The Chemistry of Mercury in the Stratosphere.

Mercury, a global contaminant, enters the stratosphere through convective uplift, but its chemical cycling in the stratosphere is unknown. We report the first model of stratospheric mercury chemistry based on a novel photosensitized oxidation mechanism. We find two very distinct Hg chemical regimes in the stratosphere: in the upper stratosphere, above the ozone maximum concentration, Hg0 oxidation is initiated by photosensitized reactions, followed by second-step chlorine chemistry. In the lower stratosphere, ground-state Hg0 is oxidized by thermal reactions at much slower rates. This dichotomy arises due to the coincidence of the mercury absorption at 253.7 nm with the ozone Hartley band maximum at 254 nm. We also find that stratospheric Hg oxidation, controlled by chlorine and hydroxyl radicals, is much faster than previously assumed, but moderated by efficient photo-reduction of mercury compounds. Mercury lifetime shows a steep increase from hours in the upper-middle stratosphere to years in the lower stratosphere.

Introduction

Mercury (Hg) is a neurotoxin with a long lifetime in the Earth system, which exchanges between the atmosphere, continents, and oceans (Driscoll et al., 2013; Lyman et al., 2020; Obrist et al., 2018). A large range of human activities result in emissions of mercury to the atmosphere, predominantly (80%–90%) as elemental mercury (Hg0) (AMAP/UN, 2019; Gustin et al., 2015), characterized by a low chemical reactivity. Atmospheric mercury deposits to ecosystems after oxidation to form water‐soluble HgI,II species (Horowitz et al., 2017) as well as in elemental form (Jiskra et al., 2021). Therefore, understanding the atmospheric mercury cycling between Hg0 and oxidized monovalent HgI and divalent HgII forms is key for accurate assessments of its global transport and input into ecosystems via surface deposition. In current mechanistic and modeling research it is assumed that only highly reactive bromine atoms (Br) or radicals (OH) can initiate the oxidative chain of gas‐phase ambient mercury (Horowitz et al., 2017; Hynes et al., 2009; Shah et al., 2021; Si & Ariya, 2018). Nevertheless, although much progress has been achieved in recent years, major aspects of atmospheric mercury chemistry are still not fully understood (Saiz‐Lopez et al., 2020).

Atmospheric Hg0 lifetime is around several months and therefore enters the stratosphere through tropical convective uplift (Horowitz et al., 2017). Mercury in the stratosphere is estimated to be about 20% of the total atmospheric Hg mass (Shah et al., 2021). Both elemental and oxidized mercury species have been measured directly in the stratosphere (Lyman et al., 2020; Lyman & Jaffe, 2012; Slemr et al., 2018; Talbot et al., 2007) or in air masses influenced by the stratosphere (e.g., Ebinghaus et al., 2007; Lyman & Jaffe, 2012; Radke et al., 2007). Early observations suggested that stratospheric mercury oxidation may be a significant sink for Hg0 since a large fraction of the resulting gaseous oxidized mercury is expected to be taken up on aerosols irreversibly and removed from the stratosphere by sedimentation to the troposphere (Lyman & Jaffe, 2012). Indeed, stratospheric mercury‐containing aerosols have also been observed, indicating that a fraction of mercury is partitioned into the particulate phase in the lower stratosphere (Murphy et al., 1998, 2006, 2014). However, more recent work reporting the most comprehensive set of measurements made over several years in the upper troposphere and lower stratosphere shows the presence of significant concentrations of stratospheric Hg0 (0.25–0.7 ng/m3) (Lyman et al., 2020; Slemr et al., 2018), which in turn implies the existence of efficient Hg reduction reactions, probably HgI,II photoreduction (Saiz‐Lopez et al., 2018, 2019). Despite this observational evidence for an active mercury redox chemical cycle in the stratosphere, most aspects of mercury chemistry in this region of the atmosphere remain unknown. This lack of fundamental knowledge prevents assessing the contribution of stratospheric mercury cycling and transport to the global biogeochemical cycle of the metal.

Here, we present a comprehensive photochemical mechanism for mercury oxidation in the stratosphere initiated by solar UV (253.7 nm) excitation of gas‐phase atomic Hg, from its unreactive 1S0 ground‐state to the highly reactive triplet state, Hg(3P1). The dominant deactivation channel of the excited triplet Hg(3P1) should be collision with molecular oxygen to yield either mercuric oxide (HgO) or, by energy transfer to molecular oxygen, returning to the ground‐state. The unstable HgO initiates a series of secondary reactions with chlorine species that ultimately yield stable water‐soluble mercury halides such as HgCl2. This new photochemical mechanism is further combined with the latest Hg chemical schemes to build the first stratospheric mercury chemistry model that describes the changes in the mercury oxidation chemistry as a function of altitude. We find that the new photosensitized chemistry dominates Hg oxidation above 35 km, with much faster rates than thermal, ground‐state oxidation in the lower stratosphere. In addition, our results indicate that the overall Hg0 oxidation to HgII in the stratosphere is controlled by chlorine and hydroxyl (OH) radical chemistry, being orders of magnitude faster than previously assumed mercury oxidation by bromine (Holmes et al., 2010; Horowitz et al., 2017; Lyman & Jaffe, 2012; Seigneur & Lohman, 2008; Slemr et al., 2018; Talbot et al., 2007).

Materials and Methods

Reaction Rate Coefficient Calculations

The rate coefficients and thermochemical values computed in this work, by a combination of electronic structure theory and Rice‐Ramsperger‐Kassel‐Markus (RRKM) theory, are included in Table 1. For electronic structure calculations we used the hybrid density functional/Hartree−Fock B3LYP method from within the Gaussian 16 suite of programs (Frisch et al., 2016), together with Dunning's quadruple‐ζ aug‐cc‐pVQZ correlation consistent basis, augmented with diffuse functions (Woon & Dunning, 1993). For Hg, the aug‐cc‐pVQZ basis set of Peterson and Puzzarini, (2005) was used. Molecular geometries were first optimized and checked for wave function stability, and their respective vibrational frequencies were calculated. The resulting geometries, rotational constants, vibrational frequencies, and energies of the stationary points on the potential energy surfaces of these reactions are listed in Table S1 in Supporting Information S1. Figure S1 in Supporting Information S1 illustrates the corresponding molecular geometries.

Table 1

New Gas‐Phase Chemistry of Hg in the Stratosphere

Reaction	∆Hr(0 K)/kJ mol−1	k/cm3 molecule−1 s−1	Source
Hg(1S0) + hv (λ = 253.7 nm) → Hg(3P1)			(Kramida et al., 2018)
Hg(3P1) → Hg(1S0) + hv (λ = 253.7 nm)		8.4 × 106 s−1	(Kramida et al., 2018)
Hg(3P1) + N2 → Hg(3P0) + N2	−21.2	5.1 × 10−11 exp (−701/T)	(Callear & Shiundu, 1987)
Hg(3P0) + O2 → Hg(1S0) + O2(3Σu +)	−27.0	1.8 × 10−10 (T/300)0.167	(Callear, 1987)
Hg(3P1) + O2 → HgO(3Π) + O	−6.1	1.7 × 10−10 (T/300)0.53	a MESMER calculation on the PES in Figure S3 in Supporting Information S1
Hg(3P1) + O2 → Hg(1S0) + O2(3Σu +)	−48.2	1.3 × 10−10 (T/300)−0.29	MESMER based on (Callear & Shiundu, 1987)
HgO + O2 → Hg + O3	−72.4	3.4 × 10−13 exp (−1993/T)	b TST
HgO + M → Hg + O + M	27.6	8.4 × 10−11 exp (−3150/T)	MESMER
HgO + HCl → HgCl + OH	−64.6	7.1 × 10−11(T/300)−1.14	MESMER calculation on the PES in Figure S4a in Supporting Information S1
HgCl + O3 → OHgCl + O2	−184.0	1.0 × 10−10 (T/300)0.5	MESMER
OHgCl + hv → HgO + Cl (67%) → Hg + Cl + O (33%)			c Absorption cross sections and AINAMD simulations
OHgCl + HCl → HgCl2 + OH	−79.4	7.9 × 10−11 (T/300)−0.916	MESMER calculation on the PES in Figure S5 in Supporting Information S1
OHgCl + CH4 → HOHgCl + CH3	−34	1.5 × 10−11 exp(−1290/T)	MESMER
HOHgCl + hv → Hg + OH + Cl (91%)			c Absorption cross sections and AINAMD simulations
→ HgOH + Cl (6%)
→ HgCl + OH (3%)
HOHgCl + HCl → HgCl2 + H2O	−100	1.3 × 10−12 (T/300)−1.60	MESMER

Master Equation Solver for Multi‐Energy well Reactions (MESMER) program based on Rice‐Ramsperger‐Kassel‐Markus (RRKM) theory. Potential Energy Surface (PES) is calculated based on Electronic Structure theory (Methods and Supporting Information S1).

Transition State Theory.

See Supporting Information S1 for computation details of absorption cross sections. The photolysis products and yields are determined by ab‐initio non‐adiabatic molecular dynamics (AINAMD) (Supporting Information S1).

For the RRKM calculations, the Master Equation Solver for Multi‐Energy well Reactions (MESMER) program (Glowacki et al., 2012) was used. Where applicable, a reaction is assumed to proceed via the formation of an excited adduct, which can either dissociate back to the reactants, be stabilized by collision with the third body into any minima on the potential energy surface, or dissociate to products. The internal energy of the adduct is divided into a contiguous set of grains (width 50 cm−1), each containing a bundle of rovibrational states. Each grain is then assigned a set of microcanonical rate coefficients for dissociation back to the reactants or to the products, using inverse Laplace transformation to link them directly to the high‐pressure limiting recombination coefficients. In the case of Hg(3P1) + O2, this was set to k ∞ = 3.0 × 10−10 (T/298)0.167 cm3 molecule−1 s−1 (Callear, 1987); for HgO + HCl, k ∞ = 2.7 × 10−10 (T/298)−0.17 cm3 molecule−1 s−1, the estimated dipole‐dipole capture rate coefficient; and for OHgCl + HCl, k ∞ = 2.0 × 10−10 (T/298)0.17 cm3 molecule−1 s−1, a typical capture rate controlled by dispersion forces. The probability of collisional transfer between grains was estimated using the exponential down model: the average energy for downward transitions 〈ΔE〉down was set to 300 cm−1, typical of M = N2 at 300 K, with a T 0.25 temperature dependence (Gilbert & Smith, 1990). The resulting temperature‐dependent rate coefficients are listed in Table 1, together with the reaction energetics.

Note that at the B3LYP level of theory, D0(HgO) is 37 kJ mol−1, which is consistent with earlier work (Filatov & Cremer, 2004). It is once electron correlation and spin‐orbit coupling is included that the smaller bond energy has been determined (Filatov & Cremer, 2004; Peterson et al., 2007; Shepler & Peterson, 2003) (see Supporting Information S1).

Computation of Cross Sections and Photolysis Products of HOHgCl and OHgCl

A computational strategy documented and calibrated in previous works to determine ultraviolet and visible (UV‐Vis) absorption spectra and cross sections of mercury compounds was reutilized here and adapted to HOHgCl, OHgCl, and Hg(OH)2 (Francés‐Monerris et al., 2020; Saiz‐Lopez et al., 2018, 2019; Sitkiewicz et al., 2019). In addition, their photolysis yields upon light absorption in the stratosphere were determined by means of state‐of‐the‐art ab‐initio non‐adiabatic molecular dynamic (AINAMD) simulations, previously applied to other atmospheric compounds (Carmona‐García et al., 2021; Francés‐Monerris et al., 2020). More details are provided in the Supporting Information S1.

Photochemical and Kinetics Model

We have developed a state‐of‐the‐art photochemical and kinetics box model to study the effect of the different chemical pathways on Hg oxidation in the stratosphere. The model is based on the latest understanding of atmospheric Hg chemistry (Francés‐Monerris et al., 2020; Saiz‐Lopez et al., 2018; Shah et al., 2021) and also includes the new chemical and photochemical scheme proposed in this study (Table 1). The model is constrained with values of pressure, temperature, O3, Br, Cl, OH, CH4, CO, and HCl concentration profiles (Figure S2 in Supporting Information S1) from the Whole Atmosphere Community Climate Model (WACCM) (Neale et al., 2013). The stratospheric Hg0 concentration value is fixed at 0.5 ng m−3 (Slemr et al., 2018), assuming a well‐mixed distribution through the column. The simultaneous differential equations describing the rate of change of each species (with rate coefficients in Table S2 in Supporting Information S1) are solved using the KPP 2.2.3 based Rosenbrock integrator (Damian et al., 2002). Photodissociation rates are calculated online using the latest cross sections with the TUV photolysis code (Madronich & Flocke, 1999). The model was run at every 5 km from the tropopause up to 60 km until steady state was reached. The model was run at 50°N 0°E for the month of June.

Results and Discussion

A Mechanism for Photosensitized Mercury Oxidation

In the chemical mechanism proposed here, atomic mercury is electronically excited from the relatively inert ground state Hg (1S0) to the highly reactive triplet state Hg (3P1) by solar radiation at 253.7 nm. In the Earth's atmosphere, these radiation conditions prevail above the stratospheric ozone layer maximum concentration (>35 km, 0.1–10 hPa, 220–260K) where the O3 Hartley band is optically thin (see Supplementary Note 1 in Supporting Information S1) and the following photochemistry can take place (Figure 1):

Figure 1

Photosensitized gas‐phase oxidation of mercury in the stratosphere above the ozone layer for altitudes higher than ∼35 km. The pseudo first‐order rate coefficients for the primary reactions of the electronically excited Hg(3P1) triplet state for these ambient conditions are shown in s−1. The major, second‐step chlorine‐dependent oxidation reactions are also included.

Primary reactions of electronically excited mercury atoms. The oscillator strength (f = 0.0243) for the transition (R1) is relatively large (Kramida et al., 2018). Natural mercury contains seven stable isotopes, with masses 196–204 u and small differences in chemical reactivity (Bergquist & Blum, 2007; Sonke, 2011). Here, the rate of mercury excitation was computed based on a single absorption line with a Doppler linewidth of 0.03 cm−1 (0.93 GHz) at a temperature of 240 K, characteristic of the lower stratosphere, and a peak cross section of σ = 7 × 10−13 cm2, derived from the f‐value. Spontaneous emission from the Hg (3P1) triplet to the ground state (R2) is spin‐forbidden and relatively slow (radiative lifetime of 119 ns, Einstein coefficient A nm = k 2 = 8.4 × 106 s−1). Thus, for the specified stratospheric conditions the rate of radiative deactivation would be ca. 50% slower than the hard collision frequency (see below). The two additional Hg (3P0) and Hg (3P2) triplet components are dark metastable states with much longer lifetimes (Garstang, 1962).

The reaction of excited atomic mercury Hg (3P1) with O2 (R3) has been studied for nearly a century, beginning with Gaviola's pioneering experiments on Hg‐sensitized photochemistry (Gaviola, 1929), and reviewed several times (Burton & Noyes, 1969; Callear, 1987; Callear et al., 1959; Noyes & Leighton, 1941; Volman, 1963). The laboratory experiments show that the final reaction products are O3 in large quantities (up to 60 molecules for each consumed mercury atom) (Volman, 1963) and HgO deposited as a solid film. However, the experiments did not establish unambiguously the path to HgO, and in particular the rate coefficient of the HgO primary reaction (R3a) could not be determined (See Supplementary Information, Part 1 in Supporting Information S1). In the present work the reaction rate coefficients were obtained from a combination of electronic structure and Rice‐Ramsperger‐Kassel‐Marcus (RRKM) calculations, as detailed in Materials and Methods and Table 1. The R3a rate coefficient yielded a value of k 3a = 1.7 × 10−10 (T/300)0.53 cm3 molecule−1 s−1. Molecular geometries, rotational constants, vibrational frequencies, and energies computed by the quantum‐chemical methods are listed in Table S1 in Supporting Information S1. The Hg‐O bond‐energy obtained in this way (27.3 kJ mol−1) is similar to previous predicted values, and in the spin‐conserving reaction the mercury oxide would be produced as HgO(3Π) with little vibrational energy, ∼6 kJ mol−1 (Supplementry Information and Figure S3 in Supporting Information S1), which is important for such a weakly bound molecule to survive in the stratospheric low‐temperature environment.

The R3b step is just the transfer of electronic energy from atomic mercury to molecular oxygen (Figure S3 in Supporting Information S1) with a rate comparable to the collision frequency (k coll ∼ 10−10 cm3 molecule−1 s−1, 240K) (Callear, 1987; Callear et al., 1959; Horiguchi & Tsuchiya, 1974; Volman, 1963). Triplet‐triplet energy transfer experiments (Hippler et al., 1978) as well as O3 synthesis cannot be explained unless the resultant excited oxygen molecule (O2 *) is produced in a high‐lying electronic state ( or (Michels, 1981)). In the present kinetic model, a value of k 3b = 1.3 × 10 −10 (T/300)−0,29 cm3 molecule−1 s−1 is calculated based on previous experimental values (Callear & Shiundu, 1987).

Reaction 4, in which Hg(3P1) is physically quenched by N2, is well characterized with a rate constant k 4 = 5.1 × 10−11 exp (−701/T) cm3 molecule−1 s−1 (Callear, 1987; Callear & Shiundu, 1987), and yields only the lowest metastable spin‐orbit multiplet Hg(3P0), which is 21.3 kJ mol−1 below Hg(3P1). As discussed above, Hg(3P1) can be quenched by O2, with two possible channels yielding HgO or Hg(1S0), respectively. Therefore, Hg(3P1) is ∼22 times more likely to be quenched by O2 than N2. Any Hg(3P0) produced by R4 can be excited back to Hg(3P1) by collisions with N2, although this is very slow at atmospheric temperatures (k(T) = 1.49 × 10−10 exp (−3243/T) cm3 molecule−1 s−1 (Callear & Shiundu, 1987). Much more likely is quenching of Hg(3P0 → 1S0) by O2 (R5), with a rate coefficient k 5 = 1.8 × 10−10 (T/300) 0.167 cm3 molecule−1 s−1 (Callear, 1987) or chemical reaction to yield HgO, by analogy with R3a. The physical and chemical initial processes of elemental gas‐phase mercury in the excited Hg(3P1) triplet discussed above are shown in Figure 1, together with the estimated rates for the specified stratospheric conditions.

Second step reactions. HgO may be reduced back to elemental mercury by reaction with O2 and by thermal‐ and photo‐dissociation:

Reaction 6 is exothermic by 73 kJ mol−1 based on a bond energy for HgO(3Π) of 27 kJ mol−1, but with a barrier of 15.4 kJ mol−1, according to our calculations (see Figure S1 in Supporting Information S1 for a diagram of the transition state geometry and Figure S4b in Supporting Information S1 for the potential energy surface). A value of k 6 = 3.4 × 10−13 exp (−1993/T) cm3 molecule−1 s−1 was computed as indicated above (Table 1). The thermal decomposition of the oxide (R7) in the gas phase is calculated to yield k 7 = 8.4 × 10−11 exp (−3150/T) cm3 molecule−1 s−1. The photo dissociation rate of the oxide back to atomic mercury has been computed based on the absorption spectrum and cross sections calculated before by high‐level quantum mechanical methods (Saiz‐Lopez et al., 2018).

We have also identified the following pathway for HgO(3Π) to form HgCl2, the most stable form of oxidized Hg in the stratosphere:

When HgO(3Π) reacts with HCl, a weakly bound HgO‐HCl complex forms initially on the spin‐conserving triplet surface (Figure S4a in Supporting Information S1), which rearranges over a slight barrier (1 kJ mol−1) to the more stable HgCl‐OH complex. This can then dissociate either to HgCl + OH (99%) or HgOH + Cl (1%), in accord with the relative energies (Figure S4a in Supporting Information S1). The rate coefficient computed here is close to the collisional limit, k 9 = 7.1 × 10 −11 (T/300)−1.14 cm3 molecule−1 s−1. For the reaction of HgCl with O3 there is no obvious barrier, and the formation of OHgCl is quite exothermic (ΔH 0 = −184 kJ mol−1). We therefore assign this reaction a rate coefficient set to a collision frequency with a T 0.5 dependence to yield k 10 = 1 × 10 −10 (T/300)0.5 cm3 molecule−1 s−1 (Table 1). Reaction 11 is also exothermic, with ΔH 0 = −79.4 kJ mol−1, and the potential energy surface (Figure S5 in Supporting Information S1) is similar to that of HgO + HCl (Figure S4a in Supporting Information S1). In this case, 99% of the reaction product is HgCl2 rather than HOHgCl, again following the more exothermic pathway. A rate coefficient of k 11 = 7.9 × 10 −11 (T/300)−0.916 cm3 molecule−1 s−1 is calculated. Photolytic dissociation of the OHgCl radical (R12) yields mainly (67%) HgO + Cl (Table S3 in Supporting Information S1), which will then recycle to HgCl2 by reactions R9–R11, as predicted by the absorption cross section (Figure S6 in Supporting Information S1) and photodissociation dynamics computed here.

An additional pathway for the formation of HgCl2 results from the abstraction reaction shown below:

The calculated rate coefficient for R13 of k 13 = 1.5 × 10−11 exp (−1290/T) cm3 molecule−1 s−1 indicates that this abstraction reaction should be quite slow at stratospheric temperatures. Thus, at 220 K the rate of the competing R11 is 2,500 times faster. In any case, even if it does happen, the resulting HOHgCl product would react with HCl to yield HgCl2 (R14) as this reaction is exothermic (−100 kJ mol−1), with a calculated rate coefficient of k 14 = 1.3 ×10 −12 (T/300)−1.6 cm3 molecule−1 s−1. The corresponding absorption cross section (Figure S7 in Supporting Information S1) and photodissociation products (Table S4 in Supporting Information S1) of HOHgCl photolysis (R15) were also calculated to evaluate its contribution to the present mechanism. Finally, the possible direct reaction of mercury oxide with H2O vapor was found to be unimportant for these specific reaction conditions (see Supplementary Note 2 in Supporting Information S1). Similarly, we studied the effect of the reaction between Hg(3P1) and H2O (Pertel & Gunning, 1959) to make HgOH followed by reaction with HCl to HgCl (Guzman & Bozzelli, 2019) and found their contribution to be negligible to total oxidized mercury (Supplementary Note 2 in Supporting Information S1).

Quantitative Model of the Chemical Cycle of Mercury in the Stratosphere

There is currently no established model of the gas‐phase chemical reactions of Hg0, HgI, and HgII species in the stratosphere, therefore, we build for the first time a quantitative photochemistry and kinetics model of stratospheric mercury (Section 2, and Supporting Information S1). The model includes the state‐of‐the‐art chemical schemes developed to explain mercury chemistry in the troposphere based on bromine‐, chlorine, and OH‐initiated Hg oxidation (Francés‐Monerris et al., 2020; Saiz‐Lopez et al., 2020; Shah et al., 2021) along with the new chemistry presented here (Table S2 in Supporting Information S1). There is a sharp transition in the metal oxidation mechanism from altitudes above the ozone layer, where the Hg‐photosensitized reactions are dominant, to those below the ozone layer, where thermal reactions take the control. These large differences are accounted for in the model, which is consistent also with established tropospheric Hg chemistry and simulations, and agrees with previous works (e.g., Shah et al., 2021). Hence, the new model represents a unified model of mercury chemistry that bridges chemistry both in the troposphere and stratosphere.

Figure 2 shows the simulation results at 50°N in June as a function of altitude, where the tropopause ranges between 12 and 14 km and the stratopause is approximately at 50 km. Below 35 km, where photosensitized oxidation does not occur, reaction with the radical OH drives the initial loss rate of gas‐phase elemental mercury (Figure 2a). In that same region (15–35 km), the overall conversion of elemental to oxidized mercury is dominated by both OH and chlorine oxidation (Figure 2c). This is because, in addition to the faster initial OH oxidation, the contribution of the OH channel is enhanced (Figure 2b) by the slow photodissociation rate of the final oxidized product Hg(OH)2, as computed here from its UV absorption cross section (Figure S8 in Supporting Information S1). Similarly, HgCl2 has a long lifetime against photolysis (Figure 2b). This results in HgCl2 and Hg(OH)2 being the dominant reservoirs for oxidized Hg in the lower stratosphere (Figure 2b; Figure S9 in Supporting Information S1). In sharp contrast, Figure 2 also shows that in the upper stratosphere, above approximately 35 km, Hg0 loss is entirely driven by the photosensitized oxidation of Hg0 to HgO, followed by reactions with HCl to form HgCl2, which is the main oxidized mercury species in the mid to upper stratosphere (Figure S9 in Supporting Information S1). Note that the initial rate of Hg oxidation by this new photosensitized chemistry in the mid‐to upper stratosphere is 103–104 times faster than the combined rates of Hg oxidation by bromine, chlorine, and OH radical (Figure 2a). The halide HgCl2, which is the main final oxidation product from the reaction of electronically excited Hg, is photolyzed rapidly back to Hg0 in the mid‐to upper‐stratosphere. Hence, Hg0 and HgCl2 exist in a pseudo steady‐state, which is tilted toward HgCl2 due to the oxidation rate being about 2 times faster than the photolysis of HgCl2. The modeled partitioning of oxidized mercury species as a function of altitude is shown in Figure S9 in Supporting Information S1. Observations of mercury species have been made in the stratosphere onboard the IAGOS‐CARIBIC aircraft and show the prevalence of about 0.5 ng m−3 Hg0 up to 4 km above the thermal tropopause (Slemr et al., 2018). If the model is initialized with this value, it predicts ∼0.1 ng m−3 of HgI,II, which matches well with the observed difference between total mercury and Hg0 (0.17 ng m−3) in this region (Slemr et al., 2018).

Figure 2

Model predictions for the different gas‐phase oxidation mechanisms of elemental Hg0 in the stratosphere developed here as a function of altitude, for 50°N, 0°E, June. (a) Initial Hg0 loss rate due to reaction with Cl, Br, and OH, represented by the corresponding pseudo first‐order rate coefficient (s−1), as a function of the concentration of these species. The initial loss rate of the excited Hg(3P1) triplet is also shown. (b) Photolysis rates (s−1) of the main HgII compounds that result from mercury oxidation. (c) Vertical concentration profile of the HgI,II species containing X = Cl, Br, OH. (d) Temperature and ozone number density.

Conclusions

The different Hg oxidation mechanisms reveal two very distinct mercury chemistry regimes in the stratosphere, one above the ozone layer where Hg(3P1) photosensitization drives fast oxidation, and the other below the ozone layer controlled by ground‐state initiated slow oxidation. As noted above, this is a consequence of the fortuitous coincidence of the absorption maxima at 254 nm of both, the mercury excitation transition Hg(1S0) → Hg(3P1) and the ozone Hartley absorption band. As shown in Figure 3, the large difference in the rates of Hg oxidation in these two stratospheric chemical regimes leads to a significant impact on the lifetime of Hg0 against chemical loss throughout the stratosphere. Thus, the Hg0 chemical lifetime ranges from years in the lower stratosphere to days and even hours in the mid‐to upper‐stratosphere (Figure 3). The long lifetime of Hg0 calculated here for the lower stratosphere, and also inferred from observations (Slemr et al., 2018), would enable the steady flow of mercury to the upper stratosphere where photosensitized oxidation is extremely efficient. The calculated Hg0 lifetime of hours in the upper stratosphere is only comparable to the atmospheric mercury depletion events measured in the Arctic springtime boundary layer (Schroeder et al., 1998). However, unlike the localized and episodic nature of these Arctic events, the mechanism proposed here indicates that the lifetime of elemental mercury against chemical loss would be of only hours throughout the global mid to upper stratosphere.

Figure 3

The lifetime of elemental mercury against chemical loss as a function of altitude in the stratosphere for 50°N, 0°E, June. In the middle to upper stratosphere Hg oxidation is dominated by fast reactions of the UV‐excited Hg(3P1) triplet state, which transform elemental mercury directly into HgII compounds (see inset), resulting in short chemical lifetimes of elemental mercury in the day‐hour range. However, in the lower stratosphere, Hg triplet excitation at 254 nm by solar radiation is blocked by the ozone Hartley band with maximum absorption at the same wavelength. Therefore, ground‐state unreactive atomic Hg0 is slowly oxidized by thermal reactions with Cl, Br, and OH species to yield HgI intermediate compounds (see inset). In addition, secondary photoreduction processes revert many of the oxidized products back to elemental mercury, resulting finally in Hg0 chemical lifetimes in the years range in the lower stratosphere.

Our study shows that the chemical cycling of mercury in the stratosphere is much more active and complex than previously anticipated. The new chemical scheme presented here results in stratosphere mercury oxidation rates orders of magnitude faster than currently assumed, which change sharply as a function of altitude. Atmospheric Hg oxidation has so far been considered to be initiated only from the Hg ground electronic state, which also requires to proceed via intermediate HgI species (Shah et al., 2021). We show here that stratospheric Hg oxidation can also proceed by an alternative Hg photosensitized chemistry, allowing efficient oxidation of Hg0 directly to HgII. The novel mercury chemical and photochemical framework reported herein has broad implications for the stratospheric mercury residence time and its global dispersion. This new fundamental chemistry provides for the first time a quantitative mechanism of mercury chemistry in the stratosphere and reveals this region of the atmosphere as a unique environment for Hg chemical cycling.

Supporting Information S1

Click here for additional data file.