Unexpected quenching effect on new particle formation from the atmospheric reaction of methanol with SO₃

Significance

New particle formation (NPF) is an important global phenomenon, contributing nearly half of the cloud condensation nuclei in nature. Today, NPF is believed to be mainly promoted by low-volatile species formed in atmosphere. Herein, we show that in certain cases, the formation of low-volatile species could undermine NPF. Specifically, we identify previously unreported catalytic reactions between alcohols and SO₃ which yield low-volatile organic sulfates. Rather than being a promoter to NPF, the low-volatile organic sulfates can compete for consuming SO₃, thereby disfavoring H₂SO₄ formation. Such unexpected quenching effects on NPF are most likely to occur in dry and polluted regions with abundant alcohols, illustrating the importance in understanding the interplay between nucleation precursor formation and subsequent NPF.

Abstract

Despite the high abundance in the atmosphere, alcohols in general and methanol in particular are believed to play a small role in atmospheric new particle formation (NPF) largely due to the weak binding abilities of alcohols with the major nucleation precursors, e.g., sulfuric acid (SA) and dimethylamine (DMA). Herein, we identify a catalytic reaction that was previously overlooked, namely, the reaction between methanol and SO₃, catalyzed by SA, DMA, or water. We found that alcohols can have unexpected quenching effects on the NPF process, particularly in dry and highly polluted regions with high concentrations of alcohols. Specifically, the catalytic reaction between methanol and SO₃ can convert methanol into a less-volatile species––methyl hydrogen sulfate (MHS). The latter was initially thought to be a good nucleation agent for NPF. However, our simulation results suggest that the formation of MHS consumes an appreciable amount of atmospheric SO₃, disfavoring further reactions of SO₃ with H₂O. Indeed, we found that MHS formation can cause a reduction of SA concentration up to 87%, whereas the nucleation ability of MHS toward new particles is not as good as that of SA. Hence, a high abundance of methanol in the atmosphere can lower the particle nucleation rate by as much as two orders of magnitude. Such a quenching effect suggests that the recently identified catalytic reactions between alcohols and SO₃ need to be considered in atmospheric modeling in order to predict SA concentration from SO₂, while also account for their potentially negative effect on NPF.

Numerous species have been shown to participate in the formation of new atmospheric aerosol particles from vapor via nucleation. However, the prevailing nucleation mechanisms often underestimate the rates of new particle formation (NPF) (1–3), suggesting that there are still many unidentified nucleating species in the atmosphere, as well as undiscovered nucleation mechanisms. Alcohols, which are released by plants, biomass combustion, and industrial emission, are abundant in the troposphere (4–9). However, their contribution to NPF was thought to be negligible due to weak hydrogen-bonding interactions (10–12) between alcohol molecules and known important nucleation precursors such as sulfuric acid (SA) and dimethylamine (DMA) (13, 14).

Many recent studies have suggested that atmospheric species generated from gas-phase chemical reactions can play an important role in the NPF process (15–19). Alcohols are active species in the atmosphere, and they can be involved in many chemical reactions (20–22). Shen et al. (23) showed fluorescence evidence that methanol (MO) can react with SO₃ with a relatively high rate constant, although the reactant concentrations used in the laboratory were much higher than those in the ambient environment, while little water was present in the experiment. The reaction between MO and SO₃ has received increasing attention recently due to two reasons: 1) A product of the reaction––methyl hydrogen sulfate (MHS)––is less volatile than MO, which may stabilize the atmospheric clusters and promote NPF (23, 24). 2) The reaction can cause appreciable consumption of SO₃, which would lower the abundance of SA [produced via the reaction of SO₃ with H₂O (25–28)] in the atmosphere. Since SA is the most important nucleation precursor, lower SA concentration would have an impact on the nucleation rate. Such a quenching effect originating from specific atmospheric reactions was recently recognized as a key factor influencing the number of aerosol particles, as shown from measured mixtures of atmospheric vapors (2). Hence, the reaction between SO₃ and MO may play either a positive or a negative role in NPF, depending on the relative importance of the two opposing effects elucidated above. Besides the potential implications on NPF, the MHS produced by the reaction of MO with SO₃ can have negative health effects, e.g., irritating eyes and the upper respiratory tract.

Currently, emissions from power plants is considered the primary source of MHS (29), although secondary sources are less explored. In addition to detection near power plants, MHS is also widely detected near city freeways (30), as well as in rural areas (31), with atmospheric concentration reaching up to 5.0 × 10⁷ molecules cm⁻³. For example, during the smog episode in Los Angeles (32), very high atmospheric concentration of MHS of up to 1.1 × 10¹¹ molecules cm⁻³ was reported. Hence, exploration of the atmospheric secondary source of MHS and the role of chemical formation of MHS in NPF is of fundamental importance. An improved understanding of the MO–SO₃ reaction and the corresponding chemical formation mechanism of MHS are also important to assess the effects of abundant small alcohols on NPF. Here, the MO–SO₃ reaction in the ambient condition was studied by using high-level density-functional theory and Born–Oppenheimer molecular-dynamics simulations. In addition, the combined effects of this reaction and the subsequent nucleation process are evaluated using the Atmospheric Cluster Dynamics Code (ACDC) (33, 34).

Results and Discussion

Reaction Mechanism of MO with SO₃.

The gas-phase MO–SO₃ reaction is investigated at the coupled cluster with singles, doubles, and perturbative triples [CCSD(T)]-F12/cc-pVDZ-F12//M06-2X/6–311++G(3df,3pd) level of theory. In the presumed reaction (without catalyst) of MO–SO₃ (Fig. 1A), the oxygen atom of MO can react with the sulfur atom of SO₃ to form an ester, followed by simultaneous proton transfers from MO to SO₃. The Gibbs free-energy barrier of this reaction is 20.93 kcal mol⁻¹, while the corresponding effective rate coefficient (1.20 × 10⁻¹⁹ cm³ molecule⁻¹ s⁻¹, see SI Appendix, Table S1) is approximately six orders of magnitude lower than the value measured experimentally [(1.17 ± 0.16) × 10⁻¹³ cm³ molecule⁻¹ s⁻¹, 295 ± 2 K) (23)], indicating that a direct reaction between MO and SO₃ is unlikely to occur under atmospheric conditions.

Fig. 1.

Gibbs free-energy (in kcal mol⁻¹ at 298.15 K) profiles for the gas-phase reaction of MO and SO₃ at the CCSD(T)-F12/cc-pVDZ-F12//M06-2X/6–311++G(3df,3pd) level of theory. (A) uncatalyzed, (B) MO-catalyzed, (C) MHS-catalyzed, (D) SA-catalyzed, and (E) H₂O (W)-catalyzed gas-phase reaction between MO and SO₃. The structures of the corresponding reactants (R), transition states (TSs), and products (P) are also shown. The gray, white, red, and yellow balls represent C, H, O, and S atoms, respectively.

Notably, the hydroxyl group of the reactant MO and the sulfate ester group of the product (MHS) can act as both hydrogen-atom donors and -acceptors to promote proton transfer, suggesting that MO and MHS may be able to catalyze the MO–SO₃ reaction. As shown in the computed energy profiles (Fig. 1 B and C), the free-energy barriers of the reactions with either catalyst are lowered significantly, to 3.06 kcal mol⁻¹ (MO) and 2.68 kcal mol⁻¹ (MHS), respectively. Under the experimental concentrations ([MO] = 4.3 × 10¹⁴ molecules cm⁻³, [SO₃] = 3 × 10¹² molecules cm⁻³), the effective rate coefficient for the MO-catalyzed reaction is 5.23 × 10⁻¹² cm³ molecule⁻¹ s⁻¹, which is close to the experimental value of 1.17 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹. This result thus explains the experimental findings and provides a possible reaction pathway for the MO–SO₃ in the atmosphere. The details of the effective rate coefficient for the MHS catalyzed reaction are shown in SI Appendix, Part 1 and Table S1.

Important common atmospheric species (SA, DMA, and H₂O) have been detected in the atmosphere (14). These species can also act as relatively strong hydrogen-atom donors/acceptors, thereby promoting various proton transfer reactions (35, 36), and possibly catalyzing the MO–SO₃ reaction. Therefore, the catalytic effects of SA/DMA/H₂O on the MO–SO₃ reaction were also investigated. As shown in Fig. 1D, the free-energy barrier of the reaction catalyzed by SA is 2.15 kcal mol⁻¹, which is even lower than that with MO or MHS as catalysts. The reaction catalyzed by DMA is expected to be barrierless since our Born–Oppenheimer molecular-dynamics (BOMD) simulation shows that the MO–SO₃ reaction catalyzed by DMA (SI Appendix, Fig. S1A) can occur in approximately 1 ps. Lastly, the Gibbs free-energy barrier for the H₂O-catalyzed MO–SO₃ reaction is 6.62 kcal mol⁻¹. Since the atmospheric concentration of H₂O is several orders of magnitudes higher than that of SA or DMA, the MO–SO₃ reaction catalyzed by H₂O may still be appreciable despite its relatively high free-energy barrier.

As shown in Table 1, H₂O is actually the most effective catalyst for the MO–SO₃ reaction. Furthermore, the rate of the MO–SO₃ reaction catalyzed by H₂O is comparable to that of the H₂O–SO₃ reaction (commonly viewed as the dominant pathway by which SO₃ is consumed), especially under relatively dry (e.g., winter) and highly polluted conditions with low H₂O concentrations and high MO concentrations (see the yellow and red areas in Fig. 2, and more details are shown in SI Appendix, Part 2 and Tables S2–S4). Overall, the MO–SO₃ reaction can be viewed as an important secondary source of atmospheric MHS.

Table 1.

Concentrations of the catalysts ([M], molecules cm⁻³), the effective reaction rate coefficients (k_eff, cm³ molecule⁻¹ s⁻¹), and the reaction rates (v, molecules cm⁻³ s⁻¹) for the formation of MHS from the MO–SO₃ reaction catalyzed by M (M = SA, DMA, or H₂O) calculated at the CCSD(T)-F12/cc-pVDZ-F12//M06-2X/6–311++G(3df,3pd) level of theory and at 298.15 K

M	[M], molecules cm−3	keff, cm3 molecule−1 s−1	v, molecules cm−3 s−1
Uncatalyzed	—	1.20 × 10−19	1.20 × 10−3
SA	1.0 × 107	9.42 × 10−20	9.42 × 10−4
DMA	1.0 × 109	2.47 × 10−14	2.47 × 102
H2O	1.0 × 1015	8.50 × 10−12	8.50 × 104

[SO₃] = 1.0 × 10³ molecules cm⁻³, [MO] = 1.0 × 10¹³ molecules cm⁻³.

Fig. 2.

Ratios of reaction rates (v₁/v₂) for the MO–SO₃ reaction catalyzed by H₂O (v₁) and the H₂O–SO₃ reaction catalyzed by H₂O (v₂) at varying MO and H₂O concentrations ([MO], [H₂O], molecules cm⁻³) in the atmospheric concentration range.

Note that gas-phase reactions catalyzed by water molecules in the atmosphere is frequently proposed in the literature (27, 37–40). However, the equilibrium constants of hydrate formation between most potential reactant molecules and water (37) are usually very low due to their low binding energies, rendering the catalytic gas-phase reactions unimportant in the atmosphere. SO₃ is an exception due to its high water affinity. Indeed, many gas-phase reactions involving SO₃ can be efficiently catalyzed by water (27, 38).

Note also that the reactions between SO₃ and other important atmospheric species bearing alcoholic hydroxyl groups, such as ethanol, glycolaldehyde, isopropanol, and glycolic acid, are also studied (SI Appendix, Table S5 and Fig. S2). The Gibbs free-energy barriers of these H₂O-catalyzed reactions are in the range of 4.54 to 7.24 kcal/mol, and these energy-barrier values are close to that of the H₂O-catalyzed MO–SO₃ reaction. Hence, the identified reaction mechanism with SO₃ appears to be a general one and can be applicable to other species with alcoholic hydroxyl groups.

Opposing Roles of MO–SO₃ Reaction in the Nucleation Process.

The above calculations show that the MO–SO₃ reaction could feasibly occur under the catalysis of H₂O, which might be an important secondary source of MHS in the atmosphere. Since the produced MHS can form stronger hydrogen bonds than the reactant MO (see the molecular electrostatic potential analysis in SI Appendix, Part 3 and Fig. S3), and the atom-in-molecule topological analysis for the most stable clusters involving MHS shows that MHS can form relatively strong hydrogen bond with SA and DMA (SI Appendix, Part 4 and Table S6), the MO–SO₃ reaction may have implications for atmospheric aerosol nucleation. In many previous studies, chemical reactions involved in nucleation precursor formation and the subsequent nucleation process are usually considered and modeled as two independent stages. If the competitive chemical reactions that are involved in nucleation precursor formation are not considered, the concentrations of the precursors would be constant, for example, the concentration of SA. In this scenario, the participation of MHS into SA-DMA–based NPF would enhance the particle formation rate (SI Appendix, Fig. S4), a prediction similar to the previously reported enhancement effect of SFA on SA-DMA–based NPF (41). However, if the chemical reactions involved in nucleation precursor formation are considered, the MO–SO₃ reaction effectively competes with the H₂O–SO₃ reaction for consuming SO₃, thus lowering the concentration of SA in atmosphere.

The decrease in the SA concentration and the clustering processes involving MHS need to be simultaneously modeled in order to understand the realistic role of the MO–SO₃ reaction in atmospheric nucleation. To this end, the combined process was simulated (Fig. 3; further details are given Part 5 in SI Appendix): 1) The competitive chemical reactions involved in the formation of potential nucleation precursors (MHS formation from the MO–SO₃ reaction and SA formation from H₂O–SO₃ reaction), and 2) the subsequent clustering process involving MHS, SA, and DMA. Since accurate atmospheric concentration of SO₃ is still unclear due to the highly reactive nature of SO₃, the simulated system starts from the reaction between SO₂ and ·OH (SI Appendix, Fig. S5 and Table S7), providing a constant source of SO₃ (SI Appendix, Fig. S6) for accurately evaluating the competing formation of MHS and SA.

Fig. 3.

Simulated processes. The brown arrows represent the competitive nucleation precursor formation processes, and the black arrow represents the clustering process.

As shown in Fig. 4, at moderate atmospheric concentrations of SO₂ (5.0 × 10¹⁰ molecules cm⁻³) and ·OH (5.0 × 10⁶ molecules cm⁻³) (42–45), the presence of MO up to 1.0 × 10¹³ molecules cm⁻³ [i.e., abundant MO in highly polluted regions (7, 9)] can produce significant amounts of MHS, but simultaneously suppresses SA production by 21∼87% as [H₂O] varies from 1.0 × 10¹⁶ to 1.0 × 10¹⁵ molecules cm⁻³, corresponding to relatively dry regions (SI Appendix, Table S8). The SA concentration is almost unaffected by the presence of MO in relatively humid areas for [H₂O] > 1.0 × 10¹⁶ molecules cm⁻³ (SI Appendix, Fig. S7). The degree of [SA] reduction with the increase of MO concentration is approximately equal at different concentrations of SO₂ and ·OH (SI Appendix, Fig. S8). Therefore, the influence on SA concentration is important only in dry and polluted areas. It is noteworthy that in practical applications, the SA concentration is often directly predicted from SO₂ concentrations (44, 46), without accounting for other reactions competing for SO₃. Hence, taking the competing MO–SO₃ reaction into account in such calculations would improve the prediction of the SA concentrations.

Fig. 4.

MHS concentration ([MHS], solid lines) and SA concentration ([SA], dashed lines) (molecules cm⁻³) predicted when accounting for both the MO–SO₃ reaction and the H₂O–SO₃ reaction at different H₂O concentrations ([H₂O], 1.0 × 10¹⁵, 5.0 × 10¹⁵, and 1.0 × 10¹⁶ molecules cm⁻³) versus the MO concentration at 280 K. [SO₂] = 5.0 × 10¹⁰ molecules cm⁻³ and [·OH] = 5.0 × 10⁶ molecules cm⁻³.

Although MHS can enhance the rate of SA-DMA–based NPF, the formation reaction of MHS can also decrease the concentration of the most important nucleation precursor, SA. As such, the effects of the MO–SO₃ reaction on the particle nucleation rate (J) were further investigated by considering simultaneously the relevant formation reactions (SO₂-·OH, MO–SO₃, H₂O–SO₃), and the subsequent clustering process, under the realistic atmospheric concentrations and temperature. As shown in Fig. 5A, J decreases by up to two orders of magnitude as [MO] increases from 1.0 × 10¹⁰ molecules cm⁻³ to 1.0 × 10¹³ molecules cm⁻³ (within the range of atmospheric concentrations) (4–9). The reduction in J can be attributed to the [SA] reduction, and to the fact that MHS has a weaker ability to form hydrogen bond compared to SA, as illustrated by the molecular electrostatic potential analysis (SI Appendix, Part 3 and Fig. S3). Fig. 5B shows the nucleation rate when the sum ([SA]+[MHS]) is kept constant. Again, J decreases as the ratio of the MHS concentration to the sum of [SA] and [MHS] (i.e., [MHS]/([SA]+[MHS])) increases, which further indicates that MHS is a relatively weaker nucleation precursor than SA.

Fig. 5.

Particle formation rates (J, cm⁻³ s⁻¹). (A) J with varying MO concentrations under different H₂O concentrations (1.0 × 10¹⁵, 5.0 × 10¹⁵, and 1.0 × 10¹⁶ molecules cm⁻³) at 280 K. [DMA] = 1.0 × 10⁹ molecules cm⁻³, [SO₂] = 5.0 × 10¹⁰ molecules cm⁻³, and [·OH] = 5.0 × 10⁶ molecules cm⁻³. (B) J with varying ratios of [MHS]/([SA] + [MHS]) at 280 K. [DMA] = 1.0 × 10⁹ molecules cm⁻³, [SA] + [MHS] = 1.0 × 10⁷ molecules cm⁻³.

As shown in SI Appendix, Fig. S9, the number concentration of large clusters containing MHS and DMA molecules is approximately one order of magnitude lower than that of large clusters containing SA and DMA molecules. In addition, when comparing clusters with the same numbers of acid molecules and the same numbers of base molecules (with MHS counted as an acid), the Gibbs free energies of formation and evaporation rates (SI Appendix, Table S9) for most of clusters including MHS ((SA)_m·(DMA)_n·(MHS)_l, 0 ≤ m + l ≤ 3, and 0 ≤ n ≤ m + l) are higher than those of the corresponding pure SA-DMA–based clusters [(SA)_m+l·(DMA)_n, 0 ≤ m + l ≤ 3, and 0 ≤ n ≤ m + l], also indicating that the stabilities of most clusters including MHS are lower than those of pure SA-DMA–based clusters. As a result, the nucleation promoted by MHS (MO–SO₃) cannot offset the nucleation suppressed by the [SA] reduction (H₂O–SO₃). Therefore, the competitive formation reactions of nucleation precursors under atmospheric conditions (concentration and temperature) need to be considered when evaluating their overall roles in nucleation events.

To our knowledge, MHS is one case where competing formation reactions of nucleation precursors show a marked quenching effect on NPF. Other sulfates formed from atmospheric abundant and larger species with alcoholic hydroxyl groups and many other functional groups tend to be much less volatile than H₂SO₄, and thus could potentially promote rather than quench the NPF. In our previous study of the formation of SFA (due to NH₃–SO₃ reaction) on NPF, the competitive formation of SFA and SA were not considered (41) because the quenching effects of the NH₃–SO₃ reaction on the SA concentration and the SA-DMA particle formation rate are negligible (SI Appendix, Tables S10 and S11). In other words, not every competitive formation process of precursor has a strong quenching effect on SA-DMA–based atmospheric aerosol nucleation––the presence or absence of quenching depends on the details of the concentrations and properties of the relevant chemical species under the realistic atmospheric conditions.

Nucleation Mechanism of SA-DMA-MHS–Based Clusters.

Finally, the nucleation mechanism of SA-DMA-MHS–based clusters at different [H₂O] is investigated (details are shown in Part 3 of SI Appendix). As shown in Fig. 6, MO can compete with H₂O for SO₃ to form MHS. Overall, the nucleation involves two pathways: 1) The SA-DMA nucleation pathway and 2) the SA-DMA-MHS nucleation pathway. In both pathways, the (SA)₁·(DMA)₁ cluster is initially formed from monomers of SA and DMA. In the SA-DMA nucleation pathway (blue arrows), SA-DMA clusters grow by the stepwise addition of either SA or DMA, as shown in previous studies (41, 47). In the SA-DMA-MHS nucleation pathway (green arrows), MHS combines with the (SA)₁·(DMA)₁ cluster first, and then this cluster grows by the stepwise addition of SA or DMA. In addition, MHS can also combine with a larger (SA)₂·(DMA)₂ cluster. As shown in the inset table of Fig. 6, the SA-DMA-MHS pathway contributes to overall particle formation by 25% at [H₂O] = 1.0 × 10¹⁶ molecules cm⁻³ and by 53% at lower [H₂O] = 1.0 × 10¹⁵ molecules cm⁻³. Therefore, the effect of MHS becomes remarkable in dry and polluted areas with relatively low H₂O concentration and high MO concentration, e.g., in winter in northern China. In such conditions, the extensive participation of MHS in the nucleation of atmospheric aerosol may have serious environmental effects.

Fig. 6.

Main formation pathways of clusters at 280 K. The green arrows represent the SA-DMA-MHS nucleation pathway, and the blue arrows represent the SA-DMA nucleation pathway. [DMA] = 1.0 × 10⁹ molecules cm⁻³, [MO] = 1.0 × 10¹³ molecules cm⁻³, [SO₂] = 5.0 × 10¹⁰ molecules cm⁻³, and [·OH] = 5.0 × 10⁶ molecules cm⁻³. The contributions to particle formation rates (J, cm⁻³ s⁻¹) at different H₂O concentrations are indicated in the inset table.

Conclusion

In conclusion, we identified a catalytic reaction that was previously overlooked in the atmosphere, namely, the chemical reaction between methanol and SO₃, catalyzed by SA, DMA, and water molecules. This reaction leads to the gas-phase formation of MHS, thereby providing a potential secondary source of MHS in the atmosphere. The possible strong quenching effect of methanol on atmospheric nucleation, especially in dry areas with relatively high concentrations of alcohols, is also reported. We show that MHS can play two opposing roles in NPF: the quenching effect due to competing formation reactions of nucleation precursors and the promotion of the subsequent nucleation processes. In either case, both MHS itself and the active participation of MHS in atmospheric aerosol nucleation can be harmful to human health. Finally, the identified reaction mechanism with SO₃ appears to be generalizable to other species involving alcoholic hydroxyl groups. Considering the two opposing effects of these reactions with SO₃ may be important not only to accurately predict SA concentrations from the source of SO₂ but also to model the effect of these species on atmospheric NPF.

Methods

The molecular structures of the reactants, products, and the transition states (TSs) were all optimized using the M06-2X (48, 49) functional with the 6–311++G(3df,3pd) (50) basis set. The vibrational frequencies were calculated to confirm the absence of imaginary frequencies in all stable structures, and the TSs possess only one imaginary frequency (SI Appendix, Table S12). In addition, to confirm that each TS connects the reactant to the corresponding product, intrinsic reaction coordinate calculations were performed. Single-point energy calculations were carried out at the CCSD(T)-F12/cc-pVDZ-F12 (51–53) level of theory based on the optimized geometries at the M06-2X/6–311++G(3df,3pd) level of theory using the ORCA 3.0.3 program package (54, 55). Detailed calculations regarding the effective reaction rate coefficients are shown in Part 1 of SI Appendix. Details of BOMD simulations are given in Part 6 of SI Appendix.

To search for the lowest-energy structures of the clusters considered, the ABCluster (56, 57) program was employed to generate initially guess structures. Combinations of different levels of theory was used to further optimize the structures (see details in Part 6 and SI Appendix, Fig. S10 and Table S13). All of the quantum-chemistry calculations were carried out with the Gaussian 09 (revision A.01) program package (58).

The ACDC (33, 34) was used to simulate the cluster formation process with structural, thermodynamic, and kinetic data generated by quantum-chemistry calculations as the inputs. The time dependence of the concentration of each cluster was determined by integrating the birth–death equations (33) numerically using the ode15s solver in the MATLAB-R2013a program (59) (see details in Part 6 of SI Appendix).

Data Availability.

All data are available in the manuscript and in SI Appendix.