Impact of Eliminating Mercury Removal Pretreatment on the Performance of a High-Level Radioactive Waste Melter Offgas System

DWPF melter offgas deposits and OGCT liquid

Deposits of solids on melter offgas system piping have occurred periodically throughout DWPF operation. Deposits have typically been found at the quencher inlet, the SAS inlet, in the vapor lines to the SAS, in the offgas condenser condensate line, and on the HEME filter. After about 5 years of radioactive operation, the steam to the SAS was stopped to reduce the amount of new water returned to the waste tank farm. Beginning at this time, plugging of the HEME filter increased significantly, resulting in biannual replacement; previously, the HEME filter had never been changed. Several years later, the steam to one SAS was restarted due to even worse plugging.

Melter offgas system solid deposit samples from the quencher inlet and SAS inlet were taken and analyzed (Bibler, 2003; Zeigler and Bibler, 2007). The locations of these deposits are shown in Fig. 3. An OGCT sample was also analyzed (Fellinger, 2004). These deposits become significant enough to require periodic removal. The average concentration of mercury, chloride, and the most abundant elements in these samples are summarized in Table 1. Only soluble Cl was measured in the samples; the total chlorine was estimated from the measured soluble amount and the amount that would be present in Hg₂Cl₂ with the measured Hg values. The melter feed values are average measured values except that (1) Cl was calculated from the sludge feed and (2) Hg was calculated from the typical Hg removal of 0.1 wt% in the SRAT. The concentrations are also shown on a basis normalized against Fe, which is the most abundant sludge species.

OPEN IN VIEWER

FIG. 3.

Locations of quencher and steam atomized scrubber deposits.

The Hg in the deposits was assumed to be mostly Hg₂Cl₂. Soluble mercury was measured only for the SAS deposits. X-ray diffraction indicated that the SAS inlet samples contained significant amounts of crystalline Hg₂Cl₂. Qualitatively, the SAS deposits were grayer than the quencher deposits, consistent with the presence of calomel, which disproportionates slightly to Hg⁰ (grayish) and HgCl₂. The major chloride species in both deposit samples was Hg₂Cl₂. Fe, Si, and Na are expected to be the predominant species in the offgas due to particle entrainment. Species known to be most volatile are B, Cl, F, S, and Hg. The most volatile species either form vapors or what is called semivolatile species. Semivolatile species are usually chloride, sulfate, or borate salts that have lower melting points than glass. The data show that Hg, Cl, F, and S are relatively greater in the SAS deposits, which is consistent with their higher volatility.

The soluble Hg contents of the SAS deposits and OGCT liquid were about 6%–10% and 5%, respectively, of the total mercury. This soluble Hg is expected to be HgCl₂, because HgCl₂ is the only likely mercury compound to have significant solubility in water (∼0.26 mol/kg water at 25°C) (Eddy and Menzies, 1940; Clever et al., 1985). Hg₂Cl₂ has an extremely low solubility (K_sp = 1.42 × 10⁻¹⁸ mol³ kg⁻³ at 25°C) (Hepler and Olofsson, 1975); it is actually more likely to disproportionate into Hg⁰ and HgCl₂ (Pokrovsky, 1995): \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Hg}_2 {\rm Cl}_2 ({\rm s}) = {\rm Hg^0} ({\rm l}) + {\rm HgCl}_2 ({\rm aq}) \qquad K = 2.35 \times 10^{-7} \ {\rm mol \ kg}^{- 1}\end{align*}\end{document}

The SAS deposits contained up to 10 times the amount of mercury as found in the quencher deposits. The temperature at the quencher inlet is about 250°C–300°C, whereas the SAS inlet temperature is about 45°C–60°C. The quencher deposits contain relatively more entrained components than at the SAS because they are scrubbed out in the quencher. The pilot-scale work previously performed by SRNL had mercury feed concentrations too low to give any useful data on mercury speciation.

ESCM at Pacific Northwest Laboratory

The ESCM melter at Pacific Northwest Laboratory was used to study mercury speciation in the offgas system (Goles et al., 1990). The offgas mercury data collected during these tests are the most complete found for glass melter operation. The melter system was similar to DWPF and consisted of a slurry-fed joule-heated melter with a venturi scrubber (quencher), condensate or quench tank (OGCT), chilled-water condenser, and high-efficiency particulate air filter. The feed composition tested was similar to DWPF and the feed was treated with formic acid. Offgas system tests were conducted at several melter plenum temperatures and several air-inleakage rates.

Chloride was present in the ESCM feed at 12 times the mercury concentration. The Cl/Hg molar ratio in the SRNL pilot system was 67–436, whereas this ratio in DWPF has ranged from 0.55 to 2.2 in the past and is currently about 0.25. The proposed DWPF operation with higher mercury gives a ratio of 0.034. Therefore, the DWPF has operated with significantly less Cl than was tested in pilot-scale operations.

The ESCM melter offgas samples were filtered, and passed through a condenser and then through three water-filled impingers. No mercury compounds were found on the filters, indicating that no solid Hg₂Cl₂ had been formed. The mercury in the impingers, and also in the condensate tank liquid, was speciated qualitatively by assuming that mercury in the solids was Hg₂Cl₂, while soluble mercury was HgCl₂. The identity of the solid mercury species was verified by X-ray diffraction. The splits of Hg species between Hg₂Cl₂ and HgCl₂ for five tests are shown in Table 2. The DWPF deposit sample ratios are also shown. No evidence of Hg⁰ was found in the offgas or condensate.

For all the data at about 750°C plenum temperature, Hg₂Cl₂ accounts for about 85%–95% of the total mercury species found except for Test 2, which had anomalous results for the impinger samples that did not agree with the OGCT samples. In Test 3 at 550°C plenum temperature, both the impinger and OGCT samples had significantly more of the more oxidized HgCl₂ present. Higher plenum temperature appears to favor formation of the less oxidized mercury compound Hg₂Cl₂. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Hg}^0 + {\rm Cl} + {\rm M} \rightarrow {\rm HgCl} + {\rm M} \qquad k_1 = 6.0 \times 10^{-11} {\rm cm}^3 \ {\rm molecule}^{-1} \ {\rm s}^{-1} \tag {1}\end{align*}\end{document}

The ESCM tests showed that the concentration of elemental mercury in the vapor phase was always approximately equal to the saturation vapor pressure at the prevailing temperature, even though no elemental Hg liquid was ever found in the offgas system. The gray color of the calomel and other solids samples was probably due to surface Hg⁰ from disproportionation. This surface Hg⁰ appears to exert the full Hg⁰ vapor pressure. In fact, even when the melter was not being fed, the vapor above the condensate remained saturated with Hg⁰.

Gas-phase oxidation literature

Numerous articles have been written on the gas-phase oxidation of elemental mercury, primarily in coal-fired power plant emissions. In power plants, the mercury and chlorine concentrations in the flue gases are significantly lower than in the melter offgas systems of interest. Power plant mercury emission levels are typically in the range 0.1–4 parts per billion by volume (ppbv) (1–30 μg/m³) (Laudal et al., 1997) and waste incinerator levels may be up to about 400 ppbv (Sliger et al., 2000). Mendelsohn and Livengood (2006) have reviewed a substantial portion of the literature on mercury chemistry in flue gas. Most articles divide the Hg emissions into insoluble (Hg⁰) and soluble (HgCl₂), but generally ignore the insoluble Hg₂Cl₂. The goal for mercury oxidation in power plant flue gases has been oxidation to the soluble HgCl₂ that can be scrubbed from the gas. This goal is contrary to that for DWPF where collection of mercury as the insoluble Hg⁰ metal might be preferred.

Homogeneous oxidation

The homogeneous reaction schemes are all similar, but none have been found to be universally applicable to all combustors. For these mechanisms, chlorine atoms are assumed to be in excess compared to mercury. The concentration of SO₂ present is usually higher than both Cl and Hg. The primary reaction step in the oxidation is the formation of HgCl from Hg atoms and Cl atoms in the presence of a stabilizing collision partner M in Reaction (1) (Horne et al., 1968; Sliger et al., 2000; Xu et al., 2003; Taylor et al., 2005):

The following subsequent reactions are slower. Reactions (2)–(5) are independent of temperature. The rate constant for Reaction (6) is at 125°C. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HgCl} + {\rm Cl} + {\rm M} \rightarrow {\rm HgCl}_2 + {\rm M} \qquad k_2 & = 4.0 \times 10^{-12} \ {\rm cm}^3 \ {\rm molecule}^{-1} \ {\rm s}^{-1} & \tag {2}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HgCl} + {\rm Cl}_2 \rightarrow {\rm HgCl}_2 + {\rm Cl} \qquad k_3 & = 1.2 \times 10^{-11} \ {\rm cm}^3 \ {\rm molecule}^{-1} \ {\rm s}^{-1} \tag {3}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Cl} + {\rm Cl} + {\rm M} \rightarrow {\rm Cl}_2 + {\rm M} \qquad k_4 & = 5.5 \times 10^{-33} \ {\rm cm}^3 \ {\rm molecule}^{-1} \ {\rm s}^{-1} \tag {4}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HgCl} + {\rm HgCl} \rightarrow {\rm Hg}_2 {\rm Cl}_2 \qquad k_6 & = 4.2 \times 10^{-10} \ {\rm cm}^3 \ {\rm molecule}^{-1} \ {\rm s}^{-1} \tag {5}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HgCl} + {\rm HCl} \rightarrow {\rm HgCl}_2 + {\rm H} \qquad k_5 = 6.3 \times 10^{-13} \ {\rm cm}^3 \ {\rm molecule}^{-1} \ {\rm s}^{-1} \tag {6}\end{align*}\end{document}

Reaction (3) is slow due to the low concentration of Cl₂ present under conditions where there is significant Cl present. The reactions of Hg⁰ with Cl₂, HCl, O₂, and O are significantly slower.

Sliger et al. (2000) found that the homogeneous oxidation was primarily governed by the formation of Cl from HCl, the rate at which the hot offgas was quenched (cooled), and by the presence of background gases involved in competing reactions. These authors also mention the fly ash or carbon-mediated reaction where HCl forms Cl₂ that can then react with Hg at temperatures lower than 300°C. The importance of Cl formation from Cl₂ has been debated.

Senior et al. (2000) suggest that a pathway through HgO may be important in systems with low chlorine concentrations, but Sliger et al. (2000) found that no HgO was formed in the absence of HCl in their test gas. Hall (1991) has suggested a mechanism with HCl and O₂ as important reactants. Edwards et al. (2001) have included oxidation by O₂ in modeling although it apparently has little effect on the overall oxidation. Models by Niksa et al. (2001) and Edwards et al. (2001) did not predict well the oxidation at <700°C. Sliger et al. (2000) predicted that the oxidation was limited to the range 400°C–700°C because there were too few Cl radicals at low temperature, while Hg⁰ was favored at high temperature. They also noted that virtually all of the oxidation reaction occurred inside their sample probe where cooling at 5,400 K/s from about 900°C occurred. They developed models that showed the extent of Hg oxidation and the equilibrium amount of oxidation during the quench. Complete oxidation was predicted in 60 ms. The kinetically controlled oxidation extent was only around 40% and was limited by the availability of Cl radicals. Fry et al. (2007) have reported that oxidation increased at higher quench rates. The species NO, NO₂, and SO₂ have been shown to have varying effects on the mercury oxidation rate (Hall et al., 1990, 1991, 1995; Galbreath et al., 2000; Niksa et al., 2001; Norton et al., 2003). NO and SO₂ have generally been found to inhibit oxidation of Hg⁰, but not in all cases.

Heterogeneous oxidation

Schofield (2004, 2005a, 2005b) has hypothesized that the main mechanism of mercury oxidation in power plant emissions occurs heterogeneously on the fly ash present and on equipment surfaces. Numerous researchers have reported the difficulty in measuring the rate of homogeneous oxidation due reactions on the surfaces of their test equipment (Lin and Pehkonen, 1999; Sliger et al., 2000; Ariya et al., 2002; Yan et al., 2005). Laudal et al. (2000) have reported that fly ash can have a large effect on speciation. Senior et al. (1996) suggest that Cl₂ formed from HCl on the fly ash surface can oxidize Hg. Edwards et al. (2001) state that heterogeneous reactions are probably dominant at lower temperature—their model drastically under-predicted mercury oxidation below 630°C.

Schofield (2004, 2005a, 2005b) proposed a mechanism that involves the interaction of mercury with surfaces. The surface intermediates HgO, HgSO₄, and HgSO₄•HgO are formed that then react with HCl to form HgCl₂. This mechanism does not require Cl₂. The presence of surface reactions in the DWPF melter system is not known. The residence time before the quench is about 0.5 s, and the amount of particulate is significantly less than the amount of fly ash in a power plant.

Oxidation in aqueous solutions

Most of the literature on aqueous oxidation concerns environmental systems such as lakes and rivers. The oxidation of elemental mercury by dissolved oxygen in the presence of chloride anions has been examined by several researchers. Other literature has addressed the unexpected oxidation of Hg⁰ in the Ontario Hydro sampling train, which uses a chloride solution to scrub oxidized mercury (Hg²⁺) (ASTM, 2008). The oxidation in this sample train will be briefly discussed as it pertains to the oxidation of mercury in a melter offgas system. Elemental mercury in oxidizing acids such as HNO₃ will be oxidized and dissolve to some extent to form Hg²⁺. Oxidation by hypochlorite and chlorite has also been reported (Zhao and Rochelle, 1999; Hutson et al., 2008).

Oxidation in dilute aqueous solutions

Magalhães and Tubino (1995) studied oxidation of metallic Hg⁰ in aqueous solutions containing NaCl. These tests were performed in agitated vessels open to the atmosphere. The reaction was monitored by following the concentration of the product \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HgCl}_4^{2-} \ ({\rm Hg}^{2+}). \ {\rm HgCl}_4^{2-}\end{align*}\end{document} , the dichloride complex of HgCl₂, is the expected product with a large excess of Cl⁻. Eventually the concentration of \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HgCl}_4^{2-}\end{align*}\end{document} was found to decrease. At this time, the metallic Hg droplets lost their characteristic brightness and started to become white on the surface. Both the drop in \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HgCl}_4^{2-}\end{align*}\end{document} and the white color were attributed to the formation of calomel on the surface of the Hg⁰ droplets: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HgCl}_4^{2-} \ ({\rm aq}) + {\rm Hg}^0 (l) = {\rm Hg}_2 {\rm Cl}_2 ({\rm s}) + 2 \ {\rm Cl}^- \ ({\rm aq})\end{align*}\end{document}

The oxidation rate was found to increase approximately linearly with chloride concentration up to about 250 g/L Cl⁻, at which point the rate decreased. At high enough Cl⁻ concentrations, the equilibrium concentration of oxygen decreases, resulting in decreased reaction rate. The pH was found to have a significant effect on the oxidation rate, with significant rate increases at lower pH; the rate increase was more than expected just by the addition of more Cl⁻. Yamamoto (1996) found that addition of the chlorides KCl or MgCl₂ also increased the reaction rate.

Magalhães and Tubino (1995) proposed the following mechanism for the oxidation that can be combined into either of the following overall reactions in acid and base solution, respectively \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}& 2 {\rm Hg}^0(l) + {\rm O}_2 \ ({\rm aq}) + 4 {\rm H}^+ + 4 {\rm Cl}^- \ ({\rm aq}) \\ & \quad = 2 {\rm HgCl}_2 \ ({\rm aq}) + 2 {\rm H}_2 {\rm O} \tag {7}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}& 2 {\rm Hg}^0(l) + {\rm O}_2 \ ({\rm aq}) + 2 {\rm H}_2 {\rm O} + 4 {\rm Cl}^- \ ({\rm aq}) \\ & \quad = 2 {\rm HgCl}_2 \ ({\rm aq}) + 4 {\rm OH}^- \tag {8}\end{align*}\end{document}

The overall reaction shows that O₂ is the active oxidant. This mechanism is consistent with their observation that the pH increased during the course of the reaction (due to formation of OH⁻). The positive effect of lower pH and higher Cl⁻ is also consistent. These authors were unable to determine if the oxidation reaction occurred in solution or on the Hg⁰ metal surface, and stated that the oxidation reaction may actually occur at the surface of the Hg⁰ droplet and that their experiments could not distinguish between these two possibilities.

The oxidation of dissolved and liquid elemental mercury in oxygenated water with chloride present was studied by Amyot et al. (2005). The reactions were performed in the dark to eliminate photochemical reactions. The chloride present was added as KCl; no chlorine with oxidizing potential such as Cl₂ was added. Total Hg concentrations were below the solubility of Hg⁰ in oxygen-free water at 25°C (about 284 nM, or 0.057 mg/L) (Clever et al., 1985). Chloride concentrations were ranged from 0 to 500 μM (17.7 mg/L). In this work, oxidized Hg was considered to be the sum of Hg(I) and Hg(II) concentrations. Oxygen concentrations were maintained by contacting the stirred test solutions with air.

In solutions containing dissolved Hg⁰ only, with no liquid elemental Hg⁰ (droplets), the rate of Hg⁰ oxidation by O₂ in the presence of chloride was essentially zero. However, rapid oxidation of Hg⁰ occurred when Hg metal was present. As determined by Magalhães and Tubino (1995), the rate of oxidation increased with increased concentration of chloride. The rate of oxidized mercury creation was also found to be a function of the mercury droplet surface area. The oxidation rate eventually decreased due to the accumulation of oxidation products on the Hg⁰ metal surface, which agrees with Magalhães' observation of calomel. The absence of O₂ resulted in no oxidation as expected.

Oxidation in the Ontario hydro sampling train

Laudal et al. (1997, 2000) reported on tests of the Ontario Hydro sampling train (ASTM, 2008), which uses three impingers containing 10 wt% KCl solution to trap oxidized (Hg²⁺) mercury. They found that the presence of Cl₂ resulted in statistically significant amounts of Hg²⁺ even though only Hg⁰ was present in the test gas. They also found that the presence of SO₂ in the test gas decreased the amount oxidized. Cauch et al. (2008) specifically tested the Ontario Hydro method to quantify the effect of Cl₂ on the measured Hg²⁺. The presence of SO₂ in the gas again eliminated the interference of Cl₂ on the Hg²⁺ measurement. Addition of sodium thiosulfate Na₂S₂O₃, a reducing agent, also eliminated the effect of Cl₂. Chlorine was postulated to be removed by these overall reactions: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm SO}_2 + {\rm Cl}_2 + 2 \ {\rm H}_2 {\rm O} = {\rm H}_2 {\rm SO}_4 + 2 {\rm HCl} \tag {9}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}2 \ {\rm Na}_2 {\rm S}_2 {\rm O}_3 + {\rm Cl}_2 = {\rm Na}_2 {\rm S}_2 {\rm O}_6 + 2 \ {\rm NaCl} \tag {10}\end{align*}\end{document}

Cauch et al. (2008) gave a mechanism for chorine oxidation of Hg⁰: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Cl}_2 ({\rm aq}) + {\rm H}_2 {\rm O} = {\rm HClO} ({\rm aq}) + {\rm H}^+ \ ({\rm aq}) + {\rm Cl}^- \ ({\rm aq}) \tag {11}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm HOCl} ({\rm aq}) = {\rm H}^+ \ ({\rm aq}) + {\rm OCl}^- \ ({\rm aq}) \tag {12}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Hg}^0 ({\rm aq}) + {\rm OCl}^- \ ({\rm aq}) = {\rm Hg}^{2+} \ ({\rm aq}) + {\rm Cl}^- \ ({\rm aq}) + 2 \ {\rm OH}^- \ ({\rm aq}) \tag {13}\end{align*}\end{document}

Zhao and Rochelle (1999) studied the absorption of element HgO from the vapor into a solution of hypochlorite and found that the active oxidizing species was more likely to be aqueous Cl₂ rather than hypochlorite ion (OCl⁻): \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Hg}^0 ({\rm aq}) + {\rm Cl}_2 ({\rm aq}) \rightarrow {\rm products} \tag{14}\end{align*}\end{document}

For the works of both Cauch et al. (2008) and of Zhao and Rochelle (1999), aqueous Cl₂ is the more likely oxidant because low pH (high H⁺) and high Cl⁻ both increased the rate of oxidation of Hg⁰.

Background

Mercury fed to DWPF melter as either Hg⁰ or HgO would be volatilized as elemental mercury vapor (Hg⁰) during the calcination/fusion process. The mercury vapor is then presumed to undergo oxidation reactions to either +1 or +2 oxidation state in the melter vapor space and downstream of the melter including the condensate tank. The only literature data found for mercury emission from a glass melter were taken during the ESCM tests at PNL (Goles et al., 1990), as shown in Table 2.

Homogeneous oxidation in the gas phase

The mercury oxidation model was developed based on the most recent kinetic data available in the literature as described above. The model was then calibrated using the ESCM data. Specifically, only those kinetic data taken under the postflame conditions of coal-fired power plants were used in this study mainly because the temperature range of these data is inclusive of those typically encountered in the melter vapor space and the offgas header leading to the quencher.

Modeling approach

With respect to the mercury speciation reactions, the ESCM and DWPF systems can be divided into three distinct reaction zones. The first (Zone 1) is the cold cap where the water portion of the slurry feed is converted into steam, and the remaining dry feed components are converted into glass and calcine gases. In this study, high-temperature thermodynamic equilibrium software called FactSage™ v6.0 was used to calculate the composition of the volatile species (Thermfact/CRCT, 2009).

The calcine gases enter the second reaction zone (Zone 2) along with the elemental mercury vapor and volatile salts such as chlorides and borates of alkali metals generated in Zone 1. Zone 2 includes the vapor space of the melter and the offgas header where species volatilized from the melter mix with steam and air and further react. Reaction Zone 3 resides inside the quencher and the condensate tank (and SAS for DWPF), where steam and volatile salts are condensed and may further react in the liquid phase. It is necessary to model all three reaction zones to have a relatively complete description of how mercury would speciate throughout the ESCM or DWPF melter offgas system. Preliminary modeling of all three zones has been performed.

Characteristics of offgas carryover

The carryover of materials into the offgas can occur via two very different mechanisms: physical entrainment and vapor phase transport (or volatilization). Both feed and glassy materials can become airborne by physical entrainment aided in part by the pulling of the exhaust blower and remain as solids throughout the offgas system. Entrained particulates with mean particle size >1 μm account for much of the particle loading in the melter exhaust, but over 90% of them are routinely scrubbed in the quencher (Colven, 1983).

Conversely, alkali salts of chloride and borate, and elemental mercury are transported into the offgas due to their volatility at the melt temperature but later condense as offgas gets cooled and further quenched. Upon condensation, these semivolatile salts, and any mercury salts formed, would become primarily submicron-sized aerosols that are difficult to remove using an ejector-venturi scrubber like the DWPF quencher. As a result, the majority of the semivolatile salts and mercury will remain in the gas stream downstream of the quencher. The peak of the particle size distribution curve shifts from >1 μm to <1 μm after quenching (Sabatino et al., 1983). Formation of molecular HgCl supports the observation that more Hg₂Cl₂ is collected in the downstream sections of the offgas system. Hg₂Cl₂ formed from the vapor would be submicron size and pass through the quencher, whereas formation of Hg₂Cl₂ in the aqueous phase is not likely to generate these submicron particles.

Model assumptions

The following simplifying assumptions were made to model the oxidation of mercury in the melter offgas system:

The composition of calcine gases produced during the melting/fusion process is at equilibrium with those of the condensed phases at 1,150°C (Zone 1).

Homogeneous gas-phase oxidation model

The elementary reactions shown in Reactions (1)–(4) were used to describe the gas-phase chlorination of mercury in Zone 2. The oxidation of mercury is initiated by the Cl atoms combining with the elemental mercury vapor via Reaction (1) to form a mercurous chloride atom (+1 oxidation state). The Cl atoms can further oxidize the mercurous chloride to mercuric chloride (+2 oxidation state) via Reaction (2). Reaction (3) was added to account for the effect of temperature on the speciation of chlorine; formation of Cl₂ is favored over Cl atoms at low temperatures, whereas formation of Cl atoms is favored at high temperatures. The formation of Cl₂ by the recombination of Cl atoms by Reaction (4), which would slow down Reactions (1) and (2) and accelerate Reaction (3), was excluded from the model, since its rate constant is many orders of magnitude smaller than the other reactions and the concentration of Cl atoms not be high enough to overcome such a large deficit in the rate constant.

The second-order rate constants used in the model were taken from a recent study using the laser photolysis/laser induced fluorescence technique (Taylor et al., 2005). It is noted that the upper temperature bounds for the rate constants of Reactions (1)–(3) are 50°C and 100°C lower than the estimated gas temperatures in the ESCM vapor space (Test 3) and the DWPF melter vapor space, respectively. As stated earlier, the temperature ranges are low because they were specifically derived under the postflame conditions of coal-fired power plants. It is implicitly assumed here that these rate constants can be extrapolated to the temperature regions of interest to this study.

The concentrations of [Hg], [HgCl], [HgCl₂], [Cl], and [Cl₂] were found as a function of time by solving the following five rate equations simultaneously: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}\frac {{\rm d} [{\rm Hg}]} {{\rm dt}} = - {\rm k}_1 [{\rm Hg}] [{\rm Cl}] \tag {15}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}\frac {{\rm d} [{\rm Cl}]} {{\rm dt}} = - {\rm k}_1 [{\rm Hg}] [{\rm Cl}] - {\rm k}_2 [{\rm HgCl}] [{\rm Cl}] + {\rm k}_3 [{\rm HgCl}] [{\rm Cl}_2] \tag {16}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}\frac {{\rm d} [{\rm HgCl}]} {{\rm dt}} = - {\rm k}_1 [{\rm Hg}] [{\rm Cl}] - {\rm k}_2 [{\rm HgCl}] [{\rm Cl}] - {\rm k}_3 [{\rm HgCl}] [{\rm Cl}_2] \tag {17}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}\frac {{\rm d} [{\rm HgCl}_2]} {{\rm dt}} = {\rm k}_2 [{\rm HgCl}] [{\rm Cl}] + {\rm k}_3 [{\rm HgCl}] [{\rm Cl}_2] \tag {18}\end{align*}\end{document} \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}\frac {{\rm d} [{\rm Cl}_2]} {{\rm dt}} = - {\rm k}_3 [{\rm HgCl}] [{\rm Cl}_2] \tag {19}\end{align*}\end{document}

All of the mercury fed is volatilized, so the initial concentration of mercury, [Hg]_o, is known, and both [HgCl]_o and [HgCl₂]_o are zero. Therefore, to solve Equations (15)–(19), the initial concentrations of [Cl] and [Cl₂] must be known, and the strategy used to find [Cl]_o and [Cl₂]_o is as follows:

Run the FactSage model to calculate the equilibrium melter exhaust composition at the measured melter vapor space temperature.

The five ordinary differential equations of the Zone 2 kinetic model were solved simultaneously using the RK4 v3.0 program (Co, 2005). The reasoning behind taking a percent approach to the equilibrium Cl₂/Cl ratio in step 5 is that the equilibrium prediction of Cl₂ concentration will be much higher than the actual value based in part on the kinetics of its formation, as evidenced by the negligibly small rate constant for Reaction (4). So, the initial Cl₂/Cl ratio is reduced to a certain percentage of what is predicted by the equilibrium model. This procedure was used to model the ESCM Test 1 and Test 3 data.

To model the DWPF offgas system, the same steps were followed, except:

For steps 1–3, DWPF conditions at 650°C vapor space temperature were used.

The basis for using the average ESCM values is the fact that the DWPF melter vapor space temperature of 650°C falls exactly in the middle of the ESCM test temperatures.

ESCM Zone 1 model

The input compositions for the FactSage equilibrium model are given in Table 3. The two main differences between Tests 1 and 3 were that (1) the melter was fed 2.8 times faster during Test 1 than Test 3 and (2) the melter vapor space was kept 200°C cooler during Test 3 by turning off the vapor space heaters. The rate of melter air inleakage was the same at 25 kg/hr in both tests. The molar ratio of Cl/Hg was 12, which means that regarding the chlorination of mercury, chloride was present in excess.

The results of the FactSage model runs at 1,150°C are shown in Table 4. As expected, 100% of the mercury fed was predicted to volatilize as elemental mercury vapor. At 1,150°C under equilibrium conditions, 100% of the chloride fed was also predicted to volatilize as either HCl or alkali chlorides at a ratio of 40:60. A negligible quantity of Cl atoms was also predicted to form but not Cl₂. It should be noted that the mercury and alkali chlorides became part of the offgas carryovers due to their low vapor pressures at the melt temperature, and the model predictions do not include any solids that are physically entrained because such a prediction is beyond the scope of an equilibrium model.

ESCM Zone 2 model

The Zone 2 model consisted of two parts. The first was the FactSage model to calculate the equilibrium compositions of the melter exhausts at the measured melter vapor space temperatures of 750°C and 550°C for Tests 1 and 3, respectively. The second part was the gas-phase kinetics model of mercury chlorination, which further adjusted the equilibrium speciation of mercury. The input for the Zone 2 FactSage model included (1) the ideal-gas output shown in Table 5, (2) free H₂O that volatilizes from the cold cap, and (3) the air inleakage to the melter.

The results of the FactSage model runs are shown in Table 5 under the heading Equilibrium. As in the Zone 1 run at 1,150°C, the 60:40 split of total chloride atoms between HCl and alkali metals, respectively, was generally maintained in Test 1, while at 200°C lower in Test 3 equilibrium favors ∼10% more chloride atoms to couple with alkali metals. The summary of gas-phase results given in Table 5 also shows that equilibrium favors (1) all of the chloride atoms that do not couple with alkali metals to be in the form of HCl with little or no Cl/Cl₂, and (2) the formation of Cl₂ over Cl even at 750°C. This is why the adjustment steps for the fraction (Cl + Cl₂) and the approach to equilibrium were necessary to force some of the equilibrium-predicted HCl into Cl and Cl₂ for the mercury chlorination reactions to proceed.

After a few trial-and-error runs, it was found that when the %Cl _x fraction was set at 13.5% and the percent approach to the equilibrium Cl₂/Cl ratio set at 1%, the calculated HgCl/HgCl₂ ratio would match the experimental value of 9:1 at these initial concentrations, as shown in Fig. 4. These results are summarized in Table 6. The kinetics of mercury chlorination are so fast that the reactions are essentially complete in 0.5 ms. Therefore, considering the fact that the estimated gas residence time in the ESCM vapor space was on the order of 5 s for Test 1, it may be concluded that the chlorination of mercury will be complete at the instant Cl and Cl₂ are formed. However, this study does not address the question of how atomic Cl and Cl₂ are formed in the first place.

OPEN IN VIEWER

FIG. 4.

Concentration profiles of mercury chlorination reactants and products for ESCM test 1.

Similarly for Test 3, it was found that when the %Cl _x was set at 20% and the percent approach to the equilibrium Cl₂/Cl ratio set at 0.22%, the calculated HgCl/HgCl₂ ratio would match the experimental value of 1:1 at these initial concentrations. As expected, the resulting initial concentration of Cl for Test 3 was calculated to be lower than that for Test 1, whereas the resulting initial concentration of Cl₂ was higher than its counterpart for Test 1, since the temperature was lower during Test 3. The calculated percent total Cl atoms not tied to the alkali metals that exist as either Cl atom or Cl₂ increased from 13.5% to 20%, as the temperature was decreased from Test 1 to Test 3. This is because the formation of Cl₂ is favored over that of Cl atom at lower temperatures, and the net effect is to increase the number of total Cl atoms in both Cl and Cl₂. These results are summarized in Table 6. Note that for both Tests 1 and 3, the total chlorinated Hg was about 86%; the remaining Hg was predicted to be Hg⁰ and HgO.

Figure 5 shows that the calculated HgCl/HgCl₂ ratio matches the experimental value of 1:1 at these initial concentrations of Cl and Cl₂. Due to a lower temperature, it took about three times as long to complete the chlorination reactions as in Test 1. However, the reactions were still complete in 1.5 ms, which is four orders of magnitude shorter than the estimated gas residence time of 14 s for Test 3. This confirms the earlier conclusion that the chlorination of mercury will be complete at the instant Cl and Cl₂ are formed. The model also predicted that at the Cl/Hg ratio of 12, 85% of the mercury fed was chlorinated either to HgCl or HgCl₂ in both ESCM tests.

OPEN IN VIEWER

FIG. 5.

Concentration profiles of mercury chlorination reactants and products for ESCM test 3.

DWPF Zone 1 model

The composition of the DWPF melter feed with no mercury removal is shown in Table 7. The molar Cl-to-Hg ratio is only 0.4, compared to 12 for the ESCM feeds, which means that there is a significant deficit in chloride, so the overall conversion of elemental mercury into HgCl and/or HgCl₂ is expected to be very low.

The results of the FactSage model run at 1,150°C are shown in Table 8. As expected, 100% of the mercury fed as HgO was predicted to volatilize as elemental mercury vapor. At 1,150°C under equilibrium conditions, 100% of the chloride fed was also predicted to volatilize as either HCl or alkali chlorides at a ratio of 25:75, respectively. Notice that the percent total Cl atoms that couple with alkali metals was predicted to be much higher than for the ESCM feeds. Perhaps due to a significant shortage of chloride atoms, neither Cl atoms nor Cl₂ were predicted to form at any concentrations. As with the ESCM results, these model predictions do not include any physically entrained solids.

DWPF Zone 2 model

As with the ESCM case, the DWPF Zone 2 model consisted of two parts. The first was the FactSage model to calculate the equilibrium compositions of the melter exhausts at the measured melter vapor space temperatures of 650°C. The second part was the gas-phase kinetics model of mercury chlorination, which further adjusted the equilibrium speciation of mercury. The input for the Zone 2 FactSage model included

the ideal-gas output shown in Table 8,

The results of the Zone 2 FactSage model runs for DWPF are shown in Table 9 under the heading Equilibrium. One notable result is that the percent total Cl atoms predicted to exist as alkali chlorides decreased from 75% of the total chloride fed at 1,150°C to zero as the melter exhaust was cooled below 850°C. By contrast, the percent total Cl atoms predicted to exist as alkali chlorides were 35%–40% even at lower temperatures for the ESCM feeds. The summary of gas-phase results also shows that equilibrium favors (1) nearly 65% of the total chloride fed to oxidize the mercury, (2) much of the remaining 35% to be in the form of HCl, and (3) the formation of Cl₂ over Cl at 550°C actual gas temperature at a ratio of nearly 6:1.

Since the nominal DWPF melter vapor space temperature was right at the midpoint of the two ESCM temperatures, the percent total Cl atoms not tied to the alkali metals that exist either as Cl atom or Cl₂ was set at 16.75%, which is the average of the two corresponding ESCM values. Likewise, the percent approach to the equilibrium Cl₂/Cl ratio was set at 0.61%, which is again the average of the two corresponding ESCM values. At the calculated initial concentrations of Cl and Cl₂, the calculated HgCl/HgCl₂ ratio was 53, which means that >98% of the chlorinated mercury would have an oxidation state of +1. These values are summarized in Table 10. Also note that with so few Cl atoms available compared to Hg, the reactions are essentially complete in 0.03 ms, and only 6% of the mercury fed would be chlorinated. As shown in Table 9, the remaining 94% of the mercury fed would exist as either Hg⁰ (90%) or HgO (4%). The concentration profiles are shown in Fig. 6.

OPEN IN VIEWER

FIG. 6.

Concentration profiles of mercury chlorination reactants and products for DWPF sludge batch 5 feed with zero mercury removal.

The measured chloride level in a recent DWPF feed sample is actually an order of magnitude lower than that of the test case used in this study. As a result, the degree of shortage in chloride in relation to mercury will be even greater in future DWPF operations. This would make the predicted level of HgCl in the actual melter exhaust <6% of the total mercury fed, whereas that of elemental mercury is expected to be >90%. Therefore, much of the mercury fed will likely condense as the elemental mercury in the offgas system.

Equilibria

The aqueous equilibria of Hg⁰, Hg²⁺, and \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Hg}_2^{2+}\end{align*}\end{document} compounds at 25°C predicted using the OLI software were compared to the best values tabulated in the literature to validate the use of OLI software (Eddy and Menzies, 1940; Hepler and Olofsson, 1975; Marcus, 1980; Clever et al., 1985, 1987; Pokrovsky, 1995; OLI Systems, 2008). The equilibrium constants calculated using StreamAnalyzer, after accounting for species activity coefficients, matched most of the tabulated values (generally for infinite dilution) within the same order of magnitude. The typical discrepancy was a factor of about 2, but factors of up to about 6 were found for some reactions. This level of agreement is within the range of values reported in the literature by different authors.

The solubility and Henry's law constants as functions of temperature for Hg⁰ and HgCl₂ were generated using StreamAnalyzer and these values were compared to literature values. The solubility and predicted vapor pressure for Hg⁰ deviated from the literature by up to a factor of 2. The solubility at 25°C was predicted to be 1.3 × 10⁻⁷ mol/kg versus the accepted value of about 2.85 × 10⁻⁷ mol/kg (Clever et al., 1985).

The solubility of HgCl₂ is predicted accurately by OLI, but the vapor pressure is 4–7 times higher than that given in the literature. Fortunately, the concentration of HgCl₂ in the present work is expected to be low, so errors in the vapor pressure should not have a significant effect on the results.

Aqueous redox reactions

OLI software has the capability of handling some reduction–oxidation (redox) reactions in the aqueous phase. This redox capability is equilibrium based, so there is no consideration for reaction kinetics. In StreamAnalyzer, redox reactions can be turned on or off for individual elements. (In the discussion below, anions are always assumed to be aqueous [aq].)

Oxidizers present in the melter offgas are Cl₂, O₂, NO, and NO₂. The roles of NO and NO₂ in Hg chemistry are not clear. Sulfur dioxide SO₂ can oxidize Hg⁰, but can also reduce Hg²⁺. It has been shown that SO₂ can suppress the oxidation of Hg⁰ by Cl₂ by reducing Cl₂ to Cl⁻ (Cauch et al., 2008). Hydrochloric acid is not an oxidizer, but provides chloride to the equilibria with oxidized Hg (as does NaCl and other chlorides). In the reactions below, the oxidation states of the elements are shown above the reactions.

Reactions with SO₂

Generally, SO₂ acts as a reductant for most species of interest, but in some cases appears to increase the amount of Hg⁰ oxidized. Hutson et al. (2008) have shown that in scrubbing of Hg⁰ using NaClO₂ as an oxidant, the addition of some SO₂ to the vapor enhances scrubbing and oxidation of Hg⁰, but that higher concentrations of SO₂ then result in less scrubbing and oxidation. Scott et al. (2003) have shown that SO₂ [S(IV)] can reduce Hg²⁺ to Hg⁰ in both the gas and liquid phases and that one of the products in addition to Hg⁰ is the oxidized species HgS, which results from disproportionation of the S(IV) species. Such a reaction could be \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}3 \ {\rm Hg}^{2+} + 6 \ {\rm SO}_2 + 8 \ {\rm H}_2 {\rm O} = & 2 \ {\rm Hg}^0 + {\rm HgS} \\& + 5 \ {\rm H}_2 {\rm SO}_4 + 2 {\rm H}^+ \tag{25}\end{align*}\end{document}

Zhao et al. (2006) have shown that SO₂ inhibits the oxidation of Hg⁰ by Cl species in the gas phase when water is present, but that no inhibition occurs when water is absent. In aqueous solutions, the formation of an Hg•S(IV) intermediate complex followed by decomposition to form Hg⁰ has been proposed to explain the reemission of Hg as Hg⁰ from scrubbers that remove Hg²⁺.

Sulfur dioxide will act as a reductant for oxidized mercury or chlorine. In OLI, Cl₂ and SO₂ preferentially react, and thus Cl₂ does not oxidize Hg⁰. This redox chemistry reflects the actual experimentally determined chemistry. Cauch et al. (2008) showed that SO₂ reduces Cl₂ to chloride in solution, while sulfur is apparently oxidized to sulfuric acid: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Cl}_2 + {\rm SO}_2 + 2 \ {\rm H}_2 {\rm O} = 2 \ {\rm HCl} + {\rm H}_2 {\rm SO}_4 \tag{26}\end{align*}\end{document}

However, excess SO₂ in OLI can oxidize Hg⁰ to HgS, while sulfur itself disproportionates to S(−2) and S(+6), as shown in Reaction (27). This reaction does not appear to actually occur in real systems. \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm Hg}^0 + 3 \ {\rm SO}_2 + 2 \ {\rm H}_2 {\rm O} = {\rm HgS} + 2 \ {\rm H}_2 {\rm SO}_4 \tag{27}\end{align*}\end{document}

Therefore, like O₂, the effect of sulfur redox in OLI must be carefully checked. In OLI, with S redox turned on, SO₂ in water disproportionates into elemental sulfur S₈ and sulfuric acid, H₂SO₄: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}24 \ {\rm SO}_2 + 16 \ {\rm H}_2 {\rm O} = {\rm S}_8 + 16 \ {\rm H}_2 {\rm SO}_4 \tag{28}\end{align*}\end{document}

Although this reaction may be thermodynamically favorable, the dissolution of SO₂ into water actually gives sulfurous acid: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland, xspace}\usepackage{amsmath, amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6}\begin{document}\begin{align*}{\rm SO}_2 + {\rm H}_2 {\rm O} = {\rm H}_2 {\rm SO}_3 \tag{29}\end{align*}\end{document}

Oxidation—ESCM tests

The homogeneous vapor-phase oxidation results from the FactSage modeling of Tests 1 and 3 were used as inputs to an approximate liquid-phase model. This composition did not contain any entrained glass or feed. For the results from both tests, the offgas composition was cooled to 25°C, resulting in aqueous condensate, a gas phase, and precipitated Hg species. Because of the effect of O₂ on the redox of mercury, the O₂ in these streams was converted mole-for-mole to inert N₂. The OLI model input species are shown in Table 11. Species specific to FactSage that are not valid species in OLI were adjusted as shown. For the OLI input, the phase does not matter, so the gas-phase and condensed-phase species were added.

The pH of actual DWPF condensate samples is approximately 2–2.5, so the pH was adjusted to 2.5 in the aqueous models by adding NaOH. The pH without adjustment was about 1.6. The temperature and pressure were assumed to be 25°C and 1 atm. Ambient temperature was chosen because even though the condensate tank temperature was higher, the samples were cooled to room temperature before analysis, so the actual speciation measured would be that at room temperature.

The target ratios of the mercury species are shown in the first column of Table 12. The FactSage output is shown in the second column. The third column shows the OLI model output using the inputs from Table 11. For both tests, the amount of Hg(I) is overpredicted. Because the proportion of chlorine present as Cl₂ at the quench temperature of around 45°C–60°C will be higher than those in Table 11 (at 650°C and 450°C), the amount of Cl₂ was adjusted in the OLI model. In Table 12, the predicted mercury speciation for the best-fit Cl₂ amount for both tests is shown in the column A; values using the average Cl₂ are shown in column B. The original and adjusted amounts of Cl₂ are shown in the bottom two rows. For both cases, increasing the Cl₂ amount to around 0.27 mM gives mercury speciation that matches the target values. This very small aqueous concentration of Cl₂ is quite reasonable. In future work, the Cl₂ concentration entering the quencher, and after cooling, will be calculated using FactSage. The calculation scheme and results are presented in a flow diagram in Fig. 7.

OPEN IN VIEWER

FIG. 7.

Flow diagram of calculation scheme.