Global impact of monocyclic aromatics on tropospheric composition

Abstract. Aromatic compounds are reactive species influencing ozone formation, OH concentrations and organic aerosol formation. An assessment of their impacts on the gas-phase composition at a global scale has been performed using a general circulation atmospheric-chemistry model. Globally, we found a small annual average net decrease (less than 3 %) in global OH, ozone, and NOx mixing ratios when aromatic compounds are included in the chemical mechanism. This inclusion of aromatics also results in CO mixing ratio increases, which cause a general decrease in OH concentrations. The largest changes are found in glyoxal and NO3, with increases in the atmospheric burden of 10 % and 6 %, respectively. Regionally, significant differences were found particularly in high NOx regime areas, with an increase of up to 4 % in O3 mixing ratios and 8 % in OH concentrations. NO3 increased by more than 30 % in several regions of the northern hemisphere, and glyoxal increased up to 40 % in Europe and Asia. Large increases in formaldehyde were found in urban areas. Although the relative impact of aromatics at the global scale is limited, at a regional level they are important in atmospheric chemistry.

. List of aromatic compounds included in this study and the respective annual emissions. This emissions are the same as in (Cabrera-Perez et al., 2016) but for higher aromatics.

Species
Emissions ( In this work a resolution of T63L31ECMWF was used, which corresponds to a horizontal reso- ). As a consequence, every simulation discussed here has identical meteorology (i.e. binary identical transport).

105
To analyze the influence of aromatic compounds on atmospheric composition, we performed a comparison between two scenarios. The baseline scenario, called REF scenario, excludes the emissions of aromatic compounds. The second scenario, called AROM scenario, includes all emissions from anthropogenic, biogenic, and biomass burning sources of the following aromatic compounds: benzene, toluene, xylenes (lumped), phenol, styrene, ethylbenzene, trimethylbenzenes (lumped), 110 benzaldehydes, and higher aromatics (as representative of aromatics with more than nine carbon atoms). Both scenarios are identical aside from emissions (in the baseline case there is no chemistry of aromatic species).
We used the Representative Concentration Pathways (RCP) inventory for anthropogenic emissions (van Vuuren et al., 2011), distributed vertically as in (Pozzer et al., 2009); the MEGAN model for TgC/yr are higher aromatics (the details of the emission factors used can be found in the supplement doi:10.5194/acp-0-1-2017-supplement). The atmospheric oxidation of aromatic compounds is performed by the MECCA sub-model (Sander et al., 2011). The complete description of the model setup-including emissions, the chemical mechanism used, and the evaluation of the AROM scenario-are included in Cabrera-Perez et al. (2016).

125
The products from the oxidation of aromatic compounds have reduced volatility, allowing them to partition into the aerosol phase and form SOA. This removal process of aromatic trace gases can significantly reduce the mixing ratios of the aromatic oxidation products. Since SOA formation is outside the scope of this work, additional channels in the chemical mechanism have been added to account for loss via SOA formation after the first oxidation step, using the yields from Ng et al. (2007).

130
The modifications in the chemical mechanism are described in the supplement (doi:10.5194/acp-0-1-2017-supplement). This approach avoids a possible overestimation of atmospheric concentrations of aromatic oxidation products but simultaneously limits the possible impact of aromatics products on the gas-phase chemistry.
3 Results/ Results discussion 135 Figure 2 shows the annual average mixing ratios of the sum of all aromatic compounds included in the numerical simulation. The mixing ratios are higher in continental areas and close to the surface.
The highest mixing ratios are found in East and South Asia, as well as in parts of Europe and the US, reaching up to ppb levels. The background mean mixing ratios in oceanic areas of the Southern Hemisphere are on the order of a few ppt.

140
In this section we compare the AROM scenario to the baseline REF scenario (i.e. AROM-REF).
3.1 Hydroxyl radical (OH)  On the upper right, surface OH relative difference between aromatic and no-aromatic scenarios expressed in %. On the lower left, OH absolute difference between mentioned scenarios. On the lower right, zonal relative differences (in %).
On the seasonal level, higher OH concentrations are found over continental areas during the winter and spring than in summer and autumn (see doi:10.5194/acp-0-1-2017-supplement). During the 165 daytime in northern hemisphere winter the relative increase in OH exceeds 50% (e.g. China and Europe). These differences correspond to absolute increases of approximately 20 × 10 4 mlc/cm 3 in China and West Asia, and of less than 6 × 10 4 mlc/cm 3 in the US and, Europe. In summer, Europe is the only region where OH concentrations increase when aromatics are included, although not exceeding 5%. The increase of OH production via NO + RO 2 through the increase of RO 2 concen-170 trations are due to aromatics oxidation. Although there is a ubiquitous decrease in NO x , this does not seem to limit OH formation. In the southern hemisphere and in oceanic areas the net effect of introducing aromatic compounds into the system results in a net depletion of OH due to the increase of CO as final product of the oxidation scheme (see Sect. 3.5.3). Figure   On the upper right, surface OH relative difference between aromatic and no-aromatic scenarios expressed in %. On the lower left, OH absolute difference between mentioned scenarios. On the lower right, zonal relative differences (in %). of the relative difference between both scenarios for the northern and southern hemispheres. We find 175 that the relative difference varies between -3.5% for winter months up to -1.5% in summer months.
The cycle is reversed for the southern hemisphere, where the changes in OH concentrations are due to transported species (mostly CO), making changes relatively homogeneous. Table 2 lists the OH concentrations for the REF and AROM scenarios, and the relative and absolute difference averaged globally in the boundary layer. We obtain an enhancement of 2-3% in the 180 atmospheric lifetime of methane, along with a decrease in OH concentration. Methane is a longlived species whose breakdown is driven by OH (Crutzen and Zimmermann, 1991); its lifetime is a measure of the OH concentration (Prather and Spivakovsky, 1990). The decrease in the OH For HO 2 a relative decrease of less than 1% in the atmospheric global burden is found. At the regional scale, only Europe and East Asia have relative increases in surface mixing ratios by more than 10%. The oxidation of carbon monoxide by OH is the main source of HO 2 (Lightfoot et al.,205 1992), serving as a buffer for OH; in addition, aromatic chemistry contains a large number of reactions leading to formation of HO 2 (e.g. the initial reactions of benzene + OH leads to Phenol + HO 2 ). The updated benzaldehyde photolysis used in this study was also more efficient than the one previously used, making it a significant source of HO 2 radicals. The small impact of aromatics observed on the HO x budget was expected, since this budget is well buffered against perturbations 210 (Montzka et al., 2011;Lelieveld et al., 2016).
Contrary to expectations, we found large increases of HONO mixing ratios in continental areas, generally with the sign of this change opposite to that of OH. These relative changes reached more than 50%, specifically in Africa, South-America, the Arabian Peninsula and South-East Asia. The reason for this pattern is explained by the photolysis of nitrophenols, which leads to HONO for-215 mation (Bejan et al., 2006;Cheng et al., 2009). In the AROM scenario, approximately 5.7% of the HONO formation is directly related to nitrophenol photolysis at surface. Nevertheless, the net effect on the atmospheric burden is a depletion of around 1.5%.

Ozone
Differences between AROM and REF of daytime annual averages of surface ozone mixing ratios are 220 shown in Fig. 4      The seasonal distribution of the relative differences (Fig. 6) shows lower amplitude than for OH, 235 but similar patterns. In the northern hemisphere, greater relative differences occurred between scenarios in the winter months than in the summer months; in the southern hemisphere this pattern is reversed. The relative differences range within 1%-2%. In the southern hemisphere a maximum difference of 2% is found over biomass burning activity regions.
Maximum increases of ozone of more than 10% occur over the northern hemisphere. In the US,

240
Europe, and China ozone mixing ratios can increase by more than 20%. Large peaks were also observed in Central Africa, due to strong biomass-burning events.
Relative changes in tropospheric ozone were found to be homogeneous within the northern hemisphere (Fig. 4, bottom right). There did not appear to be any differences between day and night, with decreases below 2% in the boundary layer and lower part of the free troposphere (up to 7 km).

245
Above 7km height, the relative differences decrease. Table 3 presents annual global mean ozone mixing ratios in the boundary layer for both scenarios-weighted by two different methods, but only mass weighted is shown, as both methods lead to the same results-as well as the relative and absolute differences. Both methods show good agreement in the mixing ratios. Relative differences are approximately 1%. During nighttime, relative 250 differences increase to 1.8%. Compared to OH, relative differences are lower for ozone, suggesting that ozone chemistry is less affected by aromatics than OH.
The simulation results show a weakening in the ozone formation; this result brings an interestingly mismatch with former studies (e.g. Butler et al. (2011)). Aromatics, especially toluene and xylenes, have significant ozone formation potentials among VOCs. It could therefore be expected an overall 255 increase in ozone formation when aromatics are introduced into the system. However, there are important differences between the work by (Butler et al., 2011)  is due to (i) the decrease in NO x mixing ratios -limiting ozone formation-, and to (ii) increasing radical production (OH, HO 2 , and RO 2 ) in ozone-depleting regimes, which enhances reactions of O 3 with HO 2 and OH. Growth in ozone mixing ratios is observed in regions of high NO x mixing ratios, where the limiting factor for ozone formation is hydrocarbon mixing ratios. The comparison between the AROM and the REF model results shows significant NO x depletion in central Africa, the Amazon forest, China, and Indonesia, with relative differences reaching approximately 10% (Fig. 7). In Europe, the US, and the Arabian Peninsula, reductions in NO x mixing ratios did not exceed 5%. Oceanic areas show small decreases, although NO x mixing ratios are sev-275 eral orders of magnitude lower than over continental territories, with absolute changes not exceeding 1 ppt. In the boundary layer, NO x mixing ratios decreased by 5% in the AROM scenario.

NO 3 and HNO 3
For annual average boundary layer NO 3 mixing ratios we find a global increase of 3% during nighttime and 6% during the daytime.
280 At the regional scale, the largest daytime changes occur in northern Africa, the Arabian Peninsula, and northern Asia, with increases of 20-30%. An increase of more than 30% was also observed over  the Tibetan plateau, although mixing ratios in this area are generally low. Decreases of up to 10% were observed in central Africa and in the Amazonian areas. Since NO 3 concentrations are low during daytime, these changes do not significantly affect atmospheric chemistry.

285
During nighttime, the difference between the scenarios are larger, with a 30% decrease in mixing ratios in central Africa and the Amazonian areas, an increase of more than 30% in the Tibetan plateau and northern Asia, and no change in northern Africa.
The net formation or depletion is defined by the competition between the amount of aromatic products consuming NO 3 versus the strength of the phenylperoxy channel, which leads to NO 3 for-290 mation. We investigated the importance of the phenylperoxy channel for O 3 and NO 3 mixing ratios.
The different channels, based on the work of Jagiella and Zabel (2007), were added to the current mechanism (Cabrera-Perez et al., 2016). Figure 8 shows the comparison for O 3 and NO 3 between the AROM scenario and an identical scenario without the phenoxy radical channels. Ozone mixing ratios increased as a result of neglecting the channel that transforms NO 2 into NO 3 , and more NO 2 is 295 therefore available for the catalytic process leading to ozone formation. The phenylperoxy channels quickly convert NO 2 into NO 3 (e.g. C 6 H 5 O 2 + NO 2 leads to C 6 H 5 O + NO 3 in the mechanism).
For NO 3 the areas with increases in the mixing ratios can be explained by the increases in ozone and NO 2 mixing ratios, leading to an increase in the NO 3 formation. These increases reach up to 30% in equatorial regions. In contrast, there are regions showing decreases by more than 30% in NO 3 300 mixing ratios; these decreases suggest a large strength of the phenylperoxy channels in certain areas of the northern hemisphere.  America. In Europe and the US, surface mixing ratios increased up to 6%. In oceanic areas changes 305 remain below 2%. The increases can be explained by the number of reactions leading to HNO 3 formation (e.g. the reactions of xylenes, trimethyl benzenes or ethylbenzene with NO 3 form HNO 3 ).

Formaldehyde
The main photochemical source of formaldehyde in the background troposphere is methane oxida-310 tion; in continental areas, VOC (including aromatic compounds) oxidation is the main source, and its main sink is reactions with OH. Comparing our two scenarios, we find a depletion in the formaldehyde surface mixing ratios (on an annual basis) in the Amazonian and central African regions-two areas that typically have higher formaldehyde mixing ratios (Figure 9). In contrast, we observed in-

Glyoxal
With respect to glyoxal, we find a global increase of 20% of the simulated mixing ratios at the surface. At the regional scale, China has the strongest absolute difference, with an increase of approximately 70 ppt (80%); followed by India and the Arabian Peninsula, where increases rise by more than 40 ppt. In Figure 10 ( eas; however, this large relative change is associated with very low mixing ratios (less than 1 ppt).
In continental areas over the northern hemisphere, increases of more than 10% are found. Relative increases are lower in the southern hemisphere than in the northern hemisphere, because the main source of glyoxal in the southern hemisphere is isoprene (Fu et al., 2008). Only in some regions of Africa and the Amazon a decrease (less than 2%) was observed; this was caused by a depletion Gg). This discrepancy is attributed to the missing mechanism of SOA formation, which has been estimated to account for approximately 20% of the total atmospheric sink (Stavrakou et al., 2009).

Carbon monoxide
The reaction chain in the oxidative process of aromatics produces carbon monoxide (CO), a relatively long-lived molecule (1-2 months). CO can travel long distances from its source, although this lifetime is not long enough to allow it to cross hemispheres (Daniel and Solomon, 1998). CO mixing ratios generally increase on the global scale, indicating a small addition to the carbon budget.

340
When comparing both scenarios, we find an increase of 3% in the atmospheric burden of CO, which corresponds to an increase of 14 Tg. The CO burden estimated by the model in the REF scenario is 546 Tg.

Sources of uncertainty
There are a number of sources of uncertainty in the estimation of the impact of aromatics on tro-345 pospheric chemistry. Firstly, emissions of aromatics have high uncertainties. One reason is that few databases provide anthropogenic speciation of aromatic VOCs. In the case of the RCP database (van Vuuren et al., 2008), speciation entails fractioning the total VOC flux into different species (a top-down approach) (Moss et al., 2008 Another source of error is due to the chemical oxidation mechanism (based on MCMv3.1), which in general overestimates peak ozone mixing ratios (e.g. by more than 15% in the case of toluene), and underestimates OH formation (up to 80%) and NO oxidation rates Bloss et al. (2005). In the case of OH, our version of the chemical mechanism includes HONO formation channels from nitrophenol 365 photolysis, which contributes to OH formation during daytime.
In the mechanism used in this work, the oxidation of benzene and toluene was taken from MCM.
For the rest of the aromatics that were included, the second oxidation products are directly linked to those of toluene. This approximation implies a less accurate representation of the oxidation of these species. Consequently, a comparison of simulation results with observations show relatively good At the global scale, OH concentrations decrease by 2-3% once aromatics are included. On the regional scale, areas with high levels of aromatics have decreases of more than 10%, while regions with large NO x mixing ratios show increases of up to 10%. This decrease in OH mixing ratios can 390 alter the VOC distribution. For example, the formaldehyde atmospheric burden decreases by 1.2%.
For ozone similar results to those of OH are found, with a global net decrease of 1%. However, the relative importance of aromatics at the regional scale can cause increases or decreases of more than 10% during the winter season.