An Analysis of the Benzene Scheme for Three Versions of the SAPRC Mechanism
Draft Report to the California Air Resources Board
Contract No. 07-730
By Wendy Goliff
January 6, 2012
Center for Environmental Research and Technology
College of Engineering
University of California
Riverside, California 92521
Abstract
Three versions of the SAPRC mechanism (SAPRC07, SAPRC11 and SAPRC11A) were used to simulate chamber experiments with benzene under both low NOx and high NOx conditions in the EUPHORE chamber, which were run as part of the EXACT campaign. Only the SAPRC07 mechanism provided predicted maximum ozone values within the stated error of observed values for the low NOx experiment. Model performance was not as good for the high NOx experiment, with only SAPRC07 predicting ozone values within measurement error of observations. Regarding the HOx species, each mechanism predicted peak radical levels close to observed values, although they underpredicted HOx concentrations during the afternoon hours. For model performance for product species such as glyoxal and measured ring-retaining products, the mechanisms were not able to predict peak concentration values close to those of observed values. Recommendations are as follows: use different ring-opening products for benzene and the other aromatics so that each degradation pathway may be tuned for each aromatic. For example, methyl glyoxal is not observed from the oxidation of benzene, and yet it is formed in amounts comparable to glyoxal by each version of the SAPRC mechanism. Also, it would be useful to compare SAPRC mechanism predictions to more product information (such as the sort provided by the EXACT campaign), especially if the modeling of particulate formation is desired.
Introduction
Benzene, the simplest of the aromatic compounds, has a relatively low reactivity in the atmosphere, with a tropospheric lifetime of 9.5 days (Atkinson and Arey, 2003) with respect to reaction with hydroxyl radical (OH). In spite of this, benzene is of interest in air toxics modeling, and has been classified as class 1 carcinogen by the International Agency for Research on Cancer (IARC). Although extensive research has been performed recently to elucidate its mechanism of photooxidation, the degradation of benzene in the atmosphere is still poorly understood.
Volkamer et al. (2002) studied the hydroxyl-initiated oxidation of benzene in the outdoor chamber EUPHORE in Spain and an indoor chamber at NIES in Japan. In the EUPHORE chamber, in which NOx levels and light conditions were representative of the atmospheric boundary layer, a phenol yield of 53.1 ± 6.6% was reported, and was independent of NOx and O2 concentrations.
Bloss et al. (2005) used data collected in the EXACT project (Effects of the oXidation of
Aromatic Compounds in the Troposphere) to update the Master Chemical Mechanism’s (MCM) aromatic scheme. The EXACT project involved the collection of a comprehensive dataset at the EUPHORE chamber for the elucidation of aromatic degradation in the atmosphere. Extensive instrumentation were used to collect measurements of O3, NO, NO2 (using DOAS as well as a NOx monitor), CO, PAN, phenol, cresols, catechols, glyoxal, methyl glyoxal, HONO, OH, HO2, actinic flux (for J(NO2)), HCHO and carbonyls. In the case of benzene, the authors found that the MCMv3.1 predictions for the O3 peak improved compared to the MCMv3 due to the increase in the yield of phenol diverting production away from radical-producing ring-opening products. However, this decrease in radical production resulted in a decrease of predicted OH compared to measured OH (termed “missing OH”) and a decrease in the oxidation capacity of the system.
Berndt and Böge (2006) studied the hydroxyl-initiated oxidation of benzene in a flow-tube reactor. In the absence and presence of NOx, they found a phenol yield of 0.61 ± 0.07. Carbonyls detected were glyoxal, cis-butenedial and trans-butenedial with formation yields of 0.29 ± 0.10, 0.08 ± 0.03 and 0.023 ± 0.007, respectively, measured in synthetic air and in the presence of NOx. In the absence of NOx, glyoxal, cis-butenedial and trans-butenedial were also detected, but with distinctly lower yields compared to the experiments with NOx.
Studies regarding the ring-opening products of the HO-benzene reaction are sparse, with main aspects of their formation uncertain due to lack of good experimental techniques for their quantification, a lack of commercially available standards, and their high reactivity. In addition to the study by Berndt and Böge (2006), Gomez Alvarez et al. (2007) investigated the photooxidation of toluene and benzene in the EUPHORE chamber. For the benzene experiments, they found yields of dicarbonyls using SPME fibers, with values of (42 ± 3) and (36 ± 2)% for the two successive experiments (September 24 and 25, 2003) of glyoxal, (17 ± 9)% for total butenedial [(8 ± 4)% cis-butenedial and (9 ± 5)% trans-butenedial (September 24, 2003)] and (15 ± 6)% total butenedial (September 25, 2003) [(7 ± 3) and (7 ± 3)% for the cis and trans isomers, respectively]. For this work, cis- and trans-butenedial were synthesized for calibration purposes. The implications of the work of Gomez Alvarez et al. 2007 were that the dicarbonyl reaction route was confirmed, and fast ring-cleavage was observed, due to a peak in observed -dicarbonyls shortly after the chamber was opened to sunlight. Also, high yields of dicarbonyls (e.g., glyoxal) imply a high formation rate of HO into the system.
In contrast to previous studies, Birdsall and Elrod (2011) did find that yields of phenol were NOx dependent. They also found that the dienedial yields were larger for benzene than for any other aromatic system, and were observed to increase significantly with NO, a result previously observed by Berndt and Boge, albeit at a significantly higher NO concentration than in the present work.
A continuing problem in chemical mechanism development is the over-prediction of ozone and underprediction of radicals in a variety of cases, especially with aromatic degradation (Bloss et al., 2005), even for the most widely known mechanisms such as the Regional Atmospheric Chemistry Mechanism (RACM), the Statewide Air Pollution Research Center mechanism (SAPRC99, SAPRC07), the Carbon Bond mechanism (CB05) and the Master Chemical Mechanism (MCM) (Bloss et al., 2005; Chen et al., 2010). In this work the chemical mechanism of benzene is explored with respect to ozone formation, radical and NOx concentrations, and product formation using the SAPRC mechanisms. To this end the chamber data acquired from the EXACT project was obtained from Dr. Pilling of Leeds University for both benzene experiments: low NOx and high NOx. In the low NOx experiment, the VOC to NOx ratio was 42 ppm/ppm, while in the high NOx experiment the VOC to NOx ratio was 5.6 ppm/ppm.
EXACT Benzene Experiments
The initial conditions for the experiments performed for the EXACT project were chosen to consider both the technical constraints (such as sufficient concentrations of VOC for accurate measurements) and to come as close as possible to atmospheric conditions (Bloss et al., 2005). The initial conditions for the benzene/high NOx and benzene/low NOx experiments are given in Table 1 (the NOx levels for the “high NOx” EXACT experiments are not as high as in the UCR database) . The chamber used for the project was the EUPHORE facility, which consists of two FEP foil hemispherical chambers with an approximate volume of 200 m3, and for which the transmission of both visible and UV light through the chamber walls is high (85–90% for wavelengths 500–320 nm, and around 75% at 290 nm). A complete description of the instrumentation used may be found in Bloss et al. (2005).
Table 1. Initial conditions used in simulations, concentrations measured before opening chamber to sunlight (Bloss et al., 2005)
Initial Concentration / Benzene/low NOx (ppm)(VOC/NOx = 42 ppm/ppm) / Benzene/high NOx (ppm)
(VOC/NOx = 5.6 ppm/ppm)
Benzene / 1.986 / 1.014
NO / 0.045 / 0.100
NO2 / 0.0022 / 0.026
HONO / 0.0001 (a) / 0.0555 (b)
O3 / 0.006 / 0
HCHO / 0 / 0
HNO3 / 0 / 0.0068
Glyoxal / 0 / 0.0005
CO / 0.751 / 0.615
(a) Estimated by authors
(b) Measured concentration.
Key compounds that were measured during the EXACT campaign fall into three groups according to their importance for the assessment of model performance. The first group contains the parent aromatic (in this case, benzene), ozone, NO and NO2. These compounds provide information on the ozone production and the oxidative capacity in the system. The second group contains compounds that deliver important information on the NOy budget, HOx production or the branching ratios for major reaction routes: HNO3, PAN, HCHO, glyoxal, methylglyoxal, cresol isomers and benzaldehyde (Bloss et al., 2005). The third group of compounds contains reaction products, such as butenedial. Typical uncertainties (2) were ±10% for NO2, aromatics, HCHO and ±5% for O3 and NO, and 26% (1) for the HOx species. (Volkamer et al., 2002; Bloss et al., 2004).
As in any smog chamber, air is lost throughout the experiment due to small leaks and withdrawal of air for samples for analysis. In the EUPHORE facility, clean air was added to compensate for this and some dilution of the reactants and products occurred as a result. To measure the dilution rate SF6 was added to the reaction mixture as an inert tracer. Dilution rates were 1.18 x 10-3 min-1 and 1.19 x 10-3 min-1 for the low and high NOx experiments, respectively (
Photolysis rates for this study were calculated using the photolysis program in Dr. Carter’s box model, using the appropriate latitude, sun declination and time start and end times. In addition, scaling factors to take into account the transmission through the walls and backscatter from the aluminum chamber floor were taken from the MCM web site.
Chamber related reactions
When modeling smog chamber experiments, it is always necessary to take into account wall effects. Current practice is to minimize wall effects through the use of inert materials and cleaning programs, but walls may still be a significant source and sink of contaminants, as well as serve as a reaction site for heterogeneous chemistry (Killus and Whitten, 1990). There are three major ways wall effects can interact with experimental photochemistry: (1) the introduction of free radicals, (2) trace NOx species can allow the formation of O3 and PAN and serves as a radical sink, and (3) organic off-gassing that convert hydroxyl radicals (OH) to HO2. An auxiliary mechanism describing the wall effects for the EUHPORE chamber was constructed for the EXACT campaign and tested with two characterization experiments using ethylene in high and low NOx regimes. Discrepancies between the model and chamber data for key reactants and products such as ethylene, ozone, NO and NO2 were interpreted as influenced by wall reactions. The auxiliary mechanism was tuned to fit the discrepancies between model and experimental data for ethylene and then was used to describe the wall effects in simulations for the aromatic experiments. The auxiliary mechanism is listed in Table 2. Wall reactions that produced less than 1% change in the maximum O3 concentration were not used in the auxiliary mechanism to minimize complexity (Bloss et al., 2005). Because ethylene degradation is well-understood, the MCM and SAPRC mechanisms perform similarly for these simulations. Therefore one may use the auxiliary mechanism constructed for the MCM for the SAPRC mechanisms as well.
Table 2. Tuned auxiliary mechanism used for EXACT campaign.
Process / Tuned reaction rateNO2 = HONO / 0.7 x 10-5 s-1
NO2 = wHNO3 / 1.6 x 10-5 s-1
O3 = wO3 / 3 x 10-6 s-1
Mechanism Evaluation
The SAPRC07, SAPRC11 and SAPRC11A mechanisms were evaluated against data collected during the EXACT campaign for the benzene experiments (high NOx and low NOx) obtained from Dr. Pilling of the University of Leeds. This data set contains information regarding the time events for adding reactants and opening the chamber roof, reactant and product information (e.g., benzene, phenol and formaldehyde), as well as nitrogen-containing species (NO, NO2 and nitric acid) and OH and HO2 concentrations (measured hydroxyl concentrations agreed well with calculated [OH] from the rate of decay of the aromatic).
SAPRC07
The low and high NOx benzene experiments conducted for the EXACT campaign were simulated using the SAPRC07 mechanism (Carter, 2010). The benzene scheme for SAPRC07 is listed in Table 3, for which the AFG1 and AFG2 species represent the highly photoreactive mono-unsaturated dialdehydes and aldehyde-ketones, and AFG3 represents the less photoreactive unsaturated diketones and di-unsaturated dicarbonyls. For the low NOx experiment, the rate of ozone formation during the simulation was slower than in the experiment, with a modeled O3(max) 7.5% lower than the experimental data (Figure 1). Modeled NO2 peaked 45 minutes later and 28% lower than the measured NO2 in the EUPHORE chamber, and modeled NO did not decrease as fast as in the chamber experiment (Figures 2 and 3). In the case of the HOx species, SAPRC07 underpredicts HO2 by 37%, and OH by 56% (Figures 4 and 5).
Bloss et al. (2005) reported a peak concentration of 22.4 ppb for glyoxal in their benzene experiment. SAPRC07 predicted a peak value of 14.0 ppb. However, SAPRC07 has a yield of 0.29 glyoxal in the BENZENE + OH reaction. Volkamer et al. (2005) report a yield of glyoxal to be 32% +/- 5% for this reaction, with negligible contribution from secondary glyoxal formation pathways. Therefore the yield of GLY was adjusted to 0.32
Table 3. Benzene scheme for the SAPRC07 mechanism.
Rate Parameters / Reactants / ProductsA = 2.33e-12
Ea = 0.38 / BENZENE + OH = / #.116 OH + #.29 {RO2C + xHO2} + #.024 {RO2XC + zRNO3} + #.57 {HO2 + CRES} + #.116 AFG3 + #.290 xGLY + #.029 xAFG1 + #.261 xAFG2 + #.314 yRAOOH + #-.976 XC
k(300) =
7.40e-11 / AFG1 + OH = / #.217 MACO3 + #.723 RO2C + #.060 {RO2XC + zRNO3} + #.521 xHO2 + #.201 xMECO3 + #.334 xCO + #.407 xRCHO + #.129 xMEK + #.107 xGLY + #.267 xMGLY + #.783 yR6OOH + #.284 XC
k(300) =
9.66e-18 / AFG1 + O3 = / #.826 OH + #.522 HO2 + #.652 RO2C + #.522 CO + #.174 CO2 + #.432 GLY + #.568 MGLY + #.652 xRCO3 + #.652 xHCHO + #.652 yR6OOH + #-.872 XC
Phot Set= AFG1 / AFG1 + HV = / #1.023 HO2 + #.173 MEO2 + #.305 MECO3 + #.500 MACO3 + #.695 CO + #.195 GLY + #.305 MGLY + #.217 XC
k(300) =
7.40e-11 / AFG2 + OH = / #.217 MACO3 + #.723 RO2C + #.060 {RO2XC + zRNO3} + #.521 xHO2 + #.201 xMECO3 + #.334 xCO + #.407 xRCHO + #.129 xMEK + #.107 xGLY + #.267 xMGLY + #.783 yR6OOH + #.284 XC
k(300) =
9.66e-18 / AFG2 + O3 = / #.826 OH + #.522 HO2 + #.652 RO2C + #.522 CO + #.174 CO2 + #.432 GLY + #.568 MGLY + #.652 xRCO3 + #.652 xHCHO + #.652 yR6OOH + #-.872 XC
Phot Set= AFG1 / AFG2 + HV = / PROD2 + #-1 XC
k(300) =
9.35e-11 / AFG3 + OH = / #.206 MACO3 + #.733 RO2C + #.117 {RO2XC + zRNO3} + #.561 xHO2 + #.117 xMECO3 + #.114 xCO + #.274 xGLY + #.153 xMGLY + #.019 xBACL + #.195 xAFG1 + #.195 xAFG2 + #.231 xIPRD + #.794 yR6OOH + #.938 XC
k(300) =
1.43e-17 / AFG3 + O3 / #.471 OH + #.554 HO2 + #.013 MECO3 + #.258 RO2C + #.007 {RO2XC + zRNO3} + #.580 CO + #.190 CO2 + #.366 GLY + #.184 MGLY + #.350 AFG1 + #.350 AFG2 + #.139 AFG3 + #.003 MACR + #.004 MVK + #.003 IPRD + #.095 xHO2 + #.163 xRCO3 + #.163 xHCHO + #.095 xMGLY + #.264 yR6OOH + #-.575 XC
Figure 1. Ozone concentrations for the benzene experiment – EXACT Campaign: measured versus modeled. The blue diamonds are observed values with 5% (2 error bars, and pink squares are predictions from SAPRC07.
Figure 2. NO2 concentrations for the benzene/low NOx experiment. Measurements (with 10% error bars) in blue diamonds and SAPRC07 predictions in pink squares.
Figure 3. NO concentrations for the benzene/low NOx experiment. Measurements in blue diamonds and SAPRC07 predictions in pink squares.
Figure 4. HO2 concentrations for the benzene/low NOx experiment. Measurements in blue diamonds and SAPRC07 predictions in pink squares.
Figure 5. OH concentrations for the benzene/low NOx experiment. Measurements in blue diamonds and SAPRC07 predictions in pink squares.
for the BENZENE + OH reaction. This gave a predicted peak GLY of 16.9 ppb. In an attempt to match model predictions to observations, the yield of GLY was then adjusted
to 0.45, the upper limit of the glyoxal yield reported by Volkamer et al. 2001. This generated a predicted peak concentration for glyoxal of 22.3 ppb, well within the experimental error of the observed peak in glyoxal concentration for the benzene experiment. However, because Volkamer et al. (2001) state that observed secondary glyoxal formation from benzene oxidation was negligible, GLY yields from the oxidation of AFG1, AFG2 and AFG3 were lowered to 0.001. Combined with a 0.45 yield of primary glyoxal from the initial benzene oxidation, this resulted in a predicted glyoxal peak concentration of 20.4 ppb, still within experimental error of the observed peak value.
There are three routes for glyoxal degradation in the atmosphere: reaction with OH, reaction with NO3 (only important in the dark) and photolysis (see Table 4). (There are 2 photolysis channels for glyoxal in SAPRC07 rather than 3 as is recommended by NASA/JPL (Sander et al., 2006). The third channel, GLY + HV = H2 + #2 CO, has a low quantum yield compared to the other two channels, so its absence is not significant in SAPRC07.). While the GLY + OH reaction products are not in agreement with the current IUPAC recommendation, this is a relatively unimportant reaction with little impact on ozone or PAN formation in the benzene/NOx degradation scheme. One photolysis channel for glyoxal leads to formation of formaldehyde (HCHO). Figure 6 shows the effects of varying glyoxal yields on HCHO formation for the benzene-low NOx experiment: higher GLY yields result in HCHO predictions that are closer to observed values. Figure 7 illustrates the predicted and measured O3 concentrations for each GLY yield described above. With higher GLY yields, the ozone formation rate is closer to observed values. Figures 8 and 9 show the OH and HO2 concentrations for each yield of glyoxal described above, with increasing glyoxal yields correlating with increasing OH and HO2 which more closely match observations.
Table 4. Degradation scheme for glyoxal in SAPRC07.
Rate Parameters / Reactants / Productsk(300) = 1.10e-11 / GLY + OH = / #.63 HO2 + #1.26 CO + #.37
RCO3 + #-.37 XC
A = 2.80e-12 Ea = 4.722 / GLY + NO3 = / HNO3 + #.63 HO2 + #1.26 CO
+ #.37 RCO3 + #-.37 XC
PF=GLY-07R / GLY + HV = / #2 {CO + HO2}
PF=GLY-07M / GLY + HV = / HCHO + CO
Product information for phenol, catechol, and nitrophenol are also available for the benzene/low NOx experiment. In SAPRC07, the model species CRES contains the grouping of phenols and cresols, so this species was compared to the sum of the catechol and phenol concentrations observed in the EUPHORE chamber (see Figure 10 for the low NOx simulation). SAPRC07 underpredicted the sum of catechol and phenol by 37% at its peak concentration. The reason for this underprediction is unclear, as the phenol yield from the benzene plus hydroxyl reaction is in agreement with literature values, and the catechol yield measured in the EXACT campaign is a small fraction (~2%) of the total phenol + catechol yield. In the case of nitrophenol, SAPRC07 (which contains the model species NPHE which represents all nitrophenols) overpredicts nitrophenol by a factor of 3.6 (see Figure 11). One possible reason for this overprediction is the yield of BZO in the CRES + OH reaction is 0.2, while the Master Chemical Mechanism v3.1 (for which the aromatic scheme is based upon the EXACT campaign) has a BZO yield of 0.06 (Coeur-Tourneur et al. 2006 measured the yield of 6-methyl-2-nitrophenol 4.7±0.8% from o –cresol). Therefore the CRES + OH reaction in SAPRC07 was adjusted to give a yield of 0.06. As a result, the NPHE peak concentration lowered by 25%, to 4.9 ppb, a factor of 2.5 higher than observed.