Size-Resolved Chemical Composition of Sub-20 nm Particles from Methanesulfonic Acid Reactions with Methylamine and Ammonia

: Particle formation in the atmosphere from gas-phase precursors has been observed around the world; however, our fundamental understanding of the key species responsible and mechanisms involved remains uncertain. Recent laboratory studies demonstrated that acid − base reactions involving methanesulfonic acid (CH 3 SO 3 H, MSA) and small alkylamines may contribute signi ﬁ cantly to new particle formation. To date, most of the investigations have been focused on particle number concentration and size distribution measurements in combination with quantum chemistry predictions of the most stable clusters. Here, we present the ﬁ rst measurements of the size-resolved chemical composition of sub-20 nm particles generated in a custom ﬂ ow reactor from the reaction of MSA with methylamine (MA) in the presence or absence of NH 3 using thermal desorption chemical ionization mass spectrometry (TDCIMS). A novel design of the TDCIMS inlet was evaluated, and the measurement of the chemical composition of particles was extended down to 5 nm in diameter, the smallest size yet reported for this method. MSA − MA particles with diameters smaller than 9 nm were found to be more acidic, with an acid/base molar ratio of 1.8 ± 0.4 (1 σ ) for 5 nm particles compared to the larger particles which were neutral. A similar acid/base molar ratio trend was observed when NH 3 was added to the MSA − MA combination. In the MSA − MA − NH 3 system, the MA/NH 3 molar ratio was higher than 1 (up to 2.6) for all particle sizes despite the much larger concentration of NH 3 in the gas phase (the gas-phase MA/ NH 3 ratio was ∼ 0.23), indicating that MA is a key component in particle formation from this system. The potential reasons for this based on previous calculations of small clusters in this system are discussed.


INTRODUCTION
−23 Competing with condensational growth is coagulation with pre-existing particles, 24,25 which is significant for sub-20 nm particles.Thus, understanding all of these processes is crucial for addressing the uncertainty associated with not only aerosol−cloud−climate interactions but also particle effects on visibility and health.
−3,26−34 While most studies to date have focused on H 2 SO 4 −ammonia/amine nucleation systems, 2,[28][29][30]34 recent investigations have demonstrated that methanesulfonic acid (CH 3 SO 3 H; MSA) can also efficiently initiate NPF in the presence of small alkylamines and water.31−33,35−38 In air, MSA is one of the oxidation products of organosulfur compounds such as dimethyl sulfide, which is widely emitted from biological ocean processes, agriculture, industry, and domestic activities. 39−45The ambient concentration of gas-phase MSA (∼10 5 to 10 7 molecules cm −3 ) can be of the same order of magnitude as that of H 2 SO 4 .46−49 In addition, the newly formed particles measured in a boreal forest 50 and in the Finnish Arctic 51 were observed to be enriched in MSA.Particulate MSA was also observed to correlate with NPF occurring in the summertime in the Arctic.52−54 Furthermore, with the current decreasing trend of anthropogenic SO 2 emissions 55−60 and the increasing rate at which polar sea ice is melting, which gives rise to increasing phytoplankton productivity and as a consequence higher dimethyl sulfide and MSA concentrations, 61−64 NPF initiated from MSA acid− base chemistry may become more important in the future.55 Ammonia and amines are important atmospheric bases and originate from a variety of sources, including animal husbandry, biomass burning, urban settings, sewage and solid waste treatment, forest floor, and marine environments.65−75 Although the concentrations of amines are typically much smaller than that of ammonia, amines have been shown both theoretically and experimentally to be much more efficient than NH 3 in NPF with both H 2 SO 4 and MSA.31,32,34,76−80 Amines were also found to replace NH 3 in H 2 SO 4 −NH 3 and MSA−NH 3 clusters and particles.,78,85−88 For example, Perraud et al. 87 demonstrated that the presence of NH 3 enhanced NPF from the acid/base-initiated chemistry between MSA and trimethylamine (TMA) up to 6 orders of magnitude. As a result, this system was extremely efficient at nucleating and growing particles to detectable sizes, while the corresponding MSA + TMA binary system was not efficient on its own. Aditionally, only a modest effect was observed for the MSA + MA system, which is already extremely efficient.Last, amines and ammonia are often found along with MSA in particles around the world.50,89−93 For instance, Lawler et al.50 detected both MSA and alkylamines in 20−70 nm nanoparticles in the Hyytialab oreal forest during a nucleation event.
An improved understanding of NPF from acid−base chemistry requires direct measurements of the size-resolved chemical composition of particles formed, especially over the diameter range from ∼2 to 20 nm.Such measurements are challenging because of the small mass loadings, low signal-tonoise ratio, and the need to remove the interferences from both larger particles and gas-phase species.To date, two approaches have shown success at measuring size-resolved, sub-10 nm diameter particle composition.The nano aerosol mass spectrometer, 94,95 equipped with a digital ion trap reflection time-of-flight mass detector, allows the elemental composition of individual nanoparticles to be measured down to 7 nm.Thermal desorption chemical ionization mass spectrometry (TDCIMS) 96−100 has been shown to determine the size-resolved composition of particles with diameters down to 6 nm with molecular information and a limit of detection of several picograms.
Recently, size-resolved nanoparticles formed from laboratory-controlled H 2 SO 4 reactions with NH 3 and dimethylamine 101,102 and HNO 3 + dimethylamine systems 103 have been studied using TDCIMS to gain insights into chemical composition through their acid/base ratios.Results from the Cosmics Leaving Outdoor Droplets (CLOUD) study indicated that 10−15 nm particles from H 2 SO 4 + dimethylamine + NH 3 were acidic but became neutral after growing above 15 nm. 101,104A similar trend was observed in laboratory flow reactor experiments examining H 2 SO 4 + dimethylamine and H 2 SO 4 + NH 3 systems where particles smaller than 12 nm were not fully neutralized in the presence of excess gas-phase base. 102Chee et al. 103 conducted nucleation and growth studies by reacting a monoprotic acid, nitric acid (HNO 3 ), with dimethylamine and found that the particles were nearly neutral above 9 nm (smaller particles were unable to be measured for this system).Here, we have extended these measurements to another acid−base system of atmospheric relevance, the reaction of MSA with MA, in the presence or absence of NH 3 .TDCIMS equipped with a new high-resolution time-of-flight mass spectrometer (LTOF mass analyzer, Tofwerk AG) was used to detect the size-resolved chemical composition of sub-20 nm particles, extending the diameter range down to 5 nm.A new feature of the instrument is the application of soft X-rays as the ionization source and the use of a new ionization reagent for the detection of NH 3 and amines.This new implementation was essential to the success of the present study, and some analytical aspects are described.From the TDCIMS measurements, the size-resolved acid/base molar ratios were determined, and possible growth mechanisms are discussed.

EXPERIMENTAL METHODS
2.1.Flow Reactor Description.A 1 m borosilicate glass flow reactor, described in detail elsewhere, 35,36,87 was used to produce nanoparticles from the reaction of MSA with MA in the presence or absence of NH 3 .The flow reactor was operated with two configurations illustrated in Figure 1 and described below.The sequence of addition of reactants was as follows: 4.2 L min −1 of dry clean air was injected in the first ring (ring A); 0.2 L min −1 of gas-phase MSA mixed with 1.8 L min −1 of dry clean air was added through the second spoke (spoke 2); and 0.2 L min −1 of gas-phase MA mixed with ∼1.0 L min −1 of dry clean air was added through the third spoke (spoke 3) through holes that face upstream.In MSA + MA only experiments (i.e., in the absence of NH 3 ), an additional 1.0 L min −1 of dry clean air was introduced through ring C. In experiments with NH 3 , the 1.0 L min −1 of air flow was moved from ring C down to spoke 1 and increased to 1.5 L min −1 , to which 0.5 L min −1 of gas-phase NH 3 was added.The stream of the NH 3 /air mixture was introduced radially so that when MSA (spoke 2) was introduced into the flow reactor, it would react with both MA and NH 3 at the same time.The total flow rates under these conditions were 8.4 L min −1 (MSA + MA only) and 9.4 L min −1 (in the presence of NH 3 ).A purge air generator (Parker−Balston; model 75-62) provided the dry clean air used in all experiments.Further details of dry clean air purification can be found in the Supporting Information.All experiments were carried out at 1 atm and 298 K and under dry conditions with the relative humidity maintained at <3% as indicated by a humidity probe (Vaisala; model HMT 838).All flows were controlled by high-precision mass flow controllers (Alicat or MKS) and were periodically confirmed with a flow calibrator (Sensidyne; Gilibrator 2).The flow reactor was cleaned regularly with nanopure water (18.2MΩ cm; Thermo Scientific, Barnstead; model 7146) and dried with dry clean air with the water jacket set at 343 K.The flow reactor was conditioned with a flow of gas-phase MSA for at least 2 days after cleaning prior to an experiment.
Particle size distributions and chemical composition measurements were performed, as described below, using a movable stainless steel sampling line (o.d.0.635 cm) located inside the flow reactor along the centerline of the reactor and placed at distances ranging from 3 to 43 cm away from spoke 2 (i.e., the MSA addition port).Based on a conversion factor determined in previous studies, 35 these distances correspond to residence times ranging from 0.64 to 9.1 s for the MSA + MA system and 0.59 to 8.5 s for the MSA + MA + NH 3 system.For simplicity, the shorter residence time is hereafter referred as ∼0.6 s and the longest residence time as ∼9 s for both systems.Note that the residence time reported here does not account for the time spent in the sampling line (∼1.1 s) as NPF and growth are expected to cease because of the losses on the walls of the sampling line.
2.2.Reactants.Gas-phase MSA was generated by passing 0.2 L min −1 of dry purge air over the pure liquid (Sigma-Aldrich, ≥ 99%) contained in a glass trap maintained at room temperature and introduced into the flow tube through spoke 2. Commercially available permeation tubes (VICI Metronics) containing pure NH 3 or MA were used to generate the gasphase bases.Each permeation tube was enclosed inside a Ushaped glass tube.The flow of air over the MA permeation tube was 0.2 L min −1 , while that of NH 3 was 0.5 L min −1 .To ensure a constant temperature, the MSA glass trap and the amine permeation tubes were immersed in a water bath maintained at room temperature (T = 293 K).The concentration of each species introduced into the flow reactor was measured routinely (details of the measurements can be found in the Supporting Information).The initial concentrations after dilution in the flow reactor for MSA, MA, and NH 3 (when present) were (2.1−2.3)× 10 11 molecules cm −3 (8.4−9.3 ppb), (1.3−1.4) × 10 11 molecules cm −3 (5.3−5.8 ppb), and 5.8 × 10 11 molecules cm −3 (23.5 ppb), respectively.These concentrations represent upper limits as they do not account for wall losses.Note that higher concentrations than ambient of reactants were required in these studies in order to fulfill the limits of detection of TDCIMS, allowing the characterization of the chemical composition of these particles with reasonable collection times.
2.3.Particle Size Distribution Measurements.Particle size distributions were continuously measured using a scanning mobility particle sizer (SMPS) consisting of a 0.071 cm impactor nozzle, a 210 Po radioactive source (NRD LLC; model P-2021), a TSI model 3080 electrostatic classifier equipped with a nanodifferential mobility analyzer (nano-DMA; TSI, Inc.; model 3085), and a butanol-based ultrafine condensation particle counter (UCPC; TSI, Inc.; model 3776).The sheath air flow was set to 15 L min −1 (recirculating mode), and the aerosol flow was 1.5 L min −1 , which provided measurements of the size distributions over the mobility diameter range of 2−64 nm.
2.4.Size-Resolved Online Chemical Composition Measurements.−102 Briefly, the sample (3 L min −1 ) was passed first through a unipolar charger (UPC) 100,105 to generate negatively charged particles that were subsequently sizeselected using a nano-DMA running in low resolution mode 106 with a sheath flow of 10 L min −1 (sheath-to-aerosol flow ratio ∼3.3).Instead of recirculating the sheath gas in the nano-DMA, dry clean air was used as the sheath flow to prevent the accumulation of gas-phase MSA, MA, or NH 3 .The particles were then directed to an electrostatic precipitator where a high voltage (+3 kV) was applied to a Pt filament to collect the particles.During collection, the filament was flushed with dry clean air to prevent the capture of gas-phase species.After collection for 1−720 min, depending on the aerosol mass concentration, the filament was translated into the ion source, where it was resistively heated using a temperature ramp up to ∼600 °C to desorb the particle components.Note that the wall of the desorption area was maintained at 50 °C during the measurement to reduce condensation of the evaporated material to the walls.The desorbed species were then ionized via chemical ionization using a soft X-ray source (Hamamatsu Photonics Ltd; photoionizer model L9490, emitting at 3−9.5 keV).The resulting ions were analyzed by an LTOF.
TDCIMS can be performed in both positive and negative ion modes: MSA was measured in the negative ion mode with (H 2 O) n O 2 − (n = 0−2) as the reagent ion, while both NH 3 and MA were measured in the positive ion mode with either (H 2 O) n H 3 O + (n = 0−2) alone or added N-methylpyrrolidone (NMP; C 5 H 9 NO; MW = 99 g mol −1 ).We previously showed 107 that NMP exhibits good sensitivity and selectivity for measuring NH 3 and amines via the formation of [M + NMP + H] + and [M + 2NMP + H] + ions (where M represents either MA or NH 3 ); this is its first application as an ionization reagent in TDCIMS.Note that the soft X-ray source illuminated the entire ionization volume of the ion source.Although chemical ionization should be the main ionization pathway, the possibility that some of the ions detected by the LTOF are produced by direct ionization by the soft X-rays cannot be ruled out.No evidence for additional chemistry was observed.
An entire TDCIMS measurement cycle included four steps: (1) cleaning mode: the filament was set to the home position and was resistively heated up to 600 °C for 1 min to clean any residues on the filament; (2) cooling mode: the filament was cooled down to room temperature for ∼30 s; (3) collection mode: as described above (1−720 min); and (4) analysis mode: the filament was moved from the collection position into the ionization source where it was resistively heated by the application of a power-controlled ac current for a total duration of 70 s.Each collection cycle was followed by a background cycle where no collection voltage was applied to the Pt filament.Background-subtracted spectra, where the spectra from the background cycle (i.e., when no particles are collected on the filament) are subtracted from the sample spectra (i.e., when size-selected particles have been collected on the filament and subsequently thermally desorbed and analyzed by the mass spectrometer), were used for quantification.Last, an "exhaust SMPS" consisting of a nano-DMA and a UCPC (model 3025, TSI, Inc.) was used to detect the particles downstream of the electrostatic precipitator.The particle mass collected on the filament was estimated from these exhaust SMPS measurements by differencing the particle size distributions obtained during the collection and background cycles.
New aspects of the instrument included the switch from nitrogen to dry clean air as the carrier gas throughout the instrument and the use of a soft X-ray source in place of the 210 Po radioactive source to generate the reagent ions (O 2 − , H 3 O + , and [NMP + H] + ).Thus, some fragmentation and ionization patterns with this new design may be different from the previous TDCIMS experiments and were remeasured in this study.
The size distributions of the particles generated in the flow tube were stable over days (Figure S1).This made it possible to explore the size-resolved chemical composition over long collection time periods.The shortest (∼0.6 s) and longest (∼9.0 s) residence times from both experiments were selected in order to produce the largest span of particle sizes, the latter being an indicator of nanoparticle age.The chemical composition of 5.2, 6.8, 9.8, 13.1, 16.6, and 19.5 nm particles was measured for each condition.The collected mass on the filament from all of the experiments in this study was between 0.16 and 4.5 ng, with collection times ranging from 1 to 720 min.For particles smaller than 5.2 nm, the sample mass was too small to be detected with 720 min collection time.For particles larger than 19.5 nm, the uncertainties increase significantly because of low particle number concentration, low filament collection efficiency, and multiply charged particles produced by the UPC.
2.5.Atomized Reference Particles.In order to determine the accurate TDCIMS acid/base signal ratio between MSA and MA and NH 3 , reference MSA−MA and MSA−NH 3 particles were generated using a constant output atomizer (TSI, Inc.; model 3076).Aqueous solutions comprising 0.5 mM equimolar concentrations of MSA and MA or MSA and NH 3 were atomized using dry purge air (36 psi) as the carrier gas.For these, liquid standards of MSA (Sigma-Aldrich; ≥ 99%), MA (Sigma-Aldrich; 40% w/w in water), and NH 3 (Aldrich; 28.7% w/w in water) were used.The flow exiting the atomizer was 2 L min −1 and was passed through two diffusion driers (TSI, Inc.; model 3062-NC) containing silica gel beads (Delta Adsorbents; 4 × 8 mesh).The flow of dry particles was then diluted with an additional 3 L min −1 of dry, clean air before being sampled by the SMPS operated in the low flow mode (i.e., aerosol flow rate, 0.3 L min −1 ; sheath flow rate, 3.0 L min −1 ) as well as by the TDCIMS.Polydisperse atomized MSA−MA particles were additionally measured using a sequential spot sampler, 108,109 followed by ion chromatography (Dionex) and liquid chromatography mass spectrometry [ultra-performance liquid chromatography−electrospray ionization−tandem mass spectrometry (UPLC−ESI−MS/MS), Waters, Quattro Premier XE].The details of these measurements can be found in the Supporting Information.

RESULTS AND DISCUSSION
3.1.Nanoparticle Size Distributions in the Presence or Absence of NH 3 .Figure 2 displays the size distributions for MSA + MA (Figure 2a) and MSA + MA + NH 3 (Figure 2b) particles generated in the flow tube at ∼0.6 and ∼9 s residence times, respectively.Under these conditions, a large number of sub-20 nm particles formed quickly, as indicated by the size distributions observed at ∼0.6 s in both systems, with a total number concentration (N total ) of (4.1 ± 0.03) × 10 6 and (5.5 ± 0.2) × 10 6 particles cm −3 , respectively.The particle geometric mean diameter (GMD) in both experiments was ∼6 nm at 0.6 s residence time (Figure 2d).Note that previous laboratory studies have demonstrated that MSA + MA is an efficient acid/base system for nucleating particles, 31,32,35,87,110 and the intent of this particular investigation was primarily the generation of a range of particle sizes in order to collect enough mass within a reasonable sampling time for sizeresolved TDCIMS measurements.Evolution of the size distribution as a function of residence time in the flow tube is presented in the Supporting Information (Figure S2 a,b).In the MSA + MA system, N total increased slightly from ∼0.6 to 2 s (Figure 2c) and then remained constant, indicating that either the limiting reactant is consumed quickly and/or the reaction is complete.The distribution of particles collected at 9 s had a GMD of 7.4 nm (Figure 2d).
Upon the addition of NH 3 (5.8× 10 11 molecules cm −3 ), NPF was systematically enhanced compared to the MSA + MA system at all residence times, with N total about double that observed in the absence of NH 3 at ∼9 s residence time.This enhancement is consistent with that observed in a previous study. 87Evolution of N total as a function of time (Figure 2c) shows that NPF was occurring continuously up to ∼8 s residence time in the presence of NH 3 .As can be seen in Figure 2d, at early residence times, the GMD was typically smaller for the MSA + MA + NH 3 system compared to the binary MSA + MA system which, combined with the increase in N tot , suggests that NH 3 promotes NPF.However, at longer residence times, in the presence of NH 3 , the GMD was slightly larger with a GMD of 8.2 nm at ∼9 s residence time.At earlier stages of growth, the concentration of gaseous precursors/ clusters is high enough to support the formation of new particles as well as the growth of newly formed particles.At later stages of growth, gaseous precursors/clusters are depleted, slowing down both nucleation and growth of particles.Note that the MSA + NH 3 nucleation system was much less efficient on its own in forming particles, with N total of only 3.2 × 10 5 particles cm −3 (Figure S3) measured at ∼9 s compared to 5.5 × 10 6 particles cm −3 for the binary MSA + MA system (Figure 2c) and 1.05 × 10 7 particles cm −3 for the MSA + MA + NH 3 system (Figure 2c).These results are consistent with the previous reports of synergy initiated by NH 3 in binary H 2 SO 4 + amine 77,78 and binary MSA + amine systems. 87.2.Size-Resolved Chemical Composition Measurements by TDCIMS.In those, the spectra from the "background" cycle (i.e., when no particles are collected on the filament) were subtracted from the "sample" spectra (i.e., when size-selected particles have been collected on the filament and subsequently thermally desorbed and analyzed by the mass spectrometer).As a result, the reagent ions are subtracted out, which yields background-subtracted spectra where the ions corresponding to the component of the particles (MA or MSA) are clearly visible.The corresponding raw spectra (before background subtraction) are displayed in Figure S4.(b) Corresponding desorption profiles of the ions related to MA, NH 3 and MSA during the "analysis mode" and the filament desorption current profile (gray trace), which is directly related to the filament temperature.
(m/z 32), [MA + NMP + H] + (m/z 131), and [MA+2NMP + H] + (m/z 230), produced from the reaction of MA with (H 2 O) n H 3 O + , [NMP + H] + , and [2NMP + H] + , respectively.The [MA + NMP + H] + adduct was the most abundant signal in the background-subtracted integrated spectra, about 2−3 times more intense than the [MA + H] + ion.For NH 3 , the dominant signal was also the [NH 3 + NMP + H] + adduct at m/z 117, with a signal about 15 times greater than that of the NH 4 + ion.No other ions were observed in the positive ion mode (a typical raw positive ion mass spectrum is shown in Figure S4a).
In the negative ion mode (Figure 3a, bottom red trace; Figure S4b), the dominant ion in the background-subtracted integrated spectra observed for MSA was the deprotonated parent CH 3 SO 3 − ion at m/z 95, followed by five additional minor ions, which included SO  112 In earlier studies, 113,114  − ion was attributed to the potential OH chemistry occurring in the ion source. 111The distribution of the different ions was similar across all the flow reactor experiments.
Figure 3b shows the current applied to the filament (gray line) and the thermal desorption profiles for ions related to MA, NH 3 and MSA during the 70 s analysis period in a typical MSA + MA + NH 3 experiment.For the first 10 s, no current was applied to the filament while it remained at room temperature.During the next 35 s, the current increased linearly from 0 to 3.4 amps, and the temperature increased to an estimated ∼600 °C.The temperature remained at its highest level for another 15 s.Finally, the current was removed from the filament, and the temperature quickly fell back to room temperature.Each signal intensity was close to zero before heating the filament, indicating that there is no evaporation for any of the species collected on the filament at room temperature.When the temperature rose, NH 3 was the first species to evaporate from the particles, followed by MA, and finally MSA.This is in accordance with the saturation vapor pressures (P sat ) of each component: P sat (NH 3 ) = 9.8 atm; 115 P sat (MA) = 3.5 atm; 112 and P sat (MSA) = 7.4 × 10 −7 atm. 116The signal returned to zero at the end of the desorption period, indicating that no residue was left on the filament after a heating cycle.The quick and thorough desorption of the species is an advantage for quantification.The subsequent background measurements (no current applied to the filament) were very clean, indicating that the contribution from the gas phase was negligible and there was no significant residue on the filament.
As described in detail in the Supporting Information (Figure S5), the background-subtracted integrated signal observed for MA, NH 3 , and MSA from MSA−MA−NH 3 particles collected (D p = 19.5 nm) at the end of the flow reactor was linear for a collection time up to 10 min for MA and MSA and for a longer time period for NH 3 (40 min).For higher mass loading, depletion of the reagent ions occurred.When the reagent ions were not depleted, the good linearity across the species indicated that the collection time did not significantly influence their detection and that the evaporation of MA, NH 3 , and MSA from the sampled particles was negligible during the collection period, which allowed quantitative measurements of the particle acid/base molar ratios.
3.2.2.Size-Resolved Acid/Base Molar Ratios.The relative sensitivity of TDCIMS for MSA, MA, and NH 3 was first investigated using atomized MSA−MA and MSA−NH 3 particles generated from equimolar MSA + MA and MSA + NH 3 solutions.Surprisingly, ammonium ions were systematically detected by TDCIMS in particles formed by atomizing an equimolar solution of MSA + MA.Independent measurements using the sequential spot sampler also exhibited high amounts of NH 4 + for those particles (Figure S6a).From measurements by both the thermal desorption chemical ionization mass spectrometer and the spot sampler, MA/NH 3 molar ratios were estimated to range from ∼2 to 4. Two studies previously reported the presence of ammonium from atomized aqueous solutions of dimethylamine and H 2 SO 4 . 101,117In these cases, it was suspected that it was due to the presence of trace amounts of NH 3 in either the nanopore water 77 used to prepare the solution or the gas supply.Another possibility is that the trace amount of NH 3 from room air permeated through the Teflon tubing containing the atomized particles.Lawler and coworkers 101 excluded the formation of NH 4 + inside TDCIMS by analyzing the aqueous H 2 SO 4 -dimethylamine solution itself, which exhibited no ammonium ions.Similarly, in the present studies, the aqueous solution used in the atomizer did not contain any detectable ammonium by IC (Supporting Information Figure S6b), although ammonium was systematically measured in particles formed by atomizing the same solution.The resulting MSA/MA molar ratio for those particles was estimated from the IC and UPLC−MS measurements to be 1.5, and the acid/base signal ratio (or sensitivity factor) recorded by TDCIMS for the MSA + MA atomized particles (F MSA/MA ) was corrected for that (see the Supporting Information for details).The final relative sensitivities (Figure S7) were 6.6 ± 1.1 for MSA/MA (F MSA/MA ) and 99 ± 15 for MSA/NH 3 (F MSA/NH 3 ).These two sensitivities were then used to convert the measured acid/ base signal ratios for particles formed in the flow tube from MSA + MA and MSA + MA + NH 3 reactions systems into acid/base molar ratios as details in the Supporting Information.
Figure 4 shows the size-resolved acid/base molar ratios in the MSA + MA nucleation from gas-phase precursors measured at the shortest (∼0.6 s) and longest (∼9 s) residence times.At ∼0.6 s, the acid/base molar ratio for particles with diameters (D p ) of 5.2 and 6.8 nm was 1.9 ± 0.2 (1σ) and 1.4 ± 0.1 (1 σ), respectively, while for larger particles (D p > 9 nm), it was close to unity.This indicates that the particles were acidic at 5.2 and 6.8 nm (i.e., contained more acid molecules than base) but became neutral as they grew to 9.8 nm.A similar trend was found for particles collected at ∼9 s residence time, indicating a similar growth mechanism over the diameter range.
Figure 5a shows the size-resolved acid/base molar ratios for the MSA + MA + NH 3 system.At ∼0.6 s residence time, the trend with particle size is similar to that for the MSA + MA system, with smaller particles (D p < 7 nm) being more acidic and larger particles exhibiting a molar acid/base ratio close to unity.Note that the particles at 19.5 nm appear to be more acidic at 0.6 s but not at 9 s.As shown in Figure 5b, the MA/ NH 3 molar ratio measured at 0.6 s was 1.1 ± 0.3 (1σ) for 5.2 nm particles and increased with diameter to reach ∼2.5 for particles larger than 10 nm.The particles measured at the longer residence time (9 s) exhibited a similar increasing trend with diameter.Note that the particle phase MA/NH 3 ratio (∼2) is much larger than the gas-phase ratio between the two bases (∼0.23), indicating that MA is the key component in NPF, although NH 3 has a synergistic effect as reported recently. 87.2.3.Discussion.A summary of the enthalpies of formation of each cluster determined in previous quantum chemical calculations 31,118 for a series of MSA−MA clusters is given in Figure 6.In this figure, only selected growth paths are represented based on those previous studies, where thick arrows correspond to the most favorable paths by which a cluster can grow (i.e., paths a, b, d, and e), and thin arrows (path c, f, and g) correspond to less favorable routes.Calculations of dissociation energies, enthalpies, and free energies have indicated that (MSA−MA) 4 is a major intermediate in the growth process.110,118 This cluster, which approaches 1 nm in size, was found to be thermodynamically stable with calculated dissociation-free energies (ΔG) of 66.8 kcal mol −1 for the (MSA−MA) 4 → 4 (MSA−MA) reaction and 20.3 kcal mol −1 for the (MSA−MA) 4 → 2 (MSA−MA) 2 reaction.The lifetime of the (MSA−MA) 4 cluster is greater than 100 picosecond (ps) at T = 300 K. 118 The same study indicated that the growth of the cluster via addition of MSA to (MSA−MA) 4 is favorable, with a Gibbs free energy of −10 kcal mol −1 for the addition reaction, 118 while that for MA addition was less favorable, with a Gibbs free energy of −5.3 kcal mol −1 for the addition reaction.Thermodynamic simulations performed at T = 300 K indicated that MSA does not evaporate from the MSA−(MSA−MA) 4 cluster in 100 ps, while MA can dissociate from MA−(MSA−MA) 4 cluster in just 3.5 ps.118 The incorporation of a second MSA molecule to the initial (MSA−MA) ion pair, forming a MSA−(MSA−MA)   cluster, was also previously predicted to be favorable (ΔH = −29 kcal/mol and ΔG = −15 kcal/mol).31 The mechanism for MA to be taken up into particles is via acid/base interactions 119 with an MSA molecule either present in the gas phase as a free species forming an ion pair, or neutralizing a cluster enriched in MSA. Howver, ΔG for the addition of MA to the MSA−(MSA−MA) cluster, resulting in full neutralization, is relatively high, between −2 and +5 kcal/ mol depending on the point of addition and the cluster structure (path c in Figure 6).31 This one-on-one growth mechanism (favorable addition of the acid, followed by a higher energetic barrier for the addition of a base, i.e., growth of the (MSA−MA) cluster following path b and then path c) was previously demonstrated to occur in the H 2 SO 4 + NH 3 system.In the latter case, similar to the MSA−MA clusters, a lower energy barrier was observed for the addition of an acid molecule to an existing cluster (close to collision limited uptake), followed by a much higher barrier for the addition of the base to yield neutralization.120,121 All these calculations are consistent with the prevalence of small acidic clusters.
The growth of the (MSA−MA) 4 unit cluster from collision with another cluster (path g in Figure 6) was found to be somewhat favorable (ΔG = −5.7 kcal/mol), where the (MSA− MA) 4 −(MSA−MA) 4 cluster is thermodynamically stable for at least 100 ps at T = 300 K. 118 More recent simulations on MSA−MA clusters using the Atmospheric Cluster Dynamics Code 122 support the findings from these earlier studies, where neutral clusters and clusters with higher numbers of MSA than MA were observed to be the most stable.Together, these calculations are qualitatively consistent with the observed higher acidity of the smallest particles in the MSA−MA system and the neutral state of the larger particles.These observations parallel experiments that examined H 2 SO 4 acid−base reactions, where the growth mechanism was shown to be not only driven by H 2 SO 4 addition (i.e., monomer−cluster collision) but also driven by cluster−cluster collisions when high concentrations of gas-phase bases are present. 26,27,123he influence of NH 3 on various MSA−base binary systems, 87 including MSA + MA, was recently investigated using quantum simulations, which show that the (NH 3 ) 2 − (MSA−MA) 2 cluster does not show any significant changes in its general structure compared to the (MSA−MA) 2 cluster.In both, the proton transfer is from the MSA to MA; however, NH 3 is shown to stabilize the cluster, consistent with the enhancement of the particle number concentration observed in the current study (Figure 2c).In this case, the preference of the proton for MA follows the gas-phase basicity, which is higher for MA (864.5 kJ mol −1 ) compared to that of NH 3 (819 kJ mol −1 ). 124Thermodynamic modeling studies indicated that the (MSA−MA) cluster exhibits a proton transfer between the acid and the base, forming a stable ion pair that could interact with an existing cluster as described above.This, however, is not the case for the (MSA−NH 3 ) cluster, which is only stabilized by van der Waals forces that are much weaker. 118,125lenius et al. 126 demonstrated that H 2 SO 4 −NH 3 clusters grow via stepwise addition of the acid, followed by the base, whereas H 2 SO 4 −dimethylamine clusters, where dimethylamine binds more strongly to the acid, grow via collision with other acid− base clusters.This is consistent with MA being much more effective than NH 3 in nucleation and growth of new particles in our experiments.This is also consistent with the observation of high MA/NH 3 molar ratios (∼2) in the particles, despite the concentration of gas-phase NH 3 being ∼4 times higher than that of MA in the experiments.

CONCLUSIONS
Reactions of MSA with MA efficiently lead to the formation of sub-20 nm particles.The size-resolved chemical composition of the nanoparticles measured with TDCIMS shows that the smallest particles (D p < 9 nm) are more acidic, while the acid/ base molar ratio approaches unity (neutral) as particles grow to larger sizes.This trend in acidity is consistent with the addition of MSA molecules to the neutral clusters being favorable thermodynamically. 31,118On the other hand, condensation of MA or NH 3 is comparatively less favorable, and these species are expected to partition back to the gas phase.Salt formation (either forming acid−base ion pair clusters in the gas phase or via neutralization by reactive uptake of the amine onto an acidic particle) can explain the larger neutral particles.The trend for MSA−amine systems is similar to previous observations reported for H 2 SO 4 + dimethylamine, H 2 SO 4 + dimethylamine + NH 3 , and H 2 SO 4 + NH 3 systems. 101,102,104inally, the addition of NH 3 to the MSA + MA system enhances particle formation.In these reactions, although the concentration of gas-phase NH 3 was about 4 times higher than that of MA, the concentration of MA was about 2 times higher in the particles.These studies expand our understanding of acid−base-initiated NPF and illustrate the importance of combining experimental and theoretical approaches to better understand on a molecular level the mechanism of particle formation and growth in the atmosphere.

* sı Supporting Information
The Supporting Information is available free of charge at h t t p s : / / p ub s .a c s. o r g / d o i / 1 0 .1 0 2 1 / a c se a r th s p a c echem.0c00120.
Additional details on the experimental methods; stability of the size distribution of particles measured in the flow reactor over time; size distributions for the MSA + MA and MSA + MA + NH 3 nucleation systems as a function of residence time; comparison of size distributions measured for the MSA + NH 3 , MSA + MA, and MSA + MA + NH 3 systems collected at ∼9 s; TDCIMS raw mass spectra obtained in the positive ion mode and negative ion mode; total integrated ion signal intensities for MA, NH 3 , and MSA as a function of collection time; independent analysis of the MSA−MA-atomized particles; size-independent signal ratio between MSA and MA, and between MSA and NH 3 , measured by TDCIMS for the atomized particles; and details on the determination of acid/base molar ratios from TDCIMS signal intensities (PDF).

■ AUTHOR INFORMATION
Corresponding Authors

Figure 1 .
Figure 1.(a) Schematic of the 1 m borosilicate flow reactor (adapted from ref 35).Flows were distributed as follows: ring A, 4.2 L min −1 (air only); spoke 2, 2.0 L min −1 (mixture of MSA diluted in air); and spoke 3, 1.2 L min −1 (mixture of MA diluted in air).For the MSA + MA system, an additional air flow (1.0 L min −1 ) was added through ring C, while for the MSA + MA + NH 3 system, the air flow was moved down to spoke 1 and mixed with the flow of NH 3 (total flow at spoke 1, 2.0 L min −1 ).(b) Simplified diagram of the sampling train including the SMPS and thermal desorption chemical ionization mass spectrometer apparatus.NMP = N-methylpyrrolidone; nano-DMA = nanodifferential mobility analyzer; UCPC = ultrafine condensation particle counter; and UPC = unipolar charger.
3.2.1.Ion Distributions Measured by TDCIMS and Their Thermal Desorption Profiles.Methylamine (MA) was detected in the positive ion mode as three different ions (Figure 3a, top blue trace), including [MA + H] +

Figure 3 .
Figure 3. (a) Representative background-subtracted integrated mass spectra in the positive ion mode (top blue trace) and negative ion mode (bottom red trace) summed over the desorption period for analysis of MSA−MA−NH 3 particles collected from the flow reactor.In those, the spectra from the "background" cycle (i.e., when no particles are collected on the filament) were subtracted from the "sample" spectra (i.e., when size-selected particles have been collected on the filament and subsequently thermally desorbed and analyzed by the mass spectrometer).As a result, the reagent ions are subtracted out, which yields background-subtracted spectra where the ions corresponding to the component of the particles (MA or MSA) are clearly visible.The corresponding raw spectra (before background subtraction) are displayed in FigureS4.(b) Corresponding desorption profiles of the ions related to MA, NH 3 and MSA during the "analysis mode" and the filament desorption current profile (gray trace), which is directly related to the filament temperature.

Figure 4 .
Figure 4. Size-resolved acid/base ratios determined for the particles in the MSA + MA system at ∼0.6 s (blue) and ∼9 s (red) residence times.The error bars represent 1 standard deviation from at least three repeated measurements.The horizontal dashed line represents an acid/base molar ratio of 1, corresponding to neutral particles.The initial concentrations of reactants are [MSA] = 2.3 × 10 11 molecules cm −3 (9.3 ppb) and [MA] = 1.4 × 10 11 molecules cm −3 (5.8 ppb).

Figure 6 .
Figure 6.Matrix representing potential growth pathways for selected MSA−MA clusters following the Gibbs free energy (ΔG) of the addition of MSA, MA, or a (MSA−MA) n cluster (n = 1, 2, or 4) based on previous quantum simulations (from refs 31 and 110).Note that the matrix only represents the mechanistic paths reported in those two studies.The thickness of the arrow represents the preference for the growth of the cluster, where a thick arrow represents a favorable route for the cluster to grow.
− with SO 3 in the analysis of H 2 SO 4 .However, because of the difference in their respective electron affinities (EAs), O 2 − ions are expected to react with SO 2 /SO 3 by charge transfer to form SO 2 − /SO 3 − ions [EAs for SO 2 (1.107 eV) and SO 3 (1.97eV) being higher than that of O 2 (0.451 eV)].
it was proposed that SO 2 − ions can immediately associate back with O 2 to form SO 4 − ions.By analogy to these studies, we propose that both ions observed here, SO 4 − and SO 5 − , correspond to [SO 2 +O 2 ] − and [SO 3 +O 2 ] − adducts, respectively.The minor HSO 4