NO2, O3 and HNO3 to the most
important rate coefficients of the DMM degradation scheme.
MeONO photolysis
OH+DMM->ALK1
OH + NO2->HNO3
OH + DMM->ALK2
OH + NO->HONO
NO2 photolysis
O3 + NO->O2+NO2
OH + MeONO -> prod.
HO2+NO->OH+NO2
HO2 + NO2->HNO4
HNO4->HO2 + NO2
OH + HCHO->H2O + H
MONO photolysis
other reactions
=
^^^^M
_jEEiES
•
J
•
•
•
~ 1
'-""- "J
^^••M
M
•'
3
•
^
• •
n
is a
a a
L
s=4> , , ,
-
-
NO
NO2
O3
HNO3
-30 -20 -10 0 10 20
relative sensitivity (%)
30
40
50
Fig. 3: Relative sensitivity of NO, NO2,O3 and HNO3 to selected reaction rate coefficients of the DMM degrada-
tion scheme (ALK1 = CH3OCH2OCH2, ALK2 = (CH3O)2CH). Experimental conditions are given in Fig. 1.
The plot illustrates, that the chemistry of the reaction system is mainly determined by
CHsONO photolysis as the only considerable OH source and the OH reactions of DMM and
NOX. Photolysis reactions (except that of methyl nitrite) are of minor importance under the pre-
sent conditions. The values for Sn of DMM and the primary reaction products lead to the same
result. It can be followed, that such well-defined conditions are an excellent basis for the vali-
dation of degradation schemes for single VOCs.
c) Dimethoxyethane (DMET)
Similar to dimethoxymethane (DMM), dimethoxyethane (DMET) is a diether which is cur-
rently used as a solvent and recently has been considered as an alternative diesel fuel. Based on
experimental data of our laboratory, a tropospheric degradation scheme for DMET in the pres-
ence of NOX was developed. The mechanism is illustrated in Fig. 4. The branching ratios given
in Fig. 4 were obtained by fitting the model to the experimental results. We assume that 10% of
the DMET will be consumed by OH attack on one of the terminal CH3 groups, leading finally to
methoxyethylformate (MEF). The major fraction of 90% reacts via OH attack to one of the CH2
groups forming an alkyl radical which, after O2 addition and NO/NO2 conversion, leads to the
corresponding CH3OCH(O»)CH2OCH3 radical. In the present mechanism, 82% of these oxy
radicals decays thermally forming methylformate (MF) and a methoxymethyl radical which will
finally lead to another MF molecule, as shown in Fig. 4.
-153-
-------
^Ox
s^
. +OH/0 » r.
^^0^ ~1^% -^X
CH20
-/^ox'
DMET
+ OH/O
^O-,
NO
^°"
2|90%
^^0^
oo*
— j-^N02
^f^^n-^ 82%
O«
products
•
+ OH
•^°^v^H
O
^CH20*
T
HO,
O H
methoxyethylformate
(MEF)
hOH
products
18%
products
methylformate (MF)
+O,
- HO2 NO
x^-
•CH2O«
Fig. 4: Degradation scheme of the OH-initiated oxidation of dimethoxyethane in the presence of NOX.
The remaining 18% of the CH3OCH(O«)CH2OCH3 degradation products are unknown at
present. A couple of reaction pathways is conceivable, which is illustrated in Fig. 5. Besides the
isomerisation leading to another alkyl radical, reaction with NO/NOa forming the corresponding
nitrites/nitrates, H atom abstraction by O2 leading to methoxyacetic acid methylester and decay
to methoxyacetaldehyde and a methoxy radical may be possible.
Fig. 6 illustrates comparisons of experimental and simulated data for selected reactants. Both
data sets are in excellent agreement.
Isom.
OH
-CH3O»
+NO/NO
methoxyaoetaldehyde
02
H02
nilrila / nitrate
methoxyacetio acid methylester methylformate (MF)
Fig. 5: Possible degradation pathways for the secondary oxy radical, CH3OCH(O)CH2OCH3.
-154-
-------
0.5
8 10
Fig. 6: Comparison of experimental (symbols) and simulated (curves) concentration-time profiles for selected
reactants. Experimental conditions: 1.00 ppm DMET, 0.45 ppm NO, 0.45 ppm methyl nitrite, 760 Torr air.
d) 1,4-Dioxane
1,4-Dioxane is a cyclic diether which is currently used as a solvent and recently has been
considered as a fuel additive. The only primary reaction product of the OH-initiated degradation
specified in the experiments is ethylene glycol diformate (EDF). The carbon yield is more than
95 mol%. The remaining few mol% are unknown. Based on the experimental results of our labo-
ratory, the reaction scheme was constructed as illustrated in Fig. 7. Fig. 8 illustrates a comparison
of experimental and simulated data for selected reactants. Both data sets are in excellent agree-
ment. It was shown, that consideration of nitrate formation with ratios of 2.5% for each peroxy
radical reaction in the model led to a much better agreement of experimental and simulated data
(see Fig. 8).
-155-
-------
0
O
1 ,4-dioxane
IXV
^°\^°°"
OH/ O2^ p y^
^^ s*^ f\\
O \ll
NC
+ NO
2.5%
nitrate
(HI)
nitrate
+ NO
v~
2.5% o (IV)
NO >-NO2
sC H2O •
0 (V)
O2 «-HO2
products
OH
ethylene diformate (EOF)
Fig. 7: Degradation scheme of the OH-initiated oxidation of 1,4-dioxane in the presence of NOX.
-. 12
I ,
Ozone
1,4-dioxane
10
t (min)
t (min)
Fig. 8: Left: Comparison of experimental (symbols) and simulated (curves) concentration-time profiles for selected
reactants. Experimental conditions: 0.87 ppm 1,4-dioxane, 0.80 ppm NO, 1.15 ppm methyl nitrite, 760 Torr
air. Right: Deviations of the simulation, if nitrate formation is not considered in the model.
-156-
-------
e) 1,3-Dioxolane
1,3-Dioxolane is a cyclic diether, which is currently used as a solvent. Based on the experi-
mental results of our laboratory, the reaction scheme was constructed as illustrated in Fig. 9. The
branching ratio for the OH reaction of 1,3-dioxolane was determined by fitting the model to the
experimental data.
In the modelling exercise the NO and NO2 profiles were sensitive to the nitrate yield. It was
found that the best fits were achieved using a nitrate formation yield of 2.5% for reaction of NO
with each of the peroxy radicals (see Fig. 9). Fig. 10 shows as an example the results of the com-
puter simulations in comparison with the experimental data. Both data sets are in excellent
agreement.
x\
\ — /
1,3-dioxolane
+ OH/O 2
52.5%
I + OH/0 2
47.5%
NO
00
+NO
2.5%
NO 2
nitrate
nitrate
nitrate
+ NO
2.5%
CH2OO«
NO «-NO2
V
products
i. OH
wall loss
products
O O
m ethylene glycol diformate
Fig. 9: Degradation scheme of the OH-initiated oxidation of 1,3-dioxolane in the presence of NOX.
-157-
-------
0.25
0.15
0
E
O5
'x
0.1
0.05
MDF
10
15
20
Time (min) Time (min)
Fig. 10: Comparison of experimental (symbols) and simulated (curves) concentration-time profiles for selected
reactants. Experimental conditions: LI 4 ppm 1,3-dioxolane, 4.50 ppm NO, 0.28 ppm NO2, 3.20 ppm
methyl nitrite, 760 Torr air.
Conclusions
In the present work, urban tropospheric degradation mechanisms for dimethoxymethane
(DMM), dimethoxyethane (DMET), 1,4-dioxane and 1,3-dioxolane were postulated and checked
by comparison with suitable laboratory data. For all compounds, experimental and modelling
results are in excellent agreement. This leads to the possibility that these reaction schemes might
be used successfully in further applications, e.g. field modelling.
The results of the sensitivity analyses carried out for the investigated reaction systems indi-
cate that photoreactor experiments using methyl nitrite as OH precursor and relatively high
concentrations of NOX are characterised by well-defined conditions, which can be modelled with
high accuracy. As a consequence, such experiments are highly suitable for mechanism validation
procedures for degradation schemes of VOC under urban conditions.
Acknowledgements
Support for this research was provided by the German Bundesministerium fur Bildung, Wis-
senschaft, Forschung und Technologic (BMBF), project "Forderschwerpunkt Tropospharen-
forschung (TFS)", contract 07TFS30. The supply of experimental data by T. Maurer, C.G. Sauer
and L.P. Thiiner (BUGH Wuppertal, FRG) prior to publication is gratefully acknowledged. We
thank W.R. Stockwell (Reno/Nevada, USA) and J.B. Milford (Boulder/Colorado, USA) for the
supply of software and very helpful discussions.
-158-
-------
References
Barnes, I., Becker, K.H. and Mihalopoulos, N., /. Atmos. Chem., 18 (1994) 267.
Brown, P.N., Byrne, G.D. and Hindmarsh, A.C., J. Sci. Stat. Comput, 10 (1989) 1038.
Gear, C.W., Prentice-Hall series in automatic computation, Vol. 17, Prentice-Hall, Englewood
Cliffs, 1971.
Madronich, S., J. Geophys. Res., 92 (1987) 9740.
Seefeld, S. PhD Thesis, Swiss Federal Institute of Technology Zurich (ETH), Switzerland, 1997.
Seefeld, S. and Stockwell, W.R., Atmos. Environ., 33 (1999) 2941.
Stockwell, W.R., Kirchner, F., Kuhn, M. and Seefeld, S., J. Geophys. Res., 102 (1997) 25847.
Stockwell, W.R., Milford, J.B., Gao, D. and Yang, Y.J., Atmos. Environ., 29 (1995) 1591.
Detailed information about the present work:
H. Geiger and K.H. Becker
Degradation Mechanisms of Dimethoxymethane and Dimethoxyethane under Urban Tropo-
spheric Conditions
Atmos. Environ. 33 (1999) 2883
C. Sauer, I. Barnes, K.H. Becker, H. Geiger, T.J. Wallington, L.K. Christensen, J. Platz and O.J.
Nielsen
Atmospheric Chemistry of 1,3-Dioxolane; Kinetic, Mechanistic and Modelling Study of OH
Radical Initiated Oxidation
J. Phys. Chem. A 103 (1999) 5959
H. Geiger, T. Maurer and K.H. Becker
OH-Initiated Degradation Mechanism of 1,4-Dioxane in the Presence of NOX
Chem. Phys. Lett, (in press)
-159-
-------
Atmospheric Oxidation of Ethers Under High and Low NOX Conditions
J. Wenger1. E. Collins1 and H. Sidebottom1, S. Le Calve2, A. Mellouki2 G. Le Bras2 and K. Wirtz3
'Department of Chemistry, University College Dublin, Belfield, Dublin 4, Ireland
2Laboratoire de Combustion et Systemes Reactifs, CNRS, 45071 Orleans Cedex 2, France
3CEAM, Parque Technologico, Paterna, Valencia, Spain
Introduction
The photooxidation of diethyl ether (C2H5OC2H5) and diisopropyl ether, (CH3)2CHOCH(CH3)2, in
the presence of NOX was studied at the European Photoreactor (EUPHORE) in May 1998. The
main objective of the experiments was to validate the degradation mechanisms that are based on
data obtained from laboratory photoreactor studies. For diisopropyl ether, the experiments
provided product yield data that differed significantly from that obtained in conventional
photoreactor studies. In order to elucidate further mechanistic information, the photooxidation of
diisopropyl ether was studied again in July 1998 in the presence and absence of NOX.
Experimental
Photooxidation studies of diethyl ether (DEE) and diisopropyl ether (DIPE) were performed in
Chambers A and B of the EUPHORE facility. Both chambers had a volume of about 195,000
litres. For DEE, photooxidation was only performed in the presence of NOX, whilst for DIPE,
reactions were performed in the presence and absence of NOX using the photolysis of HONO and
HaOa as the source of hydroxyl radicals respectively. For the photoxidation of DIPE in the
presence of NOX, two experiments were performed;
(i) "normal NOx"conditions, where a single injection of ca. 150 ppb of NO was made at the
start of the experiment
(ii) "high NOX" conditions, where several injections of NO were made during the first few
hours of the experiment to ensure that NO was present throughout the whole reaction.
Measured volumes of gaseous and liquid samples were introduced into the chambers via a stream
of purified air and H2O2 (30% w/w) was added using a nebulizer. Two mixing fans, housed in the
chambers, were used to achieve rapid mixing of reactants. The loss of reactants and formation of
products was monitored using FTIR spectroscopy (Nicolet Magna 550 spectrometer). Infrared
spectra were obtained in situ by long-path absorption, with a resolution of 1 cm"1 and using
pathlengths of 326.8 m (Chamber A) and 553.5 m (Chamber B). Additional quantitative analysis
for experiments performed in Chamber B was provided by GC-PID and HPLC. The reactants and
products were quantified using calibrated reference spectra which were obtained by introducing
known volumes of materials into the chambers.
The concentration of reactants and products decreased through chemical processes and also due to
leakage from the chambers. The leak rate was determined daily by adding an unreactive tracer gas
-160-
-------
(SF6) and measuring its loss by FTIR spectroscopy. The temperature inside the chambers was
continuously monitored by thermocouples and the intensity of sunlight was measured using the
two J(NO2) radiometers located in Chamber A. Ozone and NOX concentrations were continuously
monitored using chemiluminescent analysers.
Results and Discussion
Diethvl ether (DEE)
The oxidation of DEE hi the presence of NOX at EUPHORE yielded ethyl formate (EF),
C2H5dCHO, ethyl acetate (EA), C2H5OC(O)CH3, and HCHO as products. Then- yields are given
in Table 1 together with the results obtained from similar experiments carried out in a ~ 50 L
reaction vessel in our laboratory and available literature data. A simplified mechanism for the
atmospheric oxidation of DEE in the presence of NOX is shown in Scheme 1. Studies of the OH
radical initiated oxidation of alkanes have shown that abstraction of a secondary hydrogen atom
occurs around 95% of the time (Kwok and Atkinson, 1995). For simplification purposes,
abstraction from the CH3 group has been omitted from the mechanism shown in Scheme 1.
Table 1. Summary of the product yields for the atmospheric oxidation of DEE.
Conditions
Air/NOx
N2/O2 (3%)/H2O2/NO
Air/CH3ONO/NO
Air/HONO/NOx
EFa,b
0.72 ± 0.03
0.79 ± 0.04
0.92 ± 0.06
0.66 ±0.14
EAa'b
0.12 ±0.01
0.11 ±0.04
<0.05
0.04 ± 0.03
CH3CHOb
not detected
not detected
<0.05
0.08 ± 0.02
Reference
This work (EUPHORE)
This work (Dublin)
Wallington and Japar, 1991
Eberhard et al., 1993
EF = C2H5OCHO ; EA = C2H5OC(O)CH3
errors are twice the standard deviation and represent precision only.
-161-
-------
Scheme 1. Simplified mechanism for the atmospheric oxidation of diethyl ether in the
presence of NOX.
O,
NO2'
O-
-H / O
A(C-C)
HO,
O
+ CH,
ethyl acetate
ethyl formate
formaldehyde
The sum of the yields of EF and EA are ca. 90% for each experiment. Since abstraction of a
secondary hydrogen is expected to occur 95% of the time, this suggests that any other possible
reactions of the alkoxy radical 1, such as isomerisation, are of little importance. However, the
residual IR spectrum contains weak bands in the C=O stretching region that may be due to
products formed as a result of H-atom abstraction from the -CH3 group (e.g. C2H5OCH2CHO).
A comparison of the results from the present study with those obtained in conventional laboratory
studies is provided in Table 1. In all cases EF was identified as the major product. The EF yield
calculated from this work is consistent with that determined in small reactor studies performed in
our laboratory and that of Eberhard et al, but somewhat lower than that determined by
Wallington and Japar. Eberhard et al. also identified EA and CH3CHO as primary products, but
Wallington and Japar did not identify either of these products and set upper limits of 0.05 for their
yields. The yield of EA obtained from this work is higher than that determined by Eberhard et al.
CHaCHO was not detected as a product in this work.
-162-
-------
DiisoproDvl Ether (PIPE)
The concentration-time profile for the reaction for DIPE under "normal NOX" conditions is shown
in Figure 1. Isopropyl acetate (IPAc), formaldehyde and acetone were all detected as products.
Acetone is a relatively weak infrared absorber and was difficult to measure using FTIR
spectroscopy. However, it was detected and quantified using GC-PID and HPLC. The rate of
formation of the products, as reflected in the shape of their concentration-time profiles, is
somewhat unusual. In the early stages of the experiment, whilst NO was present in relatively high
concentration, only isopropyl acetate and formaldehyde were observed as products. However, as
the concentration of NO decreased due to reactions with peroxy radicals, acetone began to form.
Finally, under very low NO concentrations, acetone appeared to be the sole product. Clearly, two
different regimes were involved in the reaction, which depended on the presence or absence of
NO. In order to investigate these two regimes in isolation, experiments were carried out under
"high NOX" conditions and also in the absence of NOX using H2O2 as the hydroxyl radical
precursor. The products identified in the three experiments performed on DIPE are summarised
in Table 2.
Table 2. Summary of the products and then- yields for the atmospheric oxidation of DIPE.
Conditions
Air/ "high NOX"
Air/H202
Air/ "normal NOX"
Air/CH3ONO/NO
Isopropyl acetate3
1.01 ±0.02
<0.03
0.89 ± 0.09 b
1.05 ± 0.06
Acetone3
<0.10
1.05±0.11
0.93 ±0. 13 c
Not detected
References
This work
This work
This work
Wallingtonefa/., 1993
a errors are twice the standard deviation and represent precision only.
b in the presence of NO (9:00 - 12:00)
0 in the absence of NO (13:00 - 15:50)
-163-
-------
Scheme 2.
Simplified mechanism for the atmospheric oxidation of diisopropyl ether in
the presence of NOX.
OH
H20, 02
NO
2' V
NO,
A(C-C)
CH3- +
O
isopropyl acetate
formaldehyde
-164-
-------
(Aqdd)
3
I
•£S
I
(Aqdd)
-165-
-------
For the experiment performed under "high NOX" conditions, only trace amounts of acetone were
detected and IP Ac therefore accounted for virtually all of the DIPE lost due to reaction with
hydroxyl radicals. Similarly during the initial hours of the "normal NOX" experiment, the initial
yield of IPAc was also high (0.89 ± 0.09). These data agree with the findings of the only
previously published work on DIPE photooxidation (Wallington et al. 1993) and thus confirm the
proposed mechanism, shown in Scheme 2. Hydroxyl radical attack is expected to occur about
80% of the time at the >CH- group (Kwok and Atkinson, 1995). As a result, H-atom abstraction
from the -CHs group is omitted from Scheme 2 for simplification purposes.
For the experiment performed in the absence of NOX only trace amounts of IPAc were detected,
whilst the yield of acetone was 1.05 ± 0.11. Similarly during the final hours of the "normal NOX"
experiment, the yield of acetone was also high (0.93 ± 0.13). Two possible decomposition routes
that could lead to the formation of acetone are shown in Scheme 3. One possible route is through
C-O bond cleavage of the alkoxy radical yielding acetone and another alkoxy radical, which will
react with Oa to form a second acetone molecule. The other route is via a hydroperoxy-peroxy
radical reaction to form the hydroperoxide i-PrOC(OOH)(CH3)2. This hydroperoxide is similar in
structure to 1-hydroxyalkyl hydroperoxides (HOCH(OOH)R) which are known to decompose to
H2O2 and RCHO, (Kurth et al., 1991). By analogy, i-PrOC(OOH)(CH3)2 could decompose to
give isopropyl hydroperoxide and acetone.
o.
HO,
O,H
o,
OOH +
acetone acetone
Scheme 3. Possible decomposition routes leading to the formation of acetone.
The hydroperoxide and alkoxy radical channels are expected to produce one and two molecules of
acetone respectively for every molecule of DIPE consumed. Since the yield of acetone in the
absence of NOX was around 100% then it seems likely that the hydroperoxide channel was
-166-
-------
responsible for acetone formation. The HO2 radicals required for this process to occur were
formed by the reaction of H2O2 with OH radicals;
H2O2 + OH -> H2O + HO2
The hydroperoxide channel also produces isopropyl hydroperoxide. This species was not detected
as a product by FTIR spectroscopy or GC-PID. However, it is possible that its absorption
features were obscured by the strong bands due to H2O2 and that it decomposed on the separating
column. In addition, isopropyl hydroperoxide could react with OH or undergo photolysis. In the
latter case this would lead to the formation of another molecule of acetone. Further experimental
work is planned in which HPLC will be employed for the detection of isopropyl hydroperoxide.
This will therefore allow us to confirm or discount the proposed mechanism for its formation.
Acknowledgements
Financial support for this work was provided by the European Commission and the National
funding agencies in Ireland and France. The authors would like to thank Manuel Pons (CEAM)
and Lars Thuener (Wuppertal) for their assistance and valuable discussions during the course of
this work.
References
Kwok E.S.C. and R. Atkinson; Estimation of hydroxyl radical reaction-rate constants for gas-phase organic-
compounds using a structure-reactivity relationship, Atmos. Environ. 29 (1995) 1685-1695.
Wallington T.J., J.M. Andino, A.R. Potts, S.J. Rudy, W.O. Siegl, Z. Zhang, M.J. Kurylo and R.E. Huie;
Atmospheric chemistry of automotive fuel additives - diisopropyl ether, Environ. Sci. Technol., 27 (1993) 98-104.
Kurth H.H., S. Gab, W.V. Turner and A. Kettrup; A high-performance liquid-chromatography system with an
immobilized enzyme reactor for detection of hydrophilic organic peroxides, Analytical Chemistry 63 (1991) 2586-
2589.
Eberhard J., C. Muller, D.W. Stacker and J.A. Kerr; The photooxidation of diethyl ether in smog chamber
experiments simulating tropospheric conditions - product studies and proposed mechanism, Int. J. Chem. Kinet., 25
(1993) 639-649.
Kwok E.S.C. and R. Atkinson; Estimation of hydroxyl radical reaction-rate constants for gas-phase organic-
compounds using a structure-reactivity relationship, Atmos. Environ. 29 (1995) 1685-1695.
Wallington T.J and S.M. Japar; Atmospheric chemistry of diethyl ether and ethyl tert-butyl ether, Environ. Sci.
Technol., 25 (1991) 410-415.
Wallington T.J., J.M. Andino, A.R. Potts, S.J. Rudy, W.O. Siegl, Z. Zhang, M.J. Kurylo and R.E. Huie;
Atmospheric chemistry of automotive fuel additives - diisopropyl ether, Environ. Sci. Technol., 27 (1993) 98-104.
-167-
-------
Atmospheric Oxidation Mechanisms of Unsaturated Oxygenated VOCs
R. Thevenet, G. Thiault, E. V6sine, G. Laverdet, A. Mellouki, G. Le Bras
LCSR-CNRS-1C, Avenue de la recherche scientifique 45071, Orleans, — France
Introduction
Unsaturated oxygenated volatile organic compounds are an important class of
compounds which are used in different industries. These compounds are emitted to the
atmosphere where they are oxidised to produce ozone and other secondary pollutants in urban
and rural areas. The main tropospheric degradation processes for this type of compounds are
reactions with OH radicals, ozone and NO3 radicals. Kinetic and mechanistic data are needed
to better understand the atmospheric degradation mechanisms of these compounds and
therefore to assess their impact on air quality.
In this work, we have studied the O3 and OH-initiated degradation of Ethyl Vinyl
Ether (EVE, C2H5OCH=CH2) and Methyl Methacrylate (MMAC, CH2=C(CH3)C(O)OCH3).
The absolute rate constants for reactions of O3 with EVE and MMAC and that of OH radicals
with EVE have been measured. The oxidation mechanisms of C2HsOCH=CH2 and
CH2=C(CH3)C(O)OCH3 initiated by OH and O3 have been investigated in our laboratory and
atEUPHORE.
Experimental and results
Kinetic studies
OH rate constant measurement:
The pulsed laser photolysis - laser induced fluorescence (PLP-LIF) technique Was
used to measure the rate constant of the OH + EVE reaction. OH radicals were produced by
photolysis of H2O2 at "k - 248 nm. The OH temporal concentration profiles were obtained by
pulsed laser induced fluorescence. Reactions were studied under pseudo-first order conditions
with the OH concentration much lower than that of EVE ([EVE]0 » [OH]0). Under these
conditions, the OH concentration time profiles followed the pseudo-first-order rate law:
[OH]t=[OH]0e
-k't
where k' = k [EVE] + k'0
where k is the rate coefficient for the reaction of OH with EVE. The decay rate, k'0, is the
first-order OH decay rate in the absence of the ether; it is the sum of the reaction rate of OH
with its precursor (H2O2), and the diffusion rate of OH out of the detection zone.
Rate constants were determined over the temperature and the pressure ranges 230-372
K and 30-320 Torr, respectively. The obtained values were (in cm3 molecule"1 s"1) k2gg = (6.8
±0.7)xlO"u and k230-372 = (1.55±0.25)xlO"nexp[(445±13)/T]. The measured rate constants are
shown plotted in Arrhenius form in figure 1. This latter plot shows a negative temperature
dependence of the rate constant which is consistent with a mechanism proceeding mainly by
addition of OH radicals to the double bond of the ether.
-168-
-------
Ozone rate constant measurements ;
Although the major application of EUPHORE is for mechanistic studies, it was also
used to determine the rate constant for the reaction of ozone with EVE and MMAC. For the
reaction of Os with EVE, the initial concentrations of ozone and EVE were 207 ppb (5.2xl012
molecule cm"3) and 485 ppb (1.2xl013 molecule cm"3), respectively. The other reaction was
studied using the following initial concentrations [MMAC] = 1.37 ppm and [O3] = 279 ppb.
The reactions were studied in the presence of large excess of cyclohexane (> 50 ppm,
1.25x10 molecule cm ) to scavenge the OH radicals which may be produced from the
reaction of O3 with EVE or MMAC. Ozone was measured in real time by an ozone analyser.
The concentrations of EVE and MMAC were measured using FTIR and gas chromatography
techniques.
From the fitting to the obtained concentration profiles of O3 and the two unsaturated
oxygenated VOCs we determined the rate constant values at 298 K:
k(O3 + C2H5OCH=CH2) = (2.0 ± 0.4)xlO"16 cm3 molecule"1 s"1.
k(O3 + CH2=C(CH3)C(O)OCH3) = (8.7 ± 0.8)xlO"18 cm3 molecule"1 s"1.
As an example of the obtained results, figure 2 shows the experimental and fitted data
for the reaction of O3 with EVE.
Product studies
O?i reaction;
One experiment was performed for each reaction and that was at the EUPHORE
facility. The initial experimental conditions were the same as those used for the kinetic
studies. In the reaction of O3 (207 ppb) with EVE (485 ppb) (in the presence of excess of
cyclohexane, 50 ppm), the main products detected were ethyl formate (85 %) and
formaldehyde (20 %). The profiles of the reactants and products are shown in figure 3. In the
reaction of O3 (279 ppb) with MMAC (1.37 ppm)) (in the presence of excess of cyclohexane,
57 ppm), the main products detected were methyl pyruvate, CH3C(O)C(O)OCH3 (45 %) and
formaldehyde (50 %).
The observd products are the expected primary products resulting from the well
established ozonolysis mechanism:
O3 + RiR2C=CR3X ->
-------
OH reactions;
I/A first series of experiments was performed, in our laboratory, at atmospheric pressure and
298 ± 3 K, in a small Teflon bag (volume » 100 L) surrounded by UV lamps (254 nm).
Quantitative analysis were carried out using gas chromatography FID/MS and FTIR. The
experiments were performed in synthetic air, using H2O2 as the OH radical source, in the
presence and absence of NO. The main product observed, in the presence and absence of NO,
were ethyl formate ((92 ± 7) % and (83 ± 7) %, respectively) for EVE, and methyl pyruvate
(=67%) for MMAC.
2/ One experiment was performed at EUPHORE for each compound. EVE (650 ppb) and NO
(86 ppb) or MMAC (1.18 ppm) and NO (120 ppb) were introduced into the chamber in a
stream of dry air. Figure 4 shows the concentration-time profiles measured during the
photooxidation of C2H5OCH=CH2. The analysis was carried out using FTIR,
GC/FDD/PE)/ECD, NOX, O3, and CO analysers. The main observed products during the
photooxidation of EVE were ethyl formate, formaldehyde, ozone, and CO with the following
yields: ethyl formate (64 % (FID), 70 % (PID), and 79 % (FTIR)) and formaldehyde (42 %
(FTIR)). The main products observed during the photooxidation of MMAC were methyl
pyruvate, formaldehyde, ozone, and CO. The above reported product yields need, however, to
be refined to take into account the contribution of ozonolysis which was found to become
significant when sufficient ozone concentration was produced in the chamber.
Nevertheless, the obtained data, in our laboratory or at EUPHORE, show that the main
products of the OH-initiated oxidation of C2H5OCH=CH2 is C2H5OC(O)H and that of
CH2=C(CH3)C(O)OCH3 is CH3C(O)C(O)OCH3. These results are consistent with a
mechanism proceeding essentially through addition of OH radical to the double bound of the
unsaturated VOCs (already mentioned in the OH kinetic section for EVE).
Atmospheric implications
Using the rate constants for the reactions of EVE and MMAC with OH (6.8xlO'u and
2.6xlO"n respectively) and with O3 (2xlO~16 and 8.7xlO"18, respectively) and the typical
atmospheric concentrations of OH (106) and O3 (1.3xl012) molecule cm"3, the calculated
lifetimes of these two VOCs, with respect to their reactions with OH and O3, are for EVE
around 4 hours and 1 hour, and for MMAC around 11 hours and 25 hours, respectively. This
shows that these unsaturated oxygenated VOCs will be oxidised very rapidly by reaction with
OH or O3 and that will lead to the formation of carbonyls such as HCHO, C2H5OCHO and
CH3C(O)C(O)OCH3 as reported in the present work. The atmospheric impact of EVE and
MMAC is not only defined by their persistence (determined by their oxidation rate by OH or
Os) but also by the fate of its oxidation products.
References
E. Grosjean and D. Grosjean, The reactions of unsaturated aliphatic oxygenates with ozone,
J. Atmos. Chemistry, 32, 205-232, 1999
Acknowledgments: Dr K Wirtz and his co-workers for technical assistance at EUPHORE,
the European Commission, the Elf Company, for financial support.
-170-
-------
s
o
o
o
CO
1E-10 -
1E-11
•^ I
2.SE-3 3.0E-3
I r I ' I
3.5E-3 4.0E-3 4.5E-3
1/T (K'1)
Fig. 1: Plot of k(OH + C2H5OCH=CH2) vs 1/T in the temperature range (230 - 372) K.
[O3] experimental
[O3] titled
[EVE] experimental
[EVE] fitted
k= (2.0 +/- 0.4J.10•" cm'.molecule '.s
15 20
tlma (mln)
Fig. 2 : Reaction of C2H5OCH=CH2 with O3: Experimental (obtained at EUPHORE) and
fitted concentration-time profiles (EEVE]0 = 485 ppb, [O3]0 = 207 ppb).
-171-
-------
12:07:12 12:1424 122133 1228:43 12:33:00 12:43:12 12:50:24 12:57:33 13:04:43 13:12:00
TIme/hh:mm:33
Fig. 3: Concentration-time profiles measured during ozonolysis of
EUPHORE, ([C2H5OCH=CH2]o - 485 ppb, [O3]0 - 207 ppb, [cyclohexane]0 - 50 ppm)
at
EVE FID (ppb]
—*—EViPID(ppb)
I EVE FTIR (ppb)
- C- Ethyl (brrato FTIR
• EFF10(ppb)
Fomul
-------
Degradation Mechanisms for the Troposphere Oxidation of Chlorocarbons
M. Manning1, S. Le Calve2, J. Wenger1, J. Treacy1, G. Le Bras2 and H. Sidebottom1
'Department of Chemistry, University College Dublin, Belfield, Dublin 4, Irleand
2 Laboratoire de Combustion et Systemes Reactifs, CNRS, 45071 Orleans, Cedex 2 France
The distribution and fate of volatile organochlorine compounds has recently attracted considerable
attention. In particular, it has been suggested that phytotoxic chloroacetic acids found in a number of
environmental compartments arise from the atmospheric oxidation of chlorinate industrial solvents such as
trichloroethene, tetrachloroethene and 1,1,1-trichloroethane. A number of investigations have been
previously carried out to elucidate the OH radcial and Cl atom initiated oxidation of these compounds
which suggest that chloroacetyl chlorides may be formed in these reactions under certain experimental
conditions. The main objective of this work was to validate the proposed mechanisms through
potoxidation studies carried out in the presence and absence of NOX at the European Photoreactor
EUPHORE. The contributions of trichloroethene tetrachloroethene, and 1,11-trichloroethane to the
formation of chloroacetic acids in the atmosphere have been estimated and compared to the observed
levels in the various geographical regions.
-173-
-------
Session II
Heterogenous Chemistry and Modeling
Session Chair
Georges Le Bras
-------
Auxiliary Mechanisms
(Wall Models)
for UNC Outdoor Chamber
Harvey Jeffries, Kenneth Sexton, and Zac Adelman
Department of Environmental Sciences and Engineering,UNC-CH
Abstract
It has been recognized for more than 30 years that the walls of "smog cham-
bers" provided sites for heterogeneous reactions that influenced the outcome of the
homogeneous gas phase reactions. In the early 1980's several efforts were made
to elucidate the origins of these effects, with most attention focused on the hetero-
geneous formation of nitrous acid (HONO) which was subsequently released into
the gas phase. While these efforts effectively demonstrated that most likely HONO
was being formed, actual mechanisms and process rates for this phenomena have
not been developed for chambers. Instead a small set of un-real or parameterized
reactions have been used to approximate the HONO formation based upon "char-
acterization experiments." These experiments have included the simplest type of
experiments, i.e., NOX/CO, NOX/CH4, and NOx/n-butane experiments. The basic con-
ceptual model has been that the gas phase chemistries of these simple systems are
so well understood, that any failure of a reaction mechanism simulation to repro-
duce the observations from these experiments was due to chamber wall processes
or chamber-dependent initial conditions. These chamber-dependent conditions in-
cluded unintended creation of initial HONO due to injection of the initial NOX from
high concentration sources. Such chamber-dependent reactions and inputs have
often been expressed as "Auxiliary Mechanisms," one for each chamber to be sim-
ulated. When testing or evaluating a reaction mechanism in a given chamber, a
chamber-dependent mechanism is combined with a "Principle (or Core) Mecha-
nism," which is asserted to be chamber-independent. Obviously mistakes or mis-
representations in the auxiliary mechanism can introduce compensating errors in
the principle or core mechanism.
-175-
-------
Measurements in the UNC outdoor chamber and laboratory work in Europe
have shown that the simple ideas used in our Auxiliary Mechanisms are not valid. In
most modeling evaluations in the US, it has been assumed that some initial amount
of HONO (2 to 5 ppb) was created in the chambers due to injection, and that during
the experiment, NO2 was converted to HONO at a rate proportional to the product of
the radiation intensity and a "radical source" scale factor.
The chamber HONO measurements show that initial HONO is very low in the
UNC chambers (< 0.5 ppb) and that its mixing ratio increases during the experi-
ment, reaching a maximum sometime between NO-to-NO2 cross-over and just after
NOa maximum.
Recent laboratory work at Wuppertal, Germany has shown that HONO forma-
tion in systems with NOa and water vapor present is initially first order in NOa gas
phase concentration and that the yield is essentially 0.5. This is consistent with NO2
+ NOa + HaO —> HONO + HNOs, all in aqueous solution, where the rate determining
step is NOa partitioning to the aqueous phase. Furthermore, they have shown that as
the reaction proceeds, the HONO production decreases and that N2O is created in-
stead of HONO. The mechanistic explanation offered was that the increasing acidity
of the aqueous solution leads to the creation of the NO+ ion, which then oxidizes
HONO to N2O before HONO can partition out of the aqueous solution. The concept
that the "wall source" of HONO might be shut off due to a high acidity provided a
major shift in our ideas of how an auxiliary mechanism must be formulated. We
have created several new auxiliary mechanisms based on these ideas.
In simulating NOX/CO experiments, it became clear that such experiments were
far from ideal in "characterizing" the chamber wall source. This is because the only
gas-phase sink for NOX in such experiments is the formation of HONO via OH+ NO,
or the formation of HNO3, either via 'OH+ NO2 or via O3+ NO2 leading to N2O5,
which hydrolyzes to HNOa. At the beginning of the experiment, HONO is the only
source of new radicals (and the source of the HONO is NO2 reacting with wall wa-
ter), but once Oa formation starts, the photolysis of O$ becomes the major source of
new 'OH, and the formation of N2O5 provides an additional strong source of HNO3.
Analysis of the simulation mechanism shows that by the time Os formation starts,
a significant amount of HNOs has already been formed and much of this has parti-
tioned to our chamber walls. It appears that shortly after Os formation begins, the
wall source of HONO must be quenched or an excess of OH radicals is introduced
into the system. Three sets of HONO measurements in NOX/CO experiments indicate
that most HONO production stops just after Os formation occurs.
While we were able to formulate a relative simple set of gas to aqueous phase
partitioning reactions and aqueous-phase reactions to create HONO which gave ex-
cellent simulations of the initial NO oxidation rates and NO2 production rates, with-
-176-
-------
out a full acid-base, redox reaction set in the aqueous phase, we could not satis-
factorily simulate the rapid termination of HONO production, thus, we consistently
overpredicted the 63 formed in these systems.
The next step is to create a fully explicit aqueous phase, weak and strong
solution chemistry mechanism for nitrogen oxides and to couple this chemistry to
the usual gas-phase chemistry. This system will provide more testable hypotheses
in chamber experiments which can then lead to a refinement in the wall-mediated
mechanisms. Additional measurements have already been suggested, for example,
we intend to measure N2O levels in new NOX/CO experiments, as well as trying to
assess the pH of the wall water film or to measure conductivity of the Teflon film
inside the chamber.
One insight already gained is that the use a particular type of chamber exper-
iment to derive a "chamber characterization" is clearly not a valid concept, as the
actual time series of HNO3 formation would have a strong effect on the HONO pro-
duction rate from the walls. In subsequent experiments, a different time series of
HN03 would result in a different rate of HONO production than that represented by
the "parameterized" un-real reactions used in the simulations.
On the other hand, an understanding of nitrogen wall chemistry processes
could lead to a "chamber cleaning" procedure that might result in a much smaller
experiment to experiment carry over effect.
In summary we offer two puns:
Acid not what your walls do to your gases,
Acid what your gases do to your walls.
and
Too acid, or not too acid, that is the question.
-177-
-------
MEASUREMENT AND MODELING OF NOX OFFGASING
FROM FEP TEFLON CHAMBERS
William P. L. Carter
Air Pollution Research Center and College of Engineering Center for
Environmental Research and Technology, University of California, Riverside,
CA 92521
A. Introduction
The College of Engineering, Center for Environmental Research and Technology
(CE-CERT) at the University of California (UCR) is undertaking a new project to develop a
"Next Generation" environmental chamber for evaluating chemical mechanisms under lower
pollutant conditions than has previously been possible (Carter et al., 1999). A major objective of
this project is to generate mechanism evaluation data under very low NOX conditions, where the
products and mechanisms for many VOC oxidations are expected to differ from those under the
higher NOX conditions characteristic of more polluted urban areas and the current environmental
chamber data base.
One major obstacle that will limit the extent to which environmental chambers can
generate useful mechanism evaluation data under low NOX conditions is offgasing of NOX from
the walls of environmental chambers. Evidence for NOX offgasing comes from the formation of
63 in "pure air" irradiations, which cannot occur to any significant extent in the absence of NOX,
the observation of PAN formation in acetaldehyde - air irradiations, and other data (see Carter
and Lurmann, 1991, and references therein.) In addition, NOX offgasing in the form of HONO is
considered to be the most likely explanation for the "unknown chamber radical source", which
also significantly complicates mechanism evaluation using environmental chamber data (Carter et
al, 1982,1995a,b)
In this paper, we give a brief summary of preliminary analysis of previous and new data
obtained concerning NOX offgasing from environmental chamber surfaces. The emphasis will be
on NOX offgasing from indoor chambers constructed of FEP Teflon film, since this is currently
the preferred material for constructing chamber walls because of its relative inertness and good
light transmission properties. However, data from a chamber with a different type of surface will
also be presented, for comparison.
B. Methods
1. Characterization of NOX Offgasing and the Chamber Radical Source
NOX offgasing in environmental chamber systems can be evaluated using several
different methods. The most straightforward is simply measuring NOX buildup in the dark in a
chamber containing no NOX initially. There has been limited data on this in the past because of
limitations in sensitivities of NOX analyzers. However, state-of-the-art TECO NOX analyzers can
monitor NO and "NOx" species at sub-ppb sensitivity, and research instruments can obtain data
at lower concentrations.
-178-
-------
Pure air irradiations can provide a more sensitive measurement of NOX offgasing because
the amount of O3 formed is highly sensitive to the amount of NO^ offgasing occurring. The NOX
offgasing rate in the chamber can then be estimated by adjusting the offgasing rate so that model
simulations fit the chamber data. Unfortunately, O3 formation in pure air experiments is also
sensitive to background levels of CO and reactive VOCs, which are uncertain in many cases. If
the model does not have the correct amount of such VOCs (which may not all be detected), then
the amount of NOX offgasing derived by modeling will not be correct.
Modeling O3 formation in acetaldehyde - air or CO - air irradiations is provides an
equally sensitive means to assess NOX offgasing effects in environmental chamber systems,
without the confounding effects of uncertainties in the background VOC levels. O3 formation in
such experiments is equally sensitive to NOX offgasing as it is in pure air runs, but the presence of
known amounts of reactive VOCs means that uncertainties in background VOCs is not important
in affecting modeling results. Thus, these data provide less ambiguous indirect determination of
NOX offgasing. Acetaldehyde - air runs have the additional advantage of forming PAN, which
provides direct evidence for the presence of NOX in the system, and additional data for model
evaluation.
Because of the possibility that NOX offgasing is related to the chamber radical source,
measurements of the magnitude of the chamber radical source is also of interest. This can be
measured by modeling NOX - air irradiations with tracer compounds present to measure radical
levels (Carter et al, 1982), and also by modeling n-butane - NOX environmental chamber
experiments, whose results are highly sensitive to this effect. Recently, we found that modeling
n-butane results give much more consistent and reliable measurements of the chamber radical
source than tracer - NOX experiments (Carter et al, 1995b) in reactors constructed of Teflon film,
so this is the preferred type of characterization experiment for this purpose.
2. Environmental Chambers Examined
Results of experiments relevant to characterizing NOX offgasing and the chamber radical
source were examined in four different environmental chamber systems, which represent two
types of surfaces and different levels of "cleaning" between experiments and chamber history in
terms of NOX exposure. These are summarized on Table 1, and briefly discussed further below.
New Teflon Reactors. As indicated on the table, the chamber employed in this work
represents our current best effort to minimize NOX offgasing effects. The reactor was constructed
of 2 mil FEP Teflon film that was wiped with purified water and dried before construction. This
procedure was used to remove any atmospheric NOX or nitrates that may have deposited on the
walls. It is tube shaped and was constructed by heat-sealing 2 ~5' x 12' sheets together. It was
placed inside a larger tube shaped FEP reactor that was flushed with purified air. The purified air
was treated with Purafill to remove NO to below 0.05 and NOX species to below -0.2 ppb. The
detection limit of our instrument is estmated to be 0.05 ppt as NO. The inner reactor was never
knowingly exposed to NOX, other than what was offgased in the offgasing experiments.
-179-
-------
Table 1. Summary of major relevant characteristics of environmental chamber systems
examined.
Chamber
Surface
Cleaning Method
RH
This Work
New FTP Teflon film.
Wiped with H2O before
construction
TVA Chamber Well-used FEP Teflon
panels
CE-CERT DTC Well-used FEP Teflon
film
SAPRC EC
Teflon-coated aluminum
cylinder with quartz end
windows.
NOX never injected into reactor. Emptied
and filled several times between
experiments. Reactor located inside 2nd
Teflon bag flushed with clean air to
prevent NOX permeation from outside.
Flushed with lights on for two days
between runs. Typically exposed to no
more than ~50 ppb NOX.
Flushed between runs. Typically exposed
to 200-500 ppb NOX.
Evacuated between experiments.
Dry
-20%
Dry
-50%
Ozone and CO were monitored using conventional laboratory instruments used for this
purpose. NO and "NOX" species were monitored using a TECO model 42CY NO -NOX analyzer.
Note that the "NOX" measurement (also referred to as "NOC" for NO - converter) works by
converting NOX species to NO using a heated Molybdenum converter, which efficiently converts
HONO and HNO3 as well as NO2. For these experiments, the converter was located next to the
chamber with minimum sampling lines, to minimize loss of HNOs on sampling lines coming to
the converter. Because it was found that the converter had a "memory" effect when measuring
low levels of NOX, the NOX analyzer was kept on pure air when it was not sampling the reactor.
This was also used to establish the "zero" level, which is important when monitoring species at
such low levels. To minimize loss of volume to permit multi-day experiments, sampling was done
only intermittently. When monitoring NO in the presence of 63 in the irradiations, a correction
was made for loss of NO with O3 in the dark in the sample lines.
The experiments carried out in this reactor consisted of pure air irradiations and CO - air
irradiations. In each case the irradiation followed a period where the pure air or CO - air mixture
sat in the dark to permit offgasing to occur. The reactor was emptied and filled at least three times
between each experiment, which was sufficient to reduce the NOX, Os, and CO from the previous
run to be diluted to below detectable levels.
TVA Chamber. The TVA chamber is described by Simonaitis and co-workers
(Simonaitis and Bailey, 1995; Bailey et. al, 1996). As with this project, the objective of the TVA
chamber project was to conduct mechanism evaluation chamber data at very low VOC and NOX
levels. Thus, this chamber represents the current state of the art in this regard, though it is
unfortunately no longer operational. It is a relatively large indoor chamber with walls made from
"Teflon panels". When it was running, it was used primarily for experiments with -25-100 ppb
NOX, and long flushing periods were used between runs to clean the chamber. The above-
referenced reports, which can be obtained from the EPA, can be consulted for details on Hie
experimental procedures and data obtained.
-180-
-------
Unfortunately, the only experiments conducted in this chamber that could be used to
assess NOX offgasing were two acetaldehyde - air runs. Several n-butane - NOX experiments were
carried out in this chamber, and a few of these runs were modeled to estimate the magnitude of
the chamber radical source.
CE-CERT PTC. The CE-CERT Dividable Teflon Chamber (DTC) consists of two
~5000-liter reaction bags constructed of 2 mil FEP Teflon film. As discussed in various reports
(see for example Carter et al, 1997 and other CE-CERT reports that can be downloaded from
http://cert.ucr.edu/~carter/bycarter.htm), this chamber has been extensively used for assessing
ozone reactivities of various VOCs, which involves conducting multiple irradiations with NOX
levels up to ~0.5 ppm. Thus, this represents a chamber with similar surfaces as the TVA chamber
and the "clean" reactor used in this work, but with no special procedures for conducting low NOX
experiments. Pure air, acetaldehyde - air and n-butane - NOX characterization runs are periodically
carried out to assess NOX offgasing and other chamber effects.
SAPRC EC. The SAPRC Evacuable Chamber (EC) consists of a -6000-liter aluminum
cylinder with the interior surface coated with FEP Teflon with quartz end windows. The chamber
and procedures employed when operating it, are described by Carter et al (1995a). Experiments
are carried out using various NOX levels, and the chamber is evacuated between experiments. A
limited number of acetaldehyde - air experiments were conducted in this chamber to determine
NOX offgasing rates, and there is a large data base of radical source characterization runs (Carter
et al, 1995a).
3. Wall Model Employed
The physical and chemical processes responsible for NOX offgasing and radical source
effects in environmental chamber experiments are not currently understood, though complex wall
models based on aqueous chemistry and other considerations have been developed (e.g., see
Jeffries et al., 1999). For the preliminary survey in this work, we use a highly simplified wall
model to account for both of these effects. In particular, both NOX offgasing and the radical
source is represented by a continuous HONO input at a zero-order, time-independent rate that is
adjusted to fit the chamber data. Initial NO and/or HONO, which might be formed by dark
offgasing prior to the irradiation, is also added as needed to fit the data.
The wall model also must represent O$ wall losses that, though relatively slow in Teflon
film reactors, can be non-negligible in affecting Oa in multi-day experiments. This is determined
by Oi dark decay experiments. Since Os dark decay experiments have not yet been employed in
the new Teflon reactors discussed in this work, the Os decay rates measured in other Teflon
reactors under dry conditions were assumed to be applicable.
Since the "NOX" instrument (at least in the dry runs) may be responding to nitric acid,
wall loss rates assumed for HNOs may affect model simulations of NOX data. For modeling these
runs, relatively high HNOs loss rates of ~12%/hour were assumed, based on limited
measurements in the SAPRC EC. The applicability of this to the Teflon reactors is highly
uncertain, so modeling of the "NOX" data should be considered to be unreliable.
-181-
-------
C. Results and Discussion
1. NOX Dark Offgasing Rates in the New Reactor
The results of four experiments carried out when NOX offgasing was measured in the dark
for up to one day are shown in Figure 1. It can be seen that there is no significant NO offgasing,
but that there is a non-negligible buildup of "NOX" species, with an average input rate of 0.02
ppb/hour. It does not appear to decrease significantly with time (the run numbers on the Figure
caption refer to the order the runs were carried out). Note that for all these experiments the reactor
was contained in an outer reactor that was continuously flushed with pure air, and sampling the
air in the outer reactor indicted no detectable NOX species. Therefore, these results cannot be
attributed to permeation through the reactor walls.
Figure 1.
Measurements of dark NOX offgasing in the new Teflon reactor
- NOx - Run 1
- NOx - Run 2
-NOx-Run 3
-NOx-Run 4
—<3— NO - Run 1
—B— NO - Run 2
—©—• NO - Run 3
- Run 4
1.0 n
-0.2 J
)0
0.25
0.50
=M~7li....
0.75
— fcJ>B5
1.00
Days
Note that offgasing of either NO2, HONO, or HNOs could be causing the measured
"NOX" buildup. NO2 and HNOs-specific analysis methods are being developed in our laboratory
but are not yet available for these experiments. Model simulations of the pure air experiments,
discussed below, were conducted using various alternative assumptions in this regard.
2. Pure Air and CO - Air Irradiations in the New Reactor
Figure 2 and Figure 3 show results of a pure air and two CO - air experiments carried out
in the new reactor. In all three cases, significant ozone formation was observed, which could only
be modeled if non-negligible NOX offgasing is assumed. The HONO input rate used in the
chamber effects model was adjusted to fit the data, and the rates that fit the O3 data were 0.1
ppb/hour for the pure air run and 0.02 ppb/hour for the CO - air run. Note that the HONO input
rate determination in the pure air run is much more uncertain than in the CO - air run. To fit the
data for the pure air run, we also assumed a background VOC level equivalent to 1 ppm, though
this is highly uncertain.
-182-
-------
Figure 2. Experimental and model simulation results of a pure air irradiation in the new
Teflon reactor. The data were simulated using an assumed background VOC
level equivalent to 1 ppm CO, initial MONO or NO2 of 0.8 ppb, and a constant
HONO input rate of 0.1 ppb/hour.
\
140
120-
100-
80-
60-
40-
20-
0
1000 2000 3000 4000
I Experimental
Model (Initial HONO)
Model (Initial NO2)
2.0-
1.0-
0.5 |
nn -
NOC-NO
I ..t
$
S
1.00-
0.75-
0.50-
0.25-
0.00
1000 2000 3000 4000 -0.2
NOC - NO 0.8 -
0.6-
0.4-
0.2-
0.0
5 10
Minutes
15 -0.2J
1
-------
start of the run means that much if not most of the initial NOX is in the form of NO2 or HONO,
whose photolysis would result in the formation of NO. HNOs is relatively unreactive, and its
presence would not cause the observed formation of NO.
Two model simulations are shown on each figure, one based on assuming that the initial
measured NOX is all in the form of HONO, and the other assuming it is all in the form of NO2-
Note that both models give essentially the same results except for the first ~20 minutes of the
irradiation. If the NOX is assumed to be in the form of NO2, the NO formation occurs very rapidly,
and there is a corresponding rapid decrease of NOX-NO. The formation rate of NO and loss rate of
NOX-NO is slower if initial HONO is assumed, because the photolysis rate of HONO is slower
than that of NO2. The initial NOX-NO data for both experiments and the NO data for the pure air
run are most consistent with the assumption that all the initial NOX is HONO, and this is also
consistent with the assumption that the radical source is due to HONO offgasing. On the other
hand, the maximum NO concentration in the CO - air run is better fit by the model assuming
initial NO2.
In both experiments the total NOX levels were observed to increase during the first day of
the run, and then eventually level off at ~2 ppb. The model predicts that most of the "NOX" at the
end of the experiment is in the form of NO2, though this is based on assuming relatively rapid
wall loss rates of HNOs. The model tends to underpredict the final NOX levels, suggesting that
HNOs may be persisting in the gas phase longer than assumed in this model.
3. TVA Chamber Experiments
The experimental and model simulation results of the two TVA acetaldehyde - air
experiments are shown on Figure 4 and Figure 5. The model simulations shown each use HONO
input rates in the wall model that were adjusted so the simulation would fit the ozone data. In
both cases the model simulations with the NOX input rates adjusted to fit the ozone data tended to
slightly underpredict the PAN formation, though probably not by much more than the
experimental uncertainty. In the case of the 9/22/95 experiment, the model that fit the Os data
tended to underpredict the buildup of NOX during the run, though in the 6/11/96 run both the NOX
and Os were reasonably well fit by the same model. Unlike the pure air and CO - air runs with the
new reactor, discussed above, there was no increase in NO at the start of the experiments.
However, the irradiation in these TVA experiments probably started soon after the chamber was
filled with pure air, with insufficient time for significant NOX offgasing to occur.
Figure 6 shows the experimental and model simulation results of a TVA n-butane - NOX
experiment. Good fits to the data were obtained in the first 4 hours of the experiment when the
model used a radical input rate equivalent to 0.9 ppb HONO per hour. This is significantly higher
than the NOX input rates that fit the acetaldehyde - air experiments.
-184-
-------
Figure 4. Experimental and model simulation results of the TVA acetaldehyde - air run
9/22/95. The model simulation uses a HONO input rate of 0.09 ppb/hour.
PAN
120 180 240 300
TImt, mln
Figure 5. Experimental and model simulation results of the TVA acetaldehyde - air run
6/11/96. The model simulation used a HONO input rate of 0.16 ppb^our.
120 180 240 300
Time, mln
-185-
-------
Figure 6. Experimental and model simulation results of TVA n-butane - NOX experiment
6/19/95. The model simulation used a HONQ input rate of 0.9 ppb/hour.
n-Butane
60 120 180 240 300 360 420
0 60 120 180 240 300 360 420
Time (minutes)
4. DTC Chamber Experiments
Experimental and model simulation results of the DTC acetaldehyde - air experiment that
was carried out after the reactors had been extensively used for reactivity experiments are shown
on Figure 7. The results in the two paired reactors were essentially the same. (NO data are not
shown because the NO monitor used for this run is insufficiently sensitive to give useful NO data
in such runs.) The model used a HONO input rate of 0.39 ppb/hour, adjusted to approximately fit
the Os formation rate. The model also gave a good fit to toe observed NOX buildup, but tended to
underpredict PAN. However, the PAN analyzer used during this period did not tend to give very
reliable data.
A large number of n-butane - NOX experiments are carried out in this chamber to
characterize the chamber radical source. The HONO input rates that tend to give the best fits to
the NO oxidation rates in those experiments tend to vary from run to run, within the range of 0.4
to 1 ppb/hour.
5. EC Chamber Experiments
The experimental and model simulation results of the acetaldehyde - air experiment
carried out in the SAPRC EC are shown on Figure 8. The model simulation that fit the Os data
corresponded to a HONO input rate of 5.3 ppb/hour, significantly higher than observed in the all
Teflon reactors, discussed above. The model also gave a good simulation of the rate of NOX
buildup observed during the run, though tended to underpredict the PAN, to about the same
extent as observed in the other acetaldehyde - air experiments. Although the NOX readings
corresponded to an initial NOX of ~12 ppb, this is probably an interference or improper zeroing of
the instrument, since if 12 ppb of initial NOa or HONO is assumed in the model, then the model
significantly overpredicts the initial Os formation rate.
-186-
-------
Figure 7.
Figure 8.
Experimental and model simulation results of the acetaldehyde - air experiment
DTC764. The model used a HONO input rate of 0.39 ppb/hour.
,£ 0.000
o
NOx
0 60 120 180 240 300 360
PAN
0.004
0.003-
0.002-
0.001 -
0.000
0 60 120 180 240 300 360
Acetaldehyde
1.1 -
1.0-
0.9-
0.8
0.7 •
n« .
n ' D ' u i a i n-TtrrH_0
i Side A Data
a Side B Data
Model
0 60 120 180 240 300 360 0 60 120 180 240 300 360
Time (minutes)
Experimental and model calculation results for the acetaldehyde - air experiment
EC253. The model uses a HONO input rate of 5.3 ppb/hour, and an initial NOX
concentration adjusted to fit the initial 03 data.
NOx
0.06-1
0.05-
0.04-
0.03-
0.02
0.01
0.00
Formaldehyde
0 60 120 180 240 300 360
60 120 180 240 300 360 0 60 120 180 240 300 360
Time (minutes)
-187-
-------
Analysis of NOx-air (Carter et al, 1982) and modeling of n-butane - NOX experiments
(Carter et al, 1995a,b) indicate that the radical source in the SAPRC EC tends to be dependent on
NOX levels, and a standard wall model for the radical source in this chamber was developed based
on these data (Carter et al, 1982, 1995a). For the conditions of the acetaldehyde - air experiment
EC253, this wall model assumes a radical input rate corresponding to a NOX input rate of 5.3
ppb/hour, in excellent agreement with the NOX input rate that fit the Os data in the this
experiment
6. Summary of Results
Table 2 shows a summary of the results of the NOX offgasing and radical source
characterization experiments for the chambers discussed here. The NOX offgasing rates can be
seen to be the lowest in the clean, unused reactor, with the lowest NOX offgasing rates that fit the
data in the two CO - air runs being comparable to the average of the NOX offgasing rates observed
in the dark. The NOX offgasing rates that fit the data in TVA chamber runs were somewhat higher
than observed in the clean, unused reactor, but was a factor of 3-4 times lower than observed in
the DTC. The NOX offgasing rate observed in the EC was over an order of magnitude higher than
that observed in the DTC or other all-Teflon chambers, indicating that the surface of that chamber
is much more reactive. This is despite the fact that the between-runs cleaning in the EC is
probably much more thorough than in the other chambers (with the possible exception of the
TVA chambers), since the chamber is completely evacuated.
The NOX offgasing rates observed in the TVA chamber, the DTC, and the EC can be
compared with the apparent radical source input rates observed in these chambers. In the case of
the DTC and EC, the radical source rates are essentially the same as the NOX input rates, to within
the variabilities and uncertainties of the determinations. This is consistent with the hypothesis that
both the radical source and NOX input are due to the same factor, presumably HONO offgasing.
On the other hand, the apparent radical source in the TVA n-butane - NOX experiment was 6-10
times higher than the NOX input rates observed in the acetaldehyde - air runs. Either the
acetaldehyde - air runs are not representative of the conditions of the n-butane experiment, or
there is another radical source process involved in this chamber, or there is a problem with the
particular n-butane experiment that was modeled.
D. Conclusions and Near Term Work Planned
The experiments with new, unused reactors are incomplete and still underway, and the
analysis of the TVA chamber data is incomplete, so the conclusions arising from the work
discussed above must be considered to be preliminary. Nevertheless, the following conclusions
can be made:
• New Teflon film has non-negligible NOX offgasing, with the minimum offgasing rate
observed in this work being ~0.5 ppb/day. This is enough to cause significant ozone
formation over a period of a one day irradiation. The source of this NOX is unknown,
since the reactor has not been knowingly exposed to high levels of NOX in previous
experiments. The possibility of permeation is ruled out because the reactor was inside a
flushed bag with NOX levels below 0.1 ppb.
• The apparent NOX offgasing rates causing Oa formation in irradiated reactors is not
always higher than the rate of NOX offgasing in the dark. Therefore, the NOX offgasing
process may not necessarily be light induced, at least for very clean Teflon reactors.
-188-
-------
Table 2. Summary of results of modeling NOX offgasing and radical source
characterization runs in various chambers.
Run
Type
Light Intensity HONO Input
(ki, min-1) (ppb/hr)
Clean. Unused Reactor CFEP Teflon film")
Various Average for Dark Experiments (NOx input)
9/13 - 9/17 Pure Air
9/17-9/21 CO-Air
9/28-9/30 CO-Air
TVA Chamber (Low NOx Studies) (FEP Teflon film)
9/22/95 Acetald-Air
6/11/96 Acetald-Air
6/19/96 n-Butane-NOx
PTC (Well-used chamber) (FEP Teflon film)
DTC764 Acetald - Air
Various n-Butane - NOx
EC (Teflon coated aluminum with Quartz Windows')
EC253 Acetald - Air
Various n-Butane - NOx
0
0.8
0.8
0.8
0.4
0.4
0.4
0.2
0.2
0.3
0.3
0.02
0.10
0.02
0.02
0.09
0.16
0.9
0.4
0.4-1.0
5
5
• There is indirect evidence that most of the offgased NOX in the new, clean reactors is
mostly in the form of HONO. This is consistent with the observation, at least in some
chambers, that the apparent radical source is about the same as the NOX offgasing rates.
However, the radical source rate in the new Teflon reactors have not yet been measured,
and the radical source observed in at least one TVA n-butane run is much higher than the
NOX offgasing rates indicated by the data of the acetaldehyde - air runs.
• As might be expected, the NOX offgasing rates are higher in well-used reactors than in
new reactors. The multi-day light flushing procedure used with the TVA chamber
appears to reduce the NOX offgasing rate, though not to the levels observed in new, clean
reactors.
• The SAPRC evacuable chamber, which has Teflon coated metal and quartz surfaces, has
significantly higher NOX offgasing and radical source rates as the all Teflon chambers.
This indicates that this type of chamber is less suitable for conducting low NOX
experiments than all-Teflon reactors.
Work to better characterize the NOX offgasing effects from new reactors is still underway,
with the ultimate goal being not only to better understand these effects, but also to learn how to
design environmental chamber systems where these effects are minimized and predictable. The
following near term work is planned:
-189-
-------
• Luminol NOz analyzers are being developed to monitor NCh at the low levels needed to
provide useful data to evaluate NOX offgasing effects, and to determine the contribution
of NOa in the NOX offgased in the dark.
• Additional pure air and CO - air experiments are being conducted to better evaluate the
reproducibility of the NOX offgasing from new reactors. One should expect it to
eventually decrease with time.
• The effects of heat, changes in light intensity, humidity, and exposure to various NOX
species on NOX offgasing rates will be assessed.
• The effectiveness of heating, prolonged flushing in the light, various cleaning procedures,
and possibly other treatments in reducing NOX offgasing will be assessed.
• The use of other types of materials for reactor surfaces is being assessed. The data
discussed above that surfaces such as in the SAPRC EC will not be satisfactory, and
preliminary experiments in an all-glass chamber indicate that that type of chamber also
has a higher radical source (unpublished results from this laboratory). We are having
discussions with DuPont Co. about Teflon film characteristics and alternatives, and use of
reactors using other types of Teflon film will be investigated.
The results of this investigation will be used to determine the design for the "next
generation" chamber for generating mechanism evaluation data at very low NOX levels. Although
it is unlikely that zero NOX offgasing can be obtainable in practice, it is important to determine
the best chamber design and operating procedures to minimize these effects, and to understand
and be able to predictively model what cannot be eliminated.
E. References
Bailey, E. M., C. H. Copeland and R. Simonaitis (1996): "Smog Chamber Studies at Low VOC
and NOX Concentrations," Report on Interagency Agreement DW64936024 to
EPA/NREL, Research Triangle Park, NC.
Carter, W. P. L., J. H. Seinfeld, D. R. Fitz, and G. S. Tonnesen (1999): "Development of a Next-
Generation Environmental Chamber Facility for Chemical Mechanism and VOC
Reactivity Evaluation," Research proposal to the U.S. EPA, University of California at
Riverside. Available at http://helium.ucr.edu/~carter/epacham/proposal.htm.
Carter, W. P. L., R. Atkinson, A. M. Winer, and J. N. Pitts, Jr. (1982): "Experimental
Investigation of Chamber-Dependent Radical Sources," Int. J. Chem. Kinet, 14, 1071.
Carter, W. P. L. and F. W. Lurmann (1991): "Evaluation of a Detailed Gas-Phase Atmospheric
Reaction Mechanism using Environmental Chamber Data," Atm. Environ. 25A, 2771-
2806.
Carter, W. P. L., D. Luo, I. L. Malkina, and D. Fitz (1995a): "The University of California,
Riverside Environmental Chamber Data Base for Evaluating Oxidant Mechanism. Indoor
Chamber Experiments through 1993," Report submitted to the U. S. Environmental
Protection Agency, EPA/AREAL, Research Triangle Park, NC., March 20..
-190-
-------
Carter, W. P. L., D. Luo, I. L. Malkina, and J. A. Pierce (1995b): "Environmental Chamber
Studies of Atmospheric Reactivities of Volatile Organic Compounds. Effects of Varying
Chamber and Light Source," Final report to National Renewable Energy Laboratory,
Contract XZ-2-12075, Coordinating Research Council, Inc., Project M-9, California Air
Resources Board, Contract A032-0692, and South Coast Air Quality Management
District, Contract C91323, March 26.
Carter, W. P. L., D. Luo, and I. L. Malkina (1997): "Environmental Chamber Studies for
Development of an Updated Photochemical Mechanism for VOC Reactivity
Assessment," Final report to the California Air Resources Board, the Coordinating
Research Council, and the National Renewable Energy Laboratory, November 26.
Jeffries et al, 1999 - UNC wall model paper)
Jeffries, H., K. Sexton, and Z. Adelman (1999): "Auxiliary Mechanisms (Wall Models) for UNC
Outdoor Chamber," presented at the Combined US/German and Environmental Chamber
Workshop, Riverside, CA, October 4-6, 1999.
Simonaitis, R. and E. M. Bailey (1995): "Smog Chamber Studies at Low VOC and NOX
Concentrations: Phase I," Report on Interagency Agreement DW64936024 to
EPA/NREL, Research Triangle Park, NC.
-191-
-------
Experimental Techniques for Studying Surface
Chemistry in Smog Chambers
Laura T. Iraci, Jeffrey C. Johnston and David M. Golden
SRI International, Menlo Park, CA
Chemical reactions occurring on the walls of environmental chambers provide a
large fraction of the radical species which initiate gas phase smog chemistry, but the
nature and magnitude of these processes have not received focused attention. For
example, the empirical models used in conjunction with the UNC and Riverside smog
chambers contain different reaction schemes and product distributions for the adsorption
and reaction of NOa on surfaces, hi addition, several parameters are often adjusted
empirically when environmental chamber data is analyzed. The current understanding of
wall chemistry and its parameterization in models are briefly reviewed here.
Determination of the fundamental processes involved in wall reactions is needed for
prediction of heterogeneous radical generation in different chambers under a variety of
conditions. Techniques available for quantitative study of these heterogeneous processes
include Knudsen cell, flow tube, and aerosol methods. The use of these methods in our
laboratory is demonstrated, and their applicability to the study of environmental chamber
wall processes is discussed.
It has long been known that the gas-phase chemistry which occurs in an irradiated
environmental chamber cannot be reproduced in a mathematical model without the
inclusion of an initial radical source in the model. This is generally presumed to be some
(adjustable) quantity of nitrous acid (MONO) which is photolyzed to yield OH radicals at
the start of the experiment and at the start of the model run. The formation of this HONO
is generally attributed to heterogeneous formation from NOa and HbO at the wall.
An additional, constant source of OH is often added to model schemes to better fit
the experimental data. The physical basis for this OH production is less clear but may
relate to a photostimulated HONO production channel also occurring at the walls.
The results obtained from fitting chamber data thus depend significantly on the
choices made in the representation of the wall-induced radical sources. To properly
extrapolate reactions studied in smog chambers beyond the experimental conditions and
into models of the real atmosphere, molecular mechanisms of wall effects are needed.
Only with a proper accounting of the radical sources can we be certain that the correct
gas-phase reaction rates are obtained and that additional processes are not masked by the
tunable parameterizations of wall chemistry.
As a first step in achieving a molecular understanding of the wall reactions, several
studies must be undertaken:
• Identify conditions which lead to HONO production.
-192-
-------
what surfaces are active? how does wall history affect HONO production?
are there surfaces more inert than Teflon? what is the role of light?
• Quantify HONO production rate as a function of measurable chamber
parameters.
temperature, time, relative humidity, wall exposure history, hv,
• Determine if OH is directly emitted from Teflon surfaces or if a second,
photoinduced mechanism leads to HONO formation throughout an experiment.
hv needed? O2? H2O? hydrocarbons? is HONO produced in the chamber
throughout the experiment?
• Express OH production rate in terms of parameters which apply to all
chambers.
These and other studies will foster the development of an explicit wall mechanism that
can be incorporated into models in a manner analogous to the current detailed treatment
of gas phase reactions.
-193-
-------
Proposed Experiments to Characterize Wall-Induced Radical
Formation:
HONO adsorption to and desorption from Teflon
Characterize outgasing of Teflon
vs. RH, T, time, hv, NOX and HC history
watch for HONO, OH, other NOX?
What species are involved in HONO formation?
NO2 + H2O (+ hD) + surface
NO2 + H2O + O2 (+ hD) + surface
is O2 needed? what surfaces are more or less active?
Investigate mechanism using isotopes
Does the O come from NO2 or HjO?
H218O + NO2 -» H18ONO (Sakamaki et al., 1983)
Does the N come from new NO2 reacting at surface, or from old species already
trapped?
Add 15NO2 to Teflon aged with 14NO2 and/or H14NO3; look for HO15NO vs.
HO14NO. AddN18O2 and H2O; look for HON16O vs. HON18O
Available Techniques:
Two basic experimental arrangements are applicable to these studies. The first is the
traditional flow tube, which can be designed and operated with a coated wall or with
flowing aerosol particles to provide the reactive surface. A movable injector allows for
the variation of interaction time between gas and surface. A second standard technique
for studying gas-surface interactions is the Knudsen cell, which measures loss of a
reactant to wall surfaces in competition with escape from the cell.
Many analytical tools are available for the monitoring of processes in either the flow
tube or Knudsen cell arrangement. Mass spectrometry is commonly used to identify and
quantify stable reactants and products. Spectroscopic techniques such as laser-induced
fluorescence or resonance enhanced multi-photon ionization can be used to monitor
short-lived species inside the reactor. This allows for the detection of radicals which must
be measured in-situ before they are lost to downstream surfaces. Lastly, infrared
spectroscopy can be used to detect gases in a static cell, allowing longer interaction times
than those accessible with a flow tube or Knudsen cell.
-194-
-------
Movable Injector —
(Reactive Gas Inlet)
; Bath Gas Inlet
"emperature
Oortrol Jacket
Operating a coated-wall flow tube would allow the glass surface to be coated with
FEP Teflon, the same material used for chamber bags and wall coatings. The maximum
interaction time in a flow tube is on the order often seconds, and the experimentalist has
good control over conditions such as temperature, relative humidity, total pressure,
fraction of O2 present, identity of the carrier gas, and the light field available for
photolysis. Because the Teflon surface area is well-known, rate coefficients are easily
determined.
hi contrast, an aerosol flow tube is designed to continually supply fresh surface area
via particles roughly 0.1 - 1 um in diameter. The surface area can be greatly increased
over that available in a coated-wall apparatus, allowing the observation of less efficient
processes, hi addition, the Teflon particles can be pre-treated before their introduction
into the flow tube, facilitating studies of the effect of previous exposure to HNO3, NaCl,
etc. Particles can also be pre-treated with organics if desired.
-195-
-------
Valve
Gas Inlet
_l
Calibrated
Aperture
"to Mass
Spectrometer
The Knudsen cell, commonly used for uptake and kinetic measurements, allows the
stationary reactive surface to be exposed for minutes to hours. This makes the technique
especially well-suited to studies of wall saturation/passivation and regeneration
processes. Our current cell is glass coated with FEP Teflon prepared from an aqueous
suspension that is dried and annealed in place. Our cell walls would age similarly to those
of smog chamber but can easily be stripped and recoated to start with fresh Teflon.
Controlling the wall temperature is straightforward, something which cannot be said of
an aerosol flow tube arrangement. In addition, other coatings or inserts of Teflon
previously used in a smog chamber are easily accommodated.
-196-
-------
nGas Inlet
gas-tight cell
with IR windows
Walls could be coated, or Teflon wool or film could be placed in the cell.
A static cell coupled with infrared detection of the gases inside would be ideal for
long-duration studies of the loss or release of .gases from wall materials. Questions such
as: how much NOa is lost to the walls over the course of an 8 hour experiment? or how
much HONO is present the next morning? could be addressed. Also, the time evolution
of species after a step-wise change in conditions can be monitored.
In summary, many aspects of wall-induced chemistry in smog chambers can be
addressed by systematic lab studies. Several established experimental techniques are
available, and combining the strengths of multiple methods will allow us to study
problems from different perspectives. The goal of studying these processes is to develop
a physically-based understanding of reactions at the walls and the adsorption/desorption
behavior of gases present in chamber experiments. This will facilitate a realistic
extrapolation of chamber data to the atmosphere. In addition, studies of heterogeneous
HONO formation are directly relevant to polluted urban areas, which contain many
stationary and airborne surfaces and often exhibit increased levels of ambient HONO.
-197-
-------
Heterogeneous Reaction of N2O5 on Anorganic Aerosols:
The Nitrate Effect
Andreas Wahner and Thomas Mentel
Institut fur Atmospharische Chemie, Forschungszentrum Julich, D-52425 Jiilich
Heterogeneous hydrolysis of N2O5 on aqueous aerosol surfaces is an important atmospheric source of
HNOj. Reaction probabilities TNIOS of N2O5 on several aqueous inorganic substrates, like ammonium
sulfates, which compose the tropospheric aerosol, have been reported in the literature. These ymos are
typically of the order of several times 10~2 and vary relatively little for a wide range of substrates, from
pure water over deliquescent aerosols to metastable aerosols. The relative insensitivity of Jmos with
respect to the changing water activity has lead to the proposal of an ionic reaction mechanism in the
aqueous phase with reaction (R2f) is as the rate limiting step:
N205(g)
N20S(1)
2+ + H2O(1)
N2O5(aq)
NO2+ + N(V
H+ + HNO3(aq)T
(Rl)
(R2f)
(R3)
We measured ^2os on aqueous sodium nitrate, sulfate, and bisulfate aerosols, in the large aerosol chamber
at the FZ-Juelich at different relative humidities and room temperature. At a relative humidity of 50%
corresponding to a NO3~ molality of 27 we observed a YN205 of 0.0018, which is an order of magnitude
smaller than ^j2os on water, ammonium and sodium sulfate aerosol at similar ionic strength, ynaos
increases by an order of magnitude to 0.023 when the relative humidity is raised to 90% which
corresponds to a NO3~ molality of 4. These findings can be explained by the increasing importance of the
recombination reaction (R2b) with decreasing relative humidity and correspondingly increasing NO3"
molality.
NO/ + NCV -> N2O5(aq) (R2b) <
The effect of NO3" can be rationalized by a steady state consideration for N2O5(aq) and NO2+ within a thin
aerosol surface shell. The observation of a specific nitrate effect is a direct experimental indication for the
ionic uptake mechanism (R1)-(R3).
References: A. Wahner et al., J. Geophys. Res. (1998)
-198-
-------
Tropospheric Aqueous Phase Chemistry
Laboratory and Modelling Studies
Hartmut Herrmann
Institut fur Tropospharenforschung
Permoserstr.15, 04318 Leipzig
Results from recent photochemical and kinetic laboratory and modelling studies of the
formation and reactivity of aqueous phase free radicals such as OH, NO3, and C1/C12" will be
presented. Laser-based methods have been applied for the specific generation and time-
resolved detection of the above transient species in systematic studies. Photolytic radical
generation in solution, radical phase transfer, radical interconversion reactions and the
influence of organic compounds on the chemistry within the aqueous tropospheric phase will
be discussed. Results indicate that solution reactions of the above radicals may significantly
influence the net effects of chemistry within droplets and aerosols dispersed in air.
A multiphase box model coupling an advanced aqueous phase mechanism (CAPRAM 2.4) to
RADM2/RACM is applied to quantify effects of multiphase conversions. It will be discussed
how aqueous phase processes alter the oxidation capacity of the tropospheric gas phase by
uptake and release of trace gases and radicals. Current restrictions of models will be outlined.
-199-
-------
Session III
Measurement Methods
Session Chair
Ulrich Platt
-------
IN-SITU RADICAL MEASUREMENTS IN EUPHORE
U. Platt
Institut fur Umweltphysik, Univ. Heidelberg, INF 366, D-69120 Heidelberg, Germany
The investigation of chemical processes in ,smog chambers' or photoreactors is a quite old tool in
atmospheric chemistry research. In fact the factors determining urban ozone formation were empirically
known from smog chamber studies long before the photochemical theories involving free radicals (i.e.
OH) were formulated and experimentally verified.
Recent advances in measurement techniques made it possible to directly determine concentrations of a
variety free radicals during photoreactor experiments. For instance in the European Photoreactor
(EUPHORE) in Valencia successful, direct detection of several free radical species at concentrations
comparable to ambient levels was demonstrated:
• Nitrate radicals (NO3) by Differential Optical absorption spectroscopy (DOAS)
• Hydroxyl radicals (OH) by laser induced fluorescence (LIF)
• Sum of peroxy radicals (RO2) by chemical amplification
In addition also the sensitive detection of halogen oxide radicals (CIO, BrO, IO, OC1O, etc.) is possible by
DOAS. Also hydro-peroxy radicals (HO2) can be measured by chemical conversion to OH and subsequent
LIF - detection. An interesting possibility concerns the measurement of the nitrous acid (HONO)
concentration as an OH - precursor.
Compared to indirect determination from e.g. relative rate experiments direct detection of free
radicals allows better characterisation of chemical mechanisms, for instance in the degradation of
aromatic hydrocarbons. In addition absolute reaction rate constants can be measured, this has
been done for NOs + VOC reactions. Another example is the determination of the equilibrium
constant NO3 + NO2 <=> N2O5.
The various techniques employed for the measurement of free radicals in the EUPHORE
photoreactor are described, some applications, and future plans are discussed.
-201-
-------
Measurement of Peroxy Radicals in the European Photoreactor EUPHORE
First results from the project SAMPLER
M. Heitlinger. L. Hoppe, D. Mihelcic, P. Miisgen, F. Kirchner§, K. Wirtz$
InstitutJurChemiederBelastetenAtmosphare (ICG-2), Forschungszentrum Jiilich, D-52425-Jiilich, Germany. §: Ecole
Polytechnique Federate de Lausanne (EPFL), 1015 Lausanne, Switzerland.
s: Centra de Estudios Ambientales del Mediterraneo (CEAM), 46980 Paterna (Valencia), Spain.
Introduction
Peroxy radicals (HOa and ROa) are key species in the photochemical formation of ozone, peroxides
and organic nitrates in the troposphere. They arise from the oxidation of volatile organic
compounds (VOC) and CO by hydroxyl radicals (OH). The reaction of peroxy radicals with NO is
the rate-limiting step of Os production. Recent literature data for the rate coefficients for this
reaction disagree by a factor of two or more (feelers et al, 1992, Eberhard and Howard, 1997).
We studied the oxidation of i-butane (100 ppb) in the presence of 10 ppb NOX and 110 ppb HCHO
in the European Photoreactor EUPHORE. Measurements of peroxy radicals were made with two
methods: Matrix-Isolation followed by electron spin resonance MIESR and Chemical amplification
CA. The results are compared to model calculations.
Experimental
The experimental procedure was as follows: After flushing the chamber for at least 12 hours with
purified air, an FTIR background spectrum was recorded. Then, the reagent gases NOa and i-butane
and the tracer SFe were introduced in the chamber by adding known amounts of the pure gases into
a gas flow of 2 sL/min which was fed into the chamber. HCHO was added by evaporating a known
amount of paraformaldehyde into the same gas flow. After allowing for mixing, the initial
concentrations were determined. Then, the chamber was exposed to sunlight and the temporal
evolution followed for about 6 hours. Table 1 gives an overview over the initial conditions and the
instrumentation.
Table 1
Initial conditions and instrumentation
Species
i-butane
NO, NO2
HCHO
SF6
03
HO2, RO2, NO2
(HO2+RO2)
Initial value
100 ppb
11 ppb
110 ppb
26 ppb
0
0
Instrumentation
FTIR
PC-CLD
FTIR
FTIR
UV-Absorption
MIESR
CA
Time resolution
10 minutes
1 minute
1 0 minutes
1 0 minutes
1 minute
30 minutes
1 minute
Detection limit
3 ppb
lOOppt
3 ppb
0.5 ppb
2 ppb
2.5 ppt
2ppt
HCHO, i-butane and SFe were measured with FTIR. NO was measured with chemiluminescence.
NOa was converted to NO by a photolytic converter and measured by chemiluminescence. In
addition, NOa was measured by MIESR. Os was measured using UV-absorption. The photolysis
rate of NOa, J(NOa), was measured using two 2n filterradiometers pointing upward and downward,
respectively, whose spectral sensitivity matches the effective cross section of NOa. Measurements
of peroxy radicals were achieved using two methods: Matrix-Isolation Electron Spin Resonance
MIESR (Mihelcic et al, 1985, 1990) and Chemical Amplification CA (Cantrell and Stedman, 1982,
Hastie etal.,1991, Heitlinger, 1997). MIESR is an absolute method which allows speciation of HOa
and ROa (and NOa and NOs). In contrast, the CA is an indirect method which measures the sum of
HOa and ROa. The HOa calibration source used for calibration was compared to MIESR and agreed
within 15% (Heitlinger, 1997). For further details of the MIESR method, see Mihelcic et al., 1985,
-202-
-------
1990; for details of the CA and calibration source see Heitlinger, 1997. Both instruments were
mounted in the same distance from the chamber floor in specially designed flanges.
Data Quality
Figure 1 shows the comparison of the NC>2 values measured by PC-CLD and by MIESR. The high
concentrations agree within 2%, whereas the lower concentrations (which were recorded in the
presence of ca 80 ppb 03) are underestimated by PC-CLD by up to 300 ppt (or 15%). This
underestimate is presumably caused by the negative O3 interference in the photolytic converter
which reduces the NC>2 conversion efficiency.
NO2 Comparison
345
NOz MIESR [ppb]
Figure 1
Comparison of NO2 measured by PC-CLD and NOz measured by MIESR. The error bars represent
the variability of the NOa concentrations during that time. The solid line shows the expected 1:1-
behaviour. The higher concentrations agree within 2%, whereas the lower concentrations are
underestimated by PC-CLD by up to 300 ppt or 15%.
The calibration of the Oa measurement was found to agree within 3% with another instrument
which was calibrated in Julich before and after the campaign. The NO measurements are tied to a
certified gas mixture cylinder accurate within 2%.
The comparison of the radical measurements is shown in Figure 2. The data agree within 5%, which
is due to the fact that the calibration source of the CA was characterised using MIESR
(Heitlinger, 1997).
-203-
-------
Comparison of CA and MIESR
300 -
Figure 2
R = 0.9779
0 100 200 300
(HO2+RO2) [ppt] MIESR
Comparison of the radical measurements by CA and MIESR.
In order to ensure that wall loss of radicals can be neglected, the radical profile in the EUPHORE
was checked with the CA by positioning the inlet in several distances from the wall. Figure 3
shows the results of these checks. The red line shows the radical concentrations measured by the
CA, the blue line shows the distance between the chamber floor and the inlet of the CA. It can be
clearly seen that the position of the inlet has no influence on the measured radical concentrations.
The slight increase of the radical concentrations with time is due to the increasing light intensity
over that time period.
-204-
-------
Radical profile in EUPHORE
1000
— ROx
—Height above floor
10:00
10:15 10:30
Local Solar Time
Figure 3 Radical profile in EUPHORE measured with the CA.. The red line shows the radical concentrations
measured by the CA, the blue line shows the distance between the chamber floor and the inlet of the
CA. The position of the inlet has no influence on the measured radical concentrations. The slight
increase of the radical concentrations with time is due to the increasing light intensity over that time
period.
Model calculations
In ordef to compare the measurements with photochemical theory, we made model calculations
using the "Master Chemical Mechanism" described by Jenkin et al., 1997, with some minor
changes as described below.
Procedure for the model calculations
The model was initialised with the initial (dark) conditions for the time of chamber-opening and
allowed to equilibrate. A common, constant dilution factor was applied to all species. The
magnitude (ca 2%/hour) was chosen to reproduce the measured SFe concentrations. Deposition of
Os and NO2 to the chamber walls was taken as 2%/hour as derived from the lifetime of Os in the
chamber (Wirtz, pers. comm.). Besides these changes, no forcing of concentrations was applied to
the model. Forcing of the photolysis rates was achieved as follows: first, the model was used to
calculate the clear-sky photolysis rates as described by Jenkin et al., 1997. Then, all photolysis rates
were scaled with the ratio of measured J(NO2) to calculated J(NO2>. The resulting photolysis rates
were used for the model calculations. This forcing was applied every .minute. This procedure
assumes that the light attenuation by clouds is identical for all wavelengths, which is presumably
not true (Kraus, 1998). Since we found that the decrease of HCHO was initially strongly
overestimated by the model, we decreased J(HCHO) to match the HCHO concentrations.
Mechanistic changes to the model
The rate coefficient for the reaction OH + NO2 was taken from Donahue et al., 1997. The influence
of the reaction ROa + NO was studied by performing two different model runs: In model A, we
-205-
-------
'12
used a rate coefficient of k=8-10"12 cm3molek.~Is~I as suggested by Eberhard and Howard, 1997, in
model B, we used k=4-10~12 cnAnolek.'V1 as measured by Peeters et al, 1992.
Results and discussion
Figure 4 shows the time series of the measured species and the results of the model calculations.
The red lines represent the measurements performed with the Euphore equipment and the CA, the
black lines show the MIESR results. The blue lines represent the standard Model with k (RO2 +
NO)= 8-10'12 cnAnolek/V1; the green lines the standard model with k (RO2 + NO)= 4-10
cm molek.'V. The concentrations of HCHO, NO and NO2 are reproduced very good by the model,
agreement is within 5%. The agreement is slightly worse for i-butane (ca 10%), which is
underestimated by the model. One possible reasons is a slightly lower rate coefficient for the
reaction of i-butane with OH. The radical measurements of the CA reproduce the diurnal variation
of ROX (=HO2+ROa) fairly good, including short-term fluctuations of the ROX concentrations. The
absolute values measured by CA agree well with both the model and the MIESR data. This is not
surprising, given the fact that the calibration source of the CA was characterised using MIESR
(Heitlinger et al, 1997).
Both model runs yield similar values for the sum of HO2 and RO2. However, the split between HO2
and RO2 is greatly influenced by a change in the rate coefficient of RO2 + NO. Model B (k= 4 -10"12
cn&nolek/'s"1) is much closer to the MIESR results, especially for the afternoon values. In
addition, the OB concentrations predicted by Model B are in much better agreement with the
measurements than those of Model A. Our data support the value of k=4-10~12 cm3molec."1s"1
et
The values of k=8-10~12 cm3molec."1s"1
measured by Peeters et al., 1992, for t-C^Oz (the predominant organic RO2 in our experiment).
found by Eberhard and Howard, 1997, for this peroxy
radical, however, contradict this finding. Therefore, further work on the rate coefficients of this and
other RO2 with NO is required.
-206-
-------
6.0 T
50 -
r-,40
^^-T\J
0,
JBs30-|
11:00 13:00
Local solar time
15:00
9:00 11:00 13:00 15:00
Local solar time
Figure 4 Results of the experiments and model calculations. The red lines represent the measurements of
HCHO, i-butane, NO2, J(NO2), NO, ROX and O3, the black lines the MIESR results. The blue lines
represent the standard Model with k (RO2 + NO)= 8-10"12 cm3molek."ls"!; the green lines the
standard model with k (RO2 + NO)= 4-10'12 cm3moIek.'V'.
-207-
-------
Conclusions
Measurements of HOa and ROa were conducted successfully for the first time in the EUPHORE. A
comparison of the results with model calculations support the value ofPeeters et al., 1992, for the
rate coefficient of t-C^Oa + NO , namely 4 -10"12 cm3molek."1s"1. Future work will focus on the
radical chemistry of alkenes and aromatics.
References
Becker, K.-H.; The European Photoreactor EUPHORE: Design and Technical Development of the European
Photoreactor and first Experimental Results; Final Report of the EC-Project Contract EV5V-CT92-0059, BUGH
Wuppertal, Wuppertal (1996)
Cantrell, C.A., and D.H. Stedman; A Possible Technique for the Measurement of Atmospheric Peroxy
Radicals, Geophysical Research Letters, 9 (1982) 846-849.
Donahue, N.M., M.K. Dubey, R. Mohrschladt, K.L. Demerjian, and J.G. Anderson; High-pressure flow study
of the reactions OH+NOx -> HONOx: Errors in the falloff region, Journal of Geophysical Research, 102 (1997) 6159-
6168.
Eber'nard, J., and C.J. Howard; Rate Coefficients for the Reactions of Some C3 to C5 Hydrocarbon Peroxy
Radicals with NO, Journal of Physical Chemistry, 101 (1997) 3360-3366.
Hastie, D.R., M. Weissenmayer, J.P. Burrows, and G.W. Harris; Calibrated Chemical Amplifier for
Atmospheric ROx Measurements, Analytical Chemistry, 63 (1991) 2048-2057.
Heitlinger, M.; Untersuchungen zur Messung von Peroxiradikalen mittels chemischer Verstarkung, Ph.D.,
Bcrgische Universitat-Gesamthochschule Wuppertal, Wuppertal (1997)
Hofzumahaus, A., T. Brauers, U. Aschmutat, U. Brandenburger, H.-P. Dorn, M. Hausmann, F. Holland, C.
Plass-Dulmer, M. Sedlacek, M. Weber, and D. Ehhalt; Reply to Comment by Lanzendorff et al, Geophysical Research
Letters, 24 (1997) 3039-3040.
Jenkin, M.E., S.M. Saunder, and M.J. Pilling; The tropospheric degradation of volatile organic compounds: A
protocol for Mechanism Development, Atmospheric Environment, 31 (1997) 81-104.
Kraus, Alexander, Ph. D. Thesis, Cologne, 1998.
Lanzendorff, E.J., T.F. Hanisco, N.M. Donahue, and P.O. Wennberg; Comment on Hofzumahaus et al., 1996
and Brauers et al., 1996, Geophysical Research Letters, 24 (1997) 3037-3038.
Mihelcic, D., D. Klemp, P. Miisgen, H.W. Patz, and A. Volz-Thomas; Simultaneous Measurements of Peroxy
and Nitrate Radicals at Schauinsland, Journal of Atmospheric Chemistry, 16 (1993) 313-335.
Mihelcic, D., A. Volz-Thomas, H.W. Patz, D. Kley, and M. Mihelcic; Numerical Analysis of ESR Spectra
from Atmospheric Samples, Journal of Atmospheric Chemistry, 11 (1990) 271-297.
Peeters, J., J. Vertommen, and I. Langhans; Rate Constants of the Reactions of CF3O2, i-C3H7O2, and t-C4H9O2
with NO, Journal Berichte der Bunsengesellschaft Physikalische Chemie, 96 (1992) 431-436.
Schultz, M., M. Heitlinger, D. Mihelcic, and A. Volz-Thomas; A Calibration Source for Peroxy Radicals with
Built-in Actinometry Using H2O and O2 Photolysis at 185 nm, Journal of Geophysical Research, 100 (1995) 18811-
18816.
-208-
-------
Measurement of NOy and Potential Artifacts
Dennis R. Fitz
University of California, Riverside
College of Engineering
Center for Environmental Research and Technology
Background
Oxidized nitrogenous species play a critical role in the formation of ozone in the
atmosphere. For both smog chambers and ambient air, a knowledge of the mass balance
of such species is useful for the development and validation of methods to model ozone
formation. The term NOy has been developed to represent such species although the
definition depends on whether it is used by a modeler or an analyst. The modeler would
like a measure of the total reactive nitrogen oxides, or odd nitrogen, while the analyst
uses an operational definition as the response of a chemiluminescent NO analyzer after
the sample is treated with a converter that reduces more highly oxidized species to NO.
The term (NOy)i has been used to designate this operational definition from that of NOy.
These definitions, though slightly different, should be quite similar based on our
knowledge of the individual oxidized nitrogenous species typically found in ambient air.
Other definitions in general use are NOX (NO + NO2) and NO2 (NOy - NOX).
It is generally agreed that NOy consists primarily of nitric oxide (NO), nitrogen dioxide
(NO2), peroxyacetyl nitrates (PacNs, primarily peroyactyl nitrate (PAN) and peroxyproyl
nitrate (PPN)), nitric acid (HNO3), paniculate nitrate, and nitrous acid (HONO) roughly
in that order of importance. Commercial converters have been shown to readily reduce
these gaseous species while they have no efficiency for N2O or organic nitro compounds,
species which modelers do not consider photochemically reactive (Winer et al., 1974).
The reduction efficiency for paniculate nitrate may vary with converter design, although
little quantitative work has been reported in this area.
Most converters for nitrogen oxide reduction use either heated gold or molybdenum
surfaces. Although they may be designed to minimize the collection of paniculate nitrate
(PN), ammonium nitrate is the most prevalent form of PN in ambient air. The high
temperature of the converter will easily volatilize it into ammonia and nitric acid, the
former being readily reduced to NO within the converter. Other converters have been
constructed with ferrous sulfate at ambient temperature and vitrous carbon at high
temperature. High temperature stainless steel has also been used, but the temperatures
necessary for high reduction efficiency result in a significant efficiency for ammonia
reduction to NO.
(NOy)i converters are typically placed outside while the detector assembly remains in an
air-conditioned environment. This is done to insure that the sampling lines do not adsorb
nitric acid. This is a major problem when the humidity is increased when the sample is
transported to a lower temperature in an air-conditioned shelter. To prevent the
-209-
-------
adsorption of nitric acid, (NOy); sampling must be conducted without a paniculate filter
using the shortest possible PFA Teflon inlet line to the converter. If NO is also to be
measured and the switching valve remains in the analyzer, a separate sample line is run
between the inlet and the cycling valve. Calibration gases are introduced at this inlet.
Depending on the distance from the analyzer, sample flow rate, and cycling time
(between (NOy)i and NO modes) of the instrument, bypass flow may be required due to
the distance between the converter and the detector.
The objective of this paper is to describe some of the interferences, biases, and
idiosyncrasies that we and others have observed when measuring NOy. It is important to
note that much of the reported (NOy); measurement studies have focused on unpolluted
air, while we have been making measurements on relatively polluted air in which NOX
was often a minor component of the (NOy);. It should also be noted that he sub ppb
sensitivity of the new generation of commercial NO analyzers have made these
measurement artifacts more noticeable.
Interferences
One of the problems when using a chemiluminescent analyzer to measure NO is that
water quenches the chemiluminescence. This results in a small zero bias and a larger span
bias, up to 10% (Fahey et al, 1985). This raises the question of what type of air to use to
assess the zero and span response. Ideally the air should be scrubbed of all species that
would cause an instrumental response, but this is unlikely to be achieved. The method of
standard addition is an approach that has been previously employed. In this approach,
known amounts of calibration gases are added to the air being sampled by the analyzer..
The calibration gases' concentrations are much higher than the desired calibration point,
thus the flow rate of the calibration gas is small compared with the analyzer's flow rate.
This method works best in a rural environment where concentrations are not rapidly
changing. A potential solution to the rapid changes in an urban environment would be to
capture ambient air in an inert flexible container (such as a Teflon bag), mix it well, and
then use it as a diluent for a cylinder of NO calibration gas.
The potentially more important source of interference is the reduction converter.
Although these were designed to reduce NO2 to NO, it is well known that they reduce
most of the nitrogen oxides except organic nitro compounds and nitrous oxide. Due to the
high natural background of nitrous oxide, all potential converters must be shown to have
virtually no reduction efficiency for this compound. Commercial NO-NOX analyzers
generally used converters based on molybdenum. Complete conversion is obtained at a
temperature of 350 °C rather than the 500 °C or more required for stainless steel.
Although these lower temperature converters should minimize the reduction of ammonia,
we have found converter efficiencies that varied from 5 to 25% for this species as shown
in Table 1. The efficiency did not correlate with the converter temperature. A
representative of the manufacturer stated that these conversion efficiencies are normal
and that they usually rise as the converter ages due to the production of molybdenum
trioxide (Kieta, 1999). Hot spots caused by the heater design may also lead to
molybdenum decompositions (Williams, et al., 1998).
-210-
-------
Hydrocarbons have also been shown to affect converter efficiency, and a memory effect
after sampling high NOy concentrations has been observed by us and others (Fehsenfeld
et al., 1987). The memory effect results in a significant and decreasing background after
the zero is supplied to the converter after sampling polluted ambient air. We have
observed concentrations as high as 6ppb on zero air that gradually fall to zero after many
hours of sampling zero air. Bypassing the converter eliminates this memory artifact. The
memory effect can be reduced by daily bakeout at 600 °C (Fehsenfeld et al. 1987),
although this is inconvenient and can lead to early converter failure.
We have also observed a selective degradation of converter efficiencies in (NOy);
analyzers. In this case converter temperatures were adjusted to achieve near 100%
efficiency for NO2, PAN, HONO, HNO3 , and n-propyl nitrate (NPN). After the
instruments were used for one summer to measure ambient air concentrations at locations
of varying air quality they were brought together to sample air during photochemical
oxidations conducted in our environmental chamber. Figure 1 shows the results of (NOy);
measurement during one experiment. In the dark, when only NO and NO2 are in the
chamber, all of the instruments show nearly the same concentrations. After the lights are
turned on NO (measured by a chemiluminescent analyzer without a converter) drops
rapidly, the NO2 (measured by a tunable diode laser absorption spectrometer, TDLAS)
rises to a maximum then drops, and the (NOy); concentrations slowly diverge as more
oxidized (NOy); species are formed.
Catalysts consisting of heated gold tubing have been reported in a number of research
applications, particularly in rural areas (Bellinger et al., 1983). Either CO or H2 is added
to the sample stream to facilitate the reduction of (NOy)j. The conversion efficiencies of
this type of converter for NH3 and HCN was found by Fahey and co-workers (1986) to be
significant with dry air but negligible with ambient levels of water vapor. These
converters have been shown to have the potential for significant efficiency for ammonia
and cyanides and to be dependent on water, ozone concentration and previous sampling
history. Converter design and temperature profiles are also likely to affect the conversion
efficiency for these compounds. The conversion efficiency is also affected by relative
humidity and pressure of the air sampled (Crosly, 1996). Kliner and co-workers (1997)
reported a detailed laboratory study of gold and other metallic converter and found gold
to be the best material for tube construction. H2 was preferred over CO as losses of the
latter were observed on the stainless steel components of the converter and was not as
efficient at converting NPN. They reported substantial conversion efficiencies for HCN
and NH3 to NO, especially under dry conditions. Measurement artifacts due to impurities
in the CO source have also been reported (Wang et. al., 1996). Frequent cleaning is
apparently needed under ambient sampling to maintain high (NOy); conversion efficiency
and in one case full efficiency could not be restored by cleaning (Bellinger et al. 1983;
Fahey et al. 1986;Wang et al., 1996; Williams et al., 1998).
Both gold and molybdenum-based converters of different designs have been used in a
variety of (NOy); intercomparison studies (Fehsenfeld, 1987, Williams et al. 1998). These
converters have been found to generally give similar (NOy); measurements although some
groups systematically reported lower concentrations. Selective loss of nitric acid
-211-
-------
conversion efficiency has been observed for both types of converters (Williams et al.,
1998). This study also showed that ammonia conversion efficiencies ranged from zero to
twelve per cent. William et al. conclude that (NOy)i measurements require constant
attention from experienced personnel, frequent converter efficiency checks (including on-
site nitric acid and ammonia), and redundant instruments.
Other converters have been reported but not widely used. Ferrous sulfate effectively
reduces NO2 (Helas et al., 1981), but has variable efficiency for other (NOy)i components
(Fehsenfeld et al., 1987) and a significant memory effect (Dickerson, 1985). A former
NO-NOX analyzer manufacturer used a heated vitrous carbon converter, but no studies of
potential interferences have been reported.
Regardless of the type of converter used, (NOy); must be sampled without a particulate
filter. The efficiency for conversion of particulate nitrate to NO has not been quantified
and will depend on the cation and most likely a number of environmental variables. Also
complicating the evaluation is the difficulty in removing nitrogenous species from the
particulate matter or vice versa due to the equilibium between nitric acid, ammonia, and
ammonium nitrate and the affinity of the ammonia and nitric acid to adsorb on surfaces
and collected particulate matter.
Compound X
Studies where the major individual NOy species are measured (NO, NO2, PAN, PPN,
HNOs, NOs") and added together result in a summation that is typically significantly less
than the measured (NOy)i. This missing species has been dubbed compound X and the
significance of this missing species appears to be location and method dependent. Fahey
et al. (1986) speculated that this compound X was an organic nitrate other than PAN of
PPN. Further evidence of missing (NOy); species have been reported by others (Roberts,
1990; Ridley, 1991; Parrish et al., 1993; Nielsen et al., 1995).
Crosley (1996) summarized the evidence for compound X and conclusions from a panel
of (NOy)i measurement experts. They concluded that the excess (NOy)i could likely be
due to adsorption/desorption of nitric acid in the inlet to the converter. Williams et al.
(1997) observed that the missing (NOy)i correlated directly with the ratio of NOX to NOy
Os and aerosols and inversely with temperature suggesting that one or more individual
NOy species were not being measured (perhaps a compound X).
Experiments that we recently performed on dual converter (NOy)i analyzers have shown
that some or all of the compound X may be due to ammonia. These instruments were
designed to measure nitric acid by the difference between the response of the (NOy)i
channel and a similar channel from which the nitric acid is removed with a sodium
chloride coated filter. After using two instruments to measure nitric acid for a summer,
we conducted a quality control test in which the sodium chloride filters were removed
and the instruments set up to sample air from which nitric acid had been removed using a
sodium chloride coated filter. Figure 2 shows the results of measuring the difference
between the two channels for several days in Riverside, California. Rather than a
response near zero (all nitric acid has been removed prior to the sample inlet and the
-212-
-------
analyzers did not have filters to remove nitric acid), we saw negative peaks each day in
the afternoon. In the afternoon Riverside is usually downwind of a large concentration of
dairies. Significant ammonia concentrations have routinely been observed in Riverside.
Based on these findings and our laboratory testing of the conversion efficiency of
ammonia, we feel that ammonia was causing this measurement artifact. In this case, the
converters that were used on the channel scrubbed of nitric acid were more efficient in
converting ammonia than the (NOy)i converters, thus resulting in a negative concentration
when the difference is taken.
Molydenum Converter Idiosyncrasies-note that sub ppb sensitivity has exacerbated the
idiosyncrasies
• Lag time with various species- especially nitric acid, but ambient air in general.
• Zero bias or memory effect, especially for (NOy)i.
• Converter hysteresis - show Riverside HNOs scrubbed air- this may explain some of
the inconsistencies reported in the literature.
• Lowered efficiency when placing the converter at the sample inlet ahead of the flow
control capillary.
Conclusions
• There are considerable conflicting reports with respect to convert efficiencies for
(NOy)j compounds and interfering compounds.
• The use of presently used converters result in significant biases and interferences
which center on more highly oxidized species such as HNO3 .
• Better converters need to be developed.
• Until better converters are developed frequent zero checks (perhaps every hour or
two) are needed as well as routine assessment of converter efficiency for both NOy
and ammonia.
• Acid coated diffusion denuders may be useful in removing ammonia so that it does
not interfere with (NOy)i measurements. It will be necessary to thoroughly test such a
denuder to insure that it does not remove a significant amount of (NOy)i species.
-213-
-------
References
Bellinger, M.J., Sievers, R.E., Fahey, D.W., and Fehsenfeld, F.C. (1983) Conversion of
nitrogen dioxide, nitric acid, and n-propyl nitrate to nitric oxide by gold-catalyzed
reduction with carbon monoxide. Anal. Chem. 55, 1980-1986.
Crosley, D.R. (1996) NOy blue ribbon panel. J. Geophys. Res. 101 2049-2052.
Dickerson, R.R. (1985) Reactive nitrogen compounds in the Arctic. Geophys. Res. 90,
10,739-10743.
Fahey, D.W., Eubank, C.S., Hubler, G., and Fehsenfeld, F.C. (1985) Evaluation of a
catalytic reduction technique for the measurement of total reactive odd-nitrogen
NOy in the atmosphere J. Atmos. Chem. 3, 469-489.
Fahey, D.W., Hubler, G., Parrish, D.D., Williams, E.J., Norton, R.B., Ridley, B.A.,
Singh, H.B., Liu, S.C. and Fehsenfeld, F.C. (1986) Reactive nitrogen species in
the troposphere: measurements of NO, NO2, HNO3, paniculate nitrate,
peroxyacetyl nitrate (PAN), Os, and total reactive odd nitrogen (NOy) at Niwot
Ridge, Colorado. J. Geophys. Res. 91 9781-9793.
Fehsenfeld, F.C., Dickerson, R.R., Hubler, G., Luke, W.T., Nunnermacker, L.J.,
Williams, E.J., Roberts, J.M., Calvert, J.G., Curran, C.M., Delany, A.C., Eubank,
C.S., Fahey, D.W., Fried, A., Gandrud, B.W., Langford, A.O., Murphy, P.C.,
Norton, R.B., Pickering, K.E., and Ridley, B.A. (1987) A ground-based
intercomparison of NO, NOX, and NOy measurement techniques. J. Geophys. Res.
92,14,710-14,722.
Helas, G., Flanz, M., and Wameck, P. (1981) Improved NOX monitor for measurements
in tropospheric clean air regions. Intern. J. Environ. Anal. Chem. 10, 155-166.
Kita, D. (1999) Personal communication.
Kliner, D.A.V., Daube, B.C., Burley, J.D., and Wofsy, S.C. (1997) Laboratory
investigation of the catalytic reduction technique for measurement of atmospheric
NOy. J. Geophys. Res. 102,10,759-10,776. ,
Nielsen, T. (1995) Observations of particulate organic nitrates and unidentified
components of NOy. Atmos. Environ. 29,1757-1769.
-214-
-------
Parrish, D.D., Buhr, M.P., Trainer, M., Norton, R.B., Shimstock, J.P., Fehsenfeld, F.C.,
Anlauf, K.G., Bottenheim, J.W., Tang, Y.Z., Wiebe, H.A., Roberts, J.M., Tanner,
R.L., Newman. L., Bowersox, V.C., Olszyna, K.J., Bailey, E.M., Rodgers, M.O.,
Wang, T., Berresheim, H., Roychowdhury, U.K., and Demerjian, K.L. (1993) The
total reactive oxidized nitrogen levels and the partitioning between the individual
species at six rural sites in eastern North America. J. Geophys. Res. 98,2927-
2939.
Ridley, B.A. (1991) Recent measurments of oxidized nitrogen compounds in the
troposphere. Atmos. Environ. 25A, 1905-1926.
Roberts, J.M. (1990) The atmospheric chemistry of organic nitrates. Atmos. Environ.,
Part A, 24, 243-287.
Wang, T, Carroll, M.A., Albercock, G.M., Owens, K.R., Duderstadt, K.A., Markevitch,
A.N., Parrish, D.D., Holloway, J.S., Fehsenfeld, F.C., Forbes, G., and Ogren, J.
(1996) Ground-based measurements of NOX and total reactive oxidized nitrogen
(NOy) at Sable Island, Nova Scotia, during the NARE 1993 summer intensive. J.
Geophys. Res. 101, 28,991-29,004.
Williams, E.J., Roberts, J.M., Bauman, K., Bertman, S.B., Buhr, S., Norton, R.B., and
Fehesenfeld, F.C. (1997) Variations in NOy composition at Idaho Hill, Colorado.
J. Geophys. Res. 102 6297-6314.
Williams, E.J., Bauman, K., Roberts, J.M., Bertman, S.B., Norton, R.B., Fehsenfeld,
F.C., Springston, S.R., Nunnermacker, L.J., Newman, L., Olszyna, K., Meagher,
J., Hartsell, B., Edgerton, E., Pearson, J.R., and Rodgers, M.O. (1998)
Intercomparison of ground-based NOy measurement techniques. J. Geophys. Res.
103 22,261-22,280.
Winer A.M., Peters, J.W., Smith, J.P., and Pitts, J.N. Jr. (1974) Response of commercial
chemiluminescent NO-NOX analyzers to other nitrogen-containing compounds.
Environ. Sci. Technol. 8, 1118-1121.
-215-.
-------
Table 1. (NOy)j converter efficiency for ammonia (56ppb) for a variety of (NOy)i
analyzers with molybdenum-based converters.
% NH3 Conversion
14.8%
26.8%
5.2%
4.5%
10.4%
10.4%
7.7%
26.6%
5.2%
9.5%
2.5%
11.2%
Converter Temperature, °C
315
313
360
376
359
374
361
314
370
365
301
Average
Figure 1. Response of (NOy)i analyzers to synthetic polluted air generated in a smog
chamber.
417198
250.0
200.0
150.0
100.0
I
50.0
l-TDLAS, NO2,Cal1
-*-LANM. NOx,Cal1
-+-STt_2, NOy.Call
-+-STI_1,NOy,Cal1
-+-BANN, NOy.Call
BARSTOW, NOy,Cal1
-K-SVAL, NOy.Call
SOLM, NOy.Call
~—UCDC, NOy.Call
«*- AZSA, NOy,Cal1
LANM, NO.Call
-60.0
-216-
-------
Figure 2. Nitric acid channel (NA) response (ppb) to filtered ambient air with nitric acid
removed.
10.0 -
I*
-10.0 -f
-15.0
7/3/98 7/4/98 7/5/98 7/6/98 7/7/98
i 00:00 00:00 00:00 00:00 00:00
7/8/98
00:00
-217-
-------
Instrumentation for State-Of-The-Art Aerosol Measurements in Smog Chambers
David Cocker, Nathan Whitlock, Don Collins, Jian Wang, Rick Flagan, John Seinfeld
Aerosol smog chamber studies are performed to elucidate the chemical and physical
processes that occur in the atmosphere leading to the formation of fine paniculate matter. In
order to obtain this goal, the number concentration, size distribution and chemical composition of
secondary aerosol produced in an environmental chamber must be monitored as it evolves with
time. An ideal instrument would be able to obtain both the chemical and physical properties of
the aerosol at time scales comparable to the evolution of the aerosol in the chamber at infinite
resolution. A road map on the available instrumentation and the desired measurements for an
environmental chamber is laid out. Critical environmental control parameters are also discussed
as are different techniques to ensure that the reactions and secondary aerosol formation in the
environmental chamber can occur at time scales and size ranges that can be easily detected by
current instrumentation. Also mentioned are some of the new techniques being implemented at
Caltech to increase the accuracy and precision of measurements made in our chamber as well as
improvements in the time and spatial resolution of our measurements.
There are two pathways for secondary organic aerosol (SOA) formation. The first is
supersaturation of semi-volatile organics derived from gas phase oxidation of precursor
hydrocarbons which lead to homogeneous nucleation; secondly, simple heterogeneous
condensation which can occur on aerosol surfaces already present in the environment.
Homogeneous nucleation and the resulting growth immediately following the nucleation can be
extremely fast processes, often occurring on time scales much less than 1 minute. In order to
monitor the early evolution of a homogeneous, nucleation aerosol, the aerosol instrumentation
would have to be much faster than any aerosol measurement devices currently available.
Moreover, homogeneous nucleation generates molecular clusters and extremely fine aerosol
particles, below the size range typically obtained by available aerosol instrumentation. The
difficulties created by nucleation events can, however, be avoided. By providing a sufficiently
large surface area in the form of a nonvolatile, inorganic aerosol, homogeneous nucleation can be
suppressed as semi-volatile organic compounds partition onto those surfaces. This process
occurs as semi-volatile organics achieve a large enough saturation to form an initial organic layer
around the inorganic salt. Subsequent growth depresses the supersaturation below the level that
would lead to homogeneous nucleation. The growth of the seed particles is slower than that of
the nuclei so it can be followed with available instrumentation. Moreover, the smallest particles
are then determined by the seed particles, so all particles can be measured. This process is quick;
typical times for full growth range from 30 minutes to 12 hours for conditions in the Caltech
environmental chamber.
Choosing the best instrument for the measurement of the physical and chemical
processes occurring the chamber involves looking at the time scales and spatial distributions of
the aerosol inside of the chamber. Instrumentation used currently for particle size distribution
measurements includes the stepping mode differential mobility particle sizer (DMPS), the
scanning electrical mobility spectrometer (SEMS), the aerosol mass spectrometer (AMS), the
optical particle counter (OPC), and the micro-orifice uniform deposit impactor (MOUDI). Of
these instruments, the SEMS system provides us the best picture of the physical processes
occurring in the chamber because of its fast response and good size resolution in the submicron
range that dominates the aerosol dynamics. The SEMS can provide size and number
distributions on time scales of less than 1 minute over 3 decades of size with reasonable
resolution. While the OPC can also obtain data at this high rate, the extinction coefficient and
scattering properties of each individual aerosol must be known for high accuracy. Moreover, the
-218-
-------
OPC can only size particles larger than about 0.1 3m (some extend down to 0.08 ~m). Recently
developed AMS instruments, which have seen limited use in chamber studies to date, enable
chemical analysis of the aerosol, but the time resolution of these instruments is longer than the
time scale for growth occurring in the chamber. Moreover, most versions of the AMS can only
measure relatively large particles. Lower bounds on AMS measurements range from 0.04 to 0.3
Im aerodynamic diameter. The stepping mode DMA requires several minutes to measure a size
distribution, to large to follow the fast aerosol dynamics encountered in chamber studies.
MOUDI sampling times required to collect sufficient material for measurement are even larger.
Hence, the SEMS provides the best real-time measurements of aerosol physical properties for
smog chamber experiments.
Additional physical issues for environmental chamber studies of SOA include control of
temperature and relative humidity, wall effects, and homogeneity or heterogeneity of chemical
and physical aerosol properties during measurement of SOA. The vapor pressures of semi-
volatile compounds that contribute to aerosol formation are strong functions of temperature, so
the temperature of the chamber must be well characterized. Moreover, the temperature at the
point of measurement can bias the results, particularly when instrumentation is located outside of
the chamber. Another important control parameter is the relative humidity. Variations in
relative humidity strongly influences particle size and may perturb the partitioning of reaction
products between the gas and aerosol phases. The chamber volume should be large to achieve a
small surface area to volume ratio in order to minimize particle wall loss and wall effects. The
chamber should not be so large that poor mixing or thermal stratification leads to a
heterogeneous reaction environment, however.
Optimization of the SEMS enables even more powerful analysis of the physical
processes occurring in the chamber. Caltech has fully automated the SEMS system to acquire
data and process size and number distributions in real time. A computer is used to control the
flows and the voltage of the system, making any necessary adjustments 100 times a second. A
proportional-integral-differential (PID) controller maintains the flow rates of the instrument to to
within +/- 0.2%. Fast inversions allows for the data to be deconvoluted, stored and displayed in
graphical form in times of less than 1 second allowing the experimenter to follow the evolution
of the aerosol in real-time. Minimization of the lengths of tubing reduces the fluid-mechanical
smearing time of the measurements. Full size distributions are obtained over a size range
extending from 28 to 800 nm are obtained each minute.
Caltech has further modified the SEMS system to increase both the range of
measurement of the SEMS as well as the time of the scan. The size range for the SEMS
instrument has been limited at the low end by resolution and on the high end by the breakdown
voltage of air at high voltages. However, by increasing or decreasing the flow rate of the DMA,
smaller and larger diameter particles can be measured respectively. Caltech has developed a
scanning flow system that can increase the dynamic range of the measurement. While the
technique has the drawback of a more complex deconvolution of the data involving back-
calculating the trajectories of the particles to determine the system transfer function, size
distributions over greater dynamic ranges can be achieved. A limitation to the temporal
resolution of the instrument has been the delay times between classification and detection of
particles in the DMA. A majority of the delay time is currently caused in the system by delays in
the measuring time of the CNC. Caltech has developed a fast CNC that reduces the delay time of
measurement of the CNC by two orders of magnitude allowing for size distributions to be
achieved in less than two seconds.
While SEMS can provide us with adequate physical characterization of the aerosol on
the time scales of heterogeneous condensation, it is unable to provide us with chemical
-219-
-------
composition of the aerosol. Ideally, chemical compositions should be determined sufficiently
rapidly to identify any changes in composition during the evolution of the aerosol, but none of
the available techniques are capable of such time resolution. At present, we use denude/filter
apparatus to separate the gas-phase and aerosol-phase. Derivatization coupled with GC-MS
analysis facilitates identification and quantification of the chemical composition of the aerosol.
Given the low reactant concentrations in environmental chamber studies, we can only collect
bulk samples, i.e., with no size classification, one sample per hour for off-line analysis.
The tandem differential mobility analyzer (TDMA) can be used for faster inference of
changes in the chemical or physical properties of the aerosol. The technique uses a DMA to
select particles of a given size. Those particles are then subjected to an environmental
perturbation that may cause them to change size, e.g., humidification to induce growth of
hygroscopic particles. One can deduce the hygroscopicity or other properties of the aerosol, and
determine whether all particles have similar properties or whether particles with different
compositions have been produced.
Through control of the temperature and humidity of the environmental chamber, the use
of the SEMS the TDMA, and denuder/filter sampling for chemical analysis of gases and
particles, the physical growth of particles and the chemical composition of the aerosol can be
obtained. This instrumentation enables investigation of the physical and chemical
transformations occurring in the environmental smog chamber.
-220-
-------
USE OF A THERMAL DESORPTION PARTICLE BEAM MASS SPECTROMETER FOR
STUDIES OF SECONDARY ORGANIC AEROSOL FORMATION
Herbert J. Tobias, Kenneth S. Docherty, Derek E. Beving, and Paul J. Ziemann*
Air Pollution Research Center and Department of Environmental Sciences,
University of California, Riverside, CA 92521
ABSTRACT
As part of a research program focusing on studies of the chemistry of gas-to-particle
conversion, we have recently developed a new instrument for particle chemical analysis. This
instrument, which we refer to as a thermal desorption particle beam mass spectrometer
(TDPBMS), can be used for real-time, quantitative analysis of the components of organic
particles, at least within the -0.02-0.5 micrometer size range. We have also developed a
temperature-programmed TDPBMS technique to aid in compound identification. Here we
describe the operation of the TDPBMS and present results from our recent application of
TDPBMS to studies of the chemistry of secondary aerosol formation, in which we have analyzed
the composition of aerosol particles formed in environmental chamber reactions of 1-tetradecene
and ozone in the presence of alcohols, carboxylic acids, and water vapor.
INTRODUCTION
Current understanding of secondary organic aerosol formation has been developed
principally from environmental chamber studies of reactions of single VOCs with single or
multiple oxidants. Most of these experiments have only included analyses of VOC reactants and
particle size distributions, which are valuable for quantifying aerosol yield (1-4), but provide no
information on chemical processes. In some experiments the chemical composition of particles
collected by filtration or impaction have been determined by gas chromatography-mass
spectrometry (GC-MS) of solvent-extracted components, and have provided important insight
into chemical mechanisms of aerosol formation (5-8). However, while this approach yields
valuable uiformation, the technique is time consuming and is prone to sampling artifacts (9, 10).
Furthermore, many of the polar and labile compounds formed are not readily amenable to gas
chromatography without prior derivatization (8). The technique also does not yield the real-time
information necessary to follow aerosol formation processes in detail. Therefore, although much
is known about the gas-phase kinetics of the initial reactions of VOCs, and a number of studies
have provided uiformation on the volatile products and mechanisms of these reactions (11-13,
and references therein), little is known about the identity of the gaseous organic products which
undergo nucleation or condensation to form aerosol.
As part of a research program focusing on studies of the chemistry of gas-to-particle
conversion, we have recently developed a new instrument for real-time particle chemical analysis
which should help to provide some of the needed compositional information on secondary
-221-
-------
organic aerosols. We have demonstrated that this instrument, which we refer to as a thermal
desorption particle beam mass spectrometer (TDPBMS), can be used for real-time, quantitative
analysis of the components of organic particles, at least within the ~0.02-0.5 um size range (14).
We have also developed a temperature-programmed TDPBMS technique (TPTD) to aid in
compound identification (15). Here we describe the TDPBMS we have constructed and present
the results of detailed characterization studies which are necessary for understanding the
performance of the instrument. We also describe techniques we have developed for accurately
calibrating the instrument for quantitative analysis of organic particles, and the TPTD technique
for compound identification. Use of the TDPBMS for aerosol analysis is demonstrated in an
environmental chamber study of the chemistry of secondary aerosol formation from reaction of
1-tetradecene and ozone in the presence of 2-propanol.
EXPERIMENTAL
Aerosol Mass Spectrometric Analysis by TDPBMS and TPTD. Detailed descriptions
of the TDPBMS and its operation for real-time analysis (14) and temperature-programmed
thermal desorption (TPTD) (15) are presented elsewhere. The TDPBMS and associated
apparatus are shown hi Figure 1. Aerosol is sampled into the TDPBMS through a 100 um
orifice, which maintains the flow at 0.075 L/min and reduces the pressure from atmospheric to
~2 torr. Particles then enter a tube containing a series of aerodynamic lenses (16, 17), which
focus the particles into a very narrow, low-divergence particle beam that transports -0.02-0.5 um
particles from atmospheric pressure into the high-vacuum chamber with near-unit efficiency.
The operation of the lenses is simulated in Figure 2 for gas molecules and particles. After exiting
the aerodynamic lens nozzle, particles pass through two flat-plate skimmers separating three
differentially-pumped chambers and enter the detection chamber where the pressure is ~5 x 10"8
torr. The vacuum is maintained by turbomolecular pumps mounted on each chamber and backed
by an oil-free mechanical pump to reduce contaminating organic vapors in the system. Inside the
detection chamber particles impact on the walls of a V-shaped molybdenum foil (volume ~0.1
cm3), which is either resistively heated continuously at 165 ± 3 °C for real-time TDPBMS
analysis or cooled to -50°C by an external liquid nitrogen bath for collection of particles for
TPTD. The vaporization cell temperature is monitored by an attached thermocouple and
regulated by a temperature controller. After vaporization the molecules diffuse into an ionizer
where they are impacted by 70 eV electrons, and the resulting ions are mass analyzed in a
quadrupole mass spectrometer (Extrel MEXM 500, 1-500 amu mass range) equipped with a
conversion dynode/pulse counting detector.
Particle analysis by TPTD is carried out on ~1 ug of aerosol collected on the
cryogenically-cooled vaporizer. Samples are desorbed by heating at a ramp rate of ~l°C/min for
about two hours, while mass spectra are continuously recorded. During TPTD the aerosol
components desorb according to their vapor pressures, so mass spectra of individual compounds
can be extracted from time-dependent mass spectra.
Aerosol Generation and Environmental Chamber Technique. Organic aerosols of
desired compositions used for instrument characterization were generated using a Collison
atomizer. An ~0.1% (w/w) solution of the organic compounds in 2-propanol was atomized to a
mist using clean air and sent through a diffusion drier to evaporate the alcohol solvent. The
aerosol then flows through a bipolar charger and differential mobility analyzer (DMA) (18) to
-222-
-------
obtain near-monodisperse, size-selected particles for our experiments. Particles used here were
~0.15 um in diameter. Particle concentrations were measured using an aerosol electrometer.
The products of the liquid- and gas-phase reactions of 1-tetradecene and ozone in the
presence of 2-propanol were also analyzed by TPTD. The reaction was conducted in the liquid-
phase by bubbling a 1.5 L/min flow of 2% O3/O2 through a 2% (w/w) solution of 1-tetradecene in
2-propanol for 2 hours. The reacted solution was then used to create a monodisperse aerosol as
described above. The gas-phase reaction leading to formation of secondary organic aerosol was
carried out in an -7000 L Teflon environmental chamber filled with clean air and 0.3 ppm 1-
tetradecene, 1.3 ppm ozone, and 2000 ppm 2-propanol. The propanol scavenges >95% of the
OH radicals formed in the reaction (19), thereby simplifying the aerosol products by eliminating
reactions between 1-tetradecene and OH. After one hour of reaction, aerosol consisting of-0.1-
0.3 mm sized aerosol particles was sampled for -20 min into the TDPBMS without size selection
for cryogenic collection and subsequent TPTD.
RESULTS AND DISCUSSION
Particle Sampling and Vaporization. In order to calibrate the TDPBMS for
quantification of organic compounds it is necessary to characterize particle transport and
vaporization in the instrument. The efficiencies with which particles are sampled from
atmospheric pressure and transported into the unheated vaporization cell in the TDPBMS was
evaluated in a series of measurements using monodisperse aerosol particles of various sizes and
compositions. The results of these experiments are shown in Figure 3. The stated diameters are
for singly charged particles, but there is a small contribution (less than 5%) to the measurements
from larger, doubly and triply charged particles. Over the range of particle diameters from 0.02-
0.5 um the transport efficiencies were greater than 40% for all particles, but differed with particle
size and composition. The efficiency is a reflection of the width of the particle beam, with wider
beams having lower efficiencies because of particle losses at the skimmers and the entrance to
the vaporization cell. Particle losses are small between the DMA and the particle beam lens.
The decrease in efficiency with decreasing particle size is due to broadening of the beam by
Brownian motion of the particles in the lenses and nozzle, and the decrease at larger particle
sizes is probably due to decreased focusing efficiency by the lenses. Particle composition
influences the efficiency through its effect on particle shape. Lift forces acting on nonspherical
particles during nozzle accelerations cause broadening of the particle beam (16, 17), therefore,
the efficiencies are highest for spherical and highly symmetric particles, and lower for those with
irregular shapes (20). The highest efficiencies measured here were for dioctyl sebacate (DOS) [-
(CH2)4CO2CH2CH(C2H5)CH2)3CH3]2 particles, since they are liquid drops. Glutaric acid
[HOOC(CH2)3COOH], which is a solid dicarboxylic acid, also has high efficiencies, suggesting a
spherical or regular shape. The lowest efficiencies were measured for palmitic acid
[CH3(CH2)i4COOH], which is a solid, long-chain monocarboxylic acid that apparently forms
irregularly shaped particles. However, the presence of DOS in a 1:1 mixture with palmitic acid
leads to a more spherical particle, as reflected in the increased efficiencies. The efficiency curve
for adipic acid [HOOC(CH2)4COOH], which is also a solid dicarboxylic acid, increases with
particle size to physically impossible values higher than 100%. For particles larger than those
shown in Figure 3 they go well off scale. The reason for this is that some of the particles are
bouncing out of the vaporization cell with a charge that is different from when they entered,
resulting in an apparent transport efficiency greater than 100%. The effect of bounce on these
-223-
-------
measurements increases with increasing particle size because larger particles bounce more
readily, and they can carry away more charge. As one might expect, when adipic acid is mixed
with an equal mass of DOS, the measured efficiencies become more reasonable and are close to
those of a liquid drop, indicating that the DOS prevents the particles from bouncing and makes
the particle more spherical.
Although hard, crystalline particles such as (NH4)2SO4 and adipic acid can bounce out of
an unheated vaporization cell, mass spectrometric measurements made with a heated cell
demonstrate that particles completely vaporize before bouncing out. This can be seen from the
data in Figure 4, which show the mass spectral signal measured for a constant current of
monodisperse DOS and (NH4)2SO4 particles of various sizes (the quantity plotted in Figure 4 is
actually an "effective" single-particle volume that has been corrected for contributions from
larger, multiply-charged particles). The linear relationship is indicative of complete evaporation,
since bounce and incomplete evaporation would increase with particle size, leading to a less than
linear increase in mass spectral signal with increasing particle volume.
Particle Mass Spectral Analysis: Compound Quantification and Identification.
Quantification of a compound present in an air sample requires the determination of a
relationship between the mass spectral signal and the mass concentration of the compound (e.g.
ug/m of air). In the ideal case that the calibration and sample particles are spherical, the -
calibration curve would be similar to those shown in Figure 4, where the upper x-axis gives the
mass concentration of particulate compound calculated for a measured input concentration of
single-component, monodisperse particles obtained from the atomizer and DMA [both DOS and
(NH4)2SO4 particles are spheres]. If the calibration particles are not spherical, then the calculated
mass concentration will be hi error. For organic compounds that are solids, and may therefore
form nonspherical particles, a low vapor pressure organic liquid such as DOS or oleic acid
[CH3(CH)7CH=CH(CH2)7COOH] can be added to the atomizer solution to create
multicomponent spherical particles. The mass concentration of the compound of interest can
then be calculated from the relative concentration of solutes. However, this approach requires
that the mass spectrometer signal not depend on the particle matrix, but only on the quantity of
calibration compound present in the particles. The data shown in Figure 5, which are the results
of mass spectral measurements made on pure and multicomponent particles, demonstrate that
this is the case for the TDPBMS. For mixed DOS/oleic acid (liquid/liquid) and DOS/tridecanoic
acid (liquid/solid) particles, the signal measured per mass of compound sampled (uncorrected for
transport efficiency) for the multicomponent particles is within ~5-10% of the pure particle
signal, regardless of the composition. This result is reasonable, since as long as the particles
evaporate completely, have similar transport efficiencies, and have spherical shapes, then the
signal obtained when a given mass of compound is sampled into the instrument should be the
same for pure and multicomponent particles. The transport efficiencies of 0.1 um DOS/oleic
acid and 1:1 DOS/tridecanoic acid particles are 100% and 95%, respectively, indicating that both
types of particles are spherical or nearly spherical in shape.
Once a calibration curve such as those shown in Figure 4 is determined using spherical
particles, the mass spectral signal can be used to determine the concentration of compound
present in an air sample by multiplying the concentration obtained from the calibration curve by
the ratio of the transport efficiencies of sample and calibration particles. Based on the
uncertainties in the measured quantities we estimate a total uncertainty in the concentration of
-224-
-------
~20%. This value is quite acceptable for our application of the technique, but as we show below,
attaining this level of accuracy in environmental chamber studies will depend on our ability to
associate mass spectral peaks with particular compounds present in particles.
Multicomponent Aerosol TPTD. The basis for using mass spectrometry for chemical
analysis of particles is that each compound has a unique mass spectrum, "a fingerprint," which
can be compared with a particle mass spectrum to determine if the compound is present in the
particle. However, the TDPBMS mass spectra of the environmental chamber particles we
analyze are a composite of the mass spectra of all the individual particle components, as is shown
in Figure 6 for a reaction of 1-tetradecene and ozone. For this reason we developed a TPTD
technique for extracting single-compound information. During TPTD the aerosol is collected
and then slowly heated so that components desorb according to their vapor pressures. Mass
spectra of individual compounds can then be extracted from time-dependent mass spectra.
Figure 7 depicts example mass thermograms for two-component (A) and three-component (B)
test aerosols. Thermogram B resulted from a mixed aerosol of adipic acid, DOS, and stearic acid
[CH3(CH2)i6COOH] analyzed using a temperature ramp of ~0.75°C/min. For this mixture, m/z
100, 185, and 284 are specific markers for adipic acid, DOS, and stearic acid, respectively, and
indicate that these compounds thermally desorb over 10-20 minutes at this ramp rate. This is
reflected in the m/z signal traces, which in this case have well-resolved maxima. If these were
unknown compounds, it would be obvious that m/z 100, 185, and 284 should be assigned to three
different mass spectra. The m/z 60 signal would be assigned to the same mass spectra as m/z 100
and 284. Thermogram A shows similar results for a two-component aerosol consisting of
glutaric acid [HOOC(CH2)3COOH] and DOS.^
Our data reduction technique allows the creation of single-compound mass spectra from
all m/z signals bracketed within user-specified time or temperature ranges, which can be visually
determined from mass thermograms (Figure 7) or plots of peak desorption temperatures (or
times) versus m/z, as illustrated in Figure 8. Figure 9 presents examples of the mass spectra of
glutaric acid (A) and DOS extracted from the TPTD analysis of the two-component aerosol,
which match the spectra available for these compounds in the Wiley Database (not shown). The
spectra acquired for pure glutaric acid (B) and DOS (D) using the continuous vaporization mode
of TDPBMS, which are similar to the respective Wiley Database spectra, are shown in Figure 9
for comparison. These examples demonstrate how TPTD can be used to extract many of the
fragment ion peaks associated with an individual compound present in a mixture, and create a
mass spectrum for use in compound identification. Although mass thermogram peak shapes are
generally broad, component mass spectra can sti|J,be extracted as long as there is adequate
separation of peak maxima. The limitations of TPTD for compound separation were investigated
by analyzing aerosol mixtures comprised of components with various degrees of vapor pressure
differences at different temperature ramp rates. The results indicate that a minimum time
separation of ~4 minutes (or ~3-4°C) at temperature ramp rates of
-------
TPTD of Ozone-Alkene Reaction Products. The aerosol generated from the gas-phase
reaction of 1-tetradecene [CH3(CH2)i iCH=CH2] with ozone in the presence of excess 2-propanol
was sampled from an environmental chamber to demonstrate the utility of TPTD in laboratory
studies. After collection of polydisperse aerosol, TPTD at ~1.3°C/min resulted in the separation
and identification of two compounds that desorbed at 24°C and 38°C (Figure 10). The mass
spectra are presented in Figure 11. The higher vapor pressure aerosol component that desorbed
first (A) was identified as tridecanoic acid by comparison with the Wiley Database spectra (not
shown) and the pure tridecanoic acid TDPBMS spectra (B), and is an expected product of this
reaction. Mechanisms of the liquid- and gas-phase reactions of ozone with alkenes have been
reported (21, 22), and for 1-tetradecene, ozone adds to the >C=C< bond to yield an energy-rich
primary ozonide as depicted in Figure 12. This ozonide then decomposes to create formaldehyde
and an energy-rich Criegee biradical. Tridecanoic acid is formed by rearrangement of the excited
biradical. Although the ozonide can also decompose to tridecanal and a one-carbon biradical,
this pathway leads to products that are too volatile to form aerosol. A second, slightly less
volatile component (Figure 11C) was found using TPTD, but the extracted mass spectrum had no
satisfactory match in the Wiley Mass Spectral Database. However, based on the results of recent
gas-phase studies on ethene (22), and liquid-phase studies on numerous other alkenes (21, 23,
24), it is expected that the major product of ozonolysis of 1 -tetradecene in the presence of excess
2-propanol will be cc-isopropoxytridecyl hydroperoxide formed by the mechanism shown in
Figure 12. This compound would most likely have a lower vapor pressure than tridecanoic acid,
conforming with the TPTD results.
To test this hypothesis, a solution of 1-tetradecene in 2-propanol was ozonated and the
product solution was atomized and analyzed by TDPBMS. The liquid-phase reaction is known
to produce ct-alkoxyalkyl hydroperoxides in near-quantitative yield (23), so the excellent
agreement between the TPTD spectrum of chamber aerosol and the liquid-phase mass spectrum
(Figure 1 ID) is strong evidence that compound B is a-isopropoxytridecyl hydroperoxide. It is
worth noting that hydroperoxides such as this may not be amenable to GC analysis (we have
observed that they thermally decompose) and would therefore be overlooked using traditional
GC-MS identification procedures. Moreover, this environmental chamber analysis serves as one
example where TPTD is successful in identifying two aerosol components present in
concentrations differing by approximately an order of magnitude.
CONCLUSIONS
The results of this work demonstrate that the thermal desorption particle beam mass
spectrometer (TDPBMS) we have developed can be used for quantitative, real-time
measurements of organic compounds present in secondary aerosol formed in a laboratory
environment. The TDPBMS is capable of analyzing particles between ~0.02-0.5 um as long as
the mass concentration of the compound is greater than ~0.1-1 ug/m3, but this size range can be
extended in both directions for higher concentrations. The estimated uncertainty in concentration
measurements is ~20%. The instrument can be calibrated using monodisperse aerosol particles
of known composition, size, and concentration, generated using an atomizer and differential
mobility analyzer. Because the mass spectral signal measured for a given mass of a particular
compound is independent of the presence of other compounds in the particle, calibrations can be
performed using pure or multicomponent particles. However, calibrations will be most accurate
when the particles are spherical, since this allows a more accurate calculation of the particle mass
-226-
-------
from the measured mobility diameter. When calibrating with solid compounds, which can form
nonspherical particles, it is preferable to mix the compound of interest with a low vapor pressure
organic liquid such as DOS or oleic acid to make spherical, multicomponent particles.
Quantification by TDPBMS is currently limited to species with vapor pressures less than ~10"5
torr, since particle evaporation introduces error in the measurement.
The TPTD technique presented here is potentially a powerful tool for identification of
aerosol components in environmental chamber studies of atmospheric chemistry, and
complements the quantitative TDPBMS methods. When compared to other analytical methods,
this procedure minimizes sample handling since aerosol is sampled directly into the analysis
chamber over a few minutes, and reduces the possibility for decomposition of unstable reaction
products since evaporation is accomplished at low temperatures in a high-vacuum chamber.
When analyzing aerosols of equal-mass mixtures of compounds, a vapor pressure difference of a
factor of 5 or more at a temperature ramp rate of ~0.75°C/min gives sufficient separation for
accurate mass spectral identification. This is adequate resolution for many applications since, for
example, there is approximately a factor of 5 decrease in vapor pressure for each additional CH2
unit in a homologous series of monocarboxylic acids (neglecting odd-even differences in vapor
pressure) or «-alkanes. When used alongside TDPBMS, TPTD can provide valuable information
in the study of the chemistry of secondary organic aerosol formation.
ACKNOWLEDGMENTS
The authors thank the U.S. Environmental Protection Agency, Office of Research and
Development [Assistance Agreement R82-6235-010, Science to Achieve Results (STAR) grant]
and the University of California Toxic Substances Research & Teaching Program for generously
supporting this research. While this research has been supported by the U.S. Environmental
Protection Agency, it has not been subjected to Agency review and, therefore, does not
necessarily reflect the views of the Agency, and no official endorsement should be inferred. We
also thank Janet Arey and Roger Atkinson for helpful discussions.
REFERENCES
(1) Seinfeld, J. H.; Pandis, S. N. Atmospheric Chemistry and Physics; John Wiley & Sons;
New York, 1998.
(2) Odum, J. R.; Hoffmann, T.; Bowman, F.; Collins, D.; Flagan, R. C.; Seinfeld, J. H.
Environ. Sci. Technol. 1996, 30, 2580.
(3) Odum, J. R. Jungkamp; T. P. W., Griffin, R. J.; Forstner, H. J. L.; Flagan, R. C.;
Seinfeld, J. H. Environ. Sci. Technol. 1997, 31, 1890.
(4) Hoffman, T.; Odum, J. R.; Bowman, F.; Collins, D.; Klockow, D.; Flagan, R. C.;
Seinfeld, J. H. J. Atmos. Chem. 1997, 26, 189.
(5) Grosjean, D. In Ozone and Other Photochemical Oxidants; Chapter 3, National
Academy of Sciences: Washington DC, 1977; pp. 45-125.
(6) Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1997, 31, 1345.
(7) Forstner, H. J. L.; Flagan, R. C.; Seinfeld, J. H. Atmos. Environ. 1997, 31, 1953.
(8) Yu, J.; Flagan, R. C.; Seinfeld, J. H. Environ. Sci. Technol. 1998, 32, 2357.
(9) Ligocki, M. P.; Pankpw, J. F. Environ. Sci. Technol. 1989, 23, 75.
(10) McDow, S. R.; Huntzicker, J. J. Atmos. Environ. 1990, 24A, 2563.
(11) Atkinson, R. /. Phys. Chem. Ref. Data 1989, Monograph 1, 1.
-227-
-------
(12) Atkinson, R. J. Phys. Chem. Ref. Data 1994, Monograph 2, 1.
(13) Atkinson, R. /. Phys. Chem. Ref. Data 1997, 26, 215.
(14) Tobias, H. J.; Kooiman, P. M.; Docherty, K. S.; Ziemann, P. J. Aerosol Sci. Technol.
1999, in press.
(15) Tobias, H. J.; Ziemann, P. J. Anal. Chem. 1999, 71, 3428.
(16) Liu, P.; Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. Aerosol Sci. Technol. 1995,
22,293.
(17) Liu, P.; Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. Aerosol Sci. Technol. 1995,
22,314.
(18) Kinney, P.D.; Pui, D.; Mullholland, G.W.; Bryner, N.P. J. Res. Natl. Inst. Stand.
Technol. 1991, 96, 147.
(19) Atkinson, R.; Ashmann, S.M.; Arey, J.; Shorees, B. J. Geophys. Res. 1992, 97, 606.
(20) Ziemann, P. J.; Kittelson, D. B.; McMurry, P. H. J. Aerosol Sci. 1996, 27, 587.
(21) Bailey, P.S. in Ozonation in Organic Chemistry, Volume 1, Academic Press: London,
1978.
(22) Neeb, P.; Horie, O.; Moortgat, O.K. Int. J. Chem. Kinetics 1996, 28, 721.
(23) Zelikman, E. S.; Yur'ev, Y. N.; Berezova, L. V.; Tsyskovskii, V. K J. Org. Chem. USSR
(Engl. Trans/.) 1971, 7, 641.
(24) Pospelov, M. V.; Menyailo, A. T.; Bortyan, T. A.; Ustynyuk, Yu. A.; Petrosyan, V. S.
J. Org. Chem. USSR (Engl. Transl.) 1973, 9, 312.
-228-
-------
Session IV
Environmental Chamber Studies
Session Chair
Rich Kamens
-------
AEROSOL FORMATION FROM THE REACTION OF oc-PINENE AND OZONE USING
A GAS PHASE KINETICS-PARTICLE PARTITIONING MODEL
R.M. Kamens, M. Jaoui, S. Lee, C.J.Chien
Department of Environmental Sciences and Engineering,
University of North Carolina, Chapel Hill, NC 27599-7400, USA
e-mail: kamens@.unc.edu
fax:919966-7911
11-17-99, Bangkok, Thailand
ABSTRACT
A kinetic mechanism was used to describe the gas-phase reactions of cc-pinene with ozone.
This reaction scheme produces low vapor pressure reaction products that distribute between gas
and particle phases. Partitioning was treated as an equilibrium between the rate of particle uptake
and rate of particle loss of semivolatile terpene reaction products. Given estimated liquid vapor
pressures and activation energies of desorption, it was possible to calculate gas-particle
equilibrium constants and aerosol desorption rate constants at different temperatures. Gas and
aerosol phase reactions were linked together in one chemical mechanism and a chemical kinetics
solver was used to predict reactant and product concentrations over time. Aerosol formation from
the model was then compared with aerosol production observed from outdoor chamber
experiments. Approximately 10-40% of the reacted cc-pinene carbon appeared in the aerosol
phase. Models vs. experimental aerosol yields illustrate that reasonable predictions of secondary
aerosol formation are possible from both dark ozone and light/NOx a-pinene systems.
INTRODUCTION
The atmospheric chemistry of non-methane biogenic hydrocarbons has received much
attention because of their significant global emissions, high photochemical reactivity, and their
high aerosol forming potential. Although the potential of aerosol formation from terpenes was
noted as early as 1960 by Wentl, the magnitude of the natural contribution by biogenics to the
particulate burden hi the atmosphere is still not well characterized. In this paper, we will describe
the feasibility of a predictive technique for the formation of secondary aerosols from biogenic
hydrocarbons. An advantage of this approach is that it has the ability to embrace the range of
different atmospheric chemical and physical conditions that bring about secondary aerosol
formation.
EXPERIMENTAL SECTION
Gas-particle samples for this study were generated in a large outdoor 190 m3 Teflon film
chamber (Kamens et al.,1995, Fan et al., 1996). All experiments were carried out under darkness
to exclude photochemical effects. Rural background air was used to charge the chamber without
not any additional injections of oxides of nitrogen. Secondary aerosols were created by the
-230-
-------
reaction of a-pinene with Os in the chamber. Os from an electric discharge ozone generator was
added to the chamber over the course of an hour with initial concentrations ranging, depending on
the experiment, from 0.25 to 0.65 ppm. It was measured using a Bendex chemilumenesent Os
meter (model 8002, Roncerverte,WV) and calibrated via gas phase titration using a NIST traceable
NO tank. Os addition to the chamber was followed by volatilizing, depending on the experiment,
0.4 to 1 mL of liquid a-pinene into the chamber atmosphere. The gas-phase concentration of cc-
pinene was monitored with an online gas chromatograph (Shimadzu Model 8A, column: 1.5 m, 3.2
mm stainless steel packed with Supelco 5% Bentanone 34) using a flame ionization detector, and
calibrated with a known concentration of a mixture of toluene and propylene.
Gas and particle phase a-pinene products were simultaneously collected with a sampling
train that consisted of an upstream 5-channel annular denuder, followed by a 47mm Teflon glass
fiber filter (type T60A20, Pallflex Products Corp., Putnam, CT, USA) and another denuder. In one
of the experiments, a parallel sampling system consisting of a filter, followed bv a denuder was
also used. The details of the sample workup procedure and quantitative analysis have been
reported in previous manuscripts Kamens et al., 1995 and Fan et al., 1996). Carbonyl products of
cc-pinene-Ch aerosols were derivatized by O-(2,3,4,5,6-pentafluorobenzyl)hydroxyl-amine
hydrochloride (PFBHA) as described by Yu et al.,1997, and carboxylic acids, using
pentafluorobenzyl bromide (PFBBr) as a derivatizing agent as described by Chien et al., (1998).
(±)-a-Pinene, O-(2,3,4,5,6-pentafluorobenzyl)hydroxylamine hydrochloride (PFBHA),
pentafluorobenzylbromide (PFBBr), decafluorobibenzyl (internal standard for derivatization), cis-
pinonic acid, n-hexanoic acid, n-octanoic acid, hexane-l,6-dionic acid, and heptane-1,7-dionic acid
were all purchased from Aldrich (Milwaukee, WI).
In one of the experiments the particle size distribution and aerosol concentration for
particles ranging from 0.018 to 1.0 urn were monitored by an Electrical Aerosol Analyzer (EAA)
(Thermo Systems, Inc., Model 3030, Minneapolis, MN). Total aerosol number concentration was
also measured by a Condensation Nuclei Counter (CNC, Model Rich 100, Environment One
Corp., Schenectady, NY).
A GAS-PARTICLE MECHANISM
Scheme 1
JU
J^ + CO, HQ.OH.H.
O8^ CHO
norpfnonaldenyde*t
Criegeel
norpinonic
add
a-pinene
COOH
pinonic acid
As a result of new aerosol compositional information
(Jang and Kamens, 1999, Yu et al., 1998) we have
developed an exploratory model for predicting aerosol
yields from the reaction of a-pinene with ozone in the
atmosphere. Reaction pathways (Scheme 1 and 2) were
constructed from experimentally measured products
which include: pinonaldehyde, norpinonaldehyde,
pinonic acid, norpinonic acid, pinic acid, 2,2-
dimethylcyclobutane-l,3-dicarboxylic acid, and hydroxy
and aldehyde substituted pionaldehydes and hydroxy
pinonic acids. To simplify the mechanism in this study,
six generalized semivolatile products were defined: 1.
"pinald" to represent pinonaldehyde and norpinonaldehyde, 2. "pinacid" to represent pinonic and
norpinonic acids, 3. "diacid" for pinic acid and norpinic acid, (2,2-dimethylcyclobutanel,3-
dicarboxylic acid), 4. "oxy-pinald" for hydroxy and aldehyde substituted pinonaldehydes (called
oxo-substituted), 5. "P3" for 2,2-dimethylcyclobutyl-3-acetylcarboxylicacid, (a pinic acid
precursor), and 6. "oxy-pinacid" for hydroxy and aldehyde substituted pinonic acids. A last
group, frag, was employed to account for volatile oxygenated products.
CriegeeZ
+ other
products
pinic acid
-231-
-------
Scheme 2
Cadlacld
A kinetic mechanism was used to
describe the gas-phase reactions of cc-pinene
with ozone. One of the products, pinic acid,
a dicarboxylic acid, has a subcooled liquid
vapor pressure of ~10"6 mmHg at 295K,
which may be low enough to initiate self
nucleation. More volatile products such as
pinonic acid and pinonaldehyde, will not
self-nucleate, but will partition onto existing
particles. Gas and aerosol phase reactions are
linked together in one chemical mechanism
and a chemical kinetics solver (Jeffries,
1991) is used to predict reactant and product
concentrations over time. Gas phase rate
constants are available for a number of these reactions (Atkinson, 1997). These reactions were then
added to the inorganic chemistry reactions of the Carbon Bond 4 photochemical mechanism [Gery
et al., 1989). Reactions of Criegee 1 and 2 biradicals in Scheme 1 & 2, and OH attack on a-
pinene are partially described below as:
oxo-ptnonaldehyde
hydnoxyplnonahfehycte hydroxyp|non|c ack)
Criegeel -> 0.3 pinonic aicd+ 0.3 pinald + 0.25 oxy-pinaldehyde +
0.15 stablecriegl + 0.8OH + 0.5HO2 (1)
stablecriegl + HiO •> pinonic acid (2)
pinald + OH {+O2} -> pinald-oo (3)
pinald-oo + HO2 •> pinonic acid + { Os} (4)
a-puiene + OH -> ap-oo (5a)
ap-oo + ap-oo -> 0.6*pinald + HCHO + 0.2*oxypinald (5b)
Criegee2 -^0.15 stabcrieg2 +0.3 crgprod2 + 0.3 HCHO (6)
+ 0.55 oxy-pinald+ 0.8 OH + 0.5 HO2
crgprod2 + HO2 -> diacid + HCHO (7)
stabcrieg2 + H2O -> P3 + CH3OH (8)
oxy-pinald + OH-> oxy-prepinacid (9)
oxy-prepinacid + HO2 -> oxy-pinonic acid + { Os} (10)
Partitioning is treated as an equilibrium between the rate of particle up-take and the rate of
particle loss of individual semivolatile ct-pinene reaction products. The ratio of the rate constants
for the forward and backward reactions 'ko/koff *s equal to the equilibrium constant, 'Kp, for the
-232-
-------
gas-particle equilibrium of a given partitioning compound i.
lKp = '
(11)
It is assumed that secondary aerosols are liquid-like, and 'Kp can be calculated from gas-liquid
(absorptive) partitioning theory, as described by Pankow (1994).
log 'Kp = - log 'P°L- log 'yom +constant
(12)
From estimated vapor pressures, 1P°L (Kamens et al, 1999) and calculated activity
coefficients, Vom, which are close to one in these aerosols (Jang et al., 1997) 'Kp can be calculated
for the reaction products. The rate constant, koff, for loss of a compound from the liquid particle
phase to the gas phase has the form koff = 3 e"Ea/R. |3 can be evaluated and the activation energy,
Ea estimated (Kamens et al., 1999). This then permits an estimate of the rate of absorption 'kon
from the gas phase.
In the model seed nuclei provide initial liquid volumes for gas-liquid partitioning. The
product compounds that actually participate in the self nucleation process are as yet unknown.
One possibility is that the stabilized Criegee radical (stabcriegl and stabcrieg2) reacts with the
carbonyl portion of product compounds to produce an extremely low pressure secondary ozonide
product. These types of products have been observed from an Os- propylene system by Niki and
co-workers (1997) and more recently by Neeb et al., 1995) for larger molecules]. It is also
possible to form an anhydride from the reaction of the Criegee with a dicarboxylic acid (Neeb et
al., 1996). In the a-pinene case, these products would have molecular weights of-350, and be of
extremely low volatility. By techniques described previously (Kamens et al., 1999) vapor
pressures lower than 10"15 torr are estimated, and ~one second after the start of the experiments in
this study, would exceed supersaturation by -1000 fold. This suggests these compounds would
most certainly self nucleate. In our mechanism these are called "seedl" particles. We assumed
that stabilized Criegees would react with all carbonyl products. For example:
stabcriegl + pinald -> seedl
(13)
This process was not very important for the experiments in this study, given the
background nucleation levels that resulted during the injection of a-pinene into the chamber.
Predictions, using the model developed here, however, suggest a strong influence of these seeds on
particle formation when a-pinene has concentrations of ~5 ppb.Gas phase products migrate to the
particle phase and then create more mass for additional partitioning.
Once seed surfaces are available, the migration of gas phase products such as pinonic acid
(pinacidgas) on to seedl particles to give pinacid particle phase (pinacidpart) and diacid gas product
on to the newly formed pinacidpart to form pinic diacid particle phase (diacidpart) can thus be
represented as:
pinacidgas+ seedl -> pinacidpart+ seedl
pinacidpart+ diacidgas ~> pmacidpart+ diacidpart
f =19.4 min1 (14)
koff= 68 min'1 (15)
To conserve mass, the amount of diacidpartand pinacidpart mass appears on both sides of the
equation. To maintain equilibrium, both diacidpart and pinacidpart back reacts or "off-gases" from
the particle to give back gas phase pinonic acid. A similar set of reactions can be written for all of
-233-
-------
the partitioning products.
pinacidpan -> pinacidgas koff = 3.73 x 1014 exp (-9525/T) min"1 (16)
-> diacidgas koff = 3.73 x 1014 exp (-10285/T) min1 (17)
MODEL AND CHAMBER DATA COMPARISONS
Aerosol formation from the model is compared with aerosol production determined from
190 m3 dark outdoor chamber experiments. The actual rate constants and mechanism used are
suimmarized in Table 1. The main updates since our first publication of the mechanism (Kamens
et al, 1999) are:
1. An increase in the rate constant for the OH attack on pinonaldehyde from 2.4x1 0"12 cm3
molecule"1 sec"1 (as per to Kwok and Atkinson, 1995) to their more recent value of -5x10"
12 cm3 molecule"1 sec"1.
2. We replaced the reaction
(X-pinene + OH -> 0.6 pinonald + 0.1 oxy-pinald + 0.3 frag
with
cc-pinene + OH -> ap-oo
ap-oo + ap-oo -> 0.6*pinald + HCHO + 0.2*oxypinald
This was done to more easily add NO oxidation reactions by the hydroxy-peroxy-radical,
ap-oo. In addition, the previously used lumped reaction may over estimate the
pinonaldehyde yield from OH attack on a-pinene in the absence of NOx.
3. We lowered the kon values for all absorbing species because it is more appropriated to
use the the average molecular weight of the aerosol in converting from Kp units of
Given the offsetting nature of the above changes, model simulations vs. experimental data
for a-pinene, Os and aerosol yields for a warm (295K) and a cool outdoor experiment (285K)
showed little difference (Figure 1) between our originally published simulations and the updated
mechanism (Kamens et al, 1999.) Approximately 20-40% of the reacted a-pinene carbon appears
in the aerosol phase. Under cool conditions (285K), both measurements and model predictions
suggest that pinonaldehyde is the major gas phase product, while pinic acid, pinonic acid, hydroxy-
pinonaldehydes, and hydroxy-pinonic acids are the major aerosol phase products. Under warm
conditions, the predicted major gas phase product are pinonaldehyde, hydroxy-pinonaldehydes,
some pinonic acid, and a trace of dicarboxylic acid. The particle phase was dominated by pinic
acid followed by pinonic acids, hydroxy-pinonaldehydes, and hydroxypinonic acids. The
important observation is that the aerosol composition seems to differ radically between warm and
cool experiments and this is consistent with the observations of Jang and Kamens (1999).
-234-
-------
0.4
0.3
0.1
a-pinene
7 7.5 8
Time in hours
8.5
7.9 8.1
time in hours
Reacted a-pinene
. A A A A A A A
3.5
m
g/ 3
m32.5
2
1.5
1
0.5
0
8.5 9 9.5
time in hours
10.5
.^IAA*^«» **AAAAAAA
y1^ Reacted a-pinene
/ . model
*r
I
Filter mass
7.5 8
8.5 9 9.5
time in hours
Figure 1. Model and experimental data from a warm (top and bottom left panels) and cool
(top and bottom right panels) outdoor chamber experiment performed under darkness
with a 3
10 10.5
200
180
160
140
120
100
I
80
60
40
20-
TSP data vs model
model
» data
22.5
tow titan
23.;
200
180
160
140
80
60
40
20
0-
200
data
i
E
model
235 21.0 .215 220 725 210 215 240
3
Figure 3. Particle data and model for the dark reaction of 0.20
ppmV a-pinene with 0.12 ppm Os at 297K.
Figure 4. Particle data and model for the dark reaction of 0.12
ppmV a-pinene with 0.06 ppm Oa at 299K.
-235-
-------
During the summer of 1999 two relatively low concentration experiments were conducted
and as shown in Figures 3 and 4 these data are fit reasonably well the model in its current form.
Given the above mechanism, is was not difficult to implement a day time model that would
involve the oxidation of NO to NO2 and the formation of PAH analogue compounds. What is
unknown is the quantum yields for propinaldehyde type compounds, so as a default, the quantum
yields and cross sections for methyglyoxal were used. Hydroxy-nitrated a-pinene products (OH-
apNO3) and PAN analog compounds were partitioned to the particle phase as per the methodology
described above. Although these reactions are not completely described in by the dark mechanism
in Table 1, our fit to a daytime experiment with 0.97 ppm of a-pinene and 0.44 ppnm NOx is
shown in Figure 5. It is clear that are current mechanism does not characterized all of the
processes involved in the oxidation of NOx and we also over predict the amount of ozone that is
formed. What is encouraging is that we show the same trend in the ozone behavior as the data, and
model gives a reasonable fit to the fitter based particle data (Figure 6). We also show reasonable
fits to product pinonaldehyde in both the gas and the particle phases. Product model data also
suggest that nitrated compounds can dominated the particle phase
e 7 a 9 10
time in hours
Figure 5. NOx and O3 data and model fit for the daytime
reaction of 0.0.97 ppmV a-pinene with 0.44 ppm NOx on
June 9,1999. Lines are the model and heavy lines-
symbols are the data
time in hours
Figure 6. Particle data and model for the daytime reaction
of 0.0.97 ppmV a-pinene with 0.44 ppm NOx on June 9,
1999.
Table 1. Gas and particle phase reactions used to simulate secondary aerosol formation.
gas phase reactions
mirr1 or ppm"1 min
1. a-pinene+ O3 -> .4 Criegeel + .6 Criegee2
2. Criegeel -> .3 pinacidgas + .15 stabcriegl + .8 OH + .5 HO2
+ .3 pinaldgas + .25 oxy-pinaldgas + .3 CO
3. Criegee2 -» .3 crgprod2 + .SSoxy-pinaldgas +.35 HCHO
4- .15 stabcrieg2 +.8 OH + .5 HO2
1.492 exp-732/T,
IxlO6,
1000,
-236-
-------
4. stabcriegl + H2O
5. stabcrieg2 + H2O
6.P3gas + OH
7. oxy-pinald + OH
8. predi-oo + HO2
9. crgprod2 + HO2
pinacidgas
(pinalic acid) + CH3OH
predi-oo
oxy-prepinacid
-> diacid,
-> diacid,
lgas
lgas
lO.oxy-prepinacid +HO2 -> oxy-pinacid
11. OH + apine -^ ap-oo
12. ap-oo + NO -» 0.85*NO2 + 0.67*pinald+ 0.13*oxypinald
+ 0.65*HO2 + 0.17*HCHO + 0.15*OH-apNO3+0.07*acetone
13. ap-oo + ap-oo -> 0.6*pinald +HCHO + 0.2*oxypinald
14. a-pinene + NO3 •> 0.7*pinald + 0.2*OH-apNO3
{+ 0.05*oxydiNO3}
15. pinald + OH -> pinald-oo
15. pinald-oo + HO2 -> pinacidgas
17. diacidgas {walls} ->
18. oxy-pinacidgas {walls} -^
19. pinacidgas {walls} -^
20. oxy-pinaldgas {walls} ->
21. pinaldgas {walls} ->
partitioning reactions
IxlO6,
6xlO"3,
72380,
72380, .
677 exp!040/T,
677 exp!040/T,
677expl040/T
17873 @444,
5482.1
4000,
!242
lU^£*j
544 0818,
72380,
677 exp!040/T,
6xlO"7 exp2445/T,
6xlO'7 exp2445/T,
4xlQ-7exp2445/T,
4xlO"7 exp2445/T,
2.5x10"7 exp2445/T,
22. stabcriegl + pinaldgas -> seedl
23. stabcrieg2 + oxy-pinaldgas -^ seedl
24. stabcrieg2 + HCHO -> oxy-pinald
25. pinacidgas + seedl -> seedl + pinacidpart
26. pinacidgas + diacidpart "^ diacidpart + pinacidpart
27. pinacidgas + seed -> seed + pinacidpart
28. pinacidgas + pinacidpart ~> pinacidpart + pinacidpart
29. pinacidgas + pinacidpart -^ pinacidpart + pinacidpart
30. pinacidgas + oxy-pinaldpart -> oxy-pinaldpart + pinacidpart
31. pinacidgas + P3part -> P3part + pinacidpart
32. pinacidgas + oxy-pinacidpart -^ oxy-pinacidpart + pinacidpart
33. pinacidpart "^ pinacidgas
34. diacidgas + pinacidpart ~> pinacidpart + diacidpart
35. diacidpart ~> diacidgas
36. pinaldgas + oxy-pinacidpart -> oxy-pinacidpart + pinaldpart
37. pinaldparT^pinaldgas
38. oxy-pinaldgas + P3part ~> P3part + oxy-pinaldpart
39. oxy-pinaldpart •> oxy-pinaldgas
40. p3gas + oxy-pinacidpart ~> oxy-pinacidpart + p3part
41. p3part -> p3gas
42. oxy-pinacidgas + seedl -> seedl + oxy-pinacidpart
43 oxy-pinaldpart -> oxy-pinacidgas
44. diacidpart {walls} ->
45 O3 {walls)} -^
29.5,
29.5
29.5,
19.4,
19.4,
19.4,
19.4,
19.4,
19.4,
19.4,
19.4,
3.73xl014 exp-9525/T
68
3.73xl014exp-10285/T,
4.2,
3.73xl014exp-8598/T,
14.1,
3.73xl014 exp-9341/T,
13,
3.73xl014exp-9282/T,
74,
3.73xl014 exp-10353,/T
0.0008,
0.0005,
*Rate constants are at 298K. To convert the 2nd order rate constant in cm3 molecule"1 sec"1 in the
text to ppm"1 min"1, divide the rate constant in cm3 molecule"1 sec"1 by 6.77xlO"16. **exp in the
temperature dependent reactions is the natural base e, in the rate equation, k = B e ~An. The gas
constant R is imbedded in A so that A = Ea/R. ***The partitioning reactions for pinald, oxy-pinald,
diacid, and oxy-pinacid, are the same as for thepinacid, but to save space, are only given for koff
and one particle species, stabcriegl and stabcrieg2 reactions are illustrated for pinacidsas, and oxy-
-237-
-------
and HCHO, but are the same for the other carbonyls. Loss of product gases to the
chamber walls were estimated from observed pyrene loss at 297 and 27IK. References for
reactions are give in Kamens et al., (1999) were determined in this study. All reactants were
diluted from the chamber at a rate determined with an SFe tracer. A typical loss was 0.0005 min"1,
Other specific measured or estimated losses like Oa, particles and cc-pinene product to the walls,
were adjusted for the SFe loss rate.
-238-
-------
REFERENCES
Atkinson, R. /. Phys. Chem. Ref. Data, 1997 ,26,215 -290
Kwok, E.; Atkinson, R. Atmos. Environ, 1995, 29, 1685-1695.
Chien, C. J.; Charles, M. J.; Sexton, K. G.; Jeffries, H. E. Environ. Sci. Technol. 1998, 32, 299-
309.
Fan, Z.; Kamens, R. M.; Hu, J.; Zhang, J. Environ. Sci. Technol. 1996, 30, 1359-1364.
Grosjean, D.; Willams II, E. L.; Grosjean, E. Environ. Sci. Technol. 1993, 27, 830-840.
Jang, M.; Kamens R. M. Atmos. Environ. 1999 , 33, 459-474
Jeffries, H. E. "PC-Photochemical Kinetics Simulation System (PC-PKSS), Software Version 3.0",
1991, Department of Environmental Sciences and Engineering, School of Public Health,
University of North Carolina, Chapel Hill, NC 27599.
Gery, M. W.; Whitten, G. Z.; Killus, J. P.; Dodge, M. C. J. Geophys. Res. 1989, 94,12925-12956.
Jang, M.; Kamens, R. M.; Leach, K.; Strommen, M. R. Environ. Sci. Technol. 1997, 31, 2805-
2811.
Kamens, R. M.; Odum, J.; Fan, Z. Environ. Sci. Technol. 1995, 29, 43-50.
Kamens, R. M.; Jang, M.; Leach, B. K.;Chien,C. "An Exploratory Kinetics and Gas-Particle
Partitioning Model for Aerosol Formation From cc-pinene-O3 systems", Environ Sci and
Technol., in press May, 1999.
Lyman, W. J.; Reehl, W. F.; Rosenblatt, D.H; Handbook of Chemical Property Estimation
Methods, Environmental Behavior of Organic Compounds, American Chemical Society,
Wash. D.C., 1990
Moortgat,G.K.; Veyret, B.; Lesclaux,R. Chem. Phys. Let., 1989, 160,443-447.
Niki, H.; Maker, P. D.; Savage, C. M., Breitenbach, L. P. Chem. Phys. Let, 1977,46, 327-330.
Neeb, P.; Horie,O.; Moorgot, G. K. Tetrahedron Let., 1996, 37, 9297-9300.
Neeb.P.; Horie,O.; Moortgat. O.K. Chem. Phys Lett., 1995,150-156
Pankow, J. F. Atmos. Environ. 1994, 28, 185-188.
Yu, J.; Jeffries, H. E.; Le Lacheur, R. M. Environ. Sci. Technol. 1995,100, 1923-1932.
Went, F. W., Nature 1960,187, 641 -643.
Yu, J.; Flagan, R.C.; Seinfeld J.H. Envrion. Sci. Technol. 1998, 32,2357-2370.
-239-
-------
Secondary Aerosol Formation From Biogenic Precursors
David Cocker, Robert Griffin, Jian Yu, Nathan Whitlock, Rick Flagan, John Seinfeld
Global biogenic emissions from natural and agricultural plants of non-methane
hydrocarbons are an order of magnitude higher than anthropogenic emissions of organic
carbon. Biogenic hydrocarbons, composed mainly of isoprene and monoterpenes, react
quickly with atmospheric oxidants to form ozone and atmospheric aerosols. The
secondary organic aerosol yields of 14 biogenic precursors are investigated here.
Individual contributions of the nitrate radical, ozone and the hydroxyl radical to SOA
formation are deconvoluted. Additionally, the products of a-pinene and A3-carene dark
ozonolysis reactions are presented here. For more detail on biogenic SOA yields or
procedures for analysis of the chemical composition of SOA products, please refer to the
following papers:
References
Griffin, R. J., D. R. Cocker III, R. C. Flagan, and J. H. Seinfeld, Organic Aerosol
Formation from the Oxidation of Biogenic Hydrocarbons, JGR-ATMOF, 20 Feb. 1999,
104:(D3) 3555-3567
Yu, J., D. R. Cocker, R. J. Griffin, R. C. Flagan, and J. H. Seinfeld, Gas-Phase Ozone
Oxidation of Monoterpenes: Gaseous and Particulate Products. J. Atmospheric
Chemistry, Oct. 1999, V34(2) 207-258
-240-
-------
Smog Chamber Studies of Particle Formation from the Oxidation of
Terpenes
Jens Hjorth, European Commission, TP 272, Environment Institute,
JRC-Ispra, 1-21020 Ispra (VA), Italy
Important aspects of the current research on atmospheric aerosols have recently been
highlighted by Andreae and Cratzen (1997). The authors state that secondary organic
aerosols, formed by the oxidation of VOCs emitted from biogenic sources, probably account
for a significant fraction of the particulate matter in rural and remote continental air masses.
Monoterpenes (Ci0Hi6), which constitute one of the main fractions of the biogenic VOCs,
have since long been recognized as potential precursors of secondary organic aerosol in the
atmosphere (Went, 1960). Aerosol yields from 5 -100 % have been measured in laboratory
and environmental chamber studies, e.g. Hatakeyama et al. (1989); Hoffmann et al. (1997);
Kamens et al. (1999); Pandis et al. (1991); Schuetzle and Rasmussen (1978); Zhang et al.,
(1992), depending on the experimental conditions and on the species investigated. The present
study aims at providing more information on aerosol yields from terpene oxidation and on the
main chemical components in the particles.
Experimental
Studies of aerosol formation were carried out in a large outdoor smog chamber
(EUPHORE, Valencia, Spain) as well as in large Teflon bags and in a smaller (480 L)
chamber facility in Ispra, Italy. The study was mainly focussed on cc-pinene, 3-pinene and
limonene. The EUPHORE facility comprehends two hemispheric outdoor smog chambers,
each with a diameter of ~ 9 m and a volume of ~ 180 m3, both cooled at the floor. A detailed
description of the design and the analytical equipment of EUPHORE has been published by
Becker (1996).
The experiments included studies of ozonolysis reactions under dark conditions with
terpene concentrations in the range between 10 and 1000 ppbV in dry air, humidified air and
with or without addition of cyclohexane as a scavenger for OH radicals. Photo-oxidation of
terpenes was studied in EUPHORE by exposing mixtures of terpenes and NOx with initial
terpene concentrations of 100 ppbV and initial NOx-concentrations of 20 or 200 ppbV to
sunlight. The oxidation of terpenes in air, initiated by OH radicals, was also studied by
producing OH via the photolysis of methylnitrite in the chamber facility in Ispra.
The aerosol size distributions presented here were measured by a custom made
Vienna-type differential mobility analysis (DMA) system (Winkelmayr et al., 1991) coupled
to a condensation nuclei counter (TSI, model 3010). The system measured aerosol size
distributions in the range between 7 and 500 nm and was operated in dry-air conditions, thus
measuring the size of dry particles.
The hygroscopicity of aerosol particles was measured using a Tandem Differential
Mobility Analyser (TDMA) set-up where growth of particle diameter by uptake of water was
observed (Virkkula et al., 1999). Aerosol concentrations were corrected for wall-losses based
on the observed rate of decrease in particle number concentration at a late stage in the
experiments, when particle formation had ceased to take place.
Terpene concentrations were measured by gas chromatography with FID detection.
Carbonyl compounds were measured by sampling on DNPH-coated cartridges followed by
HPLC measurement of the hydrazones formed. Particles were collected on Teflon filters. The
smog-chamber aerosol extracts were analyzed either by GC-MS after derivatisation of
carboxylic acids by methylation or by a newly developed HPLC-MS" method, which has
proved efficient for polar terpene oxidation products (Glasius et al., 1999).
-241-
-------
Aerosol yields
Aerosol mass concentrations were estimated by assuming a density of 1 g/cm for the
particles that were formed. Aerosol yields were found to increase with increasing aerosol
volume, as illustrated in Fig. 1 for the case of cc-pinene. This observation is in qualitative'
agreement with theoretical work on the partitioning of semi-volatile organic species between
the gas phase and a condensed organic matter phase (Odum et al., 1996; Pankow, 1994),
cc-pinene ozonolysis with cyclohexane added
60-
50-
40-
2 30-
5? 20-
10-
0-
-10
% Yield vs. concentration
- Best fit to data
—i—
500
1000
1500
2000
2500
3000
fig/m3 aerosol
Fig. 1. Measured aerosol yields vs. measured aerosol concentrations compared to
curve obtained by applying a best fit of the unknown parameter in Equation I.
which indicates that the concentration dependence of the aerosol yield in a situation with
equilibrium partitioning between the gas and particle phase can be expressed by the equation
(I) Y=ZYi=ME(aiKomJ)/(l+Kom,iM),
where Y is the total aerosol yield, Y; is the aerosol yield of species i, M is the absorbing
organic aerosol mass concentration, CC; is the total yield of species i. K^i is the partitioning
coefficient for species i, which depends on its vapour pressure, its activity coefficient in the
condensed organic phase and on its molecular weight when the aerosol yield is expressed on a
weight-basis as it is in this work. Y^,m,\ (m3/|ig) is defined as c; aer/(M Cj g), where Cj aer and q g
are the aerosol and gas phase concentrations (ug /m3) of species i. Kom,i also depends on the
temperature, but the relatively small temperature variations (around 293 K) within these
experiments had no evident influence on the aerosol yields.
By applying a best fit to the unknown parameters in Equation I, an estimate of the
aerosol yields (on a mass-basis) under realistic, ambient conditions could be made. For an
initial aerosol concentration (M) of 2 (ig/m3 the following yields could be calculated: cc-
pinene 3.3 %, (3-pinene 1.8 %, and limonene 2.7 %. For an initial aerosol concentration of M
-242-
-------
= 10 Hg/m3, the following values were calculated: cc-pinene 8.7 %, p-pinene 6.4 % and
limonene: 10.6 %.
The aerosol yields that were measured in this study may be compared to results
presented in the literature, e.g. to the recent, comprehensive study by Hoffmann et al. (1997).
The aerosol yields from the ozonolysis of a- and p-pinene reported in this study seem to be
somewhat higher than those found in the present study, when results obtained under similar
conditions are compared.
A comparison between aerosol yields obtained by photo-oxidation experiments and
ozonolysis experiments showed significantly lower aerosol yields from photo-oxidation,
suggesting that the OH radical reaction, gives significantly lower aerosol yields than
ozonolysis. This observation is illustrated in Fig. 2, which also shows a comparison between
the aerosol yields from the ozonolysis reaction, performed under different conditions.
Concerning the comparison between photooxidation and ozonolysis it should, however, be
noticed, that the ozonolysis was studied in the absence of NO*, while NOX (particularly NO)
certainly will influence the degradation pathways of the terpenes in the photo-oxidation runs.
Thus also the influence of [NOJ on aerosol formation may be a reason for the differences
observed between the ozonolysis and photo-oxidation experiments.
Application of the tandem-DMA set-up to the study of the hygroscopic growth of
terpene-derived aerosol from ozonolysis as well as photooxidation experimental runs proved
that these were only slightly hygroscopic (Virkkula et al., 1999). The experiments were
initiated by introducing an ammonium sulphate seed aerosol. The hygroscopic growth factor
(i.e. (particle diameter in humid air)/(particle diameter in dry air)) was approximately 1.5 for
inn
sn -
P 70-
2 50-
o>
SAT\ -
HU
o
E
10
o-
c
- -
"" ~- • H
— ••
- - I , £
• -7."
- ' ,l'"'
*• ^' '
TT* „ x
• • v X
«-|X v^ v X x
'V'1 •' A
1 A
X XX
V
) 20 40 60 80 100 120
Consumed alpha-pinene (ppbV)
i alpha-pinene+ozone •
- added water A
x Photooxidation
added cyclohexane
Photooxidation
Fig. 2. Aerosol formation from ozonolysis in dry air, ozonolysis with added water,
ozonolysis in dry air with cyclohexane added as an OH scavenger and in two photo-oxidation
experiments (initial NOX was 20 ppbV).
the ammonium sulphate aerosol at 84 % RH. As the oxidation of terpenes took place and the
oxidation products were condensing on the seed particles, the growth factor decreased to
-243-
-------
around 1.10. No significant differences between the different terpenes and the different
reaction conditions (ozonolysis versus photo-oxidation) could be observed.
Chemical analysis of aerosols.
Chemical analyses of particles formed by oxidation of terpenes have been performed
and a number of components (particularly organic acids) have been identified. Dicarboxylic
acids, such as pinic acid, appear to be a characteristic constituent of terpene-derived aerosol
resulting from ozonolysis as well as photo-oxidation (Christoffersen et al., 1997, Glasius et
al., 1998). Such compounds were identified as products of a-pinene, p-pinene, limonene, 3-
carene and sabinene. The measured yields varied from below 1% up to as much as 6 %,
depending on the precursor-terpene and the reaction conditions. Also keto-acids and hydroxy-
acids, together with dicarbonyl compounds and hydroxy-dicarbonyl compounds were found
to be significant constituents of aerosol particles collected. Dicarboxylic acids were formed
also in the experimental runs where cyclohexane was added as an OH-radical scavenger in
ozonolysis experiments, thus demonstrating that there are relevant reaction pathways leading
to formation of these species that do not involve OH radical reactions.
Acknowledgements:
The following JRC-colleagues have contributed to this work:
Niels R. Jensen, Bo R. Larsen, Rita Van Dingenen, Dario DiBella, Marianne Glasius, S0ren
S0rensen, Aki Virkkula
Other participants to EUPHORE experiments were
Heike Plagens, Markus Spittler, Lars Ruppert
Physikalische Chemie, Bergische Universitat-GH Wuppertal, Germany
Osamu Horie, Peter Neeb, Richard Winterhalter
MPI fur Chemie, Atmospheric Chemistry Division, Mainz, Germany
Klaus Wirtz
Centre de Estudios Ambientales del Mediterraneo, Valencia, Spain
This work was funded within the EU shared cost actions BIOVOC and NUCVOC
References:
Andreae, M.O. and Crutzen, P.J., 1997. Atmospheric aerosols: Biogeochemical sources and role in
atmospheric chemistry. Science, 276(5315): 1052-1058.
Becker, K.H., 1996. The European photoreactor EUPHORE.
Christoffersen, T.S. et al., 1998. cfr-Pinic acid, a possible precursor for organic aerosol formation from
ozonolysis of alpha-pinene. Atmos. Environ., 32(10A): 1657-1661.
Glasius, M., Duane, M. and Larsen, B.R., 1999. Determination of polar terpene oxidation products in
aerosols by liquid chromatography-ion trap mass spectrometry. J. Chromatography, A 833:
121-135.
Glasius, M. et al., 1998. Carboxylic acids in secondary aerosols from 03 and OH oxidation of cyclic
monoterpenes. In: P.M. Borrell and P. Borrell (Editors), EUROTRAC-2 Symposium 1998,
Garmisch-Partenkirchen, Germany.
Hatakeyama, S., Izumi, K., Fukuyama, T. and Akimoto, H., 1989. Reactions of ozone with alpha-
pinene and beta-pinene in air: Yields of gaseous and paniculate products. Journal of
Geophysical Research, 94(D10): 13013-13024.
Hoffmann, T. et al., 1997. Formation of organic aerosols from the oxidation of biogenic hydrocarbons.
J. Atmos. Chem., 26: 189-222.
Kamens, R.M., Jang, M., Chien, C.-J. and Leach, K., 1999. Aerosol formation from the reaction of cc-
pinene and ozone using a gas-phase kinetics-aerosol partitioning model. Environ. Sci.
Technol., 33(9): 1430-1438.
Odum, J.R. et al., 1996. Gas/particle partitioning and secondary organic aerosol formation. Environ.
Sci. Technol., 30(8): 2580-2585.
Pandis, S.N., Paulson, S.E., Seinfeld, J.H. and Flagan, R.C., 1991. Aerosol formation in the
photooxidation of isoprene and B-Pinene. Atmospheric Environment, 25A(5-6): 997-1008.
Pankow, J.F., 1994. An absorption model of the gas/aerosol partitioning involved in the formation of
secondary organic aerosol. Atmos. Environ., 28(2): 189-93.
-244-
-------
Schuetzle, D. and Rasmussen, R.A., 1978. The molecular composition of secondary aerosol particles
formed from terpenes. Journal of the Air Pollution Control Association, 28(3): 236-240.
Virkkula, A., van Dingenen, R., Raes, F. and Hjorth, J., 1999. Hygroscopic properties of aerosols
formed by oxidation of limonene, cc-pinene, and p-pinene. JGR, 104(D3): 3569-3579.
Went, F.W., 1960. Blue hazes in the atmosphere. Nature, 187: 641-643.
Winkelmayr, W., Reischl, G.P., Lindner, A.O. and Berner, A., 1991. A new electromobility
spectrometer for the measurement of aerosol size distributions in the size range from 1 to 1000
nm. J. Aerosol Sci., 22: 289-296.
Zhang, S.H., Shaw, M, Seinfeld, J.H. and Flagan, R.C., 1992. Photochemical aerosol formation from
alpha -pinene and beta -pinene. Journal of Geophysical Research, 97(D18): 20717-29.
-245-
-------
Determination of Photolysis Frequencies and Quantum Yields for
Small Carbonyl Compounds using the EUPHORE Chamber
Klaus Wirtz
Centre de Estudios Ambientales del Mediterraneo
Parque Tecnologico, C.\ Charles R. Darwin 14,46980 Paterna, Valencia, Spain
Abstract
Small aldehydes are formed during the photochemical oxidation of many VOC's,
olefins and terpenes in urban as well as in rural areas. Photolysis and reaction with the
OH radical are the most important initiation reactions for the removal of these
compounds conducting to the formation of peroxy radicals and in the case of photolytic
decomposition, either to stable molecules or free radicals.
The photolysis frequencies for various small aldehydes were measured in the
EUPHORE Smog Chambers by photolysing the aldehydes with natural sunlight. The
actinic flux during the experiment was measured with a spectroradiometer. The decay of
aldehydes and formation of products were analysed by FTIR spectroscopy, gas
chromatography and HPLC. The major products are explained in the case of
acetaldehyde, propionaldehyde and i-butyraldehyde by a mechanism involving a primary
dissociation step which leads to the formation of free radicals. The product analysis for
photolysis experiments of butyraldehyde, pentanal, 2-Methylbutyraldehyde and 3-
Methylbutyraldehyde indicates that for these molecules two primary photodissociation
steps occur which gives either stable molecules or free radicals. Integrated quantum
yields can be calculated from the ratio of the theoretical photolysis frequency using the
measured radiation data and known absorption cross-sections by assuming a quantum
yield of unity and the measured photolytic decay rate. The results obtained can be
employed in numerical models which describe the tropospheric degradation of these
compounds in order to assess the importance of the additional radical production on the
atmospheric oxidation capacity and ozone formation potential of the precursors VOCs.
Experimental
The experiments were performed in the outdoor Smog Chamber in Valencia,
Spain. The reactor consists of a half spherical FEP (fluorine ethene propene) foil which
is highly transparent to the short wavelength sunlight in the UVB with a total volume of
about 195 m3. A detailed description of the photoreactors is given in Becker, 1996.
Small quantities of the corresponding aldehyde were introduced by an air stream into the
reactor and photolysed during most experiments in the presence of an OH tracer. A
tracer was used instead of an OH radical scavenger because the addition of high
concentration of an organic compound like cyclohexane would saturate a huge part of
the IR bands and influence the product analysis. SFe as inert inorganic compound was
introduced to measure the dilution throughout the experiment by analysing the collected
FTIR spectra.
-246-
-------
Apparatus
A FTIR spectrometer, NICOLET Magna 550, was operated with a liquid nitrogen
cooled MCT detector with a resolution of 1 cm"1. The optical path length was 553.5 m
and the spectra were recorded every 10 minutes by co-adding 550 interferograms.
Two different gas chromatographs were used to measure either the decay of the
reactants or the formation of the products. The first chromatograph, a Fisons GC-8000
was equipped with a 30 m DB-624 fused silica capillary column (J&W Scientific, 0.32
mm id, 1.8 fim film) and operated in a constant pressure mode. Two detectors were used
in series, FED and PED (GC-PED), to obtain the chromatographic signals. The second
chromatograph, a Fisons Trace-Gas-Analyser (TGA), which incorporates a cryogenic
enrichment trap, was connected to a FID detector. 200 cm3 air were collected in a
sampling loop at 120 °C and passed to a micro trap with Tenax cooled to -120°C with
liquid N2. The injection onto the chromatographic column (30 m DB-1, J&W Scientific,
0.25 mm id, 1.0 Jim film) in splitless mode is achieved by a rapid heating of the micro
trap to 240 °C.
The hydroperoxide analysis was based on the reaction of H2O2 and organic
peroxides with /7-hydroxyphenylacetic acid (POPHA) to produce a fluorescent dimer
(6,6'-dihydroxy-3,3'-biphenyldiacetic acid) using peroxidase as an enzyme catalyst
(Gab et al., 1995). The sampled hydroperoxides were separated on a Superspher 60 RP-
Select B column using an Hewlett-Packard 1050 Series isocratic pump at a flow rate of
0.5 mL/min with H3PO4 buffer as mobile phase adjusted to pH=3.5 containing 4.9xlO"8
mol/L of H2O2 to condition the column. The fluorescence of the biphenylic derivative
was measured at an excitation wavelength of 285 nm and an emission wavelength of
410 nm. Detector response was proportional to the individual hydroperoxide
concentration; quantification of the chromatographic signals was made injecting liquid
H2O2 standards prepared from serial dilutions of H2O2 stock standards (Fluka, Sigma-
Aldrich) assuming a similar response for the alkyl hydroperoxides. Air samples were
collected with a flow rate of 2 L/min using the stripping technique (Lazrus et al., 1986;
Lee et al., 1995). H2O2 and organic peroxides were stripped from the air into the
collection solution using a continuous flow glass scrubbing coil. Collection solution (a
H3PO4 buffer adjusted to pH=3.5) was pumped at 0.43 mL/min with a peristaltic pump
(205S-BA Watson-Marlow). About 0.2 mL volume samples were taken in vials in
periods varying from 5 to 15 min and analysed immediately by the HPLC-Fluorescence
technique.
The light intensity was measured during the experiments with a calibrated
spectroradiometer Bentham DM300. Special designed measurement heads with an
uniform sensitivity with respect to the incident angle of the solar light are coupled
through a quartz fibre bundle to the entrance optics of the monochromator. Two light
beams, one for the direct light and the other for the reflected light, passed
simultaneously but geometrical separated through the double monochromator.
Independent detectors measure the light for both beams. The spectra were recorded
every 5 min in the range from 290 nm to 520 nm with a spectral resolution of 1 nm
FWHM.
-247-
-------
Results and Discussion
Determination of Photolysis Frequencies and Effective Quantum Yields
All experiments were conducted at least for several hours at midday using the
highest sun light intensity in the absence of nitrogen oxides, NOX. In' order to evaluate
the photolysis frequency the loss rates obtained from the decay of the individual
aldehydes (PBD-GC data) were corrected for the dilution. There was no evidence found
from the decay rates of the tracer compounds that OH radical reactions may influence
the calculated photolysis frequencies. The measured decay rates of the tracers overlap
within the error limits with the dilution rate of the chamber. Different tracers were tested
and selected to have a high OH rate constant whereby other loss processes like wall
deposition or ozone reaction should be negligible. Isoprene as tracer has the highest OH
rate constant and could be measured with high accuracy by GC as well as by FTIR, but
the decay was influenced by the formed ozone. Cyclohexane showed the same
properties with respect to the GC detection but due to the lower OH rate constant only
high OH radical levels >105 /cm3 could be detected. Most experiments were performed
using di-n-butyl ether as tracer which does not show a significant wall deposition and
does not react with O3. The detectable OH radical level with this tracer is in the range of
5xl04/cm3.
Table 1: Photolysis frequencies and calculated effective quantum yields.
Compound
Acetaldehyde
Propionaldehyde
Butyraldehyde
Isobutyraldehyde
Pentanal
2-Methyl
butyraldehyde
3-Methyl-
butyraldehyde
2-Pentanone
Photolysis
Frequency
s'1
2.9xlO'6
l.lxlO'5
l.lxKT5
3.3xlO"5
1.6xlO'5
3.8xlO'5
1.3xlO'3
0.6xlO'6
Photolysis
Frequency
J(N02)
s-1
lO.TxlO'3
9.12xlO'3
9.92xlO"3
8.90xlO'3
9.64xlO'3
9.26x1 O'3
8.82xlO'3
7.03xlO'3
Theoretical
Loss Rate
Q-yield 1
s-1
4.92xlO'5
3.57xlO'5
5.18xlO'5
4.67x1 0'5
5.52xlO'5
5.22xlO'5#
4.69xl(T5#
0.86xl(T5
Effective
Quantum
yield
0.06
0.31
0.21
0.70
0.29
0.72
0.27
0.07
Literature
Data on
Quantum
yields
0.4 300nm&
0.1 320nm
0.12 &*
Absorption cross section from Isobutyraldehyde used.
& Atkinson etal. (1989)
-248-
-------
** Mean 300-320nm, T.J. Cronin and L. Zhu, 1998
The effective quantum yields were calculated as the ratio between the measured
photolysis frequency and the theoretical photolysis frequency assuming a quantum yield
of unity all over the absorption region of the individual carbonyl compound. The
spectroradiometer data on the actinic flux was used to evaluate the theoretical value,
which is the upper limit for the photo decomposition, according to the formula:
k (photolytic decay) = E I(A,) o(A.) (j)(A.) (s'1),
with I(A,) the actinic flux, (measured by Bentham DM300) in photons cm"2 s"'in the
wavelength interval AX, centered at A., absorption cross section a(X) base e in cm2
molecule"1 averaged over the wavelength interval AA, and centered at A, (Martinez et al.
1992) and quantum yield
-------
CH3CHO
Propionaldehyde
CH3CH2CHO
Butyraldehyde
CH3CH2CH2CHO
Isobutyraldehyde
CH3CH(CH3)CHO
Pentanal
CH3CH2CH2CH2CHO
2-Methylbutyraldehyde
CH3CH2CH(CH3)CHO
3-Methylbutyraldehyde
CH3CH(CH3)CH2CHO
2-Pentanone
CH3CCO)CH2CH2CH3
CO
H202
Methylhydroperoxide
Formaldehyde
CO
Acetaldehyde
Ethylhydroperoxide
H202
Formaldehyde
CO
Acetaldehyde
Propionaldehyde
1-Propanol
Ethene
n-Propylhydroperoxide
H202
CO
Acetaldehyde
Acetone
Isopropanol
Isopropylhydroperoxide
H202
Acetaldehyde
Butyraldehyde
Propene
n-Butylhydroperoxide (minor)
CO(?)
Acetaldehyde
CO
2-ButyIhydroperoxide
Ethene
Propene
Acetaldehyde
Isobutyraldehyde
Acetone
CO
i-Butylhydroperoxide
Ethene
Propionaldehyde
Propylhydroperoxide
Methylhydroperoxide
FTIR
HPLC-Fluorescence
HPLC-Fluorescence
FTIR
FTIR
FTIR, GC-PED
HPLC-Fluorescence,
FTIR
HPLC-Fluorescence
FTIR
FTIR
GC-PDD
GC-PID
GC-TGA
GC-PID, GC-TGA
HPLC-Fluorescence
HPLC-Fluorescence
FTIR, CO-Monitor
GC-PED, FTER
GC-PID
GC-TGA
HPLC-Fluorescence
HPLC-Fluorescence
GC-PID
GC-PID
GC-TGA, CG-PED
HPLC-Fluorescence
CO-Monitor
GC-PED
CO-Monitor
HPLC-Fluorescence
GC-TGA
GC-PED
GC-PED
GC-PED, GC-TGA
GC-TGA
CO-Monitor
HPLC-Fluorescence
GC-TGA
GC-TGA
HPLC-Fluorescence
HPLC-Fluorescence
In bold are the major products, underlined are products arising from a molecular channel.
For acetaldehyde, propionaldehyde and i-butyraldehyde all products could be
observed which can be expected from the peroxyradical cross reactions RO2+ RO2 and
RO2 + HO2 after dissociation into free radicals R + HCO. The photo decomposition of
butyraldehyde, pentanal, 2-Methylbutyraldehyde and 3-Methylbutyraldehyde yields
-250-
-------
stable molecules and free radicals. The maximum alkylhydroperoxide concentrations
measured at the end of the experiment are summarised in Table 3.
Table 3: Maximum values measured at the end of the photolysis experiments
Carbonyl
Compound
Acetaldehyde
Propionaldehyde
Butyraldehyde
i-Butyraldehyde
Pentanal
2-Methyl-
butyraldehyde
3-Methyl-
*
butyraldehyde
2-Pentanone
H202
ppb
4.7
12.2
4.5
10.0
1.2
5.7
3.5
2.5
MHP
ppb
18.62
1.5
0.4
4.9
'
4.6
EHP
ppb
153.6
1.9
36.1
0.4
PHP
ppb
47.5
7.0
i-PHP
ppb
700
i-BHP
ppb
14.23
1-BHP
ppb
3.5
2-BHP
ppb
243.7
Carbonyls which decompose into stable molecules and free radicals are marked
A numerical simulation was performed using the Facsimile code given in the MCM-
Master-Mechanism (M.Pilling, University of Leeds) in order to obtain the branching
ratio for the primary photo dissociation step for butyraldehyde. According to the
simulation results branching ratios of 0.78 for the free radical channel and 0.22 for of
the photo decomposition into stable molecules, ethene and acetaldehyde, can be
calculated. The comparison between the experimental data and the simulation is shown
in Figure 2. Using this branching ratio and the effective quantum yield measured for n-
butyraldehyde a quantum yield of 0.16 for the free radical channel and 0.05 for the
molecular channel can be determined.
-251-
-------
3.5E+13 •
• 5.0E+12
4- 4.5E+1Z
- 5.0E+11
O.OE+00
O.OE+00
Sim. Butyraktehyde
A ButyraldehydeTGA
i o Butyraldohyde GC-PID
i + Butyroldahyde FT1R
—Sim. Ethene
» Ethene TGA
—— Sim. Acetakletiyde
a AcetaldehydeTGA
—Sim. Propyl hydroperoxide
£ Propyl hydroperoxide HPLC
—Sim. Propionaidehyde
» Propionaidehyde HPLC
• Propionaidehyde FUR
5000 10000 15000
Irradiation Time/s
20000
25000
FIGURE 2: Simulation results from the butyraldehyde photolysis.
Conclusions
d Aldehydes with a chain length C4 the photo decomposition proceeds via
two channels, a molecular channel and a free radical channel. The individual
contribution for each channel depends on the molecular structure.
CD The very low n-butyl hydroperoxide and CO concentrations measured in the case
of n-pentanal indicate that the free radical channel is of minor importance for
longer chain aldehydes.
D The effective quantum yields for all compounds investigated is below unity. The
highest values were obtained for aldehydes which has a structure like R-
CH(CH3)CHO
d Alkyl hydroperoxide formation can be used to track free radicals formed from
the photo decomposition of carbonyls to determine the individual decomposition
pathways. The method will be employed to measure the effective quantum yields
for ketones.
O The effective quantum yields for longer chain ketones like 2-pentanone is quite
below unity and similar as determined for acetone (Gardner et aL, 1984).
-252-
-------
References
Atkinson R., D.L. Baulch, R.A. Cox, R.F. Hampson, Jr., A.J. Kerr and J. Troe, J. Phys.
Chem. Ref. Data 18 (1989), 881-1097.
Becker, K.H. (ed); The European Photoreactor EUPHORE. Final Report of the EC-
Project EV5V-CT92-0059, Wuppertal, Germany (1996).
Cronin, T.J. andL. Zhu, J. Phys. Chem. 102 (1998), 10274-10279.
Lazrus A.L., G.L.Kok, J.A. Lind, S.N. Gitlin, E.G. Heikes and R.E. Shetter; Anal.
Chem. 58 (1986) 594-597.
Lee M., B.C. Noone, D. O'Sullivan B.G. and Heikes; /. Atmos. Ocean. Tech. 12 (1995)
1060-1070.
Gab S., W.V. Turner, S. Wolff, K.H. Becker, L. Ruppert and KJ. Brockmann; Atmos.
Environ. 29 (1995) 2401-2407.
Possanzini M., V. Di Palo, M. Petricca, R. Fratarcangelli and D. Brocco; Atmos.
Environ. 30 (1996) 3757-3764.
Shepson P.B., Hastie D.R., Schiff H.I., M. Polizzi, J.W. Bottenheim, K. Anlauf, G.I.
Mackay and D.R. Karecki; Atmos. Environ. 25A (1991) 2001-2015.
Acknowledgements This work has been supported by the European Commission
through the project ENV4-CT97-0419. The authors want to thank the financial
contribution of the Generalitat Valenciana and Fundacion BANCAJA.
-253-
-------
Investigation of Real Car Exhaust in the
EUPHORE Chamber
P. Wiesen
Bergische Universitat Gesamthochschule Wuppertal/FB 9 - Physikalische Chemie
Gaufistrafie 20, D-42119 Wuppertal, FRG
Introduction
The use of reformulated and alternative fuels has been increasingly tested during recent
years in order to improve air quality by lowering the ozone formation of air pollution in urban
areas. However, most of these investigations were carried out rather indirectly using e.g.
artificial exhaust gas mixtures. The investigations, which are presented here were focused on
gasoline and diesel fuel formulation and its influence on atmospheric processes. The main
objective was the investigation of real car exhaust in the European Photoreactor EUPHORE in
order to improve the knowledge on atmospheric NOy chemistry, ozone formation and the
emission and reactivity of particulates.
Experimental
Gasoline exhaust gases which were sampled from an engine test bed at the Polytechnical
University of Valencia were transported to the EUPHORE chamber in 6 tedlar bags, 250 / each,
kept hi isothermal containers to stabilise the exhaust gas temperature and to prevent irradiation
by sunlight, see Figure 1.
Exhaust samples of 5 different fuel blends were used for simulation experiments (see
Table 1). The total hydrocarbon concentration of these samples varied between 40 and 90 ppmC.
Due to technical reasons during exhaust sampling the total hydrocarbon concentration of the
other exhaust samples was too low to perform simulation experiments. The experiments were
performed with a total hydrocarbon concentration inside the reaction chamber of about 400
ppbC.
Fuel Type
aromatics, % v/v
paraffins, %v/v
olefins, %v/v
MTBE, %v/v
Fuel 1
14
86
<1
0
Fuel 4
33
54
13
0
Fuel?
2
98
<1
0
FuelS
reference
32
53
15
5
Fuel 9
reformulated
18
54
22
5
-254-
-------
Table 1:
Characteristics of the different gasoline fuel blends.
Summary of tasks performed by UPV
EUPHORE
1
1 Entfne j
, ondyzerf
1 ft
--" • " It'
CnanHcxi dnym
^Mxiuu^bnxa
Fig. 1: Schematic diagram of the set-up for the gasoline fuel experiments.
The composition of the gasoline fuel types varied from a mainly aliphatic mixture
(>98%) to blends with a high percentage of aromatic compounds (>30%). Two of the gasolines,
the reference fuel (present Euro Grade fuel) and the reformulated blend, contained the fuel
additive MTBE. The composition of the exhaust gas samples was determined at the
Polytechnical University of Valencia (UPV). The exhaust gas samples consisted of a very large
number of different components, only a few main compounds had relative ratios higher than 5%
C/C.
Fig. 2: Schematic diagram of the set-up for the diesel fuel experiments.
-255-
-------
In addition to the work on gasoline fuels, a diesel fuel matrix was defined, produced and
analyzed in terms of spec, bulk properties and of composition. Different aromatic hydrocarbon
and a low sulfur content basically characterized the diesel fuels. For the experiments with diesel
fuel a simple motor test bed was set up close to the European Photoreactor EUPHORE, which
allows the operation of a commercially available small size diesel engine from a passenger car.
The diesel exhaust was injected directly into the EUPHORE chamber through a heatable and
temperature-controlled sampling line. In the EUPHORE chamber the concentration-time profiles
of several chemical species including particulates were monitored by various analytical
techniques. The schematic set-up for the diesel exhaust experiments is shown in the Figures 2
and 3.
Exhaust gas inlet system
Exhaust gas
Fig. 3: Mixing unit for the injection of diesel exhaust gases into the EUPHORE simulation chamber.
-256-
-------
Results and Discussion
The experiments in the EUPHORE simulation chamber clearly revealed the impact of
different parameters on the photochemical processes during gasoline exhaust gas oxidation:
radiation, initial NOX concentration and composition of the fuels; see Figure 4.
Ozone time profiles with different initial NOX concentrations during gasoline exhaust gas
experiments are plotted in Figure 5.
NOx
Concentration
Ozone
Formation
Fig. 4: Parameters which influence the ozone formation.
180 240
t/min
300
360
420
Fig. 5: Ozone time profiles during exhaust gas experiments with different initial NOX concentrations, reference
fiiel.
Experiments performed in EUPHORE are characterized by using natural sunlight to
irradiate the reaction mixture and initiate the oxidation processes. Therefore, the experiments
strongly depend on weather and season. While the initial NO concentration in the smog chamber
can easily be controlled, it is not possible to manipulate the radiation intensity by the use of
-257-
-------
photolysis lamps like in indoor photoreactor experiments. As a consequence, radiation intensities
at different wavelengths have to be measured carefully for every experiment.
Experiments with simple VOC mixtures indicate that MTBE does not have a significant
impact on the radical chains during the oxidation of other VOCs. The contribution to ozone
formation from MTBE oxidation only adds to the ozone formation in the system without MTBE.
If ozone formation from other compounds is already very fast, MTBE oxidation does not show a
significant impact. If ozone formation is comparably slow like from alkanes, MTBE competes
with these hydrocarbons and contributes to ozone formation.
With respect to the ozone formation potential of MTBE being similar to /-octane and much
less than that of most aromatic hydrocarbons the substitution of aromatic compounds in fuel
blends by MTBE might lead to a reduction of local ozone formation from evaporated fuel or
exhaust gas, see Figure 6.
400
350
300
•500 ppb MTBE
exhaust gas without
additional MTBE
250; exhaust gas with
additional MTBE
60
120
180 240
t/min
300 360
420
Fig. 6: Ozone formation for reference fuel exhaust gas with and without additional MTBE; 100 ppb NOX.
During future work on this topic mixtures with a few more compounds should be
investigated and experiments with similar total hydrocarbon concentrations and substitution of
aromatics by MTBE should be compared.
From the one-day simulation experiments performed during this project it cannot be
concluded whether MTBE might contribute to regional ozone formation because of its longer
tropospheric lifetime. Since MTBE and its oxidation products are water soluble and the
photolysis rates by sunlight are fairly low, deposition will probably be the most important loss
process for these compounds.
Because of the strong influence of the radiation intensity, a direct comparison between
exhaust oxidation experiments from different fuel types is only possible for very similar reaction
-258-
-------
conditions. In cases when the conditions are not very similar, reliable modeling of the
photochemical processes is required to enable the comparison of different simulation
experiments, especially those with VOC mixtures. The difficulty arising at this point is the
missing mechanistic knowledge with regard to the oxidation of aromatic hydrocarbons. As long
as the atmospheric degradation of aromatics is not fully understood it will remain infeasible to
correctly model the behavior of VOC mixtures containing aromatics. Further, the loss of NOX in
these systems and the identification of the NOZ compounds should be a matter of special interest.
From the experiments with exhaust gas samples from differently blended fuels it can be
concluded that within the same time period the ozone formation from the oxidation of paraffins
is much less than from aromatic and olefinic hydrocarbons, see Figure 7.
350
Fuel No. 8
Reference Fuel
.a
a.
a.
300 -
250 ~ Fuel No. 7
Aliphatic Fuel
O 200 :
150 •£"
100 -f-
50 |
0
0
60
Fig. 7: Ozone formation for different fiiel blends under similar conditions; 100 ppb NOX
The substitution of aromatic hydrocarbons by olefins does not significantly affect the overall
ozone formation. The use of MTBE as a fuel additive offers the possibility to reduce the
aromatic content of the blends and thus the atmospheric reactivity of the exhaust gas.
In Figure 8 typical concentration-time profiles for HONO, HNOs and SO2 are shown after
the injection of diesel exhaust into the EUPHORE chamber. The injection time was 2 min. The
engine was operated on the motor test bed under idle conditions using the European standard
reference diesel fuel. The injection of the diesel exhaust caused a rapid increase of the
concentration of the different species and then remained almost constant. After three hours a
sodium carbonate denuder was installed in the gas sampling system, which caused a significant
decrease in the HONO, HNO3 and SO2 concentrations. Surprisingly, the HONO concentration
did not go to zero when the denuder was switched on. This is probably caused by an NO2/SO2
interference in the HONO detection system. Four hours after the diesel exhaust injection the
denuder was switched off and the concentrations increased again reaching the levels measured
before the denuder was switched on. After five hours the chamber was opened and the gas
mixture was exposed to sunlight. A decrease in the HONO concentration was observed which is
in reasonable agreement with known HONO photolysis rate constants. However, the
-259-
-------
concentration values did not go to zero as expected, indicating again an NO2/SC>2 interference in
the HONO detection system. Further experiments are necessary in order to clarify this problem.
In Figure 9 typical time profiles for the SMPS volume and number concentration of
particles after the injection of diesel exhaust into the EUPHORE chamber are shown. Injection of
diesel exhaust caused a rapid increase of the SMPS volume and number concentration. Exposure
of the exhaust mixture to sunlight did not show an effect on the tune behavior.
{
IIKM:..
».__.»»*..»«.».
___«..»..
-*-6-
• »
-. ..
!•-•—•-
,LWf. 0?
* * * :
... _ >. . _»..-
i
~ -
:.»*•'
1
i
!
*;»**•
j?
>••— -
-
*
A
^ V'
*
*
*""""" »
^"*
:
* * • *
• * ' -
»„*...»„«
i-^— '•**••
i
4. :
I
I
J
I
t
Fig. 8: Typical concentration -time profiles (r.h. 50%) for HONO ( +), HNO3 (•) and SO2 (o) after the injection of
diesel exhaust into the EUPHORE chamber. The injection time was 2 min. The engine was operated under
idle conditions.
10COGO&
100600, -.--H-
VS9W526
[;;-:
•' '.SWFS VSSJtiftffi
1QGOQ
TOGO
-260-
-------
Fig. 9: Typical time profiles (r.h. 50%) for the SMPS volume and number concentration of particles after the
injection of diesel exhaust into the EUPHORE chamber.
In Figure 10 typical particle size distributions after the injection of diesel exhaust into the
EUPHORE chamber are shown. The size distribution is bimodal with a first maximum at a
particle size of 33 run and a second maximum at 107 nm directly after the exhaust injection.
After three hours the size distribution exhibits a maximum at a particle size of 90 nm which is
caused by agglomeration of the ultra-fine particles. In addition, Figure 10 illustrates a size
distribution which was obtained directly after the injection of diesel exhaust when the diesel
engine was warmed up by a different operating procedure. The size distribution exhibits a
maximum at 80 nm particle size. This clearly shows that the particulate emission is strongly
dependent on the engine operating conditions and engine pre-treatment.
%
"O
Fig. 10: Typical particle size distributions after the injection of diesel exhaust into the EUPHORE chamber.
Conclusions
The experiments performed with gasoline fuels clearly showed that the fuel composition
has a strong impact on ozone formation. The composition of the exhaust gas depends on fuel
blending and, therefore, the amounts of alkanes, alkenes and aromatic compounds in the exhaust
vary. The experiments demonstrated that exhaust gases with a high aromatic content form more
ozone within the same time than mainly aliphatic exhaust gases. The experiments carried out
with the European standard reference diesel indicate that significant HONO quantities are either
directly emitted from the engine or formed rapidly in the simulation chamber. In addition, the
diesel exhaust experiments showed that the emission of particulates is strongly dependent on the
engine operating conditions and engine pre-treatment.
-261-
-------
Further knowledge of the atmospheric oxidation of fuel components and exhaust gas is still
needed.
Acknowledgements
Financial support for this work was provided by the European Commission under contract no.
ENV4-CT95-0015 and ENV4-CT97-0390.
-262-
-------
Outdoor smog chamber experiments in Mexico
By Julio Sandoval
Abstract
Towards the end of the 1940s the air's contamination began in Mexico. It just happened
when Mexico starts its industrial growing. During these years the population increased
notably. Industrial growing in the 1950s was carried through an urbanization process
(Restrepo, 1992). Mexico City was the first because it had industries, services, government
offices, culture attractions and higher education. It was followed by suburbs of the State of
Mexico, which are close to the Mexico City. Nowadays it is called Mexico City's
metropolitan area (MCMA) and it lies in a high valley surrounded by mountains located
2240 meters above the sea level capable of holding in pollutants released by its more than
20 million inhabitants. This altitude causes a fuel's incomplete combustion because of low
oxygen concentration in the air. In addition to average low velocities (< 1.5 m/s) almost
during 7 months in a year (Martinez, 1996). Due to that, MCMA has a behavior close to a
smog chamber and has suffered during the past two decades from increasingly severe smog
events with very high ozone (O3) levels. On the other hand, the major non-methane
hydrocarbons (NMHC) components and nitrogen oxides (NOx) of the contaminated
atmospheric mixture proceed from the approximately 3 million vehicles in the MCMA and
of industrial operation (inventario de emisiones, 1996).
Taking in account this problem was necessary to do research look into the ambient air the
conditions under it was happening. Some papers were consulted in order to begin an
investigation. The ozone formation depends not only on the HC and its atmospherics
reactions, but also on the conditions of the system where the HC are reacting (Carter,
1994). So with a budge cut out, it was decided to looking for ozone control strategies under
real conditions. Several experiments were conducted to simulate potential Oa control
strategies involving the effect of adding NMHC or NOx and reductions of them in outdoor
smog chambers. It was made a similar experiment as Nelson A. Kelly carried out in Los
Angeles (Kelly and Gunst, 1990).
captive-air irradiation (CAI) was used to evaluate the effects of HC and NOx changes on
peak springtime O^ levels. The experiments were performed at the Mexican Petroleum
Institute (IMP). It is located in northwest of Mexico City near large sources of HC and NOx
emissions. At this site. The CAI experiments were run on 17 days between 3 April and 22
April in 1995.
The experimental program was designed with the primary objective of finding out the
response of maximum ozone levels, Os (max), to changes in the initial concentrations HC
and NOx. The experimental design consisted of four levels each for HC and NOx; two
levels of reduction below the ambient level (-25% and -50%). The ambient level, and one
level of increase above the ambient level (+25%). So that, 16 HC-NOx experiments were
possible each day, however only eight chambers could be operated each day. The eight HC-
NOx combination to be tested on a given day were selected to meet two objective more: to
-263-
-------
provide replicates, with identical prepared mixtures and to minimize the confounding
effects of day to day variations. In this scheme, at least one unperturbed chamber was
included each day. The selection of the chamber to be used for each experiment in an eight-
chamber block of experiments was made randomly during the study. A total number of 136
experiments were performed on 20 days. Control chamber experiments were run every day.
Those experiments comprised the largest fraction of the total.
Every day was prepared by filling the eight chambers with morning ambient air at the same
time. After that, each chamber was prepared with the experiment according to the
experimental program: (1) replacing some ambient air with clean air from a cylinder
(dilution), and /or (2) adding HC and/or NOx. The clean air diluent was HC-free air with a
few ppbC of hydrocarbons. The NOx spike gas was 10,000 ppm NO in Na and the HC
spike gas was a 39,060 ppmC in N2- The HC mixture was made of propane, n-butane,
toluene and propylene with the proportion 40.6:30.1:27.4:1.9, respectively in ppmC. This
hydrocarbons proportion is typical of MCMA. Filling Hamilton syringes from the
cylinders, and then adding the gas to the flowing stream of clean or ambient air filling the
chambers added spike gases.
All experiments were performed in 500 liters chambers constructed from 0.5 mm FEP-
Teflon. Nine chambers were mounted on nets in a 3 x 3 matrix to 1 m above a wood
platform. The middle chamber was a control chamber used for HC, Carbonyl and
Temperature measurements. The eight chamber around the control chamber were used for
HC-NOx different mixtures.
Every day, the chambers, which contained air from the day before, were evaluated at 6.00 h
in the morning for leaks. Each damaged chamber was changed, The chambers were
evacuated, refilled with 30 liters of clean air, reevacuate and then filled. After that, the
mixtures were prepared, the measurements started almost ever at 9.00 h in the morning.
The measurements reported are HC, NOx, O3, ultraviolet flux and temperature. HC samples
were taken in canisters from the control chamber beginning the experiments in the morning
and at the end of them in the afternoon. The HC samples were analyzed by using gas
chromatography. The initial HC of each chamber was calculated from the treatment it
received and the analysis of hydrocarbon in the control chamber. The NOx was measured
each 58 minutes in each chamber until 17.00 h using a Thermoenvironment 49 analyzer
equipped with a nylon filter to remove nitric acid. Ambient and chamber temperatures were
measured with aspirated thermocouple. Ultraviolet flux was measured with an Eppley
ultraviolet radiometer.
Experiments in addition to those related strategy control, were conducted to check
chambers-wall-loss and line loss determination for NOx and Os, and to study
meteorological variables like temperature an ultraviolet flux. The first two types of
experiments showed that wall and line losses were negligible. The meteorological variables
presented a linear correlation with the Os formed in surrogate HC-NOx -clean air mixtures
on different days.
All experimental database consists of hourly measurements of NOx and Oa, initial and end
HC in a control chamber, and temperature and ultra violet flux in ambient air outside the
-264-
-------
chambers. The database comprises 136 experiments run on 20 days. The initial HC and
NOx concentrations, temperature and ultra violet flux were taken in count as independent
variables; the independent variable was 63(max). This database was used to determine the
adequacy of the experimental program, to assess the experimental error, and to find out an
empirical model for O3(max).
The effects of perturbations on the O3 profiles followed a sigmoidal kinetics curve, which is
typical of an autocatalytic reaction. An initial induction period was followed by a sharp
increase to 63 (max), and then by very little change Os (max) occurred between 14.00 h and
16.00h.
To estimate the experimental uncertainty attributed to random error 50 % of the
experiments were replicates. A quantitative estimate of the random measurement error
variability as a percentage of the mean was only 4.27%.
Day to day variations in initial HC and NOx were very different. So that, the induced
perturbation expanded the experimental region slightly. It included the curvature in the
isopleths and the NOx -inhibition region. The daily variability in the control chamber was
large; HC varied from 1500 to 6000 ppbC and NOx varied from 38 to 345 ppb. The
HC/NOx ratio in the control chamber ranged from 9 to 32 with an average of 20.6. The day
to day variations in Os (max) in chambers without perturbations ( ambient morning air to
generate O3 in the chamber) ranged from 175 to 684 ppb.
To study the temperature effects on the rates of thermal reactions and ultraviolet flux
effects on the rates of photochemical reactions were performed surrogate HC-NOx-clean
air irradiation. In this experiments, perturbations of the initial HC and NOx concentration
were controlled, but the variations in temperature and ultraviolet flux were no controlled.
These important variables were assessed using two chambers containing clean air plus 1500
ppbC the three-component HC-spike mixture and 150 ppb NOx. These replicate mixture
were prepared and irradiated on 3 days. The difference between O3 (max) for same-day
replicates average only 7.5%. Those experiments allowed to confirm that O3 (max) was
directly proportional to the average temperature and ultraviolet flux of each day.
The most of the experimental dates were used to determine a functional relationship
between O3(max), and the initial HC and NOx concentrations, temperature and ultraviolet
flux. Multiple regression modeling was used for it. Some combinations of predictor
variables were investigated by including terms in the mathematical model such as HCX, and
NO*, where x and y = 0.5,1,2 and 3. Cross products and ratios, such as HCX. NO*,
HCX/NO*, and NO*/ HCX, were also considered. These terms express interactions between
HC and NOx. The effects of temperature and ultraviolet flux were investigates by including
T(avg.), T(max). u.v.(avg.) and u.v.(max) in linear form. The coefficient of determination,
R, was the statistical criterion considered. The empirical model, for OsOmax) found out,
can generate similar isopleths in shape to those generated by a mechanistic model. Thus,
the empirical model is consistent with the chemical phenomena shown to be responsible for
Os formation under laboratory conditions.
-265-
-------
Regression modeling was used to determine the empirical model for O3(max) and
considering the R2 statistics as well:
O3(max) = 195.063 + 1.12 NOX - 0.5 NO2X/HC + 0.115(Tavg - 26.98)
was chosen as the empirical model for O3(max) (the average temperature was 26.98 °C). It
contains two terms to fit the dependence of O3(max) on the initial HC and NOx precursor
concentrations, and one term to fit the temperature dependence.
Using the equation for Tavg = 26.98 °C an O3 isopleth was generated. The isopleths were
similar in shape to those generated by a mechanistic model. Thus, the empirical model was
consistent with the chemical phenomena shown to be responsible for O3 formation under
laboratory conditions.
This model predicts the NOx-inhibition region and the horizontal flattening in the lower
right hand portion region. Between the NOx-inhibition and HC-saturation regions there is a
highly curved region. Where O3(max) will respond to changes in HC and/or NOx within
the knee region. Those region provide the foundation for discussions of O3 control
strategies.
Ozone control strategies can be evaluated with the empirical model developed from the
experiments. Of course the CAI are not exact simulations of real atmosphere, because of in
the real atmosphere there are HC and NOx emissions into an air parcel throughout the day.
Thus, this difference between the chambers and the atmosphere should be unimportant for
elucidating control strategies. By the way, we focused on control strategies to reduce
O3(max) by 50%; specifically, changes in O3(max) from 600 to 300 ppb in the chambers.
Those end-points were selected because they bracket the average O3(max) observed. In the
experiments (450 ppb), and they represent a 50% decrease in O3(max).
O3(max) control strategies were evaluated for initial HC/NOx ratios of 12, 16 and 20. The
ambient HC/NOx ratio that was measured in the 90% of the control chambers. For all
calculations Tavg was taken as 26.98 °C.
O3(max) response to simultaneous HC and NOx reductions. The ozone changes are in the
HC-saturation region. 50% reductions in O3(max) required 73.5 % simultaneous reductions
in HC and NOx for HC/NOx ratios hi the range 12-20. Here the change in O3(max) is linear
with the change in precursors and independent of the HC/NOx ratio.
O3(max) response to HC reductions. These HC reductions cause changes from the HC
saturation region to the knee region. 50% reduction in O3(max) requires 95-97% reductions
in HC for initial HC/NOx ratios of 12-20. There is a marked non-linear in the response of
O3(max) to reductions in HC alone after 80% reductions.
O3(max) response to NOx reductions. These NOx reductions cause changes into the HC-
saturation region. 50% reductions hi O3(max) requires 75% roughly NOx reductions hi
NOx for initial HC/NOx ratios of 12-20. Here the change in O3(max) is linear and
independent of the HC/NOx ratio.
-266-
-------
Ambient air in Mexico City (the control chambers) usually had a HC/NOx ratio in the HC
saturation region. That means that a change in HC does not reduce la 03 significantly. But a
change in NOx moves the system into the HC saturation region where the Os is reduced in
an important way. Changes in both pollutants move the system throughout the HC
saturation region. Thus, the current emissions control on HC only may be responsible for
the lack of progress in reducing Oa in Mexico City, the best Os-control strategy is to reduce
the NOx pollutant.
Of course, while the NOx reduction strategy is the most efficient on average, no single
strategy is the best every day due to the variability in HC/NOx.
REFERENCES
Restrepo, I. (1992). "Algunos antecedentes y consecuencias del problema". La
contamination Atmosferica en Mexico, sus Causas y Efectos en la Salud. 9-13.
Martinez, L, G. (1996) "Analisis del Comportamiento de Ozono en la Ciudad de Mexico".
Tesis de Licenciatura. E.S.I.Q.I.E. Instituto Politecnico Nacional.
Inventario de Emisiones. (1996). Gobierno del Distrito Federal Secretaria del Medio
Ambiente.
Carter, W. P. L. (1994). "Development of Ozone Reactivity Scales for Volatic Organic
Compounds," J. Air and Waste Manage. Assoc. 44, 881-899.
Kelly, N.A.; Gunst, R.F. (1990). "Response of Ozone to Change in Hydrocarbon and
Nitrogen Oxide Concentrations in Outdoor Smog Chambers Filled with Los
Angeles Air." Atmospheric Environment. Vol. 24a. No. 12, pp. 299 1-3005.
-267-
-------
Session IV (continued)
Environmental Chamber Studies
Session Chair
Harvey Jeffries
-------
EPA Gas Phase Chemistry Chamber Studies
D.J. Luecken (1), and T.E. Kleindienst (2)
(1) National Exposure Research Laboratory, U.S. Environmental Protection Agency, MD-84,
Research Triangle Park, NC, 27711
(2) ManTech Environmental Technology, Inc., Research Triangle Park, NC, 27709
Extended Abstract
Gas-phase smog chamber experiments are being performed at EPA in order to develop databases
that can be used to evaluate a number of current chemical mechanisms for inclusion in EPA's
regulatory and research models. This work differs from that being performed in other
institutions because the emphasis at EPA is on examining a variety of different mechanisms,
rather than developing a single mechanism, and evaluating the appropriateness of each one for
the various types of applications of air quality models at EPA. The mechanisms that we are
currently studying include the Carbon Bond IV 99 (CB4-99) (Adelman, 1999), the SAPRC99
mechanism (Carter, W.P.L., 1999, private communication), and an updated version of the
RACM mechanism (Stockwell, et. al., 1997). In addition, the database will provide information
for studying a variety of mechanisms formulated using the newly-developed Morphecules
method (Jeffries et al.; 1999). In keeping with this objective, the experimental conditions in the
chambers are chosen to stress the mechanisms and enhance the differences among mechanisms.
By providing slightly different chamber conditions and experimental conditions than are found in
other existing smog chambers, the EPA chamber data expands the database of experiments that
are available for robust mechanism evaluation.
Two smog chambers are available, each consisting of 9000-L rectangular bags completely
constructed of either 2- or 5-mil FEP Teflon film, and housed within two opposing high-
reflectivity UV light banks mounted on an aluminum frame. The chamber has Teflon ports in a
bottom panel and a fan is used to thoroughly mix the reactants in the chamber. Cooled air from
an A/C unit is circulated in the ballast cavity and between the chamber walls and lights in order
to maintain a near constant temperature of 26 °C during the simulation. One of the chamber
frames is constructed to allow it to be rolled outside to conduct experiments in natural sunlight.
When used indoors, the chambers are irradiated by a system of 122 cm fluorescent bulbs, with
radiation in the wavelength range 300-350 nm generated with UV-340 black light bulbs, and
radiation in the wavelengths 350-400 hm generated with standard F40-BL bulbs. This
combination has been designed to match, to the extent possible, the distribution of solar radiation
between 300 and 400 nm with a fluorescent bulb system.
Prior to an irradiation, the chamber is flushed for 24 to 48 hours with ultrazero air generated in-
house with an Aadco pure air generator. Background samples are taken to verify that levels of
hydrocarbon and carbonyl compounds are negligible (>50 ppbC) prior to the addition of reactant.
Irradiations are typically performed until a peak is reached in the ozone concentration or for 12
hours. During the experiments, temperature, relative humidity, and light intensity are
continuously measured. NO and NOX are continuously quantified using a chemiluminescent
monitor with a converter that reduces NOX to NO. Because the NOX from this instrument is
-269-
-------
subject to interferences from PAN and other organic nitrates, it is a good measure of NO+NOa
only during the early stages of photochemical conversion. Ozone is continuously monitored
during the experiment with a chemiluminescent monitor. Measurements of organic precursors
and reaction products were identified by using GC with detection by flame ionization (FID),
electron capture (BCD), or mass spectrometry (MS), with samples taken at 1 to 2 hour intervals,
depending on the experiment. Peroxyacetyl nitrate (PAN) and other peroxyacyl nitrates were
measured by GC/ECD at 1-hour intervals. Measurements of carbonyl compounds were obtained
by bubbling sample air through a solution of 2,4-dinitrophenylhydrazine (DNPH) in acidified
acetonitrile, producing carbonyl hydrazones which were separated and quantified by high
performance liquid chromatography (HPLC). Concentrations of nitric acid were obtained by
sampling with nylon filters followed by extraction in 10"5 perchloric acid solution and analysis
by ion chromatography. Other instrumentation is currently being evaluated, and alternative
analytical techniques that are being developed, include the measurement of organic nitrates by
GC/ECD, determination of multifunctional carbonyl compounds with derivatization by
pentafluorobenzylhydroxyl amine (PFBHA) followed by GC/MS analysis, and determination of
hydroxylated compounds, including organic acids, by derivatization with BSTFA followed by
GC/MS analysis. Exploratory work is being undertaken to determine whether new pattern
recognition techniques utilizing UV absorption spectrometry can be used to measure NOa and
HNO2 directly.
The chamber-dependant reactions are characterized through the use of a limited set of
characterization experiments consisting of single component studies of clean air, clean air/NOx,
CO/NOx, formaldehyde/NOx, and ethylene/NOx experiments. Experiments are performed
under dry conditions and at relative humidities between 50-60%.
Earlier experiments have been performed to study ozone formation potential from a series of
reformulated gasoline and alternative fuel exhaust from new technology vehicles. These
experiments have used authentic exhaust generated from several different vehicles with a
dynamometer and have focused on ozone formation potential as a function of fuel type
(Kleindienst et al., 1994). A set of experimentally-based reactivity parameters were developed
for each fuel. Vehicles run using reformulated gas (RFG) and ethanol (E85) were found to
produce roughly equal amounts of peak ozone for similar reaction conditions, almost twice as
much as methanol-fueled (M85) vehicles, and over 4 times as much as vehicles run using
compressed natural gas (CNG). (Measurements were made for standard VOC/NOx ratios of 5.5)
Another result of this work has been the creation of auto exhaust surrogates that were prepared as
tank mixtures using whole fuel with a dozen additional components representing the major
combustion-derived components. These surrogates have been shown to reliably reproduce ozone
formation and other photochemical constituents and reactivity parameters compared to direct
automobile exhaust.
Studies of branched and straight chain Cs-Cs alkane reactions have been performed in the
chamber. Although these compounds are not extremely fast reacting, they are emitted into the
atmosphere in large quantities from automobile exhaust. For these studies we have combined
dynamic and static chamber simulations to better identify and quantify multifunctional
oxygenated compounds, organic nitrate, and secondary product formation. The static smog
chamber was used to provide moderate conversion of reactants to products without major
-270-
-------
influences due to sampling and dilution, allowing quantitative determination of the reactive
hydrocarbon loss and of the product formation rates. Simulations were also done using a
dynamic reactor consisting of a 1-L Pyrex bulb reaction vessel, which had been deactivated with
dichlorodimethylsilane, and irradiated by F40-BL fluorescent bulbs. The dynamic simulations
allowed complex organic compounds to be more easily observed and quantified by varying the
extent of the reaction.
This year's experiments focus on studying complex mixtures of hydrocarbons, including both
branched and straight chain members of each class in combinations similar to those found in
rural and urban atmospheres. These mixtures are designed to test the effect of the hydrocarbon
lumping schemes used in current chemical mechanisms versus the more explicit Morphecule-
type approaches. We are also coordinating gas-phase smog chamber experiments with studies of
aromatic and biogenic hydrocarbons performed in the PM chemistry program, by performing
gas-phase only experiments for compounds which are also studied in the particle chamber.
Future plans include using enhanced analysis techniques to more completely characterize NOy
and product formation. We are establishing a protocol for the types of comparisons that must be
done to thoroughly compare chemical mechanisms, using a Process Analysis approach. In
conjunction with our modeling of the smog chamber data and comparison of results using
different mechanisms, we hope to provide support for the clarification of uncertain reaction
parameters in atmospheric chemical mechanisms.
References:
Adelman, Z.E., A Reevaluation of the Carbon Bond-IV Photochemical Mechanism, PhD thesis,
University of North Carolina, Chapel Hill, NC, 1999.
"V
Jeffries, H., and Kessler, M., Morphecule/Allomorph Reaction Mechanisms, Final report to U.S.
EPA, contract 68D50129, 1999.
Kleindienst, T.E., Liu, F., Corse, E.W., and Bufalini, J.J. "Development of a chamber system for
determining measures of reactivity from exhaust of alternative-fueled vehicles". SAE No.
941906. Society of Automotive Engineers, Warrendale, PA, 1994.
Stockwell, W.R., Kirchner, F., and Kuhn, M. "A new mechanism for regional atmospheric
chemistry modeling", J. Geophvs. Res.. 102(D22): 25847-25879, 1997.
This work has been funded in part by the United States Environmental Protection Agency under
68-D5-0049 with ManTech Environmental Technology, Inc. It has been subjected to Agency
review and approved for clearance.
-271-
-------
Atmospheric Photochemical Degradation of 1,4-
Unsaturated Dicarbonyls
XIAOYU LIU, HARVEY E. JEFFRIES*, and KENNETH G. SEXTON
Department of Environmental Science and Engineering,
CB7400, Rosenau Hall,
The University of North Carolina at Chapel Hill,
Chapel Hill, North Carolina 27599-7400
* Author to whom correspondence should be addressed. E-mail: harvey@unc.edu
phone: 919-966-7312, fax: 919-933-2393
Introduction
The elucidation of details of photochemical reaction mechanisms for the oxidation of
aromatic volatile organic compounds remains a major problem (7). This is because there are
apparently a large variety of photooxidation products formed in the reactions. Each of these
products further reacts apparently in ways that are different from the better-understood alkane
and simple alkene monofuctional products. The total reacted carbon balance is still poor in the
aromatic systems, mostly because of lack of appropriate analytical methods for detecting the
product mixtures of those photochemical reactions.
One of the most important types of products generated in the reaction of monocyclic
aromatic hydrocarbons with the hydroxyl radical under tropospheric conditions is 1,4-
unsaturated dicarbonyls (2-4). Formation of butenedial from toluene and o-xylene, 4-oxo-2-
pentenal from toluene, o- and m-xylene, and 3-hexene-2, 5-dione from p^-xylene and 1,2,4-
trimethylbenzene have all been experimentally observed and are postulated as ring-cleavage
carbonyl products during the OH-initiated atmospheric degradation of those aromatic
hydrocarbons (2,5, 6). These species are expected to be very reactive after they are formed in the
atmosphere, undergoing reactions with OH and Os as well as directly photolyzing (2-4,7). Thus,
they may act as important sources of free radicals, promote organic aerosols, and serve as
precursors of carboxylic acids, hydroperoxides, and oxidants such as Os, PAN and
peroxycarbocyclic acids (5). Detailed studies of the fates of these highly reactive compounds,
however, are lacking. A thorough investigation of atmospheric photooxidation of these reactive
intermediates, which appear to be major secondary products, is required if we are to have
predictive knowledge of the fates of their precursors and clearly understand photochemical
reaction mechanisms for aromatics.
A few investigations of the atmospheric chemistry of unsaturated 1,4-dicarbonyls (3, 4,
7) have been reported. Using Fourier transform infrared (FTIR) absorption spectrometry, Tuazon
et al. (4) investigated the photolysis of cis- and trans-3-hexene-2, 5-dione and reactions of these
isomers with OH and O? at room temperature. Methylglyoxal and formaldehyde were found to
be the major products of the reaction with Os, but methylglyoxal was not detected as a product of
-272-
-------
the reaction of the 3-hexene-2, 5-dione with OH radicals. They also reported that the rate
constants of the cis and trans isomers of 3-hexene-2, 5-dione were (1.8+0.2)xlO~18 and
V18
(8.3±1.2)xlO'ls cmi molecule'V for the reaction with O3 at 298±2 K, and (6.3±0.6)xlO"'' and
(5.3+0.5)xlO"'' cm3molecule"'s"1 for the reaction with OH. Another compound of this category,
3,4-dihydroxy-3-hexene-2, 5-dione, has recently been examined by Wiesen et al. (7) with long-
path FTIR absorption spectroscopy. These authors argued that a hydrated vicinal polyketone,
3,3-dihydroxyhexane-2, 4, 5-trione, was probably the major product during the reaction of OH
radicals with the above enediol form of acetylformoin. They also conducted a series of reactions
of OH radicals in a 1080-L quartz-glass chamber with butenedial, 4-oxo-2-pentenal, 3-hexene-
2,5-dione, among others (3), concluding that the reaction with OH radicals and photolysis are
two major atmospheric sinks for these species. Rate constants of OH radical reactions (in unit bf
10"11 cnrWlecule'V1) are: cis-butenedial, 5.2±0.1, trans-butenedial, >2.41+0.79, cis/trans 4-
oxo-2-pentenal, 5.58±0.21, cis-3-hexene-2,5-dione, 6.9+2.1, and trans-3-hexene-2,5-dione,
4.0+0.4. Rate constants of these compounds with O3 were estimated from structure reactivity. To
the best of our knowledge, there is no experimental report on rate constants of O3 with butenedial
and 4-oxo-2-pentenal in the literature.
In this paper, we report a study in the UNC outdoor smog chamber of both OH- and
ozone-initiated photooxidation of three unsaturated 1,4-dicarbonyl compounds: butenedial, 4-
oxo-2-pentenal, and 3-hexene-2, 5-dione. Carbonyl products and intermediates produced from
reactions of these compounds with OH radicals and O3 were measured, as were time series of a
few carbonyl products. Rate constants for O3 consumption by these compounds were also
determined. Carbonyl products were analyzed by the O- (2,3,4,5,6-pentafluorobenzyl)-
hydroxylamine (PFBHA) derivative method coupled with the gas chromatography (GC) /ion trap
mass spectrometry (MS) separation and detection. This method has been successful in
identifying multifunctional carbonyls in other studies (9, 10). In daytime outdoor experiments,
we began with oxides of nitrogen and butenedial, or 4-oxo-2-pentenal, or 3-hexene-2, 5-dione in
the chamber. In nighttime experiments, ozone was injected at a constant rate into each side of the
chamber for the duration of the experiment. One side was filled with an initial amount of the
compound to be studied and about 100-150 ppmV cyclohexane as the OH radical scavenger. The
other side served as a reference for O3 concentration.
By comparison with their corresponding standards, four categories of products are found
in the OH and O3 initiated reactions, namely, simple aldehydes, dicarbonyls, unsaturated
carbonyls, and hydroxy carbonyls. Many of these carbonyls are on the EPA's hazardous air
pollutants (HAPs) list or are known to be toxic. We also observed a few compounds whose
molecular weights were obtained by mass spectra, but whose identities could not unambiguously
be determined because of lack of standards. Their possible structures are discussed in the text.
Rate constants of these 1,4-unsaturated dicarbonyls with O3 were measured. We propose
photooxidation schemes for the three unsaturated 1,4-dicarbonyls.
-273-
-------
Experimental Section
Carbonyl Reactants Preparation. Butenedial, 4-oxo-2-pentenal, and 3-hexene-2, 5-dione were
synthesized by a collaborating organic chemist from commercially available compounds, such as
furan, 2-methylfuran, and 2,5-dimethylfuran. Details of these synthesis were described elsewhere
(11). The purity of these compounds as measured by NMR is greater than 90%.
Chamber Experiments. All experiments were carried out in the UNC 300,000-liter dual
outdoor chamber located in Chatham County, North Carolina. As listed in Table 1, the daytime
chamber experiments were conducted in the presence of NOX, while the nighttime experiments
were performed hi the presence of O3 and 100-150 ppmV cyclohexane. The latter was used as
the OH radical scavenger. Cyclohexane can scavenge more than 90% of the OH radicals
produced in the Oa-dicarbonyl reaction system. As gas phase internal standard 0.11 ppmV
fluoroacetone, was also injected in each of these experiments. Detailed descriptions of the
chambers, instruments employed, analytical methods and procedures used are available
elsewhere (9,10,12). NOX, OB, and some hydrocarbons were monitored by instruments directly
connected to the sample line of the chamber. Samples for the PFBHA derivative analysis were
collected directly underneath the chamber through 0.5 m sampling lines to impingers containing
10 ml of 0.25 mg ml'1 PFBHA-H2O. The Saturn IIGC ion trap MS used a 60 m x 0.32 mm x
0.50 jjm DB-5 MS-grade, chemically-bonded, fused-silica capillary column (J & W Scientific).
Batch (integrated) samples were collected from the beginning to the end of the experiments,
while tune-series samples were taken at 30-minute intervals throughout experiment. Background
concentrations of NOX, Os, and hydrocarbons were also measured.
Quantification. Identification and quantification of products were based upon
commercially available or synthesized standards. If standards are not available, the relative
concentration defined as the peak area ratio of the m/z 181-ion peak of a carbonyl to the m/z
181-ion peak of fluoroacetone was used to represent the response of the carbonyl compound.
Volatile carbonyls can easily be injected into the chamber and calibrated either by GCs
directly connected to the sample line of the chamber or by the PFBHA derivatization method.
For high boiling point polar compounds, e.g., hydroxyacetone, glyoxal, methylglyoxal, and
glycolaldehyde, direct injection into the chamber is difficult. To solve this problem, we used a
Collison nebulizer for injection. We assumed that after injected in the chamber the relative ratio
of each compound in the solution remains the same as in the nebulizer solution. By using a
nebulizer injection internal standard, the concentration of other compounds could be calculated
according to the carbonyl's ratio and the known internal standard concentration. Cs-hydroxy
carbonyls such as 4-OH-3-methyl-2-butanone and 3-OH-3-methyl-2-butanone were found to be
appropriate standards. They are polar enough so that they will not be off-gassed when injected by
nebulizer and yet, they have sufficiently low boiling points that they can be directly injected into
the chamber.
Three sets of calibration curves were made for the PFBHA derivatization-GC ion trap
MS measurement. One was for volatile carbonyl compounds. Their calibration curves were
obtained from the PFBHA method and evaluated by comparing their quantification results with
those from GCs connected to the chamber sample lines. The second set was for polar compounds
-274-
-------
injected by nebulizer. The third set was for unstable compounds, such as glycidladehyde and
malonaldehyde. The last set was made by aqueous standards corrected for collection efficiency
and sample line loss. The slopes of these calibration curves, i.e. response factors, were used to
calculate corresponding concentrations. The measurement precision in the entire workup
procedure for the PFBHA measurement was estimated to be +20%. Time series of reactants and
products were determined by these calibration curves.
Ozone Decay Measurement. Two chamber sides enable us to calculate the O3
concentration consumed in the nighttime experiment. The second-order rate constants of O3 with
1,4-unsaturated dicarbonyls were estimated by monitoring the rate of change of ozone decay in
the presence of known concentrations of the dicarbonyls. The processes for removing ozone are
O, + wall
loss of O
3»
(1)
O3 + dicarbonyl -
products.
Hence,
d[O3] / dt = (ki+k2[dicarbonyl])[O3].
(2)
(3)
where ki and k2 are the rate constants for Reactions (1) and (2). If [dicarbonyl] » [O3]initiai, the
ozone decay is given by
-dln[O3] /dt = ki+ k2[dicarbonyl].
(4)
A plot of the ozone decay rate against the dicarbonyl concentration should be a straight line with
kz as the slope and ki the intercept.
In our experiment, the ozone was injected by alternating the O3 generator constant output
and constant flow rate between the two chamber sides at a 4-sec interval for the entire duration
of experiments. The O3 concentration consumed (A[O3]) in the reaction system from time t to
time t +At, is computed from
where A[O3]reaction and A[O3]biank are the O3 concentration changes from time t to time t +At in the
reaction chamber (O3-in-dicarbonyl) and the blank chamber (O3-in-blank).
Results and Discussion
Products in OH initiated 1,4-Unsaturated Dicarbonyl Reactions. Chamber conditions,
including air temperature, dew point temperature, Eppley total solar radiation (TSR) / Eppley
ultraviolet radiation (UV), and concentration time series of NOX and O3 in experiments
conducted on June 18, 1998 and July 2, 1998 are illustrated in Figures 1 and 2. The carbonyl
compounds that were detected are illustrated in Figures 3-5.
-275-
-------
Formaldehyde, acrolein, glycolaldehyde, glyoxal, and malonaldehyde in butenedial,
formaldehyde, methyl vinyl ketone, glycolaldehyde, hydroxyacetone, glyoxal, methylglyoxal,
and malonaldehyde in 4-oxo-2-pentenal, and formaldehyde, acetaldehyde, hydroxyacetone, and
methylglyoxal in 3-hexene-2, 5-dione have been identified and confirmed by the matching
retention time and mass spectra of their corresponding standards in the daytime experiments. A
trace amount of acetaldehyde and acetone are usually detected in the distilled water blank as well
as some glyoxal and methylglyoxal in the chamber background. Among these compounds
identified, glyoxal in butenedial, methylglyoxal and glyoxal in 4-oxo-2-pentenal, and
methyglyoxal in 3-hexene-2, 5-dione have previously been reported as the major carbonyl
products (3). The new carbonyls we detected belong to categories of unsaturated carbonyls
(acrolein, methyl vinyl ketone), hydroxyl carbonyls (glycolaldehyde and hydroxyacetone), and
dicarbonyls (malonaldehyde).
Also shown in Figures 3-5 are compounds whose molecular weights have been
determined by mass spectra but their identities have yet to be unambiguously determined
because of lack of standards. They include (M value stands for the PFBHA derivative molecular
weight) M267, M281, M283, M297, M309, M311, M323, M325, M328, M476, M490, M492,
M504, M506, and M520. The possible structures of these unidentified carbonyls with known
molecular weight are listed in Table 2. By comparing the standards available, we know M267 is
not isobutyraldehyde, butyraldehyde, methyl ethyl ketone, glycidaldehyde, methylglyoxal
(mono-derivative), or malonaldehyde (mono-derivative). M297 is not 3-hydroxy-3-methyl-2-
butanone, 4- hydroxy-3-methyl-2-butanone, and 5-hydroxy-2-pentanone. M490 is not 2,4-
pentadione. M311 is not 4-hydroxy-4-methyl-2-pentanone. M281/M476 is not mono/di-
derivative of biacetyl. M283 is not pyruvic acid and methyloxacetone Among these compounds,
some of their molecular weights are equal to that of triones or quadra-ones which may support
the point that polyketone may be important products in the photooxidation of aromatics (2, 3).
Products in Os Initiated 1,4-Unsaturated Dicarbonyl Reactions. As an example,
chamber conditions including air temperature, dew point temperature, and concentrations of Os
hi the experiment conducted on July 7,1998 are illustrated in Figure 6. In the nighttime
experiment, Os was injected into both sides of the dual chamber continuously, with only one side
containing the unsaturated dicarbonyl reactant. By comparing the time series of Os in both sides
of the chamber, the Os concentration consumed in the reaction from Eq. (5) can be calculated.
Carbonyl products detected in the Os initiated nighttime chamber experiments using cyclohexane
as the OH scavenger are summarized in Table 3. Glyoxal and formaldehyde in butenedial,
glyoxal, methylglyoxal, and malonaldehyde in 4-oxo-2-pentenal, and methylglyoxal in 3-hexene-
2,5-dione were found as the major products in these systems. Also observed in these experiments
were some carbonyls with known molecular weights but not one of the standards available. In
addition we detected cyclohexanone, which is the product of OH with the added cyclohexane.
This shows that these unsaturated dicarbonyls do produce OH radicals, probably from Criegee
biradical decomposition.
Time-series of Products. The products formed during the reaction in the UNC outdoor
smog chamber can continue to undergo further photochemical processes. It is difficult to
establish a mass balance for the photooxidation process of the unsaturated dicarbonyls studied in
this paper. When a standard was available, a time series of a carbonyl is represented by its
-276-
-------
absolute concentration in a sample, otherwise, by its relative concentration. The latter data are
still useful, especially when the time series of several related compounds are plotted together.
One can observe how reactants decay, intermediates are first produced and then consumed, and
how final products are eventually formed. Figures 7-11 present time series data of carbonyls in
4-oxo-2-pentenal and 3-hexene-2,5-dione daytime experiments. Figures 12-13 are time series of
carbonyls in the nighttime experiment of 4-oxo-2-pentenal with O3 and cyclohexane.
Plotted in Figure 7 are the daytime series of 4-oxo-2-pentenal, glyoxal, methylglyoxal,
and malonaldehyde. As 4-oxo-2-pentenal is consumed rapidly in the early morning, glyoxal,
methylglyoxal, and malonaldehyde are quickly produced, indicating that these three carbonyls
are probably primary products. After they reach maxima at about 0900 EDT, these products
themselves promptly decay to the secondary products, via either reactions with radicals in the
chamber or by photolysis. Because 4-oxo-2-pentenal is nearly totally consumed by 1000 EDT
the primary products maintain a low concentration throughout after 1000 EDT. Methylglyoxal
glyoxal, and malonaldehyde share the same pattern of formation and decay in Figure 7. While
this suggests that methylglyoxal and glyoxal share a similar formation pathway, we have been
unable to propose a formation mechanism for malonaldehyde from OH-attack followed by Oa
addition and NO-NOa conversion. More discussion of malonaldehyde formation follows later.
Shown in Figure 12 are the nighttime time series for the same compounds. Though this plot still
shows that glyoxal, methylglyoxal, and malonaldehyde are primary products, the difference is
obvious. As 4-oxo-2-pentenal slowly decays, the carbonyl products continuously build up.
Figure 12 of course does not show the loss of these compounds due to photolysis (and, because
OH is expected to be low in the Figure 12 nighttime condition, does not show significant loss by
OH attack either). The time series also help us identify different unknown isomers. As shown in
Figure 11 and 13, different chromatogram peaks showed the same molecular weight and almost
had the same time series patterns implying that they are either one carbonyl compound or
structurally similar isomers.
Ozone Reaction Rate Constants. Based on the method established in Eqs. (l)-(5), a
least-squares analysis was used to determine the rate constants for Os reactions with butenedial,
4-oxo-2-pentenal, and 3-hexene-2,5-dione to be 1.6 + 0.1 at 294-298 K, 4.8 ± 0.8 at 293-297 K,
and 3.6 + 0.3 (units in 10'18 cnrWlecule'V1 ) at 295-297 K (see Table 4). The indicated errors
are two least-squares standard deviations of the slopes of the fitted line combined with an
estimated overall uncertainty of ±20% in the PFBHA method. The carbonyls used in this study
are mixtures of cis-and tran-1, 4-unsaturated dicarbonyls. Our rate constant of Os-3-hexene-2, 5-
dione is in reasonable agreement with the literature (4). This is the first experimental
determination of kos for butenedial and 4-oxo-2-pentenal, although estimation by the structure-
reactivity method is available in the literature (3). The reaction of 1,4-unsaturated dicarbonyls
with Os involves electrophilic addition of Os to the double bond and the reactivity correlates
positively with the electronic density at the carbon-carbon double bond (13). Since the -CH3
group is an electron donator, the anticipated reactivity order should be butenedial < 4-oxo-2-
pentenal < 3-hexene-2,5-dione.Our results shows that 4-oxo-2-pentenal is most reactive and
butenedial is the slowest. This may imply that other species in the system are involved in the
consumption of OB, and thus our model in Equations (1) to (5) may overestimate the rate
constant of 4-oxo-2-pentenal.
-277-
-------
1,4-Unsaturated Dicarbonyls Photooxidation Mechanisms. Based on these results, we
can propose the photooxidation mechanisms of 1,4- unsaturated dicarbonyls shown in Schemes I
and II. Compounds with bold M values and names were carbonyls identified in experiments.
There are three major pathways available for 1,4-unsaturated to degrade: reaction with OH,
reaction with Os, and photolysis. OH radicals can add to the double bond of the unsaturated
dicarbonyls, generating hydroxy alkyl radicals. These radicals become hydroxy alkoxy radicals
via reaction with O2, followed by NO-NO2 conversion. Hydroxy alkoxy radicals can either
undergo decomposition or react with Oa via H-abstraction to generate various products. In
butenedial and 4-oxo-2-pentenal, the OH radical can also abstract a H-atom from the -CHO
group leading to the formation of unsaturated peroxy acyl radicals after reactions with 02. These
peroxy acyl radicals can either convert NO to NO2 or react with NO2 to form unsaturated
peroxyaceyl nitrate compounds that are expected to be temperature labile.
As seen hi the Scheme I and the time series data (Figure 7), glyoxal and methylglyoxal
are major primary products from these reactions. After they are formed, they quickly undergo
degradation to form secondary products such as formaldehyde, acetaldehyde, CO, peroxyacetyl
nitrate (PAN), and peroxyaceyl nitrates, etc.
We observed compounds with a derivative molecular weight of 476. Their possible
structures are C4 dicarbonyls and/or Cs triones. Because they have similar tune series patterns
(Figures 8 and 11), they may be the PFBHA derivative isomers from the same carbonyl or
structurally similar carbonyl isomers. These compounds could be formed as secondary products
from alkoxy radical decomposition in 4-oxo-2-pentenal-OH reaction. But our result shows that
M476s are formed as early as glyoxal and methylglyoxal and then decreased quickly, suggesting
that they are primary products as well. M476s have also been detected as primary products in the
OH initiated 3-hexene-2, 5-dione reaction.
In the 4-oxo-2-pentenal reaction we observed an M283 compound whose possible
structure is a C4 hydroxy carbonyl, pyruvic acid, methyloxacetone, or Cj hydroxy dicarbonyl. Lv
comparing its retention time and mass spectra with standards, the possibilities of pyruvic acid
and methyloxacetone were excluded. From Figure 8, it appears that it is a secondary product.
Thus, the primary product C* hydroxy carbonyl which formed from alkoxy radical
decomposition at 1,2 position could also be excluded. This suggests that alkoxy radical
decomposition mainly occurred at 2,3 (3,2) position.
It has been reported that 1,4-unsaturated dicarbonyl photolysis is a strong sink in the
atmosphere leading to the production of maleic anhydride, 3H-furan-2-one, 5-methyl-3H-furan-
2-one (3). According to these investigators, an OH radical concentration of (3-4) xlO7 molecules
cm"3 would be required for the OH radical reaction to compete with photolysis loss of these
dicarbonyls (14). Our PFBHA derivative method is unable to detect these anhydrides. Only a
very small amount of methyl vinyl ketone was detected in 4-oxo-2-pentenal and of acrolein in
butenedial. The time series of acrolein measured by us hi the butenedial system (not shown here,
but in ref. 11) shows that acrolein was formed earlier than glyoxal. This demonstrates that
butenedial does photolyze quickly. We did not observe, however, the formation of 4-oxo-2-
pentenal hi the oxidation of 3-hexene-2, 5-dione that was reported hi Ref. (2).
-278-
-------
We identified glycolaldehyde in the butenedial-OH reaction and hydroxyacetone in OH-
initiated 3-hexene-2, 5-dione. According to their time series, they are secondary products. One
possible pathway for glycolaldehyde formation is via acrolein reacting with OH radicals.
A Varian 3700 GC- BCD (electron capture detector) is directly connected to the chamber
sample lines and is used to detect PAN and N-containing compounds. PAN was detected in OH-
initiated 4-ox_o-2-pentenal and 3-hexene-2, 5-dione reactions. It was formed at about 1100 -1200
EDT and could be produced from products of dicarbonyls such as acetaldehyde and
methylglyoxal. Comparing PAN's measurement with the N-containing compounds measured by
the NOX analyzer, which responds to virtually all RONO2 and ROONO2 compounds (Figure 9),
we can see that within the measurement variation, the concentration of N-containing compounds
measured later in the experiment on the NOX analyzer is equal to the concentration of PAN
formed in 4-oxo-2-pentenal/NOx experiment. Note that during 0730-1100 EDT there was a huge
gap between the NOX analyzer measurement of N-containing compounds and GC-ECD
measurement of PAN. During this period, NO2 should be close to zero, because the shape of Os
time series is flat. Only about 0.35 ppm 4-oxo-2-pentenal remained in the system during this
time. Right after all the 4-oxo-2-pentenal was consumed, at about 1100-1200 EDT, O3 started to
build up again, and PAN began to get to its highest peak. This implies that some other unstable
organic nitrate(s) formed during 0730 - 1100 EDT in this system. Most likely these compounds
were unsatutated peroxyaceyl nitrates, which formed from unsaturated peroxy acyl radicals
reacting with NO2. They apparently had a short lifetime as the temperature was increasing during
this period, thus served as a temporary storage reservoir for NOX. In contrast, Figure 5 of
supporting data shows that in the 3-hexene-2, 5-dione experiments, PAN was not the only
species detected by NOx-analyzer after the parent hydrocarbon had been consumed.
Malonaldehyde as primary product was both found in OH initiated butenedial and 4-oxo-
2-pentenal reactions. Unlike methylglyoxal and glyoxal, the formation of malonaldehyde can not
be explained via rearrangement of the alkoxy radical formed after O2 addition and NO to NO2
oxidation. Rather an H-shift step appears necessary to produce this compound. After OH
addition, such an H-shift followed by decomposition could result in the subsequent formation of
malonaldehyde. For further details see ref. 11. Formation of this compound may share the same
pathway as some of the unidentified compounds such as the M476 and M283 species.
It has been argued (4) that methglyoxal was not a major product in the OH-initiated 3-
hexene-2, 5-dione reaction. Instead, isomerization was claimed to be more important than
decomposition. Our results do not confirm this argument as we did not detect the compounds that
would result from isomerization, such as Ce hydroxy triones. Our data strongly support that the
reactions leading to methylglyoxal predominate.
Since Oa was formed soon after sunrise in OH-initiated experiments, 1,4-unsaturated
dicarbonyls also reacted with Os to yield the energy-rich ozonide during the daytime
experiments. In nighttime reactions, however, with the use of high concentration cyclohexane as
the OH radical scavenger, reaction with Os is the major pathway. Reaction pathways of 1,4-
unsaturated dicarbonyls with Os are outlined in Scheme II. The energy-rich ozonide rapidly
decomposes to methylglyoxal /glyoxal and an energy-rich biradical or forms epoxy carbonyls.
The energy-rich biradicals can undergo stabilization and then react with H2O to form organic
-279-
-------
acids or carbonyls. They can also decompose to produce carbonyls, OH radicals, and other
products.
Acknowledgements
This work has been funded in part by the U.S. Environmental Protection Agency
(R824789-01-0). We thank Dr. Zuo Wang of the UNC Chemistry Department, Dr. Ramiah
Sangaiah of the UNC Department of Environmental Science and Engineering for synthesizing
compounds. We are also grateful to the Chromatography Systems Division of Varian Associates,
Inc. Walnut Creek, CA, for making available a Saturn IIGC ion trap MS for this study.
Supporting Information Available
More figures from this study are available. They include chamber conditions and O3 time
series in Oa-dicarbonyls-cyclohexane experiments, time serie^bf carbonyls in butenedial-NOx
experiment, tune series of products in 3-hexene-2, 5-dione daytime and night experiments, 63-
dicarbonyl reaction rate constants, UV spectra of dicarbonyls. Also included are documents of
synthesizing 1,4-undaturated dicarbonyls and figures of their NMR, MS spectra.
Literature Cited
1. Jeffries, H. E. Photochemical air pollution. In Composition, chemistry, and climate of the
atmosphere; Singh. H.B. Ed.; Van Nostand-Reinhold: New York, 1995, pp308-348.
2. Yu, J.; Jeffries, H. E.; Sexton, K. GAtmos. Environ. 1997, 31, 2261-2280.
3. Bierbach, A.; Barnes, I.; Becker, K. H.; Wiesen, E. Environ. Sci. TechnoL, 1994, 28, 715-
729.
4. Tuazon, E. C.; Atkinson, R.; Carter, W. Envir. Sci. Technol 1985, 19, 265-269.
5. Takagi, H.; Washida, R; Akimoto, H.; Okuda, M. Spectroscopy Letters, 1982, 15, 145-152.
6. Dumdei, B. E., and R. J. O'Brien Nature, 1984, 311, 248-250.
7; Wiesen, E.; Barnes, L; Becker, K. H. Environ. Sci. Technol. 1995, 29, 1380-1386.
8. Raber, W. H.; Moortgat, G. K. Photooxidation of selected carbonyl compounds in air: methyl
ethyl ketone, methyl vinyl ketone, methacrolein and methylglyoxal. In Progress and
problems in atmospheric chemistry; Barker, J. R. Ed.; World Scientific: New York, 1995,
pp318-373.
9. Yu, J.; Jeffries, H.E.; Le Lacheur, R. M. Environ. Sci. Tecnnol, 1995, 29, 1923-1932.
10. Liu, X. Jeffries, H. E.; Sexton, K. G. Atmos. Environ. 1999, 33, 3005-3022.
11. Liu, X. Study of atmospheric photooxidation products and mechanisms for 1,4-unsaturated
dicarbonyls and dienes, 1999, PhD thesis, Department of Environmental Science &
Engineering, University of North Carolina Chapel Hill, ftp://airsite.unc.edu/PDFs/ese-
unc/students/PhDThesis/Liu
12. Jeffries, H. E.; Fox, D. L.; Kamens, R. M Envir. Sci. TechnoL 1976, 10, 1006-1011.
13. Grosjean, D.; Grosjean, E. Int. J. Chem. Kinet, 1998, 30, 21-29.
14. S0rensen, S. Barnes, I. Photolysis of unsaturated 1,4-dicarbonyls. In The European
Photoreactor EUPHORE report 1997, Barnes, I. and Wenger, J. Ed. 1998.
-280-
-------
Figure 1. Daytime dual smog chamber experiments on June 18, 1998. 1.5 ppmV 4-oxo-2-
pentenal and 1.0 ppmV butenedial, and 0.60ppm nitrogen oxides. Top: NO, NO2+RONO2+
ROONO2 and O3 time series (NO2+RONO2+ ROONO2 measured by NOx analyzer). Bottom:
Temperature-chamber air temperature, DewPoint-chamber dew point temperature, TSR-Eppley
total solar radiation, UV-Eppley ultraviolet radiation.
Figure 2. Daytime dual smog chamberexperiments on July 2, 1998. 0.9 ppmV 4-oxo-2-
pentenal and 3-hexene-2, 5-dione, and 0.60ppm nitrogen oxides. Top: NO, NO2+RONO2+
ROONO2 and O3 time series (NO2+RONO2+ ROONO2 measured by NOx analyzer). Bottom:
Temperature-chamber air temperature, DewPoint-chamber dew point temperature, TSR-Eppley
total solar radiation, UV-Eppley ultraviolet radiation
Figure 3. Reconstructed m/z 181 ion chromatograph of the batch sample collected from a
butenedial/NOx outdoor smog chamber experiment in the daytime. Y-axis: Relative ion current
of m/z 181 ion; X-axis: Retention time in mm.
Figure 4. Reconstructed m/z 181 ion chromatograph of the batch sample collected from a 4-
oxo-2-pentenal/NOx outdoor smog chamber experiment in the daytime. Y-axis: Relative ion
current of m/z 181 ion; X-axis: Retention time in min.
Figure 5. Reconstructed m/z 181 ion chromatograph of the batch sample collected from a 3-
hexene-2,5-dione/NOx outdoor smog chamber experiment in the daytime. Y-axis: Relative ion
current of m/z 181 ion; X-axis: Retention time in min.
Figure 6. Nighttime dual smog chamber experiments on July 7, 1998. 0.9 ppmV 4-oxo-2-
pentenal. O3 and 150 ppmV cyclohexane. Top: O3 time series. Bottom: Temperature-chamber air
temperature, DewPoint-chamber dew point temperature.
Figure 7. Time series of glyoxal, methylglyoxal, malonaldehyde in the 4-oxo-2-pentenal
daytime outdoor smog chamber experiment. R: right axis; L: left axis.
Figure 8. Time series of M283, M467 in the 4-oxo-2-pentenal daytime outdoor smog chamber
experiment. R: right axis; L: left axis.
Figure 9. Time series of O3, PAN and N-Containing compounds in the 4-oxo-2-pentenal
daytime outdoor smog chamber experiment. R: right axis; L: left axis.
Figure 10. Time series of methylglyoxal, acetaldehyde, hydroxyacetone in the 3-hexene-2,5-
dione daytime outdoor smog chamber experiment. R: right axis; L: left axis.
Figure 11. Time series of M267, M476 in the 3-hexene-2,5-dione daytime outdoor smog
chamber experiment. R: right axis; L: left axis.
-281-
-------
Figure 12. Time series of glyoxal, methylglyoxal, malonaldehyde in the 4-oxo-2-pentenal
nighttime outdoor smog chamber experiment. R: right axis; L: left axis.
Figure 13. Time series of M476 in the 4-oxo-2-pentenal nighttime outdoor smog chamber
experiment. R: right axis; L: left axis.
Secheme I. Photochemical degradation mechanisms of 1,4-unsaturated dicarbonyls with OH
radicals and photolysis (Bold M value and names represent products observed in the
experiments).
Secheme II. Photochemical degradation mechanisms of 1,4-unsaturated dicarbonyls with Os
(Bold M value and names represent products observed in the experiments)
-282-
-------
E
CL
Q.
of
0)
O)
O
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
1 ' I ' I 'I ' I ' 1 '
1.5 ppm 4-oxo-2-pentenal
1.0 ppm butenedial
NO
T
T
18-Jun-98_
NO2 + RONO 2+ROONO
8 9 10 11 12
Hours, EOT
13
14 15
9 10 11 12
Hours, EOT
13
14
16
1.0
0.9
0.8
0.7
0-6 O
O
0-5 I
0.4 ~~
0.3
0.2
0.1
0.0
13
I ' I ' I ' I ' I ' I ' I ' I ' I ' I '
0.0
15 16
-283-
-------
1.0
0.9
0.8
0.7
O 0.5
4= 0.3
Z
0.2
0.1
0.0
1 ' I ' I ' I ' I ' I ' I ' T71 I ' I r
0.9 ppmV 4-oxo-2-pentenal fs 02-Jul-98
0.9 ppmV 3-hexene-2,5-dione''
(to 1.2 ppm) /
'
NO,+RONO2+ROONO2
£
J
>~
sf
0>
Q
f
V
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
:§
10 11 12 13 14 15
10 11 12 13 14 15 16
-284-
-------
co
co
CO
apAijap|euo[E|/v~
o>
CM
!•-
CM
CO
CM
OHOadd
ZSZVt
£831
CO
o>
00
o
o
o
in
uo| 121 z/ui jo
••»
UO|
o
o
o
o
CM
-285-
-------
£6ZIAI/(3Ai}eAuap-ououi)
S2£IAI/(aA!jeAuap-ououi)
CM
. f-.
CO
CM
CM
c
I 2 E
o
o
o
o
o
(O
o
o
CO
o
o
to
o
o
o
o
uo| \.2l z/ui jo
uoj
-286-
-------
auoip-g'£-auaxaH-£
|exoA|6|Ai|ia|/\|
CO
o>
CM
m
n
CN|
c
. r*-
CO
uo|
z/iu jo
uoj
-287-
-------
1.2
1.1
1.0
0.9
0.8
E 0.7
£ 0.6
-------
4-oxo-2-pentenal, (L)
malonaldehyde, (R)
9 10 11 12
Hours, EOT
' -289-
-------
1.0
0.9
0.8
Q. 0.7
D.
§ 0.6
2J 0.5
**
§ 0.4
8 0.3
0.2
0.1
0.0
0.4
4-oxo-2-pentenal, (L)
Jul 02, 1998
, i
M476, Rt=24.39, 24.46, (R)
O
o
3
O
(D
•-*
0.2 3
o
TJ
•O
0.1 3
9 10 11 12
Hours, EOT
13
14
0.0
15 16
-290-
-------
1.0
0.9
0.8
' I ' I! '\ l ' I ' I ' l ' l ' I ' I ' i '
4-oxo-2-pentpnal, (L) Jul 02, 1998
NO+RONCh +ROONO2,
0.4
9 10 11 12
Hours, EOT
-291-
-------
0.20
I ' I ' I ' I ' I ' I '
Jul 02, 1998
S-hexene-Z.S-diond, (L)
methylglyoxal, (R)
acetaldehyde, (R)
9 10 11 12 13
Hours, EOT
14 15
-292-
-------
1.0
0.9
0.8
^
3-hexene-2,5-dione, (L)
M476, Rt=23.81, (R)
10 11 12 13 14 15
' I ' I ' I ' I ' I
Jul 02, 1998
M476, Rt=24.44, (R)
M476, Rt=23.89, (R)
M476, Rt=24.39, (R)
0.20
H0.15 g
0>
(D
o
0.10 §
o
(D
3
I
0.05 O
-293-
-------
1.2
1.1
1.0
E 0.9
CL
o. 0.8
O 0.7
2 0.6
§ 0.5
u
§ 0.4
0 0.3
0.2
0.1
0.0
21
I ' I
4-oxo-pentenal, (L)
Jul 07, 1998 -
methylglyoxal, (R) ~?
glyoxal, (R)
\
malonaldehyde, (R)
22
23 24
Hours, EOT
25
0.15
0.14
— 0.13
-0.12
— 0.11
-0.10
-0.09
-0.08
-0.07
— 0.06
— 0.05
— 0.03
-0.02
-0.01
-Jo.oo
O
o
|
26
-294-
-------
I ' 1
4-oxo-pentenal, (L)
Jul07, 1998
M476, Rt=24.48, (R)
M476, Rt=23.93, (R)
\
M476, Rt=24.43, (R)
23 24
Hours, EOT
00
26
-295-
-------
CD
T3
| 4.0
-o J?O
XO&
II II CC
CM CVJ
- - x"
X X O
i
c
g
"o
2
to
JD
CD
X
^ CD
JD
r~
(D
ro ?-
'•0 CN
§ g
3O
g
•~
CD
CO
CM
O
I 1
g
'~5
-o
CD
1
CO
x"
O
-296-
-------
CO
LO
CM
O
o
I
O
X
O
73
'o
CD
O
o
CM
X
Q_
CO
O
CO
X
o
*->
a>
re
x
o>
CD
0)
2
ro
O
-CO
2
m
E
o
TO
_N
-297-
-------
w
•4-1
cu
E
"C
CD
Q.
X
111
CU
x»
E
to
JC
o
O)
o
E
CO
1_
o
o
TJ
•»->
O
o
c
o
CO
o
5
73
0)
•*-*
I
4->
CO
CO
c
CO
CO
c
T!
C
O
o
c
o
'•§
ca
CD
LU
_J
03
_
ic
If
3
to
CD
Q.
.8
W
C
o
'I
•*->
CD
O
c
o
o
•*-•
'o
Q.
CD
TJ
CD
XI
CO
x:
0
o
0
X
o
z
o
1
JE
E
Q.
3:
Q.
^-
E
Q.
Q.
p
Q
"To
Q)
TJ
"55
O
c
o
•s
(0
o
o
CD
OO
CM
CO
OO
CM
CD
O
CO
CO
oo
CM
CO
CM
h-
CM
CO
O
CO
oo
CM
O
CD
CM
06
OO
CM
x-
co
o
02
CM
O
CD
CM
00
OO
CM
^_
co
CD
CD
CM
O
CD
OO
OO
CM
•r-
CO
T —
O2
CM
0
CD
CM
OO
00
CM
^_
CO
O)
CM
O
O2
O
CD
CM
co
CD
CM
N-
CD
CM
O
CD
CM
O
CD
CM
CO
CD
CM
N-
CD
CM
CM
CD
CM
1
O2
CM
^
O2
06
02
CM
CM
CD
CM
i
T —
CD
CM
•*
CD
CM
00
O2
CM
CM
CD
CM
i
CD
CM
LO
CD
CM
O2
CM
CM
CD
CM
i
O2
CM
LO
CD
CM
CD
CM
O
CD
CM
O
CD
CM
CD
CM
CD
CM
O
CD
CM
ci
CD
CM
CD
CM
i
CM
CO CO
CO CO
CO CO
CO CO
O O CD
CO CO
CO CO
CO CO
CO CO
CD CD CD
I
TJ
•sf
CO
CD
CD
O
CO
CD
t
CD
O
Q
LU
butenedial
Oc0697
CD
CD
LO
O
O
CO
CM
CM
O
LU
_l
CQ
butenedial
Oc1397
CO
0
LO
o
o
LO
I
O
o
CD
O
Q
LU
"co
c
0
CL
cp
4
oo
CD
oo
c:
—>
CO
o
-
o
o
LO
I
o
o
CD
O
LU
_J
CQ
butenedial
Jn1898
CO
CD
OO
CD
O
CO
LO
cS
CD
O
Q
LU
OH
"CD
c
0
Q.
6
X
O
oo
CD
CM
O
— 5
CD
CD
OO
CD
O
CO
LO
0
CD
O
LU
H>
CQ
0
C
O
TJ
LO
CM"
CD
1
0
x:
CO
oo
02
CM
O
—3
O
LO
o
CO
T—
o
ci
LO
o
CM
LU
_l
CQ
0
C
CO
X
0
x:
o
^
o
CD
C.
fl\
\1J
-4—»
0
Q.
CM
6
4
00
CD
O
O
LO
O
CO
O
ci
LO
o
CM
Q
LU
cyclohexane
00
CD
O
~i
O
o
o
o
CM
O
CM
LO
O
CM
LU
ID
CQ
0
C
JO
0
x:
o
o
>,
butenedial / c
8681-nv
o
o
o
o
CM
CD
CM
LO
O
CM
Q
LU
o:
cyclohexane
oo
CD
OO
<
O
o
02
CD
O
0
1
O
CM
LU
_J
CQ
0
C
CO
X
0
t—
o
o
£?»
"35
c:
o
TJ
LO_
0
1
0
x:
CO
Au2598
o
o
T —
0
o
T —
o
CM
Q
LU
o;
cyclohexane
Au2598
o
o
02
CD
O
O
LO
O
CM
LU
ID
CQ
0
C
CO
x:
o
o
o
—
_
CD
-*-*
0
CL
O
oo
CD
LO
CO
o
o
02
CD
O
x —
O
LO
o
CM
Q
LU
K
0
CD
X
0
Q
"o
>>
cB
c
o
TJ
LO
CM
CD
0
X
0
x:
CO
oo
CD
LO
l: —
CO
0
TJ
'w
0
XI
E
CD
O
0
x:
w
TJ
TJ
CD
W
13
CD
LU
OQ
TJ
C
CO
Q
111
QL
-298-
-------
Secular Weight
w
5
f Unidentified Carbonyls with Known
o
V)
1
<•-«
o
3
i_
•*-•
0)
XI
'55
CO
o
Q.
CM
UJ
_J
CQ
f
reactants
structure
0)
X!
"53
CO
o
Q.
*
>
u.
^>
i
CO
c
0
CL
CM
6
X
o
4
o"
_co"
o
i_
CO
o
1
CL
0
o"
rated hydroxy carbonyls, C3 dicarbonyls
i
CO
CO
O
CM
h-
1 —
CD
CM
lutyraldehyde, methyl
xj
/n
carbonyls ( can not be isobutyraldehydc
•o
&
1
-*— »
CO
CO
naldehyde)
CO
ne, glycidaldehyde, methylglyoxal, or mi
o
-4— •
0
-*— »
0
Is, C4 dicarbonyls, C3 4-oxo-2-pentenal
c
rated hydroxy carbonyls, C4 epoxy carbc
-*— '
CO
CO
ZJ
0*
CO
00
•,—
oo
CM
t
CO
JQ
0
C
03
O
5 saturated carbonyl, C4 cyclic carbonyl i
o
co"
0
.0
-*-•
butenedial
4-oxo-2-pentenal
:y carbonyls but can 4-oxo-2-pentenal
y carbonyls, C3 hydroxy dicarbonyls
1
cT
00
00
CO
oo
CM
K.
2
•a
x:
in
O
0
o
co
CO
CO
0
en
CM
CO
c
o
X3
1
|
o"
CM
0
•<~
U
CM
O}
•^f
2
1 —
en
CM
-methyl-2-butanone, 3-hexene-2,5-dione
CO
X
o
1
1
0"
c
o
c
i
CM
£
"0
CO
|
CO
0
"o
lroxy-2-pentanone)
in
c
03
0
C
.0
I
in
CM
CD
C
0
X
0
CO
c
CO
o
O)
o
CO
c
o
03
o
(O
o
i-ones, C5 triones, C5 epoxy dicarbonyls
^
03
CT
o'
M-
•^
O
o
m
^
C3>
O
CO
CO
c
0
•*— '
0
CL
CM
I
O
X
o
4
6 unsaturated hydroxy carbonyl)
o
0
o
w
CO
n also be C6 hydroxy butenedial
CO
o
y triones, C5 hydroxy dicarbonyls (M31 1
X
2
•£
jz
o'
CD
^~
U
CO
o
m
^
v—
^^
CO
03
C
0
•4— «
£Z
0
CL
CM
6
X
O
4
o
-CO
0
ex
CM
i
t
X
2
4
0
•5
c.
c.
CO
o
"3
X3
co~
>,
o
_Q
8
CO
c:
0
.*— *
c.
0
CL
CM
6
X
O
4
i-ones, C6 epoxy dicarbonyls
•a
03
CT
in
O
CO
CM
x~
CO
CM
CO
0
C.
CO .2 03
c -a c:
n)7^ LO i^f
0 °H 0
9- ® 9-
^ 0 ^
O X O
X 0 X
o x: o
41 i
CO •*
co
0
c
o
T:
|
in
O
o
CO
T—
u
o
CM
m
^
in
CM
CO
1
JK:-
c:
n
CO
CO
T —
CO
CM
CO
-299-
-------
with Known Molecular Weight (continued)
U)
d Carbony
.2
*5
a>'
•o
'E
3
"8
U)
ructure:
W
_o>
.Q
'53
to
0
Q.
CM
LU
_J
CO
H
reactants
2
13
*j
U
,J3
CO
0>
JD
'53
0
Q.
5
LL
^^
>
•E
cetyl) butenedial
4-oxo-2-pentenal
.2
0
1
co
c.
o
•e
CO
_o
'•D
cf
0
.O
•1— «
0"
CD
00
CD
^J-
2
3-hexene-2,5-dione
•bonyl and C4 unsaturated hydroxy • 4-oxo-2-pentenal
CO
o
d"
_of
0
JD
CO
_o
Q.
0
o'
co"
0
c
.0
-*-*
o'
o
0
o
en
^c
^
0
c
o
LO_
CD
C
0
X
0
CO
x"™V
pentadione
-sl-
CM"
0
-t— »
o
c
c
CO
o
c
o
JD
CO
_o
CO
T3
0
C
•§
JD
"oT
c
o
TK
LfS
CM"
CD
cz
0
X
0
f~
CO
0
'o
c
CO
^
c
0
-e
CO
o
T3
T3
m
unsatun
(D
O
CM
T~
CM
O
LO
^
0
1_
CO
o
D)
'0
L.
_co
o
0
o
E
J3
I
"co
•c
0
^
X
CQ
LL
Q.
0
JZ
'o
D)
1
CO
o
0
o
E
CQ
-derivative, the second number is that
O
c
PFBHA mo
0
f*~
*
M—
O
£
D)
'0
CO
molecul
0
co
0
JD
|
C
£2
>^
0
jr
o
1
CO
0
•a
X
CQ
LL
CL
0
JZ
H—
0
-300-
-------
TABLE 3. Carbonyl Products Detected in the Batch Sample Collected from
O3-lnitiated 1,4-unsaturated Dicarbonyls Outdoor Smog Chamber Experiments
(M numbers are molecular weight of the PFBHA derivative)
Compound
Butenedial
4-oxo-2-pentenal3
3-hexene-2,5-dionea
products
formaldehyde, glyoxal, methylglyoxal
M490, M476.M506
formaldehyde, glyoxal, methylglyoxal, malonaldehyde
M267, M309, M325, M476, M502, M520
formaldehyde, methylglyoxal,
M476,M490,M504
aCyclohexanone also detected (from cyclohexane added as OH radical scavenger)
-301-
-------
TABLE 4. Comparison of Rate Constants for 1,4-Unsaturated Dicarbonyls with O3
compounds
kO3
temperature references
(10-18xcm3molecule-V1) (K)
Butenedial*
cis-butenedial
trans-butenedial
4-oxo-2-pentenal*
cis-4-oxo-2-pentenal
trans-4-oxo-2-pentenal
3-hexene-2,5-dione*
cis-3-hexene-2,5-dione
trans-3-hexene-2,5-dione
1.6±0.1
2
2
4.8±0.8
2
2
3.6±0.3
1.8±0.2
8.3±1.2
294-298
296
296
293-297
296
296
295-297
298±2
298±2
this work
(3)
(3)
this work
(3)
(3)
this work
(4)
(4)
* mixture of cis- and trans-
-302-
-------
A Modelling System for Mechanism Evaluation Using Chamber Data
Paul A. Makar
Ping Y.Li
Atmospheric Environment Service
4905 Dufferin Street
Downsview, Ontario
Canada M3H 5T4
pauLmakar @ ec.gc.ca
Gas-phase chemical mechanisms for use in regional or global tropospheric atmospheric
chemistry applications require validation against measurement data in order to assure the
modeller and policy-makers that the mechanism adequately represents the chemical processes of
the real atmosphere. One means of mechanism validation is through the use of smog-chamber
data, in which atmospheric chemistry experiments are performed under controlled conditions,
and eliminate the potential confounding effects of non-chemical processes in the ambient
atmosphere. The absence of the ambient atmosphere's confounding factors (in which some of
the chemical changes may be due to advection, diffusion, emissions, deposition, convection,
rainout and washout) allow a worthwhile comparison of chemical effects alone.
The need for a means of mechanism verification was identified as one of the components of the
Atmospheric Environment Service Unified Regional Air-Quality Modelling System
(AURAMS). This is a regional gas and particulate matter modelling system which is the subject
of current research at the AES Laboratory in Downsview, Ontario. One of the key components
of this modelling system is the gas-phase chemistry module, which includes an operational
simulation system used within the regional model itself, and a mechanism validation module to
be used in determining the validity and expected performance of a mechanism when used within
the regional model. A description of the validation module its first applications are the subject
of the current work.
The research requirements for the validation module were such that the system would have to be
capable of performing the following tasks:
(1) Assessing the accuracy of gas-phase reaction mechanisms.
(2) The use of the mechanisms in the module would have to be consistent with their use
in the regional model (with regards to lumping of emissions and calculation of
reaction rates).
(3) Provide a means of immediate verification/assessment of updates to gas-phase
chemistry, so that the regional model's chemical mechanism is always current and
up-to-date.
(4) Use recognized statistical measures of mechanism performance.
(5) Create and store input and output in a format allowing simple and direct comparison
of model and measurements - the i/o must be portable to the greatest extent possible.
(6) The validation of a mechanism and/or its components must performed in a small
amount of processing time, preferably on a vector supercomputer.
-303-
-------
Any system of mechanism validation will have five major components. These include: a
process for reading in measurement data, a conversion system whereby the raw data is converted
into a format suitable for modelling, a simulation system which numerically predicts
concentrations based on the initial conditions (starting concentrations in the chamber) and
boundary conditions (temperature, pressure, humidity, light intensity, wall effects) of each
experiment, a package for comparing the model results to data, and the creation of summary
statistics as the final product of the system.
In the AURAMS mechanism validation system, the measurement data is read in by a suite of
chamber-specific programs (UNCREAD, SAPRCREAD, DTCREAD) which also convert the
measurements to a common format, in which all data appear on the same time interval. The
concentrations of all measured species are linearly interpolated to the same time interval as the
NOX and Os measurements. The NOX and Os values usually have the finest tune resolution of the
measurement record in a typical chamber experiment - the linear interpolation results in no loss
of information. The use of linear, rather than higher order, interpolation prevents the potential
creation of spurious maxima or minima in the measurement record. The interpolation allows a
more direct comparison of measurements to model results, since the latter are constrained to
operate on a single time step. The tune-interpolated data from this stage are stored as spread-
sheet readable tune series, each row in a record being the measurements (or interpolated
measurements) specific for a given time. The speciation for this intermediate output is the same
as that of the measurement record itself. Time series records of chamber boundary conditions at
the measurement times of NOX and OB are also created at this stage.
The second stage of data conversion (program CONREAD) has several subcomponents. Any
measurements reported as total concentrations of pre-set mixtures of hydrocarbons are converted
to the detailed speciation through the use of a fractionation input file. The hydrocarbon
measurements are then lumped in the same fashion as detailed emissions in a regional model
(Middleton et al., 1990). That is, the speciation employed in the regional mechanism will in
general be simpler than the speciation in the measurement record; the measurements must
therefore be lumped in a fashion consistent with the lumping of the highly speciated emissions
data available for regional modelling. The detailed, measured species are first assigned to an
"emitted" speciation. This assignment is done on a mass-weighted basis, and provision is also
made for additional weighting factors. The intermediate, "emitted" species are then lumped into
a "model" speciation. This second stage of lumping is performed using reactivity weights
designed to preserve the total organic peroxy radical formation rate, of importance in accurately
simulating ozone formation. Again, provision is made for additional or other weighting factors,
to allow for other means of species compression. The final result of these operations is a second
data set, time interpolated to the NOX and Os measurement interval, with all hydrocarbon
measurements in the speciation of the mechanism for which validation is desired.
Following the creation of the mechanism-specific set of chamber data, simulations are performed
using the initial concentrations from these files and the chamber boundary conditions for a set of
experiments specified by the user. These are done using a third code (CHAMRUN), which
makes use of a vectorized Gear-type solver usually employed in simulations of regional or global
atmospheric chemistry (SMVGEAR, Jacobson and Turco, 1990). As part of the processing,
three important modifications are made to the data to allow for the use of this very accurate and
-304-
-------
very fast numerical solver. The constraints posed by the solver (due to its intended use in a
regional model) are a single number of integrations for all experiments, and the use of a single
time-step between updates to reaction rates for all experiments.
First, the chamber boundary conditions data, after being read in by the simulation model, are
appended so that the conditions for each experiment all have the same total number of
measurements. This is done in two stages — the current set of simulations is searched to find the
simulation with the greatest total number of measurements, and the last set of measurements in
the remaining experiments is repeated until all boundary conditions have the same total number
of measurements. Simulations are performed for this "maximum" number of steps, but only
the actual number of steps for any given experiment is retained on output.
Second, a default time step between updates to chamber boundary conditions is set (these
boundary conditions give rise to the reaction rates). The reaction rates for each experiment are
then multiplied by the ratio of the actual time between updates to this default value. The
simulation model thus operates on a default timestep, while the scaling of the rates ensures that
the same model produces the same results as if different time steps were employed for each
simulation to be performed.
Both of the above modifications can be classified as mathematical tricks which allow the use of a
regional model simulation package on chamber data with a varying number of measurements and
time-steps between data. The modifications have no effect on the accuracy of the simulations,
but allow the processing of hundreds of chamber experiments on a vector supercomputer in a
few minutes of clock time. '
CHAMRUN also automatically adds variables that track the mass through every reaction in the
mechanism under study. This allows easy process analysis of the mechanism, and comparison to
other mechanisms analyzed using the system.
The output from the simulation code takes the form of columns of model variables on the same
time interval as the interpolated measurements, in the same format as the input data. Any
spreadsheet or graphics package may be used to compare time series of model and measured
variables for any individual experiment or set of experiments. For the purposes of providing
objective evaluations of mechanism performance, a final stage of processing takes place
(program CHAMSTAT).0 This program reads the model-speciated measurement and model
simulation time series, and retrieves the maxima of ozone, NO2, formaldehyde, and PAN. The
time taken to reach these maxima, the time to the NO/NOa crossover, and the average value of
the change in (O3 - NO) per step as a function of tune, are also retrieved. These are outputted in
a final table, again as columns readable by spreadsheet codes or graphics packages.
During the course of the creation of the system, a few minor problems were encountered with a
small number of the measurement files, which prevented their further use. These are mentioned
here for the purposes of providing feedback for future chamber experiments. The problems
encountered included repeat time stamps (same time being repeated with different
measurements), different names being used for the same chemical species, and multiple
-305-
-------
measurements of the same species, with no information being given as to the expected relative
accuracy of the different measurements.
As an example use of the system, a set of 14 Ethene experiments from the UNC database were
selected for processing with the AURAMS mechanism. The conversion of the whole UNC
database to the model speciation using CONREAD required about 3 minutes of processing time
on a workstation. The corresponding simulations for the set of experiments (each of duration of
about 10 hours) required less than a minute of processing time using CHAMRUN on a vector
processor. In the next few months, the system will be used to validate the AURAMS
mechanism, and compare its performance to that of the ADOM-H mechanism which it
supercedes and the Master Chemical Mechanism.
References:
Jacobson, M.Z. and R.P. Turco, SMVGEAR: A sparse-matrix, vectorized Gear Code for
Atmospheric Models, Atm. Environ., 28, 273-284, 1994.
Middleton, P., W.R. Stockwell and W.P.L. Carter, Aggregation and analysis of volatile organic
compound emissions for regional modelling, Atm. Environ., 24, 1107-1123, 1990.
-306-
-------
Chamber Evaluation of Process Diagnostics and Photochemical Indicators
Gail S. Tonnesen
University of California, Riverside
College of Engineering Center for Environmental Research and Technology
Riverside, CA 92521
Introduction
Historically photochemical mechanisms and air quality models have been
evaluated primarily in terms of there ability to simulate observed Os data. There is an
increasing awareness that mechanism and models must also be evaluated in terms of their
ability to simulate the fundamental chemical processes that control Os formation and the
sensitivity of Os to emissions reductions (Arnold et al., 1998). Such evaluative methods
include process diagnostics which are useful for characterizing Os photochemistry
(Jeffries and Tonnesen, 1994), and photochemical indicators which can be used to
characterize O3 sensitivity to VOC and NO* (Milford et al., 1994; Sillman, 1995; Chang
et al., 1997; Blanchard et al., 1999; Tonnesen and Dennis, 1999). Both of these
approaches are currently being used in diagnostic model evaluations (e.g., Sillman et al.,
1997a,b, 1998; Imre et al, 1999). There are considerable uncertainties, however, in the
usefulness of these methods due to uncertainties in each of the many sink and source
terms that can contribute to production or loss of trace species. For example, there are
uncertainties of 20% or more in important gas phase reactions (Donahue et al, 1997; Gao
et al., 1995,1996), large uncertainties in heterogeneous chemistry (Dentener and Crutzen,
1994) and in emissions inventories, deposition rates and transport. Because of these
uncertainties, it can be difficult to use ambient data to validate model simulations of
chemical processes. Furthermore,, the particular values of photochemical indicators that
distinguish NO* sensitive or VOC sensitive conditions have only been derived from
model simulations. To date there has been no empirical validation of the usefulness of
these indicators. Here we propose that before process diagnostics and photochemical
indicators can be used with confidence in model evaluations, then: behavior must be
characterized under the relatively well controlled conditions of chamber experiments.
Each of these approaches is discussed below, and the measurements necessary to evaluate
them in chamber experiments are listed.
Definitions
Reactive odd nitrogen (NO*) is traditionally defined as the sum of NO
There are, however, several other reactive forms of odd nitrogen which participate in Os
and PM photochemistry where these include nitrous acid (HONO), the nitrate radical
(NOs), dintrogen pentoxide (NaOs), peroxynitric acid (HNO4). These species typically
have low concentrations for daytime photochemistry, but for completeness the total sum
of reactive odd nitrogen can be defined as NOAT(e.g., Dentener and Crutzen, 1994):
-307-
-------
= NO + NO2 + NO3 + 2 N2O5 + HONO + HNO4
The relatively inert forms of odd nitrogen (NOZ) can then be defined as the sum of
nitric acid (HNO3), organic nitrates (RNO3), participate nitrate (NO3-), and peroxyacetyl
nitrates (PAN):
NOZ = HNO3 + RNO3 + NO3" + PAN
And then total odd nitrogen (NOy) can be defined as
For convenience HC will be defined as the sum of all species that can react with OH to
produce O3. Thus, HC includes VOC, methane, carbon monoxide and biogenic carbon.
The family of 'odd oxygen (O*) is the sum of all species that can act as reservoirs for
atomic oxygen
O* = O3 + O'D + O3P + NO2 + 2 NO3 + 3 N2O5 + HNO4 +HNO3 + RNO3 + NO3" + PAN
Production of O* in the troposphere occurs almost exclusively by the reaction of NO with
peroxy radicals (HO2 and RO2):
(Rl)
(R2)
NO + HO2 ^ NO2 + OH
+ RO2->NO2
so gross production of Oz can be defined as the rate of reaction of NO with HO2 and RO2
Pc(0z) = hi NO H02 + kj NO R02
There is also some photochemical destruction of Oz (L0x), so net production of Oz can be
defined as:
P(0Z) = PG(0Z) -I**
This definition is identical to Rsmog as defined by Johnson (1984), and this definition is
convenient because P(O*) can be evaluated through ambient measurements as the
production of O3 plus the amount of NO oxidized to other forms of NOy. It is also
convenient to define HC as the sum of all species that can react with OH to produce Oz:
HC = VOC + biogenic VOC + CO + CH,
-308-
-------
Process Diagnostics
Figure 1 illustrates most of the important components of the chemical processes that
control the photochemical formation of Os. The first step in Os production is the
initiation of new radicals. In the models, the primarily source of radical initiation is
photolysis of HCHO and Os, with a lesser contribution by photolysis or decomposition of
other radical precursors. There is uncertainty, however, whether the these "minor" radical
sources are accurately represented in air quality models. For example, HONO may be a
significant source of radical initiation particularly in early mornings (Kleffman et al.,
1998).
OH Chain Length
/ JTp"
HO* Termination
Propagation & PG(OX)
HKVNO->NO2+OH
NOx Termination
P(NOz) - P{HNOs,PAN,ONrT)
j
PrP(OH)
NOX Chain
Length^
\
P(0X)
O3 Production
Efficiency - P(O3)/P(NOz)
Chemical
OxLoss
O3 Transport !n
OaTmnsportOut
O3 Deposition
Figure 1. Process diagram illustrating important diagnostics for characterizing
photochemical production of ozone.
-309-
-------
Radicals may be produced directly in the form of OH, or as HO2 and RO2
followed by conversion to OH via Rl or R2. OH radicals can then attack HC to produce
peroxy radical which can react with NO to produce O* in the form of NOa, and also
recreate the OH radical.
The OH propagation efficiency, Pr0n is the fraction of OH recreated for each OH
that reacts. The propagation efficiency is always less than one because radicals are also
destroyed in termination reactions that produce peroxides or that convert NO^fto NOz.
The OH chain length is the average number of times a new radical cycles through the
chain reaction until being removed hi a termination reaction, and the chain length can be
calculated as:
OH chain length = 1 + PrOH+ PrOH2
= l/(l-PrOH)
+
The OH chain length acts as a multiplier so that total OH production P(OH) is the
product of the OH chain length multiplied by OH initiation.
The result of this sequence of reactions is some gross production of odd oxygen,
some of which is lost by photochemical destruction, and some net production P(OX~)
which is initially expressed in the form of NOa. But because Oa, NO2 and NO exist in a
photostationary-state equilibrium, some of this NO2 will photolyze to produce O3 thereby
recreating an NO that can go on to produce more O*. The NO* chain length can be
defined as the average number of NO to NO2 conversions mediated by a molecule of
NOx before it is converted to inert NOZ. It is important to note that it is difficult to
explain the processes that contribute to peak OB levels simply by evaluating production of
Oa — a more complete characterization of the photochemical dynamics is obtained by
evaluating the total budget of Ox and then explaining the resulting peak Oj level as the
amount of O* that remains after oxidation of NO emissions.
Local and Cumulative Diagnostics
It is also useful to distinguish between local diagnostics and cumulative
diagnostics. Local diagnostic refers to the instantaneous chemical production rates, e.g.,
the rate of production of radicals or O*, or ratios of production rates. Local diagnostics
can not be evaluated by analyzing ambient concentration time-series of a product species
at a given site because it is impossible to distinguish the relative contributions of
chemistry and transport to changes in the concentration of the product species. Local
diagnostics can be evaluated, however, by measuring the concentrations of the reactant
species and then using the rate constants to calculate the instantaneous production rates.
An example would be using actinic flux and ambient concentrations of 03 and H2O to
estimate the local production rate of OH radicals from Os photolysis.
Cumulative diagnostics include those species concentrations that characterize
cumulative production rates integrated over a period of hours or days. Examples of these
-310-
-------
would be using the concentration of O3, NO2 and NOZ as a measure of the cumulative net
production of Ox integrated over an air parcel's trajectory.
Several field studies have also made detailed measurements of OH, HCh and RC>2
to attempt to characterize radical budgets, and hi each case the measurements have failed
to agree with the models simulations (Crosley, 1997; Stevens et al., 1997; Cantrell et al.,
1997). Thus there is a great deal of uncertainty in radical budgets, and an important
research need is to attempt to characterize radical budgets in a controlled setting.
Environmental chambers provide an ideal opportunity for characterizing radical
budgets for a number of reasons. Chamber conditions and concentrations of precursors
can be carefully controlled and varied for a range of conditions of interest; chambers use
a confined volume of air so that interactions between transport and chemistry do not
complicate the analysis; and sophisticated equipment can be quality controlled and
operated more easily in a laboratory setting compared to field experiments. An additional
advantage of studying process diagnostics hi a chamber is that the confined volume of air
allows for time integrals of measured local diagnostics to be compared directly to
measurements of the cumulative diagnostics. Finally, new chamber experiments must be
performed to evaluate process diagnostics and photochemical indicators because the
existing chamber data base is at high VOC and NO* concentrations and because the
existing data base lacks many of the necessary measurements, particularly measurements
of NO2, HNOs and peroxides, and radicals.
Wall effects will continue to be a problem in chambers studies (Carter et al.,
1982; Jeffries et al., 1990), but it may be possible to minimize such effects in a low-NO*
chamber (e.g., Bailey et al., 1996; Simonaitis et al., 1995), and hopefully wall effects can
be better characterized with the more sophisticated set of measurements that are
becoming available in chamber facilities hi the U.S. and Europe.
Photochemical Indicators
The motivation for the development of photochemical indicators is the desire for
measurements that will determine whether air quality models are faithfully simulating
real-world sensitivity of O3 to changes hi precursor emissions, hi terms of an ozone
isopleth diagram, illustrated in Figure 2 (top), indicators are used to locate ambient
conditions relative to the [O3] ridge line or the P(O*) ridgeline which are defined,
respectively, as:
where E^OX represents emissions of NO*. Alternatively, indicators can be defined for
conditions of equal sensitivity to VOC and NO*:
o, dEvoc
and
dE
voc
-311-
-------
Peakjpg] (ppb)
MBC,
160
120
40
SCO 400
Initial VQC
OH Chafo
(8 AM tp 7 PM av«.)
ISO
t20
2 00
40
200 400 ,eao sop
1200 1400
Figure 2. Response surface plots showing constant contours of Os (top) and the OH
chain length (bottom) as a function of VOC and NOx emissions levels.
-312-
-------
Indicators cannot replace models because indicators do not characterize the overall
effectiveness of a control strategy, or whether it's even possible to attain air quality
standards with a given strategy. Indicators are, however, an extremely important step in
helping to decide whether the models are useful for evaluating control strategies. The
theoretical basis for the usefulness of indicators has been derived by several approaches:
First, the extent parameter is based on work by Johnson (1994) that characterizes the
extent to which NO* has been utilized to achieve the maximum potential Os formation.
Secondly, Sillman (1995) and Kleinman et al. (1997) have derived indicators of [Os]
sensitivity based on an analysis of the steady-state radical budgets, i.e., the observation
that the rate of new radical initiation must be balanced by the rate of radical termination.
And finally, Tonnesen and Dennis have derived indicators of P(O*) and [Os] sensitivity
using an analysis of radical propagation efficiency (Tonnesen and Dennis, 1999a,b). The
latter approach is summarized next.
The increase in Os concentration due to photochemistry can be assumed to be
approximately proportional to the integral of P(O*), andP(O*) is approximately
proportional to the rate of OH attack on HC. As discussed above, production of radicals
is equal to the rate of radical initiation multiplied by the OH chain length. Because the
chainlength has a l/(l-Pron) dependence on propagation efficiency, the chain length
increases rapidly as propagation efficiency increases and it is the dominant term that
controls the rate of P(OX). As a result, the sensitivity of P(O^) to precursor emissions will
be determined primarily by the sensitivity of the OH chain length, and [Os] sensitivity
will be determined primarily by the sensitivity of the integral of the OH chainlength.
Figure 2 (bottom) shows the response for the OH chain length (integrated over a
12 hour model simulation) for the same set of trajectory model simulations that^were used
to create the [Os] response surface hi Figure 2 (top). The [O3] ridgeline is also
superimposed on Figure 2 (bottom) to illustrate its correlation with the area of maximum
OH chain length. Figure 2 shows that the [O3] ridgeline corresponds almost exactly to a
ridgeline of maximum chainlength, but there is no particular value of the chainlength that
uniquely identifies the VOC or NO* sensitive condition, for example, a chainlength of 4
could be NO*-limited (below the ridgeline), on the ridgeline, or radical-limited (above the
ridgeline). Thus, the OH chain length is not useful for distinguishing NO* sensitive versus
VOC sensitive conditions.
The individual propagation terms that control the OH chainlength, however, are
useful as indicators of [Os] and P(O*) Figure 3 illustrates major radical propagation and
termination pathways in the photochemical mechanism. The OH propagation efficiency
can be calculated approximately as the product of the fraction of OH radicals that attack
HC (/OH+HC) multiplied by the fraction of subsequent HO2 radicals that react with NO
C/ko2H-No) to recreate an OH radical. These two terms have an opposite dependence on
NO,;. As NO* increases,_/H02+NO increases but JQH+HC decreases. The result is that
propagation efficiency is maximized for some intermediate level of NO* that maximizes
the product of these two terms, thereby creating the Os ridgeline. The values of the
propagation terms can be calculated using the concentrations of the radical species, NO,
-313-
-------
OH Propagatio
HCHO + hv
HOj,+NO ,,
Radical
Initiation
O1D-f H2O
Radical Initiation Feedback due to P(O3>
[ J STOP sign Indicates radical termination
Figure 3. Radical propagation and termination pathways.
NO2, and HC (Tonnesen and Dennis, 1999) and these propagation terms are expected to
uniquely distinguish VOC sensitive and NO* sensitive conditions.
Several approximations have been made in the derivation described above. The
effects of these approximations can be numerically evaluated by using model simulations
with a Gear solver to simulate the robustness of the correlation of the propagation terms
with the ridgeline. Figure 4 illustrates the response surfaces for the propagation terms
where the [Os] ridgeline is again superimposed on each plot. Figure 4 (top) shows that a
nearly constant/Ho2+NO = 92% does uniquely identify MVlimited and radical-limited
conditions, and its value increases from about 72% at high YOG/NO* to 100% at low
VOC/NQt conditions. Similarly, Figure 4 (bottom) shows a strong correlation of/bn+HC
with the ridgeline.
-314-
-------
Q
2QQ
600 600
irtlftil ¥QC
MA to 7 PM a¥@,
teo j
tao -3
"S.
JrjWalVOC
Figure 4. Response surface for propagation pathway fractions, (top) Fraction of HC>2
reacting with NO to recreate OH. (bottom) Fraction of OH attacking HC. The bold
line represents the [O3] ridgeline.
-315-
-------
A number of indicators can be derived from this analysis for P(OX) sensitivity
(Tonnesen and Dennis, 1999a) and for [Os] sensitivity (Tonnesen and Dennis, 1999b)
including some of those previously proposed by Sillman (1995) for [Os] sensitivity. Each
of these indicators were evaluated in simulations of the Regional Acid Deposition Model
(Tonnesen and Dennis, 1999a,b). It was found that the indicators performed best for
sunny conditions with high Os levels. They were less reliable, however, for conditions
where clouds reduced the photorates or where deposition and heterogeneous chemistry
affected the concentrations of Os and radical termination products.
Although uncertainty remains in the robustness of indicators, there is nonetheless a
strong theoretical basis for their usefulness. It is potentially the most powerful method to
date for evaluating the accuracy of model simulated sensitivity of O3 to changes in
precursors. Due to uncertainties in the chemistry and other atmospheric processes, air
quality models are inadequate for characterizing indicators. Validation of the indicator
concept and determination of the particular values that correlate with the ridgeline must
be experimentally validated under controlled conditions where precursor emissions can
be perturbed and the Os response measured for a wide range of conditions. Such
experiments could be performed either in chamber experiment or in field studies using
captured ambient air.
In a chamber study, a large number of simulations could be performed to map out
the response surfaces for [Os], P(OX) and each of the indicators, as illustrated in Figure 5.
160
I
VS
Figure 5. Illustrates the range of conditions needed in a chamber experiments to map
out the response surface for Os and the indicators.
-316-
-------
NO>
sola Itea - ties* entt* dashed 11n«n -
Figure 6. Model simulations of O3 response to NO2 and HCHO injections in a
simulated "captive air" experiment.
Alternatively, a more limited experiment could be performed in a field study using three
portable chambers to capture ambient air. One chamber would be used as a null case, and
NO* or HCHO (as radical source) would be injected into the remaining two chambers
(e.g., Kelly, 1985). The system response would be determined by comparing the
changes in Os and P(O^) in each of the bags to the null case. Figure 6 illustrates the
results of a mode simulation of such an experiment. If this an experiment were "piggy-
backed" on a supersite in a field study, measurements of many of the indicators could be
obtained, and their correlation with the Os response could then be evaluated.
-317-
-------
Conclusions
There is great interest and considerable potential in using process diagnostics and
photochemical indicators to improve our confidence in the accuracy and usefulness of air
quality model simulations. Large uncertainties remain, however, in the interpretation of
the ambient data for elucidating atmospheric processes and in the robustness of these
methods. Evaluation of these methods in the carefully controlled conditions of
environmental chambers could provide a significant improvement in our ability to
evaluate air quality models.
References
Arnold, J.R.; Dennis, R.L.; and Tonnesen, G.S. (1998) Advanced techniques for
evaluating Eulerian air quality models: background and methodology. In: Preprints of the
10th Joint Conference on the Applications of Air Pollution Meteorology with the Air &
Waste Management Association, January 11-16,1998, Phoenix, Arizona. American
Meteorological Society, Boston, Massachusetts, paper no. 1.1, pp. 1-5.
Bailey, E.M.; Copeland, C.H.; and Simonaitis, R. (1996) Smog chamber studies at low
VOC and NOX concentrations. Report on Interagency Agreement DW64936024 to
EPA/NREL, Research Triangle Park, NC.
Blanchard, C.L.; Lurmann, F.W.; Roth, P.M.; Jeffries, H.E.; and M. Korc (1999). The use
of ambient data to corroborate analyses of ozone control strategies. Atmospheric
Environment (in press).
Cantrell, C.A.; Shetter, R.E.; Calvert, J.G.; Eisele, F.L.; Williams, E.; Baumann, K.;
Brune, W.H.; Stevens, P.S.; and Mather, J.H. (1997) Peroxy radicals from
photostationary state deviations and steady state calculations during the Tropospheric OH
Photochemistry Experiment at Idaho Hill, Colorado. J. Geophys. Res. 102:6369-6378.
Carter, W.P.L.; Atkinson, R.; Winer, A.M.; and Pitts, J.N., Jr. (1982) Experimental
investigation of chamber-dependent radical sources. Int. J. Chem. Kinet. 14:1071.
Chang, T. Y., D. P. Chock, B. I. Nance, and S. L. Winkler (1997) A photochemical extent
parameter to aid ozone air quality management, Atmos. Environ., 31, 2787-2794.
Crosley, D. R.( 1997) The 1993 tropospheric OH photochemistry experiment: A
summary and perspective, J. Geophys. Res., 102, 6495-6510.
Dentener, F. J., and P. J. Crutzen (1993) Reaction of N2O5 on tropospheric aerosols:
Impact on global distributions of NO*, O3, and OH, J. Geophys. Res., 98, 7149-7163.
-318-
-------
Donahue, N. M., Dubey, M. K., Mohrschladt, R., Demerjian, K. L., and Anderson, J. G.
(1997) High-pressure flow study of the reactions OH+NO* -> HONO*: Errors in the
falloff region./. Geophys. Res. 102, 6159-6168.
Gao, D., Stockwell, W. R., and Milford, J. B, (1995) First-order sensitivity analysis for a
regional-scale gas-phase chemical mechanism. J. Geophys. Res. 100, 23153-23166.
Gao, D., Stockwell, W. R., and Milford, J. B, (1996) Global uncertainty analysis of a
regional-scale gas-phase chemical mechanism. J. Geophys. Res. 101,9107-9119.
Imre, D.G.; Daum, P.H.; Kleinman, L.; Lee, Y-N.; Lee, J. H.; Nunnermacker, L.J.;
Springston, S.R.; Newman, L.; Weinstein-Lloyd, J.; and Sillman, S. Characterization of
the Nashville urban plume on July 3 and July 18, 1995. Part II. Processes, efficiencies,
and VOC and NO* limitation. J. Geophys. Res., in press.
Jacob, D., NARSTO Critical Review: Heterogeneous chemistry and tropospheric ozone,
Atmos. Environ, 1999 (in press)
Jeffries, H. E. and G. S. Tonnesen (1994) A comparison of two photochemical reaction
mechanisms using mass balance and process analysis, Atmos. Environ., 28, 2991-3003.
Jeffries, H.E.; Sexton, K.G.; Arnold, J.R.; Bai, Y.; Li, J.L.; and Grouse, R. (1990) A
chamber and modeling study to assess the photochemistry of formaldehyde. Report on
EPA Cooperative Agreement CR-813964, Atmospheric Research and Exposure
Assessment Laboratory, EPA, Research Triangle Park, NC.
Johnson, G. M. (1984) A simple model for predicting the ozone concentration of ambient
air, Proceedings of the 8th International Clean Air Conference, Melbourne, Australia, pp.
715-731.
Kelly, N. A. (1985) Ozone/precursor relationship in the Detroit metropolitan area derived
from captive-air irradiations and an empirical photochemical model, JAPCA, 35, 27-34.
Kleffmann, J., K. H. Becker, and P. Wiesen (1998) Heterogeneous NO2 conversion
processes on acid surfaces: Possible atmospheric implications, Atmos. Environment, 32,
2721-2729.
Kleinman L. I., P. Daum, J. H. Lee, Y.-N. Lee, L. Nunnermacker, S. Springston, J.
Weinstein-Lloyd, L. Newman, and S. Sillman (1997) Dependence of ozone production
on NO and hydrocarbons in the troposphere, Geophys. Res. Lett., 24, 2299-3202.
Milford, J., D. Gao, S. Sillman, P. Blossey, and A. G. Russell (1994) Total reactive
nitrogen (NOj,) as an indicator for the sensitivity of ozone to NO* and hydrocarbons, J.
Geophys. Res., 99, 3533-3542.
-319-
-------
Sillman, S. (1995) The use of NOY, H2O2, and HNO3 as indicators for
ozone-NQr-hydrocarbon sensitivity in urban locations, J. Geophys. Res., 100,
14175-14188,.
Sillman, S. (1998) NARSTO Critical Review: The method of photochemical indicators
as a basis for analyzing Os-NOx-ROG sensitivity. Submitted to Atmos. Environ.
Sillman, S.; Dongyang H.; Pippin, M.R.; Daum, P.H.; Lee, J.H.; Kleinman, L.I.; and
Weinstein-Lloyd, J. (1997a) Model correlations for ozone, reactive nitrogen and
peroxides for Nashville hi comparison with measurements: Implications for C>3-NOX-
hydrocarbon chemistry. Submitted to J. Geophys. Res.
Sillman, S.; He, D.; Cardelino, C.; and Imhoff, R.E. (1997b) The use of photochemical
indicators to evaluate ozone-NOx-hydrocarbon sensitivity: Case studies from Atlanta,
New York and Los Angeles, J. Air Waste Management Assoc. 47:1030:1040.
Sillman, S.,' D. He, M. R. Pippin, P. H. Daum, D. G. Imre, L. I. Kleinman, J. H. Lee, and
J. Weinstein-Lloyd (1998) Model correlations for ozone, reactive nitrogen and peroxides
for Nashville in comparison with measurements: Implications for Os-NCy-hydrocarbon
chemistry, J. Geophys. Res. 103, 22,629-22,644.
Simonaitis, R., and Bailey, E.M. (1995) Smog chamber studies at low VOC and NOX
concentrations: Phase I. Report on Interagency Agreement DW64936024 to EPA/NREL,
Research Triangle Park, NC.
Stevens, P.S.; Mather, J.H.; Brune, W.H.; Eisele, F.; Tanner, D.; Jefferson, A.; Cantrell,
C.; Shetter, R.; Sewall, S.; Fried, A.; Henry, B.; Williams, E.; Bagman, K.; Goldan, P.;
and Raster, W. (1997) HO2/OH and RO2/HO2 ratios during the Tropospheric OH
Photochemistry Experiment: Measurement and theory. J. Geophys. Res. 102:6379-6391.
Tonnesen, G. S. and R. L. Dennis, Analysis of radical propagation efficiency to assess
ozone sensitivity to hydrocarbons and NO*. Part 1: Local indicators of instantaneous
odd oxygen production sensitivity, J. Geophys. Res., in press.
Tonnesen, G. S., and R. L. Dennis, Analysis of radical propagation efficiency to assess
ozone sensitivity to hydrocarbons and NO*. Part 2: Long-lived species as indicators of
ozone concentration sensitivity, J. Geophys. Res., in press.
-320-
-------
Ozone Formation in Coastal Urban Atmospheres:
The Role of Anthropogenic Sources of Chlorine
Paul L. Tanaka, Sarah Oldfield, Charles B. Mullins, David T. Allen
In this communication, we present experimental results from our environmental chamber
studies to suggest that anthropogenic sources of molecular chlorine (Ck), a photolytic source of
Cl- , may contribute significantly to ozone formation in some urban environments. Many studies
have focused on the sources and chemistry of halogen atoms in pristine marine environments1"3,
often attributing a net consumption of ground-level ozone to reactions with halogen atoms,
primarily bromine and to a lesser extent, chlorine4"7. However, little attention has been directed
at determining the effect of Cl- on ozone formation in urban environments, where oxides of
nitrogen (NOX) and volatile organic compounds (VOCs) are ubiquitous and there exist many
anthropogenic sources of Ck. Large anthropogenic sources of C12 emissions include chemical
production facilities, water treatment plants, and paper production operations. Chlorine atoms,
formed by the photolysis of Cfe, react with alkanes at up to two orders of magnitude faster than
do hydroxyl radicals and can promote the formation of Os in the presence of VOCs and NOX2'3.
To understand how chlorine promotes ozone formation in a mixture of VOCs and NOX
representative of the coastal urban troposphere, we performed experiments in outdoor Teflon
environmental chambers. The chambers were approximately 2 m3 in volume with internal
volume to surface ratios of approximately 0.13 m when fully inflated. The chambers were
conditioned8'9 and subsequently prepared by flushing with zero air overnight. A commercially
available mixture of 56 hydrocarbons (Matheson "Enviro-Mat" Ozone Precursor), as well as
individual hydrocarbon reactants were used in the experiments. NOX (Praxair- NO/NO2 at a ratio
of 200:1) was also injected into the chamber while the chamber was covered with an opaque
tarp. After these reactants were allowed sufficient time to mix, C12 (Air Products and Chemicals)
was injected, the tarp removed, and gas sampling begun. Gas withdrawn from the chamber was
delivered to O3 (Dasibi 1008AH or 1003PC) and NOX (Monitor Laboratories 9841 or Columbia
Scientific Industries 1600) analyzers. These continuous measurements were collected as 5-
minute averages by Climatronics Corporation IMP 850 microloggers. Air samples for
hydrocarbon analysis were collected in 6-liter stainless steel Summa® canisters and analyzed by a
HP 5 890A gas chromatograph (GC) equipped with a flame ionization detector (FID) and/or a HP
6890 GC with a HP 5972 mass selective detector and Entech 7000 preconcentrator/cryofocuser.
A first set of chamber experiments was directed at showing whether the addition of C12 to
a mixture of VOCs and NOX representative of conditions found in Houston, TX would promote
the formation of 03. Initial reactants included a mixture of 56 hydrocarbons with a total
hydrocarbon concentration of approximately 1 part per million carbon (ppmc). Sufficient NOX
was injected to yield an initial VOC/NOX ratio of 10 ppbc/ppbv. Initial C12 concentrations were
between 0 and 47 ppbv. Each run was conducted under conditions of similar solar flux and
temperature. The data summarized in Table 1 show that the peak Os concentration ([O3]peak)
increases by up to a factor of six with the addition of C\2, and the time required to reach
0.63*[O3]peak was reduced by up to a factor of 3.5. In addition, experiments that included
injections of C12 showed significant losses of alkanes, specifically substituted alkanes. The loss
-321-
-------
of alkanes increased from less than 10 percent in experiments without added C12 to greater than
60 percent for hexane, 2,3-dimethylpentane, and octane in experiments with added C12. Losses
of alkenes and aromatic compounds were similar between runs with and without added C12
(Table 2). If the chemistry in these experiments were dominated by the reactions of hydroxyl
radical, loss of alkenes and substituted aromatics would be expected, but the C6+ alkanes would
be relatively unreactive10"15 and contribute little to the formation of ozone16. What was observed,
however, was a depletion of alkanes and an increase in ozone concentrations. This is consistent
with chlorine radical initiated oxidation chemistry.
A second set of chamber experiments were run to identify which classes of hydrocarbons
were most significantly affecting O3 production associated with the addition of C12. During these
experiments n-pentane (EM Omnisolv), propene (Matheson-C.P. grade), and benzene (EM-
reagent grade) were used to represent each of the major classes (alkanes, alkenes, and aromatics)
of hydrocarbons found in the urban troposphere.
In runs with approximately equal concentrations (in ppbc) of benzene, pentane, and
propene, no significant changes in peak O3 concentration or rate of formation were observed
with the addition of C12. However, for runs containing only pentane in air, a significant increase
in peak O3 concentration was observed and the activation time for significant ozone production
decreased by a factor of 2.
To a first approximation, these results can be explained by comparing the rates of
reaction of OH- and Cl- with each of the reactant hydrocarbon species (Table 3). The reaction
rate constant of OH- with propene is approximately an order of magnitude greater than the
reaction rate constants of OH- with pentane or benzene. However, the reaction of Cl- with
pentane and propene proceed with similar rate constants at 298K. Benzene reacts even more
slowly with Cl- than OH- at this temperature. Because anthropogenic O3 formation in the
presence of VOCs and NOX is initiated by reactions of free radicals with hydrocarbons and
subsequent conversion of NO to NO2, Cl- appears to have a pronounced effect on O3 production
and peak concentrations when there exists significant Cl- activation of alkanes relative to alkene
oxidation by OH- andCl- .
Our initial findings are presented as evidence to suggest that C12 can significantly
contribute to ozone formation in urban areas. By adding C12 to a mixture of VOCs and NOX in
air typical of urban atmospheres, we show that the rate of formation and peak concentrations of
ozone significantly increase relative to control runs without added molecular chlorine. In
addition to an increase in O3 formation, Cl- dominated chemistry is evidenced by the significant
loss of alkanes, typically not observed for chemistry dominated by OH- . Further environmental
chamber studies will be performed to determine whether these phenomena occur at various
VOC/NOx ratios and with different VOC compositions.
Acknowledgements
The authors wish to thank the Texas Natural Resource Conservation Commission for their
generous support of our investigative work. The authors would also like to acknowledge Dr.
-322-
-------
Gary Vliet and Ian Bird of the Solar Energy Laboratory at the University of Texas at Austin for
providing solar flux data.
References
1. C. W. Spicer et al., Nature, 394, 353 (1998).
2. K. W. Oum, M. J. Lakin, D. O. DeHaan, T. Brauers, B. J. Finlayson-Pitts, Science, 279, 74
(1998).
3. B. J. Finlayson-Pitts, Res. Chem. Intermed. 19,235 (1993).
4. B. J. Finlayson-Pitts and J. Pitts Jr., Science, 276, 1045 (1997).
5. B. Ramacher, J. Rudolph, R. Koppman, /. Geophys. Res., 104, 3633 (1999).
6. M. O. Andreae and P. J. Crutzen, Science, 276, 1052 (1997).
7. R. Vogt, P. J. Crutzen, R. Sander, Nature, 383, 327 (1996).
8. B. J. Finlayson-Pitts and J. N. Pitts Jr., Atmospheric Chemistry: Fundamentals and
Experimental Techniques (Wiley, New York, 1986).
9. D. Grosjean, Env. Sci. Tech., 19, 1059 (1985).
10. R. Atkinson et al., J. Phys. Chem. Ref. Data, 28, 191 (1999).
11. R. Atkinson, J. Phys. Chem. Ref. Data, 26, 215 (1997).
12. R. Atkinson, Gas-phase Tropospheric Chemistry of Organic Compounds, (Monogr. 2, J.
Phys. Chem. Ref. Data, American Chemical Society, Washington B.C., 1994) pp.1-216.
13. P. Ariya, thesis, York University (1996).
14. R. Atkinson and S. M. Aschmann, Int. J. Chem. Kin., 17, 33 (1985).
15. W. B. DeMore et al., JPL Publ. No. 97-4 (1997).
16. W. P. L. Carter, J. A. Pierce, I. L. Malkina, Atm. Env., 29, 2499 (1995).
-323-
-------
«
1
OJ
t«
Q
1
TH
,2
42
^
20
s ^*
.a *
H2
I
II
o
-------
Table 2 Fractional Losses of VOCs: Experiments with O3 Precursor
Compound
Fraction Lost Fraction Lost Fraction Lost Fraction Lost
[Cl2]o=0ppbv [Cl2]0=14ppbv [C12]0= 20 ppbv [C12]0= 47 ppbv
Alkanes
2,3-Dimethylpentane
n-Hexane
n-Octane
n-Nonane
Alkenes
1-Pentene
2-Methyl,
1-3-Butadiene
Aromatics
Benzene
Toluene
0.13
0.02
0.06
0.01
0.31
0.88
0.17
0.18
0.30
0.39
0.32
0.52
0.24
1.00
0.01
0.10
0.62
0.37
0.55
0.43
0.32
1.00
0.01
0.01
0.63
0.60
0.76
0.43
0.32
1.00
0.11
0.21
-325-
-------
TableS Rate Constants
Compound
(10'14 cm3 molecule'1 s"1) (10'14 cm3 molecule' s')
Alkanes
Methane(10)
Ethane(10)
Propane(10)
n-Butane
n-Pentane(11)
n-hexane^11^
Alkenes
Ethene
Propene
Aromatics
Benzene
Toluene
0.64
25
110
244(11)
400
545
900(io)
3,000(10)
md2)
5,960(12)
10
5,900
14,000
22,000(10)
28,000
34,000
10,700°l)
28,000(11)
0.9(13)
5,890(14)
-326-
-------
Session V
Reactivity Studies
Session Chair
Dick Derwent
-------
U.S. EPA ModeIs-3/CMAQ - Status and Applications
Ken Schere1
Atmospheric Modeling Division
U.S. Environmental Protection Agency
Research Triangle Park, NC
Extended Abstract:
An advanced third-generation air quality modeling system has been developed by the
Atmospheric Modeling Division of the U.S. EPA. The air quality simulation model at the heart
of the system is known as the Community Multiscale Air Quality (CMAQ) Model. It is
comprehensive in scope and allows for the simulation of ozone and photochemical oxidants, acid
deposition, and fine and coarse particles at spatial scales ranging from urban to regional. The
model is contained within a computational framework, Models-3 (for 3rd generation), that enables
users to interact with the modeling system through a high-level graphical user interface and also
facilitates data transmission among the components of the system and provides for analysis,
graphics, and visualization capabilities for model simulation results. The modeling system is
available from the U.S. EPA (see web site: www.epa.gov/asmdnerl/models3/), and is currently
being evaluated for photochemical oxidants and fine particles using field study databases from
the eastern United States from 1990 and 1995. The CMAQ is also being extended to include the
modeling of selected air toxics, including atmospheric mercury and atrazine (a pesticide).
Models-3 is a sophisticated computational framework for air quality modeling systems. It
has been designed and programmed using object-oriented principles (in C++ language). At the
highest level, Models-3 presents a graphical user interface (GUT) to the model user. Components
presented include the Program and Dataset Managers, for registering programs and datasets;
Science Manager and Model Builder for defining science process components and building an
executable model, Study Planner for defining and running the required preprocessors and models,
Strategy manager for defining emission control strategies and processing emissions, and Tools
Manager which is the gateway to the analysis and visualization tools. Models-3 assists the user
in setting up new model domains and applications, accessing and tracking data files, and
controlling the flow of data and model runs. Component models may be run on the same
computer platform as Models-3, or on remote computers where communications links are
maintained to the Models-3 server. A configuration currently used at the U.S. EPA has the
Models-3 server maintained on a SUN workstation with CMAQ model runs initiated on remote
CRAY supercomputers linked to the workstation through fast telecommunication lines.
The most recent release of the Models-3 framework is compatible with SUN workstations
running the Solaris2.6 operating system. Another version of the framework, capable of being
1 On assignment from the Air Resources Laboratory, National Oceanic and Atmospheric Administration,
U.S. Department of Commerce.
-328-
-------
operated on a Windows-NT computer, will become available for beta-testing in late 1999. The
fully operational version of the Windows-NT Models-3 framework is expected by summer 2000.
In addition to these two platforms, a third port of Models-3 is currently underway to a SGI
workstation (beta-version available in early 2000). The Tools Manager currently contains the
EBM-DX Explorer graphics and analysis software, and the VisSD and PAVE visualization
packages. Users and Tutorial Manuals are provided with the Models-3 package.
The CMAQ model is a set of Fortran language science codes that include the CMAQ
chemical transport (air quality) model (CCTM) and its upstream driver models and processors.
These codes are generally portable to any computer platform with available Fortran compilers.
(Fortran-77 is the current coding standard; CMAQ-2000 release will begin to use Fortran-90
constructs.) A complete set of CMAQ science documentation is available.2 A meteorological
model is required to provide hourly gridded meteorological fields to the emissions and chemical
transport model. Currently the NCAR/Penn State MM5 model is used for that purpose; other
meteorological models will be adapted in the future. The meteorological model is run outside of
the Models-3 framework and the data that it generates are then brought into Models-3 for use by
the downstream models. The MM5 is a freely-available community mesoscale model supported
and released by the National Center for Atmospheric Research in Boulder, Colorado. The
version that we have adapted for CMAQ use is MM5-v2.10, with a few of our own CMAQ-
specific modifications. It is a non-hydrostatic model, and we are using it in one-way nested
mode to feed meteorological data to the one-way nested CCTM. Four-dimensional data
assimilation is a key feature used to correct errors that typically accumulate over time during the
meteorological model run.
The Meteorology-Chemistry Interface Processor (MCIP) assimilates the data generated by
the meteorological model and performs a dynamically consistent merging of layers if the number
of CCTM layers is smaller than that of MM5. It also computes dry deposition parameters based
on an algorithm of M. Wesely for the CCTM. Presently, the MCIP also rediagnoses certain
planetary boundary layer parameters, including boundary layer depth, although in the 2000
version of the model these parameters will be passed through directly from the MM5 to the
CCTM. The fields produced from the MCIP are then provided to the emissions model and to the
CCTM.
The Models-3 Emissions Processor and Projection System (MEPPS) processes base-year
emissions, both anthropogenic and biogenic, in chemically speciated, hourly allocated, and
gridded form for the chemical transport model. The projection portion of the system estimates
emissions for a future year, based on the effects of growth and control programs. The MEPPS is
an exception to the other science codes in that it is based mainly on SAS programming code and
not Fortran, as well as making use of GIS techniques for allocating the source emissions to model
Science Algorithms of the EPA Models-3 Community Multiscale Air Quality (CMAQ) Modeling System.
EPA-600/R-99/030, March 1999, U.S. EPA, Research Triangle Park (also available from Models-3 web site:
www.epa.gov/asmdnerl/models3/)
-329-
-------
grid cells. Chemically speciated emissions profiles are cataloged by Source Classification Codes
(SCC) representing most industrial, transportation, and manufacturing activities. At present
MEPPS supports the organic compound classes from two chemical mechanisms, RADM-2 and
CB-4. As other chemical mechanisms are added to the CCTM, the emissions system will also be
extended to include the proper emissions categories corresponding to the specific chemical
mechanism. As part of MEPPS, mobile source emission factors are calculated using the „
MOBILESa model. The PARTS model calculates mobile source fine particle emission factors.
The BEIS2 is used to estimate hourly biogenic emissions of VOCs and NO. Emissions estimates
are made for the appropriate CCTM domain sizes and grid resolutions. The MEPPS-2000
version may contain the MOBILE6 and/or BEIS3 emissions models if they are operationally
ready hi tune. The Emissions-Chemistry Interface Processor (ECIP) takes the emissions outputs
from MEPPS and transforms them into CCTM-ready emissions fields, as well as calculating the
plume rise from major point source stacks.
Work is on-going to replace portions of the MEPPS emissions processor with the Fortran-
based Sparse Matrix Operator Kernel Emissions (SMOKE) processor. The SMOKE processor is
not only more advantageous because of the Fortran based portable code, but also because it is an
order of magnitude more efficient to run. The initial (Beta) implementation will occur in 1999,
with the first operational version within Models-3 completed in late 2000, subject to funding
constraints. The operational version will have associated full user documentation and quality
control modules.
As part of the SMOKE implementation in Models-3, enhancements are being made to
handle reactivity controls and projections. Reactivity control packets, by SCC source category or
by specific source, will allow for changing the VOC profile from an emissions process, including
substituting a compound of lower reactivity for a higher reactivity compound. The
implementation will also allow for a phase-in period for market penetration of substituted
compounds when making future-year emissions projections. A new reactivity control matrix is
added to the SMOKE processing along with the existing base emissions, speciation, and gridding
matrices. SMOKE is also being adapted to increase flexibility to add new pollutants as needed to
the emissions processing.
The CCTM is the principal air quality simulation portion of the modeling system. It
assimilates the meteorological and emissions data processed as described above, as well as
appropriate sets of initial and boundary conditions, and estimates ambient concentrations of
modeled pollutant species. The CCTM currently models relevant chemical species participating
in ambient photochemical oxidant and aerosol chemistry, and acid deposition. The model has
been constructed in a modular and generalized manner to facilitate its pairing with a variety of
meteorological models and its inclusion of various process modules for representing a specific
numerical solution or science process. The 2000 version will also add an initial air toxics
capability for mercury and atrazine (a pesticide).
The gas-phase chemical kinetic mechanisms now supported within the CCTM are the
-330-
-------
RADM-2 and CB-4. The SAPRC-99 mechanism is currently being added. Bill Stockwell (DRT)
has expressed an interest in working with the Models-3/CMAQ system and possibly adding the
RACM mechanism, a successor to RADM-2. In addition, a new approach to kinetic
mechanisms, the "morphecule" mechanism (developed by H. Jeffries and colleagues) is being
tested now in box models and will be ready for initial testing in CMAQ during 2001-2002. The
mechanism reader, a feature of the Models-3 framework, facilitates the inclusion of mechanism
changes or new mechanisms in the CMAQ. There are two numerical chemistry solvers available
in the CCTM. The first is a general form quasi-steady state approximation (QSSA). It is a
predictor-corrector method that makes no steady-state assumptions, and thus can be applied to
most chemical mechanisms. The other solver choice is the sparse-matrix vectorized Gear
solution (SMVGEAR) which is only applicable to vector computer platforms. This solver is
considered as the most accurate of available numerical chemical solutions. We will also be
working over the coming year to implement a version of the Hertel numerical solver, a very
efficient solver, but whose implementation will be specific to particular chemical mechanisms.
Photolysis rates are selected during model simulation from a multi-dimensional look-up table,
and they are attenuated based on cloud amount and depth. An optional process analysis and
integrated reaction rate module calculates the incremental contribution to process rates from all
major modeled processes, and performs a diagnostic analysis of all major chemical pathways.
Aqueous chemistry is performed as part of the cloud package in the CCTM, and has been
adapted from the earlier Regional Acid Deposition Model.
The chemistry and physics of fine and coarse particles are also simulated within the
CCTM. A modal approach is used to describe the size distribution of particles, with three
modes: the Aitken mode (<0.1 |j.m diameter particles), the accumulation mode (0.1-2.5 um
diameter particles), and the coarse mode (>2.5 \im diameter particles). For the fine particles,
those in the two size modes less than 2.5 um, there are 8 chemical species categories for each
mode: sulfate, nitrate, ammonium, primary anthropogenic organics, secondary anthropogenic
organics, biogenic organics, elemental carbon, and miscellaneous primary (mostly crustal
materials). The fine particles also participate in the aqueous chemistry and are dry deposited as
well. The particles in the coarse mode do not participate in the chemistry, but are transported,
diffused, and dry deposited.
The current U.S. EPA applications of the Models-3/CMAQ modeling system are to
nested domains in the eastern U.S., centered on the Northeast including the Washington-New
York corridor, as well as on the Nashville, TN area. We are modeling periods during the
summer of 1995 at which time several major field campaigns for photo-oxidants took place in
these areas. We will be diagnostically evaluating the model for its ability to simulate the key
constituents of photo-oxidant and fine particle chemistry. We are also modeling the first six
months of 1990 in which we will be assessing the CMAQ's ability to characterize acid
deposition and secondary particles against field data from that period from the eastern U.S.
We are seeking to build a community of air quality modelers around the Models-
3/CMAQ platform who are interested hi applying, evaluating, and extending the CMAQ and its
-331-
-------
associated models, and in helping to support and maintain the Models-3 computer framework to
facilitate use of the CMAQ and other models. Over the next several years we will be working
with the academic, policy, and other communities to implement the modeling system and begin
implementing the community model concept for air quality. The first of a series of annual
Models-3 Workshops is being planned for the spring of 2000.
-332-
-------
Analysis of Chamber-Derived Incremental Reactivity Estimates for
N-Butyl Acetate and 2-Butoxy Ethanol
Lihua Wang and Jana B. Milford
Department of Mechanical Engineering ,
University of Colorado, Boulder CO
William P.L. Carter
CE-CERT
University of California, Riverside. CA
Abstract
Incremental reactivity estimates for many high molecular weight organic compounds
used in consumer products are viewed as uncertain because of limited data on their reaction
mechanisms and products. This study performs a systematic uncertainty analysis of the
calculated incremental reactivities of two such compounds: n-butyl acetate and 2-butoxy ethanol.
2-butoxy ethanol is a relatively well-studied compound for which incremental reactivity -
experiments have been performed and product data are available for most reaction pathways. In
contrast, there are incremental reactivity experiments but essentially no product data for n-butyl
acetate. The uncertainty analysis accounts for uncertainties in the environmental chamber
experiments used to estimate key parameters of the 2-butoxy ethanol and n-butyl acetate
mechanisms, as well as in the parameters of the base SAPRC-97 chemical mechanism used to
calculate their incremental reactivities. Uncertainties in the 2-butoxy ethanol reactivity estimates
are lower than those estimated previously for many other volatile organic compounds (VOCs).
In contrast, the uncertainties in the n-butyl acetate reactivity estimates are at the upper end of the
range of uncertainties estimated for other VOCs. The chamber-derived parameters of the n-butyl
acetate and 2-butoxy ethanol mechanisms contribute at most about 7% of the uncertainty hi their
incremental reactivity estimates.
1. Introduction
Reactivity estimates for many high molecular weight hydrocarbons and oxygenated
organic compounds from consumer products, coatings and solvents are viewed as highly
uncertain because of gaps in understanding their oxidation mechanisms. Product studies that
would allow the mechanisms to be determined are often limited or nonexistent. For these
compounds, yields and reaction rates of radicals and stable intermediates are typically estimated
based on structural analogy with lighter compounds for which the chemistry has been tested. In
addition, incremental reactivities have been measured in environmental chambers for some of
these compounds, with mechanistic parameters sometimes estimated by fitting the chamber data.
The objective of this study is to estimate uncertainties for two representative VOCs that
are used in consumer products, for which a combination of explicit kinetic and product data,
structural analogy, and chamber-derived parameter values have been used to estimate
incremental reactivities. The compounds selected are the solvents 2-butoxy ethanol
CH2-CH2-O-CH2-CH2-OH) and n-butyl acetate (CH3-CH2-CH2-CH2-O-CO-CH3).
-333-
-------
2-Butoxy ethanol provides an example of a well-studied compound for which there is
product data for most of the reaction routes, as well as chamber reactivity data. Most of the
pathways of the 2-butoxy ethanol reaction with OH are well characterized. In the SAPRC-97
mechanism, the only parameter that is adjusted based on chamber data is the nitrate yield. N-
butyl acetate is an example of a compound for which there is chamber reactivity data but
essentially no product data. In this case, multiple parameters in the estimated mechanism are
adjusted to fit the environmental chamber data.
To estimate uncertainties in incremental reactivities for these compounds, the study
proceeds in three steps. First, stochastic programming is used to estimate uncertainties in
chamber radical source parameters, which are critical inputs to simulations of chamber
incremental reactivity experiments. Then, stochastic programming is used to estimate
uncertainties in key reaction pathway and organic nitrate yield parameters of n-butyl acetate and
2-butoxy ethanol. Chamber experiments conducted by Carter et al. (1998) in the Dividable
Teflon Chamber (DTC) at the University of California at Riverside (UCR) are used to estimate
the mechanistic parameters. Finally, the estimates of uncertainty in the chamber-derived
parameters, together with estimates of uncertainty in other SAPRC-97 parameters, are
propagated through incremental reactivity calculations using Monte Carlo analysis with Latin
hypercube sampling.
2. Methods
2.1 SAPRC-97 Mechanism and Chamber-derived Parameters
The chemical mechanism employed in this study is the SAPRC-97 mechanism (Carter et
al., 1997). SAPRC-97 is used for consistency with past work on uncertainties in chamber-
derived parameters for aromatic compounds (Wang et al., 1999). Comparisons between
SAPRC-97 and the latest available update, SAPRC-98, are given below for parameters and
incremental reactivities of 2-butoxy ethanol and n-butyl acetate.
2.1.1 2-Butoxy Ethanol Mechanism
The SAPRC-97 mechanism for 2-buxtoxy ethanol uses an OH reaction rate constant of
2.57 x 10"11 molecules cm"3 s"1 (Dagaut et al., 1988; Stemmler et al., 1997; Aschmann and
Atkinson, 1998). An uncertainty of 25% is assigned for this rate constant (Carter, 1998). The
mechanism for the subsequent reactions is generated using an automated procedure,
incorporating estimates of branching ratios for attack of OH radicals at different positions, and
branching ratios for competing reactions of the radicals that are formed (Carter, 1999). The
major reactions derived for 2-butoxy ethanol are summarized in Figure 1. In terms of model
species used in the SAPRC-97 mechanism, the overall process (with nominal parameter values)
is represented as:
OH + BUO-ETOH -» 0.887 RO2-R. + 0.113 RO2-N. + 0.136 R2O2. +
0.55 HCHO + 0.015 CCHO + 0.32 RCHO +
0.503 MEK + 0.26 PROD2 +1.136 RO2.
-334-
-------
The reader is referred to Carter (1990) for definitions of the product species. Note that PROD2 is
not in the SAPRC-97 mechanism, but was added here for the purpose of representing the
reactivities of solvent species.
As indicated in Figure 1, the following mechanistic parameters are considered in our
analysis: pi is the fraction of the initial reaction forming the CH3-CH2-CH2-CH2-O-CH[.]-CH2-
OH radical, whose subsequent reactions result primarily in formation of n-butyl formate; p2, the
fraction of the initial reaction forming the CH3-CH2-CH2-CH[.]-O-CH2-CH2-OH radical and the
formation of 2-hydroxy formate and propanal; p3, the fraction of the initial reaction forming the
CH3-CH2-CH2-CH2-O-CH2-CH[.3-OH radical, which is expected to react to form
butoxyacetaldehyde; and pN, the fraction of the initially formed peroxy radicals that form
organic nitrates when they react with NO.
The value of pi is derived based on observed 57 ± 5% yields of n-butyl formate from
Tuazon et al. (1998). The value of p2 is derived based on observed 22 ± 2% yields of 2-hydroxy
formate and 21 ± 2% yields of propanal from Tuazon et al. (1998). The value of p3 is derived
from the estimates for the other pathways. The overall yield of organic nitrates from the initially
formed peroxy radicals (pN) is estimated by fitting incremental reactivity estimates from
environmental chamber experiments. The performance of the fitted 2-butoxy ethanol mechanism
in simulating incremental reactivity experiments is illustrated hi Figure 2. Results are shown for
both the SAPRC-97 and SAPRC-98 base mechanisms, with the SAPRC-97 calculations repeated
employing both the pN value optimized using the SAPRC-98 mechanism, and the reoptimized
value derived for SAPRC-97 hi this work.
-v-12
molecules cm"3 s"1
2.1.2 N-Butyl Acetate Mechanism
The OH reaction rate constant used for n-butyl acetate is 4.20 x 10
(Atkinson et al, 1989). The mechanism for the subsequent reactions is generated using the same
automated estimation methods as employed for 2-butoxy ethanol. The major reactions derived
for n-butyl acetate are summarized in Figure 3. In terms of SAPRC-97 model species, the
overall process is nominally represented as follows.
HO. + BU-ACET -> 0.694 RO2-R. + 0.106 RO2-N. + 0.553 R2O2. + 0.2 C2CO-O2. +
0.019 CO + 0.011 CO2 + 0.011 HCHQ + 0.143 CCHO + 0.199 RCHO +
0.462 MEK + 0.254 PROD2 + 1.353 RO2. + 0.2 RCO3
As indicated hi Figure 3, five major parameters affecting product yields are considered hi
our uncertainty analysis for butyl acetate. Branching ratios of the initial OH reaction step for
CH3-CH2-CH2-CH[.]-O-CO-CH3 (ql) and CH3-CH2-CH[.]-CH2-O-CO-CH3 (q2) are estimated
based on the structure-reactivity estimation methods given by Kwok and Atkinson (1995). As hi
the 2-butoxy ethanol reaction, the overall organic nitrate yield from the initially formed peroxy
radicals (qN) is estimated from fitting incremental reactivity estimates from environmental
chamber experiments. The fraction of the alkoxy radical CH3-CO-O-CH2-CH2-CH[O.]-CH3 that
reacts with O2 (q4) rather than decomposing to CH3-CHO + CH3-CO-O-CH2-CH2 is estimated
using the alkoxy radical rate constant estimates incorporated in the mechanism generation system
-335-
-------
(Carter, 1999). Finally, the fraction of CH3-CH2-CH2-CH[O.]-O-CO-CH3 undergoing ester
rearrangement (q6) instead of isomerizing to CH3-CO-O-CH(OH)-CH2-CH2-CH2 is also adjusted
to fit chamber experimental results.
Figure 4 shows how the n-butyl acetate mechanism performs with the SAPRC-97 and
SAPRC-98 base mechanisms, with values of q6 and qN adjusted to fit the incremental reactivity
experiments. The results using the SAPRC-97 mechanism are shown using best fit parameter
values derived deterministically for SAPRC-98 as well as values derived in this work using
stochastic parameter estimation.
2.2 Chamber Data and Chamber Effects Mechanism
As indicated, the values for some of the uncertain mechanistic parameters in the 2-butoxy
ethanol and n-butyl acetate mechanisms are determined by optimization to fit chamber data. The
chamber data employed consist of a series of incremental reactivity experiments carried out
during 1996 and 1997 in the blacklight irradiated DTC (Carter et al, reports in preparation 1999).
Seven pairs of experiments, for which inputs are summarized in Table 1, were conducted for
both of these compounds. In each case, three of the pairs utilized a three-compound base
mixture including n-hexane, ethene and m-xylene. The other four pairs used an eight-compound
base mixture with n-butane, n-octane, ethene, propene, trans-2-butene, toluene, m-xylene and
formaldehyde. For each experimental pair, the test compound was added to the base mixture in
only one of the two bags of the dividable chamber.
Using experimental data to estimate mechanism parameters requires consideration of the
artifacts in the chamber. For this study, the most critical artifacts are thought to be the chamber-
dependent radical sources. Two radical source parameters, RSI and HONO-F, are considered
here. RSI represents a NO2 independent, continuous light-induced release of radicals from the
chamber walls (Carter, 1996; Carter et al., 1997). This radical source is described by the reaction
hv -> OH with reaction rate RSI x ki, where ki, the NO2 photolysis rate, is a measure of the light
intensity in the experiment. HONO-F represents the fraction of initial NO2 converted to HONO
prior to irradiation. The radical sources are estimated from n-butane-NOx or CO-NO x
experiments conducted in the DTC at about the same time as the 2-butoxy ethanol and n-butyl
acetate runs (Carter, 1996). Twelve n-butane-NOx and two CO-NOX experiments were used to
estimate the radical source parameters for the n-butyl acetate experiments; 18 n-butane-NOx
experiments were used to estimate them for 2-butoxy ethanol.
Table 1. Chamber experiments used to estimate mechanism parameters for 2 -butoxy ethanol and n-
butyl acetate
Char.Exp.11
Run ID
TestVOC
Type0 Date
B
k,
Initial Concentrations
T NO Unc. NO2 Unc.
DTC365A BU-ACET MRS 6/6/96 11 11 0.199 298 0.23 3% 0.08 11%
DTC368B BU-ACET MR3 6/11/96 11 11 0.198 299 0.25 3% 0.08 11%
DTC402B BU-ACET MR3 8/23/96 11 11 0.190 299 0.26 3% 0.10 11%
-336-
-------
DTC403A
DTC410B
DTC406A
DTC411A
DTC491B
DTC498B
DTC505B
DTC493B
DTC502A
DTC497A
DTC506A
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
MRS
MRS
R8
R8
MRS
MR3
MR3
MRS
MRS
R8
R8
8/27/96
9/10/96
8/30/96
9/11/96
5/20/97
5/30/97
6/11/97
5/22/97
6/5/97
5/29/97
6/12/97
11
11
11
11
14
14
14
14
14
14
14
11
11
11
11
15
15
15
15
15
15
15
0.189
0.188
0.189
0.187
0.221
'--0.219
0.217
0.221
0.218
0.219
0.217
299
298
299
297
298
299
298
297
299
299
298
0.23
0.21
0.08
0.06
0.28
0.27
0.28
0.23
0.23
0.09
0.08
3%
3%
4%
5%
3%
3%
3%
3%
3%
4%
4%
0.06
0.06
0.04
0.02
0.10
0.11
0.10
0.07
0.07
0.04
0.04
11%
11%
12%
14%
11%
11%
11%
11%
11%
11%
12%
Run ID
DTC365A
DTC368B
DTC402B
DTC403A
DTC410B
DTC406A
DTC411A
DTC491B
DTC498B
DTC505B
DTC493B
DTC502A
DTC497A
Test VOC
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
Initial
Test
VOC
5.88
6.30
3.79
5.15
7.60
3.69
7.72
1.72
1.15
1.08
1.11
0.53
0.86
Concentrations
Unc.
5%
5%
27%
27%
27%
27%
27%
12%
12%
12%
12%
12%
12%
N-C4
0.35
0.33
0.34
0.32
0.34
0.34
0.36
Unc.
5%
5%
5%
5%
11%
11%
11%
N-C6
0.48
0.49
0.43
0.47
0.46
0.41
Unc.
13%
13%
38%
10%
10%
10%
N-C8
0.09
0.09
0.09
0.09
0.10
0.09
0.10
Unc.
25%
15%
15%
15%
7%
12%
12%
Ethene
0.79
0.80
0.75
0.06
0.05
0.06
0.05
0.87
0.26
0.84
0.06
0.06
0.07
Unc.
5%
5%
5%
5%
5%
5%
5%
12%
12%
12%
12%
12%
12%
-337-
-------
DTC506A BUO-ETOH
0.57
12% 0.35 11%
0.10 12% 0.06
12%
Run ID
DTC365A
DTC368B
DTC402B
DTC403A
DTC410B
DTC406A
DTC411A
DTC491B
DTC498B
DTC505B
DTC493B
DTC502A
DTC497A
DTC506A
TestVOC
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BU-ACET
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
BUO-ETOH
Propene Unc.
0.05 5%
0.05 5%
0.05 5%
0.04 5%
0.05 7%
0.05 7%
0.05 7%
0.05 7%
Initial Concentrations
T-2-Bute Unc. Toluene Unc. M-Xyle
0.05 5%
0.05 5%
0.05 5%
0.04 5%
0.05 9%
0.05 9%
0.05 9%
0.05 9%
0.08
0.08
0.08
0.07
0.09
0.09
0.09
0.09
16%
25%
25%
25%
8%
10%
10%
10%
0.13
0.13
0.13
0.08
0.10
0.10
0.09
0.14
0.14
0.14
0.09
0.09
0.09
0.09
Unc
16%
16%
20%
30%
14%
14%
14%
9%
9%
9%
9%
9%
9%
9%
HCHO Unc.
0.06 40%
0.08 30%
0.07 30%
0.09 30%
0.07 30%
0.07 30%
0.07 30%
0.06 30%
'MRS stands for mini surrogate experiment; MRS full surrogate experiment; R8 full surrogate, lowNOx
experiment.
bGroup of chamber characterization experiments used to estimate radical source parameters for the
incremental reactivity experiments. See Wang (1999).
Other than the radical source parameters, the most important chamber artifacts are
expected to be the intensity and spectral distribution of the artificial lights used in indoor
chambers such as the DTC. Blacklights have an unnatural spectrum above about 320 nm. Their
relative intensity is too high in the range from 320 - 360 nm and is negligible above 400 nm.
Differences between artificial lights and sunlight can be compensated for if the spectral
distribution of the light source is characterized and the action spectra of significant photolyzing
species are known. The light intensity in the DTC experiments is measured as ki, using the
quartz tube actinometry method of Zafonte et al. (1977). A constant spectral distribution based
primarily on measurements made with a LiCor Li-1800 spectroradiometer (Carter et al., 1995a)
is also used.
-338-
-------
2.3 Stochastic Programming
Determining optimal estimates with uncertainties for chamber characterization and
organic compound mechanism parameters is a stochastic parameter estimation problem (Figure
5). The inner loop is used to provide optimal parameter estimates for a given sample of random
mechanism and experimental variables. The outer loop provides the samples. The procedure
terminates when the probability distribution functions of the optimal parameter values are
determined. Regression analysis is then used to identify the major sources of uncertainty in the
parameter estimates and thus provide guidance for designing new experiments.
The parameter estimation problem is defined in this study as minimizing the weighted
squares of the differences between the model results and experimental measurements. The
primary comparison criterion used is the incremental reactivity with respect to D(C>3-NO) of the
test compound, IR[D(C>3-NO)]. The quantity D(O3-NO)t is defined as the amount of ozone
formed plus the amount of NO consumed at time t, that is D(O3-NO)t = [O3]t - ([NO]t - [NO]0).
The incremental reactivity with respect to D(Oa-NO) at time t is determined from pairs of
experiments or simulations as:
\base
(l)
where the differences are between the experiment or simulation with the base mixture of organic
compounds and the experiment or simulation with the test compound added. A second
comparison criterion is incremental reactivity with respect to OH radical levels, which is an
important factor because radical levels affect how rapidly all VOCs present, including the base
VOC components, react to form ozone (Carter et al., 1995b). The overall OH radical level is
defined as the integrated OH radical concentration, INTOH:
INTOH, =%[OH]vdc
Estimates of INTOH are derived from the rate of consumption of m-xylene:
(2)
INTOH t =
KOH
m—xylene
(3)
where D is the dilution rate and KOHm"xyIene is the rate constant for the reaction m-xylene+OH.
Then similarly the incremental reactivity with respect to INTOH at time t is calculated from parrs
of experiments or simulations as:
(4)
[voc]test
-339-
-------
For this study, the subjective estimate was made that the experimental data for D(O3-NO) are
three times as reliable as those for ESFTOH, so the two comparison criteria were weighted
accordingly. The residuals are normalized by the maximum value in each experiment of the
absolute value of the incremental reactivity. The n-butyl acetate and 2-butoxy ethanol
parameters are optimized to give the best fits over the full duration of all of the applicable
experiments.
Uncertainty estimates for RSI and HONO-F used in estimating the n-butyl acetate and 2-
butoxy ethanol parameters must reflect how the radical source parameters vary from experiment
to experiment as well as how they respond to input uncertainties. RSI and HONO-F are thus
optimized separately for each characterization experiment and random sample of input parameter
values. Then in estimating the parameters of the n-butyl acetate and 2-butoxy ethanol
mechanisms, values of RSI and HONO-F for the kth Monte Carlo run are sampled from
distributions that account for the invariability across the experiments. The same values of RSI
and HONO-F are used for the test and base case of each experiment. The mean values and
associated variances reported below for the estimated chamber characterization parameters are
the average and variance of the experimentally averaged optimal values across all of the Monte
Carlo samples.
Successive quadratic programming (SQP) (Han 1977, Powell 1977) is used to perform
the optimization for the parameter estimation problems because of its fast convergence rate and
widespread use for chemical process applications (Biegler et al. 1983). In the SQP method, at
each iteration the original problem is approximated as a quadratic program where the objective
function is quadratic and the constraints are linear. The quadratic subproblem is solved for each
step to obtain the next trial point. This cycle is repeated until the optimum is reached. The
decision variables in this study are parameters such as pN, qN and q6 for the reactions of 2-
butoxy ethanol and n-butyl acetate. These parameters determine the product yields of the
compounds' reactions with OH, but do not show up directly in the mechanism. So, each time the
parameters being estimated are changed during the optimization routine, the corresponding
product yields for the reactions in SAPRC format are calculated according to the relationships
prescribed in the mechanism generation program.
Monte Carlo analysis with Latin hypercube sampling (LHS) (Iman et al. 1984) is used for
the uncertainty analysis loop in Figure 5. Before performing the Monte Carlo simulations, first
order uncertainty analysis is used to identify the most influential parameters. The number of
input random variables can then be limited without neglecting significant sources of uncertainty
(Yang et al., 1995). Given a specified number of uncertain input parameters, LHS further
reduces the Monte Carlo computational requirements through selective representative sampling.
2.4. Input Parameter Uncertainties for Stochastic Programming
The sources of uncertainty considered in this study include the parameters of the SAPRC-
97 mechanism and the conditions of the incremental reactivity experiments. Uncertainty
estimates for mechanism parameters are compiled primarily from expert panel reviews
(Atkinson, 1989; DeMore et al., 1994; DeMore et al., 1997; Stockwell et al., 1994). The
compilation provided by Stockwell et al. (1994) for the SAPRC-90 mechanism was updated for
this study. Uncertainty estimates for experimental conditions were estimated for this study.
-340-
-------
M-xylene, a component of the base mixture used in the incremental reactivity experiments, has
mechanistic parameters that are estimated from chamber experiments in a manner similar to the
n-butyl acetate and 2-butoxy ethanol parameters considered here. Uncertainty estimates
calculated by Wang et al. (1999) using stochastic programming were used for the chamber-
derived parameters of the m-xylene mechanism.
Uncertainty estimates for light intensity (ki), initial NO, NO2 and VOC concentrations of
the incremental reactivity experiments are listed in Table 1. The light intensity uncertainty
estimates are based on the reproducibility of the quartz tube actinometry measurements. The
uncertainties in the initial NOX concentrations reflect the span and zero calibration errors of the
Teco Model 14B chemiluminescent NO/NOX monitor and the converter efficiency for NO2.
Uncertainties in the initial hydrocarbon concentrations primarily reflect calibration errors in the
GC FID detectors. Uncertainties in temperature were also considered but found to be negligible.
Wang et al. (1999) examined uncertainties in the spectral distribution of the chamber light source
but found them to be small compared to the action spectra uncertainties for the photolyzing
species.
Before the stochastic programming runs, a first order sensitivity analysis was performed
to identify the likely influential parameters. The sensitivity analysis was performed for both the
base case and test case of each incremental reactivity experiment. First-order sensitivity
coefficients indicating the response of Os concentrations to small variations in each of 207 input
parameters or variables were calculated using the Decoupled Direct Method (Dunker, 1984).
The variables considered included 183 reaction rate constants, 15 experimental conditions, 4
chamber-derived oxidation parameters for m-xylene and toluene, the 5 mechanistic parameters
identified in Figure 2 for n-butyl acetate and the 4 mechanistic parameters identified in Figure 1
for 2-butoxy ethanol. The sensitivity coefficients were combined with uncertainty estimates for
each of the parameters according to the standard propagation of errors formula. Based on the
first-order analysis, the 38 parameters shown in Table 2 account for more than 95% of the
uncertainty in the simulated Os concentrations for all of the incremental reactivity experiments
used in this study.
Table 2. Influential parameters identified by first order sensitivity analysis
Parameter"
NO2 + hv
(light intensity)
O3+NO
O3 + NO2
HONO+hv
(action spectra)
NO2 + OH
HO2+NO
HNO4
HO2 + O3
R02 + NO
RO2+HO2
Uncertainty Reference
Wang etal. (1999)
DeMore et al 1997
DeMore et al 1997
DeMore etal 1997
DeMore etal. 1994
DeMore et al. 1994
DeMore et al. 1994
DeMore etal 1997
DeMore etal 1997
DeMore etal 1997
Coefficient
of Variance
(CT/K)
0.12
0.10
0.14
0.27
0.18
2.40
0.27
0.42
0.75
Radical
Source
Parameters
X"
X
X
X
X
X
X
BU-
ACET
X
X
X
X
X
X
X
BUO-
ETOH
X
X
X
X
X
X
X
X
X
-341-
-------
HCHO+hv
CCHO + OH
CCO02+NO
CCOO2 + NO2
PAN
C2COO2+NO2
PPN
CRES + NOj
NC4+OH
PROPENE + OH
T2BUTE+OH
T2BUTE + O3
MXYLENE + OH
PROD2 + OH
BU-ACET + OH
BUO-ETOH + OH
SC(AFG2,
MXYLENE)
SC(MGLY,
MXYLENE)
Initial NO2
Initial NO
Initial HCHO
Initial ETHE
Initial TOLUENE
Initial MXYLENE
Initial BU-ACET
Initial BUO-ETOH
RSI
HONO-F
Solvent parameters
DeMoreetal 1997
DeMoreetall997
DeMore et al 1997
DeMoreetal. 1994
Bridier et al. 1991
Grosjean et al. 1994
Stockwell et al. 1994
Grosjean etal. 1994
Stockwell et al. 1994
Stockwell etal. 1994
Stockwell et al. 1994
Stockwell et al. 1994
Stockwell et al. 1994
Stockwell et al. 1994
Carter 1999
Carter 1999
Carter 1999
Wang etal. (1999)
Wang etal. (1999)
This Study
This Study
This Study
This study
This Study
This Study
This Study
This Study
This Study
This Study
Carter 1999
0.34
0.18
0.34
0.16
0.40
0.75
0.66
0.75
0.18
0.14
0.18
0.42
0.23
1.33
0.25
0.25
0.31
0.29
-11%
~ 10%
30% - 40%
12%
16% -25%
9% - 30%
10% - 27%
12%
X
f
X
X
X
X
X
X
X
X
X
X
X
X
Yc
X
X
X
X
X
X
X
X
X
X
X
ql.qN,
q6
X
X
X
X
X
X
X
X
X
X -
X
X
X
X
X
X
X
X
X
X
X
X
pN
"When a reaction label is shown, the parameter is the rate constant for that reaction. SC indicates the
stochiometric coefficient for the product of the OH reaction of the identified organic compound
SC(product,reactant)
bX indicates the parameter is treated as a random variable in stochastic parameter estimation
0 Although first order sensitivity analysis did not find this parameter influential for ozon e concentrations, it is
treated as a random variable because it is thought to be influential for calculated incremental reactivities.
Among these influential parameters, the solvent parameters are treated as outputs, i.e.,
they are the parameters to be estimated. The other parameters identified in Table 2 are treated as
random input variables with lognormal distributions for the reaction rate constants and chamber-
derived aromatics parameters, and normal distributions for the initial concentrations.
Several uncertain input variables are influential for both the radical source parameters
and the n-butyl acetate or 2-butoxy ethanol parameters (Table 2). The relationship found
between these input variables and the radical source parameters must be maintained in estimating
the parameters for the solvents. To accomplish this, LHS samples are generated including all of
the random variables for both stages of the analysis, except for the values of RSI and F-HONO.
-342-
-------
For use in the solvent parameter estimation, the values for RSI and HONO-F are drawn from the
distributions determined in the first stage for each run in the Monte Carlo sample. The chamber-
derived aromatics oxidation parameters for m-xylene were estimated in a previous study (Wang
et al., 1999) and found to be highly correlated with reaction rate constants for NO2+OH and m-
xylene+OH. These correlations are maintained in this study through use of correlated LHS
samples (Table 3). Further discussion of the treatment of the input uncertainties is provided by
Wang (1999).
Table 3. Correlation among input parameters for chamber -derived parameter estimation
Parameter
CCOO2+NO->
SC(AFG2, m-xylene)
SC(AFG2, m-xylene)
SC(MGLY, m-xylene)
SC(MGLY, m-xylene)
COV"
0.34
0.33
0.33
0.31
0.31
Correlated Parameter
CCOO2+NO2->
m-xylene + OH ->
NO2 + OH ->
m-xylene + OH ->
NO2 + OH ->
COV
0.16
0.23
0.27
0.23
0.27
Correlation
0.7
-0.63
0.55
-0.55
0.50
' COV = Coefficient of Variation
2.5 Incremental Reactivity Calculations
The atmospheric incremental reactivity (IR) of compound j is defined as the change in
ozone associated with the addition of a small amount of the compound (A[VOCj]) to a base
mixture of VOCs, in the presence of NOX and sunlight:
lim
A[KOC,]-»0
[0,1
3l[base VOC]+A[VOCj]
l[base VOC]
(5)
In this study, the atmospheric incremental reactivities are estimated as the local sensitivity of the
predicted ozone concentration to the initial concentrations of each organic compound in a
mixture (Yang et al., 1995).
Three incremental reactivity scales (Carter, 1994) representing different environmental
conditions are evaluated in this study. The maximum incremental reactivity (MIR) scale is
calculated using NOX levels adjusted to maximize the overall incremental reactivity of the base
VOC mixture. The maximum ozone incremental reactivity scale (MOIR) is calculated for
conditions that yield the maximum O3 concentration with the base VOC mixture. The equal
benefit incremental reactivity (EBIR) is defined for the conditions under which VOC and NOX
reductions are equally effective in reducing ozone. The simulation conditions used for the MIR,
MOIR and EBIR cases represent average conditions from 39 cities (Carter, 1994).
For control strategy analyses, the relative reactivity, R_IR, of a given VOC compared to
that of a base mixture may be of greater relevance than the absolute incremental reactivity:
R_IRj =
IR,
IR,
(6)
•base VOC
-343-
-------
The base VOC mixture used in this study is the mixture of reactive organic gases initially present
or emitted in the scenarios, excluding biogenic VOCs and VOCs present aloft.
The methods used to estimate uncertainties in incremental reactivities and the uncertain
input parameters considered in the analysis are presented below.
2.6 Linear Multivariate Regression Analysis
Linear multivariate regression analysis is applied to the stochastic parameter estimation
results to identity the influence of the random variables on the optimal values of the parameters.
The same method is used to identify the major sources of uncertainty in the incremental
reactivity estimates. As explained by Wang et al. (1999) the standardized regression model is
used for this analysis unless the variance inflation factor (VIF) is greater than 3.0, indicating the
presence of multi-collinearity between the explanatory random variables. In that case, ridge
regression is used.
3. Results
3.1 Parameter Estimation for Chamber Characterization Parameters
In the first stage of this study, radical source parameters were estimated for the DTC
chamber for the tune periods during which the n-butyl acetate and 2-butoxy ethanol incremental
reactivity experiments were conducted. For the chamber characterization experiments conducted
concurrently with the n-butyl acetate experiments, the average optimal values are 0.087 ppb for
RSI and 0.272 % for HONO-F. Across these experiments, the average uncertainties (Icy) in the
optimal values are about 33% of the mean for RSI and 78% for HONO-F. Similar values were
obtained for the chamber characterization experiments conducted along with the 2-butoxy
ethanol experiments. The average optimal values for this set of experiments are 0.057 ppb for
RSI and 0.604% for HONO-F, with corresponding uncertainties of about 35% for both RSI and
HONO-F. Results for individual experiments are given hi Wang (1999). The mean values
estimated by stochastic programming are similar to those estimated deterministically by Carter
(1999). A regression analysis was not performed for RSI and HONO-F in this study, but Wang
et al., (1999) showed that optimal values of RSI for the DTC chamber are sensitive to
uncertainties in the rate parameters for NO2+OH, n-butane+OH or CO+OH, and NO2+hv (i.e.,
light intensity). The most influential parameters for the average HONO-F values in the DTC
chamber are the rate parameters for HONO+hv (action spectra), n-butane+OH, NO2+OH and
NO2+hv.
3.2 Parameter Estimation for Solvent Parameters
In the first-order sensitivity analysis qN, ql and q6 were found to be influential to ozone
formation in the n-butyl acetate experiments. Two parameters, qN and q6 are treated here as the
parameters to be estimated from the chamber experiments. In the SAPRC-98 mechanism, the
value of ql was estimated from structure-activity relationships. The only influential chamber-
derived parameter in the 2-butoxy ethanol mechanism is the nitrate yield, pN.
With the random variables listed in Table 2, the optimal value of pN for 2-butoxy ethanol
was calculated to be 0.134+0.024. In comparison, the value used by Carter hi SAPRC98 is
0.127. Ridge regression results (Wang, 1999) indicate that the 18% uncertainty calculated for
-344-
-------
the nitrate yield results primarily from uncertainly in the radical source parameters estimated for
the DTC, the formaldehyde action spectra, the NO2 photolysis rate and the initial NC-2
concentrations.
For n-butyl acetate, stochastic optimization results in values of 0.720 ± 0.223 for q6 and
0.127 ± 0.050 for qN. The mean values are about 34% and 41% higher than those used in
SAPRC-98. Regression analysis of the stochastic optimization results shows that the influential
sources of uncertainty for all of the estimated parameters include the rate constants for PPN
formation and decomposition and PAN formation, the m-xylene initial concentration and the
methyl glyoxal yield from m-xylene. The ~40% uncertainty in the optimal value for qN is also
due to uncertainties in the rate constants for NOi+OH and trans-2-butene+Os, and the initial
concentrations of NO2 and HCHO used in the experiments. Uncertainties in the rate constants
for HNC-4 decomposition, n-butyl acetate oxidation, HCHO and NO2 photolysis, RO2+HO2,
HO2+NO and CRES+NO3, and uncertainties in the initial n-butyl acetate concentration are also
among the influential sources of the 31% uncertainty in the optimal value for q6.
3.3 Incremental Reactivity Estimates
In the next step of the analysis, uncertainties in the rate parameters of the SAPRC-97
chemical mechanism, including the estimated uncertainties in qN and q6 for n-butyl acetate and
pN for 2-butoxy ethanol, were propagated through incremental reactivity calculations using
Monte Carlo analysis with Latin hypercube sampling. The input random variables used in the
Monte Carlo analysis are given in Table 4. Correlations between n-butyl acetate and 2-butoxy
ethanol parameters and other SAPRC parameters incorporated in the analysis are shown hi Table
6. Correlations between the rate constants for CCOO2+NO and CCOO2+NO2 and between
aromatics mechanistic parameters and the rate constants for NO2+OH and the aromatic +OH
reactions were also included. A complete list is given by Wang (1999). Due to the limitations of
the LHS program, only the correlations higher than 0.3 are preserved in the incremental
reactivity calculations.
The results of deterministic incremental reactivity calculations with nominal values of
SAPRC-97 parameters are listed in Table 6 for comparison with incremental reactivities
calculated with SAPRC-98 (Carter, 1998). The mean values and the associated uncertainties for
the calculated incremental reactivities for n-butyl acetate and 2-butoxy ethanol obtained from
460 LHS samples are listed in Table 7 for absolute and relative incremental reactivities.
Reaction or Coefficients
O3 + NO->
O'D + H2O ->
O'D + M->
NO2 + OH->
CO + OH->
HO2 + NO->
HO2 + HO2 ->
Coefficient of
Variance
(O~i/Ki nominal)
0.10wa
0.18 CT
0.18 ™
0.27 u;
0.27 w
0.18 ll)
0.27 ra
Reaction or Coefficients
ethene + OH ->
propene + OH ->
isopene + OH ->
l,3-butadiene + OH->
2-m-l-butene + OH ->
2-m-2-butene + OH ->
224-TM-C5 + OH->
Coefficient of
Variance
(CTi/Ki nominal)
0.11 W
0.14W
0.19 ^
0.19 w
0.18W
0.18W
0.18W
-345-
-------
HO2 + HO2 + H2O->
RO2 + NO->
RQ2+KO2->
CRES + N03->
HCHO + OH->
CCHO + OH->
RCHO + OH->
CCOO2+NO->
CCO02 + NQ2->
CC002 + HQ2->
CCOO2 + RQz^>
C2COO2 + NO2->
PPN->
PAN->
NO2 + hv -> (action spectra)1
NO3 + hv->'
O3 + hv->'
HCHO + hv^1
CCHO + hv->'
RCHO + hv->'
MEK + hv->'
benzene + OH ->
toluene + OH ->
ethylbenzene + OH
1,2,3-trimethylbenzene + OH ->
1,2,4-trimethylbenzene + OH ->
1,3,5-trimethylbenzene + OH ->
p-xylene + OH ->
o-xylene + OH->
m-xylene + OH ->
NC4 + OH->
NC6+OH->
NC8 + OH->
CYCCs + OH->
2-methylpentane + OH ->
m-cyclopentane + OH ->
methanol + OH -> b
ethanoH-OH->"
BU-ACET+OH->
qNforBU-ACET
pNforBUO-ETOH
0.27 w
0.42 w
0.75 CT
0.75 w
0.23 «>
0.18 w
0.35 w
0.34 CT
0.16 (l>
0.75 «
0.75 w
0.75 «
0.66 w
0.40 w
0.18 CT
0.42 w
0.27 (2)
0.34 w
0.34 w
0.34 w
0.42 (3J
0.27 t3J
0.18 w
0.31 w
0.3 lw
0.31 (3J
0.3 lw
0.31 «
0.23 w
0.23 w
0.18W
0.18 w
0.18(3)
0.27 w
0.23 (3J
0.27 w
0.18 w
0.18 (7)
0.25 ra
0.40 w
0.18W
MTBE + OH->"
ETBE + OH->"
ethene + O3 ->
propene + O3 ->
isoprene + O3 ->
1,3-butadiene + O3 ->
2-m-l-butene + O3 ->
2-m-2-butene + O3 ->
ALK2 + OH->
AROl + OH->
ARO2 + OH->
OLE2 + OH->
OLE2 + O3 ->
OLE3 + OH->
OLE3 + O3 ->
P1U1C
SC(AFGl,benzene)d
SC(AFG2,toluene) e
SC(MGLY,toluene) '
SC(AFG2,ethylbenzene)
SC(MGLY,ethylbenzene)
SC(AFG2,123-TMB)
SC(MGLY,123-TMB)
SC(AFG2,124-TMB)
SC(MGLY,124-TMB)
SC(AFG2,135-TMB)
SC(MGLY,135-TMB)
SC(AFG2,p-xylene)
SC(MGLY,p-xylene)
SC(AFG2,o-xylene)
SC(MGLY,o-xylene)
SC(AFG2,m-xylene)
SC(MGLY,m-xylene)
SC(AFGl,AROl)g
SCCAFG2,ARO1) E
SC(MGLY,ARO1) g
SC(AFG2,ARO2) h
SC(MGLY,ARO2) h
BUO-ETOH+OH->
q6forBU-ACET
PROD2 + OH->
0.18 CT
0.18 w
0.23 ra
0.18 w
0.35 w
0.42 v>
0.35 w
0.42 «
0.27 w
0.27 (3)
0.27 w
0.18 ^
0.42 w
0.23 w
0.42 «
0.40 w
0.33 w
0.34 w
0.31 w
0.44 w
0.63 «
0.39 w
0.36 w
0.40 (SJ
0.49 w
0.40 w
0.29 w
0.45 w
0.71 CT
0.30 w
0.43 t5)
0.33 w
0.3 1^
0.33 w
0.29 (i)
0.29 (S)
0.23 w
0.20 lft)
0.25 CT
0.31 w
1.33 CT
a. The references for the uncertainty estimates are:
(1) DeMoreetal. 1994
(2) DeMoreetal. 1997
(3) Stockwelletal. 1994
(4) Grosjean et al. 1994, Brider et al. 1991
(5) Wang et al. 1999
(6) This study
-346-
-------
(7) estimated from Carter 1998
b uncertainty is estimated for this study according to t he uncertainty classes described by (47)
c quantum yield for photolysis of model species AFG1
d product yield for model species AFG1 from reaction benzene+OH
e SC(AFG2, aromatics) represents the chamber -derived aromatics oxidation parameter B1U2 (the prod uct yield for
model species AFG2) from reaction aromatics+OH .
f SC(MGLY, aromatics) represents the chamber -derived aromatics oxidation parameter B IMG (the product yield
for model species MGLY) from reaction aromatics+OH
8 The sample values of B1 Ul, B1U2 and B1MG for ARO1 are calculated as the weighted average of the
corresponding sample values for benzene, toluene and ethylbenzene, by reactivity -weighted emission mass.
hThe sample values of B1U2 andBlMG for ARO2 are calculated as the emis sion mass weighted average of the
corresponding sample values for o -xylene, p-xylene, m-xylene, 1,2,3-trimethylbenzene and 1,3,5 -
trimethylbenzene.
1 Only uncertainty in the action spectrum is considered.
Table 5 Correlated Parameters Used in Incremental Reactivity Calcul ations
Parameter
qN for BU-ACET
qN for BU-ACET
qN for BU-ACET
q6 for BU-ACET
pN for BUO-ETOH
pNforBUO-ETOH
cov
0.40
0.40
0.40
0.31
0.18
0.18
Correlated Parameter
C2COO2+NO2->
NO2 + OH ->
PPN->
C2CO02 + NO2->
HCHO + hv ->
NO2 + OH ->
COV
0.75
0.27
0.0.66
0.75
0.34
0.34
Correlation
-0.43
0.31
0.31
-0.34
0.34
0.40
Table 6
Deterministic Incremental Reactivites for 2 -Butoxy Ethanol and n-Butyl Acetate"
voc
SAPRC-97
This Study
2-butoxy ethanol
n-butyl acetate
Base Mixture
SAPRC-98
(Carter 1998)
2-butoxy ethanol
n-butyl acetate
Base mixture
MIR
1.41
0.50
1.50
1.35
0.47
1.21
MOIR
0.66
0.32
0.55
0.55
0.26
0.42
EBIR
0.46
0.23
0.344
0.35
0.17
0.24
R MLR"
0.94
0.33
1.0
1.12
0.39
1.0
R MOIR"
1.19
0.58
1.0
1.29
0.61
1.0
R EBIR"
1.34
0.67
1.0
1.42
0.71
1.0
The units for absolute incremental reactivity are ppmO 3/ppmC.
The units for relative incremental reactivity are (ppmO 3/ppmC) /(ppmO3/ppmC of base mixture)
b R_MIR represents relative MIR, R_MOIR relative MOIR and R_EBIR rel ative EBIR.
-347-
-------
Table 7 Stochastic Incremental Reactivities for 2 -Butoxy Ethanol and n-Butyl Acetate "
voc
2-butoxy ethanol
n-butyl acetate
Base Mixture
MIR
1.12
(24%)
0.414
(37%)
1.25
(21%)
MOIR
0.59
(24%)
0.29
(34%)
0.56
(24%)
EBIR
0.40
(28%)
0.20
(41%)
0.33
(22%)
R MIR"
0.90
(16%)
0.34
(38%)
R MOIR"
1.10
(15%)
0.53
(31%)
R EBIR"
1.20
(16%)
0.60
(30%)
* Mean and (coefficient of variation). The units for absolute incremental reactivity are ppmO 3/ppmC.
The units for relative incremental reactivi ty are (ppmCVppmC) /(ppmOs/ppmC of base mixture)
b R_MIR represents relative MIR, R_MOIR represents relative MOIR, and R_EBIR represents relative EBIR.
Table 6 shows incremental reactivity values for n-butyl acetate and 2-butoxy ethanol
calculated using nominal values of SAPRC-97 parameters together with the best estimates of the
chamber-derived mechanism parameters pN, qN and q6. MIR values calculated here are close
to those calculated with SAPRC-98 (Carter, 1998), about 1.4 ppmO3/ppmC for 2-butoxy ethanol
and 0.5 ppmOs/ppmC for n-butyl acetate. The deterministic estimates of the MOIR and EBIR,
respectively, are 0.66 ppmOa/ppmC and 0.46 ppmOs/ppmC for 2-butoxy ethanol and 0.32
ppmOa/ppmC and 0.23 ppmOs/ppmC for n-butyl acetate. The mean values from the Monte
Carlo simulations shown in Table 7 are generally slightly lower than the nominal estimates given
in Table 6.
The uncertainty level for the estimated incremental reactivities for 2-butoxy ethanol is
about 25% in the MIR, MOIR and EBIR cases. For n-butyl acetate, the uncertainty level ranges
from 34 to 41% across the three cases. As found previously for most VOCs (Wang et al., 1999),
the uncertainty in the relative incremental reactivities is less than that in the corresponding
absolute incremental reactivities. For 2-butoxy ethanol, the uncertainty for the relative
incremental reactivities is about 15% for the three cases studied. The uncertainty for the relative
incremental reactivity for n-butyl acetate ranges from 30 to 38% for the three cases.
As shown in Wang (1999) the parameters identified as most influential for the absolute
incremental reactivities of 2-butoxy ethanol and n-butyl acetate include the rate constants for
their reactions with OH, the NOa photolysis rate and rate constants for PPN and PAN chemistry
and Oa+NO. The rate constant for lumped higher ketone (PROD2)+OH is highly influential to
the absolute incremental reactivities for 2-butoxy ethanol, which are also sensitive to the
chamber derived methyl glyoxal yield from the lumped aromatic species ARO2 and the HCHO
photolysis rate in the MIR and MOIR conditions. Moreover, the MIR of 2-butoxy ethanol is also
sensitive to the rate constant of HOi+NO, the RCHO photolysis rate and the chamber-derived
organic nitrate yield pN. The MER of n-butyl acetate is also sensitive to the rate constants for
ARO2+OH and Oa photolysis, and the chamber-derived parameters qN and q6. The organic
nitrate yield qN is also influential for the MOIR and EBIR of n-butyl acetate. However, the
MOIR and EBIR of 2-butoxy ethanol are not sensitive to its organic nitrate yield, pN. Instead,
uncertainty in the Oa photolysis rate and O!D reaction rate constants appear relatively important
for the MOIR for 2-butoxy ethanol. With the exception of the PROD2+OH rate constant and the
-348-
-------
mechanistic parameters for the two compounds, the influential parameters are similar to those for
most compounds that react at average or slower than average rates (Yang et al., 1995; Wang et
al, 1999).
Compared with the absolute MIR, the relative MIR for 2-butoxy ethanol is more sensitive
to the rate parameters for its reaction with OH, PROD2+OH and NO2+OH, and to the pN. The
rate parameters for 63, HCHO and RCHO photolysis and O!D reactions and the chamber-
derived aromatics parameters for the lumped species ARO2 are also more influential for the
relative MIR than the absolute MIR. On the other hand, the relative MIR of 2-butoxy ethanol is
less sensitive than the absolute MIR to the rate parameters for NO2 photolysis, PAN and PPN
chemistry. The influential parameters for the relative MOIR and EBIR of 2-butoxy ethanol
include the rate constants for its reaction with OH, PROD2+OH, O3 photolysis and PAN and
PPN chemistry.
The influential parameters for the relative incremental reactivities of n-butyl acetate
include the rate constants for n-butyl acetate+OH, NO2+OH, O!D chemistry, NO2 photolysis,
and the chamber-derived parameter qN. 0$ and HCHO photolysis rates are also influential in the
MIR and MOIR cases. Parameters related to PAN and PPN chemistry are very influential in the
MOIR and EBIR cases. The relative EBIR is especially sensitive to the rate constant for n-butyl
acetate+ OH and the value of qN, compared to the absolute EBIR.
4. Summary and Conclusions
Through formal uncertainty analysis, this study examined the uncertainties in calculated
incremental reactivities for 2-butoxy ethanol and n-butyl acetate. The analysis considers uncerta
inties in the initial conditions, radical source parameters and light intensity of the incremental
reactivity experiments used to estimate mechanistic parameters for the two compounds, and in
the other parameters of the SAPRC-97 mechanism.
The uncertainty in the chamber-derived parameters of the 2-buxtoxy ethanol and n-butyl
acetate mechanisms ranges from 18% for the organic nitrate yield (pN) from 2-butoxy ethanol to
40% for the organic nitrate yield (qN) from n-butyl acetate. The stochastically estimated value
of pN for 2-butoxy ethanol is close to that used in SAPRC-98, while the values of qN and q6 for
n-butyl acetate are about 40% and 35% higher than the SAPRC-98 values. The uncertainty in
the optimal value for qN from n-butyl acetate is primarily due to uncertainties in the initial
concentrations of m-xylene used in the experiments and the rate constants for PPN formation and
decomposition. The uncertainty in the optimal value for pN from 2-butoxy ethanol is influenced
most by uncertainty in the radical source parameters estimated for the DTC.
The absolute incremental reactivities estimated in this study for 2-butoxy ethanol are 1.12
± 0.27, 0.59 + 0.14 and 0.40 + 0.11 ppm O3/ppmC, respectively, under MIR, MOIR and EBIR
conditions. For 2-butoxy ethanol, the estimated uncertainties of about 25% relative to the mean
are comparable to those calculated previously for most VOCs with no chamber-derived
parameters in their mechanisms (Yang et al., 1995; 1996; Wang et al., 1999). The relative MIR,
MOIR and EBIR of 2-butoxy ethanol are 0.90 ± 0.14, 1.08 + 0.16 and 1.20 + 0.19, respectively.
-349-
-------
Uncertainties in these relative reactivities, which are about 15% of the mean estimates, are
comparable to or lower than those estimated for many other VOCs including relatively well-
studied light alkanes.
The MIR, MOffi. and EBIR values estimated for n-butyl acetate are 0.41 + 0.16, 0.29 ±
0.10 and 0.20 ± 0.08 ppm O3/ppmC, respectively. The uncertainties in these values, which range
from 34 to 41% relative to the mean estimates, are comparable to those calculated by Wang et al.
(1999) for aromatic compounds with chamber-derived parameters, and somewhat higher than
incremental reactivity uncertainty estimates for many other VOCs. The respective relative
incremental reactivities for n-butyl acetate are 0.34 ±0.13, 0.53 ± 0.16 and 0.60 ±0.18. The 30
to 38% uncertainties in these estimates are at the upper end of the range of estimates obtained by
Wang et al. (1999) for other VOCs without chamber-derived parameters in their mechanisms.
The absolute incremental reactivities for 2-butoxy ethanol and n-butyl acetate are
sensitive to the rate constants for their reactions with OH, NOa and O3 photolysis, and PAN and
PPN chemistry. For the MIR and MOIR of 2-butoxy ethanol, the uncertainty in the rate constant
for PROD2+ OH is also influential. The organic nitrate yield from 2-butoxy ethanol, pN,
contributes at most 1% to the uncertainty in its absolute incremental reactivities. About 3% of
the uncertainty in the absolute incremental reactivities of n-butyl acetate is attributable to its
organic nitrate yield, qN.
The relative reactivity estimates for 2-butoxy ethanol are strongly influenced by
uncertainty in the rate constants for 2-butoxy ethanol + OH, and for PROD2 + OH.
Uncertainties in the rate parameters for n-butyl acetate + OH, O3 photolysis and NO2 + OH are
most influential for n-butyl acetate relative reactivity estimates. Uncertainty in pN contributes
about 2 to 3% of the total uncertainty in the relative reactivities of 2-butoxy ethanol. About 4 to
7% of the uncertainty in the n-butyl acetate relative reactivities is attributable to qN. With the
exception of the PROD2 + OH rate constant and the chamber-derived mechanistic parameters for
the two compounds, the parameters that contribute most to the uncertainty in both the absolute
and relative reactivities of 2-butoxy ethanol and n-butyl acetate are similar to those identified for
other VOCs in previous studies (Yang et al., 1995; Wang et al., 1999).
A significant finding of this study is that most of the uncertainty in the Incremental
reactivity estimates for n-butyl acetate and 2-butoxy ethanol is attributable to parameters of the
base SAPRC mechanism. The uncertainties in the chamber-derived mechanistic parameters
specific to their reactions and in turn the conditions of the experiments used to estimate these
parameters contribute at most 7% of the total uncertainty in the reactivity estimates.
Acknowledgments
Support for this research was provided by the California Air Resources Board, under
GARB contract no.95-331. The authors appreciate the nonlinear optimization programs provided
by Professor Urmila Diewkar at University of Carnegie Mellon.
-350-
-------
References
Aschmann S.M. and Atkinson R. (1998): Kinetics of the gas-phase reactions of the OH radical
with selected glycol ethers, glycols and alcohols. International Journal of Chemical
Kinetics. 30:(8) 553-540. August.
Atkinson, R., Baulch, D.L., Cox, R.A., Hampson, R.FJr., Derr, A., Troe, J.(1989): Evaluated
kinetic and photochemical data fro atmospheric chemistry: supplement III. IUPAC
subcommittee on gas kinetic data evalution for atmospheric chemistry, Journal ofPhyical
and Chemical Reference Data, 13:1259- 1380.
Biegler, L.T. and Hughes R.R.(1983): Process optimization: a comparative study. Computers
and Chemical Engineering, 7:645-661.
Brider, I., Caralp, F., Loirat, H., Lesclaux, R. and Veyret, B.(1991): Kinetic and theoretical
studies of the reactions CH3C(O)O2+NO2+M = CH3C(O)O2NO2+M between 248 and 393K
and between 30 and 760 Torr. Journal of Physical Chemistry, 95: 3594-3600.
Carter, W.P.L.(1994): Development of ozone reactivity scales for volatile organic compounds.
Journal of the Air and Waste Management Association, 44:881-899.
Carter, W.P.L., Luo, D., Malkina, I.L. and Pierce, J.A.(1995a): Environmental chamber studies
of atmospheric reactivities of volatile organic compounds: effects of varying ROG surrogate
and NOX. Final report to Coordinating Research Council, Inc., Project ME-9, California Air
Resources Board, Contract A032-0692, and South Coast Air Quality Management District,
Contract C91323, March.
Carter, W.P.L., Pierce, J. A., Luo, D. and Malkina, L. (1995b): Environmental chamber study of
maximum incremental reactivities of volatile organic compounds. May.
Carter W.P.L.(1996): Development and application of an updated photochemical mechanism for
VOC reactivity assessment. Quarterly Report to the California Air Resources Board,
Contract NO. 92-329.
Carter,W.P.L. Luo, D., and Malkina, I.L.(1997): Environmental chamber studies for
development of an updated photochemical mechanism for VOC reactivity assessment, Final
report to California Air Resources Board, contract 92-345; coordinating Research Council,
Inc., Project M- 9; National Renewable Energy Laboratory, contract ZF-2-12252-07.
Carter, W.P.L. (1998) Updated maximum incremental reactivity scale for regulatory
applications, preliminary report to California Air Resources Board, Contract No. 95-308,
August.
Carter, W.P.L., Luo, D., Malkina, I.L. (1998) Investigation of atmospheric reactivities of
selected stationary source VOCs, Final Report for CARB Contract 95-308.
Carter (1999): personal communication
Dagaut, P.T., J. Wallington, R. Liu and MJ. Kurylo (1988) 22nd international symposium on
combustion, Seattle, August 14 - 19DeMore, W.B., Sander, S.P., Golden, D.M., Molina,
M.J., Hampson, R.F., Kurylo, M.J., Howard, C.J., Ravishankara, A.R., Kolb, M.J., and
Molina M.J.(1994): Chemical kinetics and photochemical data for use in stratospheric
modeling, Evaluation No. 11, JPL Publ., 94-26 .
DeMore, W.B., Golden, D.M., Hampson, R.F., Howard, C.J., Ravishankara, A.R., Kolb, M.J.,
and Molina, M.J.(1997): Chemical kinetics and photochemical data for use in stratospheric
modeling, Evaluation 12, JPL Publ. 97-4.
-351-
-------
Dunker, A.M.(1984): The direct decoupled method for calculating sensitivity coefficients in
chemical kinetics. Journal of Chemical Physics, 81: 2385-2393.
Grosjean, D., E. Grosjean and E.L. Williams II (1994). Thermal decomposition of PAN, PPN
and Vinyl-PAN. Air and Waste Management Association 44, 391-396.
Han, S.P. (1977). A globally convergent method for nonlinear programming. Journal of
Optimization Theory and Applications 22(3), 297-309.
Hartmann, D., A. Gedra, D. Rhasa and R. Zellner (1986) Proceedings, 4th European Symposium
on the Physico-Chemical Behavior of Atmospheric Pollutants, 1986; D. Riedel Publishing
Co., Dordrecht, Holland, 1987, p.225
Iman, R.L. and M.J. Shortencarier (1984). A FORTRAN77 program and user's guide for the
generation of Latin Hypercube and random samples for use with computer models, Rep.
SAND83-2365, U.S. Dep. of Energy, Sandia National Laboratory, Albuquerque, N.M,
1984.
McBride, S.J., Oravetz, M.A. and Russell, A.G.(1997): Cost-benefit and uncertainty issues in
using organic reactivity to regulate urban ozone. Environmental Science &
Technology/News, 31:238A-244A.
Powell, MJ.D.(1977):A fast algorithm for nonlinear constrained optimization calculations.
Dunde Conf. Numerical Anal, Dundee, Scotland.
StemmlerK., Mengon W., Kinnison, D. and Kerr, J.A.(1997): OH radical-initiated oxidation of
2-butoxyethanol under laboratory conditions related to the troposphere: product studies and
proposed mechanism. Environmetal Science Technology 31:1496-1504.
Stockwell, W.R., Yang, YJ. and Milford, J.B.(1994): A compilation of estimated uncertainty
factors for rate constants in W.P.L. Carter's detailed mechanism. Prepared for the
Automotive Emissions Cooperative Research Program under Coordinating Research
Council, Inc. Contract No. AQIRP-19-92.
Tuazon, E. C., Aschmann S. M. and Atkinson R. (1998): Products of the gas-phase reactions of
the OH radical with l-methoxy-2-propanol and 2-butoxyethanol. Environmental Science
Technology 32: 3336-3345.
Yang, Y.J., Stockwell, W.R., and Milford, J.B.(1995): Uncertainty in Incremental reactivities of
volatile organic compounds, Environmental Science Technology, 23:1336-1345.
Wang, L. (1999): Stochastic parameter estimation and uncertainty analysis applied to
photochemical air quality models, Ph.D. Thesis, Department of Mechanical Engineering,
University of Colorado.
Wang L., Milford, J.B. and Carter, W.P.L. (1999): Reactivity estimates for aromatics
compounds. Draft Report to the California Air Resources Board. Contract No. 95-331.
Zafonte, L., Rieger, P.L. and Holmes, J.R.(1977): Nitrogen dioxide photolysis in the Los
Angeles atmosphere. Environmental Science Technology, 11: 483-387.
-352-
-------
CH3-CH2-CH2-CH2-O-
CH[.]-CH2-OH
CH3-CH2-CH2-CH[.]-O-
CH2-CH2-OH
CH3-CH2-CH2-CH2-0-
CH2-CH[.]-OH
Minor
routines
CH3-CH2~CH2-CH2-
O-CH[OO.]-CH2-OH
I NO
CH3-CH2-CH2-CH2-0-
CH(ON02)-CH2-OH
N-butyl formate
CH3-CH2-CH2-CH[OO.]-
0-CH2-CH2-OH
pN
1-pN
i
Organic nitrate
2-hydroxy formate
propanal
butoxyacetaldehyde
KOH: Dagaut et alv 1988
Stemmler et al., 1996
Aschmann et al., 1998
pi, p2 and p3: estimated from
product data (Stemmler et al.,
1997, Tuazon et al., 1998)
pN: estimated from chamber data
Figure 1. Key Features of the 2-Butoxy Ethanol Mechanism Used in SAPRC.
-353-
-------
OTC491
DTC498
DTCS05
IR d(O3-NO)
0.10
DTC493
IRd(OS-NO)
0.50
0.40
0.20
20 ' If
.10 J
5 6
14.0
Experiment SAPRC-97 (S98 values)
-SAPRC-97(optpN) SAPRC-98 ''•
Figure 2. Simulation Performance for 2-Butoxy Ethanol Incremental Reactivity Experiments (4
of?)
-354-
-------
J!^^
CH3-CH2-CH2-CH[.j-
0-CO-CH3
i
, Q2
CH3-CH2-CH2- •
CH[00.]-0-CO-CH3
CH3-CH2-CH2-CH2-O-CO-CH3 + OH
-— -
^— • •
qT^-I^
CH3-CH2-CH[.]-CH2-
O-CO-CH3
'
CH3-CH2-CH[OO.]-
CH2-0-CO-CH3
— — _JX04
• — >•
CH3-CO-O-CH2-CH2-
CH[.]-CH3
1
CH3-CO-O-CH2-
CH2-CH[00.]-CH3
1
Minor
routines
Organic
nitrate
CH3-CH2-CH2-
CH[O.]-O-CO-CH3
Organic nitrate
and other
products
Organic
nitrate
• <~
1.4-
CH3-CO-O-CH2-
CH2-CH[0.]-CH3
— - -ZA -*
jecomp.
Ester
rearrangement
+0,, or
2f
decomp.
isomenzation
Limited studies: KOH: Atkinson 1989; no product data
ql, q2, q3: estimated by structure-reactivity method
q4: derived from estimated alkoxy radical rate constants
qN, q6: estimated from chamber data
Figure 3. Key Features of the N-Butyl Acetate Mechanism Used in SAPRC.
-355-
-------
DTC365
IRd(O3-NO)
0.00
DTC402
Experiment SAPRC-97 (S98 values)
-SAPRC-97(optqN,q6) SAPRC-98
Figure 4. Simulation Perfonnance for n-Butyl Acetate Incremental Reactivity Experiments (4 of
7)
-356-
-------
Distribution of
Optimal
Parameters
Stochastic
Sampling
Model Simulation
Evaluate Objective
Functions and
Constraints
Optimization
Obtain New Values of
Estimated Parameters
Samples of
Uncertain Input
Parameters
Figure 5. Stochastic Programming Approach Using Monte Carlo/Latin Hypercube Sampling and
Successive Quadratic Programming.
-357-
-------
Atmospheric Availability as a Component of the Tropospheric Ozone-Forming
Potential of Volatile Organic Compounds
Brian T. Keen, Jonathan J. Kurland and Scott P. Christensen
Union Carbide Corporation
South Charleston, West Virginia
Summary
Modeling of ozone formation requires accurate emissions inventories and chemical
mechanisms. Consideration of equilibrium vapor pressures and partition between gas phase
and aerosols suggested that many low-vapor-pressure (LVP) VOC in consumer, commercial,
and agricultural products would be present predominately in the gas phase. However, this
conclusion may not be correct in view of studies indicating alternate fates for those VOCs.
The tropospheric concentration of a VOC is affected by both the rate and extent of release
from an emission source and by the rate of removal through a variety of competing processes
(e.g. photooxidation, deposition, horizontal and vertical transport, aerosol formation). There
are many ways that compounds of low volatility, especially those which are hydrophilic, may
be prevented from entering the atmosphere or removed once they enter, but quantitative
assessments are rare.
The kinetic, mechanistic and smog chamber studies upon which calculations of MIRs are
based do not include transport to water or soil, which are present in the natural environment.
Not only the compound to be evaluated, but also its oxidation products, may not participate
in ozone formation to the extent predicted if they have other environmental fates besides
oxidation in air and advection in air.
Models need to be developed that can incorporate a chemical's entire tropospheric fate.
Environmental fate modeling, used extensively to track persistent organic pollutants, may be
applicable in assessing tropospheric ozone-forming potential of VOCs as well.
Introduction
Current EPA policy presumes the availability of low-vapor-pressure (LVP) VOCs in air1.
The assumption is based on equilibrium considerations for air-surface exchange and
consideration that during performance of EPA Method 24 certain compounds, such as
diethylene glycol monobutyl ether, may be volatilized even though they have a vapor pressure
of less than 0.1 mm Hg at 20 "C.1
However, this conclusion may not be correct because there are alternative fates for those
VOCs2'3. Compounds of low volatility, especially those which are hydrophilic, have many
competing ways (e.g., photooxidation, deposition, horizontal and vertical transport and
aerosol formation) by which they may be prevented from entering the atmosphere or
removed once they enter it, but quantitative assessments are rare.
Though some regional-scale predictive models for ozone formation incorporate information
on fate and removal, others do not. Models need to be developed that can incorporate all
that that is known about a chemical's tropospheric fate and removal. Environmental fate
-358-
-------
modeling has been used extensively to track persistent organic pollutants. Research is
needed to see what modifications are required to make it a useful tool for assessing
tropospheric ozone-forming potential of VOCs with short atmospheric lifetimes.
Environmental fate and transport models are available to examine how a chemical partitions
among air, water, soil, and sediment. These fate and transport models examine intermedia
transport rates for various diffusive and nondiffusive processes and estimates the mass
fraction of a chemical in each environmental compartment. Four processes govern the
transport of a chemical from air to water: diffusion (absorption), dissolution in rain, and wet
and dry deposition of particle-associated chemical. The troposphere is treated as an air-
aerosol mixture in these calculations with chemicals partitioning between the two phases.
We used a version of this modeling technique as a reality check for our hypotheses that the
extent to which organics in consumer products and coatings enter and remain in the air, is a
factor in ozone generating potential that may be as important as atmospheric reactivity. We
have begun to explore whether the atmospheric availability of organic compounds in
products can be determined using environmental fate modeling.
Results of Environmental Fate Modeling
To get a rough estimate of the potential removal of compounds from the atmosphere we
have used Mackay's EQC Model version 1.0, Level III5, which uses the environment of
southern Ontario, Canada as the "landscape.".
-Level II! Diagram
EQC Model v. 1.0
Level III
Chemical: 2-Butoxyethanol
100 kg/h
S.892 kg/h
Okg/h
13.1 kg/h
16.5 kg/h
-359-
-------
li-vi;! Ml f)iA(jt
-------
To roughly quantify the potential removal of compounds from the atmosphere we have
used the Level III EQC model and calculated the fraction of material either reacted in air
or advected in air, as a conservative estimate of the amount that can contribute to ozone
formation. We refer to this as the emission inventory factor (EIF) in the following tables,
since consumer product sales would have to be multiplied by this factor to get the
emissions for use in airshed models that do not incorporate environmental fate.
Calculated % VOC Contributing to Ozone Formation from Oxygenated Chemicals by the
EQC Model
Chemical
Acetone
Propylene Glycol
Methyl Ether
Ethylene Glycol
Methyl Ether
Ethylene Glycol Ethyl
Ether
Propylene Glycol
Butyl Ether
Ethylene Glycol Butyl
Ether
Dipropylene Glycol
Methyl Ether
Dipropylene Glycol
Butyl Ether
Diethylene Glycol
Butyl Ether
Ethoxytriglycol
Butoxytriglycol
Boiling Point
CO
56
120
124
135
170
171
188
229
231
256
281
Vapor Pressure
at20°C
(mm Hg)
185
8
6
4
0.6
0.6
0.2
0.03
0.01
0.01
0.00002
EIFair
% VOC
96
27
93
93
97
30
1
90
9
0
0
EiFsoil
% VOC
27
1
8
8
7
1
0
4
0
0
0
EIFaq
% VOC
12
0
1
1
1
0
0
0
0
0
0
Assumptions:
• Environment is representation of southern Ontario, Canada
• EIFair = % reaction + advection in air for 100% emission into air
• EIFsoii = % reaction + advection in air for 100% emission into soil
• EIFaq = % reaction + advection in air for 100% emission into water
Same mass transfer coefficients for all chemicalsO
-361-
-------
Calculated % VOC Contributing to Ozone Formation from Common Chemicals by
the EQC Model
Chemical
Methanol
Ethanol
Butanol
Hexanol
Methyl Ethyl Ketone
2-Pentanone
2-Heptanone
Methyl Isobutyl Ketone
Methyl Acetate
Ethyl Acetate
Butyl Acetate
Hexane
Decane
Dodecane
Hexadecane
Boiling
Point
(°C)
64.5
78.3
117.7
157.0
80
102
152
116
57
77
126
69
174
216
287
Vapor
Pressure
at 20 °C
(mmHg)
97
44
5
0.5
74
27
2
43
173
75
8
120
0.9
0.1
0.001
EIFair
%VOC
92
94
97
97
96
98
99
100
97
98
99
100
100
100
100
EIFsoii
% VOC
10
9
11
6
30
40
23
47
48
54
46
98
96
65
0
EIFaq
%VOC
1
0
2
3
8
12
13
16
14
16
21
52
48
24
0
Assumptions:
Environment is representation of southern Ontario, Canada
EIFair — % reaction + advection in air for 100% emission into air
EIFsoii = % reaction + advection in air for 100% emission into soil
EIFaq = % reaction + advection in air for 100% emission into water
Same mass transfer coefficients for all chemicals
We took the worst case, emission entirely to air, and calculated an ozone-forming potential by
multiplying the maximum incremental reactivity (MIR) by the emission inventory factor (EIF).
-362-
-------
Chemical
Acetone
C? Alkyl Ketone
2-Butoxyethanol
Dodecane
Diethylene Glycol Butyl
Ether
Boiling
Point
(°C)
56
152
171
216
231
Vapor Pressure
at 20 °C
(mmHg)
185
2
0.6
0.1
0.01
MIR
0.48
2.65
3.23
0.72
3.36
EIFair
(fraction
voc
0.96
0.99
0.30
1.00
0.09
MIR*EIF
0.46
2.62
0.97
0.72
0.30
MIRs (g of O3/g of VOC) are from W. P. L. Carter, Updated Maximum Incremental Reactivity Scale for
Regulatory Applications, Preliminary Report to California Air Resources Board, Contract No. 95-308,
August 6,1998
Note that the modeling shows striking differences in atmospheric availability among
chemical classes. Vapor pressure alone is not adequate to estimate atmospheric
availability. Volatility correlates better with Henry's Law constants and kOW. A
remarkable result is that, on this scale, the ozone-forming potential of diethylene glycol
butyl ether is less than that of acetone, which is exempted from consideration as a VOC
due to "negligible reactivity". Is this real?
There are difficulies in employing existing environmental fate modeling to atmospheric
availability for estimation of tropospheric ozone-forming potential. Most fugacity-based
environmental fate modeling has dealt with persistent organic pollutants, especially PAHs
and chlorinated hydrocarbons, which are hydrophobic and unreactive with hydroxyl
radicals. The time scale has been years. To examine the fate of hydrophilic compounds
with low vapor pressures and relatively high rates of photochemical oxidation one needs
to work with much shorter times and a limited "landscape." The assumption of
equilibrium within compartments needs to be tested. It does appear that transport rates
between compartments is fast. As discussed below application of existing environmental
fate modeling to urban ozone episodes is complicated by the existence in urban areas of
processes which are not considered in the EQC model, but may be important.
Regional modeling for estimates of ozone-forming potential related to an 80 ppb ozone
standard will be easier than modeling urban areas that are in noncompliance with the 120
ppb standard, since the landscape will more closely approximate that used in current
modeling. Daytime versus nighttime transport and reaction may need to be separated in
the environmental fate modeling if we are looking at multi-day scenarios.
The following processes have been identified as potentially important contributors to the
atmospheric availability of a VOC. Although data exist to support the inclusion of these
processes in reactivity models under specific sets of circumstances and conditions of use,
-363-
-------
it is difficult to judge a priori their overall importance on model-predicted ozone
formation. Research is needed to further evaluate the impact of these processes for
specific classes of chemicals and environmental conditions. The physico-chemical and
kinetic data needed to conduct this research are generally available, either through direct
measurement or predictive models.
A significant percentage of the hydrophilic VOCs found in consumer products may never
reach the atmosphere, but will go "down-the-drain" in aqueous solution and undergo
biological degradation7'8. Indoor atmospheric chemistry initiated by hydroxyl radicals
generated indoors may have a profound effect upon VOC emissions by oxidizing a portion
of the VOCs emitted from consumer products used indoors and limiting the amount
available for ozone formation outdoors. Adsorption onto inert materials has been
established as important in indoor air studies10. The absorption of SOC on vegetation has
been studied mainly to determine whether our food is being contaminated by air-borne
persistent organic pollutants. How vegetation cleanses the atmosphere has received less
attention. Rain-out and wash-out are important for both vapors and aerosols. Since many
oxygenated solvents and VOC intermediates with low volatilities tend to have high
Henry's law coefficients, this process may be particularly important for LVP VOCs.
Photochemical oxidation of organic molecules containing seven or more carbon atoms can
generate semi-volatile secondary organic aerosol (SOAs)". SOA formation represents a
sink for the removal of many emitted LVP VOCs.
Potential Effects on MIR Estimation
The kinetic, mechanistic and smog chamber studies upon which calculations of MIRs are
based do not include transport to water or soil, which are present in the natural
environment. Not only the compound to be evaluated, but also its oxidation products, may
not participate in ozone formation to the extent predicted if they have other environmental
fates besides oxidation in air and advection in air.
The mechanism for n-decane has several pathways for the reaction of decylperoxy
radicals, but they all lead mainly to or through decanol, decanone and (by intramolecular
hydrogen abstraction [rearrangement] of alkoxy radicals) hydroxydecanones. These
compounds have vapor pressures much lower than decane and much greater polarity.
As an example, we ran the EQC Model v. 1.0 Level III calculation on 6-hydroxy-3-
decanone emitted into air (since it is formed in air from decane). Most of it is transported
rapidly to soil. That which reaches water from air or soil is rapidly degraded so very little
reaches sediment. Accumulation is in soil.
Emission to Air
Reaction in
Advection in
Air
12.6%
2.9%
Water
22.6%
1.8%
Son
60.1%
Sediment
0.0%
0.0%
-364-
-------
DECANE
b.p. 174.1 °C
ROOH
CHs-CH^CH-CCH^e-CHs
RO2«
b.p. 211 °C
ROH <
20%
20%
CH3-OH2-C-(CH2)6-CH3
b.p. 203 °C
60%
CH3-CH2-CH-CH2-C3I2-CH2-(CH2)3-CH3
R'O2H
'HO2-
OH / O2«
I ' I
CH3-CH2-CH-CH2-CH2-CH-(C3I2)3-CH3
RO2»
NO
17%
83%
R'ON02
O OH
II I
CH3-CH2-C-(CH2)2-CH-(CH2)3-CH3
+ isomcric ketoalcohols
b.p. - 280 °C
CH3-CH2-CH-CH2-CH2-CH-{CH2)3-CH3
b.p. - 175 °C (14 mm Hg) => -305"C
An oxidation product of ethylene glycol is glycolaldehyde. The estimated MIR of ethylene
glycol, based on experiments at low relative humidity, may not reflect its true ozone-
forming potential, since both ethylene glycol and glycolaldehyde have low volatility and
high water solubility and may be removed from the atmosphere by processes other than
gas-phase oxidation.
-365-
-------
CONCLUSIONS
• Reactivity (MER) alone is an insufficient indication of the ozone-forming potential of
volatile organic compounds (VOC). Atmospheric availability is also important.
• Vapor pressure alone is not adequate to estimate atmospheric availability. Volatility
correlates better with Henry's law constants and kow.
• Environmental fate modeling, used extensively to track persistent organic pollutants, may
be applicable in assessing tropospheric ozone-forming potential of VOCs as well.
• Such modeling shows striking differences in atmospheric availability among chemical
classes.
• The challenge is to extend environmental fate modeling to smaller areas and shorter times
than normally assessed.
• The urban "landscape" may be quite different from that in larger areas. Relatively little is
known about indoor "sinks" for VOC, exchange of indoor and outdoor air, adsorption or
absorption on urban surfaces and vegetation and other aspects peculiar to an urban
environment.
• The MIR of a VOC may be lower than indicated by simple models if the products of its
oxidation have other environmental fates than oxidation in air.
REFERENCES
' G.T. Helms (1989). Definition of VOC: Rationale. Memorandum from USEPA Office
of Air Quality planning and Standards. Research triangle Park, NC.
2 R.A. Rapaport (1988). Prediction of consumer product chemical concentrations as a
function of publicly owned treatment works treatment type and riverine dilution.
Environ. Toxicol. Chem. 7, 107.
3 D.H. Bennett, T.E. McKone, M. Matthies, and W.E. Kastenberg (1998). General
formulation of characteristic travel distance for semivolatile organic chemicals in a
multimedia environment. Environ. Sci. Technol. 32, 4023.
4E. Webster, D. Mackay, and F. Wania (1998). Evaluating environmental persistence.
Environ. Toxicol. Chem. 17, 2148
5 Mackay, D. and Paterson, S. 1991. Evaluating the Regional Multimedia Fate of Organic
Chemicals: A Level HI Fugacity Model. Environ. Sci. Technol. 25: 427-436. Mackay, D.
1991. "Multimedia Environmental Models: The Fugacity Approach", Lewis Publishers
Chelsea, M.I. pp. 1-257. The EQC (Equilibrium Criterion) ModelVersion 2.10
September 1999 is available at
http://www.trentu.ca/academic/aminss/envmodel/models.html
-366-
-------
Pathways Analysis Using Fugacity Modelling of 2-Butoxyethanol for the Second Priority
Substances List (1996). Prepared for Chemicals Evaluation Division, Commercial
Chemicals Evaluation Branch, Environment Canada, by DMER and Angus
Environmental Limited.
7 J. Wooley, W. W. Nazaroff and A. T. Hodgson (1990). Release of ethanol to the
atmosphere during use of consumer cleaning products. /. Air Waste Manage. Assoc.
40, 1114.
8 Emissions of Selected VOC Compounds from the Use of Laundry and Dishwashing
Products Prepared for the Soap and Detergent Association by CH2M HILL May 1994
9 C.J. Weschler and H.C. Shields (1996). Production of the hydroxyl radical in indoor air.
Environ. Sci. Technol. 30, 3250; Weschler C.J. and Shields, H.C. (1997)
Measurements of the Hydroxyl Radical in a Manipulated but Realistic Indoor
Environment. Environ. Sci. Technol. 31, 3719-3722.
'"Inside IAQ, Spring/Summer 1997, pp 1-4, EPA/600/N-97/003.
" J. H. Seinfeld and S. N. Pandis, Atmospheric Chemistry and Physics, John Wiley &
Sons, Inc. 1998, pp 343-4
12M.E. Jenkin, S.M. Saunders and M.J. Pilling - The tropospheric degradation of volatile
organic compounds : a protocol for mechanism development. Atmos. Environ. 31 p31
(1997). The MCM v2.0 is available at http://cast.nerc.ac.uk/LIBRARY/MCM2/.
-367-
-------
CALIFORNIA'S REACTIVITY PROGRAM
Eileen McCauley
California Air Resources Board
This presentation will touch on four aspects of California's reactivity program. First, it
will cover ARB's existing Low Emission Vehicle/Clean Fuels regulation and a consumer
products regulation which is under development. Second, it will provide a very brief summary
of some of the research the California Air Resources Board (ARE) is currently funding in the
field of reactivity. Lastly, it will cover the internal reactivity team and external advisory groups.
Before discussing reactivity, however, it is necessary to provide some background on
California's VOC regulations in general. California has an effective system of VOC control
based on mass reductions. This concept is continued in California's State Implementation Plan
for Ozone which contains adopted and proposed mass-based reductions in VOC emissions for
several different source categories. Thus, it is important that any regulation that uses reactivity
provides equivalent ozone reductions to the reductions predicted from mass-based controls.
However, mass-based controls alone may not be sufficient for California to attain the federal
ozone standard. For this reason, the ARB is investigating possible uses of hydrocarbon reactivity
in our regulations. In addition, reactivity may provide a means for designing control measures
that are more cost effective and provide greater flexibility to the affected industries.
The ARB was the first regulatory agency to enact a regulation that uses reactivity in a
more complex manner than the simple two bin exemption type of regulation. In the late 1980s, a
method was needed to compare the emissions of alternatively fueled vehicles (i.e., ones that use
a fuel other than gasoline, such as compressed natural gas or methanol) to the emissions from
gasoline fueled vehicles. An advisory board was formed, and recommended the use of reactivity
to adjust the weight of emissions so that the limits reflect the ozone-forming potential of the
emissions rather than the simple mass. The Low Emission Vehicle/Clean Fuels regulation,
which was adopted in 1990, uses Reactivity Adjustment Factors (RAF) to set the limits on
vehicle emissions. A RAF is the ratio of the exhaust reactivity of the alternative fueled vehicle
to the exhaust reactivity of the conventionally fueled vehicle. The exhaust reactivity is
calculated by taking the sum of the mass fraction of each compound times the reactivity of the
compound summed over all the compounds in the exhaust.
In designing this regulation, several important scientific issues needed to be addressed.
The first was the selection of a reactivity scale to be used in the regulation. Dr. William P. L.
Carter's Maximum Incremental Reactivity (MIR) scale was chosen by ARB because it was
determined to be the most appropriate reactivity scale to complement California's NOx control
program. The MIR scale is defined in terms of environmental conditions in which ozone
production is most sensitive to changes in hydrocarbon emissions and, therefore, represents
conditions where hydrocarbon controls are most effective. As such, it complements ARB's NOx
control program, which is designed to reduce ozone under conditions that are sensitive to NOx
reductions. Another issue was the degree of uncertainty in the RAFs. Because RAFs are ratios of
reactivities, they are similar to relative reactivities. A number of studies have found that relative
reactivities have much smaller uncertainties than absolute reactivities. Work done by Yang and
-368-
-------
Milford found that uncertainties in RAFs are on the order of 15%, while uncertainties in the
associated MIRs values were closer to 30 to 70%.
ARB is currently developing the California Low Emissions and Reactivity (CLEAR)
regulation for aerosol coatings which would establish reactivity based limits as a voluntary
alternative to the mass-based VOC limits for aerosol coatings. In this proposed regulation, a
product's reactivity would be calculated as the sum of the weight percent of each VOC
multiplied by its MIR value. The regulation is currently in the development stage and two public
workshops will be held in the coming months. The dates for the workshops will be posted at
http://www.arb.ca.gov/consprod/regact/aerocoat/aerocoat.htm. It is scheduled to be presented to
the Board for consideration in the spring of 2000.
In designing a regulatory option that will include reactivity, there are a number of
challenges that need to be met. First of all, it is necessary that the regulation provide equivalent
ozone reductions as the mass-based regulation. It must be enforceable and should be flexible,
simple, and cost effective. As previously stated, there are also a number of specific technical
challenges which need to be addressed. The issue of uncertainty in the LEV regulation has been
discussed. A different approach is being considered for the proposed consumer product
regulation. For that regulation Dr. Carter was asked to assign each compound to a category or
bin based on'the degree to which he thinks the MIR value could change based on future research.
The majority of compounds in the aerosol coating inventory belong to bins which are expected to
change relatively little. The MIRs of compounds in bins with higher uncertainty may be adjusted
to reflect the higher uncertainty.
Another technical challenge that must be met is the development of accurate chemical
speciation profiles. To calculate the reactivity of a mixture, whether vehicle exhaust or a
consumer product, it is necessary to know the identities and quantities of all of the different
compounds in the mixture. The ARB is committed to improving their speciation profiles and is
currently funding research to develop improved profiles for aerosol coatings.
An enormous amount of research has been done to develop the science of reactivity to the
point where it can be used in a regulation with confidence. The ARB is just one of many funding
agencies that have supported research in the fields of atmospheric chemistry, mechanism
development, uncertainty analysis, and the calculation and/or measurement of actual reactivity
values. ARB is currently funding a number of research projects that are related to reactivity.
One such project is investigating the primary products formed from the reaction of OH radical
with three model CIO alkanes: n-decane, 3,4-diethylhexane, and n-butylcyclohexane. Another
project will develop improved methods for evaluating reactivity, as well as methods to assess the
reactivity of "sticky" VOCs that cannot be studied using standard methods. ARB is also funding
a project to determine which environmental chamber parameters contribute most to the
uncertainty in the estimate of reactivity for selected aromatic and oxygenated hydrocarbons.
A related project, which will be discussed at the October 8,1999 meeting of the
Reactivity Scientific Advisory Committee (RSAC), is a peer review of SAPRC99 by Dr.
William Stockwell. SAPRC99, Dr. Carter's latest chemical mechanism, will be used to calculate
the MIR values used in the CLEAR regulation. On the recommendation of the Reactivity
-369-
-------
Scientific Advisory Committee and Dr. Carter a peer review of SAPRC99 was funded. The
purpose of the review was to ensure that the reactivity values in the CLEAR regulation are the
best that are available with the current base of knowledge.
Other research currently underway is investigating the reactivities of several VOCs, such
as cyclohexane, cyclhexanone, octanol, and several other compounds; developing improved
speciation profiles for aerosol coatings; and compiling a public database of potentially useful
solvents with lower reactivities. Lastly, ARE is funding a study to determine reactivity values
using an urban airshed model and compare the calculated reactivity values to the MIR values.
The third aspect of ARB's reactivity program is an internal reactivity team. The team
consists of staff from several different divisions and has the goal of investigating how reactivity
may fit into existing and future board programs. The team reviews the effectiveness of ARB's
current uses of reactivity, examines the technical basis for quantifying reactivity, and provides
recommendations to the Executive Officer regarding the use of reactivity in regulatory programs.
Members of the team recently presented two papers at the Air Pollution 99 conference:
Assessment of the organic compound reactivity concept for regulatory applications in California
and Photochemical reactivity of organic compounds in central California: A grid-based
modeling study. Copies of either of these papers can be obtained from Dr. Ajith Kaduwela, team
leader, at akaduwel(a),arb.ca.gbv.
The last aspect of ARB's reactivity program is a mechanism to benefit from the expertise
of scientists and stakeholders outside the Board. ARB has formed two advisory committees, the
Reactivity Scientific Advisory Committee (RSAC) and the Reactivity Research Advisory
Committee (RRAC). The RSAC consists of six noted experts in the fields of atmospheric
chemistry: John Seinfeld, Roger Atkinson, Jack Calvert, Harvey Jeffries, Jana Milford and Ted
Russell. This committee reviews issues concerning reactivity and makes recommendations to
the Board, based on their knowledge of the science. The RRAC is comprised of over 20
representatives from industry. These committee members provide technical assistance on
reactivity related issues for consumer products and aerosol coatings. For example, the RRA.C
provided valuable suggestions which helped guide ARB's research on MIR values for
compounds found in consumer products.
In conclusion, the ARB believes that reactivity-based regulations, when properly
designed to provide ozone reduction that is equivalent to that provided by mass-based
regulations, can be an effective ozone control strategy while offering compliance flexibility to
affected industries. A large body of scientific information is available to support the
development of these reactivity-based regulations. However, to further refine the effectiveness
of this ozone control strategy, the ARB will continue to support additional research to increase
the understanding of hydrocarbon reactivity. The California Air Resources Board believes that
hydrocarbon reactivity has the potential to add significantly to its ability to protect public health
and welfare through the effective and efficient reduction of air pollution in a cost-effective
manner.
-370-
-------
Programme of Control Concepts and Measures for Ozone
(presented by the Federal Environmental Agency, Germany)
The German Programme of control concepts and measures for ozone is a result of a
research project funded by the German Federal Environmental Agency. The central issue
was the investigation of the effects of several measures to reduce ozone precursor
substances on emission levels and ozone concentrations. The whole title of the project,
elaborated by scientists of the Prognos enterprise in Basel as well as of the Meteorological
Institutes of the Free University Berlin and the University Colon, was: ..Determination and
evaluation of the effects of regional-scale and of local emission control measures on
elevated ground-level ozone concentrations in mid-summer episodes".
The project was divided into two different parts. The first part included the analysis of the
efficiency of measures at a regional, nation-wide level. This analysis was performed by using
the photochemical model systems EURAD and REM3 with meso-scale grids. The
corresponding modelling domain covered most parts of Europe with horizontal grid distances
of 30 km.
The subject of the second part was the investigation of the efficiency of measures at a local
level. This analysis was performed for three selected regions in Germany, Berlin, Dresden
and Rhein/Main, using the photochemical model systems EURAD and CALGRID/REM3 with
a higher spatial resolution. Horizontal grid distances of 2 km were applied to the model
simulations.
A broad variety of scenarios was investigated in this study. The scenarios can be divided into
long-term and temporary scenarios. Moreover scenarios for Europe, Germany and the three
regions in Germany were considered.
The scenario Trend 2005 includes emission reductions of NOX and VOC, fully achieved in
2005 by control measures which are already implemented due to existing European and
national Directives. This scenario was applied to the entire, large modelling domain (most
parts of Europe).The Reduction 2005 scenario is basically the same like Trend 2005, but for
Germany further reduction potentials concerning traffic, industry and solvent emissions are
added. Both scenarios are long-term scenarios.
Other control scenarios are temporary scenarios applied nation-wide and local to the three
selected regions. These scenarios include various restrictions like speed limits, driving bans
for cars without a regulated 3-way catalytic converter and for non-low emission diesel cars.
Another type of scenarios includes a combination of several measures. It is connected with
speed limits, other driving restrictions, temporary closure of large emission sources as well
as public appeals to business and households. These measures are assumed to be applied
in the three local regions and adjacent areas, that means in adjacent federal states.
Altogether 8 scenarios in the regional scale and nation-wide and 10 scenarios in the three
areas in Germany were considered.
The most effective scenarios in emission reductions of NOX and VOC are Reduction 2005
and the combination scenario (with a variety of measures in local regions and adjacent
areas). These two scenarios result in a reduction of NOX emissions of nearly 65 % and in a
-371-
-------
reduction of VOC emissions between 60 to 70 % (these numbers are related to Germany for
Reduction 2005 and to the corresponding region for the combination scenario).
The reduction rates concerning Trend 2005 for Germany and for the small, local areas are
similar. NOx reductions between 20 and 30% and VOC reductions around 45 % can be
achieved. Driving bans result in similar emission reductions like Trend 2005.
A temporary measure which is less effective are speed limits. NOX reductions between 1C)
and 15 % can be found. There is no significant effect on VOC emissions.
Looking at the effect of regional and local measures on simulated hourly ozone
concentrations it can be found that the effect of speed limits is the smallest. Driving bans
are a little bit more effective than speed limits.
The highest ozone reductions result for Reduction 2005 which is a long-term_scenario
applied to a larger region.
Obviously two basic results of the research project become evident: First, all regional
measures are more effective than their corresponding measures at the local level. Second,
long-term measures seem to be more effective than temporary measures.
To come to the conclusions the research project has shown that the long-term and
European-wide applied scenarios Trend and Reduction 2005 are most effective. Provided
that these scenarios are fully implemented a reduction of peak ozone concentrations by 40
% can be achieved. Temporary and local measures are less effective in decreasing high
ozone levels. For each single measure, the decrease of peak ozone concentrations would be
at most 5 %. Only the combined application of all possible local measures results in a
reduction of NOX and VOC emissions of approximately 50 %. Then, the corresponding
decrease of peak ozone concentrations would reach at most 20 %.
Long-term control measures implemented on a large region (national, international) are even
with lower overall percentage emissions reduction more effective in reducing high ozone
concentrations than temporary and locally restrictive measures.
Further information about the results of this research project can be found in the internet.
The address is: www.umweltbundesamt.de/ozon-e
-372-
-------
Session VI
New Chamber Projects
Session Chair
Bill Carter
-------
Design of SAPHIR
(Simulation of Atmospheric PHotochemistry In a large Reaction Chamber)
Andreas Wahner
Institut fur Atmospharische Chemie, Forschungszentrum Jiilich, D-52425 Jiilich
A large daylight atmosphere simulation chamber for the investigation of tropospheric
photochemistry under natural conditions (SAPHIR) is currently being build at the
Forschungszentrum Juelich. The main scientific issues planned to be investigated in this chamber
are:
1. Which chemical precursors and physical parameters determine the concentration of the
OH- radical ?
2. How do certain parameters (trace gas concentration, solar radiation) influence the degradation
of important trace gases ?
3. Which parameters influence the production and destruction of photoxidants (Ozone) and other
intermediates?
The design requirements, the actual setup, and the range of the highly specialized radical and trace gas
detection experiments will be presented.
-374-
-------
A Research Plan For a New Environmental Chamber Facility
by
Rafael Villasenor
Mexico City's air pollution is a persistent and pervasive environmental problem that
has imposed health and economic costs on society. With Mexico's expenditures on controls
calculated in several millions of pesos per year and slow progress toward compliance with
air quality standards, a critical need exits for better and more effective abatement strategies.
With the current Federal Environmental Policy on fossil fuel burning for power utilities and
industrial processes shifting from heavy fuel oils to low sulfur natural gas, it becomes
important to characterize industrial emissions as well as residential emissions that use
gaseous fuels for establishing their impact on the Mexican airshed. Concurrently, it is
recognized that an efficient strategy to abate air pollution in Mexico City consists of
improving the properties of Mexican liquid fuels for transportation. Although liquid
hydrocarbons have been substantially improved to comply with environmental regulations
not much has been done to evaluate exhaust gas emissions and their role in ozone and
aerosol formation.
The need to address the above-interrelated topics demands great efforts in reducing
air pollution. Common obstacles in this pursuit include insufficient understanding of the
relationship among the underlying scientific, economic, and social issues, and also in
addressing the problem with limited availability of resources, and infrastructure. Although
addressing all these issues collectively is out of the scope of the present investigation, much
can be accomplished by devising more prominent scientific methodologies and
experimental techniques that can shed new light in mitigating and controlling air pollution
in Mexico's populated cities. An integrated assessment will provide the opportunity to
understand the air pollution problem and its coupling to regional and global impacts.
Mexican researchers at the IMP are initiating multidisciplinary work in which the city's
environmental problems are being addressed. The major objective of this investigation is
aim at acquiring scientific knowledge to address the still prevalent high concentrations of
fine particles and ozone formation in the Mexico City Metropolitan Area (MCMA).
A two-step approach is devised to carry out the objectives of this investigation. The
first task relies on experimentation to gather enough information that can be used later on
for tuning parameters and refining chemical reaction mechanisms in ozone formation
modules embedded in advanced urban and regional models for regulatory purposes.
Modeling of important episodes and historic events will be a key element in evaluating the
impact of emission sources into the air. During the experimentation phase smog chambers
will be used for determining gas precursor contribution from both gaseous fuels and tail
pipe emissions from combusted liquid hydrocarbon fuels on ozone formation. This phase
will characterize Volatile Organic Compound (VOC's) reactivates including exhaust gas
and evaporative liquid fuel as well. The adopted methodology will be based on reactivity
indexes to estimate VOC's reactivity in controlled atmospheres inside radiated
environmental chambers. The incremental reactivities of a series of representative VOC's
-375-
-------
are expected to differ from those measured by Carter in previous studies since the VOC to
NOx ratio, VOC's speciation and radiation fluxes typical of the Mexican airshed vary
considerably from that of northern latitudes. In this line of work computer modeling of
environmental chamber measurements of incremental reactivity of VOC will be pursued in
order to evaluate detailed atmospheric photochemical mechanisms for developing ozone
reactivity scales for VOC's.
The present experiments will complement studies under high NOx maximum
reactivity conditions and enable our research group to gain sufficient knowledge in this area
of expertise. The scope of the present work is not only centered on the design and
characterization of smog chambers but also to adopt existing technologies to carry out the
activities herewith described and to study heterogeneous processes related to chamber wall
effects.
Another branch of research that the project contemplates is aerosols. Fine particle
formation within the MCMA when strong photochemical activity takes place severely
affects not only the urban airshed but also air quality of downwind rural areas. In the
Mexico City basin airborne particles contribute significantly to light scattering and
absorption and hence, visibility reduction, which is related to both light-particle interaction
and gas-to-particle processes, is drastically hinder by the intense human activity of the
region. The complex heterogeneous chemical reactions between gaseous pollutants and
suspended particles that occur during aerosol formation are not well understood. Further
information is necessary to explain aerosol formation as favored through photochemical
reaction of gases and vapors in semiarid atmospheres. Although the mechanisms
responsible for inorganic aerosol formation seem to be well characterized, very little is
known on the chemistry of organic aerosols. Mexican authorities have recognized the
urgency to address issues relevant to visibility reduction and aerosol formation in order to
improve Mexico's air quality. To resolve such problems a better understanding of the
complex heterogeneous atmospheric chemistry is required. It is therefore of paramount
interest for researchers at IMP to initiate chamber studies of aerosol formation and to
actively participate with other research groups in related projects. This collaborative effort
will fortify common areas of research and will enable us to provide the answers that are
needed to mitigate pollution in megacities.
A key element distinguishing this project from other air pollution research is its
multidisciplinary nature and the myriad areas of research that it encompasses. Air quality
modeling is the focus of the second phase of this investigation. Advanced air quality
models will be used to simulate photochemisty along with transport and formation of
aerosols. Models will be developed to study the formation of inorganic and organic
aerosols. Chemical mechanism development and evaluation for low vapor organic
compounds that are formed in photochemical atmospheres are considered as a fundamental
part of the investigation. Gas-particle equilibrium models for VOC's are fundamental to
investigate the gas to particle pathways and particle growth that characterize aerosol
formation in the basin's atmosphere. There is also a great need to study VOC accumulation
on soot particles, as the number of sources that emit soot is considerably large in.the region.
To determine the origin, relative contribution and distribution of particles and volatile
organic compounds receptor models will be used. Receptor models will aid in
-376-
-------
characterizing the VOC sources while providing a tool to evaluate the effectiveness of
pollution control strategies.
The impact of aerosols on the radiation budget will be evaluated as a first estimate
on regional climate change. Linking the local and regional air quality issues that are most
important to local decision-makers to factors affecting global change, the proposed
assessment offers a new opportunity to advance effective polices on both fronts.
With the scientific platform on which this project rests air quality models will be
successfully used to estimate future scenarios that may arise as a result of a number of
modifications or implementations on a regional scale. For instance a shift form residual fuel
oils to natural gas applied on the industrial sector. Improved liquid fuels introduced into the
market for transportation. Addition or removal of various types of emission sources hi the
area as well as the effect of adding air pollution control systems for power generation units,
or other commercial combustion applications. The project contemplates the simulation of
different scenarios to evaluate the ambient impact of aerosol when using reformulated fuels
according to a refinery system upgrade. The complementary tasks to support the project's
activities in the second phase, such as reactivity indexes and the development of explicit
reaction mechanisms will derive from a new-planed experimental facility designed for
high-volume environmental chambers.
-377-
-------
RESEARCH PLAN FOR "NEXT GENERATION ENVIRONMENTAL
CHAMBER FACILITY FOR CHEMICAL MECHANISM AND VOC
REACTIVITY EVALUATION"
William P. L. Carter
College of Engineering Center for Environmental Research and Technology,
University of California, Riverside, CA 92521
A. Introduction
A critical component for predictions of formation of ozone and other secondary pollutant
formation in the atmosphere is the chemical mechanism, i.e., the portion of the airshed models
used to represent the chemical reactions involved. This is because the chemistry is the source of
much of the complexity and non-linearity involved. Because many of the chemical reactions are
incompletely understood, these mechanisms cannot be relied upon to give accurate predictions of
impacts on emissions on air quality in the atmosphere until they have been shown to give accurate
predictions of pollutant concentrations under realistic but controlled conditions. The most cost-
effective and reliable way to test the accuracy of the chemical mechanisms is to compare their
predictions against results of environmental chamber experiments that simulate the range of
conditions in the atmosphere. If a model cannot accurately predict observed changes in pollutant
levels in such experiments, it cannot be expected to reliably predict effects of proposed control
strategies on ambient air quality.
As discussed by Dodge (1998), the current chamber data base has a number of serious
limitations and data gaps that could be limiting the accuracy of the mechanisms used in the
models to predict control strategies. Uncertainties exist concerning characterization of chamber
conditions, particularly how wall artifacts affect the gas-phase reactions (Carter and Lurmann,
1990, 1991), and inappropriate treatment of these effects could cause compensating errors in the
gas-phase mechanism (Jeffries et al, 1992). Most chamber experiments lack measurement data for
important intermediate and product species, limiting the level of detail to which the mechanisms
can be evaluated, and limiting the types of air quality impact predictions which can be assessed.
Furthermore, because of chamber background and wall effects, and because of inadequate
analytical equipment currently available at environmental chamber facilities, the current
environmental chamber data base is not suitable for evaluating chemical mechanisms under the
lower NOX conditions found in rural and urban areas with lower pollutant burdens. Because of
this, one cannot necessarily be assured that models developed to simulate urban source areas with
high NO* conditions will satisfactorily simulate downwind or cleaner environments where NOX is
low.
To address the need for improved an improved environmental chamber facility to
evaluate mechanism for O3 and PM formation, the College of Engineering, Center for
Environmental Research and Technology (CE-CERT) recently received funding to develop a
"Next Generation" environmental chamber facility for chemical mechanism evaluation and VOC
reactivity assessment. The objectives of this project are to develop the environmental chamber
facility needed for evaluating gas-phase and gas-to-particle atmospheric reaction mechanisms, for
determining secondary aerosol yields, and for measuring VOC reaction products and radical and
NOX indicator species under more realistic and varied environmental conditions than previously
has been possible. The facility will then be employed to provide data that are most relevant to
today's pollution problems and control strategy issues. The project involves a four-year program,
-378-
-------
with the first one to two years being for research on chamber design, facility development, and
chamber characterization and evaluation. The remainder of the program will involve conducting
experiments needed for model evaluation and to address issues of relevance to regulatory
assessment and control strategy development. These would include, but not necessarily be limited
to, the following:
• Determining whether current predictions of effects of VOC and NOX changes on ozone
and secondary aerosol formation are applicable to lower pollutant concentrations.
• Assessing differences among VOCs in terms of effects on ozone, secondary aerosol
formation, and other pollutants under low-NOx conditions. Current ozone reactivity
scales (e.g., Carter, 1994) for VOCs were developed for more polluted urban conditions
and may not be appropriate for lower NOx environments.
• Providing information needed to evaluate whether control strategies aimed at replacing
reactive VOCs with less reactive but more persistent compounds may adversely affect
ozone or other pollutants when they are transported downwind.
• Determining major oxidation products formed by organics when they react under low-
NOx conditions. This is important to developing scientifically-based models for low-NOx
reactions of VOCs, as well as to understanding the ultimate environmental fates and
impacts of these compounds, which in some cases may affect global climate change.
• Determining the effects of temperature on secondary pollutant formation and VOC
reactivity. Current environmental chamber facilities are not adequate to evaluate these
effects, but limited studies of temperature effects indicate that temperature effects are
probably significant.
• Determining the effects of temperature and humidity on secondary organic aerosol
formation from various VOCs. The results will be compared with data obtained using
outdoor chamber systems to evaluate the range of applicability of those data.
• Evaluating the budgets of HOX and NOy, and evaluating the usefulness of indicators of O3
and P(OX) sensitivity to precursors for conditions typical of ambient atmospheres.
• Evaluating impacts of various types of VOC sources, such as architectural coatings, on
formation of ozone, secondary PM, and other pollutants in various environments.
• Utilizing the facility to test equipment to be used for monitoring trace pollutants in
ambient air under controlled conditions where the actual pollutant concentrations, and the
history and source of the air mass being monitored, are known.
This project involves a collaboration with Dr. John Seinfeld at the California Institute of
Technology (Caltech), whose group is established as representing the state-of-the-art in using
environmental chambers to assess PM formation. As such, they complement and enhance the
group at CE-CERT, which we believe represents the state-of-the-art for using environmental
chambers for evaluating gas-phase mechanisms and assessing effects of VOCs on ozone
formation.
The specific approach that will be employed in this project will be determined as part of
the development of the research plan, which is still underway. The overall approach that was
presented in the proposal for this project is discussed below. The discussion is somewhat general
because the specifics of the research plan are still being developed. This plan will evolve as the
project is conducted and as the capabilities of the facility are determined.
-379-
-------
B. Facility Development and Evaluation
1. Evaluation of Approaches to Minimize or Control Chamber Effects
Other than analytical limitations, the main factor limiting use of environmental chambers
for mechanism evaluation at low pollutant conditions is chamber wall effects. Known chamber
effects that are taken into account during mechanism evaluation include the chamber radical
source (which is believed to be due at least in part to HONO absorption and offgasing), ozone
wall losses, NOX absorption and offgasing, N2O5 hydrolysis, excess NO to NO2 conversions
attributable to background VOC contamination, etc. (Carter and Lurmann, 1990, 1991; Carter et
al, 1995a; Gery et al, 1989). Of these, perhaps the most serious factor limiting utility of chamber
data under low NOX conditions is NOX offgasing effects, while background VOC contamination
would limit the utility of data under low VOC conditions.
Before finalizing the design and specifications for the chamber, we will investigate
approaches to minimize these effects or at least make them predictable and reproducible. Most
chambers currently used for mechanism evaluation consist of heat-sealed FEP Teflon reaction
bags, which have good UV transmission characteristics and are relatively inert. Previous
evaluations have indicated that metal is unsatisfactory, and work in our laboratories and
elsewhere indicated that chambers constructed of glass (unpublished results from this laboratory)
or Teflon coated metal (e.g., the SAPRC evacuable chamber [Carter et al, 1982, 1995a]) have
higher radical sources. Although comprehensive research may reveal that there is a superior
surface material than Teflon film, we suspect that there will probably not be time (or funds) to
carry out such comprehensive research and still have an operational chamber within the desired
time frame, which is about 1 - 1 !/2 years. Therefore, it is expected that the chamber will be
constructed of FEP Teflon film.
Assuming that this will be the case (though use of alternative materials is not ruled out),
the problem becomes how to treat or clean FEP Teflon film to minimize chamber effects and
make them predictable and reproducible. In our laboratory and at Caltech, the normal procedure
has been to flush the reaction bags between runs, carry out periodic characterization experiments
(e.g., CO — NOX and n-butane - NOX runs to measure the radical source, and pure air or
acetaldehyde - air experiments to measure NOX offgasing effects, and standard propene - NOX
runs) and replace them if anomalous characterization results indicate contamination. Although
this yields sufficiently well characterized chamber effects for reactivity experiments with NOX
levels above ~50 ppb and sufficient VOC for ozone formation, this will not be suitable for the
purpose of this program.
The approach that was employed at the TVA chamber (Simonaitis and Bailey, 1995) was
to flush the chamber for a relatively long period of time. This approach is occasionally used in
our facility, and it is often found to be successful in reducing contamination effects when
anomalous results are seen after exposure of the chamber to unusual reactants. Results of the
limited relevant characterization data given in the TVA reports indicate that this results in
significantly lower apparent NOX offgasing rates than observed in our chamber. However, an
important issue in terms of chamber productivity is the amount of flushing time required to
achieve satisfactory results. If long flushing times are required, a multi-reactor chamber facility
will be constructed to allow experiments to be carried out in some reactors while others are being
flushed.
This and other approaches for cleaning will be evaluated using small (~50-100 liter)
Teflon reaction bags, where contamination effects will be much more evident. Examples of other
-380-
-------
approaches which might be tried might include treatment with water to remove absorbed NOX and
other materials, followed by flushing with light, or perhaps even treatment with F2. Small reactors
will also be used to examine effects of temperature variation, since the chamber is expected to be
operated at a range of controlled temperatures. Obviously, any approach that will be employed
will have to be scalable to a large volume chamber at a reasonable cost.
Experience in our laboratory with a ~ 12,000-liter chamber suggests that simply
increasing the volume of the chamber can be a major factor in reducing chamber effects such as
NOX offgasing rates (Fitz et al, 1998). However, the extent to which the lower apparent chamber
effects in that reactor may be due to other factors such as use of newer reaction bags or higher
light intensity. Our existing larger chambers, which range in volume from ~1000 to ~12,000
liters, can be used to investigate the effects of chamber size on chamber effects and efficacy of
cleaning and conditioning processes, if appropriate. In any case, any cleaning or conditioning
process which gives good results for ~ 100-liter reaction bags should work at least as well in
larger chambers.
2. Chamber and Facility Design and Fabrication
Based on the results of the evaluation of prototypes and other tests, the design of one or
more larger research chambers will be developed. It may well be that more than one chamber
may be necessary, especially if it is found that flushing for long periods of time is necessary to
control chamber effects. Separate chambers may also be used for different research objectives,
such as simulations of nearly clean air, measurements of particle formation under more polluted
conditions, or assessments of VOC reactivity under more urban-like conditions using the
facilities' advanced analytical instrumentation. It is expected that chambers for low NOX and
particle research will need to be of large volume to minimize surface effects, and also to permit
analyses using instrument with high sample flow requirements. The latter will be particularly
important when using the chamber to test instrumentation designed for field use.
Another design goal would be to have a chamber that can be operated in multi-day
simulations without significant dilution or loss of volume, and without buildup of wall effects.
Multi-day experiments will be critical for simulations of rural or long range transport scenarios.
Utilizing highly purified air and avoiding outside contamination is also critical.
Exploratory experiments in our laboratory using procedures to improve air purification have
shown that this reduces some effects that were attributed to chamber walls. To avoid introduction
of contaminants from laboratory air into the chamber (through leaks or diffusion through the
Teflon walls), the reactors will be located in a "clean room" which itself is flushed with purified
air.
The first construction priority will be a high capacity air purification system that is
capable of removing CO as well as NOX and organics and other pollutants to the lowest practical
level. Removal of methane is a lower priority, but if feasible that will also be desirable. Removal
of methane, CO, and CO2 would be desirable if feasible because it would permit use of simplified
(and therefore more reliable) methods to measure total reactive carbon present. High capacity is
essential because the system will be used for flushing the clean room that will house the reactors,
as well as multiple large volume reaction bags. Reliability will also be a priority, as well as
reasonable cost. Various air purification approaches have been used at different facilities, and an
initial evaluation has been carried out in the process of developing budget estimates for this
proposal. One possibility that shows potential is the use of evaporated Qz and N2, recombined to
form synthetic air. This has the advantage of removing methane and CO2 as well as other species.
-381-
-------
Although the operating costs of this approach will be higher than an air purification system, the
initial costs may be lower, and the reliability will probably be significantly greater.
Another design issue will involve temperature control. A temperature control system will
be constructed with the design goal of holding the chamber enclosure at a constant and controlled
temperature between ~5° C or lower and -50° C or higher, to within + 0.5° C. An evaluation of
available options is needed to indicate what is feasible within the budget for this proposal.
Lighting is also an important consideration. The chamber will be constructed indoors
with artificial lights to allow for complete temperature and lighting control and adequate
characterization of chamber conditions. Xenon arc lights probably will be employed because
among the options which have been used for indoor chambers, they provide the most realistic
spectrum (Carter et al, 1995b). Although blacklights are less expensive and give an adequate
spectrum in the UV, they would be unsatisfactory for a temperature-controlled chamber because
their output is affected by temperature. However, blacklights may be employed for lighting the
reaction bags for cleaning and flushing, should that approach be employed for chamber cleaning
and conditioning.
It has not been decided exactly where the chamber facility and associated analytical
laboratory will be located within CE-CERT, since there is insufficient space in the room currently
housing the CE-CERT Atmospheric Processes Laboratory (APL) for this purpose. A space has
been identified in the high bay area near the APL which may be sufficient for this purpose.
However, it may be less costly to place the facility housing the chamber outdoors, which would
allow for more options in terms of space and configuration. An existing modular building could
be used to house at least some of the equipment, though additional structures to house the
chamber enclosure and its associated lights and temperature control system would have to be
constructed. The current budget estimate is based on assuming the latter option, which is believed
to be more cost effective, and whose cost is easier to estimate.
3. Analytical Instrumentation
An essential component of any environmental chamber facility is the instrumentation
used to measure trace pollutant levels. Much of the needed instrumentation is the same as that
needed to support experiments for ozone model evaluation at higher NOX conditions, such as
ozone and CO monitors, gas chromatographs for monitoring hydrocarbons, etc. In addition, CE-
CERT already has a tunable diode laser system for monitoring trace levels of NO2 and HNO3.
However, significant additional instrumentation would be needed to support the objectives of this
program, most of which are not presently available at current environmental chamber facilities in
the United States. Since acquisition of analytical equipment will be limited by the budget for this
program, an important subject of the initial planning workshop will be to determine the priorities
of advanced equipment needed. However, the priorities for instrumentation acquisition are
expected to be as follows.
Instrumentation for Gas-Phase Analyses:
• High sensitivity NO - NO. analyzers. This will be the first priority for instrumentation
acquisition because it will be needed in the experiments to evaluate minimizing chamber
effects and for evaluating the air purification system. Our laboratory has analyzers which
may be satisfactory for this purpose, though the need for additional or more sensitive
analyzers will have to be evaluated.
-382-
-------
High sensitivity total carbon. NMHC, and CO analyzers. These will also be priorities for
evaluating the air purification system. Our laboratories have such analyzers, but they are
not sufficiently sensitive. Note that if methane and CO2 can be removed from the
background air, a relatively simple total carbon analyzer can be used to measure NMHC
with reliability and without need for calibration, by converting everything to methane and
detecting the methane by flame ionization detection.
FID Gas Chromatographs for Monitoring Organic Species. GC/FID provides the most
sensitive, accurate, and precise analysis available for hydrocarbons and other low to
moderate volatility VOCs that can be analyzed by this method. Our laboratory already
has several such instruments, where loop sampling is used for more volatile or less
"sticky" compounds, and Tenax cartridge sampling is used for others. The present
systems at our facility will probably be sufficient for the initial evaluations, but additional
GC/FID systems will be needed once the facility becomes operational. An automated
sampling and concentration system will be acquired to maximize sensitivity, productivity
and data precision.
Tunable Diode Laser Systems (TDLAS) can be used for analysis of NO2, H2O2, HNO3,
formaldehyde, and other difficult to monitor trace species. Monitoring "true" NO2 is
particularly important, and because of interferences commercial NO - NOX analyzers are
not suitable for this purpose. Our laboratory presently has a TDLAS system that has been
used successfully to monitor NO2, though it uses somewhat older technology and lacks
the reliability for routine use. Although the present system will probably be satisfactory
for evaluation of chamber effects, additional systems for monitoring other species (whose
priorities will be determined subsequently), or a replacement to the existing system, will
probably need to be acquired later in the program.
A Differential Optical Absorption Spectrometer (DOAS) provides a means to monitor
nitrous acid, NO3 radicals, formaldehyde, glyoxal, certain aromatic compounds, and other
species with highly structured UV absorption spectra. It provides the most sensitive
available analysis for nitrous acid, which is believed to be a key species affecting
chamber radical sources and may be important in initiating radical formation when lights
are turned on in multi-day simulations. It also provides the only known means to monitor
NO3 radicals, a key nighttime species, under atmospheric conditions. Because of its
ability to monitor nitrous acid, a DOAS instrument will be useful for chamber
characterization, and will probably be a priority for early acquisition in this program.
A Gas Chromatograph / Mass Spectrometer (GC/MS) will be a priority once we begin
experiments with test compounds because it provides a means to identify as well as
quantify organic compounds which can be monitored by gas chromatography. Its primary
use will be organic product identification. It is expected that a mass selective detector
(MSD) will be employed, though the final decision concerning the instrumentation
options will be determined later in the program.
A HPLC System will also be a priority once we begin experiments with test compounds
because it provides a means to identify and quantify certain aldehydes and other
compounds that are not suitable for GC analysis. It will also be needed for analysis of
constituents of organic aerosols in experiments where these are collected.
BCD Gas Chromatographs for Monitoring PANs. Organic Nitrates, and Other Species.
GC/ECD provides a means to routinely monitor PAN analogues and monitors other
nitrogenous or halogenated compounds with high sensitivity. Although our laboratory has
a GC/ECD system, its performance is not satisfactory for quantitative work. Options for
an improved system will be investigated.
-383-
-------
• A Fourier Transform Infrared (FT-IR) system provides a means for monitoring trace
species that cannot be monitored in other ways. It is a lower priority for acquisition for
this program because it lacks sufficient sensitivity to be useful for studies carried out
using ambient or lower pollutant levels. For that reason, acquisition of a FT-IR system is
not in the budget for this program. However, it may turn out to be desirable for special
studies, verifications, or calibrations. Therefore, the possibility that the priority for
acquisition of this system for this program cannot be ruled out at this time.
Aerosol Instrumentation:
• An Ion Chromatograph will be used to analyze paniculate matter for anions following
extraction with an aqueous solvent.
• An Organic Carbon Analyzer will used to determine the organic content of paniculate
matter collected from the chamber using high purity quartz filters.
• A TSI model 3320 Aerodynamic Particle Sizer will be used to determine the particle size-
number distribution for particles greater than 0.3 urn aerodynamic diameter.
• TSI 3080L Electrostatic Classifier w 3080 Controller. 3077 Neutralizes 3081 LongDMA.
TSI 390087 SMPS Interface and Software. TSI 3010S Condensation Particle Counter
with fast scan EPROM. This combination of instruments will provide particle size
number distribution from 0.003 to 0.5 um aerodynamic diameter.
• TSI 3076 Constant Output Atomizer. TSI 3062 Diffusion Dryer. TSI 3077 Aerosol
Neutralizer. This combination of equipment will be used to add a synthetic seed aerosol
to allow SOA to condense on existing aerosol rather than self-nucleate.
It is important to recognize that organic photooxidation products that are expected to be
formed under low-NOx conditions will be different than thpse formed under higher-NOx
conditions, and among these low-NOx products there will be unstable and difficult to monitor
species such as hydroperoxides. One high-priority focus of the analytical development for this
program will be to develop and evaluate methods suitable for monitoring hydroperoxides and
other low-NOx organic product species. Where necessary, the participation of synthetic organic
chemists will be utilized to prepare authentic samples for analytical development and calibration.
Once the equipment needs are identified, the necessary equipment will be ordered and the
appropriate procedures will be evaluated for utilizing and calibrating them. They can be evaluated
utilizing the existing chambers available at CE-CERT, and intercompared with data from other
relevant instrumentation where applicable. A quality assurance plan for measurement data will
also be developed and evaluated during this period. This analytical development can be done on a
time frame parallel to the design, construction, and characterization of the new environmental
chambers.
4. Chamber Characterization and Evaluation
Once the chamber(s) are constructed, appropriate characterization experiments will be
carried out to evaluate their performance and characterize them for use for model evaluation. The
types of characterization experiments employed for this purpose are discussed elsewhere (e.g.,
see Carter and Lurmann, 1990, 1991; Carter et al, 1995a), and will include, but not necessarily be
limited to, the following:
• Light Intensity is measured using actinometry experiments and by other methods. The
NOa photolysis rate can be measured at various positions in the chamber enclosure using
the quartz tube method developed by Zafonte et al (1977), modified as discussed by
-384-
-------
Carter et al (1995a). The net NO2 photolysis rate within the reaction bag can also be
measured using the steady state method, based on the photostationary state between NO,
NO2, and O3 in the absence of other reactive compounds (Carter et al, 1995a,b). An
alternative method that has proven to be more reliable for measuring overall photolysis
rates in the chamber is to measure the C12 photolysis rate by measuring the rate of
consumption of n-butane in the presence of C12. Relative light intensities on various
surfaces within the enclosure can be measured using our LiCor Li-1800
spectraradiometer, which is also used to measure the spectrum of the light source, and
radiometers sensitive to various spectral regions can also be employed. These or other
appropriate methods may be employed for this purpose.
Temperature Characterization experiments will be conducted to evaluate the performance
of the temperature control system at various temperatures within its range of control. This
will include experiments with and without the lights, and measurements taken at various
points within the chamber enclosure.
Pure Air Experiments will be used to measure background effects such as NOX offgasing
and the presence of other reactive species in the matrix air or offgased from the walls.
Ozone formation in such experiments is highly sensitive to NOX offgasing (or NOX in the
matrix air), but is also sensitive to background VOCs (or CO) which can enhance ozone
formation by converting NO to NO2. Because O3 formation in pure air runs is sensitive to
several factors, pure air runs are best used in conjunction with other types of experiments,
which have differing sensitivities to these factors. Monitoring for trace species can also
indicate if such species are offgased from the walls during photolysis.
Aldehyde - air or VOC or CO - air experiments can also be used to measure NOX
offgasing and are useful in conjunction with pure air runs because they are much more
sensitive to other background effects. Formation of ozone or (in the case of aldehyde - air
runs) PAN provides a sensitive measure of NOX offgasing.
NOv_- air experiments are sensitive to the presence of background (or offgased) VOCs,
and thus provide another useful complement to pure air runs. The rate of NO
consumption provides a measure of the rate of NO to NO2 conversion caused by reactions
of background VOCs. They are also sensitive to chamber radical sources, and have been
used, with reactive organic tracers present to monitor OH radical levels from their rates
of consumption, to measure the chamber radical source (Carter et al, 1982). However,
subsequent analysis indicates that this method tends to give overestimates of the chamber
radical source in chambers where the radical source is low (Carter et al, 1995b), which is
expected to be the case for the chambers developed for this program. NOX air experiments
are best used in conjunction with other experiments used to measure the radical source,
so the rate of NO consumption in the NOx-air experiment can provide a more
unambiguous indication of the levels of background VOCs.
Alkane (usually n-Butane) - MX or CO - NO, have been found to provide the most
sensitive and reliable method to measure the chamber radical source. The rates of NO to
NO2 conversion in these experiments are highly sensitive to the radical source, and the
other aspects of the mechanisms that affect these observations are well characterized.
Such experiments can also be used to indicate whether the injected NOX is contaminated
by nitrous acid, which can occur if improper NOX injection procedures are employed. If
HONO is present initially, it will be indicated by relatively high NO to NO2 conversion
rates during the initial periods of the experiments.
Dark Decay Experiments are used to measure wall losses of species, such as ozone and
nitric acid, which tend to be destroyed or absorbed on surfaces. It will also be used to
evaluate whether compounds which are not expected to be lost on surfaces, such as
-385-
-------
hydrocarbons and (at least under dry conditions) formaldehyde and other oxygenates
indeed remain in the gas phase in this chamber. Dark decay experiments with low levels
of NOX will also be conducted to assess whether these species also go to the surfaces.
• Smog Simulation Control Experiments, such as propene - NOX or surrogate - NOX
experiments, will be carried out to assess reproducibility, consistency with results in other
chambers, equivalency of results in different reactors, and consistency with model
predictions for well characterized chemical systems. Such experiments are also useful for
evaluating analytical methods because the amounts of various species formed are known
or can be simulated using well evaluated mechanisms. A new set of standard control
experiments will be developed to assess reproducibility and consistency of results of
experiments at low reactant concentrations. Results of experiments using known or
previously studied systems will be compared with previous data and with model
simulations. Any discrepancies or unexpected results will be fully investigated prior to
proceeding with the remainder of the experimental program.
The above experiments are necessary for characterizing the chamber for evaluating gas-phase
mechanisms. However, since this chamber will also be used for assessing aerosol formation and
evaluating models in this regard, additional experiments are needed to characterize the chamber
for this purpose. The most important factor is the aerosol lifetime. This will be characterized by
adding seed aerosol and measuring its number concentration as a function of time. This will be
done in both the light and dark.
The characterization experiments will be carried out at varying temperatures and
humidities, representing the range of conditions expected to be used for model evaluation or
reactivity assessment experiments. The number of such experiments will depend on the
sensitivities found for the various chamber effects. In addition, since the facility will be used for
multi-day simulation experiments, appropriate multi-day characterization runs will also be carried
out. It is expected that there will be some extreme conditions (e.g., extremes of temperature, near
100% humidity, or very long irradiation periods) where the characterization data will indicate
unacceptably large or irreproducible chamber effects, or unexpected results which cannot be
modeled. Therefore, in addition to characterizing chamber conditions, such experiments will also
provide information on the range of conditions for which useful experimental data can be
obtained.
C. Model Evaluation Experiments
Once the performance of the facility and the reactors are adequately characterized, and
any discrepancies or unexpected results in control experiments have been accounted for, we will
begin using it for model evaluation experiments. Because of the special capabilities of this
facility, the focus will be on model evaluation under low-NOx conditions, and evaluation of
model predictions of temperature effects. However, the model evaluation experiments will not be
limited to these areas, and will be determined largely by scientific and regulatory needs. The
objective will be to complement and extend the existing mechanism evaluation data base to
provide the data of greatest scientific utility for evaluating models for regulatory applications,
taking advantage of the special capabilities of the facility.
The plan for specific experiments will be finalized after receiving input at the workshops
and from the Advisory Committee, and after taking into account the results of the characterization
experiments. It is expected that they would include, but not necessarily be limited to, the
following types of runs:
-386-
-------
• CO - NOv - air and Methane - NO, - air runs. Note that some of these will also be carried
out for characterization purposes. Concentrations and temperatures will be varied.
• Single organic - NO, - air runs. The organics studied will include representative alkenes
(including isoprene and other biogenics), aromatic hydrocarbons, formaldehyde,
acetaldehyde, and other representative compounds with sufficient internal radical sources
that single compound - NOX experiments provide useful data1. The compounds chosen
will include representatives of the major types of compounds in anthropogenic and
biogenic emissions. This will include compounds that are known or expected to be SOA
precursors, as discussed in the following section. Concentrations, temperature and (for
SOA-forming compounds) humidities will be varied.
• Ambient surrogate - NO, experiments will be carried out using surrogate mixtures of
varying complexity, both with and without representative biogenic compounds. Note that
some of these will be used as base cases in incremental reactivity experiments, which are
discussed separately below. For the purpose or aerosol assessment studies (see below),
surrogates with varying aerosol forming potentials will be evaluated. Concentrations,
temperature and (for SOA-forming surrogates) humidities will be varied.
Synergistic effects arising from interactions of radicals from different types of reacting
VOCs are expected to be more important under lower NOX conditions than under more polluted
conditions where most of the organic radicals react with NOX. Therefore, mixture experiments
will be important for evaluating mechanisms under realistic conditions. If experiments with
complex, atmospherically realistic mixtures give results which are not expected based on results
of the single compound runs, experiments with varying simple mixtures may be appropriate to
elucidate the synergistic effects which may be occurring.
The full array of available analytical instrumentation will be employed during these
experiments. Note that data on "true" NO2 and H2O2 will be priorities in all low-NOx
experiments. In addition to the injected organic(s), data will be obtained on as many organic
products as can be detected using the available instrumentation. A priority will be to determine
how product formation differs under conditions when NOX has been consumed compared to
products formed in the presence of NOX.
Multi-day experiments will be conducted using representatives of various types of
experiments, to obtain mechanism evaluation data over longer time periods, and also for
evaluating mechanisms for nighttime chemistry. Since artificial lights will be employed,
nighttime will be simulated by turning off the lights, and mornings will be simulated by turning
them on again. (Differences between immediately turning on and off the lights and changing the
light intensity gradually will be assessed.) NO3 radicals and HONO, which are expected to be
important species, will be monitored using DO AS.
The results of the experiments will be compared with model predictions as soon as the
data are processed and characterized for modeling. Experience has shown that this provides
important feedback for quality control and for planning the most useful follow-up experiments. If
results of a particular experimental system are as expected based on the current model, follow-up
experiments are obviously of lower priority than if unexpected results are obtained. Therefore,
modeling will be an integral part of the experimental program, as it has been with most other
environmental chamber programs at CE-CERT.
NOx-air irradiations of single alkanes and other species without internal radical sources are not
useful for mechanism evaluation because of their sensitivity to the chamber radical source.
-387-
-------
Because of the number of compounds and types of mixtures which are relevant to
ambient simulations, and the wide range of conditions which can be simulated using this
chamber, it obviously will not be possible to obtain comprehensive data on all possible systems.
Therefore, determining priorities for experimental systems that provide the most useful data for
model evaluation will be important. Immediate modeling of the data obtained, combined with
external input through the workshops and Advisory Committee as discussed above, will be
essential for maximizing the utility of this facility.
D. Experimental Determination of Secondary Organic Aerosol Yields
The researchers at the California Institute of Technology, who have conducted much of
the previous research in determining the secondary organic aerosol (SOA)-forming ability of
VOCs, will work with CE-CERT in designing a chamber appropriate for state-of-the-art aerosol
research, specifying the instrumentation needed, and defining a research plan. It is expected that
the low surface-to-volume ratio and lack of wind buffeting will provide longer particle lifetimes
than has been possible previously, and temperature, lighting, and humidity control will allow
experiments to be carried out under more controlled conditions than possible using outdoor
chambers such as that at Caltech. This will be evaluated using appropriate characterization
experiments, as discussed above.
The initial experiments will be designed to duplicate experimental systems that have been
well studied using the Caltech chamber, to see if comparable results can be obtained in the new
chamber. This will include determination of SOA yields for selected aromatics and biogenics, and
modeling the results using the approach discussed by Odum et al. (1997a,b) to see whether
similar parameters are obtained. If differing results are obtained, the sources of the discrepancies
will be investigated in consultation with the Caltech researchers. Note that this may well include
carrying out at Caltech as well as using the CE-CERT chamber, though significant experimental
work at Caltech may require additional funding or modifying the budget proposed for this
program. Assuring that consistent results can be obtained at the different chamber facilities, and
that any differences observed can be understood and taken into account, will be an important
priority, since it reflects on the general utility of chamber data for this type of research.
Assuming that the aerosol data obtained from this facility are consistent with those from
Caltech or that differences are understood, the new experiment will then be used to conduct
experiments for which this facility is best suited. This will include extending the range of
conditions under which SOA are determined for the representative aromatics, biogenics and
mixtures which were previously studied, as well as studying additional compounds which may be
of interest in regulatory applications (see below). These would include experiments with varying
reactant concentrations and variable temperature and humidities. Product analysis (both gaseous
and particulate phases) will also be a component of this research, to provide information useful
for developing and evaluating mechanistic models for SOA formation. The overall goal is to
provide information needed to evaluate models for aerosol formation under controlled and varied
conditions, and to characterize the SOA forming ability of individual VOCs of interest.
It is important to recognize that in most cases the full complement of aerosol
measurements will be carried out in conjunction with the full complement of gas-phase
measurements, to provide data for evaluating both gas-phase mechanisms and aerosol formation
model evaluations and measurements. In many cases, the SOA yield determinations can be made
while conducting appropriate types of mechanism evaluation experiments discussed in the
previous section, or while conducting reactivity experiments discussed below. Likewise, gas-
-388-
-------
phase measurements made during experiments carried out as part of aerosol-related studies can
also be used for gas-phase model evaluation. This is advantageous not only because it makes
maximum use of the facility, but also because gas-phase and aerosol dynamic processes are
interdependent, and ultimately will need to be incorporated in a unified model.
E. VOC Reactivity Assessment
Since organic compounds differ significantly in their effects on ozone formation, VOC
regulations based on considerations of relative reactivity are receiving increased attention because
of their potential as a more cost-effective alternative to mass-based controls. However, existing
reactivity scales have been developed for high-NOx conditions, and some have questioned
whether they are relevant, or even directionally correct, in very low-NOx environments. An
additional issue is the question whether replacing emissions of rapidly reacting VOCs with more
slower reacting VOCs in order to reduce ozone in urban areas may degrade air quality in
downwind areas because of the greater persistence of the slowly reacting compounds. These
issues can be addressed by model simulations, but as discussed above existing environmental
chamber data have not been adequate to evaluate the accuracy of model predictions under such
conditions.
It might be argued that since VOCs are believed to have relatively little effect on ozone
formation under low NOX conditions, assessing VOC reactivities under such conditions is not a
priority. However, since the proposed facility will provide data concerning a wide range of VOC
impacts, the term "reactivity" can be considered in a context which is broader than just ozone
impacts. Furthermore, in view of the fact that some are proposing to de-emphasize VOC
regulations in low-NOx areas because they are believed to have low or possible negative effects
on ozone, it is even more important that their other air quality impacts be accurately assessed.
That is what is meant by "reactivity evaluations under low-NOx conditions." Obviously, the
model predictions that VOCs indeed have low or negative impacts on O3 under those conditions
still need to be verified. But this is not the only, or even the most important, reason for doing this
research.
Reactivity evaluation experiments consist of determining the effects of adding (or
removing) the subject VOCs from mixtures representing the ambient environment of interest. In
this case, the experiments will be designed to simulate various low-NOx rural or regional-scale
episodes or urban areas of the future, which may be approaching or attaining the air quality
standards. This may include multi-day simulations to assess long-term effects of VOCs on
downwind air quality. Since it is expected (though not yet actually experimentally demonstrated)
that ozone levels in these scenarios will be relatively insensitive to VOCs, an important focus will
be determining the effects of the VOCs on other measures of air quality in addition to ozone. This
would include determining effects on secondary particle formation, where applicable.
Experiments will also be carried out to assess the effects of temperature on the
reactivities of selected VOCs. These would include higher NOX as well as low NOX experiments,
since temperature dependence data on VOC reactivities is highly limited. The range of
temperatures employed will represent the range of ambient conditions relevant to air pollution in
the United States, to within the capability of the facility. Note that this may include temperatures
that may be too low for significant ozone formation, but where VOCs may have other impacts
that need to be assessed.
-389-
-------
The compounds studied will include representatives of the major classes of emitted
VOCs, which include alkanes, alkenes (both anthropogenic and biogenic), aromatic
hydrocarbons, and simple oxygenates. Representatives of major classes of solvent species used in
industrial and consumer product applications also will be studied. One important example would
be studies of representative compounds present in architectural coatings, to assess the impacts of
emissions of such VOCs on low-NOx or regional environments. The specific compounds to be
studied, and the number of scenarios or conditions employed, will depend on how well the results
obtained in the initial compounds correspond to model predictions, as well as regulatory needs
concerning data on individual types of VOCs, as discussed in the following section.
The second type of study that will be carried out will simulate environments that might
result if large substitutions of current emissions with low reactivity compounds are carried out.
The specific approach employed, and substitution scenarios examined, will be determined after
consultation with the RRWG and others. However, it is expected that these would be multi-day
experiments whose results will be compared with simulations using mixtures representing current
or potential future mixtures of emitted VOCs. Since different results are expected for differing
types of low reactivity compounds that might be used, experiments will be carried out using
several representative types of such compounds.
Useful compounds to study, which are expected to yield significantly different results,
might include ethane, acetone, and a long chain n-alkane such as n-hexadecane. Ethane and
acetone have low reactivity because they react relatively slowly. Although they have similar
incremental reactivities under urban conditions, they differ in that acetone is a strong radical
source while ethane is nearly neutral in this regard. Long chain alkanes in fact react relatively
rapidly, but have low ozone reactivities under urban conditions because of their strong radical
sinks. Major increases in levels of such compounds may result in greater persistent of other
reactive pollutants.
F. Studies of Impacts of Representative VOC Sources Relevant to Regulatory
Issues
It is expected that during the course of this program studies of environmental impacts of
particular compounds or source categories may be of particular relevance to current regulatory
needs. Examples might include impacts of emissions of architectural coatings or of changes in
motor vehicle fuel reformulation. In the case of architectural coatings, it is important to determine
whether the environmental benefit strict regulation of these VOCs will be worth the economic
costs and potential loss in coatings quality that such regulations might entail. This is a particular
concern when they are used in low-NOx environments where ozone may not be sensitive to their
emissions. Results will help to determine whether their regulation needs to be a priority in the
future if significant NOX controls are implemented. Experiments simulating various scenarios
involving coatings emissions can be conducted to elucidate this. This could include determining
the effects of adding the coatings VOCs to simulated rural mixtures that might be high in
biogenic compounds, as well as simulations of mixtures more representative of urban scenarios.
CE-CERT is proposing to develop a coatings research center under separate funding, so the
appropriate expertise should be available for designing such a study to yield maximum utility and
relevance.
With regard to studies of impacts of vehicle emissions, it should be noted that CE-CERT
already has an advanced research dynamometer facility and vehicle emission expertise that can be
made available to this project. The low NOX chamber will be particularly useful for studies of
-390-
-------
impacts of very low emissions (i.e. ULEV) vehicles whose emissions are too low for useful study
using currently available chamber technology. Although the emissions from these vehicles are
low, they are not completely negligible, and will become increasingly important in the future as
other sources are controlled and higher emission older vehicles are removed from service.
G. Experimental Characterization of NOy and Radical Budgets
Uncertainty in the budget of NOy will particularly limit our confidence in model
simulations of the effectiveness of NOX reduction strategies. On the urban scale, NOX emissions
can be terminated in the form of relatively inert nitrogen compounds NOZ = HNO3 + RNO3 +
PAN. At the regional scale PAN can become a net source of NOX so that NOX is terminated as
NOZ = HNO3 + RNO3. The O3 production efficiency per NOZ (P(OX)/P(NO2)) is thought to vary
considerably as a function of both the ratio of VOC/NOX and the absolute levels of VOC and
NOX. The photochemical mechanisms most commonly used in AQMs, particularly the CB4, were
designed for use in urban scenarios with high NOX levels. For those conditions, predictions of O3
were relatively insensitive to uncertainty in the NOy budget. Even in the case of mechanisms such
as RADM2 that were designed to handle rural conditions with low NOX, there are large
uncertainties in the production of NOZ for low NOX conditions. In a recent mechanism inter-
comparison, Luecken et al. (1999) found large differences in the speciation of NOZ and in O3 per
NOZ production efficiencies, particularly for low NOX conditions. Uncertainty in the NOy budget
will become increasingly important with the increased emphasis on fine particulate matter and
regional O3 levels. Thus, it is important to account for the fate of NOX and O3 production
efficiencies per NOX at low NOX conditions, and low NOX chamber experiments will be needed to
evaluate the mechanisms for those conditions.
Uncertainty in the budget of HOX will limit our confidence in model simulations of the
effectiveness of VOC reduction strategies. Recent field studies (Carpenter et al., 1998; Wennberg
et al, 1998; Stevens et al, 1997; Crosley, 1997; Cantrell et al, 1997; Cantrell et al, 1996; Plummer
et al, 1996) have found large discrepancies between model simulated and observed HOX levels
and ratios. Thus, there remains considerable uncertainty in the budgets of HOX in current
photochemical mechanisms. We note that the magnitude of chamber wall effects are inferred
from the presence of apparent artifacts in chamber experiments, i.e., the experimental results
differed from expectations based on well accepted aspects of the photochemistry. The
discrepancies between measured and modeled ratios of HO2/OH and RO2/HO2 in these field
studies raises an important concern that real ambient processes are being subsumed in chamber
wall mechanisms. Thus, it is important to characterize radical budgets in chamber experiments as
fully as possible.
To characterize the radical budget, it is necessary to experimentally evaluate the
initiation, propagation and termination of radical species. Rates of radical initiation can be
estimated by measuring actinic flux and the concentrations of radical precursors (i.e., those
species that photolyze or decompose to produce radicals). Radical propagation efficiency can be
estimated by measuring concentrations of species that control the rates of radical propagation
(Tonnesen and Dennis, 1998a,b), and radical termination can be calculated by using kinetics data
and measuring the concentrations of species involved in termination reaction. Radical termination
can also be estimated by measuring the accumulation of radical termination products.
Several techniques exist to measure the concentrations of HOX. Tanner et al (1997) have
measured OH using ion-assisted (IA) mass spectrometry, with a lower limit of 10"5 molec/cm3 for
a 5-minute integration. Mather et al (1997) have measured OH and HO2 using low pressure laser
-391-
-------
induced fluorescence (LJDF), with an OH sensitivity = 10"6, and 1- or 5-minute integration time.
Total RO2+HO2 has been measured by the chemical amplifier technique (Cantrell et al, 1997).
We note that additional analytical methods are needed to characterize NOy and HOX
budgets, but special chamber experiments are not required.^Rather, the budget analyses should be
performed on all chamber experiments if measurements are available. Comparison of HOX
budgets in the aerosol experiments (described above) with gas-phase experiments will be useful
to investigate theories that aerosols can play an important role in peroxy radical termination
(Cantrell et al, 1996; Jacob, 1998).
H. Experimental Evaluation of Indicators of Ozone Sensitivity to Precursor
Emissions
Modeling studies have suggested that nearly unique values of particular indicator ratios
are robustly associated with [O3] and P(OX) ridgeline conditions (or conditions of equal sensitivity
to VOC and NOX) for a wide range of precursor levels. For example, Sillman (1995) found that
values of certain indicators were constant as a function of O3 and precursor levels, while
Tonnesen and Dennis (1998a,b) found small variations in the indicator value depending on the O3
and precursor concentrations. In a modeling study of the San Joaquin Valley, however, Lu and
Chang (1998) found that the values of the indicators differed from previous modeling studies, and
they suggested that the indicator values may vary as a function of environmental conditions.
Experiments in an environmental chamber will be useful for assessing the variability of
the indicator values as a function of environmental conditions. Furthermore, Tonnesen and
Dennis (1998a) found that the utility of the indicators derived from chemical processes associated
with radical propagation efficiency. Thus, it is likely that chamber artifacts that affect radical
initiation will not interfere with the experimental investigation of indicator ratios, and values
measured in chamber can be compared with values determined in modeling studies. On the other
hand, if chamber artifacts significantly affect radical propagation and termination, indicator
values measured in the chamber might not be directly comparable with ambient values.
The approach to experimentally validate the usefulness of indicators requires a series of
simulations with fixed VOC emissions (and all other inputs fixed) while NO* emissions are
incrementally changed. The ridgeline conditions for P(OJ or [O3] would be identified as the NOX
emissions level which maximized P(OJ or [O3]. The full set of indicators could be evaluated in
each series of simulations, subject only to the requirement that measurements of each of the
indicator species must be collected. Although P(OJ cannot be measured directly, it can be
determined by model simulations of the experimental conditions, or it can be calculated directly
using measurements of NO, HO2 and RO2 if these are available. The robustness and consistency
of the indicator method would then be evaluated by determining the indicator values in additional
series of experiments with different VOC, light, temperature and humidity levels. Finally, the
aerosols experiments described above can also be utilized to determine the effects of aerosols on
indicator values.
I. Evaluation of Ambient Monitoring Equipment
As discussed above, the proposed large chamber facility will provide a unique test bed
for evaluating mew monitoring equipment using well characterized chemical systems which
nevertheless are representative of field conditions. The large volume of the chamber will permit
evaluation of equipment with larger sampling requirements than are practical for use with most
-392-
-------
current indoor chambers. Most of this work would be carried out in collaboration with the
developers or intended users of this equipment, who in most cases would be expected to provide
funding for this effort. However, some of these tests can be carried out in conjunction with
experiments already being carried out for other purposes.
No specific projects of this type are described in this proposal because it is unknown
what will be the priorities for evaluation of monitoring equipment at the time this chamber is
operational. Once the facility is operational and its performance is evaluated, the availability of
this facility for evaluations of this type will be communicated to relevant researchers through
various means, including NARSTO meetings and workshops. For example, the facility could be
utilized for this purpose as part of upcoming NARSTO field projects, with the research
coordinated through NARSTO. This will be determined once the project is under way.
J. Other Studies
The projects discussed above are obviously not the only ways in which this facility can
be utilized, and it is expected that other studies will be carried out depending on regulatory needs,
interests and capabilities of collaborating researchers, and input received from the advisory
committee and the workshops. It is expected that the priorities of the program will evolve as
needs evolve, and in response to results of experiments carried out not only at this facility but at
other laboratories.
K. Schedule
This project will be carried out over a four-year period. The first six months will be
devoted to developing the work plan, holding the initial workshop and forming the advisory
committee, evaluating the experience and data obtained at other chamber facilities, and evaluating
chamber design options. Acquiring the analytical equipment and developing analytical methods
and quality assurance plans will also begin during this period, and continue throughout the first
year of the program. Chamber construction should begin during the second six months of the
program, and characterization should be completed during the first half of the second year. The
scheduling and priorities for the subsequent experiments will be determined in consultation with
the advisory committee and the funding agencies, and will depend on availability with
collaborators, where applicable.
L. Budget
The amount budgeted for this program is approximately $3 million, with approximately
half of that being for design and construction of the chamber and acquisition of necessary
analytical equipment, and the other half for researchers and staff salaries and operating the
chamber for four years
M. References
Cantrell, C.A.; Shelter, R.E.; Gilpin, T.M.; Calvert, J.G.; and Eisele, F.L. (1996) Peroxy radical
concentrations measured and calculated from trace gas measurements in the Mauna Loa
Observatory Photochemistry Experiment 2. /. Geophys. Res. 101:14,653-14,664.
-393-
-------
Cantrell, C.A.; Shelter, R.E.; Calvert, J.G.; Eisele, F.L.; Williams, E.; Baumann, K.; Brune, W.H.;
Stevens, P.S.; and Mather, J.H. (1997) Peroxy radicals from photostationary state
deviations and steady state calculations during the Tropospheric OH Photochemistry
Experiment at Idaho Hill, Colorado. /. Geophys. Res. 102:6369-6378.
Carpenter, L.J.; Clemitshaw, K.C.; Burgess, R.A.; Penkett, S.A.; Cape, J.N.; McFadyen, G.G.
(1998) Investigation and evaluation of the NOx/O3 photochemical steady state, Atmos.
Environ. 32:3353-3365.
Carter, W.P.L. (1994) Development of ozone reactivity scales for volatile organic compounds. /.
Air & Waste Manage. Assoc. 44:881 -899.
Carter, W.P.L.; Atkinson, R.; Winer, A.M.; and Pitts, J.N., Jr. (1982) Experimental investigation
of chamber-dependent radical sources. Int. J. Chem. Kinet. 14:1071.
Carter, W.P.L., and Lurmann, F.W. (1990) Evaluation of the RADM gas-phase chemical
mechanism. Final Report, EPA-600/3-90-001.
Carter, W.P.L., and Lurmann, F.W. (1991) Evaluation of a detailed gas-phase atmospheric
reaction mechanism using environmental chamber data. Atmos. Environ. 25A:2771-2806.
Carter, W.P.L.; Luo, D.; Malkina, I.L.; and Fitz, D. (1995a) The University of California,
Riverside, environmental chamber data base for evaluating oxidant mechanism. Indoor
chamber experiments through 1993. Report submitted to the U.S. Environmental
Protection Agency, EPA/AREAL, Research Triangle Park, NC, March 20.
Carter, W.P.L.; Luo, D.; Malkina, I.L.; and Pierce, J.A. (1995b) Environmental chamber studies
of atmospheric reactivities of volatile organic compounds. Effects of varying chamber
and light source. Final report to National Renewable Energy Laboratory, Contract XZ-2-
12075, Coordinating Research Council, Inc., Project M-9, California Air Resources
Board, Contract A032-0692, and South Coast Air Quality Management District, Contract
C91323, March 26.
Crosley, D.R. (1997) Tropospheric OH photochemistry experiment: A summary and perspective,
/. Geophys. Res. 102:6495-6510, 1997.
Dodge, M.C. (1998) Chemical oxidant mechanisms for air quality modeling, Critical review
paper for 1998 Ozone Assessment. Atmos. Environ., submitted for publication.
Fitz, D.R.; Carter, W.P.L.; and Cocker, D. (1998) Evaluation of the PM and ozone producing
potential of natural gas-powered vehicles. Draft final report to the California Institute for
Energy Efficiency, Lawrence Berkeley Laboratory, Agreement 4910210, October.
Gery, M.W.; Whitten, G.Z.; and Killus, J.P. (1988) Development and testing of the CBM-IV for
urban and regional modeling. EPA-600/ 3-88-012, January.
Jacob, D.J (1999): "NARSTO Critical Review Paper, Heterogeneous Chemistry and
Tropospheric Ozone," Atmospheric Environment, (in press). This manuscript can be
downloaded from http://www.cgenv.com/Narsto/critrev.papers.html.
-394-
-------
Jeffries, H.E.; Gery, M.W.; and Carter, W.P.L. (1992) Protocol for evaluating oxidant
mechanisms for urban and regional models. Report for U.S. Environmental Protection
Agency Cooperative Agreement No. 815779, Atmospheric Research and Exposure
Assessment Laboratory, Research Triangle Park, NC.
Lu, C-H, and Chang, J.S. (1998) On the indicator-based approach to asses ozone sensitivities and
emissions features. /. Geophys. Res. 103:3453-3462.
Luecken, D.J.; Tonnesen, G.S.; and Sickles, I.E., H (1999) Differences in NOy speciation
predicted by three photochemical mechanisms. Atmos. Environment, in press.
Odum, J.R.; Jungkamp, T.P.W.; Griffin, R.J.; Forstner, H.J.L.; Flagan, R.C.; and Seinfeld, J.H.
(1997a) Aromatics, reformulated gasoline, and atmospheric organic aerosol formation.
Environ. Sci. Technol., 31:1890-1897.
Odum, J.R.; Jungkamp, T.P.W.; Griffin, R.J.; Flagan, R.C.; and Seinfeld, J.H. (1997b) The
atmospheric aerosol-forming potential of whole gasoline vapor, Science, 276: 96-99.
Mather, J.H, Stevens, P.S., and Brune, W.H (1997): "OH and HO2 mesaurements using laser-
induced fluorescence," J. Geophys. Res., 102.
Plummer, D.A.; McConnell, J.C.; Shepson, P.B.; Hastie, D.R.; and Niki, H. (1996) Modeling of
ozone formation at a rural site in southern Ontario, Atmos. Environ. 30:2195-2217.
Sillman, S. (1995) The use of NOy, H2O2, and HNO3 as indicators for ozone-NOx-hydrocarbon
sensitivity in urban locations, /. Geophys. Res. 100:14175-14188.
Simonaitis, R., and Bailey, E.M. (1995) Smog chamber studies at low VOC and NOX
concentrations: Phase I. Report on Interagency Agreement DW64936024 to EPA/NREL,
Research Triangle Park, NC.
Stevens, P.S.; Mather, J.H.; Brune, W.H.; Eisele, F.; Tanner, D.; Jefferson, A.; Cantrell, C.;
Shelter, R.; Sewall, S.; Fried, A.; Henry, B.; Williams, E.; Bagman, K.; Goldan, P.; and
Kuster, W. (1997) HO2/OH and RO2/HO2 ratios during the Tropospheric OH
Photochemistry Experiment: Measurement and theory. /. Geophys. Res. 102:6379-6391.
Tanner, D. J.; Eisele, F. L.; and Jefferson, A. (1997) Ion-assisted OH instrument, /. Geophys Res.,
102:9295-9308.
Tonnesen, G.S., and Dennis, R.L. (1998a) Analysis of radical propagation efficiency to assess
ozone sensitivity to hydrocarbons and NOX. Part 1: Local indicators of odd oxygen
production sensitivity. Submitted to /. Geophys. Res.
Tonnesen, G.S., and Dennis, R.L. (1998b) Analysis of radical propagation efficiency to assess
ozone sensitivity to hydrocarbons and NOX. Part 2: Long-lived species as indicators of
ozone concentration sensitivity. Submitted to /. Geophys. Res.
Wennberg, P.O.; Hanisco, T.F.; Jaegle, L.; Jacob, D.J.; Hintsa, E.J.; et al (1998) Hydrogen
radicals, nitrogen radicals, and the production of O3 in the upper troposphere. Science
279:49-53.
-395-
-------
Zafonte, L., P. L. Rieger, and J. R. Holmes (1977): "Nitrogen Dioxide Photol-ysis in the Los
Angeles Atmosphere," Environ. Sci. Technol. 11,483-487.
-396-
-------
Appendix A
-------
Participants
1999 Combined US/German Ozone/Fine Particle Science and
Environmental Chamber Workshop
Dr. Kurt Anlauf, Research Scientist
Atmospheric Environment Service
4905 Dufferin Street
Toronto, M3H 5T4
CANADA
(416) 739-4840
(416) 739-5708
kurtanlaufiStec.gc.ca
Dr. Janet Arey, Research Chemist
University of California
Statewide Air Pollution Research Center
Riverside, CA 92521
(909) 787-5128
(909) 787-5004
ianet.arev@ucr.edu
Sara Aschmann
University of California
Statewide Air Pollution Research Center
Riverside, CA 92521
(909) 787-5128
(909) 787-5004
a5chmann@ucrac 1 .ucr.edu
Dr. Roger Atkinson, Professor
University of California
Statewide Air Pollution Research Center
Riverside, CA 92521
(909) 787-4191
(909) 787-5004
ratki'ns@ucracl .ucr.edu
Dr. Robert J. Avery
Sr. Associate - Product Issues
Eastman Chemical Co.
P.O.Box431,Bldg.280
Kingsport, TN 37662
(423)229-5409
(423) 224-0208
riaverv@eastrnan.com
Daniel C. Baker
Staff Research Engineer
Equilon Enterprises
3333 Highway 6 South, Rm. EC-129
Houston, TX 77079
(281)544-8737
(281) 544-8727
dcbaker@,equilon.com
Dr. Ian Barnes
Bergische Universitat-GH Wuppertal
Physikalische Chemie/FB 9, GauBstraBe 20
Wuppertal, D-42119
GERMANY
49-202-4392510
49-202-4392505
bames@,phvschem.uni-wuppertal.de
Prof. Dr. Karl H. Becker
Bergische Universitat-GH Wuppertal
Physikalische Chemie/FB 9, GauBstraBe 20
Wuppertal, D-42119
GERMANY
49-202-4392666
49-202-4392505
becker@,phvschern.uni-wuppertal.de
Dr. Thorsten Benter
Department of Chemistry
University of California
516 Rowland Hall
Irvine, CA 92697-2025
(949) 824-2373
(949)824-3168
tbenter(5),uci.edu
Naveen Berry
Air Quality Specialist, Pin., Rule Dev. & Area
Sources
SCAQMD
21865 E.Copley Dr.
Diamond Bar, CA 91765
(909) 396-2363
(909) 396-3324
nberrv@,aqmd.gov
-398-
-------
Derek Beving
Graduate Student
University of California
Air Pollution Research Center
Riverside, CA 92521
(909) 787-4723
(909) 787-5004
debeving@.dtrns iir-r Prfu
Dr. Heinz W. Bierman, Software Developer
c/o CE-CERT
University of California
Riverside, CA 92521
(909) 780-1418
(909) 781-5790
lieinzb(S)fcTimngtwork.com
Jack P. Broadbent
DEO/Pln., Rule Dev. & Area Sources
SCAQMD
21865 E.Copley Dr.
Diamond Bar, CA 91765-4182
(909) 396-3789
(909) 396-3802
ibroadbent(g)faqmri pr
-------
Dr. Dick Derwent
Meteorological Office
London Road
Bracknell, RG12 252
UK
44-1344-854621
44-1344-854493
rgderwent@meto.gov.uk
Basil Dimitriades, Senior Scientific Advisor
USEPA (MD-80)
Research Triangle Park, NC 27711
(919) 541-2706
(919) 541-L094
dimitriades.basil@epa.gov
Kenneth Docherty
Graduate Student
University of California
Air Pollution Research Center
Riverside, CA 92521
(909) 787-4723
(909) 787-5004
kdochertv@mail.ucr.edu
Dr. Jean-Francois Doussin
Universite Pris 7
Faculte des Sciences, 61 avenue De Gaulle
Creteil, F-94010
FRANCE
33-0-1-45171587
33-0-1-45171564
doussin@lisa.univ-paris 12.fr
Peter S. Ellis
Exxon Chemical Company
5200 Bayway Drive
Baytown, TX 77520
(281) 834-1681
(281) 834-2747
peter.s.ellis@exxon.com
Dr. Barbara J. Finlayson-Pitts, Professor
University of California
Department of Chemistry
Irvine, CA 92697-2025
(949) 824-7670
(949)824-3168
bi finl av@uci .edu
Denis Fitz
Manager, AP/SSMC
CE-CERT
University of California
Riverside, CA 92521
(909)781-5781
(909)781-5790
fitz@cert.ucr.edu
Dr. Krishna Foster
Department of Chemistry
University of California
516 Rowland hall
Irvine, CA 92697-2025
(949) 824-6347
(949)824-3168
kfosterc@uci.edu
Dr. Donald L. Fox
Dept of Environmental Sciences & Engineering
UNC-Chapel Hill
CB #7400, Rosenau Hall
Chapel Hill, NC 27599-7400
(919) 966-3054
(919) 966-2583
don fox@unc.edu
Dr. Jeff Gaffhey, Chief Scientist
Argonne National Laboratory
Bldg. 2031/ER, 9700 S. Cass Ave.
Argonne, IL 60439
(630) 252-5178
(630) 252-7415
gaffnev@anl.gov
Mike Gebel
University of California
Department of Chemistry
Irvine, CA 92697-2025
(949) 824-6915
(949)824-2261
Dr. Harald Geiger
Bergische Universitat-GH Wuppertal
Institut fur Physikalische Chemie/FB 9
BauBstr.20
Wuppertal, 49097
GERMANY
49-202-4392515
49-202-4392505
geiger@phvschem.uni-wuppertal.de
-400-
-------
David M. Golden
Senior Staff Scientist
SRI International
333 Ravenswood Ave., PS031
Menlo Park, CA 94025
(650)859-3811
(650) 859-6196
golden@,sri.com
Dr. Joyce F. Graf
Director, Environmental Science
CTFA
1101 17th Street, NW, Ste. 300
Washington, DC 20036
(202)331-1770
(202)331-1969
grafi@,ctfa.org
Susanne Grittner
Umweltbundesamt Berlin
Bismarckplatz 1
Berlin, D-14193
GERMANY
49-303-89032757
49-303-89032282
susanne. gritmer@,uba.de
Robert W. Hamilton, Research Associate
Amway Corporation
7575 Fulton Road, East, MC 50-2E
Ada, MI 49355
(616) 787-7697
(616) 787-7941
bob hamiltonfaiamwav.com
Dr. Alam Hasson
UCLA, Department of Atmospheric Sciences
7127 Math Science Bldg.
405 Hilgard Avenue
Los Angeles, CA 90095-1565
(310)206-3346
(310)206-5219
ahasson@,atmos.ucla.edu
Dipl.-Phys. Martin Heitlinger
Forschungszentrum Mich GmbH
Postfact 19 13
Mich, D-52425
GERMANY
49-2461-616930
49-2461-615346
m.heitlinger(g),kfa-iuelich.de
Prof. Dr. Hartmut Herrman
Institut fur Tropospharenforschung e.V.
Permoserstr. 15
Leipzig, D-04303
GERMANY
49-341-2352446
49-341-2352325
herrmann@.troDos.de
Dr. Jens Hjorth
Environment Institute
The European Commission
Joint Research Centre-Ispra, TP272
Ispra 1-21020
ITALY
39-0332-789076
39-0332-785837
i ens .hiorth@i re. it
Dr. John R. Holmes
Chief, Research Division
California Air Resources Board
P.O. Box2815
Sacramento, CA 95812
(916) 323-2673
(916) 327-5748
iholmes@,arb.ca.gov
Dr. Koji Inazu, Research Associate
Director of Environmental Chemistry & Engineering
Tokoyo Institute of Technology
4259 Nagatsuta, Midori-ku
Yokohama
JAPAN
Inazu@,chemenv.titech.ac. it)
Dr. Laura T. Iraci
SRI International
333 Ravenswood Avenue
Menlo Park, CA 94025
(650) 859-4608
(650) 859-6196
laura.iraci@,sri.com
Dr. Harvey Jeffries
Atmospheric and Aquatic Sciences
University of North Carolina
CB #7400, Dept. ESE
Chapel Hill, NC 27599-7400
(919)967-0160
(919) 933-2393
Harvey@unc.edu
-401-
-------
William L. Johnson, Environmental Engineer
US EPA - AQSSD (MD-15)
Research Triangle Park, NC 27711
(919) 541-5245
(919)541-8524
iohnson.william@epa.gov
Dr. Ajith Kaduwela, Staff Air Pollution Specialist
California Air Resources Board
P.O. Box 2815
Sacramento, CA 95812-2815
(916)327-3955
(916) 327-8524
aJcaduwel@arb.ca.gov
Markus Kalberer
California Institute of Technology
1200 E. California Blvd. (MS 210-41)
Pasadena, CA 91125
(626) 395-3697
(626) 585-1729
kalberer@its.caltech.edu
Dr. Richard Kamens
Prof, Environmental Sciences & Engineering
University of North Carolina, Chapel Hill
Rosemau Hall
Chapel Hill, NC 27599-7400
(919) 966-5452
(919)966-7911
kamens@unc.edu
Brian Keen, Senior Technology Manager
Union Carbide Corporation
P. O. Box 8361, Bldg. 701/250
South Charleston, WV 25213
(304) 747-4897
(304) 747-4623
keenbt@ucarb.com
Dr. Bjorn Klotz
Bergische Universitat-GH Wuppertal
Physikalische Chemie/FB 9
GauBstraBe 20, Wuppertal D-42119
GERMANY
49-202-4393534
49-202-4392505
bioem@phvschem.uni-wuDpertal.de
Dr. Udo Krischke
UCLA, Department of Atmospheric Sciences
7127 Math Science Building
405 Hilgard Avenue
Los Angeles, CA 90095-1565
(310)206-3346
(310)206-5219
krischke@atmos.ucla.edu
Dr. Jonathan J. Kurland, Senior Research Scientist
Union Carbide Corporation
P.O. Box 8361
South Charleston, WV 25303-0361
(304)747-3816
(304) 747-3752
kurlani i@ucarb.com
Eric Kwok, Air Pollution Specialist
California Air Resources Board
P.O. Box2815
Sacramento, CA 95812
(916) 322-3943
(916) 327-5621
ekwok@arb.ca.gov
MattLakin
University of California
Department of Chemistry
Irvine, CA 92697-2025
(949) 824-6915
(949) 824-2261
Dr. Georges LeBras
C:N.R:S.-LCSR
1C Avenue de la Recherche Scientifique
Orleans Cedex 2, F-45071
FRANCE
33-238-255461
33-238-257905
lebras@cnrs-orleans .fr
Dr. Yin-Nan Lee
Brookhaven National Laboratory
Environmental Chemistry Division / Bldg. 815E
Upton, NY 11973
(516)344-3294
(516) 344-2887
vnlee@bnl.gov
-402-
-------
Dr. Chung S. Liu, Deputy Executive Officer
SCAQMD
21865 E. Copley Drive
Diamond Bar, CA 91765
(909) 396-2105
(909) 396-3252
clm@aqmd.gov
Dr. Deborah Luecken, Physical Scientist
USEPA(MD-84)
Research Triangle Park, NC 27711
(919)541-0244
(919) 541-4787
luecken.deborah@epa.gov or ldg@hpcc.epa.gov
Dongmin Luo, Air Resources Engineer
California Air Resources Board
P. 0. Box 2815
Sacramento, CA 95812
(916) 323-1042
(916)323-1075
dluo@arb.ca.gov
Mark Maddox, Vice President
Kessler & Associates, Inc.
510 11th Street, SE
Washington, DC 20003
(202) 547-6808
(202) 546-5425
mmaddox@kesslerassoc.com
Dr. Claudia Maeder
Umweltbundesamt Berlin
Bismarckplatz 1
Berlin, D-14193
GERMANY
49-30-89032414
49-30-89032282
claudia.maeder@uba.de
Dr. Paul Makar
Atmospheric Environment Service
4905 Dufferin Street
Downsview, Ontario, M3H 5T4
CANADA
(416) 739-4692
(416) 739-4288
paul.makar@ec.gc.ca
Eileen McCauley
Air Pollution Specialist
California Air Resources Board
P.O. Box 2815
Sacramento, CA 95812
(916) 323-1534
(916) 322-4357
emccaule@arb .ca. gov
Dr. Jana Milford, Associate Professor
University of Colorado
Campus Box 427
Boulder, CO 80304
(303) 492-5542
(303) 492-2863
milford@colorado.edu
Dr. Michi Mochida, Post Doc
University of California
Department of Chemistry
Irvine, CA 92697-2025
(949) 824-6915
(949) 824-2261
David Morgott
Eastman Kodak Company
1100 Ridgeway Avenue
Rochester, NY 14652-6272
(716) 588-3704
(716) 722-7561
dmorgott@kodak.com
Dr. Joseph M. Norbeck, Director
CE-CERT
University of California
Riverside, CA 92521
(909) 781-5778
(909) 781-5790
i oe@cert.ucr.edu
Dr. Eduardo P. Olaguer, Environmental Specialist
The Dow Chemical Company
1803N Building
Midland, MI 48674
(517) 636-2927
(517) 638-9305
epolaguer@dow.com
-403-
-------
Grazyna E. Orzechowska
UCLA, Atmospheric Science Department
7127 Math Science Building
405 Hilgard Avenue
Los Angeles, CA 90095-1565
(310) 206-3346
(310)206-5219
eorzech@atmos.ucla.edu
Randy Pasek
Manager, Atmospheric Processes Research Section
California Air Resources Board
P.O. Box 2815
Sacramento, CA 95812
(916) 324-8496
(916) 322-4357
rpasek@arb.ca.gov
Dr. Ronald K. Patterson
Physical Science Administrator
USEPA/ORD/NERL/MD-77
Research Triangle Park, NC 27711
(919) 541-3779"
(919) 541-0239
patterson.ronald@epa.gov
Dr. Suzanne Paulson, Assistant Professor
UCLA, Department of Atmospheric Sciences
Los Angeles, CA 90095-1565
(310)206-4442
(310)206-5219
paulson@atmos.ucla.edu
Dr. James Pitts
University of California
Department of Chemistry
Irvine, CA 92697-2025
(949) 824-3756
(949)824-3168
inpitts@uci.edu
Dr.Prof.UlrichPlatt
Ruprecht-Karls-Universitat
Institut fur Umweltphysik
Im Neuenheimer Feld 229
Heidelberg, D-69120
GERMANY
49-6221-546339
49-6221-546405
nl@uphvsl.uphvs.uni-heidelberg.de
Manuel Pons
CEAM
Parque Technologico C./Charles Darwin 14
Patema (Valencia), E-46980
SPAIN
34-6-1318227
34-6-1318190
manuel@,ceam.es
Dr. Lars Ruppert
Frauhofer Institut fur Atmosphare Umweltforschung
KreuzeckbahnstraBe 19
Garmisch-Partenkirchen, D-82467
GERMANY
49-8821-183255
49-8821-183243
rupert@ifu.fhg.de
Najat Saliba
University of California
Department of Chemistry
Irvine, CA 92697-2025
(949) 824-6915
(949) 824-2261
David E. Sanders, Environmental Engineer
USEPA/QAQPS/AQSSD/OPSG/MD-15
Research Triangle Park, NC 27711
(919) 541-3356
(919) 541-0824
sanders.dave@,epa. gov
Dr. Julio Sandoval
Instituto Mexicano Del Petroleo
Eje Central Lazaro. Cardenas 152
Mexico, D.F., 07730
MEXICO
5-567-8599
5-587-7988
i sando v@www. imp .mx
Ken Schere
Chief, Atmospheric Model Development Branch
USEPA/MD-80
Research Triangle Park, NC 27711
(919) 541-3795
(919)541-1379
schere.kenneth@epa.gov
-404-
-------
Dr. John Seinfeld
Chair, Division of Engineering & Applied Science
California Institute of Technology
1201 E. California Blvd., MS 104-44
Pasadena, CA 91125
(626) 395^100
(626) 525-1729
seinfeld@caltech.edu
Keneth G. Sexton, Research Associate
University of North Carolina
CB #7400, Dept. ESE, Rosenau Hall
Chapel Hill, NC 27599
(919) 966-3932
(919)966-7911
Ken sexton@unc.edu
Prof. Dr. Howard Sidebottom
University Colege Dublin
Chemistry Department
Belfield, Dublin 4
IRELAND
353-1-7062293
353-1-7062127
howard.sidebottom@ucd.ie
Gustavo Sosa Iglesias, Environmental Researcher
Institute Mexicano del Petroleo
Ejecentral Lazaro Cardenas 152
Mexico, D.F., CP 07730
MEXICO
525-567-8599
525-587-7988
gsosa@www.imp.mx
Dr. William R. Stockwell
Associate Research Prof, Atmospheric Sciences
Desert Research Institute
2215 Raggio Parkway
Reno, NV 89512-1095
(775) 674-7058
(775) 674-7008
wstock@dri.edu
Dr. Jochen Stutz
UCLA
Department of Atmospheric Sciences
7127 Math Science Bldg., 405 Hilgard Ave.
Los Angeles, CA 90095-1565
(310) 825-5364
(310)206-5219
iochen@atmos.ucla.edu
Dr. Erick Swartz, Research Chemist
USEPA/NERL/HEASD/ACPB/MD-84
Research Triangle Park, NC 27711
(919) 541-2829
(919) 541-4787
swartz.erick@epa.gov
Carla Takemoto
California Air Resources Board
P. O. Box 2815, SSD, AQMB
Sacramento, CA 95812
(916)322-8283
(916)327-5621
ctakemot@arb.ca.gov
George Talbert, Director
Texas Air Research Center
P. O. Box 10613
Beaumont, TX 77710
(409) 880-2183
(409) 880-2397
gotalbert@aol.com
Paul L. Tanaka, Graduate Research Assistant
University of Texas at Austin
10100 Burnet Road, MC R7100
Austin, TX 78758-4497
(512)475-8872
(512)471-1720
tanaka@,che.utexas .edu
Herbert Tobias, Postdoctoral Associate
University of California
Air Pollution Research Center
Riverside, CA 92521
(909) 787-4683
(909) 787-5004
htobias@citrus.ucr.edu
Stanley Tong, Environmental Engineer
USEPA, Region IX (AIR-4)
75 Hawthorne Street
San Francisco, CA 94105
(415)774-1191
(415) 774-1073
tong.stanlev@,epa.gov
Dr. Gail Tonnensen, Manager
CE-CERT
University of California
Riverside, CA 92521
(909)781-5676
(909)781-5790
tonnensen@cert.ucr.edu
-405-
-------
Eric Vesine
LCSRCNRS
1C Avenue de la Recherche Scientifique
Orleans Cedex 2,45071
FRANCE
33-2-38-255474
33-2-38-696004
yesine@cnrs-orleans.fr
Dr. Rafael Villasenor
Institute Mexicano del Petroleo
Eje Central Lazaro Cardenas 152
Mexico, D.F., 07730
MEXICO
5-567-8599
5-587-7988
rvillase@www.imp.mx
Wolfganz Vitze
Hessische Landesanstalt fur Umwelt
RheingaustraBe 186
Wiesbaden, D-65203
GERMANY
49-611-6939250
49-611-6939555
w.vitze@hlfli.de
Andreas Warmer
Forschungszentrum Mich GmbH
Institut fur Chemie und Dynarnik der
GeosphareICG3
Stettemicher Staatsforst
Julich, 52425
GERMANY
49-2461-615932
49-2461-615346
a.wahner@kfa-iuelich.de
WeihongWang
University of California
Department of Chemistry
Irvine, CA 92697-2025
(949) 824-7714
(949) 824-3168
wanew@uci.edu
Lihua Wang, Research Assistant
University of Colorado
Department of Mechanical Engineering
Boulder, CO 80309
(303)472-4184
(303)492-3498
ljhau.wane@colorado.edu
Robert Wendoll, Director of Environmental Affairs
Dunn-Edwards Corporation
4885 E. 52nd Place
Los Angeles, CA 90040
(323) 771-3330
(323) 771-3809
Dr. John Wenger
University College Cork
Chemistry Department
Cork
IRELAND
353-21-9002454
353-21-9003014
iohn.wenger@BUREAU.UCC.IE
Dr. Peter Wiesen
Bergische Universtat-GH Wuppertal
Physikalische Chemie/FB 9, GauBstraBe 20
Wuppertal, D-42119
GERMANY
49-202-4392515
49-202-4392505
wiesen@phvschem.uni-wuppertal.de
Dr. Klaus Wirtz
CEAM
Parque Technologico C./Charles Darwin 14
Patema (Valencia), E-46980
SPAIN
34-6-1318227
34-6-1318190
klaus@ceam.es
Eugene Y. J. Yang, Air Resources Engineer
California Air Resources Board
P.O. Box 2815
Sacramento, CA 95812
(916) 324-6917
(916)327-8524
wang@arb.ca.gov
Husheng Yang
University of California
Department of Chemistry
Irvine, CA 92697-2025
(949)824-6915
(949) 824-2261
-406-
-------
Prof. Dr. FriedhelmZabel
Universitat Stuttgart
Physikalisch-Chemisches Institut
Pfaffenwaldring 55
Stuttgart, D-70569
GERMANY
49-711-6854423
49-711-6854494
f.zabel@,ipc.uni-stuttgart.de
Dr. Paul J. Ziemann, Assistant Professor
University of California
Department of Environmental Sciences
Riverside, CA 92521
(909)787-5127
(909) 787-5004
pziemannfoiucracl.ucr.edu
Bernie Zysman
Occidental Chemical Corporation
P. O. Box 344
Niagara Falls, NY 14302
(716) 278-7894
(716) 278-9297
Bernie Zvsman@,oxv.com
"US. GOVERNMENT PRINTING OFFICE: 2000-650-101-40005
-407-
-------
-------
-------
United States
Environmental Protection Agency
National Risk Management
Research Laboratory, G-72
Cincinnati, OH 45268
Please make all necessary changes on the below label,
detach or copy, and return to the address in the upper
left-hand comer.
If you do not wish to receive these reports CHECK HERED ;
detach, or copy this cover, and return to the address in the
upper left-hand corner.
Official Business
Penally for Private Use
$300
EPA/600/R-00/076
------- |