Differences in the reactivity of ammonium salts with methylamine

The heterogeneous uptake of methylamine (MA) onto (NH4)2SO4, NH4HSO4, NH4NO3 and NH4Cl was investigated using a Knudsen cell reactor coupled to a quadrupole mass spectrometer, in situ Raman spectrometer and theoretical calculations. Exchange reactions were observed between MA and NH 4NO3, (NH4)2SO4, and NH4Cl were observed at 298 K. Simple acid-base reaction for MA was found taking place on NH 4 SO4. CH3NH3NO3 and CH3NH3Cl are not stable at low pressure and have higher dissociation vapor pressure than methylammonium sulfate. The observed uptake coefficients of MA on (NH 4)2SO4, NH4HSO4, NH4NO3 and NH4Cl at 298 K were measured to be 6.30±1.03×10−3, 1.78±0.36×10−2, 8.79±1.99×10−3 and 2.29±0.28×10−3, respectively. A linear free energy relationship was found for the heterogeneous reactions between MA and NH4Cl, (NH4)2SO4 and NH4NO3. Namely, the natural logarithm of uptake coefficients of MA on these ammonium salts is linearly related to the electrostatic potential of the H atom in the NH+4 group.


Introduction
Recent field measurements suggest that organic nitrogen species may be an appreciable fraction of organic aerosol mass (Beddows et al., 2004;Chen et al., 2010;Pratt et al., 2009;Smith et al., 2010;Wang et al., 2010b).Amines are an important class of organic nitrogen species, and are frequently detected in aerosols, rainwater, and fog water (Ge et al., 2011a).They are emitted into the atmosphere from a variety of sources including livestock, biomass burning, sewage treatment, meat cooking, automobiles, industrial processes, and marine organisms (Ge et al., 2011b).The typical atmospheric concentration of alkylamines is <1-14 nmol N m −3 , compared with ∼25 ppbv of ammonia (Cornell et al., 2003).
Recently, sources and sinks of atmospheric amines have received significant attention due to their potential toxicity to humans (Gong et al., 2004), their influence on atmospheric nitrogen cycle (Cornell et al., 2003) and their possible contribution to new particle formation (Smith et al., 2010;Wang et al., 2010a).Smith et al. (2010) found that aminium salts accounted for 23 % and 47 % of the observed new particle formation in Hyytiälä in 2006 and Tecamac in 2007, respectively.With a class of strong bases, the gaseous alkylamines may undergo acid-base reactions with acids such as H 2 SO 4 and HNO 3 to form salt particles.For example, Murphy et al. (2007) observed rapid particle nucleation when amines were injected into a chamber containing gaseous nitric acid.Wang et al. (2010b) also measured the obvious uptake of methylamine (MA), dimethylamine (DMA), and trimethyamine (TMA) into 59-82 wt% H 2 SO 4 .The uptake coefficients of these amines into H 2 SO 4 are in the range of 2.0×10 −2 -4.4×10 −2 .Some amines, such as TMA and triethylamine (TEA), also significantly contribute to secondary organic aerosol formation by NO 3 (Silva et al., 2008), OH or O 3 oxidation (Murphy et al., 2007;Gai et al., 2010).In addition, Wang et al. (2010c) proposed that a Mannich reaction between amines and carbonyl compounds probably leads to the formation of nitrogen containing species with high molecular weight in secondary organic aerosols.
In field measurements, alkylamines are usually internally mixed with sulfate and nitrate ions in aerosols (Facchini et al., 2008;Pratt et al., 2009;Smith et al., 2010).It should be noted, however, that ambient ammonia concentrations tend to be an order of magnitude greater than amine concentrations (Murphy et al., 2007), and the uptake coefficient of ammonia by H 2 SO 4 , approximately 1 (Liggio et al., 2011;Shi et al., 1999;Swartz et al., 1999), is also one or two orders of magnitude greater than that of amine.Conversely, the dissociation Published by Copernicus Publications on behalf of the European Geosciences Union.Y. Liu et al.: Differences in the reactivity of ammonium salts with methylamine constants of aminium nitrate are greater than or equal to that of NH 4 NO 3 (Murphy et al., 2007).Therefore, the formation of aminium salts by acid-base reactions in the presence of typical ambient ammonia levels is not favored.These results imply that other pathways contribute to the internal mixture of organic and inorganic aminium salts found in the atmosphere.It should be noted that ammonium salts are major components of atmospheric particles.If amines are taken up onto these pre-existing particles, they may form particulate organic amines.Thus, this heterogeneous reaction would explain the formation of internal mixture of organic and inorganic aminium salts.Very few studies have investigated the reactions between DMA, TMA, and clusters of NH 4 HSO 4 and NH 4 NO 3 (Bzdek et al., 2010a, b) between TMA and NH 4 NO 3 particles (Lloyd et al., 2009), and between MA, DMA, and TMA and (NH 4 ) 2 SO 4 and NH 4 HSO 4 (Qiu et al., 2011).The uptake coefficient of TMA on NH 4 NO 3 has been measured as 2±2×10 −3 (Lloyd et al., 2009), while the uptake coefficient for MA, DMA, and TMA by (NH 4 ) 2 SO 4 has been measured as 2.6×10 −2 to 3.4×10 −2 (Qiu et al., 2011).The uptake coefficients of TMA and DMA on clusters (1-2 nm) of NH 4 HSO 4 and NH 4 NO 3 are close to unity (Bzdek et al., 2010a, b).The kinetics of these reactions in relation to the amine structure (Wang et al., 2010b) and also to the particle size of the ammonium salt (Bzdek et al., 2010a, b;Lloyd et al., 2009) have been discussed in few studies.Recently, Chan and Chan (2012) also confirmed the displacement reaction between triethylamine (TEA) and (NH 4 ) 2 SO 4 , NH 4 HSO 4 , NH 4 NO 3 , NH 4 Cl and (NH 4 ) 2 C 2 O 4 using in situ Raman spectroscopy.At present date, the reported uptake coefficients showed a discrepancy among different ammonium salts.However, it is hard to directly compare these results because of the difference in reaction conditions, such as the particle size, the type of amines and ammonium salts, and the reaction temperature used in experiments and so on.On the other hand, the relatively few kinetic data is available for one kind of amine.For example, as for MA, only one paper (Qiu et al., 2011) reported its uptake coefficient on (NH 4 ) 2 SO 4 .The uptake coefficients of MA on other salts are unknown.Thus, it is not clear yet that how the property of inorganic ammonium salts affects the reactivity for this reaction.
In this study, we investigated the heterogeneous uptake of MA on NH 4 HSO 4 , (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl to understand the effect of ammonium salts on their reactivity with MA.We established for the first time the structure-reactivity relationship between MA and ammonium salts.The surface species during uptake were investigated using an in situ Raman spectrometer.The kinetic were measured at 298 K using a Knudsen cell reactor coupled to a quadrupole mass spectrometer (QMS).A linear free energy relationship was observed for the uptake of MA onto (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl based on theoretical calculations.The environmental implications for these reactions were also discussed.

Uptake experiments
We used a Knudsen cell reactor coupled to a QMS (KCMS, Hiden, HAL 3F PIC) to conduct the kinetic experiments (Liu and He, 2009;Liu et al., 2009Liu et al., , 2010b, c), c).Briefly, the mass spectrometer was housed in a vacuum chamber equipped with a 300 L s −1 turbomolecular pump (Pfeiffer) and an ion gauge (BOC Edward).The vacuum chamber between the QMS and the Knudsen cell reactor was pumped by a 60 L s −1 turbomolecular pump for the mass spectrometer and ion gauge (both from BOC Edward) differential pumping.The Knudsen cell reactor consisted of a stainless steel chamber with a gas inlet controlled by a leak valve, an escape aperture whose area could be modified with an adjustable iris, and a sample holder attached to the top ceiling of a circulating fluid bath.The area of the escape aperture was set at 0.88 mm 2 during uptake experiments and measured according to methods reported previously (Liu and He, 2009;Liu et al., 2009Liu et al., , 2010a, b), b).The sample in the sample holder was exposed to or isolated from the reactants by a lid connected to a linear translator.The temperature of the sample holder was measured with an embedded Pt resistance thermometer and held at 298 K±0.10 K using a super thermostat and cryofluid pump (DFY-5/80, Henan Yuhua laboratory instrument Co, Ltd.).
Powder samples of ammonium salts were finely grounded and dispersed evenly on the Teflon sample holder and then out-gassed at 298 K for 8 h to reach a base pressure of approximately 5.0×10 −7 Torr.After the sample cover was closed, MA gas equilibrated with 40 wt% methylamine solution was introduced into the reactor chamber through the leak valve.The pressure in the reactor, which was controlled by the leak valve and measured using the absolute pressure transducer, was 3.0±0.2×10−5 Torr.Prior to the experiments, the reactor chamber was passivated with amines for 90 min until a steady state QMS signal was established as the samples were isolated from the gas by the sample cover.The observed uptake coefficients (γ obs ) were calculated with a Knudsen cell equation (Tabor et al., 1994;Ullerstam et al. 2003;Underwood et al., 2000).
Where, A h is the effective area of the escape aperture (cm 2 ); A g is the geometric area of the sample holder (cm 2 ); and I 0 and  being finely grounded.Their specific surface areas (N 2 -BET) were measured around 0.1 m 2 g −1 using a Quantachrome Autosorb-1-C instrument.

In situ Raman spectra measurements
In situ Raman spectra for the reaction of ammonium salts with MA were recorded on a UV resonance Raman spectrometer (UVR DLPC-DL-03), which was described elsewhere (Liu et al., 2010a).The instrument was calibrated against the Stokes Raman signal of Teflon at 1378 cm −1 .A continuous diode pumped solid state (DPSS) laser beam (532 nm) was used as the exciting radiation, with the adjustable source power ranging from 0 to 200 mW.A source power with 50 mW was used and no sample modification was observed when the sample was irradiated under the experimental conditions.The diameter of the laser spot on the sample surface was focused at 25 µm.The spectra resolution was 2.0 cm −1 .Ammonium salt powder was placed into a stainless steel sample holder and purged with 100 mL/min of simulated air (80 % high purity N 2 and 20 % high purity O 2 ) for 2 h at 298 K.Then, ca.0.5 % of MA in 100 mL/min air was carried by nitrogen from 40 % MA aqueous solution into the reactor and then in situ Raman spectra were recorded.To avoid the influence of visible light on the heterogeneous reaction, the sample was only exposed to the exciting laser for spectra collection.The exposure time for each spectrum was 10 s with 6 scans.The temperature was held at 298 K during the reaction.

Computational methods
We employ the Gaussian 09 suite of programs (Frisch et al., 2009) to investigate the electrostatic potential of the ammonium salts.Geometry optimization were performed within the premise of the density functional theory (DFT) using Becke's three-parameter hybrid functional (B3) (Becke, 1993) combined with the electron-correlation functional of Lee, Yang, and Parr (LYP) (Lee et al., 1988).6-311++G (2df, 2p) basis set was used for all the calculations.

Uptake of MA onto inorganic ammonium salts
Figure 1 shows the typical mass spectra profiles for MA uptake onto (NH 4 ) 2 SO 4 , NH 4 HSO 4 , NH 4 NO 3 , and NH 4 Cl at 298 K.According to the fragmentation patterns of CH 3 NH 2 , the fragment peak at m/e = 30 (CH 2 NH 2 ) had the strongest intensity, followed by the parent peak at m/e = 31 (CH 3 NH 2 ).Therefore, these two mass channels were scanned for MA measurement and shown in blue and red color, respectively.The mass channel of m/e = 17 was monitored for the possible product of NH 3 and show in black color for the original signal and purple color for corrected one.The fragment peaks of H 2 O (OH) and CH 3 NH 2 (NH 3 ) may have also contributed to this mass channel.To determine the contribution of H 2 O to m/e = 17, the mass channels of m/e = 18 (H 2 O) shown in olive color was scanned because gaseous MA was generated using 40 wt% MA aqueous solution.For NH 4 Cl samples, the mass channel of m/e = 36 was additionally scanned for monitoring the decomposition of NH 4 Cl.
As shown in Fig. 1a and b, the signal change of MA (m/e = 30 and 31) was classified into three stages after the (NH 4 ) 2 SO 4 sample was exposed to MA vapor.The first stage was the remarkable decrease in intensity with maximal amplitude of 67 % within the first 2 min.The second was the quick recovery stage (from 2 min to 5 min) followed by the third stage with a slow recovery rate.Signal intensity recovered about 40 % within 3 min of the second stage, and recovered a further 35 % within 30 min of the third stage.As shown in Fig. 1, when the NH 4 HSO 4 , NH 4 NO 3 , and NH 4 Cl samples were exposed to MA vapor, the signal intensity of MA (m/e = 30 and m/e = 31) only contained two stages.In the first stage, it decreased dramatically, followed by a gradual decrease to the lowest values with the maximal amplitudes of 78 %, 75 % and 38 %, respectively.In the second stage, the signal intensities of MA gradually increased with time and finally recovered about 10 %, 13 %, and 33 % at 30 min, respectively.The signal recovery should be ascribed to desorption of MA or consumption of active sites over time.Because the experiments were conducted in a steady-state, the decrease in MA signal intensity indicated the loss of MA molecules from gas phase to the surface of these ammonium salt samples.
When the (NH 4 ) 2 SO 4 sample was exposed to MA, the original profile for the m/e = 17 mass channel (Fig. 1c) was, however, quite different from that of MA, though it also included the three stages, that is, a decrease, increase and then decrease in intensity.The complicated patterns of its signal profile implied that more than one species contributed to this mass channel.As mentioned above, H 2 O, NH 3 , and CH 3 NH 2 may contribute to this mass channel.As shown in Fig. 1d, however, the signal intensity of m/e = 18 remained almost unchanged when the sample was exposed to MA vapor.So it does (for m/e = 18) when other samples were exposed to MA vapor (Figs.1h, l, and p).Therefore, the contribution of H 2 O to the m/e = 17 mass channel was ignored because the intensity ratio of m/e = 17 to that of m/e = 18 was about 0.3 in pure water vapor (NIST).In Fig. 1c, the synchronous decrease in intensity of m/e = 17, m/e = 30, and m/e = 31 in the first stage suggested that these three mass channels had the same source (CH 3 NH 2 ).The same phenomenon was observed during the whole uptake stage for MA onto NH 4 HSO 4 (Fig. 1e-g).It should be pointed out that the signal intensity of m/e = 17 in Fig. 1c exceeded its baseline in the second stage (about 3 min after the sample cover was opened) and decreased slowly with time after reaching a maximal value (about 10 min).This implied a gaseous product (NH 3 ) also contributed to this mass channel during uptaking MA onto (NH 4 ) 2 SO 4 .As shown in Fig. 1g, a continuous decrease with an amplitude around 28 % for the signal intensity of m/e = 17 was observed when the NH 4 HSO 4 sample was exposed to MA vapor.Because the contribution of H 2 O to this mass channel was not important, the decrease of the signal intensity of m/e = 17 can be mainly ascribed to consumption of MA. Figure 1k and o show the evolution of the signal intensity of m/e = 17 when NH 4 NO 3 and NH 4 Cl were exposed to MA vapor, respectively.Unlike those shown in Fig. 1c and g, significant increase in the intensity of m/e = 17 occurred intermediately at the same time as NH 4 NO 3 and NH 4 Cl samples uptake MA.The purple curves for m/e = 17 show the corrected signal originating from NH 3 after the contribution of MA to this mass channel was subtracted, using I 17 / I 30 = 0.27 (Fig. 1e and g).
Figure 2 shows the in situ Raman spectra for the heterogeneous reaction of MA with ammonium salts at 298 K.The insert graphs in purple box show the enlarged spectra in the range of 950-1080 cm −1 and those on the top right corner show the spectra in the range of 2500-3600 cm −1 .When the NH 4 NO 3 sample was exposed to MA vapor, new bands at both 1008 cm −1 and 2981 cm −1 increased with time.The band at 1008 cm −1 was ascribed to the stretch vibration of C-N in MA and the band at 2981 cm −1 was related to that of CH 3 (Purnell et al., 1976).For (NH 4 ) 2 SO 4 sample, the band at 2981 cm −1 was observed obviously while the band at 1008 cm −1 was observed as a shoulder peak because of the strong peak of SO 2− 4 (v 1 ) at 976 cm −1 (Kruus et al., 1985).It was inversed for NH 4 Cl.The changes of other bands attributing to (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl were unobservable.These results well supported the exchange reactions discussed above between MA and (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl.However, when NH 4 HSO 4 was exposed to MA vapor, the bands attributing to HSO − 4 at 1048, 1017, 884, 578, and 415 cm −1 (Kruus et al., 1985) decreased in intensity with time, accompanied with obvious increase of SO 2− 4 (978 cm −1 ), C-N (1008 cm −1 ) and CH 3 (2986 cm −1 ) in MA.The synchronous decrease in intensity of HSO − 4 with the increase of SO 2− 4 suggested the transfer of H proton from HSO − 4 to MA, then leading to the formation of SO 2− 4 .This result demonstrated that the active site for uptake of MA onto NH 4 HSO 4 was the proton in HSO − 4 .These results were also well agreement with Knudsen cell experiments.
Therefore, based on these results, we concluded that except for NH 4 HSO 4 , exchange reaction between CH 3 NH 2 and (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl occurred under these conditions.As for NH 4 HSO 4 , a simple acid-base reaction occurred and the proton in HSO − 4 should be the reactive site.These reactions can be summarized as follows.Because one (NH 4 ) 2 SO 4 molecule contains two NH + 4 ions, the exchange reaction between MA and (NH 4 ) 2 SO 4 may involve two steps.Similar exchange reactions have been observed in previous studies between DMA, TMA, and clusters of NH 4 HSO 4 and NH 4 NO 3 (Bzdek et al., 2010a, b), between TMA and NH 4 NO 3 particles (Lloyd et al., 2009), and between MA, DMA, TMA and (NH 4 ) 2 SO 4 and NH 4 HSO 4 (Qiu et al., 2011).

Stability of aminium salts
To investigate the stability of aminium salts formed on the surface, desorption experiments were performed after the completion of the uptake experiments.After the sample was isolated from the feed gas by the sample cover, the introduction of feed gas into the reactor was stopped.When a steadystate mass spectrometry signal was established, the sample cover was opened and the desorbed species were monitored online with the QMS.The escape aperture area was increased to 5.5 mm 2 from 0.88 mm 2 in some cases using the variable iris to increase the gas flow rate from the Knudsen cell reactor into the QMS detector.The mass channels were scanned as the same as that during uptake experiments.Figure 3a-d show the mass spectra of the cared desorbates from 58.2 mg (NH 4 ) 2 SO 4 , 60.5 mg NH 4 HSO 4 , 59.8 mg NH 4 NO 3 , 63.9 mg NH 4 Cl, respectively, after exposed to MA for ca.30 min.
As shown in Fig. 3a and b, no desorption of MA and NH 3 was observed at 298 K from MA treated (NH 4 ) 2 SO 4 and NH 4 HSO 4 samples even though a maximal A h (5.5 mm 2 ) was used.This indicated that the uptake of MA onto the surface of (NH 4 ) 2 SO 4 and NH 4 HSO 4 is irreversible and CH 3 NH 3 (NH 4 )SO 4 and (CH 3 NH 3 ) 2 SO 4 are stable under these conditions.This was similar to the irreversible exchange reactions between DMA or TMA and ammonium bisulfate clusters reported by Bzdek et al. (2010a).
As shown in Fig. 3c and d, when the sample cover was opened with the same A h used in the uptake experiments (0.88 mm 2 ), desorption of CH 2 NH 2 and CH 3 NH 2 from the MA exchanged NH 4 NO 3 and NH 4 Cl sample was very weak, while release of NH 3 (Figs.3c and d) was very clear.To increase the signal of the mass spectrometer, maximal A h was used through adjusting the iris.The signal intensities of the m/e = 30 and m/e = 17 mass channels increased obviously and were accompanied by a faint increase in the mass channel of m/e = 31 when the sample cover was opened.In  et al., 2000).Other ammonium salts might also undergo decomposition reaction.Therefore, the decomposition experiments for pure ammonium salts were performed under the similar conditions.The mass spectra for pure (NH 4 ) 2 SO 4 and NH 4 HSO 4 when evacuated at 1.2×10 −6 Torr and at 298 K (not shown) were the same as those shown in Fig. 3a and  b, which suggested that decomposition of (NH 4 ) 2 SO 4 and NH 4 HSO 4 are negligible under these conditions.Figure 3e shows the mass spectra of desorbates from pure NH 4 NO 3 .Observable decomposition of NH 4 NO 3 occurred under these conditions.In Fig. 3e, the I 17(NH 3 ) / I 30(NO) could be normalized to 1 because the mass channels of m/e = 30 and m/e = 17 were the result of the dissociation of NH 4 NO 3 .The normalized ratio of I 17(NH 3 ) / I 30(NO+CH 2 NH 2 ) was 0.85.This also meant that desorption of MA (CH 2 NH 2 ) explained 15 % of the m/e = 30 signal change in Fig. 3c.Therefore, the increase in m/e = 30 signal intensities (Fig. 3c) were mainly from the decomposition of the unreacted NH 4 NO 3 .Figure 3f shows the desorption of NH 3 and HCl (m/e = 36, 3 times of magnification) from pure NH 4 Cl.It was almost unobservable for desorption of NH 3 and HCl with the A h of 0.88 mm 2 ; while a weak desorption of NH 3 and HCl can be discerned when the A h was increased to 5.5 mm 2 .It meant that dissociation of both NH 4 Cl and CH 3 NH 3 Cl should contribute to the signals in Fig. 3d.According to the value of I 17 / I 30 = 0.27 and the change of signal intensity of m/e = 30 in Fig. 3d, we estimated that decomposition of CH 3 NH 3 Cl contributes 12 % to the intensity change of the m/e = 30 in Fig. 3d.Thus, we can conclude that CH 3 NH 3 NO 3 and CH 3 NH 3 Cl are not stable at low pressure and they have higher dissociation vapor pressure than methylammonium sulfate.

Kinetics of MA uptake onto ammonium salts
As discussed above, the signal intensities of m/e = 30 and m/e = 31 decreased synchronously when the sample cover was opened.Because the signal intensity of m/e = 30 was linearly correlated with that of m/e = 31 (I 31 = 0.60I 30 , R = 0.997), either can be used to calculate the γ obs according to Eq. (1).As discussed above, decomposition of NH 4 NO 3 can contribute to the mass channel of m/e = 30.However, when the signal change in Fig. 2e was compared to that in Fig. 1i, the compensation effect of NH 4 NO 3 decomposition to the signal intensity of m/e = 30 is neglectable.Thus, in the following section, the signal intensity of m/e = 30 was used for uptake coefficient calculations.
Figure 4 illustrates typical profiles of the observed uptake coefficient as a function of time at 298 K.The γ obs of MA on (NH 4 ) 2 SO 4 decreased with exposure time (Fig. 3a) corresponding to the recovery of the MA signal (Fig. 1a).In Fig. 4, the γ obs of MA on NH 4 HSO 4 , NH 4 NO 3 , and NH 4 Cl also showed a slow decline with time after the maximal value was reached.As discussed above, neither desorption of MA nor the dissociation of aminium salts formed on the (NH 4 ) 2 SO 4 and NH 4 HSO 4 surface occurred; therefore, the decrease in the uptake coefficient of MA on (NH 4 ) 2 SO 4 and NH 4 HSO 4 was ascribed to the depletion of active sites over time.As for NH 4 NO 3 and NH 4 Cl, although the exchange reactions were partially reversible, the dissociation rates of CH 3 NH 3 NO 3 and CH 3 NH 3 Cl were slower than their formation rates.Therefore, accessible active sites also decreased with time.Table 1 summaries the initial uptake coefficients of MA onto (NH 4 ) 2 SO 4 , NH 4 HSO 4 , NH 4 NO 3 and NH 4 Cl at 298 K.
Previous research has found that γ obs depends on sample mass due to the diffusion of reactant gas into the underlying layers of the multilayer powder sample (Liu and He, 2009;Liu et al., 2009 and2010b, c;Ullerstam et al., 2003;Underwood et al., 2000 and2001).Some studies have found, however, that uptake coefficient is unrelated to sample mass (Qiu et al., 2011;Seisel et al., 2005).In the present work, we performed uptake experiments on a wide range of sample masses.When the sample mass was lower than 40.0 mg, it was difficult to cover the sample holder with salt particles evenly.Therefore, uptake experiments on samples with masses lower than 40 mg were not performed.Sample mass dependence was not observed for the γ obs of MA on all of these salts.This indicates that the underlayers of the sample contributed very little to MA uptake.This phenomenon also supported the slight declines of uptake coefficient with exposure time (Fig. 4) because the reactions were confined on the surface.The same phenomena was also observed for the uptake of MA, DMA, and TMA on (NH 4 ) 2 SO 4 (Qiu et al., 2011).On the other hand, when the QMS signal intensity (I ) was calibrated with molecular flow rate (mol s −1 ) (Liu et al., 2008), the amount of MA uptake onto ammonium salts can be calculated using the integrated area shown in Fig. 1.They are 6.0×10 −7 , 1.1×10 −6 , 1.2×10 −6 , and 3.9×10 −7 moles on (NH 4 ) 2 SO 4 , NH 4 HSO 4 , NH 4 NO 3 , and NH 4 Cl, respectively.The amount of ammonium salts are 4.4×10 −4 , 5.3×10 −4 , 7.5×10 −4 and 1.2×10 −3 moles, respectively.Therefore, the ratio between uptaked amine and ammonium salts during the reaction are estimated to be 0.14 %, 0.21 %, 0.16 % and 0.03 % on (NH 4 ) 2 SO 4 , NH 4 HSO 4 , NH 4 NO 3 , and NH 4 Cl, respectively.If two ammonium ions in (NH 4 ) 2 SO 4 involve in the exchange reaction, the ratio should be 0.07 %.It means than only a very small fraction of ammonium salts, most probably confined on the surface, involves in this heterogeneous reaction.Therefore, it was reasonable to speculate that the true uptake coefficient (γ t ) was very close or equal to the γ obs .As shown in Table 1, the average γ obs of MA on (NH 4 ) 2 SO 4 , NH 4 HSO 4 , NH 4 NO 3 and NH 4 Cl at 298 K was 6.30 ± 1.03 × 10 −3 , 1.78 ± 0.36 × 10 −2 , 8.79 ± 1.99 × 10 −3 and 2.29 ± 0.28 × 10 −3 , respectively.The uptake coefficients of MA onto (NH 4 ) 2 SO 4 , NH 4 Cl and NH 4 NO 3 measured in this work are comparable to that of TMA on NH 4 NO 3 (2 ± 2 × 10 −3 ) reported by Lloyd et al. (2009), while they are two orders of magnitude lower than the values for DMA and TMA uptake on clusters of NH 4 HSO 4 and NH 4 NO 3 reported by Bzdek et al. (2010a, b).The uptake coefficient of MA onto NH 4 HSO 4 is comparable to that on sulfuric acid (Wang et al., 2010a).As for the uptake of MA on (NH 4 ) 2 SO 4 , uptake coefficients were slightly lower than the value (2.6 × 10 −2 -3.4×10 −2 ) reported by Qiu et al. (2011).Recently, Chan and Chan (2012) found that amorphous NH 4 NO 3 and NH 4 HSO 4 showed higher degree of exchange reaction to TEA than that of crystalline NH 4 NO 3 and NH 4 HSO 4 .Thus, the difference in the uptake coefficient may be ascribed to different reaction systems, different samples with different particle sizes or morphology.In the studies by Bzdek et al (2010a, b), for example, 1-2 nm clusters of bisulfate or nitrate were used, and amorphous (NH 4 ) 2 SO 4 was used by Qiu et al. (2011); while crystal samples were used in the present study.In ambient environments, however, low RH conditions may also induce potential crystallization of these ammonium salts (Clegg et al., 1998).Thus, the uptake coefficients reported here should represent the low limits for uptake of MA on ammonium salts.

Effect of property of ammonium salts on the heterogeneous reaction
Above results demonstrated that exchange reaction between MA and (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl occurred, while only acid-base reaction took place on NH 4 HSO 4 .In general speaking, these reactions can be regarded as electrophilic reaction, in which the lone-pair electron in CH 3 NH 2 was the active center.The molecular electrostatic potential (MEP) is widely used as a reactivity map displaying most probable regions for the electrophilic attack of charged point-like reagents (Politzer et al., 1981).In the reactions of a series of related molecules, A 1 , A 2 , . . ., A n , with some given reactant  B, the differences in the activation energies are linearly related to the differences in the heats of reaction, namely, a linear free energy relationship.Because electrostatic potential can be used as a measurement of thermodynamic, the electrostatic potentials at some particular sites in such a series of molecules are related to their interaction energies with B. Therefore, MEP is also used to correlate with the reaction mechanism and kinetics (Politzer et al., 1985).
Figure 5 shows the MEP of ammonium salts calculated at B3LYP/6-311++G (2df, 2p) level.The geometries for these salts were also optimized at the same level and the true minimal was confirmed without imaginary frequency.As shown in Fig. 5, the maximal value for electrostatic potential located at the H atom in the NH + 4 group and were 827.58, 1386.38, 1741.85, and 637.42 kJ mol −1 for NH 4 Cl, (NH 4 ) 2 SO 4 , NH 4 NO 3 , and NH 4 HSO 4 , respectively.Figure 6 indicates the correlation between the lnγ of MA onto these ammonium salts with the corresponding MEP.As can be seen in Fig. 6, a linear correlation between the lnγ and MEP were observed among NH 4 Cl, (NH 4 ) 2 SO 4 , and NH 4 NO 3 , with a high correlation coefficient (R = 0.991).NH 4 NO 3 with the highest MEP had the largest uptake coefficient.However, NH 4 HSO 4 obviously deviated this linear relationship.This deviation can be ascribed to the different reaction mechanism as discussed above.This result also implied that the first step for the heterogeneous reactions between MA and NH 4 Cl, (NH 4 ) 2 SO 4 , and NH 4 NO 3 should be adsorption of MA onto the NH + 4 group, in which H atom should be the active site.

Conclusions and atmospheric implications
In this work, the heterogeneous reactions of MA on the typical ammonium salts including (NH 4 ) 2 SO 4 , NH 4 HSO 4 , NH 4 NO 3 and NH 4 Cl were investigated by using Knudsen cell reactor, Raman spectroscopy and theoretical calculations.Exchange reactions between MA and (NH 4 ) 2 SO 4 , NH 4 NO 3 and NH 4 Cl were observed, while acid-base reaction was found between MA and NH 4 HSO 4 .The stability of CH 3 NH 3 NH 4 SO 4 or (CH 3 NH 3 ) 2 SO 4 were found to be stronger than CH 3 NH 3 NO 3 and CH 3 NH 3 Cl at low pressure.The observed uptake coefficients of MA on these salts can be used as the true uptake coefficients because diffusion of MA into the underlayers of powder samples was neglectable.The uptake coefficients were 6.30 ± 1.03 × 10 −3 , 1.78 ± 0.36 −2 , 8.79 ± 1.99 × 10 −3 and 2.29 ±0.28 × 10 −3 on (NH 4 ) 2 SO 4 , NH 4 HSO 4 , NH 4 NO 3 and NH 4 Cl at 298 K, respectively.As for (NH 4 ) 2 SO 4 , NH 4 Cl, and NH 4 NO 3 , the structure-reactivity relationship was established, namely, natural logarithm of the uptake coefficients were highly linear related with the MEP of the H atom in the NH + 4 group.It means property of ammonium salts have strong influence on the reactivity of MA with these ammonium salts.
Field measurements have found that aminium salts are usually internally mixed with inorganic ammonium salts in atmospheric particulate matter (Facchini et al., 2008;Pratt et al., 2009;Smith et al., 2010).In general, acid-base reactions between amines and H 2 SO 4 or HNO 3 are considered the main source of particulate aminium salts.For example, Wang et al. (2010a) measured the uptake coefficients of MA on 62-82 wt% of H 2 SO 4 to be 2.0±0.2×10−2 to 3.2±0.5×10−2 at 283 K.This is four to ten times higher than the γ obs for MA on NH 4 NO 3 , (NH 4 ) 2 SO 4 or NH 4 Cl, while it is comparable to the uptake coefficient of MA onto NH 4 HSO 4 measured in this work.When the abundance of ammonium salts in the troposphere as well as the relatively large uptake coefficients of MA on them are considered, uptake of MA onto ammonium salts should not be ignored.Of course, it should be pointed out that our experiments were carried out under dry condition and the clean surface of ammonium salts was studied.In the real atmosphere, high relative humidity (RH) and organic species on the surface might have complex influence on this reaction.Recently, Chan and Chan (2012) found that aqueous salts of (NH 4 ) 2 SO 4 , NH 4 HSO 4 and NH 4 Cl show higher degree of exchange reaction to TEA.It implies the uptake coefficients of MA on these salts might also be larger under ambient RH than these measured under dry conditions.In ambient environments, however, low RH conditions may also induce potential crystallization of these ammonium salts (Clegg et al., 1998).Thus, as the low limits for uptake coefficient of MA onto ammonium salts, the measurements in this study at least suggest that amines uptake onto pre-existing ammonium salts even under dry conditions may contribute, under particular circumstances, as high concentrations of ammonium salts and low concentration of acidic gas species, to the atmospheric source of particulate amines.The displace reactions between amines and ammonium salts should greatly modify the aerosol properties, such as increase in hygroscopicity and decrease in density, compared with ammonium salts (Qiu and Zhang, 2012).

Fig. 2 .
Fig. 2.In situ Raman spectra for heterogeneous reactions of MA with ammonium salts at 298 K.The insert graphs in purple boxes show the enlarged spectra in the range of 950-1080 cm −1 and those on the top right corner show the spectra in the range of 2500-3600 cm −1 .

Fig. 6 .Fig. 6 .
Fig. 6.Relationship between uptake coefficients of MA onto ammonium salts and the 696 electrostatic potential of H atom in NH4 + .697 698

Table 1 .
Uptake coefficients of MA onto ammonium salts at 298 K.