Elemental mercury (Hg0) is a common trace contaminant associated with corrosion of infrastructure impacting exploration, production, and processing of commercial hydrocarbons. Presently lacking is a model for the quantitative prediction of Hg concentration in reservoir fluids, sufficiently reliable for process engineering applications and design of mitigation strategies to ameliorate the potential risk of Hg presence.

In this paper, we present a thermodynamic equilibrium mineral-based model for predicting the solubility of mercury in hydrocarbons, Hg0(org), at in-situ reservoir conditions. The model is based on literature experimental data on the solubility of Hg0 in a mixture of alkanes, in equilibrium with Hg0, H2S, O2, cinnabar (HgS), and water. As the model inputs are based on the chlorite-pyrite-H2S model, its application should primarily be limited to clastic hydrocarbon-bearing reservoirs. A global data set of Hg in hydrocarbons reveals a remarkably strong association with the presence of humic coal in subsurface formations.

Assuming that pure stoichiometric cinnabar (HgS) is stable at the reducing conditions typical of hydrocarbon reservoirs (i.e., aHgS = 1) results in an overestimation of Hg0(org) solubility by up to three orders of magnitude relative to globally reported concentrations of mercury in natural hydrocarbons. A statistically robust match between model and observed concentrations of Hg0(org) was achieved using an aHgS of 0.003, consistent with reported concentrations of Hg0 from pyrite (FeS2) in coals and hydrocarbon reservoirs. The model has been validated in a case study of reservoir Hg reported in the Gorgon North-1 well, North West Shelf (NWS), Australia.

The dominant process of cinnabar precipitation is by oxidation, particularly in the near-surface environment where reduced Hg0-bearing hydrocarbons mix with shallow oxygenated or acidic surface waters. Such processes are typical of the environments where most downhole fluid samples are collected during drilling, sampling, and cleanup of exploration and development wells. This leads to the invariable conclusion that much of the particulate mercury species, specifically HgS, collected with hydrocarbon fluid samples, are metastable with respect to the dissolved Hg0(org) in hydrocarbons at reservoir conditions and should not be included in the estimation of total Hg (i.e., THg) in hydrocarbons.

This hypothesis has been confirmed by an extended well test in the Minami-Nagaoka gas condensate field, where it was observed that Hg dissolved in produced water decreased to negligible levels over time, while the Hg0(org) in the condensate liquid reached a stable value like what the new Hg0(org) solubility model would predict for in-situ reservoir conditions.

Together with carbon dioxide (CO2) and hydrogen sulfide (H2S), elemental mercury (Hg0) is the most common trace contaminant associated with corrosion of infrastructure impacting exploration, production, and processing of commercial hydrocarbons. Issues related to H2S corrosion have previously been documented by Bryndzia and Inan Villegas (2020). High fugacity of H2S (i.e., partial pressures) causes metal embrittlement in steel infrastructure and may also induce sulfide stress cracking as described by Grimes et al. 2014. Both Hg and H2S also pose significant health, safety, security, and environment risks and require special safety protocols where present. There are many implications of Hg presence for health, safety, security, and environment, which include exposure and toxicity to human health, environmental risk, infrastructure and pipeline integrity, and regulatory compliance.

The most critical problems associated with Hg are its ease of amalgamation with different metals, most notably aluminum (Al), with which it readily forms an amalgam (i.e., a solid solution of Al dissolved in Hg metal). Al-amalgam is a strong reductant and Hg can corrode aluminum via the mechanism of either amalgam or liquid metal embrittlement. Amalgamation of Al is common in gas-processing plants, including liquefied natural gas (LNG) facilities, in which the chilling components (i.e., “cold boxes” and/or heat exchangers used to separate methane from other hydrocarbon gases, nitrogen, and water) are almost exclusively made of solid Al metal.

Liquid metal embrittlement due to Hg corrosion has resulted in significant loss of life due to catastrophic failure of gas-processing infrastructure as, for example, the mercury-induced catastrophic failure of a heat exchanger at Skikda, Algeria, in 2004 and the Moomba gas plant fire in Australia (Santos 2010). The Skikda accident incapacitated three LNG lines, impacted approximately 2% of the world’s liquefaction capacity at the time, and resulted in 27 fatalities. It is estimated to have cost 900 million USD (Ouddai et al. 2012) to rebuild the facility and an additional 300 million USD in lost sales revenue. At Skikda, it was noted that the destroyed unit (Unit “40”) was the only LNG train in the Skikda gas-processing plant that did not have a mercury removal unit treating the gas before cryogenic cooling.

Although the concentration of Hg in natural gas may be considered extremely low, “its effect is cumulative as it amalgamates” (Audeh 1988). Due to the volumes of natural gas processed and the tonnages of liquid hydrocarbons handled, the quantities of mercury available to distribute throughout gas-processing facilities can be substantial. For example, a typical 10 000 t/d LNG plant would use approximately 17 million m3/d of natural gas. If this gas contained 100 µg/m3 of mercury, the plant would receive 620.5 kg/a of mercury (Carnell et al. 1995). According to Carnell et al. (2007, 2008), to avoid mercury-induced corrosion of aluminum, removal of mercury to an upper limit of 0.01 µg/m3 (volume under normal conditions) upstream of liquefaction is recommended to maintain equipment integrity within LNG plants.

Hg presence also presents an economic challenge since there is a penalty associated with processing gas condensate based on Hg concentration in the condensate crude. If not planned for and properly mitigated against, the economic penalties will obviously impair the economics and profitability of long-term producing assets containing Hg.

Once a steady-state concentration of Hg in a hydrocarbon production stream is established, it is a relatively straightforward process to design a mercury removal unit to remove Hg from the production stream. The problem lies in the fact that it often takes a considerable period of time and volume of production to realize that Hg is even present. This has transpired at a number of fields, including the Shearwater platform (EXPRO Report 2018, internal Shell report), the Kipper gas field in Gippsland in southeast Australia (Boschee 2013; Black and Saunders 2018), in the Minami-Nagaoka gas field in Niigata prefecture, Japan (Yamada et al. 2017), and most recently when Total announced the deferment of developing the ~1 Tcf Glendronach field in the west of Shetlands due to unexpectedly high levels of Hg in the gas condensate, rendering field development uneconomic (Energy Voice 2018). In Well K from the Minami-Nagaoka gas condensate field in Japan, well cleanup lasted for more than 100 days before a reliable near-steady-state Hg concentration in the gas condensate stream was achieved. This phenomenon is also commonly observed with H2S, which may take an extended period to manifest itself in the gas production stream once the exposed steel tubulars have reached saturation with respect to H2S.

An obvious solution to this problem would be to have the ability to quantitatively model and predict the expected concentration of Hg in the subsurface, at ambient reservoir pressure and temperature conditions. A statistically robust model for predicting Hg concentrations in hydrocarbons would enable design criteria to be established for Hg concentration and hydrocarbon production rates suitable for designing a mercury removal unit tailored to a specific field, including its hydrocarbon composition, Hg0 and H2S content.

An interesting conference paper published by Chevron scientists reported the observation that reservoir mercury results indicated a potential correlation of the bottomhole temperature with mercury concentration in produced fluids and that the equilibrium of mercury and mercury sulfide in the reservoir could be the basis for this correlation (Cooper et al. 2017). In this paper, we introduce a new mineral-based thermodynamic equilibrium model that for the first time enables the prediction of Hg0 solubility in both gaseous and liquid hydrocarbons based on equilibrium between coexisting hydrocarbons, native Hg0, HgS (cinnabar), H2S, O2, and water.

Most economic deposits of the mineral cinnabar (HgS) have a hydrothermal origin and are found proximally associated with organic-rich sediments, particularly black shales and recent volcanism. Native liquid mercury, Quicksilver (Hg0), is rarely, if ever, found without cinnabar, HgS. Cinnabar is the dominant and primary Hg-containing mineral from which Hg0 is produced. A consistent unique aspect of most cinnabar deposits is their intimate association with petroleum, indicating a natural affinity for mercury transport in organic-rich hydrothermal fluids (Krupp 1988; Peabody and Einaudi 1992; Fein and Williams-Jones 1997).

Numerous cinnabar deposits occur throughout the California Coast Ranges and an excellent literature summary for many deposits in this region may be found in Peabody and Einaudi (1992). The focus of their study was the Culver-Baer cinnabar deposit in the Mayacmas District of northern California. Based on petrographic and analytical data, Peabody and Einaudi (1992) concluded that other than CO2, the only components added to the system during hydrothermal alteration and mineralization were H2S, Hg, and petroleum. Isotopic and biological marker data support the hypothesis that all these components may have been derived from local sedimentary rocks in response to heating/cracking and subsequently transported to the site of deposition in a gas phase. Because there is no evidence for the introduction of any other components, they concluded that the Culver-Baer deposit may be a fossil condensate zone, where Hg0 dissolved in hydrothermally derived petroleum liquid condensed as both Hg0 and HgS during oxidation and cooling.

Source Rocks

Organic carbon-rich sediments, particularly black shales having Hg contents an order of magnitude higher than most other rock types (i.e., 1–3 ppm Hg) are frequently found in Hg ore districts [e.g., Almaden (Spain), Monte Amiata (Italy), Idrija (Balkans), Ngawha (New Zealand), and in the USA], including the Mayacmas District of northern California and Sulfur Bank in Californian and Terlingua, Texas (Yates and Thompson 1959; Stetson et al. 2009). As discussed by Krupp (1988, page 352), other Hg deposits are in coal basins which provide similar conditions [e.g., Pfalz district (Saar-Nahe Basin, southewest Germany) and Nikitovka (Donets Basin, Ukraine)]. Various fixation mechanisms for Hg in its source rocks have been proposed, including adsorption onto clay minerals and organic matter, and incorporation of Hg into the crystal structures of other host minerals, particularly sulfides (Moiseyev 1971; Beuge 1982). In soils and sedimentary rocks, Hg concentrations correlate perfectly with organic carbon contents (Andersson 1970; Beuge 1982), suggesting that the predominant fixation mechanism is by organic matter. More recent papers have focused on Hg enrichments in marine organic-rich sediments associated with oceanic anoxic events and the link with the eruption of large igneous provinces that are believed to be a source for the elevated Hg through aerial and/or aqueous dispersion (e.g., Thibodeau and Bergquist 2017 and references therein). Recent studies deploying modern analytical methods have provided new insights on the state of mercury and its distribution and phase (e.g., Hg0 in microfissures in pyrite vs. Hg2+ in lattice positions of Fe2+ in the pyrite mineral; Manceau et al. 2018).

Mercury in Hydrocarbons

As shown in Table 1, reported Hg concentrations in hydrocarbons are highly variable and can range from 450 to 5000 μg/m3 in some gas fields in northern Germany or Thailand, to 180 μg/m3 in Groningen, The Netherlands (NL), to ~10 μg/m3 in Trinidad (Carnell et al. 1995; Sainal et al. 2007; Boschee 2013; Whittenbury 2016). In general, crude or condensate liquids tend to have higher concentrations than gas. The important information in the global data captured in Table 1 is that the range of observed concentrations of Hg in hydrocarbons ranges from near zero to a maximum of ~1,200 ppb in condensate liquids (i.e., µg/kg; ~6×10−6 moles/kg = logmHg0(org) = −5.22). Even higher Hg contents are predicted for gas from northern Germany and Thailand.

Table 1

Published ranges of mercury in naturally occurring hydrocarbons.

Location
(Sainal et al. 2007)
Gas
(μg/Nm3)
Liquids
(μg/kg)
Region
(Whittenbury 2016)
Hg0 (μg/Nm3)
Europe 100–150 – Indonesia 250–300 
South America 50–120 50–100 Australia 38 
Gulf of Thailand 100–400 400–1200 NL 180–200 
Africa 80–100 500–1000 North and East Coast Trinidad 12 
Gulf of Mexico (USA) 0.02–0.4 – Northern Germany 5000 
Overthrust Belt (USA) 5–15 1–5 Oman 60 
North Africa 50–80 20–50 Thailand 2000 
Malaysia 1–200 10–100   
Indonesia 200–300 10–500   
Location
(Sainal et al. 2007)
Gas
(μg/Nm3)
Liquids
(μg/kg)
Region
(Whittenbury 2016)
Hg0 (μg/Nm3)
Europe 100–150 – Indonesia 250–300 
South America 50–120 50–100 Australia 38 
Gulf of Thailand 100–400 400–1200 NL 180–200 
Africa 80–100 500–1000 North and East Coast Trinidad 12 
Gulf of Mexico (USA) 0.02–0.4 – Northern Germany 5000 
Overthrust Belt (USA) 5–15 1–5 Oman 60 
North Africa 50–80 20–50 Thailand 2000 
Malaysia 1–200 10–100   
Indonesia 200–300 10–500   

Despite recent advances in obtaining high-quality fluid samples for the analysis of both H2S and Hg, obtaining representative samples for Hg in hydrocarbon fluids presents a significant technical challenge. Much of the observed large range of reported Hg concentrations is directly attributable to difficulties in sampling and quantifying low concentrations of Hg early in the exploration and appraisal phase of a well. Some of this is also attributable to the difficulty in quantifying the total Hg concentration (that is, THg) in a fluid sample due to the presence of many potential forms or species of Hg routinely encountered during fluid sampling in a gas-processing plant (e.g., Nengkoda and Al-Hinai 2009). Recent advances in sampling technology have attempted to address many of these issues by developing downhole sample chambers that attempt to avoid Hg loss during sampling of reservoir fluids at in-situ conditions (Dybdahl et al. 2020).

Hg0 Solubility Experiments in Liquid Hydrocarbons

Several studies have been published on the solubility of Hg0 in liquid hydrocarbons. Hg0 solubility was measured in C5-C10 up to a maximum temperature of 65°C by Clever and Iwamoto (1987). Hg0 concentration was measured using gold foil amalgamated with a saturated Hg solution and analyzed spectrophotometrically with a calibrated optical density of Hg in saturated solution of organic solvent, using a decay scheme of radioactive Hg203 spiked solvents. Miedaner et al. (2005) measured Hg0 solubility in octane and dodecane from 110 to 225°C. Hg0 solubility was determined by Hg0 mass loss to the organic solvent after the experiment. Gallup and Bloom (2010) also measured the solubility of Hg0 in C5 to n-dodecane over a temperature range of −65 to 65°C. Hg0 concentration was analyzed using SnCl2 reduction, purge, and trap on gold and electrochemical desorption of the mercury as Hg0 and quantification using cold vapor atomic fluorescence spectrometry. More recently, Marsh et al. (2016) measured the solubility of Hg0 in C5 to n-dodecane and in various alkane mixtures from 25 to 150°C. Solubility measurements of Hg0 at T > 25°C were made using a conventional absorption spectroscopic technique at λ = 253.6 nm. Results of these experimental solubility studies are shown in Fig. 1 .

Fig. 1

Solubility of Hg0 in alkanes and alkane mixtures from Clever and Iwamoto (1987), Miedaner et al. (2005), Gallup and Bloom (2010), and Marsh et al. (2016). The orange line is an extrapolation of the low temperature data from Clever and Iwamoto (1987). The blue line is the model fit to the solubility data of Miedaner et al. (2005). The red dashed line is the model fit to the solubility data of Marsh et al. (2016) (alkanes II mixture; yellow symbols).

Fig. 1

Solubility of Hg0 in alkanes and alkane mixtures from Clever and Iwamoto (1987), Miedaner et al. (2005), Gallup and Bloom (2010), and Marsh et al. (2016). The orange line is an extrapolation of the low temperature data from Clever and Iwamoto (1987). The blue line is the model fit to the solubility data of Miedaner et al. (2005). The red dashed line is the model fit to the solubility data of Marsh et al. (2016) (alkanes II mixture; yellow symbols).

Close modal

In general, linear extrapolation of the low-temperature Hg0 solubility data from Clever and Iwamoto (1987) appears to be in reasonable agreement with the high-temperature solubility data of Miedaner et al. (2005). The exponential solubility model from Gallup and Bloom (2010) is clearly inconsistent with results from most other studies. The low-temperature solubility data of Marsh et al. (2016) is an almost perfect overlay for the data of Clever and Iwamoto (1987) but at higher temperatures deviates to lower solubilities than that reported by Miedaner et al. (2005). It should be noted that there is a high degree of internal consistency in the Marsh et al. (2016) data over a significant range of alkanes and alkane mixtures. Recent thermodynamic modeling of Hg0 solubility in n-octane by Koulocheris et al. (2018), using different equation-of-state models, suggests that the Hg0 solubility data for toluene in Miedaner et al. (2005) are not consistent with many other experimental studies on which various equation-of-state models are based. Their Hg0 solubility in n-octane appears to be robust, but the veracity of their data has been challenged by Marsh et al. (2016).

Which Solubility Model Should be Used in Modeling Calculations?

Most Hg0 solubility studies were performed at relatively low temperatures (less than ~65°C), requiring significant extrapolation of models to the temperature ranges of interest (~100–200°C). Extrapolation of models based on either Clever and Iwamoto (1987) or Gallup and Bloom (2010) are therefore not considered further, despite their apparent statistical robustness. Miedaner et al. (2005) reported Hg0 solubility in octane, dodecane, and toluene at high temperatures ranging from 110 to 225°C. The only study that reports results for Hg0 solubility in alkanes and their mixtures over a significant temperature range is that of Marsh et al. (2016, alkane mixture II in their Table 3). Their low-temperature data (< 65°C) are in excellent agreement with the results of Clever and Iwamoto (1987) and imply a subtle decrease in the degree of Hg0 solubility with increase in temperature (>150°C). The Marsh et al. (2016) model is preferred at this time, as little extrapolation is required beyond 150°C to model most of the highest temperature data currently in our global data set.

Mercury Forms in Hydrocarbons

It is beyond the scope of this study to detail the numerous naturally occurring chemical forms of Hg in natural hydrocarbons. Hg occurs in four main forms in nature relevant to this work. These are typically characterized as being either dissolved or particulate components of different Hg species. The most common form of Hg occurrence in hydrocarbons is Quicksilver or dissolved metallic mercury, Hg0, and cinnabar, HgS (Ezzeldin et al. 2016). HgS is highly insoluble in water and is difficult to quantify due to its very fine-grained nature in recovered fluid samples. Organic Hg species also have very low solubility in water (Fein and Williams-Jones 1997). As shown by Fein and Williams-Jones (1997), Hg0(aq) is also likely to be very low.

It is apparent from the preceding discussion and the data in Table 2 that the solubility of Hg0 in liquid hydrocarbons is quite high. At 200°C, for example, Hg0 solubility in C8 is 821 ppm, while the solubility of Hg0 in liquid C5-C10 is 906 ppm. Comparison of the experimental data in Table 2 with reported THg data in natural hydrocarbons in Table 1 shows that the solubility of Hg0 in a range of alkane compositions is approximately three orders of magnitude greater than that observed in natural hydrocarbons. In the following section, these results are compared with solubility of Hg0 dissolved in hydrocarbons in equilibrium with cinnabar, HgS.

Table 2

Comparison of mercury solubility in alkanes C5-C10 from Clever and Iwamoto (1987) and C8 and dodecane from Miedaner et al. (2005).

C5-C10OctaneDodecane
T (°C)T (K)1000/T (K)logmppmlogmppmlogmppm
110 383.15 2.610 –3.417 77 –4.51 ± 0.03 54.2 –4.19 ± 0.07 76 
150 423.15 2.363 –2.884 262 –3.98 ± 0.09 184 –3.69 ± 0.03 240 
158 431.15 2.319 –2.790 326 –3.33 ± 0.06 821 –3.26 ± 0.09 647 
200 473.15 2.113 –2.345 906 
210 483.15 2.070 –2.251 1,126 –3.05 ± 0.04 1,002 
C5-C10OctaneDodecane
T (°C)T (K)1000/T (K)logmppmlogmppmlogmppm
110 383.15 2.610 –3.417 77 –4.51 ± 0.03 54.2 –4.19 ± 0.07 76 
150 423.15 2.363 –2.884 262 –3.98 ± 0.09 184 –3.69 ± 0.03 240 
158 431.15 2.319 –2.790 326 –3.33 ± 0.06 821 –3.26 ± 0.09 647 
200 473.15 2.113 –2.345 906 
210 483.15 2.070 –2.251 1,126 –3.05 ± 0.04 1,002 

Any model for the solubility of Hg0 to Hg0(org) includes an equilibrium constant logK term for the reaction:

Hg(org)0Hg0.
(1)

The equilibrium constant for the reaction in Eq. 1 is given by:

K(1)=aHg0/aHg(org)0,
(2)
LogK(1)=LogaHg0LogaHg0(org).
(3)

Because the activity of pure Hg0 is equal to 1, the equilibrium constant for the reaction in Eq. 1 is

LogK(1)=LogmHg0(org).
(4)

The model adopted in this study is based on the solubility data of Hg0 in alkane mixture II + C8 from Marsh et al. (2016) shown in Fig. 1. The solubility is given by:

Logm=0.26981000/T20.07821000/T1.7218,
(5)

where T is in Kelvin and for which the R2 = 0.9966.

Hg Equilibrium with Organic Liquids and H2S

The reaction of Hg0(org) with H2S and HgS was initially proposed by Fein and Williams-Jones (1997) and is given by Eq. 6:

Hg(org)0+HS+H++0.5O2HgS(cinnabar)+H2O.
(6)

This equilibrium may also be written by combining the following three reactions:

H2S(g)=HS+H+                                                        (a)+Hg(l)0+0.5O2(g)+H++HS=HgS(s)+H2O               (b)+Hg(org)0=Hg(l)0                                                                (c)=Hg(org)0+H2S(g)+0.5O2(g)=HgS(s)+H2O,
(7)

where the equilibrium constant for the reaction in Eq. 7 is given by:

LogK(7)=LogaH2O+LogaHgS(s)0.5LogfO2(g)LogfH2S(g)LogmHg0(org),
(8)

and the concentration of Hg0(org) is given by:

LogmHg(org)0=LogaH2O+LogaHgS(s)0.5LogfO2(g)LogfH2S(g)LogK(8).
(9)

As HgS(s) is a pure stoichiometric mineral, its thermodynamic activity is equal to 1. Results of previous work on the Fe-chlorite-pyrite-H2S model by Bryndzia and Inan Villegas (2020) contains all the necessary inputs to test the Hg0 solubility model in Eq. 9. The thermodynamic data necessary to calculate the equilibrium constants for the reaction in Eq. 9 are in the SUPCRT92 thermodynamic database, which is the most extensive and internally consistent thermodynamic database available for geochemical calculations (Johnson et al. 1992). Thermodynamic data in this database are valid far beyond the pressure and temperature conditions present in oil and gas fields. Input for the calculation is a balanced reaction at reservoir pressure and temperature. Water activity is a function of salinity and Bryndzia and Inan Villegas (2020) assumed a default water activity of 0.93, equivalent to a salinity of ~100 kppm NaCl equivalent brine. Also inherent in their model is a default value for the aFe-chlorite of ~0.65. The other significant input variable required to estimate the fugacity of H2S is oxygen fugacity (i.e., fO2), which may be estimated using Eq. 10 from Bryndzia and Inan Villegas (2020):

LogfO2=0.00036T2(oC)+0.26312T(oC)81.08970.
(10)

Eq. 10 has a correlation coefficient R2 = 0.9998.

Hg(org)0 Equilibrium with HgS

The reaction in Eq. 7 accounts for the ubiquitous presence of Hg0, HgS (cinnabar), and H2S in hydrocarbon liquids. Model inputs are derived from the Fe-chlorite-pyrite-H2S model as described in the preceding section. Table 3 summarizes the calculated Hg0 solubility in organic liquids for the same temperatures shown in Table 2. The results of the present study are in very good agreement with the calculated solubility of Hg0 in equilibrium with HgS at 150°C determined by Fein and Williams-Jones (1997).

Table 3

Solubility of mercury in alkane liquids in equilibrium with cinnabar (HgS) at the same temperatures as shown in Table 2. Data for C5-C10 are from Clever and Iwamoto (1987) and for C8 are from Miedaner et al. (2005).

C5-C10Octane
T (°C)logKlogm Hg0(org)ppmlogKlogm Hg0(org)ppm
110 34.95 –3.069 171 35.12 –3.239 116 
150 30.89 –2.558 556 31.00 –2.673 426 
200 26.77 –2.042 1,819 26.83 –2.104 1,580 
C5-C10Octane
T (°C)logKlogm Hg0(org)ppmlogKlogm Hg0(org)ppm
110 34.95 –3.069 171 35.12 –3.239 116 
150 30.89 –2.558 556 31.00 –2.673 426 
200 26.77 –2.042 1,819 26.83 –2.104 1,580 

The results of solubility calculations in Table 3 show that the predicted solubility of Hg0(org) in alkanes in equilibrium with HgS is in the hundreds to thousands of parts per million range, which is at least three orders of magnitude greater than is commonly observed in natural hydrocarbons (Table 1). Clearly, a better understanding of assumptions inherent in Eq. 7 is required if a rigorous predictive model for the solubility of Hg in hydrocarbons is to be developed.

The first step in this process is to reconsider the validity of Eq. 7. For example, the single underlying assumption with respect to Eq. 7 is that aHgS = 1 (i.e., it is assumed to be pure stoichiometric HgS at subsurface conditions). One possibility is that HgS may not be a pure mineral phase; for example, rather than stoichiometric HgS, the mineral phase might have a nonstoichiometric composition such as Hg1-xS (Potter and Barnes 1978) or exist as a solid solution with another sulfide phase such as pyrite (FeS2) or sphalerite (ZnS), with which HgS forms a complete solid solution series [i.e., (Zn,Hg)S] (Gabby and Eisele 2013). Assuming that aHgS ~ XHgS (where XHgS is the mole fraction of HgS in either pyrite or sphalerite), a Hg concentration of between 100 and 1,000 ppm in either pyrite or sphalerite would reduce the aHgS from unity to 10−3 or even lower (Krupp 1988).

Estimating the aHgS in Cinnabar

Using the experimental data from Potter and Barnes (1978), we investigated the range of nonstoichiometry that might exist in Hg1-xS at representative subsurface conditions. At the temperature range of interest to this study (i.e., ≤250°C), the compositional range of nonstoichiometry in cinnabar (i.e., the field of stability for Hg1-xS) is relatively limited at < 2 atomic % Hg at 200°C. Calculated activities of HgS in nonstoichiometric cinnabar (Hg1-xS) in the range of temperatures from 100 to 250°C ranged from unity in equilibrium with stoichiometric cinnabar (HgS) to a minimum value of 0.057 at 100°C to 0.340 at 250°C for nonstoichiometric cinnabar in equilibrium with sulfur. These activities of HgS, although less than unity, are still orders of magnitude too high to explain the observed concentrations of Hg0 in naturally occurring hydrocarbons. It may be concluded that in subsurface formations the aHgS is very likely <1, but what is the correct order of magnitude?

The equilibrium concentration of Hg solution species is directly proportional to the mole fraction of Hg in the solid solution phase (i.e., XHgSSph for sphalerite or XHgSPy for pyrite). The activity of HgS therefore depends greatly on what XHgS is in a sulfide phase at subsurface reservoir conditions. The dominant sulfide phase in organic-rich-shale source rocks and coals is pyrite, FeS2 (Kolker et al. 2012; Diehl et al. 2004; Yudovich and Ketris 2005). In low-sulfur coals, which are generally less enriched in Hg, only two Hg hosts dominate: Hg(org) and Hg(sulfide) (commonly pyritic). High-sulfur coals usually contain higher amounts of Hg, which is typically pyritic mercury.

The most important type of Hg enrichment in coals is due to an epigenetic, low temperature hydrothermal process; the Nikitovka (Ukraine), Warrior (USA), and Guizhou (China) basins are good examples. In the latter two coal basins, the paragenetic association Hg-As-Au-Tl is like Carlin-type Au deposits (Yudovich and Ketris 2005). The most notable Hg-rich coal basin is the Donbas-Donetsk Basin in Ukraine in which both HgS and metallic mercury (Hg0) occur. According to Yudovich and Ketris (2005), the Hg0 in Donbas was due to hydrothermal mineralization passing through sheared zones. The highest measured Hg content of pyrite found in Donbas Basin coals is 117 ppm based on analyses of 70 samples from the h8 coal seam (Dvornikov and Kirikilitsa 1987, p. 79).

Manceau et al. (2018) used high-energy-resolution XANES (HR-XANES) spectroscopy to quantitatively assess the chemical forms of mercury in pyrite from different geological settings, including coal seams. They investigated three different samples of coal, including bituminous coal from Illinois, USA, and a sub-bituminous coal from Bapung, India. Their results showed that in coals, 100% of the mercury exists as Hg2+, fully substituting for Fe2+ in the FeS2 structure. The concentration of Hg in these coal samples ranged from 0.3 ppm in a light coal up to 182.3 ppm in the sub-bituminous coal from India (Manceau et al. 2018, their Table 1).

More recent studies by Bourdet et al. (2020, 2021) have focused on the concentration and distribution of Hg and its host mineral(s) in reservoir rocks from the giant Gorgon gas condensate field on the NWS of Australia. We report here data for 22 whole-rock geochemical analyses for sulfur, total organic carbon (TOC), and Hg measured in core samples from the North Gorgon-1 well (Table 4) . Bulk whole-rock geochemical analyses were performed by ALS Laboratory in Perth, Western Australia, and combine whole-rock analysis, trace element analysis by fusion, aqua-regia digestion for the volatile trace elements, TOC, and nonorganic carbon and sulfur by combustion analysis. Approximately 10 to 20 g of rock was crushed to a powder having a maximum grain size of ~76 µm. The major oxides were analyzed on fused beads of sample which were subsequently acid digested and analyzed using inductively coupled plasma atomic emission spectroscopy. Base metals were quantified via aqua-regia digestion followed by inductively coupled plasma mass spectroscopy analysis. Total sulfur and total carbon (carbonate and organic) abundances were acquired using infrared spectroscopy after Leco furnace treatment. There was insufficient sample material for analysis of Samples 1 and 12.

Table 4

Whole-rock geochemical analyses for sulfur, TOC, and Hg measured in core samples from the North Gorgon-1 well.

SampleDepth
(m)
S
(wt%)
Fe
(wt%)
TOC (wt%)C
(wt%)
Total C (wt%)Hg (ppm)
3498.6       
3572 7.88 8.15 0.43 0.02 0.45 0.658 
3572.6 1.45 3.51 0.97 0.18 1.15 0.334 
3576.2 0.03 39.22 0.34 7.46 7.8 0.019 
3746 0.53 19.64 0.18 4.29 4.47 0.117 
3793.8 0.13 4.4 0.11 0.15 0.26 0.112 
3799.9 0.15 11.21 0.24 1.48 1.72 0.129 
3933.05 0.18 8.4 0.15 1.1 1.25 0.19 
3950 1.34 4.51 0.12 0.16 0.28 0.649 
10 3951.5 0.79 3.61 1.57 0.28 1.85 0.54 
11 3954.3 1.91 6.19 0.09 0.98 1.07 1.935 
12 4028.2       
13 4036.6 0.14 0.18 0.34 0.52 0.269 
14 4066.3 2.57 2.04 0.16 0.02 0.18 2.72 
15 4070.9 0.06 10.07 0.11 1.88 1.99 0.23 
16 4075.8 0.26 30.13 0.28 6.69 6.97 0.358 
17 4079.8 0.62 2.14 0.16 0.04 0.2 0.676 
18 4153.8 0.09 2.6 0.09 0.29 0.38 0.189 
19 4233.9 0.05 2.02 0.13 0.58 0.71 0.193 
20 4237.2 0.72 2.52 1.93 0.22 2.15 0.836 
21 4238.2 0.04 19.92 0.18 4.82 0.095 
22 4240.5 0.04 3.81 0.84 1.25 2.09 0.189 
23 4324.4 0.08 3.61 0.17 0.1 0.27 0.282 
24 4332.4 1.34 2.35 0.04 0.21 0.25 0.422 
SampleDepth
(m)
S
(wt%)
Fe
(wt%)
TOC (wt%)C
(wt%)
Total C (wt%)Hg (ppm)
3498.6       
3572 7.88 8.15 0.43 0.02 0.45 0.658 
3572.6 1.45 3.51 0.97 0.18 1.15 0.334 
3576.2 0.03 39.22 0.34 7.46 7.8 0.019 
3746 0.53 19.64 0.18 4.29 4.47 0.117 
3793.8 0.13 4.4 0.11 0.15 0.26 0.112 
3799.9 0.15 11.21 0.24 1.48 1.72 0.129 
3933.05 0.18 8.4 0.15 1.1 1.25 0.19 
3950 1.34 4.51 0.12 0.16 0.28 0.649 
10 3951.5 0.79 3.61 1.57 0.28 1.85 0.54 
11 3954.3 1.91 6.19 0.09 0.98 1.07 1.935 
12 4028.2       
13 4036.6 0.14 0.18 0.34 0.52 0.269 
14 4066.3 2.57 2.04 0.16 0.02 0.18 2.72 
15 4070.9 0.06 10.07 0.11 1.88 1.99 0.23 
16 4075.8 0.26 30.13 0.28 6.69 6.97 0.358 
17 4079.8 0.62 2.14 0.16 0.04 0.2 0.676 
18 4153.8 0.09 2.6 0.09 0.29 0.38 0.189 
19 4233.9 0.05 2.02 0.13 0.58 0.71 0.193 
20 4237.2 0.72 2.52 1.93 0.22 2.15 0.836 
21 4238.2 0.04 19.92 0.18 4.82 0.095 
22 4240.5 0.04 3.81 0.84 1.25 2.09 0.189 
23 4324.4 0.08 3.61 0.17 0.1 0.27 0.282 
24 4332.4 1.34 2.35 0.04 0.21 0.25 0.422 

Petrographic and automated energy dispersive secondary X-ray mineral maps of the samples did not reveal cinnabar in any of the core samples. Pyrite is the dominant sulfide mineral together with only trace abundances of pyrrhotite (Fe1-xS), chalcopyrite [(Cu,Fe)S2], and sphalerite (ZnS). Fig. 2 plots the bulk mercury contents of rock samples against their TOC (black dots, left axis) and sulfur contents (orange dots, right axis). Each sample is represented by two data points (TOC-Hg and S-Hg), linked by a straight line for clarity. The gray dashed lines represent isopleths of mercury concentration in organic matter or in sulfides, assuming that all the Hg is contained in these phases. The aim of the graph is to assess where the mercury resides. For example, Samples 11 and 14 both have high sulfur and low TOC values. Hg in those samples is more likely to be associated with sulfides containing an average of ~100 ppm Hg. Sample 20 has a higher TOC value and would contain ~30 to 40 ppm Hg if all the Hg were in the organic matter and ~80 to 100 ppm if all of the Hg were contained in sulfides Here, sulfides and TOC seem to show a consistent relationship, with both sulfur and mercury likely sourced from the organic matter.

Fig. 2

Bulk Hg concentration in different lithologies sampled in the North Gorgon-1 well. Vertical segments connect TOC and sulfur values from the same sample. The dashed gray lines represent isopleths of Hg concentration assuming that all of the Hg is hosted within pyrite or organic matter.

Fig. 2

Bulk Hg concentration in different lithologies sampled in the North Gorgon-1 well. Vertical segments connect TOC and sulfur values from the same sample. The dashed gray lines represent isopleths of Hg concentration assuming that all of the Hg is hosted within pyrite or organic matter.

Close modal

In-situ Hg contents in pyrite and sphalerite were acquired using laser ablation and inductively coupled plasma mass spectrometry. The maximum values of Hg in pyrite were 100 to 145 ppm, while in rare samples of sphalerite, values up to ~1,100 ppm Hg were observed. Trace levels of sphalerite with high Hg were more commonly observed in siderite-rich (FeCO3) samples (Fig. 3),. The highest reported Hg concentration in sphalerite and siderite-rich samples are found in the water leg and based on fluid inclusion and related diagenetic studies appear to have a hydrothermal origin (Bourdet et al. 2021). Fig. 3 summarizes the distribution of Hg in pyrite from the North Gorgon-1 well in the Gorgon gas condensate field.

Fig. 3

Hg abundance in sulfides from different lithologies in the North Gorgon-1 well. The data show that Hg concentrations in pyrite range from ~20 to 145 ppm and up to 450 ppm in siderite-rich rocks. Hg concentrations in pyrite appear to reach a maximum concentration of ~120 ppm Hg.

Fig. 3

Hg abundance in sulfides from different lithologies in the North Gorgon-1 well. The data show that Hg concentrations in pyrite range from ~20 to 145 ppm and up to 450 ppm in siderite-rich rocks. Hg concentrations in pyrite appear to reach a maximum concentration of ~120 ppm Hg.

Close modal

Based on published data for Hg concentrations in pyrite and assuming simple ionic substitution of Hg2+ for Fe2+ in the pyrite structure, it is possible to estimate XHgS in pyrite, using Eq. 11.

XHg=nHg/(nFe+nHg),
(11)

where XHg is the mole fraction of Hg in pyrite and n is the number of moles of each component. In this calculation scheme, XHg= XHgS in pyrite and ranges from 2.79×10−4 to 2.80×10−3 for Hg concentrations of 1 to 100 ppm Hg, respectively. If it assumed that the XHgS in pyrite is approximately equal to the aHgS in pyrite, then the calculation of Hg0(org) using the equilibrium constant for Eq. 7 will generate predictions of Hg0(org) at subsurface conditions approximately three orders of magnitude lower than values previously discussed for pure hydrocarbons (Table 3), in line with values observed in natural hydrocarbons shown in Table 1.

Concentrations of Hg0 in Natural Hydrocarbons

Despite the voluminous literature available on the solubility of Hg0 in organic solvents and naturally occurring hydrocarbons, it has proved challenging to compile a quality set of Hg0(org) data. Many published papers quote a Hg concentration without including basic subsurface properties such as reservoir pressure or temperature, let alone the density of the gas, a basic fluid property required to estimate the total concentration, by mass, of Hg0 (THg) in the hydrocarbon gas phase. Hg concentrations in gas are normally reported as µg/m3, while liquid concentrations of Hg are reported as µg/L or ppbw [i.e., parts per billion (mass)].

Testing the Hg(org)0 Equilibrium Model with aHgS in Natural Samples

The Hg0(org) solubility model given by Eq. 7 was applied to the data set used to calibrate and test the chlorite-pyrite-H2S model developed by Bryndzia and Inan Villegas (2020). This data set was used to model the fH2S in several global hydrocarbon clastic reservoirs and already contains the necessary inputs to test the model, namely T, P, fO2, fH2S, aH2O, estimated for a default XFe in chlorite ~0.65. The only input variable that is unconstrained is the aHgS. Many scenarios were tested using a range of inputs for aHgS into Eq. 9 and solving for the solubility of Hg0(org). These model results are then compared with reported values of Hg0(org) from a global suite of gas, condensate, and oils.

Fig. 4 summarizes the results of testing various models for the aHgS and its impact on the model calculation of Hg0(org). A range of aHgS values from pure stoichiometric cinnabar (a = 1) to a = 0.001 were modeled using Eq. 9. The impact of this variable on estimated logmHg0(org) is obvious. A reduction in the aHgS from 1 to 0.001 results in a reduction of Hg0(org) solubility by three orders of magnitude. The important question to answer is where do measured concentrations of Hg0(org) in hydrocarbons plot? To answer this question, we have plotted our global compilation of reported Hg abundances in hydrocarbon gas, oil, and condensate liquids on Fig. 5. The data are summarized in Table S-1 (Supplementary Materials).

Several important features in Fig. 5 warrant further elaboration:

  1. The observed ranges of Hg0(org) in natural hydrocarbons define a relatively narrow range of model cinnabar activities, aHgS between 0.01 and 0.001, and are mostly distributed around the 0.003 aHgS isopleth. This is an important discovery as it confirms that the presence of pure stoichiometric cinnabar, having an aHgS = 1, is thermodynamically not to be expected in hydrocarbon reservoirs and subsurface formations, including coals and organic-rich source rocks. These results confirm that if cinnabar is present, then it is only as a minor component hosted within another sulfide phase, most likely pyrite (FeS2) and/or sphalerite (ZnS). This has important implications for the accounting of Hg species used to estimate THg and means that the dominant species of Hg in hydrocarbons is Hg0(org), possibly with subordinate concentrations of other Hg-bearing organic moieties.

  2. The dotted black lines in Fig. 5 represent upper and lower bounds of logmHg0(org) ± 0.24 for the Hg0(org) solubility model with aHgS = 0.003, estimated for data in temperature range of 100 to 120°C. The mean temperature of this subset of data is 109 ± 0.24°C (±1σ; n = 24). The total number of data points plotted in Fig. 5 is 51, 10 of which lie outside of these bounds (i.e., the new solubility model), with an aHgS = 0.003 and ±1σ of 0.24, accounts for ~80% of the global data set. It should also be noted that the average concentrations of Hg0(org) in gas from Ordos and northern NL as well as the condensate data from northern NL also plot on the aHgS = 0.003 isopleth at their respective average temperatures (Table S-1 in Supplementary Materials).

  3. Despite large analytical uncertainties associated with samples from the NWS, the average values for the samples plotting in the temperature range of 140 to 160°C are also consistent with an average activity of cinnabar of ~0.003, though with much larger associated analytical uncertainties (Table S-1; Supplementary Materials).

  4. The Hg0(org) for the Minami-Nagaoka gas field shown in Fig. 5 has been calculated using fH2S from the chlorite-pyrite-H2S model of Bryndzia and Inan Villegas (2020) for reservoir conditions given by Kawamoto and Sato (2000). This is a moderately high-pressure and high-temperature gas condensate field hosted in a heavily chloritized and brecciated rhyolitic volcanic reservoir. The only published Hg0(org) data from this field is from Well K. The reported Hg0(org) value is in excellent agreement with measured Hg0(org) contents from many other hydrocarbon-bearing clastic reservoirs and is in reasonable agreement with the results of the thermodynamic Hg0(org) solubility model and an aHgS of 0.003.

  5. There are relatively few samples of Hg0(org) from black oils in the current global data set. One exception is an oil sample from the Norphlet Formation in the eastern Gulf of Mexico (solid red square in Fig. 5). For its reported concentration of Hg0(org) and present-day reservoir temperature (~160°C), it also plots in excellent agreement with the thermodynamic model having a cinnabar activity of ~0.003.

  6. The other striking feature captured in Fig. 5 is that much of the data has a strong subsurface association bias with coal. The source rocks for much of the Rotliegend reservoired hydrocarbons in the North German Basin and northern NL fields are known to be deeply buried Carboniferous Westphalian coals and organic-rich sediments (Krooss et al. 1995; Hoffmann et al. 2001; Gaupp and Okkerman 2011) and are also believed to be the source of Hg associated with produced gas from the Rotliegend Formation in the NL and Germany (Lokhorst 1997). However, Zettlitzer et al. (1997) attribute the high levels of Hg in gas from the North German Basin to underlying “volcanites” and hence believe the source of Hg to be volcanic in origin. Samples from the Ordos Basin (China) are reported as being gas sourced directly from deeply buried coals (Li et al. 2019, their Table 2). Yolla (Bass Strait), and many samples from the NWS also have reported occurrences of coal in their subsurface stratigraphy. The high-pressure and high-temperature sample from the Central North Sea in Table S-1 (Supplementary Materials) is intimately associated with the Pentland coal, and Salam gas condensate samples from the Western Desert of Egypt are also believed to be sourced from Jurassic coals (Keeley et al. 1990; Shafik and Saudi 2018). Coals are also reported in the subsurface stratigraphy for the carbonate-hosted gas condensate sample from the Philippines, including syn-rift sediments believed to host source rocks for petroleum fields in the north Palawan basin, Philippines (Sales et al. 1997). It is unknown if samples from eastern Gulf of Mexico, Baram Delta, and Oman have any association with coal-bearing source rocks.

  7. The upper bound of Hg concentrations in natural hydrocarbons at the approximate midpoint of temperatures in the global data set at ~130°C is logmHg0(org) ~ −5.5. This upper bound is broadly consistent with data in Table 1 (discussed previously). This is the first time that a model prediction for Hg solubility in hydrocarbons has been able to demonstrate consistency with observed Hg concentrations in natural hydrocarbons at subsurface reservoir conditions.

Fig. 4

Estimated isopleths of Hg0(org) solubility from Eq. 9 using the chlorite-pyrite-H2S model and calibration data in Bryndzia and Inan Villegas (2020). The dashed red line represents the activity of HgS1-x assuming nonstoichiometry in cinnabar based on the model of Potter and Barnes (1978).

Fig. 4

Estimated isopleths of Hg0(org) solubility from Eq. 9 using the chlorite-pyrite-H2S model and calibration data in Bryndzia and Inan Villegas (2020). The dashed red line represents the activity of HgS1-x assuming nonstoichiometry in cinnabar based on the model of Potter and Barnes (1978).

Close modal
Fig. 5

Plot of Hg0(org) in hydrocarbons from the global data set used to calibrate the aHgS in cinnabar. Samples from the Plover Formation (NWS) are shown as solid diamonds. Vertical solid lines reflect the observed ranges of reported values for the plotted average concentrations of total mercury in Table S-1 (Supplementary Materials) (i.e., THg). The solid teal circle is the mean ±1σ for samples from the NWS. The solid red circle is the mean ±1σ for gas samples from northern NL. The solid lime green square is the mean ±1σ for northern NL condensate liquids. The solid pink circle is the mean ±1σ for Ordos gas samples. Hg0(org) for Minami-Nagaoka was calculated assuming an aHgS of 0.003 at reservoir conditions (pink triangle). Well K shows the reported Hg0(org) from the Minami-Nagaoka Field (lilac diamond). For the Longtom-2 well, Hg0(org) was estimated using the reported concentration of H2S. Isopleths of aHgS were estimated using the H2S model of Bryndzia and Inan Villegas (2020). Philippines gas condensate sample is from a carbonate reef reservoir. Dotted black lines represent upper and lower bounds of logmHg0(org) ± 0.24 for an aHgS = 0.003 based on data in the temperature range of 100 to 120°C. Gorgon data (solid pink square) is the reported Hg solubility for the O Sand reservoir in the North Gorgon-1 well. The blue squares show Hg solubility estimated using the reported H2S concentration of 17 ppm, with aHgS estimated for pyrite containing 100-, 200-, and 300-ppm Hg, respectively, as discussed in the text.

Fig. 5

Plot of Hg0(org) in hydrocarbons from the global data set used to calibrate the aHgS in cinnabar. Samples from the Plover Formation (NWS) are shown as solid diamonds. Vertical solid lines reflect the observed ranges of reported values for the plotted average concentrations of total mercury in Table S-1 (Supplementary Materials) (i.e., THg). The solid teal circle is the mean ±1σ for samples from the NWS. The solid red circle is the mean ±1σ for gas samples from northern NL. The solid lime green square is the mean ±1σ for northern NL condensate liquids. The solid pink circle is the mean ±1σ for Ordos gas samples. Hg0(org) for Minami-Nagaoka was calculated assuming an aHgS of 0.003 at reservoir conditions (pink triangle). Well K shows the reported Hg0(org) from the Minami-Nagaoka Field (lilac diamond). For the Longtom-2 well, Hg0(org) was estimated using the reported concentration of H2S. Isopleths of aHgS were estimated using the H2S model of Bryndzia and Inan Villegas (2020). Philippines gas condensate sample is from a carbonate reef reservoir. Dotted black lines represent upper and lower bounds of logmHg0(org) ± 0.24 for an aHgS = 0.003 based on data in the temperature range of 100 to 120°C. Gorgon data (solid pink square) is the reported Hg solubility for the O Sand reservoir in the North Gorgon-1 well. The blue squares show Hg solubility estimated using the reported H2S concentration of 17 ppm, with aHgS estimated for pyrite containing 100-, 200-, and 300-ppm Hg, respectively, as discussed in the text.

Close modal

Deployment of the Hg Solubility Model

Deployment of the new Hg solubility model requires an estimate of both oxygen and sulfur fugacity (i.e., fO2 and fH2S) and an estimate of formation brine salinity at reservoir pressure and temperature. In the absence of reported concentration of H2S, a model value may also be used from which fH2S may be estimated using the Fe-chlorite-pyrite-H2S model of Bryndzia and Inan Villegas (2020). The workflow for estimating these properties for clastic hydrocarbon-bearing reservoirs has been discussed previously by Bryndzia and Inan Villegas (2020). By way of example, we demonstrate our preferred workflow for estimating the solubility of Hg0(org) for a suite of gas condensate samples from the NWS of Australia. The data are summarized in Table S-2 (Supplementary Materials).

Fig. 6 shows both reported and measured values of H2S from the NWS gas condensate data. An initial quality control step that we adopt is to select values of model H2S concentration that lie within ±3 ppm of the values estimated from the Fe-chlorite-pyrite-H2S model of Bryndzia and Inan Villegas (2020). The next step is to estimate the solubility of Hg0(org) using the new Hg solubility model using Eq. 9, as previously discussed. The results are summarized in Table S-2 (Supplementary Materials) and are plotted in Fig. 7 .

Fig. 6

Plot of reported vs. measured concentrations of H2S in gas condensate from subbasins on the NWS of Australia. Data are from Bryndzia and Inan Villegas (2020), their Table 1) samples in which the respective values were within ±3 ppm; samples that passed this initial quality control step were then used to estimate the solubility of Hg in the gas condensates.

Fig. 6

Plot of reported vs. measured concentrations of H2S in gas condensate from subbasins on the NWS of Australia. Data are from Bryndzia and Inan Villegas (2020), their Table 1) samples in which the respective values were within ±3 ppm; samples that passed this initial quality control step were then used to estimate the solubility of Hg in the gas condensates.

Close modal
Fig. 7

Plot of model concentrations of H2S (a) and model Hg0(org) solubilities (b) as a function of reservoir pressure and temperature in gas condensate from subbasins on the NWS of Australia. Data are from Bryndzia and Inan Villegas (2020) (their Table 1) samples in which the respective values of H2S concentration were within ±3 ppm; samples that passed this initial quality control step were used to estimate the solubility of Hg0(org) in these gas condensates. Solid lines represent pressure isobars in kpsi (1 MPa = 145.04 psi).

Fig. 7

Plot of model concentrations of H2S (a) and model Hg0(org) solubilities (b) as a function of reservoir pressure and temperature in gas condensate from subbasins on the NWS of Australia. Data are from Bryndzia and Inan Villegas (2020) (their Table 1) samples in which the respective values of H2S concentration were within ±3 ppm; samples that passed this initial quality control step were used to estimate the solubility of Hg0(org) in these gas condensates. Solid lines represent pressure isobars in kpsi (1 MPa = 145.04 psi).

Close modal

It is apparent from Fig. 7b that the Hg0(org) solubility in gas condensates on the NWS of Australia appear to cluster by subbasin, suggesting that there may be some regional geological control on their distribution. There is minor overlap between some of the subbasins, but in general, the Hg0(org) concentrations tend to be tightly clustered, as for Subbasins 1, 3, and 5, for example. The most obvious control, though, appears to be reservoir temperature of the producing formations. These results appear to validate the original observations made by Cooper et al. (2017) that reservoir mercury results indicated a potential correlation of the bottomhole temperature with mercury concentration in produced fluids and that the equilibrium of Hg0 and HgS in the reservoir could be the basis for this correlation. The new Hg0(org) solubility model presented in this paper provides a thermodynamic basis for their observation and presents a workflow on how to deploy the model and permits, for the first time, quantitative predrill prediction of the concentration of Hg0(org) in hydrocarbon-bearing clastic reservoirs.

Model Validation: Case Study from the Gorgon Gas Condensate Field, NWS–Australia

The model developed to estimate the solubility of Hg0(org) in hydrocarbons includes assumptions of mole fraction of iron (XFe) in chlorite, water activity (i.e., salinity), and estimation of oxygen and sulfur fugacity, fO2 and fH2S, respectively. These have previously been evaluated by Bryndzia and Inan Villegas (2020) and will not be repeated here. One of the biggest uncertainties in our Hg solubility model is the assumed default value for the aHgS in Eq. 9. Using data from the O Sand reservoir in the North Gorgon-1 well from Bourdet et al. (2021), it is possible to evaluate the sensitivity of Hg solubility in hydrocarbons for this reservoir. The O Sand reservoir has a formation temperature of 154.9°C and a pressure of 6578 psi (42.8 MPa). It contains a relatively lean gas condensate with a condensate gas ratio of ~10 STB/MMscf. The gas condensate contains ~14 mole percent CO2 and has a reported H2S concentration of 17 ppm. Hg content of this reservoir is reported to be 1150 ± 119 µgm/m3 (n = 15; ±). The separator gas has a reported density of 746 kg/m3. This equates to a Hg content of 1,542 ± 159 ppb (logmHg0(org) = -5.114).

Table S-2 (Supplementary Materials) shows the model solubility of Hg in the O Sand gas condensate, estimated using the Fe-chlorite-pyrite-H2S of Bryndzia and Inan Villegas (2020). The model predicts 11.3 ppm H2S for the O Sand reservoir and a Hg solubility of 422 ppb, consistent with other estimates for subbasins on the NWS (Table S-2; Supplementary Materials) but much lower than the reported value of 1,542 ppb. In Table S-2 (Supplementary Materials), we have also included model estimates based on the reported value of 17-ppm H2S in the O Sand reservoir. Using logfH2S at reservoir conditions for 17-ppm H2S and a default aHgS of 0.003 increases the model Hg solubility to 630 ppb, which is still lower than the reported value of 1,542 ppb.

We further illustrate the sensitivity of the Hg solubility model to the aHgS by estimating the solubility of Hg using the reported Hg concentrations for pyrite shown in Fig. 3. Using Hg concentrations of 200 and 300 ppm gives estimated values for aHgS of 0.0056 and 0.0085, and Hg solubilities of 1,187 and 1,793 ppb, respectively, consistent with the reported Hg solubility of 1,542 ppb. These range of values have been plotted together with reported Hg0(org) solubility for the North Gorgon-1 well in Figs. 5 and 8. Using an aHgS of 0.0074, based on a Hg concentration of 260 ppm in pyrite would match the reported Hg concentration in the O Sand reservoir. This analysis suggests that the default value of aHgS of 0.003 (~100 ppm Hg in pyrite) may be too low for general application to some of the relatively high-temperature gas condensate reservoirs on the NWS of Australia. Based on the excellent agreement between model and measured Hg solubility for the O Sand in the North Gorgon-1 well, we believe that this establishes “proof of concept” for the new Hg solubility model and conclude that it may be used to provide reasonable estimates of Hg concentration in clastic hydrocarbon-bearing reservoirs at in-situ conditions. It is also reasonable to assume a default value for the aHgS of 0.003. However, some knowledge on H2S concentration, and hence logfH2S, is the critical parameter required to apply the Hg0(org) solubility model with confidence.

Fig. 8

Phase relations in the system Fe-S-O-H-Hg0(org), calculated for the O Sand reservoir in the North Gorgon-1 well. The solid blue symbol represents the model solubility of Hg in gas condensate having an H2S concentration of 11.3 ppm. The solid yellow symbol is the model Hg solubility for the reported H2S concentration of 17 ppm. The error bars for the solid blue symbol are ±1 logfO2 unit and ±0.1 logfH2S unit, plotted at 428.06 bar (1 MPa = 10 bar) and for a reservoir temperature of 154.9°C. Dashed lines represent solubility of Hg0(org) in hydrocarbons. The red solid line represents the XFe of 0.65 chlorite isopleth from the chlorite-pyrite-H2S model of Bryndzia and Inan Villegas (2020). Red arrow indicates the impact of oxidation (i.e., increase in fO2). The impact of oxidation is to induce precipitation of HgS at the expense of Hg0(org).

Fig. 8

Phase relations in the system Fe-S-O-H-Hg0(org), calculated for the O Sand reservoir in the North Gorgon-1 well. The solid blue symbol represents the model solubility of Hg in gas condensate having an H2S concentration of 11.3 ppm. The solid yellow symbol is the model Hg solubility for the reported H2S concentration of 17 ppm. The error bars for the solid blue symbol are ±1 logfO2 unit and ±0.1 logfH2S unit, plotted at 428.06 bar (1 MPa = 10 bar) and for a reservoir temperature of 154.9°C. Dashed lines represent solubility of Hg0(org) in hydrocarbons. The red solid line represents the XFe of 0.65 chlorite isopleth from the chlorite-pyrite-H2S model of Bryndzia and Inan Villegas (2020). Red arrow indicates the impact of oxidation (i.e., increase in fO2). The impact of oxidation is to induce precipitation of HgS at the expense of Hg0(org).

Close modal

The Presence of HgS and Quantification of THg

A persistent challenge in the quantification of THg in reported hydrocarbon production streams has focused on accounting for all mercury species observed in downhole samples and various separator stages on gas-processing facilities, most notably attributed to particulate Hg (i.e., solid HgS). However, the particulate Hg may also include contributions of Hg in other solid sulfide phases such as pyrite and sphalerite, which are not normally accounted for. The large associated analytical uncertainties in reported THg values for gas condensate from the NWS of Australia bring this into clear focus (solid vertical bars in Fig. 5). What is the best way to quantify THg in hydrocarbons? In this section, the argument will be made that estimation of THg may be a lot simpler than previously considered. Many of the mercury species previously included in the estimation of THg are most likely thermodynamically metastable and are relevant only to the oxidized near-surface environment where most downhole sampling takes place. They are not representative of the redox state, mercury concentration, or dominant mercury species representative of hydrocarbons at in-situ reservoir conditions or the long-term mercury concentration in stable hydrocarbon production. The mechanisms by which mercury deposits form in the Earth’s crust are used to explain this line of reasoning.

Precipitation Mechanisms for HgS

Precipitation of any mineral from a hydrothermal fluid can only occur in response to physical and/or chemical changes in the fluid. In the case of cinnabar precipitation, two mechanisms are important in nature (Krupp 1988). The first mechanism is applicable to settings in which Hg is transported in a high-sulfide fluid as a sulfide complex species. In this case, the most efficient precipitation mechanism is oxidation of the ore fluid since this will transform the weak (i.e., largely undissociated) acid H2S into the strong acid H2SO4, thus creating an acidic pH environment where solubilities of cinnabar are significantly lower when controlled by sulfide complexes of Hg. Simultaneously, the activity of the complexing agent H2S is reduced, thus destabilizing sulfide complexes:

HgS22+2O2HgS+SO42.
(12)

The second precipitation mechanism is important where Hg is transported as a Hg0(x) species (i.e., as Hg0(aq), Hg0(g) or Hg0(org)) in reduced, sulfur-poor fluids. Again, oxidation of the ore fluid is the most efficient way to precipitate cinnabar. In the case of Hg0(org), the fluid is initially more reduced, and the stable condensed phase is liquid Hg0 rather than cinnabar. Before H2S becomes oxidized to sulfate, the stability boundary of cinnabar is crossed (Figs. 2 and 3 in Krupp 1988), thus allowing the precipitation of the sulfide:

Hg(x)0+H2S+0.5O2HgS+H2O.
(13)

In both cases, oxidation of the ore fluid is most likely brought about by contact and mixing with oxygenated near-surface waters. Consequently, cinnabar deposition occurs typically at shallow depths where the Hg is recovered mostly through surface mining. Simple cooling of solutions saturated with Hg0(aq) will precipitate either liquid Hg0 or cinnabar (due to the larger stability field of HgS at lower temperatures) depending on temperature, redox conditions, and sulfur concentrations within the depositional environment (Fig. 3 in Krupp 1988; Peabody and Einaudi 1992; Fein and Williams-Jones 1997).

It is no coincidence, therefore, that the precipitation mechanism for HgS given by Eq. 13 is the same as that shown by Eq. 7 and which forms the basis of the solubility model developed in this study. It is the oxidation of H2S that results in the precipitation of particulate HgS that is so common in downhole samples and short-term well drillstem tests, for example. The impact of oxidation is to increase fO2 and decrease fH2S, which results in the precipitation of HgS at the expense of Hg0(org) as indicated by Eq. 13 and shown by the red arrow in Fig. 8. Oxidation of H2S was also recognized as the single biggest impediment impacting sampling of hydrocarbons containing low concentrations (< 5 ppm) of H2S (Bryndzia and Inan Villegas 2020).

By this same reasoning, it is also apparent that formation of HgS is unlikely to be associated with reaction of native sulfur (S0), as the most common and stable form of sulfur in reduced hydrocarbon-bearing environments is H2S and not S0. It was argued previously that the activity of HgS at reservoir conditions is so low that HgS is unlikely to ever be present as a stable phase in hydrocarbon reservoirs. Further, as shown by Krupp (1988) and Fein and Williams-Jones (1997), the solubility of aqueous mercury species at reservoir redox conditions is lower by more than two orders of magnitude than its solubility in hydrocarbons. This means that accounting for aqueous mercury species in the estimation of THg is also likely not warranted.

Occurrence and Origin of Particulate HgS in Flowback Fluids after Well Workover

An insightful study on the potential origin(s) of particulate mercury (HgS) was performed by Yamada et al. (2017) on a well in the Minami-Nagaoka Gas Field, Niigata Prefecture in Japan. For their well (Well K), they recorded particulate Hg in flowback fluids, both condensate and brine, during an extended period of flowback lasting almost 112 days. After ~2,700 hours of flowback, particulate Hg levels were greatly reduced, most notably in the condensate. Yamada et al. (2017) also measured the dissolved Hg contents in both condensate and flowback water (Table 5).

Table 5

Dissolved mercury in condensate and flowback water during well cleanup in Well K (Yamada et al. 2017).

Dissolved Hg in Condensate and Flowback Water
Sample NumberElapsed Time (hours)Dissolved Hg (μg/L)
Condensate (CVAAS*)Flowback Water
(ICP-MS)
272.3 9.2 
212.6 12.2 
35 270.4 6.5 
39 318.5 3.0 
56 411.3 1.8 
730 142.0 2.3 
2,389 293.9 1.0 
2,679 434.1 0.5 
Dissolved Hg in Condensate and Flowback Water
Sample NumberElapsed Time (hours)Dissolved Hg (μg/L)
Condensate (CVAAS*)Flowback Water
(ICP-MS)
272.3 9.2 
212.6 12.2 
35 270.4 6.5 
39 318.5 3.0 
56 411.3 1.8 
730 142.0 2.3 
2,389 293.9 1.0 
2,679 434.1 0.5 

ICP-MS - inductively coupled plasma mass spectroscopy; CVAAS = cold vapor atomic absorption spectroscopy.

*

These results are for reference purposes only.

Their results also show that the aqueous Hg species declined to less than 1 ppb after ~2,700 hours of flowback, while the concentration of Hg0(org) appears to have stabilized in the condensate liquid (Table 5). Based on their observations, Yamada et al. (2017) concluded that particulate HgS in flowback fluids was generated by the reaction of dissolved oxygen in workover fluids with H2S and Hg0 in the natural gas because particulate mercury concentration reached a peak early in the cleanup phase and then decreased with time. It is not considered to be produced during the stable phase of production after well workover because particulate mercury concentration was very low in the late phase of cleanup flow (Table 5). The last reported Hg concentration in Table 5 is 434 ppb which is believed to be representative of Hg in stable condensate production. This value is in reasonable agreement with a value of 525 ppb predicted for this reservoir based on forward modeling using fH2S estimated from the chlorite-pyrite-H2S model and shown in Fig. 5.

Their preferred model for the formation of cinnabar, shown in Fig. 9a, involves reaction of native sulfur (i.e., S0). However, native sulfur was not reported by Yamada et al. (2017). Rather than the model in Fig. 9a, we propose that direct oxidation of H2S during sampling and well workover is the most likely cause for HgS formation, as shown in Fig. 9b. The proposed reaction in Fig. 9b is the same as Eqs. 7 and 14, the latter proposed by Krupp (1988) to explain the deposition of cinnabar in hydrothermal mercury deposits. The two equations controlling the reaction in Fig. 9a are:

2H2S+O22H2O+2S0                   (i)+Hg0+S0HgS                                   (j)=Hg0+H2S+0.5O2HgS+H2O.
(14)
Fig. 9

Models for the formation of particulate mercury (HgS) by Yamada et al. (2017) involving reaction of Hg0 and native sulfur (a), and oxidation of H2S and reaction with Hg0(org) in gas condensate (b). Figure has been modified after Yamada et al. (2017).

Fig. 9

Models for the formation of particulate mercury (HgS) by Yamada et al. (2017) involving reaction of Hg0 and native sulfur (a), and oxidation of H2S and reaction with Hg0(org) in gas condensate (b). Figure has been modified after Yamada et al. (2017).

Close modal

The equilibrium constants for the HgS forming Eq. 14 shown in Fig. 9a is therefore identical to that for Eq. 7 shown in Fig. 9b, but without the involvement of native sulfur as an intermediate reaction phase.

Yamada et al. (2017) concluded that different forms of Hg are produced in the unsteady production phase preceding well workover. This is also the stage during which most downhole fluid samples are collected. These data suggest that it is only the Hg dissolved in the hydrocarbon gas and liquids (i.e., Hg0(org) collected after well flowback and cleanup) that is representative of the THg in subsurface hydrocarbon fluids. The fundamental challenge in reliably estimating THg in a representative reservoir fluid sample is to know how much of the particulate Hg (i.e., HgS) needs to be combined with concentrations of Hg0(org) measured in the hydrocarbon fluid. It is evident that most of the Hg in the particulate mercury fraction was derived from the produced hydrocarbons, but what is not well understood at this time is how much. Thermodynamically, we do not expect HgS to be stable in the reservoir, as aHgS < 1. Ignoring its presence in surface samples for the purpose of estimating THg in hydrocarbons may therefore be a reasonable strategy to adopt.

The distribution of Hg0(org) between hydrocarbon gas and liquids that are routinely measured in separators in gas-processing facilities is mostly a function of the process engineering deployed. The partitioning of Hg0(org) will be a function of the pressure and temperature changes experienced by the hydrocarbon fluids between in-situ reservoir conditions and surface gas-processing facilities, and the ambient redox state within the separator. The presence of particulate HgS in a separator sample may simply reflect the volume fraction of Hg0(org) that partitioned into the vapor phase at ambient redox state in the separator and was oxidized to HgS. Because the particulate HgS is metastable relative to reservoir redox state does not mean that including it with that of Hg0(org) measured in the separator hydrocarbon fluid is necessarily going to be representative of the of Hg0(org) in hydrocarbons in the reservoir. Recent studies on developing Hg0 and H2S inert sample chambers have also recognized the technical challenges posed by oxidation of these reactive species on the walls of uncoated sample chambers (Enrico et al. 2020). Results of these studies suggest that direct sampling of reservoir fluids, using truly inert sample chambers, may be the only way to obtain representative in-situ concentrations of Hg0 and H2S in hydrocarbon reservoirs.

Water is a very common accessory phase that is processed concurrently with hydrocarbons and is the most likely cause for the oxidation of hydrocarbons in gas-processing facilities. The example discussed for Well K in the previous section confirms this hypothesis. Table 5 shows that with extended flowback, the Hg content in produced water decreases to a negligible value while that of Hg0(org) in the condensate increase toward a stable value similar to what the new Hg0(org) solubility model would predict for in-situ reservoir conditions.

  • A thermodynamic equilibrium mineral-based model has been developed for predicting the solubility of mercury in hydrocarbons Hg0(org) at in-situ reservoir conditions. The model is based on experimental data of Marsh et al. (2016) on the solubility of Hg0 in a mixture of alkanes and equilibrium with Hg0, H2S, O2, cinnabar (HgS), and water.

  • As the model inputs for fH2S and fO2 are based on the chlorite-pyrite-H2S model of Bryndzia and Inan Villegas (2020), its application should be limited to clastic hydrocarbon-bearing reservoirs as they normally contain the minerals required to control the fH2S and equilibrium with hydrocarbons, water, and oxygen.

  • Uncertainty around the correct value of aHgS to use in solubility calculations was resolved by comparing reported concentrations of Hg in natural hydrocarbons with thermodynamically modeled HgS activities ranging from 1 to 0.001 and over a temperature range of 80 to ~180°C. A statistically robust match between model and measured Hg0(org) was achieved for an aHgS in cinnabar 0.003.

  • Mercury concentrations of up to ~120 ppm, measured in pyrite from coals and hydrocarbon reservoirs on the NWS of Australia, are consistent with a cinnabar activity of ~0.003. Such a low aHgS confirms that pure stoichiometric cinnabar is unlikely to be a stable mineral phase at the reducing conditions typical of hydrocarbon reservoirs and is a manifestation of near-surface oxidation.

  • The new Hg0(org) solubility model with an aHgS of 0.003 accounts for ~80% of the reported of Hg0(org) values in natural hydrocarbons and attests to the veracity of the new mercury solubility model.

  • A global data set of natural hydrocarbons containing Hg shows a remarkably strong association with the presence of coals in the subsurface. This is particularly the case for gas and lean gas condensates from the Ordos Basin in China, Rotliegend reservoirs in northern NL, gas from the North German Basin, Bass Strait and Gippsland reservoirs in Victoria (Australia), gas fields in the Western Desert of Egypt, high-pressure and high-temperature gas condensate from the Central North Sea, and in many well penetrations of the Jurassic Plover Formation in the NWS of Australia.

  • Accounting for the erratic concentrations of metastable particulate HgS and soluble aqueous mercury species in downhole fluid samples is the biggest source of analytical uncertainty impacting estimation of THg in hydrocarbons. Including estimates of these Hg species, most of which are metastable relative to reservoir conditions, will only result in overestimating the risk of mercury to topside facilities, infrastructure, operators, and the environment.

We thank Shell management for permission to publish the results of this study. Discussions and guidance provided by Bruce Lockyer, Paul Martin, and Amanda Grellman (Shell Australia Pty Ltd) are much appreciated. We thank Esra Inan Villegas (SIEP Inc.) for guidance with modeling calculations.

Original SPE manuscript received for review 5 May 2022. Revised manuscript received for review 4 July 2022. Paper (SPE 212271) peer approved 18 July 2022. Supplementary materials are available in support of this paper and have been published online under Supplementary Data at https://doi.org/10.2118/212271-PA. SPE is not responsible for the content or functionality of supplementary materials supplied by the authors.

Abu El Ala
,
M.
,
Nabawi
,
M. H.
, and
Azim
,
M. A
.
2009
.
Behavior of the Mercury Removal Absorbents at Egyptian Gas Plant
.
Paper presented at the
CIPC/SPE Gas Technology Symposium 2008 Joint Conference
,
Calgary, Alberta, Canada
, 16–19 June. SPE-114521-MS. 10.2118/114521-MS.
Abu El Ala
,
M.
,
Mahgoub
,
I. S.
,
Nabawi
,
M. H
. et al. 
.
2008
.
Mercury Monitoring and Removal at Gas-Processing Facilities: Case Study of Salam Gas Plant
.
SPE Proj Fac & Const
3
(
1
):
1
9
. SPE-106900-PA. 10.2118/106900-PA.
Andersson
,
A
.
1970
.
On the Geochemical Behavior of Mercury
.
Grundförböttring
23
:
31
40
.
Beuge
,
P
.
1982
.
Paragenetische Beziehungen Zwischen Quecksilber-Mineralen Und Organischen Substanzen
,
Vol
.
C374
,
53
62
.
Freiberg
:
Forschungsh
.
Black
,
A.
and
Saunders
,
G
.
2018
.
Offshore Gas Field Development – the Ripple Effect
.
APPEA J
58
(
2
):
695
. 10.1071/AJ17157.
Audeh
,
C. A
.
1988
.
Process for Absorbing Mercury from Natural Gas
.
US Patent No. 4,717,399
.
Carnell
,
P. J. H.
,
Joslin
,
K. W.
, and
Woodham
,
P
.
1995
.
Fixed-Bed Processes Provide Flexibility for COS and H2S Removal
.
Paper presented at the
74th Annual GPA Conference
,
San Antonio, Texas, USA
.
Carnell
,
P. J. H.
,
Row
,
V. A.
, and
McKenna
,
R
.
2007
. Minimizing Mercury Emissions from Gas Processing and LNG Plants.
In
Hydrocarbon World
.
London
:
Touch Briefings
.
Carnell
,
P. J. H.
,
Row
,
V. A.
, and
McKenna
,
R. A
.
2008
.
A Re-Think of the Mercury Removal Problem for LNG Plants
,
1
8
.
London, United Kingdom
:
Johnson Matthey Catalysts
.
Clever
,
H. L.
and
Iwamoto
,
M
.
1987
.
Solubility of Mercury in Normal Alkanes
.
Ind Eng Chem Res
26
(
2
):
336
337
. 10.1021/ie00062a026.
Boschee
,
P
.
2013
.
Advancements in the Removal of Mercury from Crude Oil
.
Oil and Gas Fac
2
(
2
):
12
17
. 10.2118/0413-0012-OGF.
Bourdet
,
J. F.
,
Stalker
,
L.
,
Hortle
,
A. L
. et al. 
.
2020
.
Tracking Mercury Contaminant in the Subsurface
.
Paper presented at the
Offshore Technology Conference Asia
,
Kuala Lumpur, Malaysia
, 2–6 November. OTC-30395-MS. 10.4043/30395-MS.
Bourdet
,
J.
,
Heath
,
C.
,
Josh
,
M
. et al. 
.
2021
.
Detecting Mercury in Rocks and Fluids
.
Paper presented at the
Petroleum Exploration Society of Australia (PESA) WA Technical October Lunch Talk
.
Bryndzia
,
L. T.
and
Inan Villegas
,
E
.
2020
.
Predicting the Concentration of Hydrogen Sulfide in Hydrocarbon-Bearing Clastic Reservoirs: Introducing the Iron-Chlorite-Pyrite-Hydrogen Sulfide Model
.
SPE J
25
(
6
):
3186
3199
. SPE-201190-PA. 10.2118/201190-PA.
Cooper
,
R.
,
Lopez-Linares
,
F.
,
Wang
,
W.
et al. 
.
2017
.
Improved Early Detection of Mercury for Oil and Gas Developments
.
Paper presented at the
ICMGP2017. 13th International Conference on Mercury as a Global Pollutant. FO-033
,
Providence, Rhode Island
, 16–21 July.
Diehl
,
S. F.
,
Goldhaber
,
M. B.
, and
Hatch
,
J. R
.
2004
.
Modes of Occurrence of Mercury and Other Trace Elements in Coals from the Warrior Field, Black Warrior Basin, Northwestern Alabama
.
Int J Coal Geol
59
(
3–4
):
193
208
. 10.1016/j.coal.2004.02.003.
Dvornikov
,
A. G.
and
Kirikilitsa
,
S. I
.
1987
.
Mercury in Donetsk Basin Coals
,
155
.
Moscow
:
Nedra [Entrails Publication House]
.
Dybdahl
,
B.
,
Lie
,
S.
,
Langford
,
M
. et al. 
.
2020
.
Bridging the Gap Between Reservoir and Sample; Reducing Asset Development Risk by Using Down-Hole Mercury Trapping and Non-Reactive Sampler for Trace Component Sampling
.
Paper presented at the
SPE Asia Pacific Oil & Gas Conference and Exhibition
,
Virtual
, 17–19 November. SPE-202446-MS. 10.2118/202446-MS.
Enrico
,
M.
,
Mere
,
A.
,
Zhou
,
H
. et al. 
.
2020
.
Experimental Tests of Natural Gas Samplers Prior to Mercury Concentration Analysis
.
Energy Fuels
34
(
5
):
5205
5212
. 10.1021/acs.energyfuels.9b03540.
Gabby
,
K. L.
and
Eisele
,
T. C
.
2013
.
Selective Removal of Mercury Using Zinc Sulfide
.
Min Metall Explor
30
(
2
):
91
94
. 10.1007/BF03402410.
Ezzeldin
,
M. F.
,
Gajdosechova
,
Z.
,
Masod
,
M. B.
et al. 
.
2016
.
Mercury Speciation and Distribution in an Egyptian Natural Gas Processing Plant
.
Energy Fuels
30
(
12
):
10236
10243
. 10.1021/acs.energyfuels.6b02035.
Fein
,
J. B.
and
Williams-Jones
,
A. E
.
1997
.
The Role of Mercury-Organic Interactions in the Hydrothermal Transport of Mercury
.
Econ Geol
92
(
1
):
20
28
. 10.2113/gsecongeo.92.1.20.
Gallup
,
D.
and
Bloom
,
N. S
.
2010
.
(36e) On the Solubility of Mercury in Liquid Hydrocarbons
.
Paper presented at the
AIChE Spring Meeting and Global Congress on Process Safety
.
Hoffmann
,
N.
,
Jödicke
,
H.
, and
Gerling
,
P
.
2001
.
The Distribution of Pre-Westphalian Source Rocks in the North German Basin – Evidence from Magnetotelluric and Geochemical Data
.
Neth J Geosci
80
(
1
):
71
84
. 10.1017/S0016774600022174.
Johnson
,
J. W.
,
Oelkers
,
E. H.
, and
Helgeson
,
H. C
.
1992
.
SUPCRT92: A Software Package for Calculating the Standard Molal Thermodynamic Properties of Minerals, Gases, Aqueous Species, and Reactions from 1 to 5000 Bar and 0 to 1000°C
.
Comput and Geosci
18
(
7
):
899
947
. 10.1016/0098-3004(92)90029-Q.
Kawamoto
,
T.
and
Sato
,
K
.
2000
.
Geological Modelling of a Heterogeneous Volcanic Reservoir by the Petrological Method
.
Paper presented at the
SPE Asia Pacific Conference on Integrated Modelling for Asset Management
,
Yokohama, Japan
, 25–26 April. SPE-59407-MS. 10.2118/59407-MS.
Keeley
,
M. L.
,
Dungworth
,
G.
,
Floyd
,
C. S
. et al. 
.
1990
.
The Jurassic System in Northern Egypt: I. Regional Stratigraphy and Implications for Hydrocarbon Prospectivity
.
J Petroleum Geol
13
(
4
):
397
420
. 10.1111/j.1747-5457.1990.tb00856.x.
Grimes
,
W. D.
,
French
,
R. N.
,
Miglin
,
B. P
. et al. 
.
2014
. The Physical Chemistry Nature of Hydrogen Sulfide Gas as It Affects Sulfide Stress Crtack Propagation in Steel.
In
Presented at CORROSION 2014
.
San Antonio, Texas, USA
:
NACE International
.
Gaupp
,
R.
and
Okkerman
,
J. A
.
2011
.
Diagenesis and Reservoir Quality of Rotliegend Sandstones in the Northern Netherlands - A Review. The Permian Rotliegend of the Netherlands SEPM Special Publication No.98, Copyright © 2011 SEPM
,
193
226
.
N.p
.:
Society for Sedimentary Geology
.
Kolker
,
A.
,
Quick
,
J. C.
,
Senior
,
C. L.
et al. 
.
2012
. Mercury and Halogens in Coal – Their Role in Determining Mercury Emissions from Coal Combustion.
In
Fact Sheet 2012–3122
.
Virginia, USA
:
U.S. Geological Survey
. 10.3133/fs20123122.
Koulocheris
,
V.
,
Louli
,
V.
,
Panteli
,
E.
et al. 
.
2018
.
Modelling of Elemental Mercury Solubility in Natural Gas Components
.
Fuel
233
(
2018
):
558
564
. 10.1016/j.fuel.2018.06.077.
Krooss
,
B. M.
,
Littke
,
R.
,
Muller
,
B.
et al. 
.
1995
.
Generation of Nitrogen and Methane from Sedimentary Organic Matter: Implications on the Dynamics of Natural Gas Accumulations
.
Chem Geol
126
(
1995
):
291
318
. 10.1016/0009-2541(95)00124-7.
Krupp
,
R
.
1988
.
Physicochemical Aspects of Mercury Metallogenesis
.
Chem Geol
69
(
1988
):
345
356
. 10.1016/0009-2541(88)90045-9.
Li
,
J.
,
Han
,
Z.
,
Yan
,
Q
. et al. 
.
2019
.
Distribution and Genesis of Mercury in Natural Gas of Large Coal Derived Gas Fields in China
.
Pet Explor Dev
46
(
3
):
463
470
. 10.1016/S1876-3804(19)60027-3.
Lokhorst
,
A.
,
ed
.
1997
.
NW European Gas Atlas: Haarlem
.
Haarlem, The Netherlands
:
NITG-TNO
.
Marsh
,
K. N.
,
Bevan
,
J. W.
,
Holste
,
J. C.
et al. 
.
2016
.
Solubility of Mercury in Liquid Hydrocarbons and Hydrocarbon Mixtures
.
J Chem Eng Data
61
(
8
):
2805
2817
. 10.1021/acs.jced.6b00173.
Miedaner
,
M. M.
,
Migdisov
,
A. S.
, and
Williams-Jones
,
A. E
.
2005
.
Solubility of Metallic Mercury in Octane, Dodecane and Toluene at Temperatures between 100 and 200°C
.
Geochim Cosmochim Acta
2005
(
69
):
5511
5516
. 10.1016/j.gca.2005.06.029.
Moiseyev
,
A. N
.
1971
.
A Non-Magmatic Source for Mercury Ore Deposits?
Econ Geol
66
:
591
601
. 10.2113/gsecongeo.66.4.591.
Sainal
,
M. R.
,
Shafawi
,
A.
, and
Mohamed
,
I. A. J
.
2007
.
Mercury Removal System for Upstream Application: Experience in Treating Mercury From Raw Condensate
.
Paper presented at the
E&P Environmental and Safety Conference
,
Galveston, Texas, USA
, 5–7 March. SPE-106610-MS. 10.2118/106610-MS.
Nengkoda
,
A.
and
Al-Hinai
,
Z. M
.
2009
.
Understanding of Mercury Corrosion Attack on Stainless Steel Material at Gas Wells: Case Study
.
Paper presented at the
International Petroleum Technology Conference
,
Doha, Qatar
, 7–9 December. IPTC-13368-MS. 10.2523/IPTC-13368-MS.
Ouddai
,
R.
,
Hassane
,
C.
, and
Assia Boughaba
,
A
.
2012
.
The Skikda LNG Accident: Losses, Lessons Learned and Safety Climate Assessment
.
IJGEI
35
(
6
):
518
533
. 10.1504/IJGEI.2012.051691.
Peabody
,
C. E.
and
Einaudi
,
M. T
.
1992
.
Origin of Petroleum and Mercury in the Culver-Baer Cinnabar Deposit, Mayacmas District, California
.
Econ Geol
87
(
4
):
1078
1103
. 10.2113/gsecongeo.87.4.1078.
Potter
,
R.
and
Barnes
,
H. L
.
1978
.
Phase Relations in the Binary Hg-S
.
Am Mineral
63
:
1143
1152
.
Ren
,
Z.
,
Zhang
,
S.
,
Gao
,
S.
et al. 
.
2007
.
Tectonic Thermal History and Its Significance on the Formation of Oil and Gas Accumulation and Mineral Deposit in Ordos Basin
.
Sci China Ser D-Earth Sci
50
(
II
):
27
38
. 10.1007/s11430-007-6022-1.
Sales
,
A. O.
,
Jacobsen
,
E. C.
,
Morado
,
A. A.
et al. 
.
1997
.
The Petroleum Potential of Deep-Water NorthwestPalawan Block GSEC 66
.
J Asian Earth Sci
15
(
2–3
):
217
240
. 10.1016/S0743-9547(97)00009-3.
Santos
,
S
.
2010
.
Challenges in Understanding the Fate of Mercury during Oxyfuel Combustion
.
Paper presented at the
MEC7Workshop DLCS
,
Strathclyde University
, 18 June. www.ieaghg.org.
Shafik
,
S. A.
and
Saudi
,
A. T
.
2018
.
Explanation of Low Reservoir Quality Wells by Re-Understanding the Environmental Deposition of Lower Safa “A” Unit within Khattatba Formation in JG Field, Abu Gharadig Basin, North Western Desert, Egypt, Case Study
.
Paper presented at the
Offshore Technology Conference Asia
,
Kuala Lumpur, Malaysia
, 20–23 March. OTC-28575-MS. 10.4043/28575-MS.
Stetson
,
S. J.
,
Gray
,
J. E.
,
Wanty
,
R. B.
et al. 
.
2009
.
Isotopic Variability of Mercury in Ore, Mine-Waste Calcine, and Leachates of Mine-Waste Calcine from Areas Mined for Mercury
.
Environ Sci Technol
43
(
19
):
7331
7336
. 10.1021/es9006993.
Thibodeau
,
A. M.
and
Bergquist
,
B. A
.
2017
.
Do Mercury Isotopes Record the Signature of Massive Volcanism in Marine Sedimentary Records?
Geology
45
(
1
):
95
96
. 10.1130/focus012017.1.
Whittenbury
,
H
.
2016
.
Eliminating Elemental Mercury. LNG Industry
. www.lngindustry.com.
Yates
,
R. G.
and
Thompson
,
G. A
.
1959
.
Geology and Quicksilver Deposits of the Terlingua District, Texas
.
Geol Surv Prof Pap
:
114
. 312.
Yamada
,
J.
,
Kobayashi
,
K.
,
Shibuya
,
T
. et al. 
.
2017
.
Occurrence of Particulate Mercury in Flowback Fluids After Well Workover
.
Paper presented at the
SPE/IATMI Asia Pacific Oil & Gas Conference and Exhibition
,
Jakarta, Indonesia
, 17–19 October. SPE-186220-MS. 10.2118/186220-MS.
Yudovich
,
Y. E.
and
Ketris
,
M. P
.
2005
.
Mercury in Coal: A Review Part 1 Geochemistry
.
Int J Coal Geol
62
:
107
134
. 10.1016/j.coal.2004.11.002.
Zettlitzer
,
M.
,
Scholer
,
H. F.
,
Eiden
,
R
. et al. 
.
1997
.
Determination of Elemental, Inorganic and Organic Mercury in North German Gas Condensates and Formation Brines
.
Paper presented at the
International Symposium on Oilfield Chemistry
,
Houston, Texas, USA
, 18–21 February. SPE-37260-MS. 10.2118/37260-MS.
Manceau
,
A.
,
Merkulova
,
M.
,
Murdzek
,
M.
et al. 
.
2018
.
Chemical Forms of Mercury in Pyrite: Implications for Predicting Mercury Releases in Acid Mine Drainage Settings
.
Environmental Science & Technology
52
(
18
):
10286
10296
. 10.1021/acs.est.8b02027.
Peabody
,
C.
and
Einaudi
,
M. T
.
1992
.
Origin of Petroleum and Mercury in the Culver-Baer Cinnabar Deposit Mayacmas District
.
California: Economic Geology
:
1078
1103
.

Supplementary data