Dry deposition fluxes were calculated following the scheme and parameters of Wesely (1989) and Zhang et al. (2003). As shown in Fig. 1, the resistance model applied to characterize the dry deposition process includes the aerodynamic resistance (Ra), quasi-laminar resistance (Rb) and surface resistance (Rc). The basic equations for the flux calculations are as follows: 1Fgrd=-VdXg×10-2 2Vd=1Rgrd=1Ra+Rb+Rc, where Fgrd represents the gas deposition fluxes on various ground surfaces (mol m-2 s-1); Vd represents the deposition velocity (cm s-1); [Xg] is the averaged gas concentration (mol m-3); and Rgrd is the total resistance in the dry deposition process (s cm-1), composed of Ra, Rb and Rc. The detailed equations and parameterization scheme for the determination of Ra, Rb and Rc are provided in the Supplement. A neutral meteorological condition was assumed in the calculation. We present the key input parameters and the calculated Vd in Tables  S1 and S2, respectively.

The net flux of gas X from the gas phase to the condensed phase (Jnet, mol m-2 s-1) for one aerosol particle can be expressed as Eq. (3) under (quasi-) steady-state conditions (Pöschl et al., 2007): 3Jnet=ωγeff4Xg The effective uptake coefficient, γeff, represents the number of gas molecules taken by the aerosol particle divided by the number of those impacting onto the particle surface (Pöschl et al., 2007); ω is the mean thermal velocity (m s-1) – we use a typical value of 300 m s-1 in this study; [Xg] is the averaged gas concentration far away from the aerosol surface (mol m-3). 41γeff=1Γg+1α+1Γb As shown in Fig. 1, resistance models have been widely applied to quantify the mass transfer of gases to aerosol particles. For gas uptake on liquid droplets, following the resistance model as described by Eq. (4), the overall resistance 1/γeff is composed of three resistor terms due to gas diffusion (1/Γg), interfacial mass transfer (1/α) and bulk diffusion and reaction (1/Γb) (Pöschl et al., 2007). The conductance of gas diffusion is commonly calculated based on Γg=8Dgω-1dp-1, where Dg is the diffusion coefficient of gas X in the gas phase (m2 s-1), and dp represents the aerosol particle diameter. For large particles and very fast uptake processes, the gas diffusion process can be a limiting factor for the overall uptake (Tang et al., 2014a). For atmospheric aerosols with a diameter of ∼ 0.2 µm, the related gaseous uptake tends to be limited by the free molecular collision rate (uptake rate →ωαA[Xg]/4) (Jacob, 2000). Thus, in the following analyses, we mainly focus on the discussion of γeff, and neglect the diffusion resistance in the gas phase.

Given a mixing height of h and an aerosol surface area density of A (particle surface area per unit volume of air, µm2 cm-3), the total uptake flux of gas X by aerosols (Faer, mol m-2 s-1) is 5Faer=JnetAh=ωγeff4Ah[Xg]×10-6, where 10-6 is the unit conversion factor. We summarized the measured uptake coefficients for a variety of gas species and aerosol types at both the initial state and the steady state in Table 1 (details in Tables A1–A4). These coefficients are mainly derived from measured values in the literature, reviewed data from the IUPAC (International Union of Pure and Applied Chemistry) “Task Group on Atmospheric Chemical Kinetic Data Evaluation” (Crowley et al., 2010, 2013; Ammann et al., 2013; available at http://iupac.pole-ether.fr/, last access: January 2019) and the NASA JPL (Jet Propulsion Laboratory, Burkholder et al., 2015) (see references in Tables A1–A4). As we focus on the PBL, γeff values measured at room temperatures (∼ 298 K) are mainly presented. Gas uptakes at very low temperatures (e.g., in the polar region or stratosphere) are outside the scope of this study and should be explored in future work.

Although the initial and steady-state uptake coefficients are listed, it should be noted that the values at the initial state may not be appropriate for direct application in chemical transport models (CTMs) considering the subsequent surface saturation and depletion of reactants for several cases (e.g., on mineral dust and soot; Ndour et al., 2009; Stephens et al., 1986; Ammann et al., 1998; Kalberer et al., 1999). In general, the upper limit and lower limit are determined based on those derived using the geometric surface and the BET (Brunauer–Emmett–Teller) surface, respectively. Preferences are given to those measured at steady state using ambient aerosols, or values recommended by the IUPAC group with relatively high reliability. As shown in Table A1, variances of more than 3 orders of magnitude are found for SO2 and O3 uptake on mineral dust depending on the gas concentration and aerosol components (Michel et al., 2002, 2003; Mogili et al., 2006; Ullerstam et al., 2002, 2003; Li et al., 2006). Large discrepancies also exist for SO2 and HNO3 uptake on soot (Longfellow et al., 2000; Saathoff et al., 2001; Xu et al., 2015). For H2O2, limited measurements of γeff have been conducted for aerosols apart from mineral dust.

To help the evaluation, we define an uptake coefficient at equivalent flux γeqv. Here, γeqv is the effective uptake coefficient on aerosols when the ground flux equals the aerosol flux within the PBL. When γeff>γeqv, the aerosol surfaces are more important than the ground surfaces regarding the gas uptake and vice versa. By letting Fgrd equal Faer, we can derive the expression for γeqv as follows: 6γeqv=43VdAh×102 and at a typical mixing height of 300 m, we have 7γeqv=Vd2.25A According to Eq. (6), γeqv is proportional to Vd, and is inversely proportional to the aerosol surface area density and the mixing height. We calculated a series of γeqv values for a variety of gas species, land use categories, seasons, aerosol surface area densities (A) and mixing heights (h).

As defined, γeqv reflects the relative importance of gas uptake on aerosols compared to those on the ground surfaces. Larger γeqv indicates a higher probability for gases to deposit on the ground rather than on aerosols for further chemical reactions on surface and bulk, and vice versa. Low dry deposition velocities and high loadings of aerosols providing large amounts of surface reaction sites can benefit gas uptake on aerosols. The derived γeqv and γeff values from laboratory measurements are compared in Sect. 3.

Under typical conditions (typical A by land use, h=300 m, as illustrated above), γeqv values for O3 between 9.2×10-5 and 2.2×10-3 are determined, with the lowest value in urban areas and the highest value in the Amazon forest. The extended range is 1.4×10-5–3.8×10-2, which varies with particle area densities and mixing heights (Fig. 3). There are overlaps between γeqv and γeff for liquid organic aerosols (OAs) among all investigated typical environments, and other types of aerosols under favorable circumstances for aerosol uptake in urban areas. γeff values lie below γeqv values for other combinations of aerosol types and land use categories.

We can only expect comparable uptake between ground and aerosol surfaces of mineral dust, soot, solid organic aerosol and SSA at high aerosol loadings in urban areas (e.g., A=1400 µm2 m-3, Beijing) and/or high mixing layers (e.g., h=1.0 km). Combined with the measured uptake coefficients which lie in the range of 1.0×10-7 to 1.6×10-4 for soot, 1.1×10-5 to 3.0×10-3 for liquid organic aerosols and 1.3×10-6 to 1.0×10-4 for SSA, we can expect high uptake fluxes of O3 on these three kinds of aerosols when the corresponding γeff value is larger than 10-4 for ground surfaces other than urban.

Complexity comes from the organic aerosols, of which the phase state has a large impact on the uptake and is subject to the temperature, relative humidity and particle size (see Fig. 3) (Virtanen et al., 2010; Cheng et al., 2015). For liquid organic aerosols, the measured γeff values show large variances from 10-5 to 10-3 and corresponding γeqv values fall into this range, demonstrating that O3 uptake on aerosols is comparable to that on the ground. Thus, multiphase reactions of O3 on liquid organic aerosols should be included in atmospheric models. This is also consistent with the findings of Mu et al. (2018), which demonstrate the importance of the phase state of aerosols in multiphase reactions and the transport of polycyclic aromatic hydrocarbons to improve the model performances at both regional and global scales.

Shiraiwa et al. (2017) show the global map of the SOA (secondary organic aerosol) phase state at the Earth's surface. SOA in southern China, the Amazon forest and South Africa are mainly in the liquid phase within the PBL. For these regions, the comparable uptake fluxes for O3 on liquid organic aerosols compared with the dry deposition demonstrate the importance of aerosol uptake. Dry deposition is one of the major sinking pathways for O3 (Ganzeveld and Lelieveld, 1995). The uptakes of O3 by aerosols are expected to contribute comparable sink fluxes to dry deposition regionally. Inclusion of the O3 uptake by organic aerosols in these regions will increase the deposition rate of O3 on aerosols, affect its lifetime, and further affect the fate of HOx and NOx through chemical reactions in the gas phase.

For NO2, γeqv values are generally above the upper limit of γeff in urban, agricultural and forest environments, as shown in Fig. 4, demonstrating that ground surfaces are of greater importance than aerosols. Overlaps are found for SSA on various land use types and also for liquid organic aerosols in the urban environment.

NO2 tends to deposit on the ground surface instead of on mineral dust particles, soot and solid organic aerosols. As reviewed in Tables A1–A3, the effective uptake coefficient of NO2 on these three kinds of aerosols are at magnitudes of <10-6 under steady-state conditions. For A values ranging from 46 µm2 cm-3 (Amazon) to 1050 µm2 cm-3 (Wangdu) and at a mixing height of 300 m, γeqv values of NO2 lie between 1.4×10-5 and 1.3×10-3, and are 1–3 orders of magnitude larger than γeff on these three kinds of aerosols. Increasing the PBL mixing height and aerosol surface area may reduce γeqv values by ∼ 1–2 orders of magnitude, but they are still above the measured γeff values at steady state.

The reactive uptake coefficients of NO2 by SSA were quantified in the range of 10-6 to 10-4, demonstrating the ability of ambient sea salt aerosols to take in chemical species like NO2 (Harrison and Collins, 1998; Yabushita et al., 2009; Ye et al., 2010). The high uptake coefficients observed for SSA (6.0×10-7–3.0×10-4) are probably attributed to the reactions of Cl- with dissolved NO2 in the aqueous phase (Msibi et al., 1993; Harrison and Collins, 1998; Yabushita et al., 2009). The overlapped values of γeqv and γeff show that the NO2 uptake by SSA is comparable to the uptake by the land surface or water bodies in coastal areas; therefore, it should be taken into account in atmospheric models.

Another important process is the NO2 uptake on liquid organic aerosols (γeff in the range of 2.2×10-7–7.0×10-6) in urban areas of high A. As shown in Fig. 4, the lower limit of γeqv in urban is ∼2.2×10-6, which lies within the range of γeff. The uptake coefficients of NO2 on pure water are estimated to be around 10-7–10-6, driven by low solubility and slow hydrolysis rates (Kleffmann et al., 1998; Gutzwiller et al., 2002; Ammann et al., 2005; Komiyama and Inoue, 1980). Harrison and Collins (1998) reported a high γeff of 5.4–5.8×10-4 for NO2 uptake on ammonium sulfate aerosols at high relative humidity (RH; RH = 50 %, 85 %). The presence of reactants such as inorganics of HSO3- or phenolic compounds in aqueous aerosols can promote the uptake significantly via chemical reactions with dissolved NO2 to 10-5–10-4 (Msibi et al., 1993; Lee and Tang, 1998; Spindler et al., 2003; Ammann et al., 2005; Yabushita et al., 2009; Su et al., 2008a; Cheng et al., 2016). Multiple measurements and modeling work have also pointed out that the high alkalinity of aqueous aerosols is key to promote the reactions and further increase the NO2 uptake rates (Ammann et al., 2005; Herrmann et al., 2015; Cheng et al., 2016). Therefore, the NO2 uptake on alkaline aqueous aerosols containing organic/inorganic reactants is competitive in the urban atmosphere, and should be addressed in detail in models. In the Amazon forest, where A is too low (46 µm2 cm-3), corresponding to a γeqv value of the order of 10-3, even a high γeff value of 10-4 is not sufficient to compete with the uptake by the ground surfaces.

In summary, the NO2 uptake coefficients on liquid aerosol droplets can vary by 3 orders of magnitude with aerosol composition (10-7–10-4). On liquid organic aerosols and sea salt aerosols, the uptake can reach up to 10-6–10-4 via chemical reactions (Abbatt and Waschewsky, 1998; Ammann et al., 2005; Yabushita et al., 2009), which is significantly larger than the uptake on pure water of 10-7–10-6 (Lee and Tang, 1988; Kleffmann et al., 1998; Gutzwiller et al., 2002). For liquid ammonium sulfate aerosols, discrepancies with 2 orders of magnitude (10-6–10-4) in γeff values are found, although the reasons for this are currently unknown (Harrison and Collins, 1998; Tan et al., 2016). Considering these variances, aerosol components are important to parameterize the γeff in atmospheric models.

The calculated γeqv values of SO2 vary between 1.0×10-4 and 2.1×10-3 for land surfaces and 1.7×10-4 above water bodies under typical conditions. As shown in Fig. 4, the SO2 uptake by mineral dust is comparable to the ground uptake in urban areas, and under favorable conditions over agricultural land and water bodies. For soot, aerosol uptake is magnitudes lower than those on the ground (γeqv≥γeff); thus, this process is unimportant for SO2. For SSA, γeff values of 3.2×10-3–1.7×10-2 have been reported for SO2 at an aerosol pH of 5.4–6.6, which is high enough to compete with dry deposition over most environments (Gebel et al., 2000). Additional reactions of SO2 and O3 in alkaline solutions are found to promote the SO2 uptake and form sulfate on SSA at first stage (Laskin et al., 2003). However, aerosol acidification due to production of H+ has been suggest to quickly suppress the oxidation process in the real world (Alexander et al., 2005). We suggest including both the SO2 uptake on SSA and the aerosol acidification process in models.

The extended range of γeqv is 1.6×10-5–1.6×10-3, 5.5×10-5–2.8×10-3 and 1.9×10-5–1.9×10-3 for urban areas, agricultural land and water bodies, respectively. The γeff of mineral dust falls in this range under high aerosol loadings or high mixing heights. The wide range of γeff values for mineral dust (1.5×10-8 to 6.3×10-4) is a big challenge regarding its application in models, because it can be affected by the presence of oxidants, the phase state, the components of the tested dust and the use of surface area in calculation (Huss et al., 1982; Ullerstam et al., 2003; Li et al., 2006; Alexander et al., 2009; Zhang et al., 2018). We further discuss the SO2 uptake on mineral dust under different conditions in the following.

Under dry conditions (as reviewed in Table A1), γeff values of the order of 10-7–10-4 are measured (Goodman et al., 2001; Usher et al., 2002; Ullerstam et al., 2003; Adams et al., 2005; Li et al., 2006). IUPAC recommends an averaged value of 4×10-5 for atmospheric modeling, based on measurements using airborne aerosols (Usher et al., 2002; Adams et al., 2005).

In environments with a high RH, water can enhance or inhibit the uptake by affecting reactive sites, and this effect varies with experimental conditions (Huang et al., 2015; Zhang et al., 2018). Conversely, the uptake rate can be improved by several factors and/or aqueous chemical reactions, such as the presence of O3, H2O2 and transition metal ions (TMIs), which strongly depend on the aerosol pH (Jayne et al., 1990; Li et al., 2006; Cheng et al., 2016; Zhang et al., 2018). The initial γeff value of SO2 on pure water can reach as high as 10-3–0.1, varying with pH (Gardner et al., 1987; Worsnop et al., 1989; Jayne et al., 1990; Ponche et al., 1993). Depending on aerosol pH and oxidant concentrations, the regimes of SO2 uptake and sulfate formation may transit from a TMI- or H2O2-dominated regime to a NO2- or O3-dominated regime (Cheng et al., 2016). In this case, the SO2 uptake on aqueous aerosols is expected to play a dominant role over dry deposition under specific circumstances, such as haze events (He et al., 2014; Cheng et al., 2016), which should be quantified by combining in situ measurements and atmospheric modeling.

As shown in the examples in Table S4, several model schemes adopt a γeff value of ∼10-4 (Liao and Seinfeld, 2005; K. Wang et al., 2012), around 1 order of magnitude higher than the measured values under low RH conditions (Usher et al., 2002; Ullerstam et al., 2003; Adams et al., 2005; Li et al., 2006). For example, in Liao and Seinfeld (2005), γeff is 3.0×10-4 for RH < 50 %, and 0.1 for RH ≥ 50 % (see Table S4 with references). Under low RH conditions, the uptake coefficient commonly used in models is based on the dry deposition measurement of SO2 on calcareous soils, cements and Fe2O3, rather than on laboratory measured γeff values recommended by IUPAC. The reason for this divergence is unclear, and we are in favor of using the IUPAC recommended γeff (e.g., Zhu et al., 2010, as shown in Table S4). The high uptake coefficient in models under high RH conditions is based on two assumptions: fast oxidation of SO2 by O3 in the aqueous phase, and high alkalinity in the dust aerosols. Thus, this γeff value should be applied with the caveat that these prerequisites have been fulfilled, especially when extending it for other type of aerosols (Zheng et al., 2015).

N2O5, HNO3 and H2O2 demonstrate their high uptake ability on atmospheric aerosols, as shown in Fig. 5. For N2O5, the similar or higher values of γeff over γeqv demonstrate that the multiphase uptake by all types of aerosols is as important as or even more important than dry deposition. The N2O5 uptake by aerosols has been widely included in models (Bauer et al., 2004; Liao and Seinfeld, 2005; K. Wang et al., 2012). The uptake of HNO3 and H2O2 by mineral dust and HNO3 by SSA are important given the overlap between γeff and γeqv; thus, this uptake should also be characterized in atmospheric models in detail.

For N2O5, the measured uptake coefficients are 4.8×10-3–0.20 for mineral dust, 4.0×10-5–6.3×10-3 for soot and 6.4×10-3–3.9×10-2 for SSA, which are comparable to or 1–2 orders of magnitude higher than the calculated γeqv values of 9.3×10-4–2.1×10-2 under typical conditions (details in Tables A1–A4). For other kinds of aqueous aerosols, e.g., ammonium sulfate aerosols with high RH, N2O5 can also be taken up very efficiently with γeff values of 10-3–10-2 (Kane et al., 2001; Schötze and Herrman, 2002; Hallquist et al., 2003; Badger et al., 2006). The importance of N2O5 and HNO3 uptake by aerosols has been sufficiently addressed in previous studies (Evans and Jacob, 2005; Liao and Seinfeld, 2005; Stadtler et al., 2018). In current CTMs, γeff of N2O5 is explicitly calculated as a function of temperature and RH, for which the relation was determined from laboratory experiments (Kane et al., 2001; Bauer et al., 2004; Liao and Seinfeld, 2005).

The extended ranges of γeqv for HNO3 is 1.5×10-4–1.5×10-2 (urban), 1.5×10-4–7.7×10-3 (agricultural land), 4.2×10-4–3.7×10-1 (Amazon forest) and 7.0×10-4–7.0×10-2 (water), which are within or below the range of γeff for mineral dust and SSA. The higher γeff of 1.0×10-3 to 0.21 for mineral dust and of 5.0×10-4 to 0.25 for SSA demonstrated the more important role of aerosol uptake than of the ground surfaces. The uptake of HNO3 on soot and solid organic aerosols appears to be less important. The HNO3 uptake on mineral dust has been implemented in current models with an uptake coefficient of 0.1, or between 1.1×10-3 and 0.2, which is consistent with the range of experimentally determined γeff values reviewed in this study (Liao and Seinfeld, 2005; K. Wang et al., 2012).

The study of the uptake of H2O2 by aerosols is rather limited compared with the other aforementioned trace gases. The reported γeff values on dust and ambient aerosol samples suggest that aerosol uptake is more important than that by the ground surface. The measured uptake coefficients of H2O2 on mineral dust are in the range of 1.0×10-5–9.4×10-4 and overlap with the calculated γeqv of 1.5×10-4–3.0×10-3 under typical conditions. Ambient aerosols collected in urban areas show similar γeff values of H2O2 (8.1×10-5–4.6×10-4) to mineral dust (Wu et al., 2015). The aerosol chemistry of H2O2 in the troposphere is complex and unclear (Liang et al., 2013; Li et al., 2016). In some cases, a net emission of H2O2 from aerosol surfaces has been speculated instead of an uptake or adsorption as a result of HOx radicals cycling (Liang et al., 2013; Li et al., 2016). Most models only parameterize the H2O2 uptake by dust particles (Dentener et al., 1996; K. Wang et al., 2012). The uptake by other aerosol types has not been considered due to limited experimental data. Thus, more laboratory kinetic measurements are needed. As ambient aerosol samples show a γeff similar to that of dust particles (Wu et al., 2015; Pradhan et al., 2010ab; Zhou et al., 2016), we suggest adopting the γeff of dust particles and applying it to all aerosol types before new kinetic data become available.

Notably, there is a large variability in the reviewed γeff of SO2 uptake by dust particles (as discussed in Sect. 3.2). For SO2 uptake by dust particles, differences of more than 3 orders of magnitude are found for its uptake by mineral dust (10-8–10-4, steady state), which may be attributed to several factors such as the experimental particle substrates, co-existing oxidants (O3, H2O2 and NO2), RH, measurement techniques and the surface area used in data processing (Ullerstam et al., 2003; Li et al., 2006; Huang et al., 2015). For example, a γeff of 1.6×10-4 was derived for SO2 uptake on Al2O3 powder (Usher et al., 2002). The uptake coefficient was reduced by 1 order of magnitude to 1.6–6.6×10-5 using ambient aerosols of Chinese loess/Saharan dust (Usher et al., 2002; Ullerstam et al., 2003; Adams et al., 2005), indicating that the particle substrate is key in investigating SO2 uptake. Similarly, through cross-comparisons between the other different investigations shown in Table A1, we anticipate that the above factors can all contribute to this large discrepancy. As recommended by IUPAC, an uptake coefficient of 4×10-5 based on airborne measurements is suggested for use in models under low RH conditions. For high RH, we suggest determining γeff using information on aerosol pH due to the high correlation between these factors, as illustrated in Sect. 3.3.

For NO2 uptake on liquid aerosol droplets, differences of 3 orders of magnitude are found (10-7–10-4), which vary significantly with aerosol composition. On pure water, the uptake is measured at 10-7–10-6 (Lee and Tang, 1988; Kleffmann et al., 1998; Gutzwiller et al., 2002). On liquid organic aerosols and sea salt aerosols, the uptake can be effectively accelerated to 10-6–10-4 via multiphase reactions (Abbatt and Waschewsky, 1998; Ammann et al., 2005; Yabushita et al., 2009). For ammonium sulfate aerosols, large discrepancies of 10-6–10-4 for the initial γeff are found, although the reasons for this are currently unknown (Harrison and Collins, 1998; Tan et al., 2016). Based on the reviewed measurements, we suggest using a relatively high uptake coefficient (∼10-4) for aqueous aerosols containing reactants, and a lower value (<10-6) for other cases.

Measurements of the effective uptake coefficients revealed the instantly fast uptake at the initial state and the gradual decline due to the saturation of surface reaction sites and loss of reactive substances (Hanisch and Crowley, 2003). The uptake at the initial state can be orders of magnitude faster than that at the steady state for aerosols (see Tables 1 and A1–A4). The timescale for reaching surface saturation/equilibrium is dependent on the reaction system. For a gas–aqueous particle surface, the timescale to establish equilibrium for the investigated species is less than 1 s (Seinfeld and Pandis, 2006, 554–557). For dust particles, it can take hours for complete saturation (Judeikis et al., 1978; Goodman et al., 2001). Fine particles with diameters < 10 µm have lifetimes of several days in the atmosphere (Prospero, 1999; Lee et al., 2009). Thus, using uptake coefficients at steady state maybe more representative in models, unless we can assume that the uptake process is not limited by surface accommodation and reactions (like HNO3; Goodman et al., 2000), typically when the gas concentration is low enough that the surface passivation is negligible compared with the lifetime of aerosols in the atmosphere (Hanisch and Crowley, 2003). Gas uptake on fresh aerosols may reach or even surpass the level of the ground near emission sources. Using a uniform uptake coefficient in atmospheric models may not be enough to reflect the deactivation process of the multiphase gas uptake during aerosol aging, considering the large range of γeff values varying with time.

In addition, the γeff values are measured and reported based on the geometric surface and/or the BET surface. Differences of more than 3 orders of magnitude are derived depending on whether or not the pores within the microstructure of solid aerosol surface are considered (see Table A1). Using the same method to calculate the available surface area may reconcile these differences (Tang et al., 2017). In this study, γeff values with a revised BET surface are generally used as the lower limit, and those using the geometric surface are used as the upper limit. Whether or not the BET area is used as a correction in the calculation of γeff or not remains a discrepancy when applied in models (Hanisch and Crowley, 2001a, b; Underwood et al., 2001ab). This discrepancy from measurements may come from the differences in experimental samples (airborne particles vs. powder). To solve this issue, more studies on the reactive surface area for ambient aerosols are needed to guide the data processing and model parameterization.

We can conclude that the phase state is a crucial factor influencing the uptake rates. The uptake rates of O3 and NO2 in liquid organic aerosols are 1–3 orders of magnitude higher than on solid/semi-solid surfaces. In regions with high RH conditions and sufficient sources of organic compounds (e.g., the Amazon forest and southern China), the gas uptake is anticipated to have a considerable effect on concentrations. The effect is yet to be evaluated in combination with further model simulations.

Measurement of uptake by ambient aerosols is crucial to reconcile laboratory experiments and modeling results, especially for gases that have undergone limited investigation (e.g., H2O2). Currently limited work has been undertaken to address the uptake of H2O2 by aerosol particles other than mineral dust (Liao and Seinfeld, 2005; Pradhan et al., 2010a, b; K. Wang et al., 2012; Zhou et al., 2016). Because ambient aerosol samples show a γeff value comparable to that of dust particles, we recommend similar γeff values of 1.0×10-5–9.4×10-4 for H2O2 uptake by other types of aerosol, which will lead to a larger sink in the atmospheric budget of H2O2.

Considering the complexity of multiple factors affecting the uptake rates, such as temperature, RH, gas concentration, aerosol pH and aerosol state (fresh or aged), establishing a look-up table for γeff based on the available factors mentioned above should be a feasible way to implement the gas uptake processes in atmospheric models (Mu et al., 2018).

There are limitations to the comparisons conducted in this study. We use a unified thermal velocity (300 m s-1) for all gases, which will introduce positive biases of +4 % to +30 % for O3, NO2, SO2, HNO3 and H2O2, and a negative bias of -24 % for N2O5 in calculations of γeqv at the same temperature. The ambient parameters to calculate the dry deposition velocities (temperature and radiation) refer to the standard meteorological database for construction in northern China (Zhang, 2004), which may introduce uncertainties for analyses of other areas. In addition, the variability of aerosol surface area in each environment can also contribute to the variability of γeqv. We mainly focused on the uptake fluxes at room temperature (∼ 298 K). The gas uptakes at very low temperatures (e.g., in the polar region and stratosphere) are outside the scope of this study but should be further explored concerning its potentially large impact. The real ambient multiphase processes are much more complex than the laboratory measurements; nevertheless, they use airborne aerosols. Ambient online measurements of γeff will favor the model parameterization and improve our understanding of the multiphase processes within the PBL in the real world (Li et al., 2019). Moreover, more gaseous and aerosol species such as VOCs and bioaerosols should also be investigated (Zhou et al., 1996; Wagner et al., 2002; Fried et al., 2003; Beck et al., 2013; Li et al., 2014, 2016; Ouyang et al., 2016; Liu et al., 2017; Meusel et al., 2017).