In recent years, gaseous amines have attracted increasing attention due to theoretical, laboratory, and field measurements indicating that amines may considerably enhance particle formation and growth (Kurtén et al., 2008; Nadykto et al., 2011, 2014; Almeida et al., 2013; Berndt et al., 2010; Zhao et al., 2011; Erupe et al., 2011; Chen et al., 2012; Wang et al., 2010a; Yu et al., 2012) and affect secondary organic aerosol (SOA) formation (De Haan et al., 2009, Myriokefalitakis et al., 2010; Williams et al., 2010). Amines are organic compounds and derivatives of ammonia wherein one or more hydrogen atoms are replaced by a substituent such as an alkyl or aryl group. About 150 amines have been identified in the atmosphere; the most common and abundant amines being the low-molecular-weight methylamines such as monomethylamine (MMA), dimethylamine (DMA), and trimethylamine (TMA) (Ge et al., 2011a). Concentrations of amines can exceed several parts-per-billion by volume (ppbv) near their sources (Ge et al., 2011a; Schade and Crutzen, 1995) but are expected to be low farther away as a result of their short lifetime due to oxidation by OH (Carl and Crowley, 1998) and uptake by particles (Qiu and Zhang, 2013).

While amines are stronger bases than ammonia and ternary H2SO4-H2O-amine clusters are more stable (Kurtén et al., 2008; Nadykto et al., 2011, 2014; Almeida et al., 2013), the relative role of amines versus ammonia in enhancing particle formation in the atmosphere is yet to be determined (Zollner et al., 2012). This is because the concentration of amines in the atmosphere is generally much lower than that of ammonia (by 2–3 orders of magnitude or more) (Ge et al., 2011a; Hanson et al., 2011). Recent measurements taken during the CLOUD (Cosmics Leaving Outdoor Droplets) chamber experiments at CERN (European Council for Nuclear Research; Almeida et al., 2013) indicate that a [DMA] of above ∼5 parts-per-trillion by volume (pptv) enhances nucleation substantially, but enhancement drops significantly as [DMA] decreases below that level.

In order to determine the contribution of ternary nucleation involving amines to atmospheric particle production, it is critical to know the concentrations of key amines and their variations in the atmosphere. Due to their high reactivity and low concentrations, measurements of gaseous amines in the background atmosphere are very limited (Ge et al., 2011a). Several studies show [DMA] is below 1 pptv in urban areas (Grönberg et al., 1992a, b) while a couple of other studies observed [DMA] around a few pptv in rural and coastal areas (Hanson et al., 2011; VandenBoer et al., 2011, 2012; Van Neste et al., 1987; Gibb et al., 1999). Although TMA is generally more abundant (Ge et al., 2011a), the concentration of TMA needed to substantially enhance nucleation remains to be studied.

In addition to in situ measurements, numerical modeling is also needed to integrate the various processes controlling amine concentrations and ultimately assess the impact of amines on global nucleation, aerosol properties, and climate. While limited measurements of amines are available, modeling of global amines is basically nonexistent. Myriokefalitakis et al. (2010) explored the potential contribution of amines emitted from oceans to SOA formation, assuming total amine emissions to be one-tenth of the oceanic ammonia emissions. Myriokefalitakis et al. (2010) neither considered amines from continental sources nor presented concentrations of gaseous amines over oceans. In the present work, we aim to simulate the global distributions of gaseous amines in the air with a global chemistry transport model. The key processes controlling amine concentrations (including emission, transport, oxidation, deposition, and aerosol uptake) are considered and the simulated results are compared to the limited measurements available.

The methods of the present study (including sources, sinks, and model representation) are described in Sect. 2. The modeling results, comparisons with measurements, and sensitivity studies are given in Sect. 3. Section 4 contains the summary and discussion.

The results presented below are based on a 1 yr simulation (October 2005–December 2006, with the first 3 months as spin up) using GEOS-Chem v8.3.2 + APM, with the kinetic condensation of low, volatile secondary organic gases from successive oxidation aging taken into account (Yu, 2011). The horizontal resolution (latitude by longitude) is 2∘ × 2.5∘ and there are 47 vertical layers in the model (surface to 0.01 hpa).

Table 1 shows global annual mean emissions, sinks (due to oxidation, uptake, and dry/wet deposition), and burdens for ammonia, MMA, DMA, and TMA. Sinks and burdens of methylamines under four different uptake coefficients (γ = 0.03, 0.01, 0.001, and 0) are given. The global ammonia emission flux for 2006 based on GEOS-Chem is 5.8 × 104 Gg N yr-1, about 15 % higher than the estimation of Schade and Crutzen (1995). The MMA, DMA, and TMA emissions fluxes assumed in the present study (96.2, 38.3, and 196.0 Gg N yr-1, respectively) are also 15 % higher, as the same ratios of methylamines to ammonia emission fluxes given in Schade and Crutzen (1995) are used. The 15 % difference is within the estimated methylamine emission uncertainty of ∼30 % (Schade and Crutzen, 1995).

As an example for the spatial distribution of emission fluxes, Fig. 1 presents the horizontal distributions of DMA emissions assumed in the present study. As mentioned earlier, we approximate the spatial distribution and seasonal variations of methylamine emissions following those of ammonia. Again, this should be considered as a first-order approximation, as the emission rates of amines from various sources may be quite different from those of ammonia. With the understanding of this limitation, we can see in Fig. 1 that DMA emission rates are in the range of ∼0.2–10 kg N km-2 yr-1 over major continents and below 0.2 kg N km-2 yr-1 over oceans. For MMA and TMA, the absolute emission fluxes are a factor of 2.5 and 5.1 higher (Table 1). In Fig. 1 we also marked the locations of sites where some kind of methylamine measurements are available, as summarized in Table 2. It should be noted that sites A, B, D, and G are close to each other and overlap in Fig. 1. Similarly, sites E and F overlap in Fig. 1. Sites J and K are the same location but measurements were taken during different time periods. A comparison of simulated and observed methylamine concentrations is discussed later.

It can be seen from Table 1 that gas phase oxidation and aerosol uptakes are dominant sinks for methylamines (Table 1). As expected, the uptake sinks are sensitive to γ when γ > ∼0.001 and the oxidation becomes more important when γ is smaller. The change of γ from 0.03 to 0.001 increases the modeled global burdens of methylamines by a factor of ∼2.7. Further decrease of γ from 0.001 to 0 has relatively small effects on the predicted burdens. Dry and wet deposition accounts for 11–14 % and 25–35 % of the sinks when γ = 0.03 and γ = 0, respectively. The global burdens of MMA, DMA, and TMA are respectively from 0.07 to 0.27 Gg N, 0.03 to 0.08 Gg N, and 0.24 to 0.72 Gg N as γ changes from 0.03 to 0. The ratios of ammonia burden to that of methylamines (MMA + DMA + TMA) range from 74 (γ = 0) to 236 (γ = 0.03). The burdens are roughly but not strictly proportional to emission fluxes because of the difference in the oxidation rates and deposition velocities (which also depend on molecular weights).

Figure 2 shows the simulated horizontal distributions of the annual mean DMA oxidation and uptake lifetime (τ, calculated as the ratio of the burden in each gird box to the corresponding sinks associated with oxidation and uptake) and concentration ([DMA]) in the model surface layer (0–150 m above surface) under two aerosol uptake coefficients: (a, b) γ = 0 (i.e., oxidation only) and (c, d) γ = 0.03 (uptake by sulfuric acid particles). The corresponding zonally averaged vertical distributions of τ and [DMA] are given in Fig. 3. The oxidation only condition (i.e., no aerosol uptake) leads to a DMA lifetime of 5–10 h in most parts of lower and middle latitude regions, from the surface to the upper troposphere. The oxidation lifetime is relatively long (from 10 to >200 h) over the high latitude regions due to low OH concentrations there. The aerosol uptake with γ = 0.03 (upper limit, corresponding to the uptake by sulfuric acid particles) shortens the lifetime of DMA by ∼30 % over oceans and much more over the major continents, resulting in a DMA lifetime of less than 1–2 h over central Europe, eastern Asia, and the eastern US (Fig. 2c). Our sensitivity study indicates that τ values decrease with increasing γ when γ > 0.001 but become relatively insensitive to γ when γ < 0.001, as oxidation dominates the lifetime under this condition.

As a result of a short lifetime, high values of [DMA] are generally confined to the source regions (Figs. 1, 2b, d). Depending on the uptake coefficients, [DMA] in the surface layer over major continents is in the range of 0.1–2 ppt when γ = 0.03 (Fig. 2d) and 0.2–10 ppt when γ = 0 (Fig. 2b). [DMA] decreases quickly with altitudes, with zonally averaged values dropping below 0.1 ppt a few hundred meters above the surface (Fig. 3b, d). [DMA] over oceans are below 0.05 ppt and these DMA are emitted from marine organisms (Fig. 1) rather than transported from continents. [DMA] over polar regions is below 0.01 ppt (Figs. 2, 3) due to the lack of emissions there (Fig. 1).

The annual mean horizontal and vertical distributions of MMA and TMA concentrations ([MMA], [TMA]) under two γ values (0.03, and 0) are shown in Figs. 4 and 5. As a result of identical emission spatial distributions (assumed) and short lifetimes, [MMA] and [TMA] have similar spatial distributions as those of [DMA]. [MMA] is generally a factor of ∼2.5 higher than [DMA], reaching 0.2–5 ppt when γ = 0.03 (Fig. 4c) and 0.5–20 ppt when γ = 0 (Fig. 4a) in the surface layer over major continents. While the oxidation rate of MMA is smaller than that of DMA, its deposition velocity is larger. As a result, the [MMA] to [DMA] ratio is close to the ratio of the corresponding global emission fluxes. In contrast, both the oxidation and deposition velocities of TMA are smaller than those of DMA, leading to a larger [TMA] to [DMA] ratio (∼8) than the corresponding ratio of emission fluxes (∼5). [TMA] in the surface layer over major continents reaches 1–10 ppt when γ = 0.03 (Fig. 5c) and 2–50 ppt when γ = 0 (Fig. 5a). Similar to [DMA], [MMA] and [TMA] decrease quickly with altitudes, down to <0.1 ppt above ∼800 mb (Figs. 4b, d, 5b, d).

Figure 6 compares the simulated [MMA], [DMA], and [TMA] with measurements at the sites listed in Table 2 and marked in Fig. 1. The modeling results under four γ values (0.03, 0.01, 0.001, and 0) are given. It should be noted that the model results in Figs. 2–5 are annual mean values, while most of the methylamine data are from various field measurements that lasted everywhere from less than 1 day to a few months (Table 2). Owing to large seasonal variations, the model results corresponding to the months of the observations are used for comparisons with observations in Fig. 6. The vertical bars in Fig. 6 (for γ = 0.03 and 0 cases only) define the simulated ranges of monthly mean concentrations of methylamines.

Based on very limited measurements currently available (Table 2), [DMA] in urban areas is smaller than those in rural and coastal areas while [MMA] and [TMA] in these regions do not show a systematic difference. Over the Arabian Sea, measurements of two periods differ by a factor of 5 for [DMA] and by a factor of 10 for [TMA], indicating a large temporal variation in [DMA] and [TMA] concentrations at some locations. It is clear in Fig. 6 that the model predictions of methylamines are substantially lower than the limited observed values available, with normalized mean bias (NMB) ranging from -57 (γ = 0) to -88 % (γ = 0.03) for MMA and TMA, and from -78 (γ = 0) to -93 % (γ = 0.03) for DMA. [MMA] and [TMA] are relatively closer to observed values, especially when γ < ∼0.001. It appears that the simulated [DMA] values are close to the measured values for the three urban sites (A, B, and C) (Fig. 6b).

It is unclear how much the underestimation is associated with the spatial (2∘ × 2.5∘ model grid box with a depth of ∼150 m versus measurements at given sites near the surface) and temporal (model monthly mean versus measurements of a few days to a few weeks) average. The seasonal variations of simulated concentrations of methylamines are generally within a factor of 2–5. As we can see in Figs. 2–5 and Table 1, concentrations of methylamines are roughly proportional to the emission fluxes. Methylamine emissions in certain regions could be much larger, while in other regions much lower than those shown in Fig. 1. Due to the short lifetime of these amines, long-range transport is not important, and thus the observed amine concentrations (together with their lifetime) can be used to estimate the emission strength in the region. If the measurements are representative and reflect the real methylamine concentrations, the underprediction of methylamines by 1–2 orders of magnitude in some sites (Fig. 6) may indicate that the methylamine emissions in these regions are 1–2 orders of magnitude larger than those assumed in this study (Fig. 1, Table 1), at least around the sites of the measurements. Apparently, long-term measurements of amines at more locations are needed to evaluate the potential importance of amines.

As a result of the substitution by one or more organic functional groups, amines have a stronger basicity than ammonia and may participate in new particle formation in the atmosphere. To integrate the various processes controlling amine concentrations and understand the concentrations of key amines and their spatiotemporal variations in the atmosphere, we simulate the global distributions of amines in the air with a global chemistry transport model (GEOS-Chem), focusing on methylamines (MMA, DMA, and TMA) in this study.

We showed that gas phase oxidation and aerosol uptakes are dominant sinks for methylamines. The uptake sinks are sensitive to γ when γ > ∼0.001 and the oxidation becomes more important when γ is smaller. The oxidation only (i.e., no aerosol uptake) leads to a methylamine lifetime of 5–10 h in most part of low and middle latitude regions, from the surface to the upper troposphere. The oxidation lifetime is relatively longer (>10–50 h) over the high latitude regions due to low OH concentrations there. The aerosol uptake with γ of 0.03 reduces the lifetime of methylamines by ∼30 % over oceans and much more over the major continents, resulting in methylamine lifetimes as short as 1–2 h over central Europe, eastern Asia, and eastern US. Depending on γ values, [DMA] in the surface layer over major continents is in the range of 0.1–2 ppt when γ = 0.03 and 0.2–10 ppt when γ = 0, and much smaller than over oceans are polar regions (<0.01–0.05 ppt). Compared to [DMA], [MMA] is generally a factor of ∼2.5 higher while [TMA] is a factor of ∼8 higher. Concentrations of methylamines decrease quickly with altitudes, with zonally averaged values dropping below 0.1 ppt above the boundary layer.

The simulated concentrations of methylamines are substantially lower than the limited observed values available, with normalized mean bias (NMB) ranging from -57 (γ = 0) to -88 % (γ = 0.03) for MMA and TMA, and from -78 (γ = 0) to -93 % (γ = 0.03) for DMA. The underestimation cannot be explained by the possible uncertainty in the uptake coefficients and long-range transport. The concentrations of methylamines are roughly proportional to their emission fluxes, and thus the model underprediction by 1–2 orders of magnitude at some sites may indicate that the methylamine emissions in these regions are 1–2 orders of magnitude higher than those assumed in this study. It should be noted that methylamine measurements are very limited and subject to large uncertainty as well because of their low concentrations and short lifetime.

Amines have been suggested to be the most likely compound to sequester carbon dioxide and there exists concern about the potential impacts of substantial increases in future amine emissions (Nielsen et al., 2012). Our study indicates that the impact of amine emissions from carbon sequestration is likely to be local rather than global as a result of their short lifetime. The low concentrations of amines away from source regions (<0.1–1 ppt) suggest that the impact of amines on global new particle formation may be quite limited. Nevertheless, amines can exceed a few ppt over the main source regions and thus may substantially enhance new particle formation. It should be noted that about 150 amines have been identified in the atmosphere and amines of different kinds are likely to have different abilities in stabilizing prenucleation clusters. It is important to identify those amines with abundant concentrations in the atmosphere and study their ability in enhancing new particle formation. We would like to emphasize that the present global simulations of methylamines are subject to uncertainties associated with emissions, uptake coefficients, and chemistry. Further laboratory study, field measurement, and numerical modeling are needed to advance our understanding of spatiotemporal distributions of key amines and to evaluate their contributions to new particle formation and growth in the global atmosphere.