c/EPA
EPA/600/R-10/050
April 2010
Temporal and modal characterization of DoD source air toxic emission
factors
Final Report
SERDP Project Number: WP/CP1247
Brian K. Gullett
Air Pollution Prevention and Control Division
National Risk Management Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Prepared for
Strategic Environmental Research and Development Program
Department of Defense
901 North Stuart Street, Suite 303
Arlington, VA 22203
EPA Interagency Agreement # RW-96-92280901-0
RW969398430, RW969394437, RW969398433, RW969398434, RW969398435.
-------
EPA/600/R-10/050
April 2010
Table of Contents
List of Tables iv
List of Figures iv
List of Acronyms viii
Keywords: x
Acknowledgements xi
Abstract xii
Executive Summary xiii
Objectives xvii
1. Background 1
1.1 Emission Factors 1
1.2 Monitoring for Organic Toxics 1
1.3 Related Real-time Technologies for Monitoring of Trace Organic Pollutants 2
1.4 Related REMPI-TOFMS Studies on Vehicle Exhausts 2
1.5 Monitoring for Metals 3
1.6 Optical Path Methods and Field Measurements 3
2. Description of Equipment and Methods 5
2.1 REMPI-TOFMS 5
2.1.1 Laser Systems 7
2.1.2 Valve Inlet Systems 9
2.1.3 Time of Flight Mass Spectrometer 10
2.1.4 REMPI-TOFMS Instrument 11
2.1.5 Operating Procedures REMPI-TOFMS 12
2.2 LIBS 14
2.3 ORS 15
3. REMPI-TOFMS: Field Ready Development and Performance Evaluation and
Improvement 17
3.1 Laser Systems 17
3.1.1 Continuum Laser 17
3.1.2 OPOTEK Laser 17
3.2 Pulsed Valve Operation 18
3.3 Data Acquisition Software 19
3.4 Calibration of REMPI-TOFMS System 20
4. Sampling from DoD Sources 21
5. Source 1: Validation of REMPI-TOFMS Measurements on a U.S. Marine Corps
Diesel Generator 22
5.1 Experimental section 22
-------
EPA/600/R-10/050
April 2010
5.2 Results and Discussion 23
5.2.1 Steady State Diesel Generator Results 23
5.2.2 Transient Diesel Generator Results 31
6. Source 1: U.S. Marine Corps Diesel Generator Air Toxic Emission Characterization 34
6.1 Experimental 34
6.2 Results and Discussion 37
6.2.1 Steady State Emissions 37
6.2.2 Emissions during Startups 37
6.2.3 Emissions during Load Variation 41
7. Source 2: Real-Time Measurement of Trace Aromatics during Operation of Aircraft
Ground Equipment 44
7.1 Experimental 44
7.1.1 AGE 44
7.1.2 Operating and Sampling Procedures 45
7.2 Results and Discussion 46
7.3 ORS Measurements during AGE sampling 56
7.3.1 Experimental Design 56
7.3.2 ORS Instrument-Retro reflector Distance 57
7.4 Data Processing 57
7.4.1 ORS Results 58
8. Source 3: Verification Results of REMPI-TOFMS as a Real-Time PCDD/F Emission
Monitor 61
8.1 Materials and Methods 61
8.1.1 Boiler testing 61
8.1.2 REMPI-TOFMS Testing 62
8.1.3 Indicator Compounds 63
8.2 Results and Discussion 64
8.2.1 Pre-ETV results 64
8.2.2 ETV Results 64
9. Source 4: Sampling from MWC Flue Gas 68
9.1 Portsmouth Naval Shipyard Waste Combustor 2004 68
9.2 Experimental Approach 68
9.3 Test Matrix 69
9.4 Results 70
9.4.1 General On-site REMPI-TOFMS Instrument Performance 70
9.4.2 REMPI-TOFMS results 70
9.4.3 Method 0023 and Method 0010 Results 73
9.5 Portsmouth Naval Shipyard Waste Combustor 2006 73
9.6 Materials and Methods 75
11
-------
EPA/600/R-10/050
April 2010
9.7 Results 76
10. Source 5: Emission Responses from HMMWVs, the Ml Abrams Tank, and the
Bradley IFV 79
10.1 Experimental 79
10.1.1 Platforms Tested 79
10.1.2 Test Protocols 80
10.1.3 Sampling Approach 80
10.2Results and Discussion 82
10.2.1 WV Cycle 83
10.2.2 HWFET Cycle 85
10.2.3 Velocity/Gradient Cycle 85
10.2.4 Startups 86
10.2.5 Bradley and Abrams 87
10.2.6 Number Size Distributions (NSDs) 90
10.3Emission Correlations 90
11. Source 6: F-15 and F-22 Aircraft Engine Emissions 93
11.1 Experimental Method 93
11.1.1 Aircraft/Engine 93
11.1.2 Testing Venue 94
11.1.3 Exhaust Sampling 95
11.1.4 REMPI-TOFMS Sampling Approach 95
11.1.5 REMPI-TOFMS Data Analysis Procedure 96
11.2Results 97
11.2.1 F-15 Tests 97
11.2.2 F-22 Tests 100
12. Conclusions 108
13. References 110
14. Appendix A: List of Scientific/Technical Publications 116
14.1.1 Journal Articles 116
14.2Oral and Poster Presentations 116
in
-------
EPA/600/R-10/050
April 2010
List of Tables
Table 4-1. Source testing 21
Table 6-1. EPA sampling methods 35
Table 6-2. Air toxic emission factors determined by jet REMPI-TOFMS, OLGC, and
EPA reference methods, and listed in AP-42 39
Table 6-3. RA of the three measurement methods during parallel sampling 40
Table 7-1. Emission factors for BTEX compounds during idle and full load operation.
Comparison with published values 52
Table 7-2. Emission factors for PAH compounds during idle and full load operation.
Comparison with published values 53
Table 7-3. Emission factors for aldehydes compounds during idle and full load
operation. Comparison with published values 54
Table 8-1. Candidate TEQ surrogate compounds from pre-ETV Method 0010
sampling 63
Table 9-1. R2 forPCDD/F TEQ based on 36, 5-min samples at the MWI 75
Table 9-2. PCDD/F TEQ emission rates from HRGC/HRMS analyses 76
Table 10-1. Emission factors of M1097 and Ml 114 HMMWVs 84
Table 11-1. Engine power settings tested 94
Table 11-2. Summary of F-15 engine tests 97
Table 11-3. Test schedule for the F-22 engine tests 101
Table 11-4. A comparison between Summa canister, REMPI-TOFMS and PTR-MS data
with the PTR-MS average calculated from provided Battelle data and
canister sampling time 102
List of Figures
Figure 2-1. REMPI-TOFMS instrument 5
Figure 2-2 (a, b, c). REMPI ionization principles 6
Figure 2-3. Top View of the Movable Inlet Mounting Plate 10
Figure 2-4. Schematic ofLRlO compact 19" rack-mount reflectron TOFMS 11
Figure 2-5. Two views of large REMPI-TOFMS system 11
Figure 2-6. Compact REMPI-TOFMS instrument 12
Figure 2-7. Detection of multiple analytes using wavelength-dependent ionization 13
Figure 2-8. Transient benzene concentrations detected in vehicle exhaust while running
a dynamometer-based RS 13
Figure 2-9. REMPI-TOFMS mass spectrum at benzene (78) wavelength as observed in
vehicle exhaust 14
Figure 2-10. Schematic setup of LIBS system 15
IV
-------
EPA/600/R-10/050
April 2010
Figure 2-11. Setup ORS during sampling 16
Figure 3-1. Wavelength emitted from laser with increased ambient temperature
(increase in temperature for scan 1 to 9) 18
Figure 3-2. Snapshot of data acquisition software in operation 20
Figure 5-1. U.S. Marine Corps tactical utility generator 23
Figure 5-2. A 3D survey scan (mass, wavelength, ion signal) 24
Figure 5-3. Time-of-flight mass spectrum recorded at (A) optimum styrene (287.7 nm)
and (B) Benzene (259.0 nm) ionization wavelength in diesel generator
exhaust 25
Figure 5-4. Comparison of 1+1 REMPI wavelength spectrum recorded for toluene in
diesel generator exhaust and from calibration gas standard (the latter scaled
to equal intensity) 26
Figure 5-5. Monitored REMPI-TOFMS signal for reference gas (C6D6) and benzene in
exhaust diesel generator 26
Figure 5-6. On-line REMPI-TOFMS ion signal traces of nine individual target analytes,
labeled with mass, chemical name and used, most favorable ionization
wavelength 27
Figure 5-7. Recorded ion signals during nine repeats of the analyte cycle 28
Figure 5-8. Observed day to day variations in estimated concentrations of BTEX-like
analytes in exhaust diesel generator 29
Figure 5-9. Comparison between extractive and REMPI-TOFMS concentrations for
three exhaust analytes 30
Figure 5-10 (a, b). (a) Real-time transient benzene emissions following a cold start, (b)
Normalized ion signal traces for single aromatic ring (thin lines) and double
aromatic ring (fat lines) analytes 32
Figure 5-11. Transient REMPI-TOFMS results 33
Figure 6-1. Cold and hot start-up emissions of benzene, naphthalene, CO, and CO2 with
exhaust temperature 38
Figure 6-2. Temporal temperature and particle-bound PAH concentrations in the diesel
exhaust 41
Figure 6-3. Load change emissions of benzene, naphthalene, CO, and CO2 with exhaust
temperature 42
Figure 6-4. Temperature distribution and particle-bound PAH concentrations under
varying loads 43
Figure 7-1. Aircraft ground equipment outside EPA facilities 45
Figure 7-2. REMPI-TOFMS time-resolved trace of cold startup, idle (low) load, and
high load conditions 47
Figure 7-3. REMPI-TOFMS wavelength-resolved trace during idle (low) load condition.
48
Figure 7-4. Comparison of (selected) AGE wavelength spectra versus the TO-14
calibrated gas standard 49
-------
EPA/600/R-10/050
April 2010
Figure 7-5. Comparison between GC/MS and REMPI-TOFMS analysis of summa
canister gases 50
Figure 7-6. Comparison between REMPI-TOFMS measurements at low load and TO 15
method analyses by two laboratories 50
Figure 7-7. REMPI-TOFMS and CO and CO2 CEM concentrations during a series of
startups (the initial startup was a cold start) and shutdowns (for refueling),
operating at high, high, then idle load. Range of data (barely visible) are
shown for benzene 51
Figure 7-8. Comparison of startup responses of benzene and naphthalene for the AGE
and diesel generator 56
Figure 7-9. ORS measurement configuration 57
Figure 7-10. Comparison of the reference spectrum of formaldehyde (blue trace) to a
measured spectrum of the turbine plume (red trace) 59
Figure 7-11. Comparison of the reference spectrum of ethylene (red trace) to a measured
spectrum of the turbine plume (blue trace) 59
Figure 7-12. Comparison of the reference spectrum of n-octane (red trace) to a measured
spectrum of the turbine plume (blue trace) 60
Figure 7-13. Comparison between CEM and ORS measurements 60
Figure 8-1. Marine package boiler at EPA facilities 62
Figure 8-2. Pre-ETV phase determination of MClBz as a PCDD/F TEQ surrogate 65
Figure 8-3. REMPI-TOFMS mass spectrum of MClBz with inset of relevant mass range.
0.1 ppb for a 4 min averaging period, S/N = 3 65
Figure 8-4. ETV phase comparison of TEQ from Method 23 and MClBz prediction
from REMPI-TOFMS measurements 66
Figure 8-5. Comparison of M-10 and time-integrated REMPI predictions of PCDD/F
TEQ with actual M-23 measurements 67
Figure 9-1. Comparison between real time results for PAHs acquired with REMPI-
TOFMS at identical Boiler A and Boiler B for two 4-h intervals that were
typical of the normal (steady state) operation 71
Figure 9-2. Emissions of PAHs during shutdown of boiler. Break in data during filter
change 71
Figure 9-3. A 3-dimensional qualitative depiction PAH emissions during the startup of a
boiler (at ~t= 10 min) 72
Figure 9-4. Emissions of selected PAHs and monochlorobenzene during startup of
boiler (at-9:00 AM) 72
Figure 9-5. Comparison between REMPI-TOFMS time averaged concentrations and
those from conventional Method 0010 sampling during the course of the
sampling campaign 74
Figure 9-6. Mass spectrum recorded at monochlorobenzene wavelength (~ 270.8 nm) 74
Figure 9-7. PCDD/F TEQ, conventional 1,2,4 triClBz and mass 18- REMPI-TOFMS
ion signal over the coarse of the four day sampling campaign 77
Figure 9-8. Recorded REMPI-TOFMS ion signals for mass 180/182/184 78
VI
-------
EPA/600/R-10/050
April 2010
Figure 10-1. Emission transients for the M1097 and Ml 114 HMMWVs. A: West
Virginia and B: Highway HWFET cycles 80
Figure 10-2. Steady-state benzene, naphthalene, methylnaphthalene, CO, and CO2
emissions from the M1097 and Ml 114 HMMWVs during a velocity and
gradient stepped cycle 86
Figure 10-3. Cold start emissions of benzene, naphthalene, methylnaphthalene, CO, and
CO2 and PM size distribution for the M1097 (left) and Ml 114 (right)
HMMWVs 87
Figure 10-4. Steady state organic emissions at analyte-specific wavelengths, A: Ml
Abrams, low and high idle conditions, and B: Bradley IFV, idle 0 and 2
trials. Inset: real-time variance of organics at low idle 88
Figure 10-5. Cold and warm start emissions of benzene, naphthalene, and
methylnaphthalene for the Bradley IFV and Ml Abrams 89
Figure 10-6. Benzene, naphthalene, and methylnaphthalene emissions during a shutdown
of the Ml Abrams 89
Figure 10-7. Steady state PM size distribution for the M1097 HMMWV (top) and Ml 114
HMMWV (bottom) during a stepped velocity and gradient cycle 91
Figure 10-8. Steady state REMPI, ELPI, and CEM data for the M1097 (left) and Ml 114
(right) 92
Figure 11-1. Trimpad and location of probe and instrument trailers with respect to aircraft 95
Figure 11-2. Concentration profiles during F-15 engines testing 98
Figure 11-3. Concentration profiles during second sequence of F-15 engine testing 98
Figure 11-4. Concentration profiles during the 30 minute test 2B1 99
Figure 11-5. Complete 2B series 99
Figure 11-6. Concentration during first sampling of F-22 engine 103
Figure 11-7. Concentrations during 4A and 4B sampling of F-22 engine 103
Figure 11-8. Concentrations for other REMPI-TOFMS detectable analytes 104
Figure 11-9. Comparison between REMPI-TOFMS and PTR-MS 104
Figure 11-10. Benzene and PAHs response during afterburner test 105
Figure 11-11. Comparison REMPI-TOFMS with PTR-MS for benzene and naphthalene.
106
Figure 11-12. Comparison between time averaged REMPI-TOFMS benzene data and
Summa canister results 106
Figure 11-13. Benzene concentrations as function of thrust level F-22 engine 107
Figure 11-14. Benzene emission factor (ug/g carbon burned) for F-22 engines 107
vn
-------
EPA/600/R-10/050
April 2010
AFB
AGE
APCS
APIMS
APU
AVG
AWMA
AZ
BTEX
CAA
CEM
CLS
CO
C02
DNPH
DoD
DOE
ECD
ELPI
EPA
ETV
FID
FTIR
FWHM
GC
GC/MS
GHz
HAP
HEPA
HMMWV
HPLC
HRGC
HRMS
HWFET
IFV
LACEA
LfflS
LRMS
LTA
MHz
MS
MWC
ND
NDIR
List of Acronyms
Air Force Base
Aircraft Ground Equipment
Air Pollution Control System
Air Permit Information Management System
Auxiliary Power Unit
Average
Air and Waste Management Association
Arizona
Benzene, Toluene, Ethylbenzene, and Xylenes
Clean Air Act
Continuous Emission Monitors
Classical Least Squares
Carbon monoxide
Carbon Dioxide
2,4-Dinitrophenylhydrazine
Department of Defense
Department of Energy
Electron Capture Detector
Electrical Low Pressure Impactor
Environmental Protection Agency
Environmental Technology Verification
Flame lonization Detector
Fourier Transform Infrared Spectroscopy
Full Width Half Maximum
Gas Chromatography
Gas Chromatography/Mass Spectrometry
Gigahertz
Hazardous Air Pollutants
High Efficiency Particulate Arresting
High Mobility Multipurpose Wheeled Vehicle
High Performance Liquid Chromatography
High Resolution Gas Chromatography
High Resolution Mass Spectroscopy
Highway Fuel Economy Test
Infantry Fighting Vehicles
Laser Applications to Chemical and Environmental Analysis
Laser Induced Breakdown Spectroscopy
Low Resolution Mass Spectrometry
Low Temperature Ashing
Megahertz
Mass Spectrometry
Municipal Waste Combustor
Non Detect
Non-dispersive Infrared Analyzer
Vlll
-------
EPA/600/R-10/050
April 2010
NIST
NO
NSD
OLGC
OP-FTIR
OPO
ORS
PAC
PAH
PAS
PCDD
PCDD/F
PIC
PM
PTFE
PTRMS
PTR-MS
RA
RDF
REMPI
REMPI-TOFMS
RPM
RSD
RTF
RWS
S/N
SDA
STP
SVOC
TEF
TEQ
TIC
TOFMS
U.S.
UV
VOC
National Institute of Standards and Technology
Nitric Oxide
Number Size Distributions
On Line Gas Chromatography
Open Path Fourier Transform Infrared
Optical Parametric Oscillators
Optical Remote Sensing
Path Average Concentration
Polycyclic Aromatic Hydrocarbons
Photoelectric Aerosol Sensor
Polychlorinated Dibenzodioxins Dibenzofurans
PolyChlorinated Dibenzo-p-Dioxin and polychlorinated
dibenzoFuran
Products of Incomplete Combustion
Particulate Matter
PolyTetraFluoroEthylene
Proton Transfer Reaction Mass Spectrometer
Proton Transfer Reaction - Mass Spectrometry
Relative Accuracy
Refuse Derived Fuel
Resonance Enhanced Multi Photon lonization
Time of Flight Mass Spectrometry
Revolutions per Minute
Relative Standard Deviation
Research Triangle Park
Roadway Simulator
Signal Noise
Spray Dryer Catwalk Area
Standard Pressure
Semi Volatile Organic Compounds
Toxic Equivalency Factor
Toxicity Equivalent
Tentatively Identified Compounds
Time of Flight Mass Spectrometry
United States
Ultraviolet
Volatile Organic Compounds
IX
-------
EPA/600/R-10/050
April 2010
Keywords:
Resonance-enhanced multiphoton ionization spectroscopy
REMPI
Time of flight mass spectrometry
TOFMS
Laser-induced breakdown spectroscopy
LIBS
Open path measurements
Air toxics
Air pollutants
Measurement
Emission factors
Diesel generator
Auxiliary power unit
Municipal waste combustor
HMMWV
Abrams tank
Bradley Infantry Fighting Vehicle
F-15
F-22
-------
EPA/600/R-10/050
April 2010
Acknowledgements
The investigators and authors would like to extend their appreciation and acknowledge to the
SERDP organization for funding this research program, and would also like to thank the
following organizations and individuals that contributed to the success of this project:
Drs. Abderrahmane Touati and Lukas Oudejans of Arcadis U.S., Inc. enacted the testing
programs and performed the organic measurements throughout this project.
Drs. Harald Oser and Michael Coggiola of SRI International were critical providers of the
resonance enhanced multi photon ionization (REMPI) technology and able consultants
throughout the program.
Dr. Andrzej Miziolek of the Army Research Laboratory was the primary motivator of the
laser induced breakdown spectroscopy (LIBS)-related work.
Drs. Shannon Serre and Emily Gibb-Snyder (United States (U.S.) Environmental
Protection Agency (EPA), National Homeland Security Research Center) provided
critical expertise with the LIBS instrument and experimentation.
Messrs. Whaley, Huffman, and Winberry from U.S. Marine Corps at Camp Lejeune for
loan of the tactical generator
Col. Steven Aylor, SMSgt David Caldwell, and Sgts. Thomas Clemens, George Garnot,
and James Scaccia from Pope Air Force Base (AFB) for the loan of the AGE.
Dr. Ken Cowen from Battelle Memorial Institute
Messrs. Michael Barnett and Jeff Landrum from the Southern Public Service Authority
Power Plant (Portsmouth, Virginia)
Messrs. William Bolt, Jason Jack, and Gregg Schultz from the Aberdeen Test Center,
Maryland, for access to vehicles and the Roadway Simulator (RWS).
Dr. Howard Mayfield from Tyndall AFB, Panama City, Florida
XI
-------
EPA/600/R-10/050
April 2010
Abstract
This project tested three, real-/near real-time monitoring techniques to develop air toxic emission
factors for Department of Defense (DoD) platform sources. These techniques included:
resonance enhanced multi photon ionization time of flight mass spectrometry (REMPI-TOFMS)
for organic air toxics, laser induced breakdown spectroscopy (LIBS) for metallic air toxics, and
optical remote sensing (ORS) methods for measurement of criteria pollutants and other
hazardous air pollutants (HAPs). Conventional emission measurements were used for
verification of the real-time monitoring results. The REMPI-TOFMS system was demonstrated
on the following:
a United States U.S. Marine Corps (USMC) diesel generator,
a U.S. Air Force auxiliary power unit (APU),
the waste combustor at the Portsmouth Naval Shipyard, during a multi-monitor
Environmental Technology Verification (ETV) test for dioxin monitoring systems,
two dynamometer-driven high mobility multi-purpose wheeled vehicles (HMMWVs),
an idling Abrams battle tank,
a Bradley infantry fighting vehicle (TFV), and
an F-15 and multiple F-22 U.S. Air Force aircraft engines.
LIBS was tested and applied solely to the U.S. Marine Corps diesel generator. The high detection
limits of LIBS for toxic metals limited its usefulness as a real time analyzer for most DoD
sources. ORS was tested only on the APU with satisfactory results for non-condensable
combustion products [carbon monoxide (CO), carbon dioxide (CO2)] but with limited success on
condensable volatile organic by-products. This program demonstrated the ability to measure
trace aromatics with REMPI-TOFMS in harsh environments and with a high degree of accuracy
and precision.
xn
-------
EPA/600/R-10/050
April 2010
Executive Summary
The Department of Defense (DoD) quantifies air pollutant emissions from its facilities and
weapon platforms in order to identify potential sources needing remediation and to comply with
base permitting requirements. These pollutants include the so-called criteria pollutants such as
nitrogen oxides and particulate matter (PM) as well as trace air toxics. This information further
enables DoD to employ measures such as adoption of preventive operational modes or
equipment substitution in order to minimize disruption of training and operational activities.
Information on source air toxics is particularly limited, primarily due to lack of automated
methods of sample analysis, underscoring the need for innovative approaches, instruments, and
methods to achieve these ends. Since a large portion of DoD sources are mobile sources that
operate under non-steady state modes, most of the current extractive methods cannot resolve
modal or temporal changes in emissions that result from such type of sources, leading to gaps in
the overall DoD air toxics emissions inventory. The United States Environmental Protection
Agency (EPA), under a DoD Strategic Environmental Research and Development Program,
"Source and Ambient Air Monitoring for DoD Operations (WP/CP1247)," initiated and
completed a research project to develop and validate the latest state-of-the-art technologies for
these measurements. These technologies would provide modal and time resolved measurements
of air toxics emissions from various types of point- and mobile-sources. Such information could
be used, for example, in the Air Force's Air Permit Information Management System (APIMS),
an emission inventory system currently used by Hill AFB and being adopted DoD-wide, as well
as to improve EPA's AP-42 emission factor system. This report presents the results of this
project.
The project objective was approached through the applied development of a combination of three
unique, versatile, field-ready, real or near real-time monitoring techniques that can measure trace
air toxic levels: resonance enhanced multi photon ionization time of flight mass spectrometry
(REMPI-TOFMS) for organic air toxics, laser induced breakdown spectroscopy (LIBS) for
metallic air toxics, and optical remote sensing (ORS) methods for verification and measurements
of criteria pollutants and other hazardous air pollutants (HAPs). Conventional emission
measurements were used for verification of the real-time monitoring results and assessment of
their accuracy. A fast sampling method that separates particles from gases was successfully
integrated in an overall sampling scheme designed to conserve the integrity of the sample, avoid
any particle/gas partitioning, while bringing the quantity of target compounds above the
detection limit of the REMPI-TOFMS measurement technique.
Satisfactory development of the REMPI-TOFMS system, from laboratory to a field-ready real or
near real time monitor of trace organic aromatic pollutants system, was performed at the EPA
research facility in Research Triangle Park (RTF), North Carolina. The system has been
demonstrated on the following:
a U.S. Marine Corps (USMC) diesel generator,
a U.S. Air Force auxiliary power unit (APU),
the waste combustor at the Portsmouth Naval Shipyard, during a multi-monitor
Environmental Technology Verification (ETV) test for dioxin monitoring systems,
Xlll
-------
EPA/600/R-10/050
April 2010
two dynamometer-driven High Mobility Multi-purpose Wheeled Vehicles (HMMWVs),
an idling Abrams battle tank,
a Bradley infantry fighting vehicle (IFV), and
an F-15 and multiple F-22 U.S. Air Force aircraft engines.
LIBS was tested and applied solely to the U.S. Marine Corps diesel generator. The high detection
limits of LIBS for toxic metals limited its usefulness as a real time analyzer for most DoD
sources, resulting in the withdrawal of the U.S. Army Research Laboratory from the SERDP
project. ORS was tested only on the APU with satisfactory results for non-condensable
combustion products [carbon monoxide (CO), carbon dioxide (CC^)] but with limited success on
condensable volatile organic by-products thereby eliminating the opportunity to directly compare
the ORS and REMPI-TOFMS technology.
During the SERDP funding period 2003-2008, the following (in approximate chronological
order) achievements were made:
A large laboratory scale size REMPI-TOFMS system has been applied to the exhaust of
an USMC diesel generator. Steady state benzene, toluene, ethylbenzene, and xylenes
(BTEX) emissions on the order of 100 ppb were observed. Cold start benzene emissions
were observed on the order of 14 ppm, which lasted ~ 20-30 seconds. The REMPI-
TOFMS results were successfully compared to measurements obtained by conventional
certified EPA methods, and validated the system as responsive and functional with
complex mixtures. LIBS results from the USMC diesel generator study were found to be
unsatisfactory due to the lack of sensitivity for real time detection of metals in the
exhaust gas and limited detection of metals on soot loaded filters collected during the
sampling of the exhaust. LIBS plus low temperature ashing (LTA) showed positive
results, improving metal discrimination from highly carbonaceous backgrounds.
However, the examined DoD sources to date did not show sufficient metal emissions to
be quantified by LIBS with LTA.
Testing by SRI International on a compact REMPI-TOFMS indicated a minimal sacrifice
of system quality and so the small system was approved for design and construction by
SERDP. A first compact REMPI-TOFMS system was delivered to EPA in April 2004.
REMPI-TOFMS emission measurements were completed on an Auxiliary Power
Unit/Aircraft Ground Equipment (APU/AGE). This was a turbine engine compressor
from Pope AFB (type A/M32A-95). Modal-dependent results for benzene, toluene,
ethylbenzene, o-, m-, p-xylenes, and styrene showed excellent agreement with standard
EPA sampling methods and gas chromatography/mass spectrometry (GC/MS) analysis.
The ORS system was applied to this source, and measurements for criteria pollutants
such as CO2 and CO emissions were completed. In addition to these species, absorption
bands for formaldehyde, ethylene, and aliphatic mixtures that originated from the fuels
were found in the measured spectra.
A prototype compact REMPI-TOFMS developed under another project was taken into
the field (December 2004) to the Portsmouth Naval Shipyard waste combustor and tested
xiv
-------
EPA/600/R-10/050
April 2010
on the flue gas prior to the air-pollution-cleaning-device. The sampling system platform
consisted of the REMPI-TOFMS system and a novel rotating filtering system designed to
conserve the integrity of the sample and avoid or minimize any particle or gas
partitioning. Real time detection of various small poly cyclic aromatic hydrocarbons
(PAHs) and monochlorobenzene in the flue gas of the combustor was accomplished.
Results from parallel conventional sampling for dioxins and furans and semivolatiles
were used to establish correlations between the aforementioned analyte classes.
The design for the SERDP-funded REMPI-TOFMS field unit was submitted in February
2004 as a required SERDP "Go/No Go" deliverable. The design consisted primarily of a
compact laser system and rack mounted TOFMS for organics measurements, with two
distinct inlet types for increased flexibility. The simpler inlet system could be used with a
fixed frequency laser system.
The compact REMPI-TOFMS system was delivered to EPA in May, 2005 after an initial
evaluation at SRI International.
The REMPI-TOFMS system participated in an EPA ETV test at the EPA laboratories in
RTF, NC. The ETV test verified the performance of four different detection systems for
chlorinated dioxins, trace air toxics. This was the first fully international ETV test, with
participants from Japan, Austria, and Germany complementing the EPA-SRI
International participation. The ETV verification reports are available at
http://www.epa.gov/etv/vt-ams.html#dems.
The capabilities of REMPI-TOFMS towards detection of higher-chlorinated benzenes
were extended after the addition of a second (fixed) wavelength option for a two color
REMPI approach on one of the two compact systems. Results showed minimal loss in
REMPI-TOFMS sensitivity with increasing level of chlorination while maintaining the
ability to separate individual isomers.
The REMPI-TOFMS instrument was taken to the Aberdeen Test Center's RWS for
sampling of exhausts from two different (Ml097 and Ml 114) HMMWVs as driven on
various simulated roadway profiles as well as steady state velocity profiles. It included
parallel sampling for CO, CO2 and PM [via an electrical low pressure impactor, (ELPI)].
Additional data were obtained from the exhausts of Ml Abrams and Bradley track
vehicles under start-up, shutdown, and stationary idle conditions. A journal paper
detailing the performance of REMPI-TOFMS in characterizing real-time air toxic
emissions during the Aberdeen RWS tests has been submitted to Atmospheric
Environment entitled "Transient PAH, PM, CO, and CO2 Emission Responses from
HMMWVs, the Ml Abrams tank, and the Bradley Infantry Fighting Vehicle" by Brian
Gullett, Lukas Oudejans, and Abderrahmane Touati (2009).
Two compact REMPI-TOFMS instruments were taken into the field (December 2006) to
the Portsmouth Naval Shipyard waste combustor and tested on the flue gas before the gas
entered the air-pollution-cleaning-system. One instrument monitored PAHs while the
other (two-color) REMPI-TOFMS instrument was focused on real time detection of
1,2,4-trichlorobenzene as a potential indicator of dioxin toxicity based on previously
obtained results at the same combustor facility.
xv
-------
EPA/600/R-10/050
April 2010
The REMPI-TOFMS instrument was deployed to Tyndall AFB, Florida, for sampling of
jet engine exhausts in collaboration with Battelle Memorial Institute. Emission factors
were obtained for individual engines from an F-15 jet aircraft and four F-22 aircraft as
operated on a trimpad. Results were consistent and found to be in good (benzene) to fair
(naphthalene) agreement with proton transfer reaction MS (PTRMS), another real time
detection method for these compounds. A comparison of the time averaged REMPI-
TOFMS data with conventionally sampled summa canisters was good across the whole
(benzene) dataset. The results are currently (July, 2009) being written into a paper prime-
authored by Battelle.
xvi
-------
EPA/600/R-10/050
April 2010
Objectives
This project applied state-of-the-art trace pollutant detection technologies for the determination
of emission factors for Department of Defense (DoD) sources. Results from these technologies
were to be verified via comparisons with conventional extractive sampling measurement
techniques. The ultimate results will provide temporal and spatial measurements of the air toxics
at tested DoD sources. The state-of-the-art technologies are:
Jet Resonance Enhanced Multiphoton lonization (REMPI)-Time of Flight Mass
Spectrometer (TOFMS)
Laser Induced Breakdown Spectroscopy (LIBS)
Optical Remote Sensing (ORS)
The United States Environmental Protection Agency (EPA) regulates emissions of air toxics
under the EPA - Clean Air Act (CAA) and DoD needs to address the potential impacts of these
regulations on its operations. Identification of potential DoD sources which contribute to ambient
air toxic levels, mobile sources in particular, will permit DoD to devise strategies to control and
minimize emissions of air toxic pollutants from its facilities and from its on-road and non-road
sources. Currently, the major DoD air emission database for toxic air compounds is very limited
in scope since measurement methods for many of the 188 air toxics listed in the CAA have not
yet been developed; further, when these measurement methods have been developed, most of
them do not produce temporally and spatially resolved measurements. Spatially and temporally
resolved emission measurements are needed to assess their impact, if any, on ambient air toxic
levels, and to determine what operational modes contribute significantly to these emissions.
xvn
-------
EPA/600/R-10/050
April 2010
1. Background
1.1 Emission Factors
Source-specific emission factors are often required in order to comply with state reporting
regulations or permit requirements which requires a quantitative emissions inventory of more
than 500 species of toxic air pollutants (Levy et al., 1993). Preference is usually given to
sampled information rather than use of generic emission factors. However, both sampling data
and emission factors are limited in breadth since relatively few sources have been characterized
for the approximately 188 defined hazardous air pollutants (HAPs) which include both organic
and metallic compounds (EPA, 2004). This fact reflects the difficulty and cost of assessing a
multitude of emissions from a myriad of sources. A further complication of emissions
quantification is that many of these methods were validated on specific sources and are of
uncertain universal applicability. Very few of the current extractive methods can resolve modal
or temporal changes in emissions that can be related to, for example, load changes or start-ups.
For some sources these non-steady-state emissions are suspected to be a significant portion of the
total air pollution emitted, and hence, their quantification may be important towards determining
average emission factors, pollutant exposure, and mode-specific pollutant minimization.
Protocols for source emissions characterization are a function of sampling and analytical
limitations. Analytical detection limits also determine sampling protocols, primarily related to
sample volume collection time necessary to avoid non-detects. For HAPs, these limitations can
be particularly influential, since HAP concentrations are typically significantly lower than other
pollutants such as criteria pollutants NOX, CO, and 862. Hence, sampling protocols for HAPs
typically require long-term, steady state monitoring and this requirement prevents observation of
HAP transients.
1.2 Monitoring for Organic Toxics
Analyses for trace organic HAPs typically require a 4 h extractive sample taken on an annual or
less frequent basis and lengthy laboratory analyses which are not conducive to prompt feedback.
From a regulatory or public interest viewpoint, this results in an infrequent and potentially
minimally representative monitoring scheme. The conventional methods also provide little
assurance that subsequent emissions remain controlled, especially during periods of transient
upsets, such as startups and shutdowns, when emissions are typically higher.
Real time detection, high sensitivity, and high selectivity are three key requirements toward
process control for combustion-derived HAPs. Recent technological advances in measurement of
HAPs show promise for continuous, real time monitoring. This raises the possibility of
minimizing their formation through feedback to the combustion process control, assurance to the
public of compliance, and minimization of over-design of gas cleaning systems. Since the typical
ppq-level concentrations of the toxic HAP congeners are beyond the detection limits (~ 100 ppt)
-------
EPA/600/R-10/050
April 2010
of developing monitoring technologies, two strategies have been adopted: (1) use of higher
concentration indicator, or surrogate, compounds to infer HAP concentrations of trace HAP
compounds such as polychlorinated dibenzodioxins/dibenzofurans (PCDDs/Fs), and (2)
concentration of emissions on a sorbent followed by a purge and analysis. Indicator compounds
are determined by conventional sampling and analysis (GC/MS) and are chemically similar
compounds to the target HAPs. These indicator compounds may be precursors to PCDDs/Fs or
pollutants formed in similar, parallel reactions. The universality of these indicator compounds
across facility types remains to be investigated, but they likely vary somewhat depending on the
plant type, the waste and fuel types, and plant operating and combustion conditions.
The presence of a real time monitor raises the first possibility to provide operational feedback to
minimize the formation of HAPs, such as polychlorinated dibenzo-p-doxin and polychlorinated
dibenzofuran (PCDD/F), as well as characterization of emissions during operational transients.
This would be an important capability for distinguishing effects of operating modes or air
pollution control equipment failures on HAP emissions. Indeed recent work (Gass, 2002; Gullett
et al., 2006) has demonstrated seven-fold increases in PCDD/F emissions during 1 h combustor
shutdowns and startups, consistent with other work (Gross et al., 2004; Neuer-Etscheidt et al.,
2006) in which PCDD/F raw gas levels increased by one to two orders of magnitude during
transient combustion conditions. The extent to which these transient emissions may affect short-
and long-term stack emission values and, hence, compliance issues, is undetermined. The rapid
variation of PCDD/F, as well as other co-pollutants, due to transients, fuel changes, and
operating variations suggests that fast on-line monitoring is necessary in order to effect changes
in operating conditions that will reduce or prevent conditions favorable to PCDD/F formation.
Work with REMPI-TOFMS at an industrial hazardous waste incinerator was successfully able to
monitor aromatics including monochlorobenzene during the introduction of barrels of liquid
hazardous waste (Heger et al., 2001), finding transient evolution of the pollutants. The ability to
measure facility transients for indicator concentrations with REMPI-TOFMS and to correlate
their levels with toxicity equivalent (TEQ) values has yet to be established.
1.3 Related Real-time Technologies for Monitoring of Trace Organic Pollutants
Only a few studies exist in which low molecular weight hydrocarbons have been analyzed in
detail for diesel trucks, air ground equipment or Army vehicles. The typical extractive method
with GC or GC/MS analysis does not detect transient emission events, such as a cold starts, since
the timescales of the event are much shorter than the typical sampling time of several minutes for
extractive sampling. GC/MS also prevents full speciation of some isomers as in the case of
xylenes. Developing technologies such as SPI-TOFMS or PTR-MS are able to detect most
BTEX compounds as present in gasoline automobile exhaust in real time. However, these
methods lack isomer selectivity and the generally encountered detection limits for SPI-TOFMS
are insufficient to detect a fairly large subset of the aromatic compounds present in modern
vehicle exhaust gas flows.
1.4 Related REMPI-TOFMS Studies on Vehicle Exhausts
The REMPI-TOFMS technique combined with a supersonic expansion has been developed over
the last decade as an isomer selective, sensitive and real time monitor of aromatic organic
compounds. The sensitivity and rapidity of REMPI-TOFMS outperforms current extractive
-------
EPA/600/R-10/050
April 2010
methods, enabling characterizations of transients and immediate feedback to the operator of the
emissions source. The application of REMPI-TOFMS to vehicle exhaust so far has been limited
to studies on exhaust gas emissions from gasoline engines and diesel engines (Franzen et al.,
1995, Weickhardt et al., 1993, Boesl, 1998) emphasizing the real time and selective capabilities
of REMPI-TOFMS without providing extensive emission factors. Laser MS work by Frey et al.
(1995) examined time-resolved concentrations of benzene, toluene, xylenes, and
trimethylbenzene under dynamic engine operation, such as load and speed changes as well as
misfires and fuel mixture changes. Their compact mass spectrometer had a sensitivity of 1 ppm
for aromatic hydrocarbons and a sampling period of 20 ms. Other related work (Nagel et al.,
1996) examined formaldehyde and acetaldehyde emissions. Boesl et al. (1998, 2000) reviewed
developments in REMPI with a time-of-flight mass spectrometer (REMPI-TOFMS) to measure
engine transients on a spark-ignited, four-cylinder gasoline engine coupled to an electronic
dynamometer. Extreme fluctuations of benzene, toluene, and xylenes concentrations were
observed in the exhaust for the first 60 s of operation, after which levels dropped and stabilized.
1.5 Monitoring for Metals
LIBS is a simple, laser-based technique that characterizes the elemental composition of aerosols,
liquids, gases, and solids in real time with a single laser pulse. LIBS instruments typically use a
short pulse-duration laser (~ 3-10 nsec in duration and 20-150 ml of pulse energy) that is focused
through a lens onto a surface or into the volume to be analyzed. The high energy pulse creates a
small plasma at the focal point of the system optics during the pulse. The resulting high
temperature plasma (~ 20,000 K) is sufficient to vaporize, atomize, and electronically excite a
small amount of the sample matter (pg to ng). These excited atoms and molecular fragments
decay primarily by emission of photons whose wavelength spectrum is characteristic of the
atoms in the plasma. The light from the emitting atoms is collected using standard optical
techniques and dispersed in a monochromator or spectrograph depending on the design of the
detector. The resulting frequency spectrum is a fingerprint of the elemental composition of the
sample. Through calibration the intensity of each peak in the spectrum can be related to the
quantity of the element giving rise to the spectral feature.
LIBS development, pioneered by Dr. David Cremers of Los Alamos National Laboratory, has
shown the ability to detect many metals, including beryllium, lead, chromium, and uranium.
ADA Technologies, Inc. has adapted Dr. Cremers' technology in several instruments with the
demonstrated capability of measuring beryllium at levels below 2 ug/m3 in air with fieldable
units. While LIBS is typically considered an in situ measurement, it can also be used to analyze
solids (e.g., Be at 0.1 ug on a filter), enabling it to analyze PM-laden filters. This will be
important to this effort, as many metals are expected to be below detection limits for real time
sampling.
1.6 Optical Path Methods and Field Measurements
Researchers have developed strategies over the past decade to make measurements of gaseous
and particulate pollutant emissions from agricultural operations and transportation activities
using novel ORS and conventional point sampling techniques. This has included open-path
Fourier transform infrared (OP-FTIR)) and laser-based technologies to estimate gaseous and PM
emission fluxes and to map air pollution. OP-FTIR has the capability of identifying and detecting
a wide range of gases and is an accepted technique to measure gaseous air toxics and volatile
-------
EPA/600/R-10/050
April 2010
organic compounds (VOCs). OP-FTIR instruments pass an infrared light along an open optical
beam path up to 1 km long to measure and identify chemical contaminants directly in the field.
This yields real-time data (<1 minute per sample) for multiple chemical species (25 of the 33
UAT compounds) typically with ppb detection levels. Detection limits for chemicals measured
by OP-FTIR systems will vary depending on the chemical species, atmospheric conditions
(humidity and temperature), and whether interfering compounds are present. Typical system
detection limits for a 100 meter separation between the transmitter/receiver telescope and the
retroreflector is from 0.1 to 15 ppb for most infrared-active chemicals. Ambient values for many
of the air toxics of interest are at or below these levels but these values represent integrated
concentrations collected over significant periods of time and cannot be temporally resolved. In
other words, using conventional ambient measurement methods, one cannot differentiate
between low background levels that remain constant and short-duration "spikes" in concentration
that occur infrequently during the entire sampling period. The strength of ORS techniques lies in
their ability to measure changes in pollutant concentration on a near-real time scale. Our recent
work, as well as developmental work done by instrument manufacturers indicates that for many
of the air toxics of interest, event-related source emissions will be measurable using OP-FTIR.
-------
EPA/600/R-10/050
April 2010
2. Description of Equipment and Methods
Aside from the conventional continuous emission monitors (CEM) for real time measurement of
CO, CC>2 and C>2, three complementary technologies were used during this research project,
namely REMPI-TOFMS, LIBS, and ORS. A description of these three methods is provided in
the next sections.
2.1 REMPI-TOFMS
The REMPI-TOFMS instrument consists of multiple components as shown in Figure 2-1.
Exhaust
Time-of-Flight Mass Spectrometer
MicroChannel plate Detector
REMPI-TOFMS
Figure 2-1. REMPI-TOFMS instrument.
REMPI is tailored as a wavelength-selective method of creating ions for mass separation and
detection in a TOFMS. The REMPI approach to ionize molecules was established many years
ago in academia for identification of reaction intermediates and end products. As the acronym
-------
EPA/600/R-10/050
April 2010
alludes to, the ion yield is enhanced when the laser wavelength is found to be in resonance with
an intermediate molecular transition. Depending on the wavelength used, such intermediate state
can be reached in a single step or with multiple albeit non-resonant photon steps. Depending on
the internal energy level configuration, ionization may occur after absorption of single or
multiple ions with the same initial color (one-color REMPI) or a different color (two-color
REMPI). Ionization schemes are typically denoted as a combination of two numbers that are
characteristic for the number of photons to reach the intermediate resonant state (first number)
plus the number of photons to reach ionization (second number). For example, 1+3 REMPI
means that one photon is needed for reaching the intermediate state and three photons are needed
to create an ion from this intermediate state. The REMPI method of ionization as used during
this SERDP project is always 1+1 REMPI (using a single wavelength) or 1+1' REMPI (using
two wavelengths as indicated by the ' ).
In general, the lowest electronic state for small aromatic organic analytes (Si) is located between
4.1 and 5.0 eV. This corresponds to a wavelength range between 300 and 250 nm [ultraviolet
(UV) range of the spectrum] when this lowest electronic state is reached using a single photon or
alternatively between 600 and 500 nm (visible light of the spectrum) when using two photons to
reach this electronic state. Since in the latter case the intermediate electronic state is reached
through a non-resonant process, the ionization efficiency is typically much lower while much
more laser energy is needed to accomplish such two photon absorption. Therefore, with the main
objective to create a highly sensitive detection method, the two photon absorption to reach the
intermediate state was not considered. The subsequent absorption of a second UV photon leads
to ionization of the analyte if the ionization energy is below twice the photon energy as
illustrated in Figure 2-2 (a). This is the case for the 1+1 REMPI ionization scheme. If the
ionization energy is larger than twice the photon energy [Figure 2-2 (b)], no ionization occurs
and the analyte will not be detected using TOFMS. This barrier can be overcome by using a
second, more energetic shorter wavelength photon [Figure 2-2 (c)]. This is the case for a 1+1'
REMPI ionization scheme.
Ionization Continuum
A
Ionization Continuum
Ionization Continuum
A
a b
Figure 2-2 (a, b, c). REMPI ionization principles.
-------
EPA/600/R-10/050
April 2010
So far, the REMPI process has been described as if the intermediate electronic state is a single
discreet state. In reality, each electronic state, including the electronic ground state (So) consists
of numerous discrete states due to the internal vibration and rotation of the analyte. This leads to
an energy level structure that is unique for each molecule. Consequently, when the wavelength of
the absorbing photon is changed, multiple discrete wavelengths will exist within a fairly short
wavelength window where ionization can take place. Such wavelength scan yields an
absorption/ionization spectrum that can be considered as a fingerprint of the target analyte and
are even isomer specific.
There are four main reasons why a wavelength spectrum may not show these transitions as
discrete, sharp, lines. First, the lifetime of the excited state may be extremely short
(microseconds or less) which results in a (lifetime) broadening of the energy transitions, or
spectral lines. Second, insufficient cooling in the expansion can broaden transitions. Third, the
actual ionization scheme may involve ionization via a higher electronic (SN; N > 1) state with a
higher density of available states, and fourth, the laser source of ionization cannot be fine tuned
over a short wavelength range due to its own linewidth. The latter aspect appears when a
comparison is made between the two laser systems that were used throughout this SERDP
project. The combination of these four reasons means isomer selectivity cannot always be
accomplished using REMPI. This is especially the case for heavier aromatic analytes like larger
PAHs.
REMPI as applied here is a soft ionization technique. Therefore, no significant fragmentation
takes place and a mass spectrum essentially consists of the parent ion only.
The single color 1+1 REMPI process at the lowest wavelength used of 250 nm also excludes
detection of molecules with ionization energies above 9.2 eV. Consequently, interferences with
aldehydes, for example, can be ignored since most have ionization energies well above this
value. Note that the ionization efficiency of higher order REMPI processes is significantly
smaller than that of 1+1 REMPI. It also requires higher power densities that are not present in the
unfocussed laser beam. For the same reasons, the far more abundant exhaust gases such as
nitrogen, steam and CC>2 cannot be ionized and, therefore, do not interfere.
2.1.1 Laser Systems
Optical parametric oscillators (OPO) are solid-state devices which use non-linear optical
conversion to provide tunable laser light over a very broad wavelength range covering the visible
and IR spectrum. In combination with a frequency doubling module, UV light can be created
from the OPO signal beam. The technology eliminates the need for handling, replacement and
swapping of multiple dyes that are used in the nearest competitive laser technology, the dye
laser. By doing away with the need to change between multiple different laser dyes in order to
scan across a broad tuning range, the OPO provides an extremely flexible system and facilitates
measurements that are not otherwise possible to perform in a timely manner.
Two laser systems were used during the SERDP funding period. The initial tests were performed
with a Continuum Sunlite EX OPO with FX-1 UV frequency extension. This OPO is pumped by
the third harmonic (355 nm) of a Powerlite Precision 9000 Nd:YAG laser system. Its two-stage
-------
EPA/600/R-10/050
April 2010
design combines a narrowband oscillator with a high efficiency optical parametric amplifier to
produce coherent light between 445 and 1750 nm. A computer control system maintains
optimum crystal performance (phase matching) while moving over wide wavelength ranges. UV
output energy, using the signal beam output and the frequency doubler, is between 5 and 10 ml
across the tuning range, with between 5 ml and 7 ml in the 250-280 nm range. The laser spectral
line width was tested by the manufacturer to be ~ 0.2 cm"1.
The major drawback of this Continuum laser system is its size; it just fits on a 3'x 8' table which
makes it too large for field deployment. Therefore, the smaller REMPI-TOFMS systems (about
2' x 4') utilized a more compact laser system. OPOTEK (Carlsbad, CA) manufactures efficient
compact and widely tunable solid-state laser systems based on its patented OPO designs. The
OPOTEK system is designed to be compact, and rugged. Their current Vibrant II-355nm with
frequency-doubling laser system has sufficient power for sensitive measurements and produces
continuously tunable light across a wide range of wavelengths to accommodate the detection of a
large variety of species. This system is also compact and rugged enough to be used during field
experiments. The standard Vibrant 11-355 nm laser system from OPOTEK is a compact all-in-
one tunable laser system which can generate tunable output from 210 nm to 2.6 jim, based on
OPO technology. Due to the all-solid state nature of this system, there is no need to circulate or
change laser dyes that have typically been used to generate tunable light at these wavelengths.
The system consists of the Nd:YAG pump laser, associated second and third harmonic
generation, OPO, UV extensions, and control electronics all on one rigid frame. This packaging
not only leads to a compact unit, but also significantly adds to the ruggedness of the product by
eliminating the alignment problems that come from the scattered discrete components usually
found in this type of system. These attributes made this system an ideal starting-point for the
integration into a field capable system. The UV output energy is between 1.5 and 3.5 ml across
the tuning range, with between 2.0 ml to 3.5 ml in the 250 to 280 nm range used during our
preliminary studies. The laser spectral linewidth was tested by the manufacturer to be
approximately 4 cm"1. The assessment of the increase in linewidth with respect to the large laser
system is part of the evaluation.
When using a single color wavelength, the use of the REMPI method is restricted to analytes
with an intermediate electronic state with an energy value above half that needed to create an
ion. In the case of multiple chlorinated aromatics, the intermediate state lies below half the
ionization energy and no ions are formed. In order to overcome this limitation, one of the
OPOTEK laser systems was modified to include the generation of a higher energy laser photon
as a second color / wavelength option for ionization. This is accomplished by generating 213nm
(fifth overtone of fundamental of Nd:YAG) laser light through the mixing of residual 355 nm
and 532 nm laser light onto a BBO crystal. This method generates 2 to 3 ml of this highly
energetic laser light which is combined with the tunable UV from the OPO system using a
dichroic mirror. With the combination of the tunable UV and fixed wavelength 213 nm laser
light, detection of higher chlorinated benzenes (up to hexachlorobenzene), chlorinated furans (up
to tetrachlorofuran) have been confirmed with only moderate losses in sensitivity with increased
chlorination.
-------
EPA/600/R-10/050
April 2010
2.1.2 Valve Inlet Systems
Under atmospheric conditions, individual (ro)-vibrational lines are broadened due to collisions
with abundant molecules like nitrogen, oxygen and air. Consequently, wavelength spectroscopy
under these conditions will result in unresolved features which lead to low or no wavelength
selectivity. This applies for recorded wavelength spectra when using an effusive inlet source.
Such loss in selectivity can be avoided by the use of a pulsed valve inlet system. The adiabatic
expansion into vacuum creates a supersonic jet in which rotational and vibrational cooling of the
target analytes takes place. Consequently, the population in the electronic ground state that was
originally spread over a large number of (ro-) vibrational states is transferred into only the lowest
(ro-) vibrational states. This cooling results in more discernible spectra, enhancing the selectivity
in comparison with effusive-source ionization spectra. In addition, since the ion signal is
proportional to the population in the initial state, transitions starting from this smaller set of
populated states appear stronger than in the case where the population is spread over a larger
number of initial states.
All results described in this report are obtained using Parker-Hannifm Corporation Series 99
Pulse Valves that are mounted in various ways as described here. During the first part of the
SERDP research efforts with the large time of flight mass spectrometer, four valves were
mounted onto a single stainless steel block. Because the valve itself serves as a vacuum seal
between the sample inlet line and the ion source chamber, removing the valve necessitates
venting the entire instrument. While a sliding gate valve could be incorporated between the exit
of the pulsed valve and the ion source, such an arrangement adds undesirable distance between
the exit orifice and the ionization region. As an alternative to this approach, SRI International
had previously developed a configuration that incorporates four pulsed valves on a single sliding
mount. This arrangement allows any one of the four valves to be aligned with an entrance
channel to the source region while keeping the remaining three valves isolated from the vacuum,
and hence free to be removed and replaced as necessary. However, the size of this four-valve
configuration restricts its use to the larger ion source chamber.
Because of the migration of the laser ionization mass spectrometer to a smaller platform, a new
multi-valve design was required. In order to incorporate a combined valve/GC capillary inlet unit
on this instrument, it was necessary to have the manufacturer modify the valve-mount vacuum
flange. The new flange design, which is shown in Figure 2-3, features a larger, flat surface in
close proximity to the ionization region. Even with the modified flange design, the new valve-
mounting configuration can only accommodate two pulsed valves or one pulsed valve and one
capillary inlet. The two inlet types are mounted onto a single, wedge-shaped plate that pivots
near its apex. A system of slots and precision spring-loaded guides allows the plate to rotate ±30°
about its midpoint while maintaining adequate compression of a high-temperature Kalrez o-ring
seal placed around the sample entrance orifice on the vacuum flange. When the valve mounting
plate is in either of its extreme positions, the exit aperture of the pulsed valve or capillary is
precisely aligned with the sample entrance orifice. A positive detent mechanical lock secures the
valve plate in either position, and ensures reproducible location. In addition, a third locking
position is provide midway between the two extremes that effectively seals the vacuum chamber
and allows either of the valve or the capillary to be removed. The design incorporates several
other important features. The valve seat has been modified to significantly reduce the dead
volume between the sample gas and the exit orifice. Rather than introducing the sample through
-------
EPA/600/R-10/050
April 2010
the conventional coaxial entrance channel, which results in essentially no flow and a large dead
volume, the sample can be introduced through a small entrance channel cut near the sealing
poppet. A corresponding exit channel is also provided on the opposite side of the valve thus
allowing for continuous sample flow through the pulsed valve independent of its opening
function. The use of a flowing sample results in the rapid removal of organic samples by the
carrier gas flow, while also substantially reducing the dead volume of the valve. All results in
this report were obtained with these modified valve systems unless explicitly noted.
Spring-loaded guide
General Valve series 99
Spring-loaded pivot
Figure 2-3. Top View of the Movable Inlet Mounting Plate.
An alternative method of adiabatic expansion can be obtained using a continuous inlet system
with a small orifice (10-20 micrometer) as opposed to 0.5mm for the pulsed version. This
approach was tested for determination of the optimum orifice opening (by SRI International) but
has not been evaluated under field sampling conditions.
2.1.3 Time of Flight Mass Spectrometer
The first experiments supported by SERDP were performed using a Jordan reflectron time of
flight mass spectrometer. It has a mass resolution as high as 4000 in combination with the
supersonic jet inlet system. The size of the system prevented field operation (taking up a volume
of 6' x 4' x 3') and the mobile REMPI-TOFMS was therefore built around a 19" rack mounted
TOFMS. While the initial goal was to design a compact mass spectrometer from standard
materials, the need for very high level ion-optical engineering to obtain maximum mass
resolution in a small package led to a commercial solution. An ideal solution is found in the IL-
R10 In-Line Reflectron Time-of-Flight Mass Spectrometer package from Stefan Kaesdorf
Industries (Germany). This all-in-one unit, shown in Figure 2-4, contains the source region,
10
-------
EPA/600/R-10/050
April 2010
Reflectron TOFMS, ion detector, electronics, and high-vacuum pumps in a single, compact
standard 19" rack-mount package.
Figure 2-4. Schematic of LR10 compact 19" rack-mount reflectron TOFMS.
2.1.4 REMPI-TOFMS Instrument
The three major components (laser, inlet valve and time of flight mass spectrometer) previously
described make the REMPI-TOFMS instrument. Figure 2-5 shows the setup of the laboratory-
only large instrument while Figure 2-6 shows the system that can and has been deployed during
multiple field sampling trips.
Figure 2-5. Two views of large REMPI-TOFMS system.
On the left is the time of flight mass spectrometer and on the right a view of the large table with
OPO laser system
11
-------
EPA/600/R-10/050
April 2010
in
Figure 2-6. Compact REMPI-TOFMS instrument.
2.1.5 Operating Procedures REMPI-TOFMS
In contrast to the high voltage power supply for the Jordan TOFMS, the high voltage supply for
the Kaesdorf reflectron TOFMS was found to be extremely stable with no noticeable drift of
voltage values with time. Consequently, the mass calibration of the TOFMS was performed only
when significant changes were made to the laser beam alignment through the ionization
chamber. The mass calibration uses internal gas standards (usually benzene, 78 m/z and
trimethylbenzene, 120 m/z). These two sets of mass and time-of-flight arrival time, and the
constraint that the arrival time starts at t = 0.0 us for zero m/z, are used to generate two constants
for a second order polynomial fit to the data which is satisfactory up to mass 250. The calibrated
time of flight versus mass curve is stored in the data acquisition software.
REMPI-TOFMS can be operated in three different data acquisition modes:
Monitoring one or a range of masses as function of ionization wavelength
Monitoring one or a range of masses as function of time at one specific wavelength
Monitoring specific masses as linked to specific wavelengths
An example of a wavelength spectrum for multiple masses (75-200 m/z) is shown in Figure 2-7.
12
-------
EPA/600/R-10/050
April 2010
*>.
Figure 2-7. Detection of multiple analytes using wavelength-dependent ionization.
In Figure 2-7, the REMPI-TOFMS instrument was measuring analytes as present in the exhaust
gas from a diesel generator. Depending on the degree of resolution required and the actual length
in nanometers of the scan, these wavelength scans may take up to 30 minutes to complete.
Consequently, such wavelength scan can only be determined when the emissions are constant
(on the time scale of 30 minutes). This is the case for calibration gases or during steady state
operations of combustion systems. Clearly, this is not the case during accelerations or changes in
loading of an engine as shown in Figure 2-8. In such case it would have been impossible to
obtain a reliable wavelength spectrum. This is an example where it is more advantageous to
monitor multiple masses with one single wavelength.
120
100-
C
o
1?
I
o
O
0)
c
0)
0)
m
80-
60-
40-
20-
10 12 14 16 18 20 22 24 26
time (min)
Figure 2-8.
Transient benzene concentrations detected in vehicle exhaust while running a
dynamometer-based roadway simulator.
13
-------
EPA/600/R-10/050
April 2010
One advantage of lower selectivity in wavelength is that at a specific wavelength, it is possible to
ionize and detect multiple analytes with different mass. An example of this is shown in Figure
2-9 where at the most favorable wavelength for benzene detection multiple PAH-like analytes
are detected. As discussed earlier, overlap in absorption spectra provides information on a
selected subset of analytes. Confirmation of the analyte identity should be performed with
wavelength scans.
240
Figure 2-9. REMPI-TOFMS mass spectrum at benzene (78) wavelength as observed in
vehicle exhaust.
The third method of data acquisition has become available only after conclusion of all field tests
that were performed with the REMPI-TOFMS instruments. The data acquisition software uses a
list of predetermined mass and wavelength combinations that are unique for detection of each
analyte individually. The data acquisition system runs automatically through such sequence of
wavelengths and acquires the ion intensity for the associated masses. This provides a much more
hands-off approach that can be utilized in the detection of multiple analytes under "steady state"
engine combustion conditions. This approach has been tested on a multi component calibration
gas with satisfactory results. Further improvements could include changes in detector voltage
("gain"), using predetermined values (passive control) or based on observed intensity active
control by enhancing the gain until a specific ion signal is present.
2.2 LIBS
In laser-induced breakdown spectroscopy, LIBS, a pulsed Nd:YAG laser beam is focused
through a lens onto a substrate and plasma is formed. This plasma vaporizes and atomizes the
sample, exciting the constituents of the sample. Light emitted from the plasma is characteristic of
the interrogated sample. This light is collected with a spectrometer containing a charge coupled
device, and a broadband spectrum (200-980 nm) is generated. Figure 2-10 shows the setup as
used for interrogation of filter deposits as collected during vehicle sampling.
14
-------
Broadband Spectrometer
Nd-YAG
1064nm
OOLIBS
software
Sv
Tocusing/
collecting optics
Plasma «=:
Spark
EPA/600/R-10/050
April 2010
Figure 2-10. Schematic setup of LIBS system.
2.3 ORS
ORS measurements provide information on a path-integrated concentration of HAPs in
emissions from, possibly, a diesel turbine engine. The OP-FTIR can also be set up to collect
information on PM in the engine exhaust. The OP-FTIR spectrometer is designed for both fence-
line monitoring applications, and real-time, on-site, remediation monitoring and source
characterization. An infrared (IR) light beam, modulated by a Michelson interferometer is
transmitted from a single telescope to a retro reflector (mirror) target as shown in Figure 2-11.
The returned light signal is received by the single telescope and directed to a detector. The light
is absorbed by the molecules in the beam path to the retro reflector and again as the light is
reflected back to the analyzer. Thus, the round-trip path of the light doubles the chemical
absorption signal. One advantage of OP-FTIR monitoring is that the concentrations of a
multitude of infrared absorbing gaseous chemicals can be detected and measured simultaneously,
with high temporal resolution.
15
-------
Retroreflector,
EPA/600/R-10/050
April 2010
30m
Exhaust
Stack
Figure 2-11. Setup ORS during sampling.
16
-------
EPA/600/R-10/050
April 2010
3. REMPI-TOFMS: Field Ready Development and Performance Evaluation
and Improvement
At the onset of this SERDP project, a large REMPI-TOFMS system existed at the EPA facility
which was suitable for emission characterization for sources located on-site. The experiences
with this system were used to develop a smaller and field-ready unit. The evaluations of the
critical instrument parts that were part of the SERDP project are described here.
3.1 Laser Systems
The performance of a laser system is in general sensitive to the environmental conditions in
which it is placed. Although ultra-stable laser systems (both in laser energy output and absolute
wavelength) exist, the laser systems used during the SERDP funded projects had limitations in
their stability, especially considering their application in less favorable field conditions with
potentially larger ambient temperature fluctuations and the presence of PM (dust) in air.
3.1.1 Continuum Laser
The large Continuum laser system is located in a laboratory equipped with an air conditioning
system. However, slight changes in ambient temperature were found to have only a minimal
effect on the emitted laser energy. The phase matching angles of the second and third harmonic
crystals needed remote control adjustments only during the first 2-3 hours of operation). No
appreciable wavelength shifts were noted over the course of a day of measurements. This in part
due to the fact that the wavelength selection in this laser system is not determined by the OPO
crystal (phase matching) angle but rather by a grating which is much less sensitive to
temperature fluctuations. Phase matching angles adjustments of all crystals inside this
Continuum Sunlite OPO system are performed using the software installed with the instrument.
3.1.2 OPOTEK Laser
Temperature changes inside the laser system, especially during the warm-up period, results in
noticeable changes in laser energy as well as actual wavelength emitted from this laser. The
changes in laser energy are due to changes in the phase matching angles of the second and third
harmonic generating crystals which tend to shift when the temperature changes. Adjustments
from the outside of the laser system can be made to correct for these changes and do not create a
major concern in the performance of the laser system. OPOTEK has informed us that a more
stable version with feedback control is now available but such improvement has not been
implemented in the existing laser systems. Adjustments of the phase matching angle of the
frequency doubling module can be performed using the software control of the OPOTEK laser
system. Figure 3-1 shows an example of laser wavelength drift as a consequence of increased
temperature inside the laser system. This is observed when the wavelength-dependent ion signal
is monitored for a calibration gas. A suggested improvement to the manufacturer would be to
elevate the temperature of the OPO crystal to well above ambient temperature in order to reduce
these shifts. It should be noted that the application of this laser system for semi-high resolution
17
-------
EPA/600/R-10/050
April 2010
MS (HRMS) in combination with TOFMS is at the edge of its wavelength stability capabilities.
Alternatively, a temperature stabilization of the air inside the laser system would be possible
(without turbulence preferably). Note that the observed wavelength shifts were obtained despite
the presence of a working air conditioning unit that was connected to the dust cover under which
the laser is placed. The main function of the air conditioning unit was to maintain a relatively
constant temperature inside the dust cover, independent of the ambient outside temperature.
Fortunately, the drift in wavelength, if present, is constant across the typically used wavelength
range of operation (255 to 290 nm). Therefore, it was implemented in all sampling procedures
with the REMPI-TOFMS system to frequently monitor an internal standard, deuterated benzene,
and correct established wavelengths of target analytes by the drift in wavelength for the internal
standard.
ro
c
D)
30-
25-
20-
15-
258.1 258.2 258.3 258.4 258.5 258.6
Wavelength (nm)
258.7 258.8
Figure 3-1. Wavelength emitted from laser with increased ambient temperature (increase
in temperature for scan 1 to 9).
Wavelength scans over extended ranges were found to result in an oscillating laser energy output
which can be corrected by simultaneous recording of the laser energy at the exit window of the
ionization chamber. This oscillation is primarily due to the use of high order polynomials to fit
the OPO and UV stepper motor positions as controlled by the software. Deviations in motor
position from this fit at intermediate wavelength values are oscillatory in nature resulting in non
optimal phase matching angles, hence less optimal laser energy performance.
3.2 Pulsed Valve Operation
The performance of the pulsed valve system was expected to be a weakness in the whole
REMPI-TOFMS system since (1) it cannot be heated above 200-210 °C, (2) it would need
frequent cleaning following exposure to "dirty" samples (e.g., high concentrations, high PM,
"sticky" gases), and (3) it could be subject to fast wear and tear on the pulsating poppet made of
Vespel material. Since most of the applications funded under SERDP were related to the
detection of more volatile analytes, it was not necessary to heat the valve system continuously to
18
-------
EPA/600/R-10/050
April 2010
such high temperature. Good results were obtained at valve temperatures of 150-180 °C. Also,
implementing a 10 |j,m inline filter prevents passage of ultrafme PM which otherwise can pass
through standard fiber filters, reaching the poppet in the pulsed valve system. The swivel
mounted double valve systems worked properly. However, the number of equipment connections
to the inlet side of the valve makes switching from one valve to the other in case of failure more
cumbersome. The swivel valve design was also modified during the funding period to avoid
vacuum leaks under applied stress to the pulse valve unit when connections to sampling line and
equipment were present.
3.3 Data Acquisition Software
Time of flight spectra are commonly recorded using a digital oscilloscope. The data acquisition
during this project made use of a fast digitizer card inside a PC. Measurements performed on the
large REMPI-TOFMS instrument used a 500 megahertz (MHz) digitizer card (Signatec) while
the small REMPI-TOFMS instruments all include a PC with a 1 gigahertz (GHz) digitizer card
from the same manufacturer. Laser control was performed using provided software (Continuum)
or provided Lab View virtual instruments (OPOTEK). A snapshot of the data acquisition
software which was in part developed during this SERDP project by SRI International is shown
in Figure 3-2. Data acquisition stores data files to the hard disk that include time, wavelength,
ion signal, and multiplier detector voltage. Depending on the number of averaged laser shots, a
data acquisition period as fast as 0.1 s, however, in reality 10 or 20 laser shots are typically
averaged before data is written to hard disk. The data acquisition software has the capability to
immediately calculate the integrated area under a time of flight ion peak. This works very well
under controlled conditions such as time or wavelength dependence of an analyte in the presence
of a limited number of nom-interfering analytes such as in multi-analyte calibrated gas mixtures.
Under fluctuating emission conditions with a larger number of unknown analytes such as in the
exhaust of a combustion system, it is more advantageous to record and store the whole mass
spectrum (within set mass limits) and perform post-analysis processing in order to asses if
interfering events such a changes in baseline signal occurred. Although this method of data
acquisition is more demanding (large data files and post-data processing) it is much better for
data quality assurance with the additional benefit that all data is saved. As part of the SERDP
project, a custom macro was written for use with the data analysis software (Origin 6.1 and
higher, OriginLab Corp.) which simplifies the post-data analysis processing.
19
-------
EPA/600/R-10/050
April 2010
Figure 3-2. Photo of data acquisition software in operation.
3.4 Calibration of REMPI-TOFMS System
External calibration of the detected ion signals was typically performed at the beginning and at
the end of the day. This was accomplished by monitoring the response to a 100 ppb calibrated
gas mixture in nitrogen containing 14 aromatic compounds (Scott Specialty Gases). The
concentration of a target analyte is derived from the ratio between its ion signal intensity and that
of the same analyte as present in the calibration gas mix, normalized for possible laser energy
changes. Concentrations of those exhaust analytes not present in the calibration mix were derived
from the ionization probability ratios between benzene and such analytes as found by Cool and
coworkers (Velazquez et al., 1998). In the case of concentration calibration of compounds that
are in their crystalline form at room temperature, such as PAHs, a diffusion-tube-based delivery
system was designed and applied to generate a known concentration of a specific analyte in the
gas phase. The absolute concentration of this delivery system was verified independently for
each analyte of interest using external high resolution GC (HRGC) combined with low resolution
MS (LRMS) analysis.
System performance monitoring is provided by addition and measurement of fully deuterated
benzene (C6D6) since it is unambiguously detected and is not an exhaust product. A linear
REMPI-TOFMS response has been established over several orders of magnitude of
concentration (ppt to ppm range) for all aromatic compounds of interest. Since the response to all
compounds is linear with concentration, the measured response to the added CeDe was used as an
indicator for REMPI-TOFMS system performance.
All recorded ion signals were corrected for possible changes in laser energy during post analysis
processing of the experimental data.
REMPI-TOFMS has the capability to individually identify a large set of aromatic compounds
and this can be accomplished in real time with a 1 s resolution at a high pptv level concentration.
No sample clean-up or extraction is needed and consequently, qualitative results are available
instantaneously.
20
-------
EPA/600/R-10/050
April 2010
4. Sampling from DoD Sources
This report presents the results obtained with REMPI-TOFMS, LIBS, ORS, online-GC and other
standard EPA methods during the sampling from multiple sources. Table 4-1 summarizes the
research efforts that were taken during this SERDP project to obtain emission factors for
multiple relevant sources.
Table 4-1. Source testing.
Source
Number
1
2
4
3
5
4
6
Chapter
7,8
9
11
10
12
11
13
Source
Diesel Generator
Aircraft Ground
Equipment
Municipal Waste
Combustor (MWC)
Boiler /ETV
DoD vehicles
MWC
Military Aircraft
Date
04/2004
05/2004
12/2004
09/2005
10/2006
12/2006
10/2007
Location
EPA, RTP, NC
EPA, RTP, NC
Portsmouth, VA
EPA, RTP, NC
Aberdeen, MA
Portsmouth, VA
TyndallAFB, FL
Used REMPI-TOFMS
System
Large
Compact
Compact
Compact (two systems)
Compact
Compact (two systems)
Compact
21
-------
EPA/600/R-10/050
April 2010
5. Source 1: Validation of REMPI-TOFMS Measurements on a U.S. Marine
Corps Diesel Generator
REMPI-TOFMS has been applied to the exhaust of a diesel generator in order to provide
extensive emission concentrations of aromatic compounds, both under steady state conditions as
well as during transient events, such as a cold start. REMPI-TOFMS results were compared
qualitatively and quantitatively with extractive sampling techniques utilizing GC/MS analyses,
providing insights to the applicability and precision of the method.
5.1 Experimental section
The diesel generator tested was a 60 kW Tactical Utility (Class 2) Generator set provided by the
United States Marine Corps (Figure 5-1). It was placed outside and the exhaust was ducted past
the sampling ports into a point exhaust vent of the building. All ducts were thermally isolated
and kept under negative pressure to prevent fugitive emissions from entering the building. The
diesel generator exhaust was sampled sub-isokinetically through a Vi" (6.35mm) stainless steel
line at typical flow rates of 2 L/min [at standard temperature and pressure (STP)]. A micro fiber
filter, kept at 150 °C, was placed in the sample line in order to avoid soot reaching the valve inlet
system of the REMPI-TOFMS instrument. A flexible 9 m long, 1A" (1.27 cm) diameter, silicon
steel transfer line transferred the exhaust gas to the REMPI-TOFMS instrument. This line was
maintained at 150 °C while sampling exhaust gases in order to avoid adsorption of the organic
compounds to the walls. Upstream from the filter, 0.1 L/min of 1 ppm fully deuterated benzene
(CeD6) in nitrogen was added as a dynamic spike to obtain typical concentrations of 60 ppb CeD6
in the total exhaust stream. CeD6 was chosen as an internal calibration gas for REMPI-TOFMS
since its mass of 84 m/z and ionization wavelength will not interfere with the detection of
aromatic compounds. It is not present in the diesel exhaust gas itself, and has an ionization
wavelength within the commonly used wavelength scanning range (250-360 nm). The
concentration of CeD6 was also monitored to indicate the potential for adsorptive loss of VOC
during soot loading on the filter surface (sampling bias).
22
-------
EPA/600/R-10/050
April 2010
Figure 5-1. U.S. Marine Corps tactical utility generator.
Most of the steady state diesel generator exhaust measurements were recorded over three non-
consecutive days within a 7-day period. The purpose of the repetition was to (1) evaluate diesel
generator exhaust emission levels at different days and (2) to verify the reproducibility of the
REMPI-TOFMS results on a day-to-day basis. Emission data were collected and calibrated for a
selected group of organic compounds using the method described previously. Multiple
repetitions of this method provide information about how many samples should be taken to
obtain representative and accurate emission factors using REMPI-TOFMS. Concurrently, co-
located diesel generator exhaust samples were taken with two EPA established extractive
analysis methods, being Method 0040 (EPA, 1996a) and Method 0010 (EPA, 1986a), in order to
verify the accuracy of the concentrations as determined by REMPI-TOFMS. All steady state
results were obtained during a 75% loading ("full load") of the diesel generator. With the major
exhaust analytes now established, the cold and warm starts of the diesel generator verified the
ability of REMPI-TOFMS to perform fast, real time measurements of such transient events.
5.2 Results and Discussion
5.2.1 Steady State Diesel Generator Results
A survey wavelength scan of the exhaust gas stream in Figure 5-2 illustrates the complexity of
aromatic compounds in diesel generator exhaust. Each mass spectrum plotted is the average over
10 single-shot mass spectra at a set laser wavelength. The wavelength was scanned in 0.01 nm
steps. Traces of ion signals at mass 78, 92, 94, 104, 106, and 120 are identified as single aromatic
ring compounds while the dominant traces at mass 128 and higher are methylated-PAHs. The
single peak at mass 84 (A,=257.8 nm) is CeD6 as added to the diesel generator exhaust as an
instrument monitor. With the diesel generator running at full load condition, the wavelength was
tuned over the range where all BTEX compounds as well as many PAHs are ionized. No major
ion signals, other than those due to minimal fragmentation, were detected below mass 78
(benzene) for any wavelength within the 250 to 360 nm wavelength window. At low (less than
m/z = 128) mass, the 3D graph shows contributions of all BTEX compounds as well as other
compounds. At a wavelength optimized for benzene detection, numerous additional ion signals
are detected as illustrated in the mass spectrum in Figure 5-3. The inset in (B) shows the lower
part of the mass spectrum recorded. Methylated benzenes (0) and methylated naphthalenes (*)
23
-------
EPA/600/R-10/050
April 2010
are indicated; the former group is not ionized at the higher wavelength. Major "ghost" mass
peaks due to instrument anomaly are indicated ( ) below the mass spectra. There appears to be a
characteristic series of mass peaks with poor spectral resolution that can tentatively be assigned
to methylated naphthalenes. Another series that can be identified are methylated benzenes. A
more extensive analysis of the mass spectra recorded with this TOFMS for diesel generator
exhaust and calibration standards revealed that the molecular ion peak of each individual target
analyte is accompanied by a second peak of lower intensity which appears approximately 1.25 to
2.75 mass units lower, depending on the mass of the target analyte. This is not an ion signal due
to fragmentation but rather an anomaly of the TOFMS instrument. Since there was no clear
dependence found with all applied voltages in the TOFMS, signal intensity, as well the position
of the laser beam in the ion source, this anomaly is or may have been due to an impedance
mismatch in the signal circuit that occurred during these experiments. It clutters the observed
mass spectrum as in Figure 5-2; however, since most of these peaks do not line up at an integer
mass unit, they can be ignored in the analysis.
Figure 5-2. A 3D survey scan (mass, wavelength, ion signal)
24
-------
EPA/600/R-10/050
April 2010
z.u -
| 1-6-
1 1-2J
^0.8^
ro
g) 0.4-
to
n n
jl Jiii JiU w
HJU»A
ul
A
LjJl^ll^^
2.8-
2.4^
-g 1.6-
(13
.s> 0.8-
to
g 0.4 -|
0.0-
80
100
120
140 160
m/z
180
200
220
20
40
60
B
100
120
140 160
m/z
180
200
220
Figure 5-3. Time-of-flight mass spectrum recorded at (A) optimum styrene (287.7 nm)
and (B) Benzene (259.0 nm) ionization wavelength in diesel generator exhaust.
Before concentrations can be derived from the ion signals in the diesel generator exhaust, it is
necessary to verify if the spectroscopy of the target analytes matches with that of a calibration
standard taken under identical experimental conditions. Figure 5-4 shows as an example the
spectrum of toluene to illustrate that there are no detectable spectral interferences at the
concentration level present in the diesel generator exhaust. To verify if a steady state diesel
generator condition results in stable emission concentrations the REMPI signals of the detected
benzene in the exhaust, and CeDe, as added to the exhaust, were recorded over a one-hour period.
Both REMPI signals were found to be nearly equal in stability; the average REMPI signal for
benzene had a relative standard deviation (RSD) a = 6. 1% while for CeD6, a = 5.2% as
illustrated in Figure 5-5. This observed stability of diesel generator exhaust emissions under
steady state conditions simplifies the subsequent determination of emission levels for other
aromatic compounds.
25
-------
EPA/600/R-10/050
April 2010
256 258 260 262 264 266 268 270
Wavelength (nm)
Figure 5-4. Comparison of 1+1 REMPI wavelength spectrum recorded for toluene in
diesel generator exhaust and from calibration gas standard (the latter scaled
to equal intensity).
In 18-
'c 16-
= 14.
f! 12-
^ 10-
£5 8-
.9> 6-
W 4.
o 2^
n _
CD r
~X 14 J:
0
2 12-
8 10-
P 8-
CD
iS 6:
CD 4-
^ 2-
n-
JuLtiUbuJ^^
Wwfflrl^^^ ni/pfrrW
deuterated benzene internal standard
) 10 20 30 40 50 6
^₯4^^^
benzene from diesel generator exhaust
Time (min)
Figure 5-5. Monitored REMPI-TOFMS signal for reference gas (C6D6) and benzene in
exhaust diesel generator.
26
-------
EPA/600/R-10/050
April 2010
Concentrations were derived from approximately 1 min averages of recorded ion signals between
changes over nine predetermined wavelengths for phenol, p-xylene, m-xylene, o-xylene, 1,2,4-
trimethylbenzene, toluene, naphthalene, benzene, and deuterated benzene. Figure 5-6 shows the
raw data results for such a nine wavelength sequence that includes the above mentioned target
analytes. For simplicity, only the parent ion signals for the target analyte within each 1 min
interval are shown; traces for all individual masses (range m/z = 75 - 300) are recorded during
the whole sequence and stored to a computer hard disk. As can be seen, all ion signals are
constant within the 1-minute intervals, an additional indication of stable emission concentrations,
even on this short time scale. The ion signals have not been corrected for pulsed laser energy
fluctuations that are visible in the noise of the ion signal. Since the laser energy detector is not
fast enough for single laser pulse energy detection, the 1 s average of the laser energy (i.e., 10
laser pulses) was continuously recorded and stored for further normalization. On Day 1, the nine-
target analyte sequence was repeated eight times as shown in Figure 5-7, which shows the
complete mass traces. The observed ion signals remained constant for each individual compound
over the 1.5 h of data acquisition. RSD in the 1 min averages typically range between 5 and 12%.
The variation in the 1 min averages of each of the nine selected analytes, after correction for
laser energy differences, ranged from 2 to 9%, well within the standard deviation of the 1 min
averaged ion signals. Similar results were obtained for the external calibration of the diesel
generator exhaust REMPI-TOFMS ion signals. This demonstrates that over the 1.5- hr timeframe
of data acquisition, variations in emission levels in the diesel generator exhaust are much smaller
than the accuracy with which REMPI-TOFMS can measure the concentrations. It is also in
agreement with the previously discussed constant benzene emissions over an extended period of
time. Since the variation in the observed 8-fold repeat of ion signals is comparable with the RSD
of a single 1 min average, the number of repeats can be reduced to three. The final repeat of the
nine target analyte sequence was obtained after a clean filter substitution. No changes in ion
signals were observed for all nine studied analytes indicating that over a time period of more
than 1.5 h, no bias is detected towards absorption of these target analytes on the soot loaded filter
surface.
14
12-
3T lo-
00
cw
§
6-
4-
2-
0
S B
"S s"
Vi
23456
time (min)
10
Figure 5-6. On-line REMPI-TOFMS ion signal traces of nine individual target analytes,
labeled with mass, chemical name and used, most favorable ionization
wavelength.
27
-------
EPA/600/R-10/050
April 2010
10
20
30 40 50
time (min)
60
70
80
Figure 5-7. Recorded ion signals during nine repeats of the analyte cycle.
To verify if the diesel generator exhaust emission pattern is constant over several days and to
assess if the response of the REMPI-TOFMS instrument changes from day to day, the nine
wavelength sequence measurements were repeated over three non-consecutive days within a
seven day period. Analyte exhaust concentration levels appeared relatively consistent with each
other, indicative of steady state diesel generator emission conditions on a day-to-day comparison.
Variances ranging from 9 to 30% exist, however, for the investigated target analytes as shown in
Figure 5-8. An analysis showed that most of the variation in the daily comparison of the
individual target analytes was already present in the daily variation of detected ion signals from
the diesel generator exhaust (13-25%) while the calibrated gas mix ion signals were fairly
constant (5-12%) over the course of the three days of experiments. This suggests that the largest
changes in emission levels are due to daily changes in the steady state diesel generator emissions,
the latter being larger than the intraday variation (1-10%).
28
-------
EPA/600/R-10/050
April 2010
180-
160-
£" 140-
t-\
L_L
B 120-
~ 100-
"cS
f oU -
0)
0 60-
O
O 40-
20-
0-
_ld
-
T
=
^
/
^
£
/
',
',
'/
1
I
s
s,
S
S
s
^
?
>
y
>
/
/
/
>
/
/
/
/
_;
_L
\
s
s
s
\
\
\
s
s
s
s,
\
s
s
s
s
\
s
\
s
s
s
\
\
s
T
=
:
=
^H phenol
I I p-xylene
I I m-xylene
J^l I o-xylene
i
|T
I
=
^n
I
\
\
\
S
S
\
S
\
\
\
s
s
s
/
/
/
/
/
/
',
/
/
/
',
/
''.
s
s
s
s
s
s
s
s
s
s
s
s
N
s
\
s
s
s
s
s
s
s
s
s
V/TA 1,2,4tmbz
IX\NM toluene
V77A naphthaler
1\NX1 benzene
1 I styrene
?
/
/
/
/
/
/
/
₯
\
s
s
s
s
s
N
S
s
s
s
\
s
_i
7
',
/
',
^
/
',
',
/
'<_
\
\
\
\
\
\
^
\
S
s
s
\
s
s
s
s
s
s
N
^
I_
-
=
3
Day 1
Day 2
Day3
Figure 5-8. Observed day to day variations in estimated concentrations of BTEX-like
analytes in exhaust diesel generator.
Information regarding the emission level of other observed analytes in the diesel generator
exhaust was extracted from separate measurements. Styrene (m/z = 104) and ethylbenzene (m/z
= 106) were detected and identified based on their agreement with literature spectra. On the other
hand, aniline (m/z = 93) was noticeably absent in the diesel generator exhaust. An upper value of
the aniline concentration is determined to be 0.3 ppb, based on the similar ionization probability
of aniline in comparison to benzene, being only a factor 1.67 lower (Velazquez, 1998). Note that
this is not the REMPI-TOFMS instrument detection limit for aniline but rather the minimum
concentration that would have been detectable in the diesel generator exhaust without significant
ion signal overload on the detector at other masses. In fact, all aromatic compounds detected and
discussed so far have detection limits in the low parts per trillion (ppt) range using this REMPI-
TOFMS instrument as observed prior to this study.
The qualitative analysis of the PAHs as observed in the diesel generator exhaust is obscured by
the presence of a multitude of methylated naphthalenes. The identification of m/z = 128 as
naphthalene is based on the spectral agreement of the Si < So band near 301.6 nm with literature
(Velazquez, 1998). Note that for naphthalene, this band around 301.6 nm has significantly lower
intensity than the (SN <- Si; N>1) bands between 250 and 278 nm. This example illustrates that
naphthalene, like many other PAHs, can be readily detected at 266 nm using a fixed wavelength
laser system, but that for analyte verification a tunable laser system is necessary. Following the
naphthalene assignment, it is provisional to assign the mass 142, 156, 170 and 184 ion signals to
methylated naphthalenes up to tetra methylnaphthalene. Minor mass contributions were observed
at 266 nm for mass 152 and 154 that could be attributed to acenaphthylene and acenaphthene.
Fluorene (m/z = 166) in diesel generator exhaust was confirmed by the agreement with the
spectral features around 296 nm (Zimmermann et al., 1994). Assignment of m/z =178 was
complicated by the fact that the individual identification of the most likely PAHs, namely
anthracene or phenanthrene, requires a two-color ionization scheme (1+T REMPI) and was
beyond the scope of this work. The extractive sampling GC/MS analysis showed that only
29
-------
EPA/600/R-10/050
April 2010
phenanthrene was detected and no anthracene. With this additional information available, the
mass 178 peak was assigned to phenanthrene. A low intensity mass peak was detected at mass
202 with no clear spectral features as recorded in an additional wavelength scan (250-360 nm)
that can be attributed to either pyrene or fluoranthene. Without further 1+1' REMPI experiments,
a distinction can not be made as to which isomer is detected. The concentrations of the observed
PAHs in diesel generator exhaust were obtained from a comparison between the established
calibration curves for each PAH using the sample delivery system and the diesel generator
exhaust ion signals.
The diesel generator exhaust emission results obtained by REMPI-TOFMS were compared with
those from standard extractive and GC/MS analyses, taken in parallel and at the same location in
the exhaust gas flow. Since m- and p-xylene co-elute in the GC/MS analysis, the REMPI-
TOFMS based concentrations for these isomers were added for comparison. Figure 5-9 shows
the good agreement that is observed for the three day average benzene concentration (163 ± 14
ppb with REMPI-TOFMS and 189 ± 24 ppb with GC/MS) with daily differences in measured
concentrations varying between 1 and 25%. For toluene and the summed m- and p-xylene
isomers, the agreement is only fair, as the REMPI-TOFMS results are up to 60% different from
those determined using extractive sampling and GC/MS analyses. REMPI-TOFMS and GC/MS
analyses measured, respectively, 83 ± 15 ppb and 120 ± 19 for toluene and 69 ± 16 ppb and 126
±10 ppb for the summed m- and p-xylene. The cause for this discrepancy is unknown. The
extractive GC/MS analysis of the diesel generator emissions appears to have a similar level of
variation in daily average concentration when observed in parallel with integrated, daily average
REMPI-TOFMS measurements.
250<
&
Q. 200-
g
2 150-
*-»
I
8 100-
-I*
w
CD
§ 5°*
CD
Q
Extractive Sampling
REMPI-TOFMS
T
I
123 123 123
benzene toluene m+p xylene
Figure 5-9. Comparison between extractive and REMPI-TOFMS concentrations for three
exhaust analytes.
30
-------
EPA/600/R-10/050
April 2010
5.2.2 Transient Diesel Generator Results
So far, results have been presented revealing the constant emission levels under steady state
diesel generator exhaust conditions. Since start-up of a cold engine includes a period of
incomplete combustion, it was expected that emissions during start-up would not be constant.
Emissions were recorded during the start-up of the diesel generator at the beginning of the day
("cold start"). As illustrated in Figure 5-10 (a), at t = 40 sec, the diesel generator was started and
brought directly to the "full load" operational condition. With the laser wavelength optimized for
benzene (A, = 259 nm) detection, an initial sharp spike of- 10 s in benzene concentration was
detected, as well as in several other BTEX and PAH concentrations. The benzene peak intensity
represents a transient maximum concentration of 14 ppm, which is 90 times larger than the
steady state emission levels (1 min average) recorded after 5 minutes of diesel generator run
time. Similar transients, but with significantly different durations, were observed for all detected
PAHs and methyl-substituted PAHs of which naphthalene and two methylated naphthalenes are
shown in the insert of Figure 5-10 (a). Again, the emission levels after 5 minutes are equal to the
steady state emission levels reported previously for all detected compounds. Following cold start,
the increase in ion signal, hence concentration, of these larger mass compounds is slower than
for benzene due to the 150 °C temperature used for the sampling equipment. The maximum
emission level of 1.5 ppm for naphthalene was found to be a factor of 14 higher than the steady
state emission level that day (109 ppb). The high selectivity of REMPI-TOFMS prevents the
study of all individual BTEX compounds during the same cold start so there is no complete
information available about specific peak emissions for all BTEX compounds. Benzene peak
emissions from cold starts of the diesel generator on other days were found to be in good
agreement, with less than 20% variation in the peak emission concentration. The integrated area
under the transient benzene ion signal following the cold start in Figure 5-10 (a) was calculated
to be a factor of 28 higher than that of the steady state emissions during an equal time interval of
15 seconds. Consequently, the amount of benzene emitted at a cold start (first 15 seconds) is
equal to the amount emitted over a 7 min time period under steady state conditions. With the
laser wavelength positioned where toluene is effectively ionized (A, ~ 265.8 nm), similar results
were obtained for a different cold start. In this case, the toluene transient peak was found to be
about a factor of 23 higher than the steady state emission concentration. Figure 5-10 (b) shows
the normalized ion signals recorded in parallel for other masses during this cold start. The higher
sampling temperature of 200 °C resulted in an equal incline (in 4-6 seconds) for all masses,
excluding sampling bias of higher mass analytes. The tail end of the curve, however, appears to
be different depending on the molecular structure. All single aromatic ring molecules decay
faster to their respective steady state values while the two-ring methylated-naphthalenes show a
slower decay. The figure has a bias towards this statement since the concentrations of the latter
group of molecules remained higher than those for the former group, or, in other words, the
sharp increase following a cold start is less intense for (methylated-) naphthalenes than for single
aromatic ring molecules. Taking this into account by using the decay rate from the peak to half
the value between peak and steady state level, the decay rates are still a factor of three slower.
The explanation behind this observation can be attributed to the diesel generator combustion
process where apparently the transition from a process of incomplete combustion (cold-start) to
more or less complete combustion (steady state) is faster for single aromatic ring analytes than
for double aromatic ring analytes. Similar "memory effects" have been observed for PAHs in
31
-------
EPA/600/R-10/050
April 2010
waste incinerators using REMPI-TOFMS (Heger et al., 1999). No evidence was found for
significantly delayed (in terms of minutes) PAHs emissions.
The sampling time of REMPI-TOFMS for each data point is 1 second (10 laser shots average ion
signal). This is clearly sufficient to resolve the 7 sec (full width half maximum, FWHM)
emission peak during cold startup. In contrast with that, the extractive sampling technique using
Tedlar bags (EPA Method 0040) has to sample for several minutes to get sufficient information
and the GC/MS analysis revealed no elevated emission values for the first 2-3 minutes that
includes cold startup. Apparently, these short events go unnoticed in the extractive sampling
technique.
naphthalene
methylnaphthalene
dimethylnaphthalene
double ring aromatics
single ring aromatics
30
60 90
time (sec)
120
150
Figure 5-10 (a, b). (a) Real-time transient benzene emissions following a cold start, (b)
Normalized ion signal traces for single aromatic ring (thin lines) and double
aromatic ring (fat lines) analytes.
Similar transient results were obtained for start-up events when the diesel generator was already
at its operating temperature. Figure 5-11 shows benzene and naphthalene concentrations during a
triple repeat of these warm restarts. The benzene emissions were found to be a factor of 2.3
lower (6 ppm versus 14 ppm) in comparison to the cold start, while for naphthalene this ratio was
a factor of 5-6 lower (250-300 ppb versus 1.5 ppm). This indicates that naphthalene (and
probably all PAHs) is predominantly formed during a significant period of incomplete
combustion, such as a cold start, in comparison to the shorter periods of incomplete combustion
with a warm restart. Again, only a real time detection method, such as REMPI-TOFMS, can
provide this information.
32
-------
EPA/600/R-10/050
April 2010
u
Q
OD
*^3
0
J
7-
^ 6:
xB> 5~
% ₯ 4-
(L) .J5
N -a o
G 2 3~
(L) 43
& % 2-
o
C 1 -
0
^ n "
u -
^^
M 1 | 1 1 1
1
j\
1 1 1 1
+_
1 '
-A.
1 | , , ,
^
:
'7T7TPO
Z^IilvU
^P^OFF
300 -,
g
5
3
'
-^200-
2l50-
cd
^ 100-1
CL>
g 50^
o
O
0
5 10 15 20
time (min)
25
Figure 5-11. Transient REMPI-TOFMS results.
Top frame: Loading profile during warm restart test schedule. Middle frame: Transient
benzene emissions following three restarts as recorded at the wavelength for benzene
detection. Bottom frame: Transient naphthalene emissions.
Besides a sharp rise in emissions during a restart, the shutdowns of the diesel generator were also
accompanied by transient, yet smaller, increases of benzene and naphthalene concentrations
before dropping to zero. This is probably due to the additional emission of small amounts of
unburned hydrocarbons in the final movement of the engine piston without combustion of the
diesel fuel. Note that the relatively slow signal decay upon shutdown is due to the large duct
volume from which REMPI-TOFMS continues to sample, even when the diesel generator was
shutdown. For comparison, the signal decay time for the sampling of BTEX compounds at 2
L/min including the 9 m transfer line at 150 °C was found to be 2-3 s while the decay time that
includes sampling from the duct volume as shown in Figure 5-11 was 10-12 s.
33
-------
EPA/600/R-10/050
April 2010
6. Source 1: U.S. Marine Corps Diesel Generator Air Toxic Emission
Characterization
This chapter continues the work described in Chapter 7 (Oudejans et al., 2004), by using
REMPI-TOFMS to determine HAP emission factors for a diesel generator, both for steady state
and transient operations, such as start-ups. Temporally-resolved and mode-relevant emission
factors were determined for the most prevalent aromatic air toxics and compared to cumulative
samples collected simultaneously via conventional EPA Methods. These measurements were
compared to emission factors available in EPA databases to determine the efficacy of the
REMPI-TOFMS method as well as the need for real time emission factors. Other online
measurement techniques for VOC, fine particles, particle-bound PAHs, and O2, CO2, and CO
were used in this study to complement the gaseous organic air toxics measurements and to
highlight potential relationships.
6.1 Experimental
Air toxic emissions were characterized for a Tactical Utility Generator (Class 2, MEP6) provided
by the United States Marine Corps. This is a Kurz & Root diesel-engine-driven, JP-8-fired,
tactical, skid-mounted unit, Model MEP006A. The engine is a liquid-cooled, 6-cylinder, 4
stroke, and turbo-charged diesel. The generator is a 50/60 kW, brushless, air cooled, rotating
field generator. The maintenance records are dated June, 1973; presumptively this is its initial
year of operation. The hours of operation are unknown. The load bank was a 100-kW, variable
load, Avtron model. The diesel generator set was placed outside the research facility, and the
exhaust was connected to a 25.4 cm diameter duct of 350 cm length with sampling ports
connected to the various sampling equipment. The exhaust gas was drawn past the sampling
ports, through a dilution tunnel, and connected to a point exhaust on the building. The sampling
duct was thermally insulated and under negative pressure to prevent any fugitive emissions.
Various operational loads on the generator were established via connection to a variable, pre-set,
100-kW Avtron load bank.
The organic air toxic emissions were characterized for cold and hot starts, different engine loads,
and steady state operating conditions. Measurements were made using the REMPI-TOFMS, an
on-line GC (OLGC) system (75 m RTX624 column), a photoelectric aerosol sensor (PAS 2000),
an ELPI, and extractive EPA certified methods. Emission rates were determined for steady state
conditions at a nominal load of 75% of the total load, or 45 kW power output, at least 30 minutes
after start-ups. Data were taken for three consecutive days and reported as average daily
concentrations and standard deviations thereof.
The REMPI-TOFMS results were compared with reference data, obtained concurrently with the
conventional EPA sampling/analytical methods listed in Table 6-1, as well as the OLGC. Steady
state emission factors for the target organic pollutants obtained with the reference methods were
compared to the integrated values obtained with the REMPI-TOFMS system.
34
-------
EPA/600/R-10/050
April 2010
Table 6-1. EPA sampling methods.
Class of
Targets
VOCs
Semi Volatile
Organic
Compounds
(SVOC)
Carbonyl
CO
02
CO2
EPA
Methods
M-0040
M-8260b
M-0010
M-8290a
M-0011
M-8315a
Method 10A
Method 3A
Reference and Web-Link
SW-846 "Test Methods For Evaluating Solid Wastes, Physical/Chemical Methods,"
Office of Solid Waste, Method 0040 "Sampling Of Principal Organic Constituents From
Combustion Sources Using TEDLAR Bags," 1996.
http://vwvw.epa.qov/epaoswer/hazwaste/test/pdfs/0040.pdf
Method 8260b "Volatile Organic compounds by Gas Chromatography/Mass
Spectrometry (GC/MS)," 1996.
http://www.epa.qov/epaoswer/hazwaste/test/pdfs/8260b.pdf
SW-846, Test Methods For Evaluating Solid Wastes, Physical/Chemical Methods, Office
of Solid Waste, Method 0010 Modified Method 5 Sampling Train, 1986.
http://www.epa.qov/epaoswer/hazwaste/test/pdfs/0010.pdf
http://www.epa.qov/epaoswer/hazwaste/test/pdfs/8290a.pdf
U.S. Environmental Protection Agency, Method 8290a "Semi volatile Organic
Compounds by HRGC/High Resolution Mass Spectroscopy (HRMS)," 1998.
SW-846 "Test Methods For Evaluating Solid Wastes, Physical/Chemical Methods,"
Office of Solid Waste, Method 001 1 "Sampling for Selected Aldehydes and ketone
Emissions from Stationary Sources," 1996.
http://www.epa.qov/epaoswer/hazwaste/test/pdfs/001 1 .pdf
SW-846 "Test Methods For Evaluating Solid Wastes, Physical/Chemical Methods,"
Office of Solid Waste, Method 831 5a "Determination of Carbonyl Compounds by High
Performance Liquid Chromatography (HPLC)," 1996.
http://www.epa.qov/epaoswer/hazwaste/test/pdfs/8315a.pdf
Determination of Carbon Monoxide Emissions in Certifying Continuous Emission
Monitoring Systems at Petroleum Refineries, http://www.epa.qov/ttn/emc/promqate/m-
10a.pdf
Determination of Oxygen and Carbon Dioxide Concentrations in Emissions from
Stationary Sources http://www.epa.qov/ttn/emc/promqate/m-03a.pdf
Samples from the diesel generator exhaust were taken for the REMPI-TOFMS through a 0.64 cm
stainless steel line at typical flow rates of 2 L/min (at STP). Since the REMPI-TOFMS measures
vapor phase analytes only, the sample was collected under sub-isokinetic conditions and filtered
through a micro-fiber filter (at 150 °C) to avoid PM reaching the valve inlet system of the
REMPI-TOFMS instrument. The filter was a Unique Product International heated filter element
(Model FLT-1584A) comprised of a glass/polytetrafluroethylene (PTFE) fluorocarbon micro-
fiber composite, with a paniculate efficiency of 99.9% at 1.0 um. A flexible, 9 m long, 1.27 cm
diameter, Silico-Steel transfer line conveyed the exhaust gas to the REMPI-TOFMS instrument.
This line was kept at 150 °C while sampling exhaust gases in order to minimize adsorption to the
line walls.
An OLGC system (Ryan et al., 1998) was used for this project to aid in the identification of
volatile organic and organo-chlorinated compounds. The OLGC system contains a heated sample
delivery system, a purge and trap sample concentration system, and the GC analytical system.
The sample concentration device is a Tekmar LSC-2000 thermal desorption unit, modified to
accommodate the direct collection of combustion samples from the stack or from a Tedlar bag.
35
-------
EPA/600/R-10/050
April 2010
The GC analytical system is an HP 5890 series II GC equipped with both a flame ionization
detector (FID) and an electron capture detector (BCD).
On-line monitoring of the diesel exhaust PM size distribution was performed using a Dekati Ltd.
ELPI. Particles from the flue gas were sampled through a unipolar corona charger. The electric
current carried by the charged particles into each impactor stage is measured in real time by a
sensitive multi-channel electrometer. The particle collection into each impactor stage is
dependent on the aerodynamic size of the particles. Measured current signals are converted to
aerodynamic size distribution using particle size dependent relations describing the properties of
the charger and the impactor stages.
The particle-bound PAHs were monitored using a real-time PAS 2000 from EchoChem
Analytics (Burtscher, 1992; Dunbar et al., 2001). This analyzer photoionizes PAHs on the
surface of particles while the gas molecules and particles themselves remain neutral. The PAH-
laden particles from the sample flow pass through a quartz tube with an excimer lamp mounted
around it. The lamp operates at high frequency and high voltage, ionizing the surface of the
particles. The charged particles then flow through a short tube to remove all negatively charged
particles. Subsequently, the charged particles are collected on a filter element which is mounted
in a Faraday cage. An electrometer measures the ion current associated with the charged particles,
which is proportional to the concentrations of the PAHs (three or more aromatic rings) bound on
their surface.
The steady state emission results obtained by REMPI-TOFMS for organic compounds were
compared with those obtained via EPA standard Methods 0040 (volatiles), 0011 (carbonyls), and
0010 (semi-volatile PAHs) and the OLGC technique. The method comparison used emission
data obtained during nine sampling runs that occurred on three consecutive days. Since the
reference methods are cumulative sampling techniques that provide a time-averaged
concentration value, direct comparisons can only be made with the integrated values obtained
with REMPI-TOFMS. Only compounds that had concentrations above the detection limit for all
three measurements techniques were used. Since m- and p-xylene co-elute in the GC/MS
analysis, the REMPI-TOFMS-based concentrations for these isomers were summed for
comparison. Sub-isokinetic sampling for REMPI-TOFMS (to avoid particles in the detector) was
equivalent to the isokinetic reference methods; even the semi-volatile PAH compounds were all
found within the gas phase fraction, allowing direct comparison of the REMPI/TOFMS with the
reference method measurements.
The system performance of the REMPI-TOFMS was evaluated based on the relative accuracy
(RA) measure (EPA, 2000a). The results presented here only provide a relative comparison of
this technique with promulgated EPA techniques based on a limited number of parallel sampling
tests (N = 5 and 3 for BTEX and naphthalene compounds, respectively). The RA is defined in
this work as the absolute mean between the measured values of a specific compound determined
by the REMPI-TOFMS system and the values determined by the reference method (RM) plus the
2.5% error confidence coefficient of a series of tests divided by the mean of the reference
method.
36
-------
PI
' '
EPA/600/R-10/050
April 2010
\d + CC
* _ xlOO
KM (1)
where ' ' is the absolute value of the mean of the difference between /' pairs of data defined as
i
, ,
^^^ \/~1/~1\
n i=l in which n is the number of parallel data points, I I is the absolute value of the
'0.
.
confidence coefficient defined as ^" in which d is the standard deviation and £0.975 is the
Student's t distribution for 0.975, and RM'is the average of the reference method samples.
6.2 Results and Discussion
6. 2. 1 Steady State Emissions
Steady state emission measurements were compared between on-line REMPI-TOFMS, OLGC,
and the standard reference methods that use extractive sampling and analysis for volatile and
semi-volatile organic air toxic compounds (Table 6-2). Results were also compared with an
emission factor for a related source (large stationary diesel) published in an EPA database (EPA,
1993), albeit under a different test cycle. In general, the emission factors measured by REMPI-
TOFMS versus the reference methods vary by less than a factor of 2, which demonstrates the
accuracy of the REMPI-TOFMS as an organic air toxic emission monitor. Generally good
agreement was observed for the REMPI-TOFMS benzene emission measurements when
compared to the OLGC results and the corresponding EPA reference method (Table 6-3).
While additional tests would likely have improved the RA, these values are still noteworthy; for
the m- and p-xylene and toluene, the RA was determined to be around 60-80%. Higher
naphthalene concentrations were observed with the REMPI-TOFMS compared with the
reference method, resulting in a RA of more than 110%.
In general, standard deviations of the REMPI-TOFMS-measured compounds are in agreement
with the corresponding standard deviations of the respective compounds measured with the
OLGC or the reference methods. This suggests that the precision of each method is relatively the
same.
6.2.2 Emissions during Startups
Cold and hot startups of the diesel engine caused immediate, sharp emission peaks for benzene
and other BTEX compounds, as well as for all gas-phase PAHs and methyl-substituted PAHs as
measured by REMPI-TOFMS. The very short response time of the REMPI-TOFMS for each
data point (1 s for a 10-laser-shot average ion signal) was sufficient to resolve the transient
emissions for each individual target organic air toxic; results for benzene and naphthalene are
illustrated in Figure 6-1. The cold start benzene peak intensity (-15 ppm), which represents the
maximum benzene concentration measured, was found to be about 100 times larger than the
steady state emission levels (~ 150 ppb, 1 min average) recorded after 5 minutes of diesel run
time. Similarly, the cold start naphthalene concentration (~ 1.5 ppm) was 14 times larger than its
steady state emission level (-110 ppb, 1 min average). This concentration behavior is consistent
37
-------
EPA/600/R-10/050
April 2010
with the presence of unburned fuel during the initial fuel-rich combustion reported in Boesl et
al., (1998, 2000). The CO, CC>2, and temperature traces (Figure 6-1) also reflect the changes in
combustion conditions as the engine cylinders heat up and combustion efficiency improves.
Benzene (ppmji
t t
c
S
co
2
o
O
c
ro
CO
I
8=
O
-------
Table 6-2.
EPA/600/R-10/050
April 2010
Air toxic emission factors determined by jet REMPI-TOFMS, OLGC, and EPA reference methods, and listed in
AP-42.
Compounds
Benzene
Toluene
Ethylbenzene
Xylenes
m,p-xylenes
m-xylenes
p-xylenes
o-xylenes
Styrene
Chloro benzene
Hexane
Formaldehyde
Acetaldehyde
Acrolein
Naphthalene
Acenaphthylene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benz(a)anthracene
Chrysene
Benzo(b)fluoranthene
Benzo(k)fluoranthene
Benzo(a)pyrene
lndenol(1 ,2,3-cd)pyrene
Dibenz(a,h)anthracene
Benzo(g,h,l)perylene
MW
(g/gmole)
78
92
106
106
106
106
106
106
104
113
86
30
44
56
128
152
154
166
178
178
202
202
228
228
252
252
252
276
278
276
REMPI-TOFMS
Average
kg/kWh
1 .31 E-06
7.93E-07
1 .03E-06
1 .20E-06
7.64E-07
6.48E-07
1 .36E-07
4.20E-07
5.00E-07
3.63E-08
ppb
156
80
90
106
67
57
12
37
44
3
Stdev
%
9.3
23.2
d
23.2
27.6
26.4
36.5
15.1
12.9
a
a
1 .51 E-06
110
25.2
a
a
2.48E-07
6.38E-07
14
33
c
c
a
a
a
Non-Detect (ND)
ND
ND
ND
ND
ND
ND
ND
OLGC
Average
kg/kWh
1 .28E-06
6.84E-06
6.66E-06
1.17e-06
ppb
152
142
120
a
103
a
Stdev
%
16.5
15.0
10.8
10.1
a
a
b
<2.42E-09
<1 .85E-09
<0.2
<0.2
a
a
Reference Method
Method
M0040
M0011
M0010
Average
kg/kWh
1 .57E-06
1.18E-06
4.80E-07
ppb
187
119
42
Stdev
%
19.0
21.3
15.1
a
1 .44E-06
126
14.0
a
a
a
b
2.66E-08
9.25E-08
4.58E-06
8.60E-08
3.86E-07
1.10E-06
5.66E-08
8.46E-09
1 .43E-07
4.34E-07
<8.96E-09
1 .22E-08
3.54E-08
2.22E-09
2.47E-09
<8.96E-09
<8.96E-09
<8.96E-09
<8.96E-09
<8.96E-09
<8.96E-09
2.2
10.0
1384
18
63
80
3.4
0.5
8.0
22.6
O.4
0.6
1.6
0.1
0.1
<0.3
O.3
<0.3
<0.3
<0.3
<0.3
23.1
14.3
58.0
50.8
48.7
11.0
14.3
7.9
3.2
10.6
15.6
16.1
45.3
37.1
AP-42 (d)
Average
kg/kWh
1 .06E-06
4.26E-07
2.97E-07
1 .22E-07
3.90E-08
1 .22E-08
2.01 E-07
1 .43E-08
7.25E-09
1 .98E-08
6.32E-08
1 .90E-09
6.24E-09
5.74E-09
9.63E-10
2.37E-09
1 .72E-09
<3.37E-10
<3.98E-10
<6.41E-10
<5.36E-10
<8.61E-10
ppb
126.0
43.0
26.0
36.9
8.2
2.0
14.6
0.9
0.4
1.1
3.3
<0.1
0.31
0.26
0.04
0.10
<0.06
O.04
O.04
O.07
<0.06
<0.03
Notes: (a) immeasurable with this technique; (b) not attempted; (c) one data point; (d) EPA, 1993; ND = non-detect.
39
-------
EPA/600/R-10/050
April 2010
Table 6-3. RA of the three measurement methods during parallel sampling.
Method
Compounds
benzene
toluene
m-,p-xylene
naphthalene
REMPI-TOFMS
Average
ppb
156
80
67
110
sd
%
9.3
23.2
27.6
25.2
OLGC
Average
ppb
165
142
103
NAb
sd
%
16.5
15.0
10.2
NAb
RMa
Average
ppb
182
116
121
80
sd
%
19.0
21.3
14.0
3.2
RA
REMPI-TOFMS/RM
OLGC/RM
%RM
36
59
79
114
36
39
34
NAb
aRM = reference method+
bNot available
Similar transient concentration phenomena were obtained for hot start events. Figure 6-1 also
illustrates the concentrations response of benzene and naphthalene during a triple repeat of the
hot start-ups. However, peak concentrations were lower and durations to reach steady state
concentrations were generally shorter than with the cold starts. The transient concentration
responses to changes in operating conditions are a function of their mechanism of formation and
their mass transfer properties to the detector. Concentration levels are a result of in situ formation
processes and survival of unburned parent fuel. For pollutants whose concentrations are
dependent predominantly on formation reactions inside the cylinder, transient concentrations
may last only few seconds. Compounds that are present as a result of unburned fuel droplets are
more closely related to the time for the engine block temperature to reach thermal equilibrium.
This time is source dependent, could last hundred of seconds (Rakopoulos et al., 1998), and will
be shorter for warm startups in which the combustion chamber walls already exceed ambient
temperatures.
The rate at which pollutants retreat from the peak concentration to the steady state level may also
be, in part, an artifact of their transfer time to the detector. Tests changing the sample line
temperature from 150 °C to 200 °C resulted in a 1 to 3 s reduction of the cold startup peak width
for benzene and toluene and a 35 s reduction (from 76 to 41 s) for naphthalene (the peak width
duration is defined from cold startup until double the steady state concentration). This
phenomenon is likely due to transfer line wall adsorption of the target analyte and, hence, is a
function of the pollutant's vapor pressure.
REMPI-TOFMS was able to resolve emission concentrations during start-ups, as well as the
differences between the two types of start-ups (cold and hot), because of its high sampling
frequency. Standard sampling methods required extended sample collection time with Tedlar
bags and were unable to resolve these transients. The OLGC showed elevated concentrations
during start-ups, but since the sampling time was set at 5.6 min, this period was well in excess of
the transient peak durations observed by REMPI-TOFMS.
No solid-bound PAHs were observed by the PAS during the cold and hot startups (see Figure
6-2). This may be misleading for two reasons. First, combustion-derived PM may be contained
within a mist of small, re-condensed droplets of unburned fuels. These droplets are not easily
40
-------
EPA/600/R-10/050
April 2010
photo-ionized and remain neutral, hence, undetectable by the PAS. Second, there also may be
extensive dilution of the PAH mass due to the large number of particles present at startup. The
solid-bound PAHs first appear after the transient peaks of the benzene (for example), and their
concentrations steadily subside with run time.
350-
300-
250-
O
100-
50-
r&Ap
Linpo°-
- Temperature
-O- Solid-Phase PAHs
#-
120
100
80
60
40
20
0 200 400 600 800 2000 2250
Run Time (Seconds)
2500
*
©
o
"-ii
2
-i'
I
o
O
I
0.
Figure 6-2. Temporal temperature and particle-bound PAH concentrations in the diesel
exhaust.
6.2.3 Emissions during Load Variation
The modal (load change) characterization of the targeted air toxics, solid-bound PAHs,
temperature, and CO/CO2 were determined by the REMPI-TOFMS, the PAS analyzer, duct
thermocouples, and the continuous emission monitor (CEM) system, respectively. Seven 10 kW
increments in load, from no load to 60 kW, resulted in benzene, naphthalene, CO, and CC>2
concentrations shown in Figure 6-3 while particle-bound PAH measurements are shown in
Figure 6-4. The relatively higher concentrations observed during the no-load condition are likely
due to the same cold startup phenomena observed earlier. The no-load conditions resulted in
higher aromatic and CO concentration than the low-load (10 kW) condition. Further increases in
load resulted in compound-specific concentration trends: benzene increased in concentration at
higher loads, whereas naphthalene showed little apparent change. Methylated naphthalenes (not
shown), determined only with REMPI-TOFMS, also increased. Increased pollutant
concentrations are likely due to the higher fuel to air ratio at higher loads, also reflected by the
increase of the concentrations of CO and CO2 (Figure 6-3) and the decrease in 02. When the
emission concentrations are evaluated at a constant oxygen level at the stack, minimal variation
between load levels (from 10 to 50 kW) is noted (about 12% Sd for CO and less than 0.6% Sd for
CO2). The no-load and full load conditions had slightly higher CO emission rates that the other
41
-------
EPA/600/R-10/050
April 2010
load conditions. The exhaust gas temperature at the sampling location increased linearly with the
load due to an increase in the heat release rate and lower air/fuel ratio in the combustion
chamber.
Load
300n
250-
200-
150-
100-
50-
0
No Load ; 10kW ; 20 kW
- benzene (ppb;)
30 kW
40 kW
50 kW
60 kW
naphthalene (ppb)
10
20
30 40 50
Run Time (min)
60
70
Figure 6-3. Load change emissions of benzene, naphthalene, CO, and COi with exhaust
temperature.
42
-------
EPA/600/R-10/050
April 2010
380
O
10
20
30 40 50
Run Time (min)
60
70
Figure 6-4. Temperature distribution and particle-bound PAH concentrations under
varying loads.
43
-------
EPA/600/R-10/050
April 2010
7. Source 2: Real-Time Measurement of Trace Aromatics during Operation
of Aircraft Ground Equipment
This chapter describes work that extends the REMPI-TOFMS emission factor determinations to
a U.S. Marine Corps gas turbine, contrasting the HAP concentration trends during startups,
shutdowns, and load changes and demonstrating the ability to monitor these trace pollutants.
Concentrations were compared to published emission factors available in EPA databases and the
referenced literature to determine the efficacy of the REMPI-TOFMS method.
Two gas turbine units using JP8 fuel, included in the general category of aerospace ground
equipment (AGE), were previously characterized (Gerstle et al.,1999). The AGE classification
system includes aircraft support equipment such as air compressors, floodlights, bomb lifts,
turbines, generators, and heaters. Auxiliary power units (APUs) (GTCP85-180 and GTCP165-1)
from C-130H and C-5A/B aircraft, respectively, were tested at constant power settings, or about
120 kg fuel/h. Ambient air samples were taken to allow for background correction. The authors
found from testing with these and aircraft engine sources that of the 120 or so compounds
identified, only a few compounds were universally detected, suggesting that emission
characterization is fairly engine or source specific. Benzene, toluene, and xylenes were prevalent
and formaldehyde was over 90% of the aldehydes/ketone present. The authors mentioned the
possibility of reducing the characterization effort by monitoring a few surrogate compounds, but
were unable to test this theory due to an insufficient database (lack of replicates on a single
source).
7.1 Experimental
7.1.1 AGE
The exhaust of a turbine engine compressor, USAF type A/M32A-95 (Large Aircraft Starting
unit), was sampled with the REMPI-TOFMS system. This AGE (Figure 7-1) is used to furnish
pneumatic power for ground support of aircraft systems. Its primary mission is to start engines
for a variety of aircraft. The turbine engine, fuel and electrical systems, as well as the air delivery
system are enclosed inside a four-wheeled, towable cart. The cart contains a 300 L metal fuel
tank which holds JP-8, a kerosene-based fuel (MIL-T-83133) used as the standard military fuel
(U.S. Army, 2001). It is comprised of paraffins, olefins (< 5% by vol), and aromatics (< 25% by
vol) with less than 0.3% sulfur (S) by mass and a minimum net heat content of 42,700 kJ/kg
(Kimm et al., 1997). Annual fuel consumption per AGE unit is typically low (on the order of 750
L/year) but the large number of units make AGE a high category of total usage. The JP-8 fuel
headspace above the liquid layer of a nearly empty fuel barrel (at 25 °C) was sampled into the
REMPI-TOFMS sampling line. Concentrations of target analytes were determined by calibrating
the measured ion signals with those from a TO-14 Aromatics Subset (Supelco) gas standard
cylinder mixture containing 100 ppb of aromatic analytes (BTEX).
44
-------
EPA/600/R-10/050
April 2010
Figure 7-1. Aircraft ground equipment outside EPA facilities
7.1.2 Operating and Sampling Procedures
The AGE was operated for two days to characterize its emissions under idle and full load
conditions as well as start up and shut down scenarios. The idle and full load operating periods
ranged from 1 to 4 h duration, interspersed with refueling times. Consecutive, ~ 2 h cycles of
start up and shut down were completed over a 6-7 h sampling day.
REMPI-TOFMS sampling was conducted using 5 cm diameter ports located about 30 cm below
the top of a 4.9 m high, 0.84 m x 0.28 m rectangular carbon-steel duct covered with 7.6 cm
fiberglass insulation. This duct was built to convey the AGE exhaust away from building air
intakes. A 2 L/min isokinetic sample was pulled through a 15 m, 0.95 cm (inner diameter) heated
(150 °C) transfer line to a heated (150 °C), filter. Filters were exchanged at least three times per
day (1 to 3 h sampling time) to minimize capture of the target analytes on the carbon-laden filter.
Beyond the filter, a 3 m long, 1.25 cm diameter silico steel coated transfer line (at 150 °C)
conveyed the sample to the REMPI-TOFMS instrument. A small quantity of 1 ppm CeDe in N2
calibration gas was added upstream of the filter as an internal calibrant. REMPI-TOFMS
sampled directly from the 2 L/min exhaust with a slipstream of approximately 1 mL/min.
Fourteen pre-determined wavelengths in the A, = 258 - 275 nm range were selected, based on
established wavelength scan libraries, for their high probability of detecting discrete, individual
(isomer selective) compounds. This wavelength changing sequence method was performed
during steady state idle (low) and full load AGE operational modes where changes in
concentration were expected to be gradual. The REMPI ion signal of a target analyte at its
specific wavelength was recorded for 1 minute before switching to the next wavelength. This 15
minute sequence was repeated multiple times within the sampling time period as defined by the
conventional extractive sampling methods. In addition, cold/hot starts and shutdowns were
followed in real time at the most effective wavelength for toluene ionization (267 nm) without
additional wavelength changes in order to eliminate losses in data acquisition during wavelength
switches (~ 2 s).
45
-------
EPA/600/R-10/050
April 2010
Conventional extractive sampling techniques, based on EPA Methods, were used to validate the
integrated REMPI/TOFMS measurements taken over the same sampling time. VOCs were
sampled in parallel with the REMPI-TOFMS, using standard extractive sampling for volatiles
from the duct into stainless-steel Summa canisters and analyzed by GC/MS (EPA 1999) for 60
target compounds and for unknown VOCs as tentatively identified compounds (TICs) using the
National Institute of Standards and Technology (NIST) spectral database (NIST, 2005). Nine
samples from the turbine stack were taken during the two days of sampling, four at idle load and
five at full load. All samples were extracted from a 0.65 cm (1/4 inch) port through a 0.65 cm
(1/4 inch) Teflon tube. All summa canister samplings were started with an initial vacuum of 76.2
cm (30 inches) Hg and ended with 1.2 to 0 cm Hg (3 to 0 inches of Hg) Two summa canisters
were also used as a feed into the REMPI-TOFMS instrument as well as the GC/MS for further
validation of the methods.
PAHs were sampled using EPA Method 0010 (EPA, 1986a) and Method 5 (EPA, 1986b)
sampling, and analyzed via a modified EPA Method 8270 (EPA 1996b) using a toluene
extraction solvent and deuterated and fluorinated pre-sampling spikes. Formaldehydes and other
carbonyl group compounds were sampled via EPA Method 0011 (EPA, 1996c). This method
uses a sample train configuration adapted from EPA Method 5, wherein the sample gas is
collected in an aqueous, acidic 2,4-dinitrophenyhydrazine (DNPH) solution, and analyzed via
EPA Method 8315A (EPA, 1996d) using high performance liquid chromatography (HPLC).
Results of the carbonyl groups were provided to develop additional emission factors for the AGE
and not for the purpose of the REMPI/TOFMS validation.
Emissions factors were calculated by dividing the concentration of the compound by the amount
of fuel used, or g/kg of fuel. Graphical results are reported as the value + 1 standard deviation, a.
Tabular results are reported as RSDs divided by the average value, or a / average (AVG).
Comparisons between measurement methods are made with matched pairs of each target analyte
using a paired t test.
7.2 Results and Discussion
REMPI-TOFMS traces during AGE startups, low load, and high load operation show complex
emissions of varying concentrations. Figure 7-2 shows a 3D time-resolved trace of the emissions
during a cold start (commencing at 2 min, 20 s) followed by about 5 min of low load operation.
The concentration is proportional to the ion signal strength and the mass values highlight the
different compounds that are observed. The cold start results in a significant peak of toluene (m/z
= 92) as well as a number of other methylated naphthalenes (m/z = 128 to 184). The trace shown
is optimized for toluene detection at A, = 267.5 nm. At this wavelength, methylated naphthalenes
are also efficiently ionized while benzene and other single ring aromatics are marginally
detected. The initial concentrations decline within 1 min, then are mostly constant during the low
load operation. At 6 min and 53 s a change to high load shows signal reduction across the whole
mass range, indicating a large drop in the measured concentrations. The lower concentrations are
expected due to dilution from the increased throughput air in the turbine and increased
combustion efficiency.
46
-------
EPA/600/R-10/050
April 2010
toluene
benzene
Att= 2 min,20s:
Coldstartto idle load
Att = 6 min,53s:
Idle to high load
Figure 7-2. REMPI-TOFMS time-resolved trace of cold startup, idle (low) load, and high
load conditions.
Note: A = 267.5 nm, optimized for toluene detection, but not calibrated externally.
A wavelength scan of the steady state, low load emissions (Figure 7-3) shows discrete spectra for
a number of individual compounds including toluene, phenol, styrene, and individual xylenes
isomers. When the wavelength spectra from two selected mass numbers (m/z 92 and 106) from
Figure 7-3 are compared to those obtained with a calibrated gas mixture (Figure 7-4), no
wavelength interferences are observed, allowing for unambiguous concentration determinations.
Also note that isomer separation for xylenes is possible using the moderately broad spectral line-
width of the laser system. Larger PAH analytes, such as methylated naphthalenes, are visible in
Figure 7-3 but their characteristically broad wavelength spectra obscures spectroscopic details
that could otherwise have been used for individual isomer identification. The observed emission
compounds have little overlap with a headspace analysis of the JP-8 fuel drum (not shown),
suggesting that the majority of the operating emissions are due to secondary organic byproduct
formation rather than emission of unburned fuel. The predominance of benzene, toluene, and
phenol in the emissions are not noted in the fuel headspace, which is dominated by xylenes and
trimethylbenzene concentrations.
Analysis of the AGE emission traces enables determination of major compound concentrations.
With the laser set to a wavelength specific for each compound, real time integrated
concentrations were determined during idle and high load operation. These REMPI-TOFMS-
derived concentrations were compared to standard EPA sampling and analysis methods. REMPI-
TOFMS results were first compared against a standard analysis of emission gases gathered into a
common Summa canister during low load operation. Figure 7-5 shows two replicate tests of
REMPI-TOFMS BTEX measurements and two tests via standard GC/MS, both from the Summa
canisters. Good agreement is observed, validating the REMPI-TOFMS quantification with the
standard methods.
47
-------
EPA/600/R-10/050
April 2010
.trimethylbenzenes, 120
xylenes, 106
styrene, 104
phenol, 94
Figure 7-3. REMPI-TOFMS wavelength-resolved trace during idle (low) load condition.
The REMPI-TOFMS on-line measurements were then compared against emissions sampled in
parallel via EPA standard Method TO-15 (EPA, 1999). Figure 7-6 shows a comparison of idle
(low) load emissions from two sequential testing phases, with the Phase I and II TO-15 analyses
done by a commercial laboratory then an EPA laboratory, respectively. Duplicate Phase I
REMPI-TOFMS measurements predominantly fall within the + 1 standard deviation range of the
two GC/MS analyses. No apparent explanation was found for the wide disparity of the
commercial laboratory analyses. The REMPI-TOFMS results were reasonably precise suggesting
the actual emission variations were minimal over the short sampling period (~ 30 min). In Phase
II, duplicate REMPI-TOFMS results agree quite well with the TO-15 standard analysis provided
by our in-house laboratory.
Comparison of the time-resolved REMPI-TOFMS-determined concentrations at high and idle
(low) load versus the CO and CO2 concentrations (Figure 7-7) shows no apparent time-resolved
relationship between combustion efficiency and trace organic emissions. Post-startup REMPI-
TOFMS results show benzene and toluene declining to about 30 to 20% of their original values
over a 5 h period while other organics, CO, and CO2 appear relatively constant. These declines
occur throughout several startup/high load cycles, showing progressively declining organic
emissions with time. Here, however, changes in CO2 level appear to be inversely related to a few
compounds, most notably benzene, toluene, and phenol.
48
-------
EPA/600/R-10/050
April 2010
266
Mass 92, toluene
Reference Spectrum toluene
268
270
272
274
276
m-xylene
p-xylene
Mass 106
Reference Spectrum
xylenes + ethylbenzene
266 268 270 272
Wavelength (nm)
274
276
Figure 7-4. Comparison of (selected) AGE wavelength spectra versus the TO-14
calibrated gas standard.
49
-------
EPA/600/R-10/050
April 2010
250
200
150
.a
a.
o
is
r 100
-------
EPA/600/R-10/050
April 2010
.a
a.
o.
o o
§
o
O
Q.
Q.
J
* l
A. 5.
i
i
High
Load
1 I
'co o '
-o
i
i
1
*m
m
-*W
High
Load
benzene toluene
phenol styrene
p-xylene m-xylene
o-xylene ethylbenzene
1 ,2,4 trimethylbenzene
1 ,3,5 trimethylbenzene
naphthalene fluorene
"
i c
'OT 0
-a
T
1
'
m
High ,
Load
I 1
OT 0 '
~o
xx
* *
* * -
** -
j^
Idle
c
B
(D
500 -
400
o
300
200
n 400
10:00 11:00 12:00 13:00 14:00
Time (h:min)
15:00
16:00
17:00
Q.
Q.
Figure 7-7. REMPI-TOFMS and CO and CO2 CEM concentrations during a series of
startups (the initial startup was a cold start) and shutdowns (for refueling),
operating at high, high, then idle load. Range of data (barely visible) are
shown for benzene.
Emission factors determined by REMPI-TOFMS as well as standard methods TO-15 for BTEX
compounds, 0010 for PAHs, and 0011 for aldehydes are shown in Tables 7-1, 7-2, and 7-3,
respectively. Data are presented for both idle load and high load steady state emissions.
Comparisons are also shown for published emission factors from related AGE sources operating
under full (high) load.
Idle load emission factors of BTEX compounds (Table 7-1) at 23 kg of fuel per hour are
typically about two orders of magnitude higher than full load emission factors at 78 kg of fuel
per hour. This suggests that emissions from operation of AGE units under idle conditions may be
significantly underestimated if they are calculated simply from fuel consumption and the full
load emission factors.
51
-------
EPA/600/R-10/050
April 2010
Table 7-1. Emission factors for BTEX compounds during idle and full load operation. Comparison with published values.
AGE Type
Operating Condition
Sampling Technique
VOCs
Benzene
Toluene
Ethylbenzene
m-Xylenes
p-Xylene
o-Xylene
m,p-Xylenes
Xylenes
Styrene
1,3-Butadiene
A/M32A-95 (this work)
Idle
TO-15
REMPI-TOFMS
g/kg of fuel
Value
9.8E-01
4.1E-01
1.0E-01
a
a
1.7E-01
2.4E-01
3.5E-01
2.1E-01
2.0E+00
RSD (%)
70.1
74.1
81.6
a
a
45.4
84.1
82.8
95.5
0.1
Value
7.0E-01
2.7E-01
1.2E-01
4.6E-02
1.1E-01
8.3E-02
1.6E-01
2.4E-01
1.7E-01
a
RSD (%)
1.1
0.8
5.7
1.4
1.1
1.2
1.2
1.2
1.4
a
Full
TO-15
REMPI-TOFMS
g/kg of fuel
Value
8.6E-03
5.0E-03
2.3E-03
a
a
5.4E-04
3.4E-03
3.8E-03
3.8E-04
2.6E-03
RSD (%)
88.6
107
135
a
a
97.8
120.0
105.0
47.4
97.4
Value
3.3E-03
2.8E-03
7.1E-05
8.1E-05
1.8E-04
7.5E-05
2.6E-04
3.4E-04
4.8E-04
a
RSD (%)
1.7
1.0
18.3
6.2
3.6
5.6
4.4
4.7
4.5
a
GTCP85-180d
GTCP-165-1d
Full
TO-15
g/kg of fuel
Value
1.5E-02
4.4E-03
1.2E-04
a
a
3.3E-04
2.4E-03
2.7E-03
1.9E-04
b
Value
3.9E-02
1.9E-02
8.8E-04
a
a
1.2E-03
4.9E-04
1.7E-03
2.3E-03
b
(a): Immeasurable with this technique
(b): Non-detect
(c): Detected but not quantified due to potential wavelength overlap
(d): Gerstle et al. (1999)
52
-------
EPA/600/R-10/050
April 2010
Table 7-2. Emission factors for PAH compounds during idle and full load operation. Comparison with published values.
(b): Non-detect
(c): Detected but not quantified due to potential wavelength overlap
(d): Gerstle et al. (1999)
AGE Type
Operating Condition
Sampling Technique
PAH Compounds
Naphthalene
Acenaphthylene
Acenaphthene
Fluorene
Phenanthrene
Anthracene
Fluoranthene
Pyrene
Benzo(a)anthracene
Chrysene
Benzo(b+k)fluoranthene
Benzo(a)pyrene
lndeno(1 ,2,3-cd)pyrene
Dibenz(a,h)anthracene
Benzo(ghi)perylene
A/M32A-95 (this work)
Idle
M-0010
REMPI-TOFMS
g/kg of fuel
Value
1.56E-01
6.07E-03
RSD
(%)
3.9
48.0
b
1.66E-02
9.00E-03
5.10E-04
1.47E-03
2.22E-03
1.48E-04
2.19E-04
3.02E-04
9.00E-05
7.85E-05
b
1.15E-04
18.7
15.4
38.9
28.0
0.2
17.5
10.3
20.4
54.3
8.2
b
28.1
Value
7.97E-02
b
RSD
(%)
2.04
c
1.20E-02
5.24E-03
b
0.06
0.03
c
b
Full
M-0010
REMPI-TOFMS
g/kg of fuel
Value
5.60E-04
b
RSD
(%)
16.1
b
2.96E-05
2.11E-05
b
56.3
66.7
b
b
Value
2.05E-04
b
RSD
(%)
3.21
b
6.27E-05
1.13E-04
b
4.69
21.51
GTCP85-180d
GTCP-165-1d
Full
M-0010
g/kg of fuel
Value
b
Value
5.55E-03
b
53
-------
EPA/600/R-10/050
April 2010
Table 7-3. Emission factors for aldehydes compounds during idle and full load operation. Comparison with published values.
Operating Condition
Sampling Technique
Aldehydes Compounds
Formaldehyde
Acetaldehyde
Acrolein
Propionaldehyde
Idle
M-0011
Value
1.63E+00
4.10E-01
2.17E-01
8.00E-02
RSD (%)
29.9
35.7
21.4
37.0
Full
M-0011
Value
8.29E-03
1.70E-03
1.86E-04
1.74E-04
RSD (%)
43.2
50.8
80.7
35.4
Full
M-0011
Value
2.03E-02
2.09E-03
3.04E-04
b
Value
1.88E-02
5.62E-03
b
b
(b): Non-detect
54
-------
EPA/600/R-10/050
April 2010
Minimal differences (less than 2x) are observed between REMPI-TOFMS and TO-15 methods
for determining emission factors of BTEX compounds during idle (low) load operation (Table
7-1). The R2 values for the TO-15 and REMPI-TOFMS compound matched pairs are 0.99 and
0.84 for idle and high load conditions, respectively. For the idle tests, two of seven paired
compounds (benzene and ethylbenzene) were beyond the 95% confidence level for no
differences between the methods. During high loads, the most abundant compounds, benzene
and toluene, are within a factor of three. However, differences for some of the lower
concentration compounds exceed lOx, with REMPI-TOFMS recording lower values. Three of
the seven paired compounds (benzene, o-xylene, styrene) exceeded the criterion, indicating
differences in the methods' concentrations. The range of the TO-15 data (n=2) results in high
RSDs (commercial laboratory analyses only). This variation does not show on the simultaneous
REMPI-TOFMS results.
Published BTEX emission factors (Gerstle et al., 1999) for similar-type AGE units at full load
are predominantly higher than REMPI-TOFMS results (Table 7-1). Nonetheless, their matched
pair compounds with REMPI-TOFMS are correlated at R2 = 0.85 and 0.95 for the GTCP85-180
and the GTCP165-1 units, respectively.
PAH emission factors show few differences between REMPI-TOFMS and Method 0010
determinations (Table 7-2) both for idle and full load.
As with the BTEX compounds, PAH emission factors are at least an order of magnitude higher
for idle than for full load. The one published PAH emission factor (Gerstle et al. 1999) for
naphthalene is about SOX the REMPI-TOFMS value at full load. Aldehydes emission factors
(Table 7-3) by Method 0011 (Aldehydes are immeasurable by REMPI-TOFMS conditions
employed here since their ionization energy exceeds 10.5 eV) again show two orders of
magnitude higher emission factors for idle than full load; emission factors are generally
comparable to published values (Gerstle et al., 1999).
Finally, Figure 7-8 compares the benzene and naphthalene startup responses for the AGE and
diesel generator. The AGE maintained fairly constant concentrations of benzene and
naphthalene, while the generator showed rapid declines by about 1 min. The ability of REMPI-
TOFMS to discern distinctive startup responses in these units makes it an important instrument
for characterizing and distinguishing source emissions under varying conditions.
55
-------
EPA/600/R-10/050
April 2010
benzene AGE emissions
naphthalene AGE emissions
benzene DG emissions
naphthalene DG emissions
0.0
-1
Time (min)
Figure 7-8. Comparison of startup responses of benzene and naphthalene for the AGE
and diesel generator.
7.3 ORS Measurements during AGE sampling
An ORS system was added to the AGE test plan to perform with an OP-FTIR spectrometer
(manufactured by IMACC, Inc.). A second measurement using the IMACC OP-FTIR was
operated in radiance mode to collect data from the gas turbine exhaust.
7.3.1 Experimental Design
One OP-FTIR with one retro reflector was used for data collection. The retro reflector was set up
on scaffolding approximately 7 meters high, located adjacent to the engine exhaust stack. The
retro reflector was placed on a tripod and elevated so that it is slightly higher than the top of the
stack. The OP-FTIR instrument was mounted on a tripod on top of a vertical structure
approximately 2 meters high. The instrument was located such that the physical path length
between the instrument and the retro reflector was approximately 20 meters (see Figure 7-9). The
spectra were measured as single beams (I) that were signal averaged for 30 seconds, and
collected with a spectral resolution of 0.5 cm"1. The OP-FTIR operated in static scanning mode,
collecting data for several hours over a two day period. The data is stored as interferograms on
the data collection computer, and backed up to CD-ROM. The background spectra (lo) were
created synthetically using OP-FTIR S/W.
56
-------
EPA/600/R-10/050
April 2010
Turbine
Plume
Retro-
reflector
PlumeWidth =0.5 m
OP-FTIR
Figure 7-9. ORS measurement configuration.
7.3.2 ORS Instrument-Retro reflector Distance
The physical distance between the OP-FTIR instrument and the retro reflector was measured
using a Topcon, Inc. model GTS-21 ID theodolite. This distance is required for converting the
products of incomplete combustion (PIC) data to path average concentration (PAC) values. PAC
is equal to PIC divided by the optical path length. The optical path length is two times the
physical distance, accounting for travel to and from the retro reflector.
7.4 Data Processing
The concentrations were determined by a regression fit to reference spectra using the Beers Law
expression:
I/Io =A= -LoglO(aCL),
where:
I = single-beam measurement
lo = background spectrum
A = absorbance spectrum
a = molecular absorptivity (extinction coefficient)
C = concentration of the absorbing molecule
L = total path-length of the infrared beam through the molecular plume
The regression equation is written as:
Af(v) = ZrAr(v) + a + b v + s(v)
Data analysis was performed by the chemometric regression method that is referred to as
classical least squares (CLS) and is written as:
-------
EPA/600/R-10/050
April 2010
Af(v) = ZrAr(v) + a + b v+ s(v)
Where:
Af(v)= measured field absorbance spectrum
Ar(v) = reference spectrum of chemical species, r
a is a scalar that corrects for the baseline error
b is a scalar that corrects for the baseline slope
s(v) is the error or residual term
The summation is over all species, r, that absorb in the region of analysis.
7.4.1 ORS Results
The 2-days of data gathering using the ORS equipment resulted in the measurements of criteria
pollutants such as CO and CO2 as well as other compounds found in the absorption spectra such
as formaldehyde, ethylene, and aliphatic mixture, the later which presumably originated from the
fuel; n-octane is a major component of the aliphatic mixture and was used as a surrogate for the
entire mixture. Other components include n-nonane, n-heptane, n-hexane and probably some
bent-chain species.
Data validation was performed by comparing the reference spectrum of a target species to the
measured spectrum from the turbine plume. The trace emissions for formaldehyde, ethylene, and
octane are presented in Figures 7-10, 7-11, and 7-12, respectively.
The results of the ORS measurements were compared with stack measurements for CO and CO2,
obtained using regular CEM instruments. The results were comparable for CO; however, the
ORS results were not sensitive enough to differentiate between the high and low loads for CO2 as
illustrated in Figure 7-13. Furthermore, drift of the ORS concentrations from those of the CEMs
was observed throughout the day. This was due to an insufficient path length of the beam in the
relatively small plume as well as effects of diurnal temperatures changes. The ORS system was
not used for subsequent tests due to its limitations for measurement of point source emissions.
58
-------
EPA/600/R-10/050
April 2010
cc.
0.05
880 900 920 940 960
Wavenumber (cm-1)
980
1000
1020
Figure 7-11. Comparison of the reference spectrum of ethylene (red trace) to a measured
spectrum of the turbine plume (blue trace)
59
-------
I
6
i/>
.a
oi
2800
EPA/600/R-10/050
April 2010
2850 2900 2950
Wavenumber (cm-1)
3000
Figure 7-12. Comparison of the reference spectrum of n-octane (red trace) to a measured
spectrum of the turbine plume (blue trace)
12000
o
250
0
09:36 10:48 12:00 13:12 14:24 15:36 16:48 18:1
Time (EOT)
* C°2(CEM> CO(CEM)
A C02 (Open Path) 0 CO (Open Path)
Figure 7-13. Comparison between CEM and ORS measurements
60
-------
EPA/600/R-10/050
April 2010
8. Source 3: Verification Results of REMPI-TOFMS as a Real-Time
PCDD/F Emission Monitor
In 2005, REMPI with TOFMS was used in an international ETV test program monitoring levels
of PCDD/F in a hazardous-waste-firing boiler (Cowen, 2006). Prediction of PCDD/F levels was
based on prior determination of an indicator compound and intercorrelation of the indicator and
PCDD/F concentrations. This chapter reports the determination of indicator compounds and the
performance of REMPI-TOFMS during the ETV tests.
8.1 Materials and Methods
8.1.1 Boiler testing
An 860 KW capacity, 3-pass fire-tube, marine package boiler (Superior Boiler Works, Inc.),
firing #2 fuel oil was used as the combustion gas source (Figure 8-1). The flue gas from the
boiler passes through an exhaust duct to a manifold and then on to an air pollution control system
(APCS). The APCS consists of a natural-gas-fired secondary combustion chamber, a fabric filter,
and an acid gas scrubber to ensure proper removal of pollutants. The fuel oil was doped at a
constant ratio of a surrogate hazardous waste (1,2-dichlorobenzene, DCB) and a source of metal
catalyst (copper naphthenate) to promote PCDD/F formation. Fuel flows were measured with a
total liquid volume meter. The amount of injected copper was set to simulate that of a refuse-
derived fuel (RDF) ash (100 mg per kg of ash), conditions in which PCDD/F formation is
favorable. The feed-rate of the mixture was varied to achieve a high (200-500 ppm) and low (10-
50 ppm) target HC1 concentration in the flue gas with the expectation of generating relatively
high and low PCDD/F concentrations, respectively. Test runs of various durations were
conducted under each set of operating conditions. Four 4-hours and two 8-hour sampling periods
on successive days (i.e., totaling 16 hours per sample) were used to assess short-term and long-
term accuracy of the measurements.
The test campaigns sampled for PCDDs/Fs and semi-volatile PAHs using EPA methods 23
(EPA, 1996d), and 0010 (EPA, 1996a), respectively. Both Method 23 and Method 0010 consist
of a 125 °C heated probe, 125 °C heated box containing a filter, water-cooled condenser, water-
cooled XAD-2 resin cartridge, impinger train for water determination, vacuum line and pump,
and a dry gas and orifice meter.
61
-------
EPA/600/R-10/050
April 2010
Figure 8-1. Marine package boiler at EPA facilities.
The test campaigns sampled for PCDDs/Fs and semi-volatile PAHs using EPA methods 23
(EPA, 1996d), and 0010 (EPA, 1996a), respectively. Both Method 23 and Method 0010 consist
of a 125 °C heated probe, 125 °C heated box containing a filter, water-cooled condenser, water-
cooled XAD-2 resin cartridge, impinger train for water determination, vacuum line and pump,
and a dry gas and orifice meter.
All emission measurements for this work were taken prior to the APCS from a horizontal section
of the exhaust duct (20 cm diameter steel pipe) sufficient in length and free of flow disturbances
so that PM can be sampled in accordance with standard sampling requirements. Stoichiometric
ratios were verified through monitored C>2 and CC>2 emission concentrations. Continuous
emission monitors (CEMs) included four gas analyzers: high and low range CO (Rosemount
Analytical Model 880 non-dispersive infrared analyzer (NDIR), Range = 0-500 ppm), 62
(Rosemount Analytical Model 755 R, Range 0-25%, calibrated range 0-10%), and CO2
(Rosemount Analytical Model 880 NDIR, Range 0-20%).
8.1.2 REMPI-TOFMS Testing
Samples for the REMPI-TOFMS were taken through a 0.95 cm outer diameter stainless steel line
at typical flow rates of 2 L/min (all flows are reported at 101.325 kPa and 273.15 K). A flexible,
9 m long, 1.27 cm diameter, Silico-steel transfer line conveyed the exhaust gas to the REMPI-
TOFMS instrument. The line was kept at 150 °C while sampling exhaust gases in order to
minimize adsorption to the line walls. The sample was collected under sub-isokinetic conditions
62
-------
EPA/600/R-10/050
April 2010
and filtered through a micro-fiber filter (at 150 °C) to avoid PM reaching the valve inlet system
of the REMPI-TOFMS instrument. The REMPI-TOFMS underwent an in-situ calibration at the
start of each test using fully deuterated benzene, CeDe, added to the exhaust stream. Deuterated
benzene can be readily detected without spectral or mass interferences and is not an exhaust gas
component.
8.1.3 Indicator Compounds
The use of indicator compound measurements to predict PCDD/F values requires a pre-
determined correlation analysis to find the best indicator compound or compounds and to
determine its mathematical relationship with the PCDD/F measure. These indicator correlations
are likely to be source-specific, requiring an initial series of sampling tests for determining
compound concentrations from which to run correlation analyses. Selection of the PCDD/F
indicator compound was based on a "pre-ETV," six-test measurement campaign of 31
compounds (Table 8-1). These 31 compounds were selected, in part, based on the general
understanding that lower chlorinated compounds are more readily detectable with REMPI-
TOFMS than the higher chlorinated compounds (Zimmermann et al., 1999). These pre-ETV tests
replicated the high and low dopant loads that would occur under the subsequent actual ETV
study. A correlation analysis was run on the results between the PCDD/F TEQ and other semi-
volatile compounds sampled by EPA Method 0010 (EPA, 1996a). The highest correlation, R2,
between the REMPI-detectable compounds and the TEQ values, and one that has a positive (+)
coefficient, provides an indicator candidate.
Table 8-1. Candidate TEQ surrogate compounds from pre-ETV Method 0010 sampling.
Compounds:
2-MCDD
2,7- + 2,8-DiCDD
2,3-DiCDD
2,3,7-TriCDD
1-MCDF
3-MCDF
2-MCDF
naphthalene
2-Methylnaphthalene
1-Methylnaphthalene
2-Chloronaphthalene
1-Chloronaphthalene
Acenaphthylene
Dibenzofuran
Fluorene
Phenanthrene
Fluoranthene
Pyrene
Chlorobenzene
1,3-Dichlorobenzene
1,4-Dichlorobenzene
1,2-Dichlorobenzene
2-Chlorophenol
3- + 4-Chlorophenol
2,4-Dichlorophenol
2,5-Dichlorophenol
2,3-Dichlorophenol
2,6-Dichlorophenol
3,4-Dichlorophenol
Compound concentrations are determined by monitoring pre-determined, compound-specific
optimal wavelengths. After measurement of the emissions, an external calibration of the detected
ion signals was accomplished with a 100-ppb calibrated gas mixture in nitrogen containing 14
aromatic compounds, including monochlorobenzene (MClBz). The concentration of a target
analyte was derived from the ratio between its ion signal intensity and that of the same analyte
present in the calibration gas mix, normalized for any laser energy changes.
Time-integrated concentrations for the indicator compound, obtained with REMPI-TOFMS, are
entered into the predictive correlation equation obtained from the pre-ETV tests to determine the
63
-------
EPA/600/R-10/050
April 2010
TEQ estimate. This value is then compared with the results from the cumulative M23 sampling
methods. Sub-isokinetic sampling for REMPI-TOFMS to minimize particles reaching the filter
in front of the detector was assumed equivalent to the isokinetic sampling reference methods:
under lean firing conditions all of the volatile and semi-volatile compounds would be found
within the gas phase fraction, allowing direct comparison of the REMPI/TOFMS with the
reference method measurements.
The accuracy of REMPI-TOFMS during the ETV tests was determined by comparing results
from nine runs obtained with isokinetic sampling via Method 23 (EPA, 1996d). Simultaneous
extractive Method 23 samples were taken before and after the REMPI-TOFMS sampling port.
The correlation-derived TEQ estimation was compared with these EPA Method 23 results.
8.2 Results and Discussion
8.2.1 Pre-ETV results
Sampling during the pre-ETV phase with EPA Method 0010 (M10) resulted in measurable
concentrations for the compounds listed in Table 8-1. The PCDD/F TEQ value was derived from
a predetermined correlative relationship with the best single predictor chosen from amongst over
30 compounds consisting of chlorophenols, chlorobenzenes, and PAHs. A single-predictor
regression model using these M10 concentrations versus the EPA Method 23 TEQ
concentrations (ranging from 0.90 ng TEQ/Nm3 to 25.50 ng TEQ/Nm3 resulted in a best-fit
model (R2 = 0.7420) for TEQ using MClBz. MClBz was previously detected in the flue gas of an
incinerator using REMPI-TOFMS (Thanner et al, 1998, Zimmermann et al., 1999) and by SPI-
IT-TOFMS (Kuribayashi et al., 2005). Figure 8-2 shows the fit of a linear regression model of
TEQ on MClBz, based on the six lean-fire test runs with a 0.95 confidence belt for TEQ
predicted from the fitted model, TEQ = 1.23 + 53.85 MClBz.
8.2.2 ETVResults
In the ETV field test, REMPI-TOFMS monitored MClBz in real time to derive a concentration
for use in the predictor equation. The MClBz detection limit [signal/noise (S/N) = 3, 10 s signal
averaging time] was 1 ppb which was on occasion higher than the encountered concentration in
the exhaust gas. An extended time averaged mass spectrum at the optimal wavelength for MClBz
detection as shown in Figure 8-3 makes quantification possible. The calculated MClBz
concentration for this mass spectrum was 0.4 ppb. A total of nine tests with sampling durations
between four to sixteen hours were conducted, resulting in standard method PCDD/F
concentrations from 0.9 to 6.0 ng TEQ/m3.
64
-------
EPA/600/R-10/050
April 2010
a
LU
Q
Q
25-
20-
15-
10-
5-
0-
Pre-ETV DataB
Linear Fit of Data1_B
Upper 95% Confidence Limit
Lower 95% Confidence Limit
0.0 0.1 0.2 0.3 0.4 0.5
MClBz (|jg/m3)
Figure 8-2. Pre-ETV phase determination of MClBz as a PCDD/F TEQ surrogate
0.18-
~
§ 0.14-1
S 0.12H
"ro
§) 0.10-1
o 0.08-
LJJ
o:
0.06-
0.04-
0.02-
0.00'
35
CIBz
100 102 104 106 108 110 112 114 116 118 120
220
Figure 8-3. REMPI-TOFMS mass spectrum of MClBz with inset of relevant mass range.
0.1 ppb for a 4 min averaging period, S/N = 3.
65
-------
EPA/600/R-10/050
April 2010
The comparison of PCDD/F TEQ determined by the standard EPA Method 23 (M-23) and that
determined by the predictor equation using REMPI-TOFMS time-integrated measurements of
MClBz is seen in Figure 8-4. For the high waste-firing-rate with a range of PCDD/F TEQ values
between 3.9 and 6.0 ng TEQ/m3, the REMPI-TOFMS-predicted TEQ values had a relative
difference of 26% with the standard, EPA Method 23 results. At the low waste-firing-rate (0.9 to
1.6 ng TEQ/m3) the relative difference increased to 219%. This decrease in predictive capability
was attributable to the chosen predictor, MClBz, being below its GC method detection limit for
the lower firing rate. The uncertainty in the MClBz measurements was confirmed by evaluating
the waste feed rate versus the MClBz concentrations: no correlative agreement was apparent at
the lower concentrations, suggesting that the standard sampling and analytical methods for
MClBz were unreliable at these concentrations. This is more easily seen in Figure 8-5 where M-
10 and REMPI measurements have similar predictions of PCDD/F TEQ, but depart from the M-
23 results when PCDD/F TEQ has lower values. Improvements in the selection of the model
predictor and use of a second predictor in a multivariate, 2-predictor model would likely improve
the correlation (e.g., adding 2-methylnaphthalene improves R2 from 0.74 to 0.93) and the ability
of REMPI-TOFMS to determine PCDD/F TEQ.
w
M)
a 6
5-
4H
I
N
PQ
3-
o
Low DiClBz Rate
High DiClBz Rate
]» Sampled PCDD/F: M23 (ng TEQ/m3)
Figure 8-4. ETV phase comparison of TEQ from Method 23 and MClBz prediction from
REMPI-TOFMS measurements
66
-------
EPA/600/R-10/050
April 2010
,
'
1
A
9
4
<
J
1
k
>
:.
1
A % (REMPI Predicted Results)
% (M-23 Results)
% (M-10 Predicted Results)
' 1 1
r^
A
:
1
A
1
' <
1
(
k
1
A
\
\
>
L
O
LU
I-
y.
Q
Q
O
Q.
34567
ETV2005 Run Number
10
Figure 8-5. Comparison of M-10 and time-integrated REMPI predictions of PCDD/F
TEQ with actual M-23 measurements.
67
-------
EPA/600/R-10/050
April 2010
9. Source 4: Sampling from MWC Flue Gas
Two field measurement campaigns have been performed at the same MWC site using REMPI-
TOFMS instruments to monitor pre-determined target compounds in real time in the flue gas
prior to the air pollution cleaning system. These campaigns were conducted under challenging
conditions due to the presence of dust and large ambient temperature changes.
9.1 Portsmouth Naval Shipyard Waste Combustor 2004.
The primary objective of the first MWC study was the field demonstration of the REMPI-
TOFMS technology to measure target organic compounds identified as potential indicators for
PCDD/F TEQ determinations. Prior to the field study, the REMPI-TOFMS was found to
measure easily, reliably, and potentially online or through a short pre-concentration step,
PCDD/F TEQ indicators include compounds such as low chlorinated PCDD/F compounds,
chlorobenzenes, chlorophenols, and PAHs. The main objective of this first test at this MWC was
to test and validate the REMPI-TOFMS instrument in the field using the results from an initial
MWC sampling campaign that provided correlations for potential indicators.
This initial campaign of this project focused primarily on extractive flue gas samplings prior to
the air pollution control system (APCS), which consists of a spray dryer and baghouse, and to a
lesser extent at the stack. The primary sampling locations were the spray dryer catwalk area
(SDA) at the exit of boilers 1 and 2. These locations were chosen to provide concentrations of
the target compounds of interest that were sampled in a relatively reasonable time, and analyzed
by the EPA-RTP Organic Laboratory using HRGC/ LRMS. A potentially multivariate, nonlinear,
time-variant model was developed based on the initial sampling campaign (in 2003). The model
relates the gas-phase concentration of one or more indicator compounds from sub-isokinetic
samples from SDA2 with the PCDD/F TEQ and PCDD/F total (from SDA2-iso). This model
was used to develop a test matrix that would optimize field-sampling data collection in the first
field demonstration of the REMPI-TOFMS instrument. Validation of the system was performed
using extractive methods for dioxins and furans. The main classes of compounds that were
targeted in this project were primarily mono- to octa-chlorinated dioxins and furans (PCDD/Fs),
chlorinated phenols (CIPhs), chlorinated benzenes (ClBzs), and PAHs.
9.2 Experimental Approach
The main objective of this test program was to develop potential organic indicator compounds
that can mimic PCDD/F TEQ emissions, and test the validation of REMPI-TOFMS to produce
sensitive, time resolved measurements of these target indicator compounds.
To further examine REMPI-applicable indicator compounds, samples were taken from a MWC
with multiple RDF fired boilers (>400 Mg/day) over a 9 day period for PCDD/F and other
combustion by-products. Samples were drawn from the flue gas duct prior to the flue gas
cleaning system; hence, concentrations do not represent stack emissions. Over the course of nine
68
-------
EPA/600/R-10/050
April 2010
days, 23 sampling periods were completed. Sampling for PCDD/F included 1 h Method 0023
(M23) samples and three 5-min M23 samples taken throughout a 1 h sampling period. Sampling
for semi-volatiles (e.g., monochlorobenzene, phenol) was accomplished with REMPI-TOFMS
and by standard extractive sampling via Method 0010 (Ml0). These runs paralleled the M23
sampling periods, and should be comparable. M10 samples were analyzed for semi-volatiles; in
this case only the back half values are reported (in ug/m3, dry). These back half semi-volatile
values should be directly comparable to the REMPI-TOFMS determinations, as they are both
post-filter (akin to what the XAD "sees") measurements. It should be noted that M10 samples
were 1 h in duration whereas REMPI was on-line; hence many of the REMPI-detectable semi-
volatiles reported by M10 were not above the on-line concentration detection limits of REMPI-
TOFMS.
9.3 Test Matrix
The test matrix presented here refers to the first campaign of testing with the REMPI-TOFMS
system. The proposed time sequence of the initial data was based on a statistical analysis of the
initial MWC data. The time sequence of those sampling events recorded on the same axis as the
CO stream data showed some evidences that the largest PCDD/F TEQ values observed followed
a period of identifiable CO instability reflected by an upsurge of CO spikes above a 600 ppm
threshold level. The sampling strategy for this testing campaign was to capture and identify
delayed effects of the critical CO period on the PCDD/F formation with a minimal loss of
instantaneous PCDD/F TEQ and other surrogates compounds in the process of averaging over
any particular sampling time. Six identifiable phases of boiler operation were proposed in this
sampling campaign to be characterized experimentally:
Phase 1. Normal operation: This testing phase will start after normal operations that would have
lasted at least 6 hours after a startup without any sign of probable shutdown
Phase 2. High frequency CO spiking without boiler shutdown: This sampling phase will
correspond to an identifiable period of CO instability due probably to a plug in the feed
system. This transitional period last in general less than 2 hours.
Phase 3. Boiler leading to shut-down: This phase is expected to occur within 2 hours of phase 2
conditions. The shutdowns can be unscheduled, and due to high CO hourly averages or
initiated by the boiler.
Phase 4. Actual boiler shutdown phase
Phase 5. Boiler early startup: This sampling phase will start just after the boiler startup is
initiated after a scheduled or unscheduled shutdown
Phase 6. Boiler late startup: This phase will be performed about 3 hours after the startup, after
the early startup phase is completed.
69
-------
EPA/600/R-10/050
April 2010
9.4 Results
9.4.1 General On-site REMPI-TOFMS Instrument Performance
The first field study of the first compact REMPI-TOFMS instrument (December 2004) was
satisfactory based on the relative fast setup time (~ 6 hours) between arrival onsite and the first
detection of flue gas constituents and the near 90% coverage of real time (2s sampling period)
REMPI-TOFMS data during parallel conventional Method 0010 and Method 0023 sampling. In
one occasion, the sampling periods were delayed in order to have REMPI-TOFMS available for
parallel sampling after a system shutdown. This occurred after ultrafme fly-ash passed through
the inline filter (or was pulled into the system during a swap of filters) and got trapped in the
pulsed valve inlet system hereby obstructing the pulsed sample flow. The exchange of filter
elements using a single housing unit resulted in a temporary loss in PAH signal intensity most
likely due to the initial low temperature of the new filter element. Preheating a clean new
element did resolve this issue and was implemented later in the sampling campaign.
9.4.2 REMPI-TOFMS results
The focus of REMPI-TOFMS during this sampling campaign was on the real time detection of
PAHs and monochlorobenzene, if possible in real time. Over the course of a seven day period,
the flue gas of the MWC was sampled with REMPI-TOFMS for a total of 58 hours. In parallel,
23 EPA Method 0010 (semi-volatiles) and 0023 (PCDD/Fs) samples were taken. After day 2, all
sampling equipment was relocated to a different boiler ("B") area due to unscheduled extended
maintenance on the first boiler ("A"). This provided an opportunity to compare boiler
performances under normal, "steady state" conditions. As illustrated in Figure 9-1, there are
large differences in emissions of especially naphthalene over the course of a 4 hour time period.
PAH emissions for boiler "B" were generally much higher during the shutdown of a boiler as
shown in Figure 9-2. The observed transient naphthalene spikes are most likely due to residual
smoldering waste with flare-ups. It should be pointed out that PAH concentrations are elevated
under conditions of lower flow velocity in the duct from which the sample is taken which tend to
enhance the concentration but not the actual emissions at that time.
More dramatic changes in PAH concentrations occur during the period leading up to a shutdown
of a boiler (B) as well as during the startup. An example of emissions during the startup of the
boiler is shown in Figure 9-3.
A similar transient response was observed during another startup of the same boiler. In that case,
the REMPI-TOFMS wavelength was toggled between detection of PAHs and
monochlorobenzene providing partial real time data for both as shown in Figure 9-4.
70
-------
300 n
30-,
25-
20-
15-
10-
5-
0
EPA/600/R-10/050
April 2010
-naphthalene
fluorene
- phenanthrene/anthracene
pyrene/fluoranthene
Boiler B
**
o-
15:00 16:00 17:00 18:00 19:00 11:00 12:00 13:00 14:00 15:00
Plant Time (h:min), Day 2 Plant time (h:min), Day 5
Figure 9-1. Comparison between real time results for PAHs acquired with REMPI-
TOFMS at identical Boiler A and Boiler B for two 4-h intervals that were
typical of the normal (steady state) operation.
25-,
BOILER OFF LINE
naphthalene
fluorene
phenanthrene/anthracene
pyrene/fluoranthrene
0-
19:45 20:00 20:15 20:30 20:45 21:00
Plant Time (h:min)
Figure 9-2. Emissions of PAHs during shutdown of boiler. Break in data during filter
change.
71
-------
EPA/600/R-10/050
April 2010
80
Figure 9-3. A 3-dimensional qualitative depiction PAH emissions during the startup of a
boiler (at ~ t=10 min).
naphthalene
fluorene
phenant
pyrene
_Q
CL
Q.
O
U»
ro
-------
EPA/600/R-10/050
April 2010
shows such comparison in the case for naphthalene which was readily detected with the REMPI-
TOFMS instrument.
What is also clear from all results is that flue gas contains multiple PAH-like compounds as
visible in a typical mass spectrum recorded at the wavelength for monochlorobenzene detection
(Figure 9-6).
Since the single color REMPI-TOFMS ionization method is more efficient for PAHs than for
chlorinated aromatics, it is extremely difficult, if not impossible, to efficiently detect higher
chlorinated aromatics such as trichlorobenzene in the presence of more abundant analytes with
equal mass.
9.4.3 Method 0023 and Method 0010 Results
An analysis of M23 PCDD/F TEQ values with simultaneously-sampled M10 results in
significant correlations for multiple semi-volatile compounds (Table 9-1). The best 5-min
sampling duration predictors for this facility, based on Table 9-1 results in combination with
typically greater spectral resolution and sensitivity for trichlorobenzenes versus
tetrachlorobenzenes, appear to be 1,2,4-trichlorobenzene, 1,2,3,4-tetrachlorobenzene, and 1,3,5-
trichlorobenzene. The detection limit for 1,2,4 trichlorobenzene with 2-color REMPI (established
post first MWI test) was about 1 ppb, a value commonly found in incinerators. An additional
analysis examined these correlations under the assumption of a 20 and 40 minute time lag
between the concentration of the indicator and that of the PCDD/F TEQ. As can be seen in Table
9-1, no convincing improvement in correlation is found with the time-lagged indicators,
suggesting that the increase in the indicator compound is due to a parallel or related mechanism
to the formation of the PCDD/F TEQ compounds. These results suggest that indicators are likely
system-specific, requiring final selection by testing at each plant, albeit from a list of fairly likely
and common candidates.
9.5 Portsmouth Naval Shipyard Waste Combustor 2006
The introduction of a two color REMPI approach towards ionization of an analyte increased the
accessibility of higher chlorinated aromatics for real time measurement. This enhancement,
among others, resulted in a second sampling campaign at the same MWC (December 2006) with
the objective of sampling 1,2,4-trichlorobenzene (124-TrClBz), preferably in real time, as the
dioxin toxicity indicator compound. REMPI-TOFMS sampled 124-TrClBz using a two color
ionization method.
Work reported here shows the first REMPI-TOFMS measurements for a specific PCDD/F
indicator compound (1,2,4-trichlorobenzene) along with simultaneous conventional sampling for
PCDD/F TEQ values.
73
-------
EPA/600/R-10/050
April 2010
120-
100-
.0
Q.
Q.
C
o
'-t-t
2
"c
CD
O
c
o
O
80-
60-
40-
20-
1 hour extractive Method 0010
1 hour averaged jet REMPI-TOFMS
e
$
«
4-1
.2,
Q.
i
S
I
t
S
«
B
O
o
>.
c
ra
S
,
c
Q.
i
$
-------
EPA/600/R-10/050
April 2010
Table 9-1. R for PCDD/F TEQ based on 36, 5-min samples at the MWL
12DiCIBz
13DiCIBz
14DiCIBz
123TrCIBz
124TrCIBz
135TrCIBz
1234TeCIBz
1235TeCIBz
1245TeCIBz
PeCIBz
HxCIBz
TEQ
Lag=0 n=36
0.482
0.551
0.074
0.558
0.709
0.703
0.727
0.642
0.702
0.619
0.391
TEQ
Lag=20
n=24
0.516
0.499
0.355
0.520
0.414
0.444
0.518
0.529
0.398
0.604
0.557
TEQ
Lag=40
n=12
0.594
0.215
0.010
0.642
0.195
0.116
0.559
0.655
0.132
0.586
0.599
The 0.95 critical values for testing the null hypothesis of "no significant correlation" for n= 36, 24 and 12 are 0.108, 0.164,
and 0.332, respectively. Lag = time duration in min between event and concentration correlation.
9.6 Materials and Methods
REMPI-TOFMS was employed at an MWC with multiple refuse-derived-fuel (RDF) fired
boilers to sample flue gas concentrations after the boiler chamber and prior to the air cleaning
devices. The boiler was operated under steady state conditions, shutdowns, and startups. Boiler
shutdowns would normally occur with operating problems such as jams in the fuel feeding
system or for routine maintenance but were initiated for these tests to observe the pollutant
response during transient operating conditions. Flue gas sampling for REMPI-TOFMS was
accomplished with a glass-lined, stainless steel sampling probe coupled to a heated (T = 170 °C)
line. A heated filter (T = 150 °C) prior to the REMPI inlet prevented PM from clogging the
pulsed inlet valve. The filter was changed daily to minimize adsorption or desorption phase bias
of target analytes. Sampling was sub-isokinetic to minimize particle collection. REMPI was
operated in a 2-color, 2-photon mode. The wavelength of the first laser, 284 nm, was set in
resonance with the origin of the Si< So transition for 1,2,4-trichlorobenzene (TrClBz), a
suspected PCDD/F indicator compound. A fixed 213 nm wavelength completed the ionization
process. Continuous CO, O2, and CO2 measurements were taken from the plant's post-boiler,
pre-stack monitors. The possible presence of concentration gradients within MWC ducts were
evaluated for their potential detrimental effects on establishing concentration correlations.
Simultaneous measurements were taken at 30% and 50% of the cross-duct width to check for
stratification of PCDD/F concentrations. In parallel to the REMPI-TOFMS measurements, 74
PCDD/F samples (modified EPA Method 23) were taken over a four day period. Typically, three
5-min duration samples were collected over 1-h periods during steady-state, shutdown, and
startup conditions. There was approximately 15 min between samples. The samples were
analyzed by HRGC/LRMS for all three TrClBz (1,2,3-; 1,2,4-; and 1,3,5-TrClBz) compounds
75
-------
EPA/600/R-10/050
April 2010
and by HRGC/HRMS for the 17 toxic equivalency factor (TEF) weighted compounds that
comprise the PCDD/F TEQ measure. A field blank and an XAD sorbent blank showed
minimal levels of 1,2,4-trichlorobenzene compared to the lowest measured sample level. The
choice of units from which to compare compound concentrations depends on the purpose of the
comparison. The various operational modes at the facility result in different RDF feed-rates,
boiler temperatures, and flue gas flowrates. For example, comparison in units of mass per waste
feed of steady state emissions versus those during facility shutdowns could make the latter
exceptionally high if pollutant production is more related to mass of waste on the grate or the
amount of deposits on the boiler. In this paper, concentrations are compared on a time basis to
understand how different process options at the plant affect the rate of PCDD/F production.
9.7 Results
Pre-APCS PCDD/F TEQ concentrations are reported in Table 9-2. Transient operating
conditions resulted in higher average PCDD/F TEQ values and RSDs by almost 4-fold. Peak
concentrations of PCDD/F TEQ exhibit at least a 10-fold increase during startups over those of
the average steady state values. Integrating the mass concentration rate curve suggests that the
TEQ levels during a combined shutdown and startup period are ~ 25 percent of the plant's
projected daily PCDD/F TEQ concentrations. The effect on plant stack emissions, however,
remains to be determined.
Table 9-2. PCDD/F TEQ emission rates from HRGC/HRMS analyses.
Operating Condition
Steady State (SSTS#)
Shutdown (SSHD#)
Startup (SSUP#)
Four Day Averages (ug TEQ/min)
High
208
349
1262
Low
109
98
181
Average
154
203
529
RSD
23%
64%
86%
Simultaneous M23 samples taken across the duct showed less than 10% RSD between the
values, confirming that measurements of PCDD/F indicator compounds are amenable to
correlation analyses. GC-measured di- and tri-chlorinated benzene isomers showed excellent
correlation with PCDD/F TEQ. Two of the three diClBz isomers showed an average correlation
(R2) with PCDD/F TEQ of 0.85. All three of the triClBz isomers, including target 1,2,4-triClBz,
had an average R2 of 0.90 with PCDD/F TEQ. These results suggest that any one or combination
of these target ClBz isomers would make excellent indicators. Figure 9-7 summarizes the 5-
minute average concentrations of PCDD/F TEQ, 1,2,4-triClBz using GC/MS and 1,2,4-triClBz
values using REMPI-TOFMS.
76
-------
EPA/600/R-10/050
April 2010
I
I
o
o
LU
Q
Q
-PCDD/FTEQ
-1,2,4-TriCIBz
REMPI Mass 180
-o.
Day1 -
r * A . i T
L^JAfc:-
i-*-«-*>«J_*=t=*-*=
-------
EPA/600/R-10/050
April 2010
times. Since all combustion shutdown/startup operations do not necessarily occur under the same
conditions, and our sampling intervals would not necessarily have been at the same exact onset
of the process changes, consideration of a single sampling period lag seems reasonable. These
results demonstrate the ability of REMPI-TOFMS to provide real time feedback on correlative
PCDD/F levels and offer promise for operational feedback to minimize emissions.
120-
'g 100-
JD
S 80-
ro
o)
OT 60-
o
3= 40-
20-
0-
-i on IVIZO S
mass 182
mass 184
M23 sampling times
Shutdown
M23 sampling times
Steady State
I
_ J
ampling times
Startup
\
v_
17:00 18:00 19:00 20:00 21:00 22:00 23:00
Time (h:min)
Figure 9-8. Recorded REMPI-TOFMS ion signals for mass 180/182/184.
The vertical bars indicate the shutdown (black) and startup (blue) events.
78
-------
EPA/600/R-10/050
April 2010
10. Source 5: Emission Responses from HMMWVs, the Ml Abrams Tank,
and the Bradley Infantry Fighting Vehicle
Real time emission measurements of criteria pollutants, organic air toxics, and particles were
made on two U.S. Army High Mobility Multi-purpose Wheeled Vehicles (HMMWVs), a
Bradley Infantry Fighting Vehicle (IFV), and an Abrams Ml battle tank. The HMMWV
emissions were measured while running on a dynamometer-based roadway simulator under the
West Virginia and EPA Federal Highway Economy Test (HWFET) cycles, and under steady
state, constant velocity conditions. REMPI-TOFMS emissions from cold and warm starts of the
HMMWVs, the Bradley, and Abrams were recorded.
The U.S. Army currently has over 99,000 tactical High Mobility Multi-purpose Wheeled
Vehicles (HMMWVs) and over 12,000 tracked vehicles such as the Ml Abrams tank and
Bradley IFV (https://www.osmi sweb.army.mil/, 2008) that account for approximately half of the
DoD vehicle fleet (http://www.globalsecurity.org/military/systems/ground/hmmwv-recap.htm).
Some emissions data are available for the HMMWVs, primarily criteria pollutants, but none are
known for the tracked vehicles.
10.1 Experimental
10.1.1 Platforms Tested
Criteria and organic pollutant emissions were characterized during transient operation of four
U.S. Army weapon platforms at the U.S. Army Aberdeen Test Center, Roadway Simulator
(RWS). The RWS is the world's largest dynamometer, supporting vehicles ranging from 2,300
kg (5,000 Ib), 2-axle light vehicles (up to 192 km/h or 120 mph) to 27,300 kg (60,000 Ib),
tandem-axle tractor trailers (Schultz et al., 2005). Two versions of HMMWVs were tested for
emissions on the RWS. These included a heavy HMMWV (M1097 Al) and a turbine engine, up-
armored HMMWV (Ml 114). The HMMWV is currently the main transport vehicle for the DoD.
These 6.2 L, General Motors vehicles consume almost 15 million gallons of JP-8 each year,
making it the top-consuming vehicle type (17%) in the DoD inventory (Kemme et al., 2006). The
6.5 liter Turbo Diesel 142 kW (190 hp) up-armored Ml 114 version has a higher payload
capacity (1,043 kg) for a gross vehicle weight of 5,489 kg (GlobalSecurity.org, 2008a) and is
equipped with a heavy armor protection package instituted during the Iraq war. Both of these
vehicles were tested with two driving cycle protocols (described later) and stepped
velocity/gradient cycle.
Two additional weapons platforms, an Ml Abrams tank and a Bradley IFV, were tested at the
facility but not on the RWS due to weight and track restrictions. The 62 metric ton Ml Abrams
main battle tank is manufactured by General Dynamics Land Systems and powered by a
Honeywell AGT 1118 kW (1500 hp) gas turbine. The Allison hydrokenetic transmission has four
forward and two reverse gears. The Bradley is powered by a 447 kW (600 hp) Cummins VTA-
903T water-cooled 4 cycle diesel engine. It is a fully armored, fully tracked vehicle designed to
79
-------
EPA/600/R-10/050
April 2010
carry mechanized infantry into close contact with the enemy (GlobalSecurity.org, 2008b). Both
vehicles were fueled with JP-8, a standard military fuel (U.S. Army, 2001).
10.1.2 Test Protocols
Two standardized driving cycles and one combined velocity and gradient cycle were used to test
both HMMWVs. The West Virginia highway driving cycle (WV) was developed originally from
local delivery tractor-trailer activity logged by West Virginia University (Nine et al., 2000) and
described elsewhere (Clark et al., 1999). The WV cycle represents the travel on four-lane
highways of trucks to and from distribution depots located outside the city delivery areas [see
bottom panel, Figure 10-1 (A)]. The EPA Highway Fuel Economy Test (HWFET) driving
schedule [U.S. Department of Energy (DOE)/EPA, 2008] represents a mixture of rural and
interstate highway driving with a warmed-up engine, typical of longer trips in free-flowing
traffic [see bottom panel, Figure 10-1 (B)]. The M1097 HMMWV was run once on both the WV
and HWFET cycles while the Ml 114 had triplicate runs on both cycles. Both HMMWV types
were run once on the stepped velocity and gradient cycle. Transient emissions of the Ml Abrams
and the Bradley IFV were sampled directly from the exhaust during startup, stationary idle,
increased idle, and shutdown without actual movement of the vehicles.
Black: M1097; Red: M1114 HMMWV
Black: M1097; Red: M1114 HMMWV
10 15 20
Time (min)
1800
1200
600
0
120
Benzen
o
O
i Naphthalene
|>^V^^
4x103
2x105
r^Jw^J^
Time (min)
Figure 10-1. Emission transients for the M1097 and Ml 114 HMMWVs. A: West Virginia
and B: Highway HWFET cycles.
10.1.3 Sampling Approach
A 15.2 cm (6 in.) ID. stainless steel sampling pipe was attached to the HMMWV tailpipes via a
high temperature rubber adaptor to convey the emissions to the analyzers. The exhaust
80
-------
EPA/600/R-10/050
April 2010
temperature was measured at or below about 273 °C and the gas pipe was heat traced to about
200 °C. No dilution air was added, and the flow was monitored with an annubar pitot tube.
Exhaust for the REMPI-TOFMS system, the particle measurement systems, and the continuous
emissions monitors (CEMs) was sampled using separate heated lines maintained at about 125-
150 °C.
The Bradley and Abrams idle, cold start, and warm start emissions were sampled at a rate of 1
L/min by means of a 0.6 cm (H-inch) diameter stainless probe inserted inside their exhaust line.
In all studies, a slip stream of the sample exhaust was sampled through the pulsed valve of the
REMPI-TOFMS system at a rate of 1 mL/min.
The exhaust gas criteria pollutants were analyzed using a self-contained CEM bench that
included two analyzers using EPA promulgated Method 10A (EPA, 2000b) for CO [low CO:
RosemountNGA 2000, high CO: California Analytical Instrument and two analyzers using EPA
Method 3 A (EPA, 2006), one for oxygen (O2: Rosemount NGA 2000) and one for CO2
(California Analytical Instrument)]. Each analyzer had a daily, 3-point calibration and a 3-point
bias check before the start of the test, followed by a 3-point system calibration and bias check at
the end of the test. Gas standards were introduced via the bias check port at the sample line inlets
during pre-sampling CEM performance checks. The CEM is linked to a data acquisition system
set for a sampling frequency of 10 s"1.
The REMPI-TOFMS instrument measured aromatic organic air pollutants in real-time, parallel
to the standard CEMs. BTEX, styrene, and gaseous PAHs, are among the non-exclusive list that
were targeted for measurement. An approximately 1 L/min slipstream was pulled from the main
sampling pipe through a 15 m long, 0.95 cm (3/8 in.) diameter Silico-Steel coated line at 150 °C
towards the REMPI-TOFMS inlet. A glass microfiber filter (Unique Heated products) mounted
directly downstream from the main sampling pipe prevents PM from reaching the REMPI-
TOFMS instrument. Two filter housings were operated in parallel in order to have the capability
to replace a filter element on an hourly basis without interruption of the sampling. The pulsed
inlet valve (modified General Valve Series 99) of the REMPI-TOFMS instrument pulls
approximately 1 mL/min into the ionization chamber of the TOFMS.
Real time particle size distributions were measured with a Dakati ELPI. The ELPI is a 12-stage
impactor with a single pre-stage knockout, measuring particle size distributions ranging from 30
nm to 10 |im at a frequency of 1 Hz. The ELPI generates a particle size distribution by charging
the particles based on geometrical diameter prior to them entering into a cascade impactor. The
charged particles land on the impactor stages based upon their inertia and their charge is
converted into a particle number and mass, given a known density. The sample was conveyed via
a non-static, conductive tubing, diluted with two 10:1 VKL (Palas) venturi-based dilutors in
series (for a 100:1 dilution), totaling 10 L/min at ambient temperature and pressure. Dilution
ratios are verified by using measurements of pre- and post-diluted CO and CO2. Daily quality
control checks on the ELPI were performed by zeroing the electrometer using a flush of high
efficiency particulate arresting (HEPA) purified air and by monitoring the current profiles of the
ELPI while it samples room air and HEPA purified room air.
81
-------
EPA/600/R-10/050
April 2010
Longitudinal and lateral vehicle velocities and forces as well as engine rotation and vehicle speed
data were provided by the RWS control system.
Cold start emissions were evaluated in this study using a regression model (Heeb, 2001; as in
Heeb and Weilenmann et al., 2005) which consists of a linear extrapolation to the ordinate of the
idle state portion of the cumulative target compound (CO, CC>2, benzene, naphthalene,
methylnaphthalene, and others) evolution. The ordinate value is the target compound mass
amount emitted from a single cold start event.
The repeatability of the RWS facility operations and pollutant measurements were evaluated
using the t-test statistical approach. RA is used to compare differences in the means of the target
measurements between two cycles. All cycles within one type of driving profile (HWFET or
WV) are compared with respect to the first cycle. The RA (%) is defined as follows:
RA = -
RM (2)
where:
d
= the mean of the absolute values of the differences between measurements taken at the
same time during a complete cycle
t0 975 = the lvalue,
Sd = the standard deviation of the differences between measurements taken at the same time
during a complete cycle
RM = the mean of the reference cycle
n = number of measurements taken with a rolling average of 10 s during a complete cycle
10.2 Results and Discussion
REMPI -TOFMS and conventional CEMs were used to compare emissions from the diesel-
powered heavy Ml097 HMMWV with the armored-upfitted, turbine-driven Ml 114 HMMWV.
Concentration values of benzene, naphthalene, methylnaphthalene, CC>2, and CO as well as the
total particle density (N), applied power (HP) and velocity (v) are shown for both vehicles on the
WV and HWFET protocols, in Figure 10-1 (A) and Figure 10-1 (B), respectively.
The RWS facility operations had excellent velocity repeatability between cycles for each driving
cycle tested. The relative velocity accuracy for the HMMWV Ml 114 vehicle was 0.04% and
0.44% HWFET cycles, and 0.2% and 0.5% for the WV cycles. The RA for both HWFET and
WV triplicate cycles, for the criteria compounds measured using conventional CEMs varied from
2.4% to 8.1% for CO2, and 7.4% to 31% for CO. The RA for the trace compounds concentrations
measured by the REMPI-TOFMS system varied from 15% to 79% for benzene, 31% to 109% for
naphthalene, and 47 to 214% for methylnaphthalene. The lower reproducibility of the emissions
for naphthalene and methylnaphthalene are due to the compounding effect of their relatively low
82
-------
EPA/600/R-10/050
April 2010
concentration (sometimes just above the detection limit of the REMPI-TOFMS system) and the
variance in the engine performance. It should be noted that the first two of the triplicate tests
were run successively while the third test was performed later (4-5 h) on the same day with the
lower reproducibility always between the non-successive runs.
10.2.1 WVCycle
REMPI PAH measurements for the armored (Ml 114) HMMWV (red traces) on the WV cycle
showed benzene peaks of about 120 ppb with a median value of 20 ppb [Figure 10-1 (A)].
Naphthalene and methylnaphthalene peaks of 6 and 12 ppb, respectively, were accompanied by
median values of < 1 ppb. The average emission factors for benzene, naphthalene, and
methylnaphthalene for the WV cycle were 325, 14, and 29 jig/km (see Table 10-1).
83
-------
EPA/600/R-10/050
April 2010
Table 10-1. Emission factors of M1097 and Ml 114 HMMWVs.
Type of Engine
Vehicle Speed
CO2
CO
benzene
naphthalene
methylnaphthalene
styrene
phenol
1,2,4trimethylbz.
p-xylenes
m-xylenes
1,3,5trimethylbz
o-xylene
toluene
ethylbenzene
M1097 HMMWV (CO/CO2 g/km), Others Air Toxics (ng/km)
32
km/h
231
0.38
246
64
128
54
60
64
27
71
61
38
97
24
48
km/h
206
0.75
1163
244
386
160
116
145
66
195
145
127
302
70
64
km/h
268
1.07
3228
673
1573
785
421
627
196
791
540
501
1132
329
80
km/h
356
0.84
3248
727
1658
718
428
526
218
265
641
476
1641
526
80 km/h -
1.0%
grade
398
1.64
5271
1133
1726
2165
1536
587
412
739
749
354
1376
70
80 km/h -
1.5%
grade
418
1.74
7732
1197
1463
1055
978
400
289
466
428
337
1735
221
80 km/h -
2.5%
grade
419
1.73
NA
592
577
231
158
482
367
HWFET
361
1.23
5096
1001
2072
WV
380
0.74
3482
564
1184
NA
NA
Type of Engine
Vehicle Speed
CO2
CO
benzene
naphthalene
methylnaphthalene
styrene
phenol
1,2,4trimethylbz.
p-xylenes
m-xylenes
1,3,5trimethylbz.
o-xylene
toluene
ethylbenzene
M1114 HMMWV(CO/CO2g/km), Others Air Toxics (ng/km)
32
km/h
383
0.61
177.3
23.2
21.8
3.5
7.7
3.9
5.4
19.4
8.8
10.6
37.8
15.5
48
km/h
354
0.37
110.5
9.1
4.2
2.6
6.4
2.6
3.3
15.7
5.2
7.2
30.3
2.9
64
km/h
350
0.28
116.1
5.4
6.8
0.9
6.8
1.8
1.7
6.3
4.0
5.7
20.9
0.6
80
km/h
421
0.21
28.8
3.0
6.5
1.4
9.8
2.7
1.1
2.6
2.8
1.6
3.2
0.6
80 km/h -
1.0%
grade
522
0.20
129.1
2.0
3.7
80 km/h -
1.5%
grade
581
0.19
161.3
2.0
2.7
80 km/h -
2.5%
grade
647
0.18
225.5
2.0
2.3
HWFET
493
0.42
275.1
12.7
19.8
WV
435
0.55
324.5
13.6
28.8
NA
84
-------
EPA/600/R-10/050
April 2010
The presence of naphthalene is expected; its presence in diesel fuel emissions is due to
incomplete fuel combustion (24%) as well as pyrosynthetic processes (76%) from compounds
such as methylnaphthalene (Rhead and Pemberton, 1996) or small unsaturated hydrocarbons
(Badger and Novotny, 1963). Approximately 0.5% of the naphthalene in diesel fuel has been
observed to have survived combustion (Rhead and Pemberton, 1996). Clearly, the emission
factors for most of the BTEX compounds and PAHs found in the exhaust system are dependent
on multiple factors such as their initial concentration in the parent fuel, their survivability in the
combustion chamber, combustion efficiency of the engine, potential reformation in the cooling
zones, and load. CO and CO2 were found to correlate poorly with benzene, naphthalene, and
methylnaphthalene emissions for the Ml 114 (R2 always less than 0.30), eliminating their use as a
correlative indicator of air toxics levels.
The M1097 had much higher emissions than the Ml 114, reflecting the turbine adoption in the
latter (see Table 10-1). The diesel M1097 exhibited significant quenching of the fuel combustion
on the cold cylinder walls, resulting in considerable emissions. For example, the benzene peak
value for the M1097 was about 1600 ppb versus about 120 ppb for the Ml 114. CO emissions
similarly were higher with the M1097. The two vehicles also had distinct, pollutant-specific
responses to the WV cycle. For example, the 15-25 min trace of the Ml 097 showed relatively
high, sustained naphthalene and methylnaphthalene emissions that were not evident during the
same period with the Ml 114. The average emission factors for benzene, naphthalene, and
methylnaphthalene for the WV cycle reflect a 1-2 orders of magnitude higher emissions than the
emissions from the Ml 114 engine. Correlations between REMPI-TOFMS detected analytes and
CO, CO2 were found to be only marginally better for the M1097 (R2 up to 0.59) than the Ml 114.
10.2.2 HWFET Cycle
As with the WV cycle, tests of the M1097 and Ml 114 on the FIWFET cycle showed the M1097
to have significantly higher overall air toxic emissions. The emission factors, presented in Table
10-1, for both vehicles show dependence with the driving cycle; however, the results are within
the accuracy of the cycle reproducibility. Again, CO and CO2 were found to correlate poorly
with benzene, naphthalene, and methylnaphthalene emissions for the Ml 114 (R2 always less than
0.34). Correlations were slightly better for the M1097 (R2 up to 0.48).
10.2.3 Velocity/Gradient Cycle
Steady state emissions from a prescribed protocol of velocity and gradient changes show
distinctive trace emissions from the Ml 114 and M1097 (Figure 10-2).
With higher velocity, the M1097 produces higher CO and trace emissions, while gradient
increases (at 80 km/h) appear to have no effect. The Ml 114 shows minimal effect of velocity
and gradient increases. Trace organic emissions and CO tend to decrease with increased velocity
but this is believed to be due to increased load and engine warm up. As already observed during
the WV and HWFET cycles, observed concentrations with REMPI-TOFMS for trace analytes
were found to be much lower in the Ml 114 exhaust when compared to the M1097 results. In
fact, the highest emissions for the Ml 114 were temporary peaks after speed changes, especially
from idle to 32 km/h.
85
-------
EPA/600/R-10/050
April 2010
M1097
Benzene Napthalene
Methylnaphthalene
200
150-
M1114
0 10 20 30 40 50 60 70 140 150
Time (min)
170 180
Benzene Napthalene
Methylnaphthalene
30 40 50
Time (min)
Figure 10-2. Steady-state benzene, naphthalene, methylnaphthalene, CO, and COi
emissions from the M1097 and Ml 114 HMMWVs during a velocity and
gradient stepped cycle.
10.2.4 Startups
The fast time response of REMPI-TOFMS allows for documentation of trace emissions during
startups. Figure 10-3 shows pollutant-specific responses for the M1097 and Ml 114.
The M1097 cold startup shows benzene at ~ 230 ppb which rapidly (within 60 s) tails off to a
steady state level around 50 ppb. This highlights the responsiveness of the REMPI system as
well as the characteristics of the cold start emissions. The other trace pollutants, naphthalene and
methylnaphthalene, show distinctive traces from that of the benzene, increasing over 60 to 120 s
only to decline slightly to a steady state value. A similar benzene spike is observed for the
Ml 114, although its period is less than 15s, suggesting that benzene peaks are indicative of fuel
benzene vaporization during cold engine fuel starts. Calculation of the cold start benzene
emissions for the M1097 and Ml 114 FDVIMWVs indicate, respectively, that a 2.5 min and 5.5
min period of steady state emissions are equivalent to those of a single cold start. Similar time
equivalents were found for naphthalene to be 0.9 min and 7.5 min for the M1097 and Ml 114
FDVIMWVs, respectively.
86
-------
EPA/600/R-10/050
April 2010
250-
Napthalene
Methylnaphthalene
300-
01234
Time (min)
Napthalene
Methylnaphthalene
1 2
Time (min)
Figure 10-3. Cold start emissions of benzene, naphthalene, methylnaphthalene, CO, and
CO2 and PM size distribution for the Ml 097 (left) and Ml 114 (right)
HMMWVs.
10.2.5 Bradley and Abrams
Figure 10-4 shows the 30 s average, steady state idle emissions from the Abrams and Bradley,
respectively, of twelve organic compounds analyzed by REMPI. REMPI was sequentially set to
each compound's specific ionization wavelength throughout a 15 min period. Two types of idle
measurements are reported for the Abrams (low and high idle) as well as the Bradley ("idle 0"
and "idle 2" settings). The Bradley's concentrations are about 100-1000-fold higher than those of
the Abrams. The Bradley's emissions are most abundant for the substituted naphthalenes and
benzenes while the Abrams shows its highest concentrations in the lighter phenol, toluene, and
benzene compounds. This may be due to a higher percentage of unburnt fuel in the Bradley
exhaust versus the Abrams exhaust.
87
-------
EPA/600/R-10/050
April 2010
Figure 10-4. Steady state organic emissions at analyte-specific wavelengths, A: Ml Abrams,
low and high idle conditions, and B: Bradley IFV, idle 0 and 2 trials. Inset:
real-time variance of organics at low idle.
The Bradley and Abrams cold starts had quite dissimilar peak benzene concentrations at 600 and
35 ppb, respectively (Figure 10-5). These peaks did not persist more than about 30 s and, like the
M1097 and Ml 114, are derived from the unburnt fuel. The naphthalene and methylnaphthalene
traces don't exhibit such sharp startup peaks, perhaps reflecting an origin from combustion
byproducts rather than as unburnt fuel. The Bradley's methylnaphthalene concentrations were
high, at almost 500 ppb even 6 min past the cold start. The naphthalene and methylnaphthalene
emissions on the Abrams were very low, less than 2 ppb. The warm start on the Bradley showed
similar pollutant-specific trends to its respective cold start. The Abrams appeared to have higher
methylnaphthalene emissions on the warm start than during the cold start This observation is
biased since the response was observed in addition to residual methylnaphthalene (at much
higher concentration) in the sampling line after the shutdown preceding the warm start.
The Abrams tank exhibits an interesting and repeatable emission profile during shutdowns.
Figure 10-6 show that the Abrams undergoes a 3 min process to shut down in which the
concentration of benzene undergoes a double peak followed in time by methylnaphthalene and,
to a lesser extent, naphthalene.
88
-------
1400
1200
g-1000
o 800-
O 600
-------
EPA/600/R-10/050
April 2010
10.2.6 Number Size Distributions (NSDs)
Figure 10-7 reports steady state NSDs for the Ml097 and Ml 114 HMMWVs through the
stepped velocity and gradient cycle of Figure 10-2. The cold start and the lower velocities on the
M1097 tended to result in high particle numbers for all six median particle diameters between 39
and 484 nm as compared to the higher velocities. The NSD showed little shift in particle
diameter for all velocities, showing a maximum particle diameter at around 100 nm. Time-
resolved (every 15s) NSDs for the Ml097 showed no size shifts from 0 to 90 s, although the
maximum particle diameter decreased 50% (not shown). The Ml 114 showed somewhat lower
particle counts than the Ml097 but had about the same maximum particle diameter. Its response
to velocity and gradient changes were quite different than the Ml097, with higher particle
emissions with velocity increases and a distribution shift to higher particle diameters at higher
roadway gradients. This is seen more easily in Figure 10-8, which shows number concentrations
versus speed and gradient changes, as well as CO, CO2, and organics.
Both vehicles' PM size distributions are in the range of ultra-fine inhalable particles peaking at
around 100 nm diameter typical of diesel emissions found in other studies. Although particle
number density is not regulated in the U.S. and is proposed in Europe, its significance is very
important because of the pulmonary response to respiratory infections as well as adverse
cardiovascular events. The particle number densities per km found in both vehicles are in the
order of 1014. These are 100-1000 times higher than proposed European particle number
emission standard for diesels of 5 x 1011 per km (Good, 2007).
10.3 Emission Correlations
Correlations between organic compounds, vehicle parameters [engine revolutions per minute
(RPM), power (hp)], criteria pollutants (CO, CO2), 62, and NSDs by stage were analyzed across
the various operating cycles to provide insights into pollutant mechanisms, surrogates for
emission predictions, and operational modes to minimize pollutant generation. This method was
used by Schulz et al. (1999) to gain insights into combustion mechanisms of pollutant formation.
Highlights of these correlations are discussed below.
The Ml 097 concentrations of toluene, benzene, naphthalene, and methylnaphthalene were the
most closely correlated of the 12 organics with RPM, hp, CO, CO2, and O2. Across the range of
steady state test conditions, these four compounds had an average R2 of 0.82. The Ml 114,
however, was poorly correlated for these four compounds with R2 = 0.40. The o-, m-, p-xylenes
for the Ml 114, however, had an average correlation of R2 = 0.82. Benzene provides an
illustrative example; the average correlation for benzene with vehicle parameters and criteria
pollutants for the M1097 is 0.82. For the Ml 114, these R2 values are 0.04 and 0.23 for two days
of testing.
For the M1097, the 12 organics had an average intra-correlation of R2 = 0.60. This was highest
for naphthalene at R2 = 0.89 and lowest for ethylbenzene, with R2 = 0.22. The Ml 114's organic
intra-correlation was lower than the Ml097's at R2 = 0.57; the lowest was for benzene at R2 =
0.04 (ethylbenzene was 0.58) and the highest at R2= 0.85 for o- and m-xylenes.
90
-------
EPA/600/R-10/050
April 2010
2.0x10"
1.5x10"
c
ra
o.
Q
D)
o
1.0x10"
-0 5.0x10k
0.0
M1097|
j^
/
v?
I
1
1
1
1
\
V
--3
--4
'-6
-T-8
-4-8
«-8
8
_
2 km/h
8 km/h
4 km/h
0 km/h
0 km/h + 1 .0 %
0 km/h + 1 .5 %
0 km/h + 2.5 %
10
100 1000
Particle Diameter (nm)
1.5x10
1.0x10"
a.
a
D)
o
5.0x10
0.0
10000
4
/
//*
//
///
//
J
*/ ^
/ /i\
I
»>
»
\
\\
1
xj L
\ V*
V,^
P^
.A
T
*
4
>
32 km/h
48 km/h
64 km/h
80 km/h
80km/h+ 1.0%
80 km/h + 1 5%
80km/h+ 2.0%
10
100 1000
Particle Diameter (nm)
10000
Figure 10-7. Steady state PM size distribution for the M1097 HMMWV (top) and Ml 114
HMMWV (bottom) during a stepped velocity and gradient cycle.
For the M1097 and Ml 114, the NSDs by stage had an average intra-correlation R2 of 0.84 and
0.88, respectively, for stages 3 (0.1193 (j,m) and larger. This suggests that particle NSDs were
similarly altered (if at all) by changes in velocity, cycle type, and incline gradient. The fine
particle stages on the M1097, but not on the Ml 114, were highly correlated with the other stages.
From this limited dataset, it appears that good correlations between pollutants and operating
conditions do not always exist. Use of these correlations would significantly aid vehicle emission
characterizations by allowing one to predict levels of unmeasured pollutants from more easily
measured pollutants. However, these correlations must be established beforehand on a source-
and pollutant-specific basis.
91
-------
EPA/600/R-10/050
April 2010
- Styrene
-Phend
124 trmbz
- p-xylene
m-xylene
-135 trmbz
o-xylene
toluene
ethylbenzene
benzene
naphthalene
Methylnaphthalene
CO * IT)
velocity (MPH) and incline
2 §
g
velocity (MPH) and incline
Figure 10-8. Steady state REMPI, ELPI, and CEM data for the Ml 097 (left) and Ml 114
(right).
92
-------
EPA/600/R-10/050
April 2010
11. Source 6: F-15 and F-22 Aircraft Engine Emissions
This chapter describes the results of a field study using multiple extractive sampling techniques
as well as real time detection techniques to measure gaseous emissions from F-15 and F-22
fighter aircraft. Tests were performed between October 12 and 18, 2007 on the trim pad facility
at Tyndall AFB in Panama City, Florida. Measurements were made on two different F100
engines, and eight different Fl 19 engines mounted on in-use aircraft. 54 test runs were
performed at engine power levels that ranged from idle to full "military power", with additional
test runs at full augmented power (afterburner). The approach adopted for these tests involved
extractive sampling at a distance of approximately 20-25 nozzle diameters downstream of the
engine exit plane with real-time measurements using REMPI-TOFMS and a PTRMS (Battelle).
Integrated air samples were also collected and analyzed for VOC.
11.1 Experimental Method
11.1.1 Aircraft/Engine
The engines tested included the F100 engine associated with the F-15 Eagle and the F-16 Falcon
aircraft and the Fl 19 engine associated with the F-22 aircraft. The Pratt & Whitney F100 engine
is the primary engine in the worldwide F-15 fleet and has gained overwhelming use with more
than 85% of the world's air forces that fly the F-16. With more than 6,800 active installed
engines worldwide and over 16 million flight hours, the F100 series represents a large proportion
of the current fighter aircraft inventory. The F100 is an axial-flow turbofan produced in four
variations: F-100-PE-100, F-100-PW-200, F-100-PW-220, and F-100-PW-229. The engines
tested during this measurement campaign were F-100-PW-100 engines with normal thrust of
12,420 pounds, rising to a maximum thrust of 14,670 pounds at full military power. Maximum
afterburning thrust is 23,830 pounds.
The F-22 aircraft incorporates a pair of new, higher thrust engines, the Pratt & Whitney Fl 19-
PW-100, which is designed for efficient supersonic operation without afterburner. The Fl 19
engine develops more than twice the thrust of current engines under supersonic conditions, and
more thrust without afterburner than conventional engines with afterburner. Each F-22 is
powered by two of these 35,000-pound-thrust-class engines.
The experimental design called for duplicate test cycles to be performed on each test engine with
individual test runs conducted at each of the engine power settings shown in Table 11-1 (nominal
percentage of maximum thrust in parentheses). Due to the extreme conditions encountered in
sampling the exhaust at afterburner power levels, these tests were performed only on the final
day of testing.
93
-------
EPA/600/R-10/050
April 2010
Table 11-1. Engine power settings tested.
F-15
Ground idle
Low intermediate
High intermediate
Full/Military
(65-70%)
(80%)
(85%)
(91-93%)
F-22
Idle
Approach
Intermediate
Sub-military
Afterburner
(10%)
(20%)
(70%)
(80%)
(150%)
11.1.2 Testing Venue
The engines were tested at Tyndall AFB in Panama City, Florida. Tyndall is home to both the
Air Force Research Laboratory Airbase Technologies Division and the 325th Tactical Fighter
Wing. The 325th Operations Group of the 325th TFW is the focal point for all F-15 and F-22
initial pilot training. Operations Group maintenance personnel service and prepare aircraft for
flight. They also troubleshoot all mechanical problems that could prevent the aircraft from
sustaining the mission. As part of their maintenance role they operate and use two noise-
suppressing, "hush" houses and a trim pad for testing engines preflight. Their schedule typically
involves testing 25-30 engines or aircraft per month and their facilities including the trim pad
have recently been upgraded to handle the extreme requirements of the new, high thrust F-22
aircraft. The engines that were tested were mounted in their associated aircraft, and the aircraft
were tied down using existing anchors at the trim pad. The aircraft that were tested were
representative F-22 or F-15 planes rotated out of the training program at Tyndall AFB.
The trim pad (Figure 11-1) is an outdoor facility consisting of a circular concrete pad
approximately 35 meters in diameter. A concrete wall approximately 4 meters high encloses the
circumference of the trim pad with the exception of approximately 30 meters of the southeastern
face. During testing the aircraft were tied down using a tail hook at the center of the trim pad
with the nose of the aircraft facing the opening in the wall. An instrumented trailer housing the
extractive measurement equipment was positioned on the trim pad approximately 20 meters off
the right wing tip. A 16-foot van truck was situated near instrumented trailer, and housed the
REMPI-TOFMS and supporting equipment.
94
-------
EPA/600/R-10/050
April 2010
Figure 11-1. Trimpad and location of probe and instrument trailers with respect to
aircraft
11.1.3 Exhaust Sampling
Extractive sampling from the exhaust plume was conducted using a stainless steel probe located
downstream on the engine exit plane (see Figure 11-1). The extractive sampling probe was
positioned 23 meters behind the exhaust exit plane for tests at idle through military power, and at
38 meters for afterburner tests. The probe inlet height was 3 meters above ground level. The base
of the probe was secured to the trim pad with several concrete anchoring bolts to prevent
movement of the probe during engine operation. Heavy duty chain was used to stabilize the top
of the probe and was anchored to the trim pad in front of and approximately 45° on each side of
the probe axis.
The extracted sample was continuously transferred through heated, electrostatically protected
Teflon tubing to a heated sampling manifold inside the instrumented laboratory. Exhaust was
drawn from the manifold to a set of real-time monitoring instruments for continuous
measurement of CC>2, CO, nitric oxide (NO), nitrogen dioxide (NO2), total organic carbon, and
individual VOC. Integrated samples (adsorbent cartridges and canisters) also were collected for
laboratory analysis to determine carbonyl compounds and VOCs.
11.1.4 REMPI-TOFMS Sampling Approach
A 25' heated line (150 °C) with a 1/8" SS insert was connected to the main sampling manifold
inside the Battelle instrument trailer. An approximately 1.5 L per min flow was pulled through
this line. The flow rate was verified visually by monitoring the pressure drop upstream from the
REMPI-TOFMS pulsed valve. Marginal changes in flow rate do not affect recorded ion signals
but are important during pre and post test calibration using dynamic spiking with calibration gas
standards. Real time instrument performance was monitored using deuterated benzene that was
added to the main exhaust gas stream at ~ 5 ppb concentration. Ion signal for deuterated benzene
95
-------
EPA/600/R-10/050
April 2010
was verified at least once every hour for a 30s period. At the start and end of each day, a small
amount of TO-14 calibration gas was added to the pulled gas stream (no aircraft engine running
at such times) as a dynamic spike. This calibration gas was used for absolute calibration of the
TO-14 analytes (as far as included in the sampling procedure) and verification of predetermined
REMPI transition wavelengths. Calibration of the mass spectrum was performed and verified
daily using afore mentioned TO-14 calibration gas standard. The exhaust flow was also used to
measure CO and CO2 concentrations in the exhaust of the tested jet engines, independently from
Battelle's measurements. The sampling procedure as established by Battelle prior to the field
sampling campaign did not allow for detection of exhaust gases during the startup and shutdown
of any jet engine. In fact, a reverse flow purge was applied during such events to minimize the
possibility of unburned jet fuel being sampled (and trapped) inside the sampling line.
REMPI-TOFMS signals were recorded using the following approach: Based on the test plan,
each jet engine was tested twice for each thrust level. Therefore, the first time around, the laser
wavelength was tuned to the optimal wavelength for detection of benzene. This wavelength is
also suitable for detection of PAHs and is useful for real time detection of transients without loss
of data during wavelength switches. During the second sampling of the same thrust level, the
wavelength was changed over a pre-determined set of nine wavelengths, including the benzene
wavelength in order to collect data for (at least) nine analytes.
In general, mass spectra from 75 to 180 m/z were recorded for every 20 laser shots resulting in a
~ 2 s sampling period. The high-mass detection level was established to reduce the total amount
of data collected and was determined by the lack of ion signals at higher mass number under
conditions where real time low mass responses were clearly visible. Although this approach
results in large datasets (more than 200 MB after 30 minutes), it also ensures the capability to
verify if recorded ion signals are true ion signals or signals recorded due to changes in baseline
signals as a consequence of changing ambient conditions inside the 16 foot truck enclosure
where the REMPI-TOFMS instrument was positioned.
A daily time synchronization of all equipment ensured the accuracy of all sampling periods.
11.1.5 REMPI-TOFMS Data Analysis Procedure
Under sampling conditions with ion signals well above the real time detection limit of ~ 1 ppb,
the integrated area under a mass peak relates directly to the concentration by comparison of this
integrated area with that of a calibrated gas mixture integrated area recorded under the same
experimental condition. Across this jet engine exhaust sampling study, recorded ion signals at a
2 s sampling period were frequently found to be near or below the 2 s detection limit of many of
the target analytes. Therefore, time averaging was utilized to enhance the signal to noise ratio by
reduction of the (random) noise. Time (30 s or 60 s average) mass spectra were obtained in post-
data collection processing that represent the average mass spectrum at a specific wavelength per
30 or 60 s time interval.
Changes in the baseline signal will cause a bias in the results. To eliminate such bias, the average
integrated area of two adjacent masses with no visually verified ion signal was subtracted from
the integrated area for the mass of interest. The average concentration was than obtained through
direct comparison with the integrated area under a dynamically spiked air sample with TO-14
96
-------
EPA/600/R-10/050
April 2010
calibration gas. The presented REMPI-TOFMS results are focused on the sampling times with
conventionally sampled (Summa) canisters.
11.2 Results
11.2.1 F-15 Tests
Table 11-2 (obtained from Battelle) describes the followed sequence in sampling. Note that the
ground idles (tests No. 1-A-l and 1-B-l) were not performed since the exhaust plume did not
reach the probe due to strong crosswinds that morning.
Table 11-2. Summary of F-15 engine tests.
Date
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
10-12-07
Test No.
1-A-2
1-A-3
1-A-4
1-B-2
1-B-3
1-B-4
2-A-1
2-A-2
2-A-3
2-A-4
2-B-1
2-B-2
2-B-3
2-B-4
Tail No.
031
031
031
031
031
031
031
031
031
031
031
031
031
031
Engine
2
2
2
2
2
2
1
1
1
1
1
1
1
1
Nominal
Thrust %
80
85
92
80
85
92
70
80
85
92
70
80
85
92
Start
canister
sampling
9:00:52
9:16:45
9:31:43
10:19:16
10:54:19
11:09:14
13:05:35
13:20:23
13:36:54
13:51:26
14:13:15
15:25:10
15:40:40
15:55:09
Stop
canister
Sampling
9:10:52
9:26:45
9:41:43
10:49:16
11:04:19
11:34:14
13:15:35
13:30:23
13:46:54
14:01:26
14:43:15
15:35:19
15:50:40
16:20:09
Probe
Location
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Figures 11-2 through 11-5 show the concentration profiles during the one day of testing on an F-
15 airplane. The 1A and IB series in Figure 11-2 show very minimal response for benzene,
naphthalene, and methylnaphthalene. Figure 11-3 shows the results for the 2A series. In the early
afternoon, the wind conditions were more favorable than in the morning to detect the exhaust
plume, even at the idle conditions. This resulted in more favorable detection of both REMPI
analytes as well as CC>2. A clear correlation is visible between the observation of benzene, and
CC>2, due to the occasional interference of cross winds. Note that the large benzene transient
between 2A2 and 2A3 sampling occurred when the jet engine was changing thrust levels
confirming that fast changes in engine conditions (revving up and down in rpm) result in
transient emissions. In comparison with the first data set shown in Figure 11-2, the CC>2 emission
levels are considerably larger due to more favorable detection of the plume.
Figure 11-4 also shows data (inside blue box) for other BTEX-like target analytes (styrene,
phenol, p-xylenes, m-xylenes, o-xylenes, toluene) collected during a sequence of pre-determined
changes in REMPI laser wavelength. Detection of these analytes occurs only when the plume
reaches the probe as indicated by the relative high CC>2 concentration. Clearly there is a high
correlation between any REMPI-TOFMS detected analyte and the CC>2 concentration due to the
fact that the exhaust plume does not always reach the probe.
97
-------
F-15 exhaust Test 1A and 1B series
\
(16ppb)
.1A2. |1A3, .1Afc
Benzene
Naphthalene
Methylnaphthalene
Canister sampling periods
1B2
EPA/600/R-10/050
April 2010
3000
2000
1000
o
O
CM
O
O
09:00 09:30 10:00 10:30 11:00 11:30
Time (h:min)
Figure 11-2. Concentration profiles during F-15 engines testing.
F-15 exhaust Test 2A series
Benzene
Naphthalene
Methylnaphthalene
Canister sampling period
13:00 13:10
13:20 13:30 13:40
Time (h:min)
13:50 14:00
Figure 11-3. Concentration profiles during second sequence of F-15 engine testing.
98
-------
EPA/600/R-10/050
April 2010
F-15 exhaust Test 2B1 series
10-
,2B1
- Benzene
- Naphthalene
Methyl naphthalene
Canister sampling period -
14:10
Toluene
Phenol
Styrene
p- m- o- Xylenes
14:20 14:30
Time (h:min)
14:40
3000
Figure 11-4. Concentration profiles during the 30 minute test 2B1.
10-
F-15 exhaust Test 2B series
2B2 i ,2B3 i |2B4
Benzene
Naphthalene
Methylnaphthalene
Canister sampling period
il iiiWwn
4000
3000
Q.
2000
1000
£=
0)
O
£=
O
O
CM
O
o
14:20 14:40 15:00 15:20 15:40 16:00 16:20
Time (h:min)
Figure 11-5. Complete 2B series.
99
-------
EPA/600/R-10/050
April 2010
11.2.2 F-22 Tests
Table 11-3 shows the test schedule for the F-22 engine tests.
Figure 11-6 shows the CO2 and REMPI-TOFMS concentration data with the REMPI wavelength
set to detection of benzene and PAHs. Data for other analytes at other wavelengths recorded
during the 3B series resulted in non-detects for a 2s sampling period except for the low thrust
levels. The largest transient benzene concentration was recorded immediately after the switch
from sampling the back-flushed air to sampling of the aircraft exhaust. On this occasion, this
happened shortly (less than 1 minute) after the restart of the jet engine. The transient benzene is
therefore likely due to residual unburned fuel fumes that reached the sampling port. Changes in
thrust level happen well before and after each sampling period and were accompanied by
transient emissions that were not captured in the conventional samples.
Figure 11-7 shows the REMPI-TOFMS data for the 4A and 4B series. The interesting
observation here is the sharp increase in benzene concentration to a constant value between the
4A3 and 4A4 as well as after the 4A4 conventional samples. During this time, the aircraft engine
speed was reduced to a (unknown) low thrust level which agrees with the observed drop in CO2
concentration. However, the observed benzene concentrations with REMPI-TOFMS were much
higher than expected based on the observed CO2 concentration for lower thrust levels that were
recorded earlier in the sequence. Figure 11-8 complements Figure 11-7 with data for other
analytes that were sampled in the 4B series which are a replicate of the 4A series.
Here again, some analytes are non-detects when considering a 2 s sampling time, especially for
the higher thrust levels. Similar data responses were recorded during Test 5A and 5B and the
results are tabulated in Table 11-4.
For the 6A and 6B series, a comparison has been made with the PTR-MS data as measured by
Battelle and provided for comparison. Figure 11-9 has such comparison for the benzene and
naphthalene trace. For benzene, the agreement is very good, both in time response as well as
absolute intensity. In the case of naphthalene (as well as other BTEX-like analytes; not shown)
the quantitative comparison is fair at best with a consistently lower concentration measured with
REMPI-TOFMS than with PTR-MS.
A comparison between REMPI-TOFMS, PTR-MS, and (Summa) canister data as tabulated in
Table 11-4 is possible for only a limited number of points since (1) the canister data is
sometimes below the detection limit (2) no canister data is available for the 7A1 series and (3)
naphthalene is not analyzed using canisters.
If the canister data is considered to be the reference standard then there are seven data points in
Table 11-4 where the PTR-MS has a concentration value above the detection limit of the canister
as opposed to 0 data points for REMPI-TOFMS. This would suggest (independently from the
REMPI-TOFMS results) that the PTR-MS results are in general too high. A comparison between
REMPI-TOFMS and PTR-MS for of the obtained benzene concentration is good; others (where
available) fair with up to a factor 20 difference in average concentrations recorded for
naphthalene.
100
-------
EPA/600/R-10/050
April 2010
Table 11-3. Test schedule for the F-22 engine tests.
Date
10-15-07
10-15-07
10-15-07
10-15-07
10-15-07
10-15-07
10-15-07
10-15-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-16-07
10-17-07
10-17-07
10-17-07
10-17-07
10-17-07
10-17-07
10-17-07
10-17-07
10-17-07
10-18-07
10-18-07
10-18-07
10-18-07
10-18-07
10-18-07
Test No.
3-A-1
3-A-2
3-A-3
3-A-4
3-B-4
3-B-3
3-B-2
3-B-1
4-A-1
4-A-2
4-A-3
4-A-4
4-B-1
4-B-2
4-B-3
4-B-4
5-A-4
5-A-3
5-A-2
5-A-1
5-B-4
5-B-3
5-B-2
5-B-1
6-A-4
6-A-3
6-B-4
6-A-1
6-A-2
6-B-3
6-B-2
6-B-1
7-A-1
8-A-3*
8-A-A/B
9-A-3
9-A-A/B
9-B-3
9-B-A/B
Tail No.
045
045
045
045
045
045
045
045
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
43 FS
041
041
041
041
041
041
041
041
041
035
035
035
035
035
035
Engine
2
2
2
2
2
2
2
2
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
1
1
1
2
2
2
2
Nominal
Thrust %
10
20
70
80
80
70
20
10
10
20
70
80
10
20
70
80
80
70
20
10
80
70
20
10
80
70
80
10
20
70
20
10
10
30
150
70
150
70
150
Start
canister
sampling
9:53:23
10:07:44
10:22:30
10:39:32
12:00:08
13:29:32
14:36:40
14:52:03
9:08:18
9:22:40
9:38:12
9:57:29
10:17:39
10:33:23
10:48:44
12:40:50
13:02:49
13:19:10
13:36:43
14:22:00
14:38:46
14:54:28
15:09:55
15:23:19
9:15:29
9:29:44
9:54:09
10:16:47
10:32:47
12:06:59
12:20:32
12:35:54
12:51:28
10:10:30
10:18:00
12:16:09
12:28:44
14:40:22
14:52:33
Stop
canister
sampling
10:03:23
10:17:44
10:29:55
10:49:32
12:23:40
13:39:32
14:46:40
15:18:15
9:18:18
9:32:41
9:48:13
10:07:29
10:27:39
10:43:23
10:58:44
12:50:50
13:12:49
13:29:10
13:42:40
14:32:02
14:48:46
15:04:28
15:19:55
15:31:05
9:25:32
9:32:49
9:55:20
10:26:47
10:40:08
12:16:59
12:30:32
12:45:54
12:54:40
10:18:00
10:19:00
12:26:09
12:29:50
14:50:22
14:53:29
Probe
Location
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Near
Far
Far
Far
Far
Far
Far
101
-------
EPA/600/R-10/050
April 2010
Table 11-4. A comparison between Summa canister, REMPI-TOFMS and PTR-MS data
with the PTR-MS average calculated from provided Battelle data and canister
sampling time.
Test Number
6B3
6B2
6B1
7A1
Analyte
benzene
toluene
styrene
m- + p- xylenes
ethylbenzene
naphthalene
benzene
toluene
styrene
m- + p- xylenes
ethylbenzene
naphthalene
benzene
toluene
styrene
m- + p- xylenes
ethylbenzene
naphthalene
benzene
toluene
styrene
m- + p- xylenes
ethylbenzene
naphthalene
REMPI
(PPb)
0.05
0.03
ND
0.11
0.02
0.02
0.34
0.04
0.06
ND
ND
0.03
1.67
0.35
0.15
0.15
0.10
0.09
0.55
Not sampled
ND
Not sampled
Not sampled
0.06
PTR-MS
(PPb)
0.25
0.57
0.46
0.46
0.73
0.41
0.82
0.88
0.55
0.64
0.83
0.52
3.07
1.96
1.48
2.01
1.59
0.88
0.98
1.16
0.43
1.08
1.34
0.61
Canister
(PPb)
<0.36
<0.29
Not available*
<0.53
<0.23
No data
0.89
<0.29
Not available*
<0.53
<0.23
No data
3.84
0.92
Not available*
<0.35
<0.15
No data
Not available
Not available*
Not available*
Not available
Not available
No data
' No styrene data received from Battelle; should have been analyzed from Summa canister
102
-------
EPA/600/R-10/050
April 2010
-Q
CL
I
o
O
60 -,
50-
40-
I so ^
20-
10-
0- -
- Benzene
- Naphthalene
Methylnaphthalene
Canister sampling
10:00
11:00
12:00 13:00
Time (h: min)
14:00
-i 6000
5000
4000 E
Q.
a.
c
3000 £
to
'E
CD
2000
o
O
1000
15:00
Figure 11-6. Concentration during first sampling of F-22 engine.
60
50-
a. 40
Q.
C
O
30-
c
8
o 20
O
10-
0-
Benzene
Naphthalene
Methylnaphthalene
09:00 09:20 09:40 10:00 10:20
Time (h: min)
10:40
11:00
Figure 11-7. Concentrations during 4A and 4B sampling of F-22 engine.
103
-------
EPA/600/R-10/050
April 2010
5 _
I 4-
ntration (p
CO
0)
O
§ 2-
0
1 -
10
{HjU
4B1 ^62
WWN%K»
1 1
* Jl%
00 10:10 10:20 10:30 10:40
. 4B3
i to, L iiL jjijk
"w^fmw
J
Toluene
Phenol
Styrene
Mass 106
ilV*
10:50 11:
3000
Q.
Q.
- 2000
- 1000
o
O
o"
O
Time (h: min)
Figure 11-8. Concentrations for other REMPI-TOFMS detectable analytes.
After movement of the (Battelle) sampling equipment to the location further from the engine
nozzle, three afterburner tests were performed. Figure 11-10 shows the REMPI-TOFMS trace
during the third afterburner test. REMPI-TOFMS data is available for the first and second
afterburner test, however, the CO2 CEM trace is only complete for the last afterburner test. All
three afterburner tests are similar in response.
35
30
25
20
15
R 10
~Z 5
o
'g 0
'c 9
8
o
0 6H
3-
.663
6B2
6B1
-REMPI Benzene
PTR-MS Benzene
. 7A1
- REMPI Naphthalene
- PTR-MS Naphthalene
11:50 12:00 12:10 12:20 12:30 12:40 12:50 13:00
Time (h:min)
Figure 11-9. Comparison between REMPI-TOFMS and PTR-MS.
104
-------
EPA/600/R-10/050
April 2010
75-
70-
65-
60-
.
^55-
-R so-
f 45:
.§ 40-
1 35-
0 30-
o
o 25-
o
° 20-
15-
10-
5-
0-
14
A/vA ftA
V^^WjvA/
r V
»V^ ^*^
/^
.
_
Q
CO
2
_
J,
r uuu
6000
5000 "§
£L
Q.
4000 g
1
3000 g
c
o
O
2000 c.
O
O
\ A M AT V 1 10°°
L^J^J/V^fl
51 14:52 14:53 14:54 14:55 14
u
56
Time (h:min)
Figure 11-10. Benzene and PAHs response during afterburner test.
Figure 11-11 shows the comparison between REMPI-TOFMS and PTR-MS. In order to make
such comparison, the REMPI-TOFMS data was revised to reflect a 10s sampling period to match
the PTR-MS data. Here again, the comparison for the benzene data is very good while the
naphthalene data is good in time dependence but fair in absolute concentration value.
Average benzene concentrations as obtained with REMPI-TOFMS (see Table 11-3) were found
to compare very well with the canister data as shown in Figure 11-12. The linear fit is to the
black square data only. One outlier, namely the 9B AB (afterburner) value (red square in Figure
11-12) had a 6 ppb concentration according to the summa can data and was non-detect for
REMPI. This outlier can be explained by a slight (2-4 sec) difference in actual sampling times of
the canister and REMPI during which a strong transient benzene spike may have occurred during
the afterburner-to-idle transition (see Figure 11-10). When such a shift is included, the REMPI
average concentration during this test becomes 6 ppb, in accord with the summa can data.
Figure 11-13 shows the average benzene concentrations measured with REMPI-TOFMS during
the canister sampling at the four thrust levels and four engines with duplicates. When corrected
for average CO2 concentration, the emission factor in g/g carbon burned can be derived as shown
in Figure 11-14.
The observed trend of lower emissions with higher thrust levels is consistent across the data set.
However, emissions for the lowest thrust level are not very reproducible, even for the same
engine. This could suggest that the runtime history of the engine prior to sampling at the lowest
nominal thrust level changes the emission level.
105
-------
EPA/600/R-10/050
April 2010
40-
35-
30-
0.
S 25-
£=
O
ro 20-
4'
§ 15-
o
O
10-
5-
Benzene REMPI
Naphthalene REMPI
Benzene PTR-MS
Naphthalene PTR-MS
14:50 14:52 14:54 14:56
Time (h:min)
14:58
Figure 11-11. Comparison REMPI-TOFMS with PTR-MS for benzene and naphthalene.
Q.
Q.
5-
B 4-
o>
O
c
O
O
3-
Experimental data
Linear Fit
95% Confidence Limits
1234567
REMPI Benzene concentration (ppb)
Figure 11-12. Comparison between time averaged REMPI-TOFMS benzene data and
Summa canister results.
106
-------
EPA/600/R-10/050
April 2010
o
O
7.0-
6.5-
6.0-
5.5-
5.0-
4.5-
3.5-
3.0-
2.0.
1.5-
0.5.
0.0.
-0.5-
Plane 1
Plane 1
Plane 2
Plane 2
Plane 2
Plane 2
Plane 3
Plane 3
Engine 1
Engine 1 repeat
Engine 1
Engine 1 repeat
Engine 2
Engine 2 repeat
Engine 1
Engine 1 repeat
1234
Thrust Level (% of max)
Figure 11-13. Benzene concentrations as function of thrust level F-22 engine.
Plane 1 Engine 1
Plane 1 Engine 1 repeat
A plane 2 Engine 1
T Plane 2 Engine 1 repeat
Plane 2 Engine 2
* Plane 2 Engine 2 repeat
* Plane 3 Engine 1
Plane 3 Engine 1 repeat
10
20 30 40 50 60 70 80
Nominal Thrust Level (%)
90 100
Figure 11-14. Benzene emission factor (ug/g carbon burned) for F-22 engines.
107
-------
EPA/600/R-10/050
April 2010
12. Conclusions
This project has developed a technology termed REMPI-TOFMS for real time detection of trace
aromatic organic compounds in complex (combustion exhaust) matrices. During this project the
REMPI-TOFMS technology was transformed from a laboratory type, large frame research
instrument to a compact system that was taken into the field and used to characterize trace
aromatic air toxics. The REMPI-TOFMS system versatility was demonstrated by sampling on a
broad range of sources from an MWC to an Ml Abrams battle tank. The system also proved
portable, functional in harsh environments, reproducible, sensitive, and quick to respond, making
it ideal for characterization of multiple source types and compound types.
Operation of the LIBS system for detection of metals showed that its use was limited by the
presence of carbonaceous soot. An LTA procedure prior to analysis looked promising but was
not pursued due to the departure from the project of the LIBS cooperator, the U.S. Army
Research Laboratory. Testing with ORS instrumentation did not prove fruitful, as daily drift of
the criteria pollutant readings (compared to on-line CEMs) lent doubt to the value of these
measurements for point source measurement.
The low ppbv detection limit of REMPI-TOFMS is sufficient for detection of BTEX and small
PAHs in exhaust gas streams from vehicles in real time. This is a capability not previously
accomplished on such a wide variety of sources and to such an extent. Only in cases of
significant exhaust dilution (high load AGE and higher thrust levels of jet engines) were the real
time detection limits of the instrument insufficient to provide real time results. Variation of the
system procedures allowed collection of an averaged response up to 1 minute in duration to
greatly enhance the sensitivity of the instrument, resulting in minimal species non-detects.
Emission factors for many small organic aromatic compounds in exhausts from a diesel
generator, aircraft ground equipment, waste combustor, two HMMWVs, Abrams and Bradley
vehicles, and F-15 and F-22 aircraft engines have been reported for the first time. Strong and
significant transient emissions appear during changes in engine conditions, including shutdown
and restarts. Steady state conditions were, in general, found to be constant over an extended
period of time. REMPI-TOFMS was able to clearly distinguish effects of engine type (diesel,
turbine) on emission characteristics.
In critical operations, the REMPI-TOFMS instrument can be used to provide operational
feedback, improving efficiency and reducing emissions. The speed and sensitivity of the REMPI
instrument allows transient and steady state characterization of target pollutants emissions that
can be used in assessing or developing new exhaust gas treatment systems for mobile sources or
in developing emission factors by use-mode. For reasonably high ambient levels of air toxics,
REMPI-TOFMS can also be used as an ambient air monitor. Its recent application to a near-road,
highway emissions study proved its usefulness in open source sampling. Both of these
108
-------
EPA/600/R-10/050
April 2010
applications, point source or area source, are appropriate for REMPI-TOFMS and, thus, the
technology could prove useful in a diagnostic and characterization capacity.
109
-------
EPA/600/R-10/050
April 2010
13. References
Badger, G. andNovotny, J., 1963. Mode of Formation of 3,4-Benzopyrene at High
Temperatures. Nature 198, 1086
Blumenstock, M.; Zimmermann, R.; Schramm, K.-W.; Kettrup, 2000. Influence of combustion
conditions on the PCDD/F-, PCB-, PCBz- and PAH-concentrations in the post-combustion
chamber of a waste incineration pilot plant. Chemosphere 40, 987-993.
Blumenstock, M.; Zimmermann, R.; Schramm, K.-W.; Kettrup, 2001. Identification of surrogate
compounds for the emission of PCDD/F (I-TEQ value) and evaluation of their on-line real-time
detectability in flue gases of waste incineration plants by REMPI-TOFMS mass spectrometry.
Chemosphere 42, 507-518.
Boesl, U., Nagel, H., Weickhardt, C., Frey, R., Schlag, E., 1998. Vehicle exhaust emission time-
resolved multicomponent analysis by laser mass spectrometry. In: Meyers, R.A. (Ed), The
Encyclopedia of Environmental Analysis and Remediation, Volume 87. John Wiley, New York,
5000-5022.
Boesl, U., Heger, H. , Zimmermann, R., Piiffel, P., Nagel, H., 2000. Laser mass spectrometry in
trace analysis. In: Meyers, R.A. (Ed), Encyclopedia of Chemical Analysis. Wiley, New York,
2087-2118.
Burtscher, H., 1992. Measurement and characteristics of combustion aerosols with special
consideration of photoelectric charging and charging by flame ions. Journal of Aerosol Science
23, 549-595.
Clark, N., Daley, J.J., Nine, R.D. and Atkinson, C.M., 1999. Application of the new City-
Suburban Heavy Vehicle Route (CSHVR) to Truck Emissions Characterization. SAE Paper
1999-01-1467.
Cowen K., Kelly T., Dindal A., Willenberg Z., Riggs K., 2006. Environmental Technology
Verification Report: SRI International Jet-REMPI (Resonance Enhanced Multi-Photon
lonization), Battelle.
Dunbar, J.C., Lin, C.-L, Vergucht, I, Wong, J., Durant, J.L., 2001. Estimating the contributions
of mobile sources of PAH to urban air using real-time PAH monitoring. Science of the Total
Environment 279, 1-19.
Franzen, J., Frey, R., Nagel, H, 1995. Fast monitoring of motor exhaust components by resonant
multi-photon ionisation and time-of-flight mass spectrometry. Journal of Molecular Structure
347, 143-52.
110
-------
EPA/600/R-10/050
April 2010
Frey, R., Nagel, H., Franzen, J, Rikeit, H.-E., 1995. Time-resolved measurement of individual
aromatic hydrocarbons in automotive exhaust at transient engine operation. SAE Technical Paper
Series No. 951053, Society of Automotive Engineers International, Warrendale, PA.
Gass, H., Liider, K., Wilken, M., 2002. PCDD/F- Emissions during cold start-up and shut-down
of municipal waste incinerator. Organohalogen Compounds, 56, 193
Gerstle, T., Virag, P., Wade, M.D., Brown, P.P., 1999 Aircraft Engine and Auxiliary Power Unit
Emissions Testing: Final Report. Addendum. Fl 19-PW-100 Engine Emissions Testing Report;
USAF, IERA, IERA-R2-BR-TR-1999-0006, March 1999.
GlobalSecurity.org, 2008a,
http://www.globalsecurity.org/military/systems/ground/images/ml 114-up-armored-specsOl .jpg
(accessed January 2009)
GlobalSecurity.org, 2008b,
http://www.globalsecurity.org/military/systems/ground/ml-specs.htm (accessed January 2009)
Good, Phillip, "An Overview of European Emissions Legislation for Light Duty Vehicles,"
International Conference on Transport and Environment Milan, 19-21 March 2007.
http://transportenv07.jrc.ec.europa.eu/pdf/D/D 1 19 Good P An%20overview%20of%20%20
EU%20Emissions%20Legislation.pdf (accessed January 2009)
Gross, M.; Cernuschi, S.; Palini, E.; Lodi, M.; Mariani, G. 2004. PCDD/Fs release during normal
and transient operation of a full scale MSWI plant. Organohalogen Compounds 66, 1227.
Gullett, B. K. and Wikstrom, E., 2000. Mono- to tri-chlorinated dibenzodioxin (CDD) and
dibenzofuran (CDF) congeners/homologues as indicators of CDD and CDF emissions from
municipal waste and waste/coal combustion. Chemosphere 40, 1015-1019.
Gullett, B.; Touati, A.; Oudejans, L.; Ryan, S.; Tabor, D., 2006. Transient PCDD and PCDF
concentrations in an MWC. Organohalogen Compounds 68, 143.
Gullett, B.; Touati, A.; Oudejans, L.; Tabor, D. 2007. AWMA Hazardous Waste Combustor
Conference, Charleston, SC, March 13-14, 2007.
Heeb, N., Forss, A.-M., Wilhelm, P., 2001. Comparison of methane, benzene and alkyl benzenes
cold start emission data of EURO I and EURO II three-way catalyst passenger cars and light
duty vehicles from time resolved exhaust gas analysis. Proceedings International Symposium
Transport and air pollution, Boulder Colorado USA, 17-19 September 2001.
Heger, H. J.; Zimmermann, R.; Dorfner, R.; Beckmann, M.; Griebel, H.; Kettrup, A.; Boesl, U.,
1999. In-Line Catalytic Derivatization Method for Selective Detection of Chlorinated Aromatics
with a Hyphenated Gas Chromatography/Laser Mass Spectrometry Technique: A Concept for
Comprehensive Detection of Isomeric Ensembles. Analytical Chemistry 71, 46-57.
Ill
-------
EPA/600/R-10/050
April 2010
Heger, H., Zimmermann, R., Blumenstock, M., Kettrup, A., 2001. On-line real-time
measurements at incineration plants: PAHs and a PCDD/F surrogate compound at stationary
combustion conditions and during transient emission puffs. Chemosphere 42, 691-696.
Kemme, M.R.; Brown, W.T., Frame, E.A., and Alvarez, R.A. 2006. Diesel-Powered Equipment
Properties and Activity Database for DoD Off-Road Sources, SERDP Project WP-1338,
ERDC/CERL TR-06-6, March 2006.
Kimm, L., Sylvia, D., Gerstle, T., Virag, P., 1997. Proceedings of the Air & Waste Management
Association Annual Meeting, June 8-13, 1997, Toronto, Ontario, Canada.
Kuribayashi S., Yamakoshi H., Danno M., Sakai S., Tsuruga S., Futami H., Morii S., 2005. VUV
single-photon ionization ion trap time-of-flight mass spectrometer for on-line, real-time
monitoring of chlorinated organic compounds in waste incineration flue gas. Analytical
Chemistry 77, 1007-1012
Levy, Jr., J.A., Schawitsch, T.L., Thoreson, H.F., Walter, J.K., 1993. Estimation of toxic air
pollutant emissions and development of a supporting information management system for
Vandenberg Air Force Base. 86th Annual Meeting & Exhibition, Paper 93-TP-30B.02, Air &
Waste Management Association, June 13-18, Denver, CO.
Nagel, H., Frey, R., and Boesel, U., 1996. On-line analysis of formaldehyde and acetaldehyde in
non-stationary engine operation using laser mass spectrometry. SAE Technical Paper Series No.
961084, Society of Automotive Engineers International, Warrendale, PA.
Neuer-Etscheidt, K.; Nordsiek, H.; Liu, Y.; Kettrup, A.; Zimmermann, R., 2006. PCDD/F and
Other Micropollutants in MSWI Crude Gas and Ashes during Plant Start-Up and Shut-Down
Processes. Environment Science Technology 40, 342.
Nine, R.D., Clark, N.N., and Norton, P., 2000. Effects of Multiple Driving Test Schedules
Performed on Two Heavy-Duty Vehicles", SAE Fuels & Lubricants Meeting, Baltimore, Oct.
2000, SAE Paper 2000-01-2818.
NIST Mass Spec Data Center, S.E. Stein, director, "Mass Spectra" in NIST Chemistry
WebBook, NIST Standard Reference Database Number 69, Eds. P. J. Linstrom and W.G.
Mallard, June 2005, National Institute of Standards and Technology, Gaithersburg MD, 20899
http://webbook.nist.gov.
Oudejans, L., Touati, A., Gullett, B., 2004. Real-Time, On-Line Characterization of Diesel
Generator Air Toxic Emissions by Resonance Enhanced Multi-Photon Ionization Time of Flight
Mass Spectrometry, Analytical Chemistry 76(9), 2517-2524.
Rakopoulos, C.;Mavropoulos, G., Hountalas, D., 1998. Modeling the Structural Thermal
Response of An Air-Cooled Diesel Engine Under Transient Operation Including a Detailed
Thermodynamic Description of Boundary Conditions. Society of Automotive Engineers, SAE
Technical Paper Series No. 981024.
112
-------
EPA/600/R-10/050
April 2010
Rhead, M. and Pemberton, R. Sources of Naphthalene in Diesel Exhaust Emissions. Energy &
Fuels 10, 837-843, 1996.
Ryan, J., Lemieux, P., Preston, W., 1998. Near-real-time measurement of trace volatile organic
compounds from combustion processes using an on-line gas chromatograph. Waste Management
18,403-410.
Schultz, G., long, I, Kefauver, K. and Ishibashi, J., 2006. Steering and Handling Testing Using
Roadway Simulator Technology. International Journal of Vehicle Systems Modeling and
Testing, Vol. 1, No. 1/2/3, pp. 32-47.
Schulz, H., De Melo, B., and Ousmanov, F., 1999. Volatile Organic Compounds and Particulates
as Components of Diesel Engine Exhaust Gas. Combustion and Flame 118, 179-190.
Thanner R., Oser H., Grotheer H.-H., 1998. Time-resolved monitoring of aromatic compounds in
an experimental incinerator using an improved jet-resonance-enhanced multi-photon ionization
system jet-REMPI. European Mass Spectrometry 4, 215-222
U.S. Army, JP-8, The Single Fuel Forward, Armaments Command Research, Development, and
Engineering Center, MI, 2001. http://www.mfp.usmc.mil/TeamApp/G4/, "JP-8 Single Fuel Info
Compendium
U.S. Environmental Protection Agency, SW-846 On-line, Test methods for evaluating solid
wastes, Physical/Chemical methods, Office of Solid Waste, Method 0010 Modified Method 5
Sampling Train, 1986a, at URL http://www.epa.gov/epaoswer/hazwaste/test/pdfs/0010.pdf,
accessed on November 12, 2003.
U.S. EPA Method 5 "Determination of Parti culate Matter Emissions from Stationary Sources,"
found in 40 CFR part 60, appendix A, 1986b. http://www.epa.gov/ttn/emc/promgate/m-05.pdf
(accessed January 2009)
U.S. Environmental Protection Agency, 1993. Emission factor documentation for AP-42 section
3.4, large stationary diesel & all stationary dual fuel engines.
http://www.epa.gov/ttn/chief/ap42/ch03/bgdocs/b03s04.pdf (accessed January 2009).
U.S. Environmental Protection Agency, SW-846 On-line, Test methods for evaluating solid
wastes, Physical/Chemical methods, Office of Solid Waste, Method 0040 Sampling of Principal
Organic Constituents from Combustion Sources using TEDLAR Bags, 1996a, at URL
http://www.epa.gov/epaoswer/hazwaste/test/pdfs/0040.pdf (accessed January 2009).
U.S. EPA Method 8270 in SW-846 Test Methods for Evaluating SolidWaste, Physical/Chemical
Methods, Office of Solid Waste, 1996b.
http://www.epa.gov/epawaste/hazard/testmethods/sw846/pdfs/8270d.pdf (accessed January
2009)
113
-------
EPA/600/R-10/050
April 2010
U.S. EPA Method 0011 in SW-846 Test methods for evaluating solid wastes, Physical/Chemical
methods, Office of Solid Waste, 1996c.
http://www.epa.gov/epawaste/hazard/testmethods/sw846/pdfs/0011 .pdf (accessed January 2009)
U.S. EPA Method 8315a in SW-846 Test methods for evaluating solid wastes, Physical/Chemical
methods, Office of Solid Waste, 1996d.
http://www.epa.gov/epawaste/hazard/testmethods/sw846/pdfs/8315a.pdf (accessed January
2009)
U.S. EPA M-23A, Test Method 23 A, Sampling method for poly chlorinated dibenzo-p dioxins
and polychlorinated dibenzofuran emissions from stationary sources. EPA-SW 846 On-Line, 3rd
ed., 1996d. Available at:
http://www.epa.gov/epawaste/hazard/testmethods/sw846/pdfs/0023a.pdf.
U.S. EPA Method TO-15 "Determination of Volatile Organic Compounds (VOCs) in Air
collected in Specially-Prepared Canisters and Analyzed by Gas Chromatography/Mass
Spectrometry (GC/MS)", 1999. http://www.epa.gov/ttn/amtic/files/ambient/airtox/to-15r.pdf
(accessed January 2009)
U.S. Environmental Protection Agency, 2000a. Performance specification 2 specifications and
test procedures for SO2 and NOx continuous monitoring systems in stationary sources. 40 CFR
Ch.l Pt.60, App. B, Spec.2. http://www.epa.gov/ttn/emc/specs/prompspec2.html (accessed
January 2009).
U.S. EPA, 2000b, Method 10A: "Determination Of Carbon Monoxide Emissions In Certifying
Continuous Emission Monitoring Systems At Petroleum Refineries,"
http ://www. epa. gov/ttn/emc/promgate/m-1 Oa.pdf.
U.S. Environmental Protection Agency, 2004. EPA original list hazardous pollutants
http://www.epa.gov/ttn/atw/188polls.html (accessed March 2005).
U.S. EPA, 2006, Method 3 A, "Determination of Oxygen and Carbon Dioxide Concentrations in
Emissions from Stationary Sources (Instrumental Analyzer Procedure),"
http ://www. epa. gov/ttn/emc/promgate/m-03 a.pdf
Velazquez, J.; Voloboueva, L. A.; Cool, T. A., Selective Detection of Dibenzodioxin,
Dibenzofuran and Some Small Poly cyclic Aromatics. Combustion Science Technology 134,
139-163.
Weickhardt, C.; Boesl, U., 1993. Time resolved trace analysis of exhaust gas by means of laser
mass spectrometry. Ber. Bunsenges. Phys. Chem. 97, 1716-19.
Zimmermann, R.; Boesl, U.; Weickhardt, C.; Lenoir, D.; Schramm, K.-W.; Kettrup, A.; Schlag,
E.W., 1994. Isomer-selective ionization of chlorinated aromatics with lasers for analytical time-
of-flight mass spectrometry: First results for polychlorinated dibenzo-p-dioxins (PCDD),
biphenyls (PCB) and benzenes (PCBz). Chemosphere 29, 1877-88.
114
-------
EPA/600/R-10/050
April 2010
Zimmermann R., Heger H., Blumenstock M., Dorfner R., Schramm K.-W., Boesl U., Kettrup A.,
1999. On-line measurement of chlorobenzene in wast incineration fl ue gas as a surrogate for the
emission of poly chlorinated dibenzo-p-dioxins/Furans (I-TEQ) using mobile resonance laser
ionization time-of-flight mass spectrometry. Rapid Communications Mass Spectrometry 13,
307-314.
115
-------
EPA/600/R-10/050
April 2010
14. Appendix A: List of Scientific/Technical Publications
14.1.1 Journal Articles
L. Oudejans, A. Touati, andB.K. Gullett, "Real-Time, On-Line Characterization of
Diesel Generator Air Toxic Emissions by REMPI-TOFMS ", Anal Chem., 76, 2517-2524
(2004)
Brian K. Gullett, Abderrahmane Touati, Lukas Oudejans, and Shawn P. Ryan, "Real-time
emission characterization of organic air toxic pollutants during steady state and
transient operation of a medium duty diesel engine ", Atmosph. Envir., 40, 4037-4047
(2006)
Brian Gullett, Abderrahmane Touati, and Lukas Oudejans, "Use of REMPI-TOFMS for
real-time measurement of trace aromatics during operation of aircraft ground
equipment", Atmosph. Envir., 42, 2117-2128 (2008)
Brian Gullett, Lukas Oudejans, Abderrahmane Touati, Shawn Ryan, and Dennis Tabor,
"Verification results of jet resonance-enhanced multiphoton ionization as a real-time
PCDD/Femission monitor", J. Mater. Cycles WasteManag., 10, (2008) 32-37
submitted to Environmental Science & Technology, entitled "Transient PAH, PM, CO,
and CO 2 Emission Responses from HMMWVs, the Ml Abrams tank, and the Bradley
Infantry Fighting Vehicle" by Brian Gullett, Lukas Oudejans, and Abderrahmane Touati
(2009).
14.2 Oral and Poster Presentations
Results obtained during this project were presented to (international) audiences at the following
conferences:
Annual conferences:
o Partners in Environmental Technology Technical Symposium & Workshop,
Washington, DC, annual, 2000-2007 (Posters)
o International Karasek Conference, annual, 2001-2008 (oral and/or poster)
Individual conferences:
o Seventh International Congress on Toxic Combustion By-Products, NIEHS,
Research Triangle Park, NC (2001) (Oral)
o 26th Annual EPA-Air & Waste Management Association Information Exchange
EPA, RTF (2001) (Oral)
o 22nd International Symposium on Halogenated Environmental Organic Pollutants
and POPs (Dioxin 2002), Barcelona, Spain (2002) (Oral)
o 27th Annual EPA-Air & Waste Management Association Information Exchange
EPA, RTF (2002) (Oral)
o Eighth International Congress on Toxic Combustion By-Products, NIEHS, at
University of Arizona, Tucson, AZ (2003) (Poster)
116
-------
EPA/600/R-10/050
April 2010
o Laser Applications to Chemical and Environmental Analysis (LACEA) conference,
Annapolis (2004) (Poster)
o 3rd International Conference on Combustion, Incineration/Pyrolysis and Emission
Control (3rd i-CIPEC), Hangzhou, China, (2004). (Oral)
o Air and Waste Management Association (AWMA) Symposium on Air Quality
Measurements and Technology, Durham, NC (2006) (Oral)
o 10th International Congress on Combustion By-Products and their Health Effects,
Ischia, Italy (2007) (Oral)
117
------- |