EPA/600/A-97/036
For Presentation at the Air & Waste Management Association's 90th Annual
Meeting & Exhibition, June 8-13,1997, Toronto, Ontario, Canada
97-RP139.04
Atmospheric Chemistry Studies of Exhaust from Vehicles Operating with
Reformulated Fuel
T.E. Klelndienst and D.F. Smith
ManTech Environmental Technology, Inc., P.O. Box 12313, Research Triangle Park, NC 27709
F.M. Black and S.B. Tejada
National Exposure Research Laboratory, U.S. Environmental Protection Agency, Research Triangle
Park, NC 27711
Introduction
In many areas of the United States, ozone concentrations continue to be higher than that specified by the
NAAQS. In spite of substantial improvements, automobile exhaust still remains one of the major sources
of ozone precursors, reactive organic gases and oxides of nitrogen. One strategy for dealing with this
emission source is to use fuels that have lower reactivity than conventional gasoline. The degree of
ozone reduction expected from these fuels has only been estimated using reactivity scales which are
largely based on modeling results.
The U.S. Congress, in the 1990 Clean Air Act Amendments [1] has required further reduction of
emissions from new motor vehicles, introduction of environmentally favorable reformulations of
conventional diesel and gasoline fuels, and introduction of alternative fuels such as methanol, ethanol, and
natural gas with associated vehicle technologies. Gasoline reformulation is designed to result in mass
(and reactivity) reductions of both ozone-forming volatile organic compounds (VOCs) and toxic
compounds (formaldehyde, acetaldehyde, polycyclic organic matter, benzene, and 1,3-butadiene). The
Alternative Motor Fuels Act of 1988 (AMFA) also mandated the introduction of alternatives to
conventional vehicle and fuel technologies. In addition to changes and improvements in fuels,
considerable efforts have been made by automotive manufacturers to produce vehicles that generate
lower levels of exhaust and evaporative emissions. Many of these new technology vehicles are
specifically designed to use alternatives to conventional fuels. As new fuels and vehicle technologies are
introduced, the distribution of compounds introduced into the atmosphere, particularly, the urban
atmosphere, could change considerably.
With lower levels of exhaust and evaporative emissions from new vehicles, their associated reactivity
(i.e., their ability to produce ozone) has been found to be lower than that from present-day vehicles,
according to computer models. Whereas kinetic and mechanistic experiments have been conducted on
individual fuel components of both reformulated and alternative fuels, little effort has been expended to
examine effects from the entire exhaust. Information on the photochemical transformations of these
emissions is vital for determining tropospheric lifetimes, effects on ozone formation, and possible
1

-------
generation of toxic and genotoxic reaction products in the atmosphere.
The formation of ozone and other toxic compounds is frequently studied in the laboratory using smog
chambers. Research in studying reactivities of complex mixtures typically uses surrogate mixtures to
represent automotive exhaust and/or evaporative emissions. This approach permits controlled additions
of the initial reactant mixtures to the chamber. However, there are two potential shortcomings to this
approach. First, there is no guarantee that the surrogate mixture adequately represents the distribution of
reactive constituents present in the exhaust. Second, the formulation of a realistic surrogate mixture is
not necessarily straightforward for fuels having a complex composition. For fuels, such as compressed
natural gas (CNG), having a few major constituents, surrogates which represent vast majority of the
reactivity can be produced by mixing one to two dozen organic constituents in the proper ratio in a high
pressure cylinder. However, for gasoline-based fuels more sophisticated approaches for surrogate
development must be undertaken.
We have recently assembled a facility for conducting direct irradiations of exhaust from vehicles operated
on a dynamometer. The approach has permitted a direct measure of reactivity parameters (related to the
maximum incremental reactivity) for complex exhaust mixtures. Controlled irradiations of these mixtures
has shown remarkable reproducibility, much better than originally anticipated. This high degree of
experimental reproducibility has permitted us to examine in detail direct comparisons between surrogate
mixtures designed to represent automotive exhaust and automotive exhaust itself. The reported study
describes experiments that have been conducted to examine the reactivity and major product constituents
from direct irradiations of complex exhaust mixtures. These experiments will serve as the basis for
producing realistic surrogates for automotive exhaust. The approach is demonstrated with exhaust
generated with a reformulated gasoline (RFG). A complex surrogate representing this exhaust was
designed and formulated to allow the initial conditions to be generated reprodudbly. A series of
irradiations were then undertaken to compare reaction profiles for the adjusted exhaust mixtures.
Experimental Section
The dynamometer/photochemistry chamber system used in this study was previously developed to
developed to examine the reactivity of exhaust from vehicles operating on alternative fuels. While the
main features of the apparatus have been previously described, some of the salient and additional aspects
of the experimental design are included below.
Apparatus
A schematic diagram for the experimental apparatus required to conduct these studies is shown in
Figure 1. The primary components for this system are the dynamometer (not shown) for generating the
required exhaust emissions, the photochemical chamber with associated apparatus for reactant
introduction and product sampling, and the analytical instiumentation for making chemical and physical
measurements.
Automobile exhaust was generated in an adjoining area from vehicles operated on a chassis
dynamometer. The dynamometer used in these studies was a twin-roll, electric chassis dynamometer
manufactured by Burke-Porter (Grand Rapids, MI). The entire exhaust from the automobile was
transferred through a heated, stainless-steel line to the dilution tunnel. The exhaust was diluted at 20 m3
min'1 with a constant volume sampler (CVS) using ambient air to reduce exhaust concentrations and to
2

-------
equilibrate the hot exhaust gases. Exhaust samples for injection into the chamber were collected from the
dilution tunnel with the use of five Metal Bellows pumps (Sharon, MA; Models MB-302 and MB-151)
operating in parallel. Exhaust was collected in 200 L Teflon bags for transfer to the photochemical
chamber.
The reaction chamber is an 9,000-L rectangular chamber constructed of 2-mil Teflon film. Samples are
taken from fittings mounted on a Teflon plate at the bottom of the chamber. A fan was installed to
achieve active mixing of the reactants and pure compounds injected into the chamber. Ultrazero air was
used in the chamber as the base air matrix and for as dilution during the irradiations. The air was
produced in-house with an Aadco pure air generator (Aadco, Inc., Model 737). This air typically had
NOx levels less than 1 ppbv and hydrocarbon levels less than 50 ppbC. The irradiation chamber is
typically operated with 2-3 L min"1 dilution air to compensate the volume loss due to sampling.
The irradiation source for indoor experiments was based on a system of 122-cm fluorescent bulbs.
Radiation in the range 300-350 nm was generated with UVA-340 (Q-Panel, Cleveland, OH) fluorescent
blacklight bulbs; radiation at wavelengths 350-400 nm was generated with standard UV bulbs (F40-BL).
The mixture of bulbs was deigned to match, to the extent possible, the distribution of solar radiation
between 300 and 400 nm. In most cases, the distribution of bulbs were 50% UVA-340 and 50% F40-
BL. It has been found that different lots of the UV-340 bulbs can yield different absolute radiation
intensities and therefore each lot is checked with a spectral radiometer. The spectral distribution of the
light setting was measured with a portable spectroradiometer (LI-COR, Inc., Model LI-1800). During
the experiment, the light intensity was continuously monitored with an integrating radiometer (Eppley
Laboratory, Inc. Newport, RI) that measured radiation between 300 and 385 nm.
Materials and Surrogate Formulation
Exhaust was generated from a Dodge Caravan with a V-6,3.3 L engine. This vehicle was selected based
on other program requirements. The vehicle was fueled with California Phase II reformulated fuel. The
major physical characteristics of this fuel are given in Table 1. Hydrocarbons and NOx for this study
were derived from collected automobile exhaust generated during the first two cycles (i.e., the first 124 s)
of Bag 1 of the FTP. This mixture was relatively hydrocarbon rich compared to exhaust mixtures
obtained from other gasoline based vehicles. These exhaust characteristics provided a unique opportunity
to measure exhaust reactivity without a substantial addition of hydrocarbon to bring the mixture into a
photochemically active regime, that is having a sufficiently high reactive organic gas (ROG)/NOx ratio
initially present. In some cases, NO from a 100 ppmv in air cylinder mixture (National Specialty Gases,
Raleigh, NC) was required to produce equivalent photochemical starting conditions. This cylinder was
also used in irradiations of the hydrocarbon surrogate mixture.
The hydrocarbon surrogate was formulated to match the hydrocarbon composition of the RFG exhaust
mixture described above. The exhaust and associated surrogate comprised 4 major components
important to ozone formation - the fuel itself the major combustion-derived alkenes (Cj- C4), Cy - C,
aromatic hydrocarbons, and - Cj carbonyl compounds. Calculations were initially performed to
establish the reactivity of the RFG exhaust using principles of incremental reactivity [2]. The reactivity-
weighted compositional difference between the fuel and the exhaust was then determined. Appropriate
amounts of pure organic compounds were then used to make adjustments to the hydrocarbon mixture to
better represent the reactivity of the exhaust. According to the calculations, better than 95% of the
i

-------
reactivity was represented by the resultant surrogate mixture. To produce the experimental mixture the
whole fuel and appropriate levels of 16 individual organic components were added to an evacuated
vessel. The vessel was then filled with ultrazero air to produce a mixture in the high part-per-million
level. Table 2 provides overall characteristics of the surrogate mixture produced and includes the identity
of the 25 most abundant compounds along with the distribution of the surrogate according to
hydrocarbon class and number. The contents of the vessel could then be used for reproducible chamber
experiments with the addition of formaldehyde, carbon monoxide, NOx, and water vapor. When this
RFG hydrocarbon surrogate is irradiated in the chamber in the presence of NOx, it was found that the
mixture produced 99% of the ozone found in the exhaust after a similar 12 h irradiation.
Sampling and Instrumentation
Gas-phase samples woe taken through a 200-mL Teflon manifold that was connected to the chamber
with 6.4-mm Teflon tubing. The sampling line was inserted 45 an above the chamber floor to ensure that
a well-mixed and representative sample was obtained from the chamber. Nitric oxide (NO), total NOx,
and ozone were measured with continuous monitors and techniques as previously described [3,4].
Temperature and relative humidity (RH) were measured with an Omega digital Thermo-Hydrometer
(Model RH411). The temperature-humidity sensor was placed in an insulated Teflon tube that was in-
line with samples taken from the manifold. Air samples drawn past the temperature and RH sensor were
in close proximity to the irradiation chamber but shielded from the UV lights. This approach permitted
the air temperature of the chamber samples to be accurately measured while avoiding radiative heating
from fluorescent bulbs. To prevent excessive chamber heating from the fluorescent bulbs, cooled air
from an auxiliary air conditioner in the laboratory was circulated in the ballast cavity, as well as between
the lights and the chamber walls. This configuration allowed the temperature of the mixture in the
chamber to be maintained between 25 and 27 °C during the irradiation.
Direct concentrations of vapor-phase water in the chamber were measured continuously by using a fast
response water analyzer (LICOR Corp., Lincoln, NE; Model LI6262). This analyzer could provide more
accurate and precise measurements of the water level in the chamber than could be achieved with the
combination of temperature and relative humidity indicated above. The instrument operates by IR
absorption and concentrations above 500 ppmv can be accurately determined. For these measurements, a
sample flow of 0.5 L/min was used and dry nitrogen was used as the reference cell gas.
Measurements of organic precursors and reaction products were obtained from the chamber as a function
of irradiation time. Speciated hydrocarbons were measured by GC with detection by flame ionization
(FID), electron capture (ECD), or mass spectrometry (MS) [5], Sample were cryogenically collected and
separated on a methyl silicone column (Restek Corp. RTx-t, 60 m * 0,32 mm, 1-^m film thickness)
which was connected serially to a Carbowax column (Restek Corp. Stabilwax, 30 m * 0.32 mm, 0.5-^m
film thickness). This approach substantially improves the peak shape of polar compounds produced
during the irradiation and permitted separation of methanol from C4 hydrocarbons.
Peroxyacetyl nitrate (PAN) and other peroxyacyl nitrates were measured by GC/ECD. The column used
for analysis was of moderate polarity, trifluoropropyl methyl silicone (Restek RTx-200; 30 m x 0.53 mm,
l-/zm film thickness) giving an elution order similar to those obtained with a standard Carbowax column.
The column, operated at room temperature, provided baseline separation for PAN and peroxypropionyl
nitrate (PPN) and mitigated moisture problems. The carrier gas for the analytical column was ultrahigh
purity hydrogen with detection using a ®Ni ECD.
4

-------
Measurements for carbonyl compounds were obtained by bubbling the effluent through a solution of 2,4-
dinitrophenylhydrazine (DNPH) in acidified acetonitrile. The derivatization reactions produce carbonyl
hydrazones which were separated and quantified by HPLC (Milton Roy, Model CM4000). The column
used in the analysis was a Zorbax ODS-1 column with detection of the separated hydrazones by UV
absorption at 360 nm. A ternary gradient, composed of acetonitrile, methanol and water, was used to
provide reasonable separation of intermediate-weight carbonyl compounds [6].
Absolute concentrations of nitric acid were obtained through nylon filter technique. Approximately
100 L of the chamber effluent was drawn through a 47-mm, l-jim nylon membrane filter (Gelman
Sciences, Inc., Ann Arbor, MI) at a flow rate of 5 L/min. Filters were extracted in 15 mL of 0.01 mM
perchloric add solution. The extract was analyzed by ion chromatography (Dionex, Inc., Sunnyvale, CA)
for nitrate ion. The analysis used a standard anion column (AS4) with a HC037C03" eluent and
conductivity detection.
Identifications of /-butyl formate were performed principally by GC/MS. The instrument
(Hewlett-Packard Model 5970) was tuned with perfluorotributylamine (PFTBA, or FC-43). An initial
library search of the mass spectrum for TBF was conducted using the Wiley/NBS Mass Spectra Library.
Pure samples of TBF were procured for matching spectra and chromatographic retention time. Full-scan
electron impact pi) spectra (70 eV) were then collected for mass units 35-250 to validate the standard
spectrum, particularly with respect to the principal mass ratios.
Characteristic ions were selected from the mass spectra for TBF for the quantitative SIM runs. No
problems were associated with this process because each of the components was chromatographically
well resolved. With the cryogenic sampling system, 100 mL gas samples were collected, then flash
injected into the GC/MS. Each ion was integrated for 150 ms. Standard bags containing known
quantities of TBF were run each day usually at the beginning and end of each set of experiments to
determine calibration factors and check the instrument performance. For the SIM mode, the TBF ions
used for analysis were m/e = 59 and 87. In most cases, concentrations were quantified with the 59 ion
and the 87 ion was used for confirmation. Thus, the presence of these ions On the correct ratio) at the
appropriate retention time served as the criteria for validating the presence of TBF. Due to the presence
of water in the photochemical system, the full 100 mL sample could not be collected in many instances.
However, sample collection volumes of 25 mL also yielded an excellent signal-to-noise ratio for typically
measured sample concentrations detected during the irradiation (1-5 ppbv). An examination of the
instrumental noise indicated a detection limit of 0.03 ppbv (S/N=3) for the full 100 mL collection volume.
Procedures
Exhaust from test vehicles was generated using a conventional dynamometer operated under the Federal
Test Procedure (FTP). The exhaust samples were injected into a smog chamber and irradiated under
conditions which simulate ambient urban atmospheres. The experimental procedures generally followed
those previously described [4], Prior to irradiations, the chamber was typically flushed with ultrazero air
for 48 h. Background samples for detailed hydrocarbons and carbonyl compounds verified negligible
levels of these compounds prior to reactant addition.
Automobile exhaust samples were collected into 200-L Teflon bags during the first two cycles of the
FTP, that is, the first 124 s of Bag 1. During this period, there is a substantially higher hydrocarbon
emissions before the catalyst becomes operative. After collection, the exhaust content of the Teflon bags
were immediately iqected into the irradiation chamber using a metal bellows pump. The initial chamber
mixture was targeted at an ROG/NOx ratio of 5.5. The volume of exhaust injected into the chamber was
5

-------
typically limited by the level of N0X present in the exhaust sample. Exhaust was injected to bring the
initial chamber concentration of NOx to the target value of400 ppbv. If the ROG/NOx ratio was below
the target ratio of 5.5, surrogate was added to raise the mixture to this target ratio. If the ROG/NOx
ratio was greater than 5.5, NO was added to the chamber to lower the ratio to the appropriate level. In
either case, this approach rendered an initial ROG concentration of 2.2 ppmC.
Once the reactant mixture had been added to the chamber, initial samples (i.e., time zero irradiation
samples and parameters) were taken and the irradiation source was turned on. Physical parameters and
inorganic concentrations (i.e., NO, NOx, 03) were monitored continuously during the irradiation.
Samples for peroxyacyl nitrates were taken at 1-h intervals. Detailed NMHC, carbonyl, and nitric add
samples required direct operator intervention aid were taken at 2 h intervals. GC/MS measurements for
TBF were made at the beginning, midpoint, and at the end of the irradiation. The total irradiation time
was typically 12 h or the time required to reach a peak in the ozone concentration.
Results
The results are given for three experiments (referred to as IR109, ER.110, and IR111) where RFG exhaust
was irradiated. For two of the experiments, greater than 89% of the hydrocarbon was derived from the
exhaust mixture. In IR111, all of the hydrocarbon initially present in the chamber was derived from the
exhaust. As noted, experiments were conducted at an initial ROG/NOx ratio of 5.5. The ratio was
selected to accentuate differences in reactivity from the composition of the hydrocarbon mixture. At
substantially lower ratios (e.g., <3), these mixtures have unacceptably low reactivity for radiation levels
that can be achieved in the chamber. At substantially higher reactivities, the system becomes NOx-Iimited
and the rate of oxidant formation becomes far less dependent on the ROG distribution.
Table 2 provides the overall initial chamber conditions for each of the experiments. In each experiment,
the target conditions (ROG and NOx) were largely achieved. IR111 shows a slightly higher ROG/NOx
ratio than the other two experiments. In each of these runs water was added to the initial mixture
through a humidification system. At the temperature of the irradiations, these added water
concentrations correspond to approximately 50% relative humidity. The distributions of the major
hydrocarbon classes are given and show nearly equivalent paraffin and aromatic abundances, which are
approximately twice that of the olefin. The oxygenate (i.e., MTBE) and carbonyl abundances together
are less than 10%of the total ROG.
As indicated in Table 3, IR109 required a 0.117 ppmC (5.4%) increase of the hydrocarbon surrogate to
give the desired initial ROG/NOx ratio; Experiment IR110 required a 0.230 ppmC (10.9%) increase of
the surrogate. Major features for the composition of the hydrocarbon surrogate representing RFG
exhaust are given in Table 2. The surrogate shows excellent conformity to the exhaust in both
composition and reactivity. However, there were some compositional differences between the surrogate
and the exhaust. Alkynes, which are generated exclusively in the combustion process (particularly
acetylene), were not added to the surrogate mixture due to their negligible effect on the reactivity. By
contrast, MTBE was substantially higher in the surrogate than in the exhaust, since it is removed during
the combustion process. In the exhaust, MTBE represents approximately 2% of the total emitted carbon
in the 124 s bag. As in the case of the alkynes, MTBE has a minor influence on the total reactivity.
Finally, the exhaust produced approximately 50% greater carbonyl mass than found in the surrogate.
Total carbonyl abundance in the exhaust was never larger than 3%, however. Thus, for purposes of these
irradiations the hydrocarbon surrogate is considered to be an excellent representation of the exhaust,
6

-------
particularly given the low levels of surrogate required to adjust the initial ratio.
Figure 2 presents time series profiles for the major inorganic species in the photochemical system which
include nitric oxide, nitrogen dioxide, and ozone. The photochemical reactions are typically initiated by
the photolysis of compounds such as formaldehyde or nitrous acid. Both compounds can serve as a
primary source of OH which oxidizes organic compounds present. Nitrous acid can be formed at the
chamber walls from a heterogeneous reaction between N02 and water and photolyses to give OH
directly. The conversion of NO to N02 is clearly seen in the figure. Once NO is removed from the
system ozone increases until it formation rate is equivalent to the loss rate. The overlaid data shows the
excellent agreement for the family of profiles for the three experiments.
The major set of reactive compounds formed in photochemical systems are carbonyl compounds,
especially the aldehydes, formaldehyde and acetaldehyde. Time series profiles for these compounds are
given in Figures 3(a) and 3(b), respectively. For both compounds, the rate of formation exceeds the rate
of removal due to OH reaction and/or photolysis until an irradiation time of 4 h and leads to the observed
peaks in concentration. The acetaldehyde data shows high experimental consistency for the 3
experiments. By contrast, the formaldehyde is considerably more scattered and is probably more sensitive
to difference in the distribution of precursors. Figure 3c gives the data for acetone, which next to
formaldehyde is the second most abundant carbonyl compound formed in these irradiations. Again in this
case, the run-to-run consistency for this compound is remarkably good. Acetone is formed from
precursor hydrocarbons which have two methyl groups in the alpha position with respect to alkoxy
radical intermediate, such as the alkoxy radical formed from the oxidation of isobutylene.
Termination reactions form inorganic and organic nitrates which are found as stable products in most
photochemical systems. Figure 4 shows the time-series profiles for the two major nitrogen containing
products, nitric acid and peroxyacetyl nitrate (PAN). These are also the major NOy compounds found in
the troposphere. Both plots show excellent agreement among the experiments for the first six hours of
irradiation with moderate divergences toward the end of the radiation period.
The major product for the oxidation ofMTBE (the major oxygenated component of RFG) is tertiary-
butyl formcte (TBF), From previous studies [7], this compound has been found to form with a reported
yield ranging between 0.6 - 0.7 and thus was examined during this study. During these experiments,
measurements for TBF were made for the preirradiation mixture, at the midpoint of the irradiation, and at
the ozone peak. TBF concentrations for all preirradiation samples were below the detection limit of
approximately 0.03 ppbv. Table 4 gives both the reactive losses for MTBE and formation yield for TBF.
IR110 and IR111 were quite consistent with previous kinetic results even given the substantial
uncertainties from these measurements. The 11 h sample from run IR110 was lost due to an excessive
injection of water onto the GC column. In these measurements, the major uncertainties for the MTBE
loss included the low level ofMTBE present initially in the mixture (ca. 10 ppbv), its relatively slow rate
of reaction with OH, and corrections for required for dilution. TBF measurement uncertainties included
difficulties in producing an accurate calibration at the 10 ppbv level and calibration reproducibility in the
mass spectrometer.
Discussion
The photooxidation of RFG exhaust results in oxidation of NO toM)2 followed by the formation of
ozone once the NO level is sufficiently low to render the NO + ozone reaction as unimportant. The rate
7

-------
of NO removal in the system is an excellent indicator of the reactivity of the system. This value generally
ranges from 1 ppbv/min for slowly reacting mixtures to greater than 5 ppbv/min for mixtures that show
high reactivity. In these experiments, the values ranged between 2.4 and 2.9 ppbv/min. These values
have the most meaning in comparison with the reactivity from other systems.
Several features typify the composition ofRFG exhaust, the two main features being relatively high
abundance of M1BE and it major combustion product, isobutylene. At present, the photochemistry of
MTBE is thought to be fairly well understood [7, 8, 9]. MTBE is removed from the atmosphere mainly
by its reaction with the hydroxyl radical (OH). However, the removal rate is relatively slow with an OH
+ MTBE rate constant of 3 x 10"12 cm3 molec s"1. This value leads to an atmospheric lifetime of
approximately 3 days, assuming an average OH concentration of 0.05 pptv. The mechanism for OH
reaction with MTBE is through hydrogen atom abstraction from the methyl or l-butyl group of MTBE
with abstraction from the methyl group favored by a factor of three. Abstraction from the methyl group
leads to the formation ofTBF. The preponderance of evidence from mechanistic studies suggest that the
yield from the reaction is 67%. The experiments in this study have yielded measurable quantities ofTBF
as expected from prior studies. However, given extremely low initial concentrations of 10 ppbv MTBE
and the slow removal rate relatively little conversion of MTBE was found to occur. Not withstanding
this situation, positive identification for TBF was made by mass spectrometry in the selected ion mode.
TBF yields in this study were consistent with previous measurements of 68%, although the precision here
is substantially poorer than previous reported measurements, particularly for the amount of MTBE
removed during photooxidation upon which the yield measurements rely.
The MS-SIM measurements provided excellent precision for TBF, as noted earlier 0.03 ppbv under the
sampling conditions for this study. Measurements of the initial chamber composition thus provided a
sensitive measure of the amount ofTBF present in the combustion mixture. For the initial ROG
concentration of 2.2 ppmC Mid the previously stated detection limit, less than 0.008% of the exhaust
mixture would consist ofTBF. Moreover, some additional GC/MS measurements were made from the
undiluted exhaust which also does not show the presence ofTBF in the nonirradiated exhaust. Thus, it is
highly unlikely that TBF observed in environmental samples result directly from the combustion ofRFG
fuel.
Isobutylene (2-methylpropene) is a highly reactive compound formed from the combustion of MTBE.
Detailed hydrocarbon measurements indicates that the compound comprises approximately 5% of the
total ROG and approximately 25% of the olefin component in the exhaust. This compound reacts with
OH radicals with a rate constant of 5.2 x 10'" cm3 molec"1 s"1 and thus having an atmospheric lifetime of 7
h under the conditions previously specified. OH reacts with isobutylene by adding to the double bond of
the molecule. While reaction to the most substituted carbon is preferred, addition at other ate will lead
to the same products. Following normal photooxidation mechanisms and by analogy with ethylene and
propylene, the alkoxy radical formal following OH addition is expected to decompose to give equimolar
quantities of acetone and formaldehyde. Since acetone is substantially less reactive than either
formaldehyde or acetaldehyde, isobutylene is expected to be less reactive than either ethylene and
propylene which each form 2 aldehydic molecules for each reacted molecule. The relatively high acetone
levels observed during the photooxidation suggests that isobutylene is an important yet not sole precursor
to this compound.
The initial isobutylene concentrations for the exhaust irradiations were 37.1,41.4, and 35.5 ppbv for
IR109, IR110, and IR111, respectively. Figure 3c shows corresponding time series profiles for the
formation of acetone. Tins plot shows a rapid rise in the acetone concentration over 4-6 h which is most

-------
likely due mainly to reaction from one or more olefins, with isobutylene being the most reasonable
contributor given its high initial concentration. Since isobutylene is completely removed by 6 h, further
increase in the acetone concentration is probably due to the oxidation of moderate molecular weight
alkanes such as isopentane or 2,2,4-trimethylpentane which has tertiary carbons.
Secondary reactions of acetaldehyde leads to the formation of PAN. Compounds, such as methyl glyoxal
or biacetyl, can also formed acetyl radicals following photolysis or reaction with OH. Thus, these
molecules can also serve as a precursor for PAN. Methyl glyoxal and biacetyl are typically generated
following reaction of OH with aromatic compounds. As Table 3 indicates, aromatic hydrocarbons
(particularly toluene, xylenes and the trimethylbenzenes) are present at yields approaching 40% of the
total carbon. Product distributions from aromatic compound reactions with OH are poorly understood.
However, their rapid reactions with OH (for >Cj aromatics) and the potential for products to form
radicals make these compounds highly reactive. However, these compounds also have a propensity for
reacting with N02 to remove NOx from the system and careful studies under a broad range of conditions
must be undertaken to evaluate the reactivity of aromatic compounds.
As seen from the initial conditions in Table 3, moderate variability is present even under highly controlled
conditions. Even with such variability, compensating factors appear to raider the inorganic parameters
largely invariant to the specific conditions provided that the initial levels of NOx and total ROG are
consistent. For example, the peak ozone level in the three experiments are all within 3%. The agreement
is even closer when the values are normalized to the initial NOx present. Reactivity factors (e.g.,
-d[NO]/dt and d[03]/dt) can also be calculated from the inorganic profiles. These factors allow
comparisons in the reactivity of different real-world exhaust mixtures. Further consideration of these
issues will be left for a separate forum.
Conclusions
This study has been conducted to examine the atmospheric chemistry an reactivity of real-world exhaust
mixtures. Considerable effort has been expended in establishing an accurate representation of the initial
conditions for these experiments. Use of exhaust from the first 124 s of the FTP for the vehicle utilized
permitted irradiations of the exhaust mixture without the substantial addition of a surrogate compound.
Even with the moderate variability observed in these experiments, the formation of oxidant compounds
such as ozone and PAN are quite reproducible. Reactive intermediates, particularly formaldehyde, are
considerably less reproducible in their time-series profile. A major differences between RFG exhaust and
that produced with standard 1990 industry average gasoline is the presence ofMTBE and isobutylene.
The oxidation ofMTBE produces a high yield of TBF which have been reconciled on a semiquantitative
basis in this study. However, the conversion ofMTBE was extremely slow in the chamber, as it is in the
ambient atmosphere. Likewise, isobutylene produces acetone as a major product for which their is
reasonably good mass balance between the reactive removal of isobutylene and the formation of acetone.
Since acetone reacts extremely slowly with OH in the ambient atmosphere, it could distribute widely over
the upper troposphere and serve as an additional source of radicals following photolysis.
9

-------
Acknowledgments and Disclaimer
The authors wish to acknowledge the efforts of Eric Corse and Ned Perry of ManTech Environmental,
Inc. for experimental operation in the chamber and dynamometer facilities. Analytical support was also
provided by Kimberley Wood, Colleen Loomis, and Tholeathcus Grantham of Clean Air Vehicle
Technology Center, Inc. The authors also wish to acknowledge Michael Gurevich (Department of
Energy) and Brent Bailey (National Renewable Energy Laboratory). Partial support for this project came
from the U.S. Department of Energy.
The information presented in this document has in part been funded by the U.S. Environmental Protection
Agency. It has been subject to the Agency's peer and administrative review. Mention of trade names or
commercial products does not constitute endorsement or recommendation for use.
References
1.	Public Law 101-549 (1990) An Act to amend the Clean Air Act.
2.	Carter, W.P.L. (1994) Development of ozone reactivity scales for volatile organic compounds. J. Air
Waste Manage. Assoc., 44,881-899.
3.	Kleindienst, T.E., D.F. Smith, E.E. Hudgens, R.F. Snow, E. Perry, L.D. Claxton, J.J. Bufalini, F.M.
Black, and L.T. Cupitt (1992) The photooxidation of automobile emissions: Measurement of the
transformation products and their mutagenic activity. Atmospheric Environment, 26A, 3039-3053.
4.	Kleindienst, T.E., F. Liu, E.W. Corse, and J .J. Bufalini (1994) Development of a chamber system for
determining measures of reactivity from exhaust of alternative-fuel vehicles. SAE Technical Paper
No. 941906, Society of Automotive Engineers, Warrendale, PA
5.	Smith, D.F., T.E. Kleindienst, E.E. Hudgens, and I.J. Bufalini (1994) Measurement of organic
atmospheric transformation products by gas chromatography. Intern J. Environ. Anal Chem54,
265-281.
6.	Smith, D.F., T.E. Kleindienst, E.E. Hudgens (1989) Improved high-performance liquid
chromatographic method for artifact-free measurements of aldehydes in the presence of ozone using 2,4-
dinitrophenylhydraane. J. Chromatog. 483,431-436.
7.	Japar, S.M., T.J. Wellington, F.O. Rickert, and J.C. Ball (1990) The atmospheric chemistry of
oxygenated fuel additives: /-butyl alcohol, dimethyl ether, and methyl /-butyl ether. Int. J. Chem. Kinet.
22,1257-1269.
8.	Smith, D.F., T.E. Kleindienst, E.E. Hudgens, C.D. Mclver, and J.J. Bufalini (1991) The
photooxidation of methyl tertiary-butyl ether. Int. J. Chem. Kinet., 23, 907-924.
9.	Tuazon, E C., W.P.L. Carter, S.M. Aschmann, and R. Atkinson (1991) Products of the gas-phase
reaction methyl tert-butyl ether with the OH radical in the presence of NOx. Int. J. Chem. Kinet, 23,
1003-1015.
10

-------
Table 1. Fuel specification for RFG used in the study.
Specification
Value
Specification
Value
Specific Gravity
0.7385
Distillation, °C.

RVP, psi
7.06
IBP
39
Sulfur, wt %
0.0036
10%
56
MTBE, v %
12.4
50%
93
Alkane, V%
56.0
90%
148
Alkene, V %
5.8
¦
EP
191
Aromatic, V %
25.8
Lower Heating Value,
112,240
BTU/gal

-------
Table 2. Results of hydrocarbon analysis for the RFG hydrocarbon currogate; (left) top 25
^rago^ds^^^on^i^dM^i^^^jenencdirtri^tionjacco^^to_^^on_^^Aer^_
Hydrocarbon	Surrog Cone" fl Distribution	% of Total
for 2.2 ppmC H	Hydrocarbon
Toluene
0.1800
Alkanes: C4
1.07
Methyl /-butyl ether
0.1636
C5
7.92
m+p-Xylene
0.1621
C6
7.12
Isopentane
0.1463
C7
10.21
2,2,4-T rimethylpentane
0.1302
C8
12.30
Ethylene
0.1113
C9
1.65
Isobutylene
0.0972
>C9
0.48
2,3-Dimethylpentane
0.0790
Alkenes: C2
5.06
Ethylbenzene
0.0675
C3
2.94
Propylene
0.0646
C4
5.05
o-Xylene
0.0613
C5
1.45
2-Methylpentane
0.0521
C6
1.29
1,2,4-Trimethylbenzene
0.0471
C7
0.84
1 -Methyl-3-ethylbenzene
0.0437
>C7
0.95
2,3,4-T rimethylpentane
0.0416
Aromatics: C6
0.98
2,4-Dimethylpentane
0.0384
C7
8.18
3-Methylpentane
0.0343
C8
13.36
3-Methylhexane
0.0316
C9
7.56
2-Methylhexane
0.0296
>C9
2.35
2,3-Dimethylbutane
0.0260
Total Alkanes:
40.8
H-Pentane
0.0259
Total Alkenes:
17.6
1,3,5-Trimethylbenzene
0.0226
Total Aromatics:
32.4
Benzene
0.0215
Total Oxygenates:
7.9
/i-Butane
0.0214
Total Carbonyls:
1.2
1 -Methyl-4-ethyIbenzene
0.0194
Unknowns:
0.1

-------
Table 3. Initial conditions for RFG exhaust iiradiations. RFG surrogate additions indicated.
Parameter
mi 09
mi io
mm
NO, ppbv
402
396
397
NOx, ppbv
234
342
360
ROG, ppmC
2.17
2.11
2.22
Surrogate, %
5.4
10.9
0
ROG/NOx
5.39
5.33
5.59
Water, ppmv
16,300
15,200
15,300
HCHO, ppbv
14.2
18.7
10.3
CHjCHO, ppbv
3.3
4.2
2.8
CO, ppmv
8
8
10
Sum Alkanes, %C
32.4
33.6
36.0
Sum Alkenes, %C
19.3
21.5
20.1
Sum Alkynes, %C
3.6
4.1
6.7
Sum Aromatics, %C
39.7
36.1
32.3
Sum Oxygenate, %C
2.4
4.1
6.7
Sum Carbonyl, %C
2.4
2.9
2.6
Table 4. Concentrations (ppbv) of /-butyl formate (TBF) for measured reactive losses of MTBE.
TBF concentrations from selected ion mass spectrometry (m/e=59) and initial concentrations were
below the detection limit of approximately 50 pptv. MTBE concentrations were based on a per
caibon response factor from propane and its losses were corrected for dilution.
Experiment MTBE loss MTBE loss TBF formed TBF formed
	(6h)	OJJO	(6j0	(11 h)
IR109	2.0	4.2	1.8	4.9
IR110	2.4	3.9	1.9	—
IR1U	3.2	6.2	2.2	4.1

-------
Carrier Row Control
Water
Ari&riyzdr
ECD FID
Sample
Heated Row
Valve Control Pump
Mass Spec
and Data
Systeii
Ozone
Monitor
Temperature
Control


Sample Lines

Nylon Filter
UW-Wi
9,000
Dry Test Meter

Row
Controller
250-L Teflon Bag
for Exhaust Addition
x 7
UV Radiation
Implngerfbr
1 j DNPH Analysis
Fignre 1. Schematic of chamber irradiation system for alternative-fuel automobile exhaust.

-------
0.8
0.6 -
Q.
3
c
o
S 0.4 -
c
8
c
o
o
0.2 -
Ozone
NOx-NO
8
10
12
Irradiation Time (h)
Figure 2. Time series profiles for the major inorganic components ozone, NO, and NOy-NO.

-------
100
£ 80
Q.
a.
c
.2
I
«*¦»
c
©
u
c
o
O
O
X
o
60
40
20
(a)
~
4c
~
IR109 IR110 IR111
¦ ~ *
40
8
10
12
30
2
Q.
3
c
.2
I
§ 20
c
o
O
O
110
CO
x
o
60
§50
GL
&
I 40
1
1 30
f 20
c
o
(b)
.~
*
~
*
~
$ ¦
IR109 IR110 IR111
¦ ~ *
6
8
10
12
(C)
110
~
*
~
¦*
~
*
L
IR109 IR110 IR111
¦ ~ *
6	8
Irradiation Time (h)
10
12
Figure 3. Time series profiles for selected carbonyl compounds formed through termination
reactions during the irradiations (a) formaldehyde, (b) acetaldehyde, and (c) acetone.

-------
[R109
IR110
200
R109
4	6	8
Irradiation Time (h)
Figure 4. Time series profiles for the major NOy compounds formed through termination reactions
during the irradiations (a) peroxyacetyl nitrate; PAN, (b) nitric add; HN03.

-------
TECHNICAL REPORT DATA
1. REPORT "NO.
EPA/600/A-97/036
2.
i
4. TITLE AND SUBTITLE
Atmospheric Chemistry of Exhaust from Vehicles
Operating with Reformulated Fuel
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
7. AUTHOR(S!
Tadeusz Kleindienst, David Smith, Frank Black, and
Silvestre Tejada
8.PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
10.PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68 D5 0049
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
U.S. Department of Energy
Washington, DC 20585
13.TYPE OF REPORT AND PERIOD COVERED
Proceedings of Annual Meeting
of the Air & Waste Management
Association
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The Alternative Motor Fuels Act (AMFA) of 1988 requires the Department of Energy (DOE) to purchase alternative fuel
motor vehicles for use in government fleets, and further requires DOE to cooperate with the Environmental Protection
Agency (EPA) and the National Highway Traffic Administration to evaluate potential safety, fuel economy, and emissions
implications of these alternative technologies. EPA and DOE initiated cooperative study of alternative fuel vehicle
emissions and their implications for air quality in 1991. This paper describes studies of the atmospheric photochemistry of
motor vehicle exhaust emissions with reformulated gasoline fuel. Alternative fuel (e.g., methanol, ethanol, and natural gas)
benefits must be contrasted with those of reformulated gasoline. The results of indoor (using an array of fluorescent lights to
simulate sunlight) chamber irradiations of vehicle exhaust are reported. Time profiles of NOx, ozone, PAN, nitric acid, and
aldehydes are reported. Tertiary butylformate, a transformation product of MTBE emissions, is also examined.
17. KEY WORDS AND DOCUMENT ANALYSIS
a., DESCRIPTORS
b.IDENTIFIERS/ OPEN ENDED
TERMS
C.COSATI



16. DISTRIBUTION STATEMENT
RELEASE T0 POTLIC

19. SECURITY CLASS (This
Report)
UNCLASSIFIED
21.NO. OF PAGES

20. SECURITY CLASS (This
Page)
UNCLASSIFIED
22. PRICE

-------