Recent Advances in Atmospheric Chemistry of Mercury

Abstract

Mercury is one of the most toxic metals and has global importance due to the biomagnification and bioaccumulation of organomercury via the aquatic food web. The physical and chemical transformations of various mercury species in the atmosphere strongly influence their composition, phase, transport characteristics and deposition rate to the ground. Modeling efforts to evaluate the mercury cycling in the environment require an accurate understanding of atmospheric mercury chemistry. We focus this article on recent studies (since 2015) on improving our understanding of the atmospheric chemistry of mercury. We discuss recent advances in (i) determining the dominant atmospheric oxidant of elemental mercury (Hg⁰); (ii) understanding the oxidation reactions of Hg⁰ by halogen atoms and by nitrate radical (NO₃); (iii) the aqueous reduction of oxidized mercury compounds (Hg^II); and (iv) the heterogeneous reactions of Hg on atmospherically-relevant surfaces. The need for future research to improve understanding of the fate and transformation of mercury in the atmosphere is also discussed.

1. Introduction

Mercury (Hg) is one of the most toxic metals present globally in the environment [1]. Due to the rather long lifetime of atmospheric mercury, once mercury compounds are released into the atmosphere, they can be transported around the globe. As such, they not only have local impacts but also regional and global implications [2,3].

Hg is a Group IIB transition metal with an atomic number of 80 and a closed shell electronic configuration (5d¹⁰ 6s²). Elemental mercury (Hg⁰) is the only liquid metal at room temperature and pressure. Mercury has been widely used in electrochemistry, in optical spectroscopy, in liquid mirror telescopes and also in medicine. However, tragic outbreaks of mercury-induced diseases have occurred in many areas of the world over the years, particularly in Japan and Iraq [4]. Mercury toxicity depends on its chemical species, with methylmercury being highly toxic. Humans and wildlife are exposed to mercury mainly through fish consumption. Exposure to mercury can damage the human nervous system, cause cardiovascular diseases in adults, impede cognitive development in children, as well as affect the reproduction of fish, mammals and birds ([5] and references therein).

Atmospheric deposition has been identified as a significant pathway for Hg transport into the aquatic and soil environment. It has been estimated that 5500–8900 tons of mercury are emitted per year into the atmosphere, with three major natural, anthropogenic and re-emitted sources [2]. Natural sources of Hg emission include volcanoes, soils, forests, natural waters, and the largest anthropogenic source of Hg emission is coal burning in coal fueled power plants [6]. Previously-deposited mercury can be re-emitted to the atmosphere through a series of physical, chemical and biological transformations in the environment [7,8]. Large uncertainties remain in existing emission inventories, particularly for natural and re-emission sources [9].

Figure 1a illustrates a simplified picture of mercury transformations in the environment, particularly on various environmental surfaces, and Figure 1b focuses on the atmospheric chemistry of mercury. In the atmospheric environment, mercury presents predominantly in oxidation states of 0 and +2 [10]. Atmospheric mercury exists in the gas phase, in atmospheric droplets, or in the particulate phase. Mercury speciation significantly affects the rates of dry and wet deposition and subsequently, the atmospheric lifetimes of different Hg species [11]. For example, gaseous elemental mercury (Hg⁰) has been estimated to have an atmospheric lifetime of about one year [4,12], which render sit subject to long-range transportation across the globe. However, atmospheric oxidation of Hg⁰ to HgBr₂ will make mercury more subject to deposition, due to the low solubility of Hg⁰ and high solubility of HgBr₂ in atmospheric droplets. The deposition of oxidized mercury can make Hg potentially become available to biota [4]. Field studies have demonstrated the fast oxidation of Hg⁰ to Hg^II in the Arctic and Antarctic regions after polar sunrise, a phenomenon known as Arctic mercury depletion events (AMDEs). The explanation of this fascinating phenomenon will rely on fully understanding the redox reactions of atmospheric Hg [13]. On the other hand, atmospheric Hg models are essential tools for interpreting field observations and evaluating pollution control policies [10]. The further advancement of atmospheric Hg modeling depends on accurate kinetic data on the oxidation and reduction reactions of various mercury species in the atmosphere. However, many important processes involved in the transformation and deposition of atmospheric Hg are yet to be identified or quantified [14]. Global modeling studies also suggest that large uncertainties exist in our current knowledge of the global cycling of mercury [15].

The chemistry of atmospheric mercury was previously reviewed in detail by Lin and Pehkonen [16], Ariya et al. [17], Subir et al. [18], Lin et al. [14] and Ariya et al. [4]. Several excellent contributions have been made to our understanding in this area since 2015. The objective of this article is to discuss the current understanding of atmospheric mercury chemistry, with more focus on recent laboratory, theoretical, field and model studies since 2015 on the interconverting reactions between Hg⁰ and Hg^II. Finally, future research is recommended in order to further advance our knowledge of the dynamic transformations of atmospheric mercury.

2. Chemical Redox Pathways in the Gas Phase

The oxidation of Hg⁰ is a very crucial step for the removal of Hg from the atmosphere. Understanding the oxidation reactions of Hg⁰ is important for estimating the atmospheric lifetime of mercury. To date, the proposed oxidation pathways of Hg⁰ in the gas phase include the oxidation of Hg⁰ by O₃, H₂O₂, OH radical, NO₃ radical and by various halogen species [14].

Table 1. Current accumulated knowledge on the chemical redox pathways of Hg in the gas phase.

Gas Phase Reaction	Diluent gas a (T = 298 K)	Rate Coefficient b (cm3 molec−1 s−1)	References
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{\mathbf{-}}\mathbf{+}\mathit{B}{\mathit{r}}_{}\mathbf{\to }\mathit{H}\mathit{g}\mathit{B}{\mathit{r}}_{}$	Air, N2, 1 atm	(3.2 ± 0.3) × 10−12	[19]
	Air, NO, 1 atm	9 × 10−13	[20]
	N/A, 1 atmc	1.01 × 10−12exp(209.03/T)	[3]
	N/A, 1 atm	2.07 × 10−12
	N/A (180–400 K)	1.1 × 10−12(T/298K)−2.37	[21]
	N/A, 1 atm	1.1 × 10−12
	N2, 1atm (243–298 K)	(1.46 ± 0.34) × 10−32 × (T/298)−(1.86 ± 1.49) cm6/molec2/s	[22]
	N2, 1 atm	(3.6 ± 0.9) × 10−13
	Ar, 1 atm (260 K)c	1.2 × 10−12	[23]
	Air, 1 atm	(1.6 ± 0.8) × 10−12	[24]
$\mathit{H}\mathit{g}\mathit{B}{\mathit{r}}_{}\mathbf{+}\mathit{M}\mathbf{\to }\mathit{B}\mathit{r}\mathit{H}\mathit{g}{\mathit{M}}_{}$ M = NO2 or HO2	N/A (220–320 K)	k ([M], T)d	[25]
$\mathit{H}\mathit{g}\mathit{B}{\mathit{r}}_{}\mathbf{+}\mathit{B}{\mathit{r}}_{}\mathbf{\to }\mathit{H}\mathit{g}\mathit{B}{\mathit{r}}_{\mathbf{2}}$	CF3Br, 0.26 atm (397 K)	7 × 10−17	[26]
	N/A, 1 atm (180–400 K)	2.5 × 10−10(T/298K)−0.57	[21]
	N/A, 1 atm	2.5 × 10−10
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}\mathit{C}{\mathit{l}}_{}\mathbf{\to }\mathit{H}\mathit{g}\mathit{C}{\mathit{l}}_{}$	Ar, 0.93 atm (383–443 K)	(3.2 ± 1.7) × 10−11	[27]
	Air, NO, 1 atm	6.4 × 10−11	[20]
	Air, N2, 1 atm	(1.0 ± 0.2) × 10−11	[19]
	N/A, 1 atmc	1.38 × 10−12exp(208.02/T)	[3]
	N/A, 1 atm	2.81 × 10−12
	N2 (243–298 K)	(2.2 ± 0.5) × 10−32 × exp [(680 ± 400)(1/T − 1/298)] cm6/molec2/s	[28]
	N2, 1 atm	5.4 × 10−13
	N2, 1 atm	1.2 × 10−10	[29]
	Air, 1 atm	(1.8 ± 0.5) × 10−11	[24]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}{\mathit{O}}_{\mathbf{3}}\mathbf{\to }\mathit{H}\mathit{g}{\mathit{O}}_{\mathbf{\left(}\mathit{s}\mathbf{\right)}}\mathbf{+}{\mathit{O}}_{\mathbf{2}}$	N/A, 1 atm (293 K)	4.2 × 10−19	[30,31]
	N/A, 1 atm (293 K)	4.9 × 10−18	[31,32]
	Air, 1 atm (293 K)	1.7 × 10−18	[33]
	N2/O2, 1 atm (293 K)	(3 ± 2) × 10−20	[34]
	N2, 1 atm	(7.5 ± 0.9) × 10−19	[35]
	Air, 1 atm	(6.4 ± 2.3) × 10−19	[36]
	N2, 1 atm	(6.2 ± 1.1) × 10−19	[37]
	Air, 1 atm	(7.4 ± 0.5) × 10−19	[38]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}\mathit{O}{\mathit{H}}_{}\mathbf{\to }\mathit{H}\mathit{g}\mathit{O}{\mathit{H}}_{}$	Air, 1 atm	(8.7 ± 2.8) × 10−14	[39]
$\mathit{H}\mathit{g}\mathit{O}{\mathit{H}}_{}\mathbf{+}{\mathit{O}}_{\mathbf{2}}\mathbf{\to }\mathit{H}\mathit{g}{\mathit{O}}_{\mathbf{\left(}\mathit{s}\mathbf{\right)}}\mathbf{+}\mathit{O}\mathit{H}$	N/A, 1 atm (343 K)	(1.6 ± 0.2) × 10−11	[40]
	Air, 1 atm	<1.2 × 10−13	[41]
	N/A, 1 atm (180–400 K)	3.2 × 10−13(T/298K)−3.06	[21]
	N/A, 1 atm	3.2 × 10−13
	Air/N2, 1 atm	(9.0 ± 1.3) × 10−14	[42]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}{\mathit{F}}_{}\mathbf{\to }\mathit{H}\mathit{g}{\mathit{F}}_{}$	N/A, 1 atmc	9.2 × 10−13exp(206.81/T)	[3]
	N/A, 1 atm	1.86 × 10−12
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}{\mathit{I}}_{}\mathbf{\to }\mathit{H}\mathit{g}{\mathit{I}}_{}$	N/A, 1 atm (180–400 K)	4.0 × 10−13(T/298 K)−2.38	[21]
	N/A, 1 atm	4.0 × 10−13
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}{\mathit{F}}_{\mathbf{2}}\mathbf{\to }\mathit{H}{\mathit{g}}^{\mathbf{2}\mathbf{+}}\mathbf{+}\mathit{P}\mathit{r}\mathit{o}\mathit{d}\mathit{u}\mathit{c}\mathit{t}\mathit{s}$	Air, 1 atm	(1.8 ± 0.4) × 10−15	[36]
	N2, 1 atm	≤(1.27 ± 0.58) × 10−19	[43]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}\mathit{C}{\mathit{l}}_{\mathbf{2}}\mathbf{\to }\mathit{H}\mathit{g}\mathit{C}{\mathit{l}}_{\mathbf{2}}$	Air, N2, 1 atm	(2.6 ± 0.2) × 10−18	[19]
	Air, 1 atm	(2.5 ± 0.9) × 10−18	[36]
	N2, 1 atm	4.3 × 10−15	[29]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}\mathit{B}{\mathit{r}}_{\mathbf{2}}\mathbf{\to }\mathit{H}\mathit{g}\mathit{B}{\mathit{r}}_{\mathbf{2}}$	Air, N2, 1 atm	<(0.9 ± 0.2) × 10−16	[19]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}\mathit{C}\mathit{l}{\mathit{O}}_{}\mathbf{\to }\mathit{H}\mathit{g}\mathit{C}\mathit{l}{\mathit{O}}_{}$	N2, 1 atm	1.1 × 10−11	[29]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}\mathit{B}\mathit{r}{\mathit{O}}_{}\mathbf{\to }\mathit{H}\mathit{g}\mathit{B}\mathit{r}{\mathit{O}}_{}$	Air, NO, 1 atm	(3.0–6.4) × 10−14	[20]
	N2, 1 atm	(1–100) × 10−15	[44]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}\mathit{N}{\mathit{O}}_{\mathbf{3}}\mathbf{\to }\mathit{H}\mathit{g}{\mathit{O}}_{}\mathbf{+}\mathit{N}{\mathit{O}}_{\mathbf{2}}$	N2, (5–10) × 10−3atm(294 K)	<4 × 10−15	[39]
	Air, 1 atm	<7 × 10−15	[36]
$\mathit{H}{\mathit{g}}^{\mathbf{0}}{}_{}\mathbf{+}{\mathit{H}}_{\mathbf{2}}{\mathit{O}}_{\mathbf{2}}\mathbf{\to }\mathit{H}\mathit{g}{\mathit{O}}_{}\mathbf{+}{\mathit{H}}_{\mathbf{2}}{\mathit{O}}_{}$	N/A, 1 atm	≤4.1 × 10−16	[45]
	N2, N/A (293 K)	<8.5 × 10−19	[46]

^a T = 298 K, unless otherwise stated; ^b The unit is cm³ molec⁻¹ s⁻¹ unless otherwise stated; ^c Temperature range is unknown. ^d$k\left(\left[M\right], T\right)=\left(\frac{k^0\left(T\right){\left[M\right]}^{}}{1+\frac{k^0\left(T\right)\left[M\right]}{k^{\infty }\left(T\right)}}\right){0.6}^{\left(\frac{1}{1+(log{\left(\frac{k^0\left(T\right)\left[M\right]}{k^{\infty }\left(T\right)}\right)}^2}\right)}$ where k⁰(T) and k^∞(T) values are tabulated in [25].

lists the gas-phase oxidation reactions of Hg⁰ and the reported rate coefficients. As shown in Figure 2 in Subiret al. [18], large variations exist in the current reported rate constants of gas-phase oxidation reactions of Hg. Here we discuss recent advances in the understanding of the oxidation of Hg⁰ by halogen atoms and NO₃ radicals, as well as determining the major oxidant(s) in the atmosphere.

2.1. Br-Initiated Oxidation of Hg⁰

The oxidation of Hg⁰ by Br occurs via a two-step process with HgBr as the intermediate. Previous quantum calculations by Dibble et al. [47] demonstrated that BrHg could react with the abundant radicals in the atmosphere to form stable BrHgY compounds where Y is NO₂, HO₂, ClO, or BrO. Previous laboratory studies reported apparent rate coefficients in the range of (3.6 ± 0.9) × 10⁻¹³ to (3.2 ± 0.3) × 10⁻¹² cm³ molec⁻¹ s⁻¹, at 1 atm and 298 K [19,20,22]. Theoretical studies estimated the rate constants to be from 9.8 × 10⁻¹³ to 2.1 × 10⁻¹² cm³ molec⁻¹ s⁻¹, at 1 atm and 298 K [3,21,23]. Theoretical and experimental rate constants for mercury oxidation by Br and Cl atoms at 298 K have previously been compared in detail in Table A.2 by Subir et al. [18].

The first kinetic study of the reactions of BrHg with NO₂ and HO₂ was recently reported by Jiao and Dibble [25], using computational chemistry. Quantum calculations were performed to obtain the rate constants and product yields for the oxidation reactions of BrHg with NO₂ and HO₂. The rate constant for the oxidation of HgBr by NO₂ was larger than that for the oxidation of HgBr by HO₂ (T = 200–320 K). The fate of HgBr replied more on the concentration ratio of [NO₂]/[HO₂] in the atmosphere than the ratio of their oxidation rate constants. While the addition reaction dominated the reaction of HgBy by HO₂, the addition of HgBr by NO₂ was competed by a-reduction reaction to form Hg + BrNO₂ (up to 18% of the oxidation). The reaction product of the oxidation of HgBr by NO₂ was computed to be syn-BrHgONO under atmospheric conditions. Because no experimental studies on the reaction of HgBr of NO₂ and HO₂ have been previously reported in the literature, the rate constants computed in this study are significant for future model and laboratory studies [25].

Sunet al. [24] used a relative rate technique with ethane and propene as references and determined the rate coefficients for Hg⁰ + Br• reactions to be (1.6 ± 0.8) × 10⁻¹² cm³ molec⁻¹ s⁻¹, at 100 kPa and 298 ± 3 K. The determined rate coefficient was in the midst of previously reported rate constants. Measurements using a scanning mobility particle sizer by Sunet al. [24] also showed that Hg products largely existed as wall deposits in agreement with the previous product analysis by Ariyaet al. [19]. It has been suggested that the presence of aerosols in the Aitken mode during the oxidation reaction may be initiated by the vapor nucleation of mercury-containing products.

2.2. Cl-Initiated Oxidation of Hg⁰

Similar to Hg⁰-Br oxidation, the oxidation of Hg⁰ by atomic Cl was believed to occur via a two-step chlorination process with HgCl as the intermediate [19,24,28]. Theoretical calculations indicated that after the initializing step to form the HgCl intermediate, the secondary oxidation of HgCl could be carried out by NO₂, HO₂, ClO, or BrO [47]. Recently, Sunet al. [24] determined the rate coefficients using a relative rate technique with ethane and 2-chloro-propane as references. The rate coefficient for the Hg⁰ + Cl• reaction was reported to be (1.8 ± 0.5) × 10⁻¹¹ cm³ molecule⁻¹ s⁻¹, at 100 kPa and 298 ± 3 K, which was in agreement with three earlier laboratory studies by Horne et al. [27], Spiceret al. [20] and Ariyaet al. [19] (1.0–6.4 × 10⁻¹¹ cm³ molecule⁻¹ s⁻¹). However, the reported rate coefficient was about two orders of magnitude larger than that determined by Donohoueet al. [28] (5.4 × 10⁻¹³ cm³ molecule⁻¹ s⁻¹) and one order of magnitude smaller than that determined by Byunet al. [29] (1.2 × 10⁻¹⁰ cm³ molecule⁻¹ s⁻¹). The determined rate constant was one order of magnitude higher than the only theoretical estimation by Khalizov et al. [2] (2.81 × 10⁻¹² cm³ molecule⁻¹ s⁻¹).

More studies are needed to reduce the uncertainties in the kinetic estimate of the oxidation reactions of atmospheric Hg. To render it more complex, as the existing reactions show the importance of surfaces in catalysis and heterogeneous reactions [4,48], further studies on atmospherically-relevant surfaces are recommended.

2.3. Oxidation of Hg⁰ by NO₃

The experimentally-determined rate constant for the oxidation of Hg⁰ by NO₃ has been previously reported to be <4 × 10⁻¹⁵ cm³ molecule⁻¹ s⁻¹ by Sommaret al. [39] and <7 × 10⁻¹⁵ cm³ molecule⁻¹ s⁻¹ by Sumner et al. [36] at 1 atm and 298 K. However, based on the new HgO thermochemistry, Hynes et al. estimated this reaction to be highly endothermic and, thus, suggested that this oxidation pathway is not viable in the atmosphere [14,49]. Furthermore, theoretical calculations by Dibble et al. [47] suggest that NO₃ do not form strong bonds with Hg⁰ and therefore is unlikely to initiate gas-phase Hg⁰ oxidation reactions.

A recent study by Peleg et al. [50] provided field evidence of the possible participation of nitrate radicals (NO₃) in the nighttime oxidation of Hg⁰ in the atmosphere. In their study, the role of NO₃ inthe nighttime oxidation of Hg⁰ in the atmosphere was evaluated by measuring the concentrations of NO₃, Hg⁰ and Hg^II continuously during a six-week period in the urban air shed of Jerusalem, Israel during the summer of 2012. The R² average of 0.47 indicated a strong correlation between nighttime [Hg^II] and [NO₃], while correlations of nighttime [Hg^II] with other environmental variables were either weak or absent. Detailed analyses implied that NO₃ radicals may be involved in the Hg⁰ oxidation reaction in the atmosphere. Previous theoretical calculations suggested that NO₃ may be unlikely to initiate Hg⁰ oxidation [47,49]; therefore, NO₃ may be involved in the secondary oxidation reaction of unstable Hg^I species [50]. More laboratory and theoretical studies are required to assess the role of NO₃ in secondary reactions of Hg⁰ oxidation in the real atmosphere, including on atmospherically-relevant surfaces.

2.4. Dominant Gaseous Oxidant for Hg⁰: O₃/OH, Br or Others?

Previous studies have assumed O₃ and OH to be the major Hg⁰ oxidants in the atmosphere [51,52,53,54,55,56]. Despite theoretical doubts concerning the thermal stability of gaseous HgOH and HgO products, many modeling results using O₃/OH as the main atmospheric Hg oxidants showed good agreement with the observed Hg⁰ concentration and wet deposition flux. However, more complex atmospheric Hg oxidation reactions, particularly heterogeneous reactions involving O₃ and OH, have been implied in recent studies ([57] and references therin).

Atomic bromine (Br) has been suggested to be the dominant oxidant of Hg⁰ in the marine boundary layer and in the Arctic [58,59,60]. A previous model study by Holmeset al. [61] implied that Br could be the major oxidant of Hg⁰ in the global atmosphere. The existence of BrO radical has been reported in the upper troposphere [62,63] and over the southeastern US [5,64]. Due to the rapid exchange between Br and BrO radicals [65] and the much slower oxidation rate of Hg⁰ by BrO [18] the field observations of BrO seem to support the dominant oxidation of Hg⁰ by Br in the atmosphere. Shah et al. [5] found that sensitivity simulations using the GEOS-Chem chemical transport model by either increasing Br concentrations, or using a faster rate constant for the oxidation reaction of Hg⁰ byBr, resulted in a better agreement between the modeled results and the aircraft observations. Yet, consistent BrO measurements do not exist in the lower troposphere where humans and biota exist. Moreover, note that the majority of chemical compounds are more concentrated and diversified in the boundary layer, which leads to much more complexity in chemistry, physics and biology due to various interfacial processes. The field studies by Gust in and co-workers during the last decades also suggested that Hg⁰ may be oxidized not just by Br or BrO, but also by various oxidants in the atmosphere. The composition of oxidized mercury showed that Hg⁰ is subject to oxidation processes with various oxidants and not necessarily Br. The assumption of using Br as a universal global oxidant for atmospheric Hg may need further evaluation [66].

The most recent modeling studies indicated more complex Hg chemistry in the atmosphere, and multiple oxidants may be significant under various atmospheric conditions. Ye et al. [67] developed a box model including up-to-date atmospheric Hg chemistry and used it to study the oxidation of Hg⁰ at three different locations in the northeastern United States. The simulated diurnal cycles of Hg^II agreed well with the observations. Model results showed that Hg⁰ oxidation was dominated by O₃ and OH at the coastal and inland sites during the day and Hg⁰ oxidation initiated by H₂O₂ was significant at the inland site during the night. In the marine boundary layer (MBL), the model simulations indicated that Br/BrO were the major oxidants of Hg⁰ at midday while O₃ became the dominant oxidant for the remainder of the day.

Travnikovet al. [57] compared simulation results from four state-of-the-art chemical transport models with field data from various global and regional monitoring networks. It was found that models using the Br oxidation mechanism correctly simulated the observed seasonal variation of the Hg^II/Hg⁰ ratio in the near-surface layer but failed to predict the observed wet deposition maximum in summer at monitoring sites in North America and Europe. Models applying OH chemistry successfully predicted both the observed amplitude of seasonal changes and the periods of maximum and minimum values, but did not catch the maximum Hg^II/Hg⁰ ratios observed in spring. Models using O₃ chemistry could not predict the observed large seasonal variation of either Hg oxidation or wet deposition flux.

Gencarelliet al. [68] simulated the deposition, transport and chemical interactions of atmospheric Hg over Europe for the year of 2013. The outputs of 14 model sensitivity tests were compared with field data from 28 monitoring sites. In general, good agreement was achieved between the model results and the observations. However, the observed deposition in precipitation was significantly underestimated when employing either the O₃ or OH reaction mechanism alone. Using the Br oxidation mechanism overestimated Hg^II at the ground level and produced a lower overall Hg wet deposition than the simulations using both O₃ and OH as atmospheric oxidants for Hg⁰. These model results revealed that the filed data could not be reproduced using the oxidation of Hg⁰ by O₃, OH or Br alone, which indicated a more complicated oxidation mechanism of atmospheric Hg.

Bieseret al. [69] performed a model comparison study evaluating the impact of oxidation schemes and emissions on atmospheric mercury. The models under study successfully simulated the concentration distribution of total Hg and Hg⁰ in the troposphere. It was found that the agreement between the observed Hg^II patterns and the model results employing different chemistry schemes seemed to depend on altitude. Although models using the Br oxidation scheme well simulated high concentrations of Hg^II in the upper troposphere, models applying O₃ and OH chemistry better estimated elevated concentrations in the lower troposphere. However, more studies are needed to confirm this conclusion due to the possible significant influence of model results by the physical and chemical parameters used in these models.

Recent model studies seem to suggest that multiple oxidants are likely involved in the oxidation of atmospheric Hg⁰ dependent on seasons and locations. However, whether Br, O₃/OH or multiple oxidants are the major oxidants of Hg⁰ in the global atmosphere is unclear. To address this important question, more studies on reducing the large uncertainties in rate constants, understanding the heterogeneous oxidation reactions of Hg⁰, as well as improving the treatment of chemical mechanisms in atmospheric Hg models and the accuracy of mercury emission inventories are needed [8,18,48].

3. Chemical Redox Reactions of Hg in the Aqueous Phase

To date, the proposed chemical oxidation pathways in atmospheric droplets include the aqueous-phase oxidation of Hg⁰ by O₃, OH, chlorine (HOCl/OCl⁻) and bromine (Br₂/HOBr/BrO⁻). The proposed chemical reduction pathways of mercury in the aqueous phase relevant to environmental conditions include the reduction of Hg^II by sulfite, photo-reduction of Hg(OH)₂, photo-reduction of Hg^II by HO₂ and photo-reduction of Hg^II-dicarboxylic acid complexes. The obtained rate constants and proposed mechanisms for these reactions were summarized in

Table 2. Current accumulated knowledge on chemical redox pathways of Hg in the aqueous phase.

Reactant(s)	Rate Constants	T(K)	pH	Potential Mechanism	Reference
Identified Aqueous Reduction Pathways of Hg2+
Hg2+ + sulfite (aq)	0.6 s−1	299	3.0–4.84	$HgS{O}_3\to H{g}^{+}\stackrel{+HS{O}_3^{-}or+H{g}^{+}}{\to }H{g}^0+products$	[70]
	0.0106 ± 0.0009 s−1	298	3	$HgS{O}_3+n{H}_{2}O\to [Hg(S{O}_3)\cdot (H_{2}O)n]\to H{g}^0+products$	[71]
	0.013 ± 0.007 s−1	298	7	Same as above	[72]
	<10−4 s−1	299	3.0–4.84	$Hg{(S{O}_3)}_2^{2-}\to H{g}^0+products$	[70]
Hg(OH)2	3 × 10−7 s−1	293	7	$Hg{(OH)}_2\stackrel{hv}{\to }[Hg{(OH)}_2\ast ]\to Hg{(OH)}^{•}+{}^{{.}_{•}}OH\to H{g}^0+products$	[73]
HgS22−	~10−7 s−1	298		Not available	[73]
Hg2+ + HO2	1.7 × 104 M−1 s−1	298		$C_2{O}_4{}^{2-}+2O_2\stackrel{h\upsilon }{\to }2O_2{}^{•-}+2C{O}_2$ $O_2^{•-}+H^{+}\to H{O}_2^{•}$ $H{O}_2^{•}+H{g}^{2+}\to H{g}^{+}+O_2+H^{+}$ $H{O}_2^{•}+H{g}^{+}\to H{g}_{}^0+O_2+H_{}^{+}$ $H{O}_2^{•}+H{O}_2^{•}\to H_2{O}_2+O_2$	[74]
	Not available			Intramolecular 2e− transfer via Hg2+-oxalate complex	[75]
Hg2++Dicarboxylic acids (C2–C4)	(1.2 ± 0.2) × 104 M−1 s−1(Oxalic) (4.9 ± 0.8) × 103 M−1 s−1(Malonic); (2.8 ± 0.5) × 103 M−1 s−1(Succinic)	296	3.0	Mainly intramolecular 2e− transfer via Hg2+-dicarboxylate complexes	[76]
Identified Aqueous Oxidation Pathways of Hg0
Hg0 + O3	(4.7 ± 2.2) × 107 M−1 s−1	298	4.5–9.5	$H{g}_{}^0+O_3\stackrel{}{\to }HgO+O{H}_{}^{-}+O_2$ $HgO+H^{+}\stackrel{}{\to }H{g}^{2+}+O{H}^{-}$	[77]
Hg0 + •OH	2.0 × 109 M−1 s−1	298		$H{g}^0+{}^{.}•OH\to H{g}^{+}+O{H}^{-}$ $H{g}^{+}+{}^{.}•OH\to H{g}^{2+}+O{H}^{-}$	[74]
	(2.4 ± 0.3) × 109 M−1 s–1	298		$H{g}^0+•OH\to •HgOH$ $•HgOH+O_2+H_{2}O\to Hg{(OH)}_2+H^{+}+O_2^{-•}$ $•HgOH+•OH\to Hg{(OH)}_2$ $2•HgOH\to H{g}_2{(OH)}_2\leftrightarrow H{g}^0+Hg{(OH)}_2$	[78]
	5.5 × 109 M–1 s–1			Not available	[79]
Hg0 + aqueous bromine	0.28 ± 0.02 M–1 s–1 0.27 ± 0.04 M–1 s–1 0.2 ± 0.03 M–1 s–1	294–296	2, 6.8, 11.7	$HOBr+H{g}^0\to H{g}_{}^{2+}+B{r}_{}^{-}+O{H}_{}^{-}$ $OB{r}^{-}+H{g}_{}^0\stackrel{H^{+}}{\to }H{g}_{}^{2+}+B{r}_{}^{-}+O{H}_{}^{-}$ $H{g}^0+B{r}_2\to H{g}^{2+}+2B{r}^{-}$	[80]
Hg0 + HOCl/OCl−	(2.09 ± 0.06) × 106 M–1 s–1	Ambient		$HOCl+H{g}^0\to H{g}_{}^{2+}+C{l}_{}^{-}+O{H}_{}^{-}$ $OC{l}^{-}+H{g}_{}^0\stackrel{H^{+}}{\to }H{g}_{}^{2+}+C{l}_{}^{-}+O{H}_{}^{-}$	[81]
	(1.99 ± 0.05) × 106 M–1 s–1

. Recent advances include the aqueous photoreduction of Hg^II-organic complexes and the effects of environmental variables on the aqueous reduction of Hg^II by sulfite.

3.1. Field Evidence for the Reduction of Hg^II

The first field observation of the sunlight-induced reduction of Hg^II in the atmosphere was recently reported by Foyet al. [82]. In their study, concentrations of Hg⁰, Hg^II and Hg_(p) were monitored hourly over four winter months in a remote, high-altitude location. This study site is absent of local anthropogenic sources, which is ideal for studying atmospheric Hg chemical transformations. The parameters of a chemical box model required to reproduce the observations were determined using an optimization algorithm. It was found that the presence of a photolytic reduction reaction previously observed in laboratory studies was needed in order to match the model results with the field observations. The results suggested that the reduction reaction needs to be included in atmospheric Hg models in order to improve simulations of mercury deposition in the atmosphere.

3.2. Photoreduction of Hg^II-Organic Complexes

Hg^II could form strong complexes with organic ligands such as reduced sulfur groups [74,83,84,85]. The photoreduction of Hg^II in the presence of dissolved organic matter (DOM) has been widely documented in various natural water systems [86,87,88,89,90] and may also occur in atmospheric droplets and organic aerosols (OA) [91]. Recent model studies by Horowitz et al. supported the occurrence of the reduction of Hg^II-organic complexes in the atmosphere. Horowitz et al. [91] incorporated the updated chemical mechanism for atmospheric Hg into the GEOS-Chem global model and found that the inclusion of aqueous-phase photoreduction of Hg^II-organic complexes reported by Si and Ariya [76] in GEOS-Chem models, was critical for reproducing the observations. The major Hg^II deposition to the global oceans and the relatively low observed wet deposition of Hg over rural China may be due to different reduction rates of Hg^II with organic aerosols at various geographic locations.

3.3. Direct Reduction of Hg^II by Sulfite

The aqueous-phase reduction of Hg²⁺ with sulfite is believed to be a process relevant to atmospheric droplets. The first-order rate constant determined by van Loon and her co-workers [71] has been widely used in current atmospheric Hg models. The effects of pH (1–7), temperature (274–318 K) and Hg^II sources (Hg(NO₃)₂ or HgO) on the aqueous-phase reduction rate were recently examined by Feinberget al. [72] to better understand this reduction pathway. The results showed that the reduction could occur in the pH range of 1–7. The activation parameters of the aqueous HgO reduction by sulfite at pH 1 and 3 with T = 274–318 K, was in good agreement with the previous study [71]. The reduction rate at pH = 7 decreased with increasing ionic strength, especially with Hg(NO₃)₂. No statistical difference was found between the reduction rate constants of Hg(NO₃)₂ and HgO, which suggested that the reduction of Hg^II by sulfite may be independent of the Hg^II species in the aqueous phase. The results indicated the possible occurrence of this reduction reaction under various environmental conditions and, thus, the necessity of its universal inclusion in atmospheric Hg models.

4. Heterogeneous Redox Reactions of Hg

As shown in Figure 1b, gaseous mercury species can adsorb on atmospheric surfaces and then undergo desorption, dissolution to atmospheric droplets or surface-enhanced (photo)chemical reactions [48]. Despite the potential significant role of heterogeneous Hg reactions on atmospheric Hg chemistry and model simulations [37,92], scare data is available on mercury reactions and equilibrium processes on atmospheric surfaces such as aerosols [48]. A systematic understanding of the surface chemistry of Hg is extremely difficult [48,93] due to the varying concentration, size distribution and composition of aerosols at different locations, times and meteorological conditions [94,95]. Nevertheless, several recent laboratory studies focus on understanding the complex heterogeneous reactions between atmospheric mercury and various aerosols. The major findings in these studies are summarized in

Table 3. Current accumulated knowledge on heterogeneous chemical reactions of Hg in the atmosphere.

Reactants	Surfaces	Major Findings	References
Hg2+ + organic acids	0.1 g/L iron oxides particles or 0.01 g/L ambient aerosols	The presence of iron oxides or ambient aerosols enhanced the rate. Proposed Mechanism: $\rangle F{e}^{3+}-OH\stackrel{+OA}{\to }\rangle F{e}^{3+}-OA\stackrel{h\nu }{\to }{R}^{•}\stackrel{+H{g}^{2+}}{\to \to }H{g}^0$	[96]
HgCl2	Synthetic NaCl aerosols	Significant reduction of HgII observed upon UV, visible or a simulated solar radiation. The presence of iron in synthetic NaCl aerosols inhibited the reduction rate.	[97]
HgCl2	Coal fly ash or synthetic aerosols	The average half-life was estimated to be 1.6 h under clear sky atmospheric conditions from the reduction rates on three diverse fly ash samples; The reduction rate on low sulfate/low carbon fly ash was the fastest of the fly ash samples under study. The reduction rates on synthetic aerosols of carbon black and levoglucosan were about the same as those on coal fly ashes; The presence of adipic acid in synthetic aerosols significantly increased the reduction rate. The soluble constituents of fly ash may be important for the reduction.	[98]
Hg2+ + sulfite	Fly ash	The presence of Cumberland and Shawnee fly ash samples inhibited the reduction rate. The reduction was observed in Cumberland fly ash without sulfite, possibly due to the richness of sulfur in these fly ash samples.	[72]
HgCl2, HgBr2, Hg(NO3)2, HgSO4	Fe(110), NaCl(100) and NaCl(111)Na	The reduction was highly favorable on Fe(110) and NaCl(111)Na surfaces. The desorption of reduced Hg required either no energy input on the NaCl(111)Na surfaces or ~0.5 eV of external energy on the Fe(110) surfaces. The reduction of many HgII species may proceed on metallic Fe and NaCl surfaces.	[99]

.

Recently, Feinberget al. [72] performed the first investigation of the kinetics of heterogeneous Hg^II reduction by sulfite (Na₂SO₃) in the presence of fly ash using UV absorption spectroscopy. Compared with the corresponding homogeneous reduction rates of Hg^II by sulfite, the addition of fly ash samples from Cumberland Power Plant (Tennessee, USA) and Shawnee Fossil Plant (Kentucky, USA) reduced the reduction rates by c.a.45% and 95%, respectively. The observation of the Hg^II reduction in presence of the fly ash samples from Cumberland Power Plant (Tennessee) without added Na₂SO₃, may be due to the richness of sulfite in these fly ash samples. The existence of a large proportion of nanoparticles in the fly ash samples suggested that there were adequate surfaces for heterogeneous chemical reactions to occur. These results indicated the need to incorporate heterogeneous Hg reduction reactions into various environmental models of mercury.

Previous laboratory studies showed that the heterogeneous reduction reaction of Hg^II could occur on iron and sodium chloride aerosol surfaces, which are important components of atmospheric aerosols [97]. Recently, theoretical calculations by Taceyet al. [99] supported the experimental results. In this study, theoretical calculations were performed for the heterogeneous reduction reactions of HgCl₂, HgBr₂, Hg(NO₃)₂ and HgSO₄ on clean Fe(110), NaCl(100) and NaCl(111)^Na surfaces. Here, Fe(110) was the most thermodynamically stable and, thus, the most abundant surface on metallic iron aerosols. The NaCl(100) facet is composed of neutrally charged layers with both Na and Cl atoms exposed on the surface, while NaCl(111)^Na surfaces are charged layers exposing only Na atoms. The results indicated that the heterogeneous reduction reactionthatgeneratesHg⁰ is highly favorable on Fe(110) and NaCl(111)^Na surfaces. The desorption of reduced Hg required either no energy input on the NaCl(111)^Na surfaces or ~0.5 eV of external energy on the Fe(110) surfaces. The results suggested that many oxidized mercury species can be heterogeneously reduced on metallic Fe and NaCl surfaces and the photochemical reaction on the aerosol surfaces may be a necessary step to catalyze the reaction.

Kurienet al. [100] used various iron(oxyhydr)oxide (γ-Fe₂O₃, α-FeOOH, α-Fe₂O₃ and Fe₃O₄) nanoparticles as proxies for reactive components of mineral dust and determined the uptake coefficients for the heterogeneous reaction of Hg⁰_(g) on these nanoparticles. Upon irradiation (λ = 290–700 nm), the Hg⁰_(g) uptake kinetics significantly increased on γ-Fe₂O₃, α-FeOOH, α-Fe₂O₃ nanoparticles, but not on the Fe₃O₄ surface at P = 760 ± 5 Torr and T = 295 ± 2 K. The effect of radiation on the uptake of Hg⁰_(g) by α-Fe₂O₃was retarded by relative humidity. The variation in the uptake behavior of the iron(oxyhydr) oxides nanoparticles was due to their different band gaps. More studies are needed to improve our understanding of such reactions.

Such research presents the need for further studies of the heterogeneous chemistry of mercury, because elemental mercury and many types of oxidized mercury are likely adsorbed and undergo (photo)chemical reactions in the presence of abundant atmospheric surfaces, such as particles and clouds.

5. Future Research Directions

Despite research progress in the understanding of the atmospheric processes of Hg, there are still major knowledge gaps in laboratory, theoretical and modeling studies, as well as field measurement. These gaps include but are not limited to mercury chemical speciation in the field, better kinetic and laboratory studies under various environmental conditions, and more consistent and sophisticated theoretical and modeling integration. We herein focus on the future research needs implied in recent studies since 2015.

Despite the potential importance of the oxidation of Hg⁰ by O₃/OH in the atmosphere, large uncertainties exist in the current gas-phase reaction rate constants and more studies need to be carried out to reduce these uncertainties in the first place. More studies are needed to evaluate the contribution of heterogeneous processes to the obtained rate coefficients for the oxidation of Hg⁰ by O₃ and/or OH, in order to fully understand the discrepancy between the consistent experimental values and theoretical studies. Secondly, recent theoretical studies indicated that the oxidation of Hg⁰ could be initiated by Br- or Cl-atoms and then the secondary oxidation of the HgX (X= Br or Cl) intermediate could be carried out by NO₂, HO₂, ClO, or BrO. More laboratory studies are welcome to confirm this oxidation mechanism. Future field work and model sensitivity studies will also provide valuable insights into the viability of these reactions in the atmosphere. Since several studies supported the possible significant role of Br in the oxidation of Hg⁰ in the atmosphere, more kinetic and mechanistic studies are required to reduce the discrepancy in the reported rate constants for the oxidation of Hg⁰ by halogen atoms.

Recent model studies seemed to suggest that multiple oxidants are likely involved in the oxidation of atmospheric Hg⁰ dependent on seasons and locations. However, whether Br, O₃/OH or multiple oxidants are the major oxidants of Hg⁰ in the global atmosphere is unclear. Accurate measurements of vertical tropospheric concentration profiles of the species involved, such as various atmospheric oxidants, Hg⁰_(g) as well as detailed chemical compositions of oxidized mercury using diverse techniques, are critical for verifying the significance of various oxidants in Hg removal in the atmosphere globally.

Recent field observations and model studies supported the occurrence of reduction reactions in the atmosphere. Furthermore, their results supported the hypothesis that the reduction of Hg^II–organic complexes may play an important role in atmospheric Hg cycling besides sulfite-mediated reduction and photo-reduction of Hg(OH)₂. More kinetic and mechanistic studies of Hg^II reduction pathways under environmentally-relevant conditions are needed. A better understanding of the reduction of Hg^II by organic compounds will require studies on the possible reduction pathways as well as on the quantification of various organic compounds in atmospheric droplets and aerosols.

Another important knowledge gap is the understanding of the redox reactions of Hg on various environmental surfaces such as aerosol, water, ice, snow, soil, and vegetative surfaces [48]. Despite the experimental difficulties caused by the variability in the size and composition of aerosols, several recent studies using fly ash or model aerosols have provided valuable information on understanding heterogeneous reactions of Hg on aerosols. The measured reaction rates are likely important in the chemical transformation of mercury in the atmosphere and the incorporation of these recent laboratory data in future model studies is essential to reduce uncertainties in current atmospheric Hg models. Such endeavors will benefit from the identification and quantification of oxidized mercury compounds, which has been a major challenge in atmospheric Hg research. There are novel instruments including mercury mass spectrometry [101], which can be used to provide such information. Further complementary analytical innovations to accurately quantify mercury-containing compounds in the atmosphere and atmospheric interfaces are needed.

In the light of the Minamata convention, we encourage a more integrated multi-disciplinary approach to comprehend mercury transformation, dynamics, speciation and remediation. Such integration is required to translate sound science to sound policy and regulations.