Mercury capture by native fly ash carbons in coal-fired power plants

Abstract

The control of mercury in the air emissions from coal-fired power plants is an on-going challenge. The native unburned carbons in fly ash can capture varying amounts of Hg depending upon the temperature and composition of the flue gas at the air pollution control device, with Hg capture increasing with a decrease in temperature; the amount of carbon in the fly ash, with Hg capture increasing with an increase in carbon; and the form of the carbon and the consequent surface area of the carbon, with Hg capture increasing with an increase in surface area. The latter is influenced by the rank of the feed coal, with carbons derived from the combustion of low-rank coals having a greater surface area than carbons from bituminous- and anthracite-rank coals.

The chemistry of the feed coal and the resulting composition of the flue gas enhances Hg capture by fly ash carbons. This is particularly evident in the correlation of feed coal Cl content to Hg oxidation to HgCl₂, enhancing Hg capture. Acid gases, including HCl and H₂SO₄ and the combination of HCl and NO₂, in the flue gas can enhance the oxidation of Hg.

In this presentation, we discuss the transport of Hg through the boiler and pollution control systems, the mechanisms of Hg oxidation, and the parameters controlling Hg capture by coal-derived fly ash carbons.

1. Introduction

The control of mercury at United States coal-fired power plants has been one of the objectives of the most recent generation of clean air rules from the US Environmental Protection Agency [1-2]. Provisions of the U.S. Environmental Protection Agency’s Clean Air Interstate Rule (CAIR) [3] and Clean Air Mercury Rule (CAMR) [4] would have dictated limits the amount of Hg emissions from coal-fired power plants. CAIR would have indirectly cut Hg emissions by calling for increased flue-gas desulfurization of power plants in the eastern United States, while CAMR set strict guidelines for emissions throughout the country. Both would have been enacted by cap-and-trade guidelines. However, in 2008, CAIR¹ and CAMR were vacated by the United States Court of Appeals District of Columbia Circuit [6-7], the latter upheld by the United States Supreme Court [8]. Consequently, in February 2009 the Obama administration’s Department of Justice decided to not pursue appeals of the rulings and instead will draft new rules for mercury control. CAIR remains in place until the US EPA issues new guidelines [9]. Some utilities are proceeding with planned construction of FGD units, while others will wait for clarification of the regulations [10].

Engineered solutions, such as the injection of halogenated activated carbon into the flue gas stream [11] have been proposed to capture Hg from the flue gas. Unburned carbon in fly ash will also adsorb varying amounts of the Hg in the flue gas stream. In this paper we discuss the nature of native carbon in fly ash, the controls on Hg capture, and the boiler and ash-collection parameters which influence Hg capture.

2. Mercury in coal

The most fundamental control on the amount of Hg in fly ash is the amount of Hg in the feed coal. The amount of Hg in feed coals varies considerably, for example, for US coals the average Hg content is nearly 0.20 μg/g [2, 12], while the delivered coal to US utilities, based on the US Environmental Protection Agency’s 1999 Information Collection Request, is about 0.10 μg Hg/g (1999 data) [2]. Differences arise from the overall average being based on a wide range of coals, many not currently mined, while the delivered coal average reflects mining practices, coal beneficiation, and utility coal quality specifications, among other variables, all of which can, intentionally or not, serve to reduce Hg in the delivered coal.

If economics and mining conditions permit, Hg in the power plant feed coal can be avoided by not mining Hg-rich lithologies [13-17]. Mercury can be reduced through beneficiation prior to delivery to the power plant [12, 18] or through rejection of coarse and/or dense particles by pulverizers [14, 17, 19-20]. The pulverizers rejects, or “pyrites” as known in the power industry, typically account for less than 1% of the feed coal going to the pulverizers, but can contain >10% of the total Hg in the delivered coal [19].

Hower et al. [20-21] showed that Hg content varied significantly between sites of Pennsylvanian coalbeds in eastern and western Kentucky and Indiana. Mercury in coal has been found to be associated with pyrite and marcasite [23-28]. Mercury has also been found in calcite and chlorite [27]; gold minerals [29-30]; clausthalite and other Pb and Se minerals [31-32]; cinnabar, metacinnabar, and native Hg [33]; getchellite [34]; and kleinite and organic complexes [35]. Yudovich and Ketris [36] provided an extensive review of Hg chemistry in coal, including summaries of the Soviet and Russian literature on the subject. Ding et al. [24], in an electron microprobe study, had proven Hg concentrations in pyrite ranging from 200-4700 μg/g. Hower et al. [28], using a scanning electron microprobe, demonstrated that concentrations in marcasite as low as 39 μg Hg/g could be detected. All other Fe-sulfide grains analyzed were within three times the limit of detection; therefore, the values were not significant.

The basic lesson from all of these studies is that Hg is not observed at consistent levels within individual coalbeds, as noted above, but it also does not occur at consistent levels within Fe-sulfides or is it only found in Fe-sulfides. Instead, it is present in other minerals and in organic combination. Therefore, attempts at simple correlation between Hg and pyritic sulfur are flawed, not only due to the problems in comparing elements present at differences of about five-orders-of-magnitude, but also due to Hg not always being associated with sulfides.

3. Evolution of Unburned Carbon in Coal-Fired Power Plants

A voluminous literature exists regarding the nature of the char intermediate of coal combustion processes. No attempt will be made here to exhaustively review this literature. Rather, the focus will be on background literature of most relevance to Hg capture by the native char intermediate formed in pulverized coal fired boilers, which are the most important anthropogenic source of Hg emissions. Mercury adsorption on coal-derived carbon occurs at temperatures below 300 °C, which correspond to the flue gas cooling zone where carbon combustion has become very slow, so the challenge of describing the carbon sorbent at the point of Hg capture is almost identical to the problem of describing the unburned carbon in ash.

The unburned carbonaceous fraction of coal fly ash has received considerable attention over recent years with respect to its role in boiler efficiency, combustion system operation and the subsequent beneficial use of the ash. Perversely, the same properties that may favor Hg capture (high carbon content, high surface area, and fine particle size distribution) are those that lead to problems in ash utilization, both phenomena being ultimately related to adsorption, as will be seen. It is useful when considering the possible role of unburned carbon in Hg capture, to briefly recount its role in some other important adsorption processes.

Unburned carbon has a significant impact on potential beneficial use of fly ash. If present in large enough amount, this impact is always negative. Generally speaking, utilities will try to run their boilers such that fuel burnout is as high as possible, and hence, unburned carbon in the ash will be as low as possible. There are, however, issues of boiler design, coal selection and optimization with respect to a variety of emissions requirements that result in an unburned carbon content of around a percent to a few percent by mass in the fly ash from a typical power station. The economic and design considerations that lead to this incomplete combustion generally outweigh the incentives to reduce unburned carbon below a few percent, even if this would make the ash more attractive for beneficial use.

The major beneficial use of coal fly ash in the U.S. is as a pozzolanic additive in concrete [37]. In such use, the presence of unburned carbon can pose a problem associated with adsorption of air entrainment additives (Fig. 1) in the concrete mix [39-40], as well as in certain cases cause problems with color and/or concrete-mix water requirements and behavior. Consequently, commercial processes have been developed for the post-combustor removal of this unburned carbon by burnout in a downstream process (the CBO process of PMI Technologies, LLC) or by electrostatic processes (Separations Technologies, LLC and Tribo Flow Separations, LLC). Tuning the boiler to reduce unburned carbon is indeed possible, but at the expense of an increase in NOx emissions [41]. This is entirely expected, as it is known that air staging for reduction of NOx historically resulted in higher levels of unburned carbon in the ash. Hence, subtle changes in boiler operational parameters can potentially result in significant changes in unburned carbon amounts and properties. Such changes could naturally also impact Hg adsorption capacity of a fly ash. A review of various environmentally driven coal combustion technology changes was prepared by Beer [42].

Figure 1.

Sketch of competitive surfactant adsorption that contributes to the deleterious effect of unburned carbon on fly ash utilization when used in concrete as a replacement for Portland cement. The hydrophobic fraction of the carbon surface, not occupied by oxygen-containing functional groups, provides adsorption sites for the hydrophobic portion of the surfactant, or air entraining admixture, and reduces its availability to stabilize air bubbles necessary for freeze-thaw resistance in concrete. Adapted from Hachman et al. [38].

In certain cases, some of the unburned carbon in a fly ash might actually be soot [40, 43], and thus be of a different origin than the usual unburned carbon derived from pyrolytic processes in the solid fuel phase. This ultrafine carbon is believed to be much more active towards air entrainment agents than the usual unburned carbon fraction found in most ashes [40]. This is because very fine, highly dispersed carbon presents a readily available surface for adsorption. Thus, it is also important to have a complete understanding of the nature of unburned carbon for the prediction of adsorptive Hg capture. It is uncommon to find significant amounts of soot material in a well-tuned boiler ash, but if it exists, it could potentially play a significant adsorptive role beyond what its mass fraction might suggest. For a comprehensive review of various aspects of soot in coal combustion, see Fletcher et al. [44].

The unburned carbon in fly ash can also serve as a carrier for other species that are associated with the combustion process itself. This carbon has been cited as a possible site for adsorption of polycyclic aromatic hydrocarbons (PAH) that are produced during combustion, since carbon is a strong sorbent for PAH [45]. The levels of PAH associated with a fly ash are, however, a very strong function of the temperature and excess oxygen used in combustion [46], and the levels from pulverized coal combustion power plants are typically quite low (Wornat et al. [47] found levels of typical 2- to 3-ring PAH in the range of ppb and below). In fact, these inherent PAH levels are usually sufficiently low, suggesting that coal combustion fly ashes might themselves be useful for binding (adsorbing) PAH from other sources and residing in marine sediments [48]. Fly ash has also been considered as an inexpensive adsorbent for water or soil treatment, such as for adsorption of aromatic acid dyes [49].

Selective catalytic reduction (SCR) or selective non-catalytic reduction (SNCR) processes for control of NOx controls involve injection of ammonia or urea into fly ash-containing combustion gases, following the main combustion process. If the ammonia is not fully consumed in the process, there exists a problem of “ammonia slip” and some of the unconverted ammonia is adsorbed onto the fly ash. There may also be, in certain instances, ammonia addition to enhance the performance of electrostatic precipitators used to collect the fly ash. Any ammonia adsorbed on fly ash can become an environmental problem when the ammonia is released upon contact of the fly ash with water (see references in Gao et al. [50]).

SCR processes can influence Hg oxidation (e.g., Lee et al. [51], and discussed elsewhere in this review). Ammonia has also been implicated in increasing the release of Hg by leaching of fly ash [52], creating a need to understand the linkage between ammonia sorption and Hg. There have been relatively few fundamental studies of the adsorption process of ammonia on fly ash (e.g., Turner et al. [53]). Pure ammonia has been found to follow more or less standard physisorption behavior on fly ash carbon surfaces, but it has also been concluded that the high concentrations seen in actual field samples reflects a more complicated chemistry [54].

These studies serve to emphasize that unburned carbon is known to play an important role in adsorption of other species, in addition to Hg, which are present in coal combustion systems. These adsorption processes follow, in many respects, the expected pattern in which the high surface area unburned carbon phase predominates in sorption processes, when compared with the low surface area mineral (inorganic ash) phase. Hence, the remainder of this section will focus on development of available carbon surface during coal char burnout. In this section, the focus is on microscopic surface area, that is, the surfaces contained in submicron and nanometer scale porosity. Section 4 of this review is concerned with carbon morphology on a larger or macroscale, in which petrographic techniques provide information. Both scales are important- the microscale for establishing the overall extent of available surface, and the macroscale for determining how readily available this surface is for species present at the nominal geometric surface of the ash (or unburned carbon) particles.

3.1 Evolution of Coal Char Structure during Burnout

There have been numerous studies that have examined the nature and amount of unburned carbon from pulverized coal combustion systems [43, 55-84]. This selection of papers is by no means complete, and primarily more recent studies, and those with a tie to Hg capture, have been emphasized. There are some general conclusions that may be drawn from the above studies.

• Utility boilers operate at quite high carbon conversions, most often well over 99% [85]. Typical unburned carbon in fly ash ranges from near a percent by mass to a few tens of percent, with well-tuned boilers typically giving ashes with unburned carbon well below 10%.

• Unburned carbon levels in ash depend in a complicated way on boiler type, operation, and fuel selection. Power stations that use a range of coals, often including internationally traded coals, have found systematic fuel-to-fuel differences, which are believed to result from differences in grinding and classifying behavior, char yield, char reactivity, and char morphology as affected by coal rank, type, and mineralogy.

• There are a variety of different carbon types that may be found in the ash. The generally predominant form is char derived from pyrolysis of the coal, which leads to discreet, carbon-rich particles clearly visible in ash under the optical microscope (Fig. 2). Soot also forms in the fuel-rich, near-burner zones, but, due to its ultrafine size, burns out readily in the oxygen containing post-flame zones unless it fails to contact oxygen by burner imbalance or poor mixing of secondary, tertiary, or overfire air [40, 43] and thus soot, the nanoscale carbon form, is not commonly a major contributor though may play an adsorptive role in some special cases [75].

• Related to the above, there is always a distribution of unburned carbon particle sizes in the ash [67] and a distribution of unburned carbon in different size fractions of the ash [55, 78]. It is quite common to find a high fraction of unburned carbon mass in relatively large particles with a high carbon content [55, 67, 78], though this is not necessarily always the case. Figure 3 shows examples of size distributions for the total fly ash, while Fig. 4 shows the distributions for the carbon component alone. The unburned carbon is typically significantly coarser than the mineral component of the ash (Fig. 3) due to differences in the natural grain sizes of the two components in the feed coal. In this data set (Fig. 4), the carbon size distribution is also coal rank dependent, with low rank coals showing a larger percentage of the total carbon in the largest two size categories (> 180 um). A number of factors may contribute to this trend, but it is likely that the dominant effect is the higher overall burnout for low rank coals – when a polydisperse population of char particles burns in a diffusion influenced regime, the size distribution becomes courser as small particles burn out leaving larger particles increasingly concentrated in the unburned carbon fraction of ash.

• There is general agreement that carbon surface area does play some role in adsorption of various species present in the combustion environment. Figure 5 shows that the surface areas of the carbon fractions themselves are typically 20-70 m²/g-carbon for unburned carbon derived from bituminous coals and 300-400 m²/g-carbon for unburned carbon derived from lower rank (non-softening) coals [58]. The possible range of surface areas is strongly imprinted by the choice of coal, but there is an influence of combustion environment. The implication is that final burnout is taking place under partial mass transfer limited (Zone II) conditions [66]. Figure 6 provides evidence that the full-scale combustion process occurs in a regime influenced by oxygen diffusion. The surfaced area of unburned carbon in ash can be greatly increased (the carbon can be “activated”) by subsequent reaction with oxygen at low temperatures.

• The total sorptive area of a fly ash will be determined mainly by its carbon content, since the mineral portion normally has very low surface area (< 1 m²/g-inorganic ash) [58].

• Within the coal-derived unburned carbons there exist a wide range of different morphologies whose presence and structures depend upon the nature of the starting coal and the combustion conditions.

Figure 2.

Optical micrograph of fly ash sample from full-scale utility boiler. Arrows show discreet carbon-rich particles that make up the major fraction of unburned carbon in most samples. Adapted from Kulaots et al. [67].

Figure 3.

Particle size distribution of whole fly ash samples from full-scale utility boilers for nine different parent coals. Adapted from Kulaots et al. [67].

Figure 4.

Particle size distribution of the unburned carbon fraction of fly ash sampled from full-scale utility boilers for nine different parent coals. Adapted from Kulaots et al. [67].

Figure 5.

Surface areas per unit mass of carbon in a collection of fly ash samples from full-scale utility boilers. Adapted from Gao et al. [50].

Figure 6.

Development of surface area in full-scale unburned carbon samples through post-capture air oxidation at low-temperature in the laboratory. Adapted from Kulaots et al. [66].

The key features of interest with respect to the potential of unburned carbon to contribute to Hg capture are its amount, size distribution, surface area, and surface chemistry, all of which vary through the combustor as the char burns out. It has been well established that the amount of carbon in fly ash generally plays some role in determining its capacity to sorb Hg [19, 70-72, 75, 77, 79, 86-87], since carbon is the main sorptive species in the ash. There are clearly issues of morphology and surface chemistry that are known to play important roles as well [64, 75, 78-80]. In this chapter on the combustion process, we focus on the amount of carbon that is left in a fly ash, its size distribution, and its surface area. Other important factors such as char surface chemistry, the role of heteroatomic species (especially the halogens), the nature of the contacting of the ash, and Hg (especially temperatures and residence times) are covered in other sections of this review.

The overall outline of the coal combustion process is generally understood. The reader can refer to many different summaries of the process, such as those by Essenhigh [88] or Smoot [89]. As coal is first heated in the boiler, it devolatilizes, giving off tars and gases. Depending upon particle size and coal composition, as well as on the mixing characteristics of the boiler, combustion may initially take place in the vapor phase in a volatile-rich phase, or the particle might ignite on its surface [88]. Here, there will be no further consideration of volatiles combustion, but rather an emphasis on factors determining the final char structure. It is useful to note, however, that depending upon the details of the combustion system, soot formation is strongly tied to volatiles chemistry. If the burnout of the soot is incomplete, there can be an influence of volatiles processes on unburned carbon in fly ash.

During the pyrolytic phase in the bituminous coals, the organic matter will typically soften, whereas in the case of low-rank coals, the organic matter will generally behave in a thermosetting (non-softening) manner. The fluidity of the pyrolyzing organic matter largely determines the observed morphological characteristics of the char, and different macerals within the same coal will behave differently (some may soften, others not). If the particles soften during pyrolysis, then a fluid with low amounts of “free volume” obtain, and upon resolidification, there is no mechanism for large amounts of porosity to be re-created. Within a single coal, the unburned carbons from inertinite macerals have been shown in one case to have the lowest density and surface area, followed by those that are isotropic and anisotropic [69].

The actual burnout of the char will follow processes dictated by both the reactivity of the char as well as the mass transfer characteristics in the boiler. Porous solid-gas reactions are often characterized by three reaction regimes: kinetic control (sometimes called Zone I), pore diffusion control (Zone II), and film diffusion control (Zone III), with the shorthand zone nomenclature adopted from the work of Walker et al. [90]. It was once commonly assumed that the temperatures of combustion, and thus rates, are sufficiently high in pulverized coal combustors that the combustion mostly takes place under Zone III conditions. This view is, however, not consistent with observations. The more generally accepted view is now that pulverized coal chars burn under Zone II conditions, at least in later, lower-temperature regions of a boiler where the final char burnout is occurring, and in this regime the carbon reactivity does appear to impact observed results [66, 85, 91].

Added to these time-dependent processes are significant three-dimensional spatial variations in full-scale boilers that make the modeling of coal combustion processes in the field exceptionally complicated. The chemistry of the pyrolytic process already involves large numbers of distinct chemical species, and the combustion processes add many more (not all of which are even firmly established). Onto this must be added the complexities of heat transfer and fluid flow, and the physical processes in the solid phases (including both the organic phase as well as the ash phase). Summaries of some of the relevant combined computational fluid dynamics (CFD)-combustion models have been published [89, 92-93]. There remain important questions regarding the ability of the CFD-based models to correctly predict unburned carbon values at the level of accuracy that might be required for predicting Hg capture [94-95].

A different approach to modeling the char burnout process is that embodied in the so-called CBK (Char Burnout Kinetics) model [96]. In this model, the focus is on the problem of carbon burnout, and the aim was particularly the prediction of the late stages of carbon burnout. The CBK model is not intended to be a complete combustor model, as are the others alluded to above, but rather, to be a stand-alone computation of char burnout, provided that the a particular system can somehow be characterized with respect to relevant temperature-time characteristics (something that is inherent in the CFD models). The CBK model includes a statistical kinetics description of char combustion, as well as the process of annealing. It also includes inhibition of the kinetics by an ash layer. It has been implemented in another code designed to predict unburned carbon levels in fuel switching scenarios, again without explicit reference to fluid flow or transport processes in a full-scale boiler [97]. The model has also been extended by Stephenson [98], and in modified form by Cloke et al. [99] and Wu et al. [100]. It has also been grafted onto a CFD approach, in a two-step calculation [94, 101-102]. Another integration into a CFD code has been described [103], but comparison of predictions to actual full-scale combustor results were not available. A different, detailed particle-level combustion model has recently been presented, that seeks to predict the development of surface area and particle morphology under both Zone I and Zone II conditions [104]. Again, this latter model seeks to predict behavior assuming information is separately available on the particle environment.

There is a significant challenge for any effort at modeling coal combustion with an eye towards predicting the adsorptive properties of unburned carbon. First, the accurate prediction of unburned carbon levels, at the desired level of accuracy, is beyond what can be expected from any of the CFD-based computation models at the present time. Factors of two in unburned carbon predictions might be hardly noticed in models aimed at capturing the main features of heat transfer and gas phase processes, but will make a large difference in determining Hg-sorption capacity. Beyond this, not even the more detailed burnout models attempt to get into the prediction of porosity or into the details of fragmentation (prediction of unburned carbon particle size distributions). Both features are known to be important in adsorption processes. Prediction of porosity development during combustion and gasification involves a significant effort [105-106]. This is in addition to the effort that would be needed to describe the formation of bubble and void structures in softening coals [107-108], or the fragmentation of particles during combustion [84, 109].

4. Fly ash carbon morphology

The classification of fly ash carbons has gone through a number of iterations through the years [60-62, 110-115]. Based on those discussions, and on the philosophy behind coke petrography [116], Hower et al. [117] developed a system suitable for the bituminous-coal-derived fly ashes of the eastern US. Unlike the morphology-based descriptions of Bailey et al. [62], their system relied more on the optical properties of the carbons, dividing the neoformed vitrinite- and semi-inertinite-derived carbons into isotropic and anisotropic cokes or chars (Figure 7). Inertinite-derived carbons pass largely intact through the boiler (Figure 8), although some alteration can be detected (Figure 9). As noted by Hower et al. [118], “although fly ash inertinite is likely derived from inertinite in the coal, it cannot be said that isotropic and anisotropic chars are derived strictly from vitrinite.” Relatively unburned, albeit slightly devolatilized, coal is rare (Figure 10). Hower and Mastalerz [119] developed a system which combined aspects of the Hower et al. [117] optical-properties system and the morphology-based system of Bailey et al. [62] and the International Committee for Coal and Organic Petrology [120] Hower et al. [118] later expanded the definitions to include coal of lower and higher rank than the bituminous coals of the original system.

Figure 7.

Inertinite (bottom) with anisotropic coke from combustion of high volatile A bituminous central Eastern Kentucky coal. (scale = 270 microns along long edge).

Figure 8.

Variety of inertinite forms from power plant burning high volatile A bituminous Central Appalachian coal. (scale = 330 microns on long edge).

Figure 9.

Inertinite from a power plant burning high volatile A bituminous central Eastern Kentucky coal. Note incipient breakdown of inertinite cell walls. (scale = 270 microns along long edge).

Figure 10.

Partially burned coal with devolatilization features, remnant liptinite, and oxidation rims along particle edges and along fractures. (scale = 330 microns on long edge).

Coal rank is an important factor in determining the type of carbon in fly ash. Vitrinite/huminite (nomenclature for vitrinite/huminite and inertinite group macerals can be found at ICCP (1998, 2001) and Sykorova et al. (2005)) from low-rank coals generally does not form the isotropic and anisotropic carbons illustrated above for bituminous coal-derived fly ash. Upon heating, low-rank coals do not undergo thermoplastic transitions. Rather, the chars derived from low-rank coals are dominated by forms that are devolatilized, but without the devolatilization vesicles of higher-rank coals (Figure 11). Bituminous coals will generally swell and undergo thermoplastic transitions upon heating to 300-400 °C [116]. In the boiler, such plasticity leads to neo-formed (generally) vitrinite-derived structures, as seen on Figure 4.1, and to carbon cenospheres. In coals of any rank, coal included within rock may be carbonized, but it will generally not be combusted nor exposed to the flue gas, therefore, the carbon would not be accessible for Hg capture (Figure 12).

Figure 11.

Isotropic char from the combustion of subbituminous western US coal. (scale = 200 microns along long edge).

Figure 12.

Vitrified outer boundary of particle. The interior has been thermally altered; note high reflectance carbons. From Tennessee power plant burning medium-sulfur, high volatile A bituminous Central Appalachian coal. (scale = 330 microns on long edge).

Coals of semi-anthracite rank and higher generally do not display thermoplastic properties. Vesiculated chars have been observed in anthracite-derived fly ash carbons and porous chars have been noted in meta-anthracite-derived chars [70-71, 118]. Both isotropic and anisotropic chars have been seen in fly ashes from high-rank coals. (Figure 13)

Figure 13.

Fly ashes from Portuguese power plant burning anthracite. Sample courtesy of Bruno Valentim, University of Porto. Scale bar is 25 microns on both images. a/ Anisotropic coke (across diagonal) and other anthracite-derived carbons. b/ Vitrinite-derived anisotropic carbon (right center), perhaps without significant alteration from feed coal, and other anthracite-derived carbons.

A fullerene-like carbon form has been found in emissions from power plants as well as in the form of a deposit on glassy particles in bituminous-coal-derived fly ash [43, 74, 78, 124-127]. To date, low- and high-rank-coal-derived fly ashes have not been investigated for the presence of this carbon form. As seen in Figure 14 (Figure 3 from Hower et al. [74]), the carbons can form bridges, presumed to be rather fragile, between the glassy Si-Al spheres dominating this and most fly ashes. The small, few-nm, dark spots in Figure 14 are metal grains. The chemistry of these grains will be discussed in section 6.

Figure 14.

High-resolution transmission electron microscope (HRTEM) images of C-rich nano-clusters. Large dark round bodies in 4.8a and 4.8b and at extreme upper right of 4.8c are Si-Al glass fly ash particles. Carbon surrounds Si-Al particles and acts as a bridge between Si-al grains. Few-nm dark spots within carbon in 4.8d are metal grains. (scale: a – 0.2 microns; b – 0.1 micron; c – 20 nm; d – 10 nm)

5. Hg chemistry in boilers and APCDs

The principal Hg input stream to a coal-fired boiler is the coal. The form of Hg in the coal is either mineral-bound (often in the sulfide phase) or organically associated, as discussed in Section 2. In either case, in the high temperature of the flame in a coal-fired boiler, all the Hg is expected to be in the vapor phase [128-129]. Mercury is expected to be in the gaseous elemental form in the flame, according to equilibrium calculations [128, 130]. In a coal-fired boiler, the flue gases cool from high temperature (~1600 K) to the temperature at the inlet to the air pollution control devices (APCDs) (400-450 K) with a cooling rate of ca. 300 K/sec [130]. Given the elements that are typically present in coal and air, thermodynamic equilibrium predicts that Hg will be HgCl₂ at the temperature of the inlet to the air pollution control device ([128, 103]. However, the oxidation of elemental Hg to HgCl₂ is kinetically limited in coal-fired boilers [130]; therefore, some fraction of the Hg is oxidized at the inlet to the APCDs.

Theoretical models have been developed to explain the speciation Hg in coal-combustion flue gases. Transformation of elemental to oxidized Hg involves both gas-phase (homogeneous) reactions [131-137] and gas-solid (heterogeneous) reactions [138-142].

As a result of chemical transformations in the flue gas, Hg enters the APCDs as a mixture of species. The methods available to measure Hg in coal flue gas can distinguish among gaseous elemental, gaseous oxidized, and particulate-bound Hg. Each of these different forms of Hg behaves differently in APCDs. The extent of Hg oxidation or conversion to particulate-bound Hg depends on the flue gas composition, the amount and properties of fly ash, and the flue gas quench rate.

A survey of the types of APCDs on coal-fired electric utility boilers in the US was carried out using NETL’s 2007 Coal Power Plant Database, which includes data from the DOE EIA-767 database [143]. Table 1 summarizes the major types of APCDs on coal-fired utility boilers in terms of total boiler capacity; the data have been subcategorized by type of APCD and by the rank of coal burned.

Table 1.

Distribution of air pollution control devices on coal-fired power plants in the U.S by total capacity in MW.

	Bituminous	Subbituminous	Lignite
FGD-SCR (all PCDs)	68,204	6,518	0
SDA (all PCDs)	5,552	8,711	1,320
C-ESP (except FGD-SCR)	81,289	88,357	10,714
FF	9,942	12,712	1,376
H-ESP	16,615	13,204	0
Other	535	7,534	50
Total	182,138	137,037	13,460

The most common particulate control devices (PCD) in U.S. coal-fired utility power stations are electrostatic precipitators (ESP), used with or without flue gas desulfurization (FGD) for SO₂ control. Most ESPs are “cold-side” or C-ESP, that is, they operate at temperatures in the range of 400-500 K. Subbituminous- and lignite-fired plants have predominantly cold-side ESPs without any desulphurization equipment (65% and 80%, respectively). A small number of ESPs are “hot-side” or H-ESP, and operate at 620 to 670 K. In an ESP, there are a number of electric fields arranged in series with respect to the gas flow. Particles are collected on charged plates in each field and fall into a hopper below. Typically each field of an ESP has an associated hopper. Ash collection takes place serially, with most of the ash being collected in the first field and lower amounts in subsequent fields.

A fabric filter (FF) is the other common PCD, which may be used alone or with a spray dry adsorber (SDA), the latter to remove SO₂. Fabric filter collectors have a series of compartments, each containing a number of bags on which the ash is collected. Each compartment has a hopper where the ash is collected upon cleaning the bags. Bags are cleaned by mechanical agitation or pulsing with air. Compartments are cleaned when the pressure drop across the bags reaches a certain level; ash collection in a fabric filter does not follow such a distinctive pattern as in an ESP.

Ten percent of the low-rank-fired plants have fabric filters and 10% of the subbituminous-fired plants have hot-side ESPs. Many boilers have a selective catalytic reduction (SCR) unit for removal of nitrogen oxides (NOx).

As part of EPA’s Information Collection Request (ICR), data were collected on Hg speciation and Hg removal across air pollution control devices at 83 full-scale power plants. For power plants in the EPA study, the percentage of Hg that was oxidized at the inlet to the PCD increased with increasing coal chlorine content [144]. The amount of elemental Hg at the inlet to cold-side ESPs was less than 20% as long as the coal Cl content was greater than about 500 μg/g. Bituminous coals from the U.S. typically would have greater than 500 μg/g Cl [2]. Thus, bituminous coal-fired power plants are expected to have higher amounts of oxidized Hg in the flue gas.

The average Hg removal across the wet FGDs in this dataset was well over 50% for bituminous coals but only about 30% for subbituminous coals [144-145]. The amount of Hg removed across a wet FGD depends largely on how much oxidized Hg is present at the scrubber inlet [146]. SCR systems have been observed to oxidize Hg in coal-fired power plants [147-149] and the range of Hg oxidation observed in plants firing bituminous coals was 30% to 98%. Limited data from plants burning subbituminous coal suggests much lower amounts of Hg oxidation across SCRs in these plants. Subbituminous coals and lignites have lower Cl than bituminous coals [2]. Modeling of Hg oxidation across SCRs [77] has demonstrated that coal Cl content is one of the key factors affecting Hg oxidation across SCRs.

The combination of an SCR and a wet FGD scrubber can remove 90% or more of the Hg (from input to stack), if there is sufficient oxidized Hg at the scrubber inlet. Withum [150] measured Hg removal on eight bituminous coal-fired boilers with SCR and FGD; average Hg removal (coal to stack) on these boilers varied from 65% to 97%.

A review of some of the data collected from full-scale systems showed that Hg removal across ESPs in pulverized-coal fired boilers burning bituminous coals appeared to be related to the LOI or unburned carbon in the fly ash [77]. Sulfur in the flue gas has been shown to negatively affect capture of Hg by carbon, and the effect is most pronounced where both LOI and coal chlorine are high, and is consistent with the notion that sulfur fills reactive sites on fly-ash carbon [140]. Data reported by the Canadian Electricity Association [151] for two units equipped with cold-side ESPs showed that Hg capture across the C-ESP decreased as coal sulfur increased.

Sjostrom et al. [152] estimated the LOI of the fly ash from ICR plants equipped with fabric filters. Good Hg removal was observed across fabric filters in the ICR database for boilers burning bituminous and subbituminous coals, an average removal 70 and 84%, respectively. Little removal was observed across fabric filters on plants firing lignite coal. Sjostrom et al. [152] did not observe a correlation between Hg removal across fabric filter and LOI, coal Cl, or temperature.

In summary, APCDs collect Hg via two pathways: removal of particulate-bound Hg in particulate control devices and removal of gaseous oxidized Hg in flue gas desulphurization (FGD) scrubbers or spray dryer absorbers (SDAs). The behavior of Hg in air pollution control devices is summarized in Figure 15.

Figure 15.

Behavior of mercury in common air pollution control devices in coal-fired power plant.

6. Distribution of mercury in fly ash collection systems

The distribution of trace elements within fly ash collection systems is, first, dependent upon the concentration of elements in the feed coal [16-17, 130, 153-168]. Following combustion, the partitioning of volatile trace elements, such as Pb, As, and Zn, within the electrostatic precipitator (ESP) or baghouse or fabric filter (FF) array is a function of the temperature of the flue gas at the collection point and the ESP or FF row [14, 19, 114-115, 153, 169-175]. Specifically, trace element concentration will generally increase towards the back rows of the collection system, coincident with a decrease in the flue gas temperature and a decrease in the particle size of the fly ash and a concomitant increase in the fly ash surface area. The latter is a consequence of the first rows of the pollution-control system scalping off the coarser fly ash particles, leaving the finest particles in the final rows (Figure 16).

Figure 16.

Size distribution of fly ash collected by economizer, mechanical (cyclone), and electrostatic precipitators at a Kentucky power plant burning southeastern Kentucky high volatile A bituminous coal.

Capture of Hg by fly ash varies from the behavior of other volatile elements [16, 69, 72, 75, 130, 159-160, 176-184]. Li et al. [79], however, in a study of seven different bituminous coals burned in a utility 100-MW boiler, the same unit studied by Sakulpitakphon et al. [185], could not discern a relationship between the amount of feed-coal Hg and the fly ash Hg. They [79] did not mention whether the analyzed coal was the delivered coal or the pulverized coal. Aside from the variations in the amount of Hg in the boiler feed coal (section 2), Hg concentrations in fly ash have proven to be largely a function of the (a) amount of carbon, (b) the flue gas temperature at the point of collection, (c) the composition of the flue gas, and (d) the type of fly ash carbon, including variations dependent upon the rank of the feed coal. Each of these variables will be discussed below.

6.1 Variation in mercury capture by amount of fly ash carbon

Within a single field of an ESP or FF (that is, within the ash collected in the row of the hoppers corresponding to that field), Hg can be correlated with the amount of carbon in the fly ash [14, 16, 19, 174, 185-186]. Hower et al. [174] analyzed Hg from wet-screened² fly ash collected in consecutive months from the same ESP row at a Kentucky power plant. They found that Hg was highly correlated to fly ash carbon (Figure 17), with the important caveat that both the flue gas temperature at the collection point and the feed-coal source were similar between the two collections. Hower et al. [184] extended the correlation to the mechanical separation prior to the FF array and to the FF hoppers at a western Kentucky power plant. Among all of the variables investigated by Li et al. [79], fly ash carbon was one of the stronger correlations to Hg capture. They only obtained one ESP fly ash sample for each of the feed-coal burns studied; therefore, their investigation could not discern carbon vs. Hg trends within or between ESP rows for each of the coals.

Figure 17.

Fly ash carbon versus mercury in wet-screened fly ashes from two collections of the same row of a Kentucky power plant burning a high volatile bituminous Illinois Basin coal blend (after Hower et al. [174]).

The conversion of boilers to low-NOx combustion systems through the 1990’s resulted in a general increase in carbon in the post-NOx-conversion fly ash [189], although this was not always the case [190]. The fly ashes obtained from ESPs from several of the conversions were later analyzed for Hg and showed generally good intra-row Hg vs. fly ash carbon correlations, including the near coincidence at zero of the extrapolated Hg and fly ash carbon (Figure 18).

Figure 18.

a/ Pre- and post-NOx conversion Hg vs. ultimate analysis carbon for two power plants burning Illinois Basin coal. b/ Detail of low-C portion of plant MC plot. Steeper regression line is for samples with less than 5% C. Shallower regression line is for all samples. (modified after Hower et al [19])

6.2 Variation in mercury capture by flue gas temperature

Many of the studies cited in section 6.1 also addressed the variation by flue gas temperature at the collection point.

The Hower et al. [184] study of an Appalachian-low-S-coal-fired power plant with a three-row mechanical/ five-row baghouse (FF) ash collection system, mentioned above, is notable for the difference in Hg levels between the two parts of the system (Figure 19). The mechanical-separated fly ash has higher carbon content than the FF fly ash but, owing to the higher flue gas temperature, the Hg content of the mechanical fly ash is significantly lower than the FF fly ash.

Figure 19.

Fly ash carbon versus mercury for mechanical and baghouse (fabric filter) collection for two units burning high volatile A bituminous Central Appalachian coal at the same power plant. (after Hower et al. [184]).

Multi-year studies at a power plant with a five-row-ESP-ash-collection system, also burning central Appalachian coal, demonstrated a clearer relationship between Hg vs. C within ESP rows, as well as a clear relationship with ESP row as a proxy for flue gas temperature. With the exception of two hoppers on the fourth ESP row, the same hoppers were sampled in both 2004 and 2007. The fifth row hoppers were empty on both sampling dates. Figure 20 (based on unpublished data, University of Kentucky Center for Applied Energy Research) demonstrates: (1) a general increase in Hg in the ash from the later fields of the ESP, implying a lower flue gas temperature; and (2), for both years, an increase in Hg with fly ash carbon. The differences in Hg in rows 3 and 4 for 2007 versus 2004 is a function of the amount of fly ash carbon, perhaps among other unquantified parameters, and not to the amount of Hg in the feed coal, 0.09 μg/g for both years.

Figure 20.

Fly ash carbon versus mercury for the first four rows of a five-row electrostatic precipitator array at a Kentucky power plant burning high volatile A bituminous coal. Points represent sample collections in 2004 and 2007. Unpublished data from University of Kentucky Center for Applied Energy Research.

In separate studies at a 200-MW utility boiler with (initially) a two-row mechanical and three-row ESP ash collection system, Mardon and Hower [14] and Hower et al. [186] demonstrated the complexity of the Hg-carbon relationships for up to three rows of the ESP for five different collections from 2001 to 2007 (Figure 21). For the row-by-row collections, as best as possible, collections from each row followed a straight path through the ESP array. Two important caveats must be considered in examining this data. First, the coal source, while always from the central Appalachians, usually southeastern Kentucky, did change throughout the period examined as the company purchased some of their coal on the spot market. Second, for collections after 2002, the utility modified the collection system through the bypassing of the mechanical hoppers. For the 2004 and 2007 collections, all of the post-economizer fly ash passed directly to the ESP’s. In contrast to many other plants we have examined, carbon is relatively more abundant in the later rows of the ESP than in the first row [170, 191]. Figure 22 illustrates the relationship for all first-row and second-row ESP’s sampled in Kentucky in the 2002 and 2007 pentannual collections.³ In sharp contrast to the single-ESP-row examples, the many permutations of feed-coal Hg, feed coal halogen content, flue-gas temperature, among other variables, obscure the relationship between Hg and fly ash carbon

Figure 21.

Fly ash carbon versus mercury for a three-row electrostatic precipitator array at a Kentucky power plant burning high volatile A bituminous coal. Points represent five sample collections from 2001 to 2007. Unpublished data from University of Kentucky Center for Applied Energy Research.

Figure 22.

a/ Fly ash carbon versus mercury for the first and second rows of electrostatic precipitators and baghouses at Kentucky power plants burning a variety of bituminous and subbituminous coals. Points represent sample collections in 2002 and 2007. Unpublished data from University of Kentucky Center for Applied Energy Research. b/ Detailed view of low-C/low-Hg corner of (a).

In contrast to the latter examples, Li et al. [79] neglected flue gas temperature as a determining factor in the difference between the Hg content of mechanical and ESP fly ashes in the 100-MW power plant in their study, simply noting that unburned carbon was not likely to be a source of the variation in Hg between the mechanical and ESP fly ashes.

6.3 Variation in mercury capture with carbon type and feed coal rank

As noted in section 4, there are distinct differences in carbon forms dependent upon the rank of the feed coal. This can be further complicated by the mix of coal ranks in some power plant blends; for example, the power plant in the Hower et al. [184] study now burns a Wyoming subbituminous/ Colorado and Utah high volatile bituminous blend. Further, the blending of coal with non-coal carbon sources, such as tire-derived fuel and petroleum coke complicates the relationships [192-194].

Carbons from low-rank coals have proven to be highly efficient in Hg capture [67, 80, 195-198]. Considering fly ash Hg capture as a function of the amount of carbon in the fly ash, low-carbon Bulgarian fly ashes sourced from low-rank coals had a greater tendency to capture Hg than did bituminous-sourced Kentucky fly ashes [198]. Goodarzi and Hower [80] also demonstrated that the Alberta subbituminous-coal-derived fly ashes in their study proved to have higher Hg than the bituminous-coal-derived fly ashes. With any study comparing fly ashes from coals of different ranks, other factors, such as the chemistry of the feed coals and the engineering parameters within the respective power plants, pose complications. Indeed, Goodarzi and Hower noted that the Cl content of the feed coals was a factor in the Hg capture.

Hower et al. [72] and Maroto-Valer et al. [68-69] separated carbon from the fly ash derived from the combustion of a blend of eastern Kentucky high volatile A bituminous coals. The final step in the separation process involved the isolation of high-gradient density centrifugation splits from <1.32 to 2.15-2.30 g/cm³. With the important caveat that none of the splits contained pure carbon forms, the fractions dominated by anisotropic coke had a higher BET surface area and, with one exception, the highest Hg content of the fractions investigated. The exception, with the highest Hg content of all of the eight fractions analyzed, was dominated by isotropic coke. Since the samples all contained a mix of forms, the authors could not be absolutely certain that the dominant carbon form was also the dominant form with respect to Hg capture. Hill et al. [199], using fly ashes from a variety of sources, indicated that isotropic coke had a greater propensity towards Hg capture than the other fly ash carbon forms.

Hower et al. [74] found Hg associated with Fe-rich metal inclusions, perhaps in the form of Fe spinels, within the fullerene-like carbons in the fly ash from a 220-MW power plant burning an eastern Kentucky high volatile A bituminous coal (from study by Mardon and Hower [14]). The limits of resolution of the high-resolution transmission electron microscope (HRTEM) did not allow discernment of the degree of association between the Hg and the Fe-rich metal inclusions. Fly ashes derived from coals of other ranks have not yet been investigated, so we do not know if coal rank is a factor in the development if this type of carbon. In addition, we do not know the controls of coal Fe content on the development of the metal inclusions.

Suárez-Ruiz et al [71] and Suárez-Ruiz and Parra [70] extended the relationship between Hg and fly ash carbon to anthracite-derived fly ashes. As with the studies of bituminous coals [68-69, 72], they found a positive correlation between Hg capture with both the amount of anisotropic carbon and with the BET surface area. López-Antón et al. [64] included an anthracite-derived fly ash in their investigation of fly ash carbon petrology, BET surface area, and Hg capture. BET surface area per unit carbon, expressed as the loss-on-ignition, decreased from the subbituminous-derived fly ash, through the bituminous-derived fly ash, and to the anthracite-derived fly ash. Mercury retention, expressed as mg Hg/g sorbent, was highest in the bituminous-derived fly ash and lowest in the subbituminous-derived fly ash. In contrast, Kostova and Hower [198] found subbituminous-derived fly ashes to have greater affinity for Hg than Appalachian-bituminous-derived fly ashes.

7. Mechanisms of Hg capture by carbon

Various factors have been described that determine the extent of Hg capture on carbons. The importance of the acid gases in determining the reactivity of the carbon has provided valuable clues to the mechanism of Hg chemisorption and binding on carbon. Most of the mechanistic concepts were developed from studies with activated carbons, which have an isotropic nature, whereas unburned carbons represent a variety of isotropic and anisotropic forms. But the similarities in behavior of activated carbons could result from the same mechanisms operating with some of the unburned carbon [140].

Young and Musich [200] produced the seminal report on acid gas effects with Hg on a carbon surface; exposing carbons to gas-phase Hg also containing HCl and SO₂. Although oxygen-containing groups on the carbon surface were suggested as being responsible for Hg sorption, no one has been able to correlate Hg capacity or reactivity with oxygen functionalities [201]. Since HCl is the exclusive form of halogen in flue gas at least at temperatures at which Hg capture occurs, we need to understand its role in the capture mechanism. Since HCl is not an oxidant, and Hg is oxidized on carbon in the absence of HCl [202], it is clear that neither HCl nor Cl₂ derived from HCl is responsible for oxidation of Hg on the surface. The key to understanding the role of HCl was the finding that sorption experiments conducted in low amounts of or no HCl experienced an induction effect, an initial period of time where the reactivity to Hg oxidation develops [203]. The oxidation is promoted not only by the addition of HCl, but any other acid, including small amounts of sulfuric acid [202-203]. A detailed oxidation mechanism was hypothesized to explain this acid promotion effect [203]. The mechanism involves addition of a proton or Lewis acid to a zig-zag carbon edge site to form a carbenium ion, which represents the oxidation site for the Hg or other gas components, such as SO₂. Later, the mechanism was somewhat modified to include the role of NO₂ in helping to promote the oxidation site [204]. The combination of HCl and NO₂ in the gas was a very effective promoter of Hg oxidation. The current thought is that multiple charges in the aromatic system avoids extensive delocalization and concentrates the charge and, thus, the oxidation potential at the protonated edge site [205-206].

Application of the acid promotion mechanism for Hg oxidation to the case of unburned carbon requires graphene edge structures on the carbon surface. This requirement is met for the isotropic carbons, but perhaps not as many sites are present as for an activated carbon. The requirement is more difficult to meet for an anisotropic carbon, since there will be more planar graphene carbons and fewer edge structures.

The effect of acid gases on the binding of the oxidized Hg to the carbon surface is also important to consider. Early studies elucidated the interactions of SO₂, NO₂, and moisture on oxidized Hg binding to activated carbon [141, 207-209]. These publications hypothesized a competition model for Hg capture, wherein the oxidation of SO₂, mainly by NO₂, and subsequent hydrolysis on the carbon surface produced sulfuric acid. Accumulation of sulfuric acid displaces the bound Hg²⁺, resulting in emission of mainly HgCl₂ from the binding sites. The extent to which this reaction occurs on unburned carbon that is collected in an ESP is unknown, but likely does contribute to the capture results under certain circumstances.

Another important point that must be considered for Hg sorption by unburned carbon is that the halogen adducts with carbon are thermodynamically stable at high temperatures. Thus, halogen (as HCl, Cl, or Cl₂) will react with carbon in a high temperature zone, so that the resulting promoted carbenium sites will be available for reaction with Hg when the particles have transited to lower temperature zones where the Hg is able to form stable compounds. This is different from the case where activated carbon is injected at a lower temperature, and thus lower surface areas or lower numbers of potential sites are compensated by an earlier promotion effect. Rapid Hg oxidation and capture for capture on unburned carbon may occur because the carbon does not have to wait to accumulate active sites. However, the early availability of oxidation sites also applies to SO₂ oxidation on the carbon surfaces. Thus, sulfuric acid forms more rapidly on the unburned carbon sites and poisoning occurs earlier, possibly explaining why attempts to demonstrate Hg capture in beds comprising materials collected from particulate control devices have not been successful.

The effects of temperature on Hg capture on carbons are generally negative, implying less capture occurs at higher temperatures. At control device temperatures, physisorption of elemental Hg does not occur at all, but chemisorption involving formation of Hg²⁺ compounds does occur, and the temperature effect is the result of a complex set of equilibrium and kinetic factors involving the oxidized Hg species. As expected, the rate of oxidation increases with temperature [210] and, initially, the oxidized Hg is bound to the carbon. Thus, for lower capture to occur as temperature increases, the desorption rate of Hg must increase with a steeper slope, so that the net Hg captured is less at higher temperature. The bound Hg is oxidized, but the released Hg may be oxidized and/or elemental, so at least two mechanisms must be considered for desorption: 1) release of oxidized Hg via an oxidative mechanism using NO₂, and 2) release of elemental Hg via hemolytic dissociation of the carbon-Hg bond [211]. Either mechanism likely occurs continuously, but relative rates are dependent on gas composition and temperature.

8. Prediction of mercury capture by carbon

Numerous studies have been published on the interpretation of data from a variety of scales ranging from bench-, pilot-, to full-scale indicating that unburned carbon (UBC) in fly ash can capture Hg in coal combustion flue gases (section 6) [16, 64, 78, 87, 178, 212]. Although not all studies are in agreement, there is a general consensus within the literature that the extent and inherent mechanism of Hg adsorption is influenced by the coal type, boiler type, temperature and quench rate, amount of UBC, surface area and porosity of UBC, surface functional groups of UBC, and flue gas composition. Several studies have directly indicated that Hg content in fly ash is directly correlated with increasing UBC content [69, 87, 173, 178, 213]. Hasset and Eylands [178] indicate that Hg capture on the inorganic components of fly ash is low compared to those of UBC, but Hower et al. [19] report that the presence of inorganic fly ash increases the complexity of the interrelationship of all of the factors that influence Hg capture.

Temperature effects influence Hg capture indirectly through UBC formation in the high temperature environment of the boiler to the quenching environment of the flue gas. Smaller particles, characteristic of those fed into a PC boiler, sinter more easily in the boiler and produce fly ash UBC particles with lower surface areas. The temperature of the boiler and residence time of the coal particles in the boiler can influence the extent of sintering further dictating the particle size and external surface area of the UBC particles produced in the fly ash. Investigations of Lu et al. [78] revealed that the Hg content of the Powder River Basin fly ashes produced from a cyclone boiler (PRB-CYC) was approximately eight times higher than Eastern Bituminous fly ashes from the PC boiler (EB-PC). However, Hower et al. [187] noted that the correlations drawn by Lu et al. [78] should be questioned based upon their small sample size of just five from cyclone boilers and four from PC boilers. Hower et al. [19] have carried out studies that indicate a wide range of fly ash carbon contents from cyclone to PC boilers, with several of the PC fly ashes having a higher UBC content than those of the cyclone boilers. Often not discussed in detail is the change in shape that the particles undergo upon exposure to high temperature. The nonuniform temperature distribution of coal upon combustion may lead to stresses within the particles leading to the curvature associated with the graphene structures [78]. Although this particular feature has not yet been correlated to Hg capture, it could be an important aspect since it is well known that carbon nanotubes exhibit higher reactivity compared to uniform graphene sheets due to the change in the electronic structure of their active sites from their curvature.

Influencing internal surface area and porosity of UBC particles is the oxidation potential of a given particle. In general, low-rank coals are easier to activate with oxidizing gases to increase porosity versus high-rank coal [214] Another aspect to consider are low-NOx burners which are known to produce high levels of UBC (sections 3 and 6.1). The peak flame temperature is reduced in these burners and this may influence the extent of sintering that the coal particles undergo and determine the extent to which UBC is formed in the boiler. Adsorption is an exothermic process; therefore, as flue gases are quenched it is expected that Hg adsorption will increase [77, 212, 215].

Fly ashes derived from lignite, subbituminous, bituminous, and anthracite have all been investigated for Hg capture, with the majority of the studies focusing on blended coals. The time-temperature profile of a coal particle in the boiler dictates the fly ash particle morphology and chemistry. In a PC boiler, approximately 70 wt. % of the coal particles are less than 75 μm and therefore burn quite easily compared to coarser particles of a cyclone boiler where approximately 100 wt. % are less than 6350 μm [216] which take a longer time to burn and form ash. A higher degree of sintering can induce the formation of graphitic carbon which has a lower surface area and number of active sites [78]. Lu et al. [78] examined fly ash from both Eastern bituminous (EB) and Powder River Basin (PRB) coal burned in PC and cyclone boilers, respectively and determined fly ashes of EB have surface areas of about 50 m²/g while those of PRB are about 150 m²/g. Külatos et al. [67] reported surface areas for UBC of bituminous-derived fly ash of between 20 and 80 m²/g and of subbituminous-derived fly ash between 230-400 m²/g. Lopez-Anton et al. [64] found that, in contrast to elemental Hg, oxidized Hg capture correlated with surface area. However, with elevated levels of Hg (0.4 μg/ml), a synthetic gas mixture to simulate the flue gas environment, and a sorbent temperature of 120 °C, their [64] studies did not take place in a realistic coal combustion flue gas environment. Even the effects of temperature on Hg capture on carbons are generally negative, that is less capture occurs at higher temperatures (section 6.2). The sorption is entirely chemisorption at control device temperatures, and the rate of oxidation increases with temperature [210]. Thus, the reason for lower capture at higher temperatures must be that the desorption rate of Hg is higher at the higher temperature. The UBC particles are at least an order of magnitude lower in surface area than the typical commercially available activated carbons, e.g., Norit Americas activated carbon for Hg removal has a surface area of 600 m²/g [217] and Calgon HGR has a surface area of 1000 m²/g.

Serre and Silcox [75] carried out a series of Hg adsorption experiments by passing Hg⁰ diluted by N₂ through both fixed- and fluidized-bed reactors containing various fly ash and activated carbon samples. As with Lopez-Anton et al.’s [64] investigations, the experiments took place with Hg levels an order of magnitude higher than typically found in coal combustion flue gases and at a fixed temperature of 121 °C. They found that the level of Hg⁰ adsorbed was directly proportional to the carbon content within each of the fly ashes tested. Additionally, Serre and Silcox [75] tested Calgon HGR and found that it adsorbed twice the amount of Hg as any of the fly ashes investigated. The fly ashes with the highest surface area tested was derived from the Clark and Huntington power plants. The coal source was not specified, but the each fly ash type had surface areas of 65.1 and 63.8 m²/g, respectively. The adsorption ability of a given fly ash particle can be associated with its BET surface area which increases from inertinite, isotropic coke, to anisotropic coke [72]. However, many of the fly ash adsorption studies carried out fail to discuss the coal from which the fly ash was derived. Future studies would benefit from deeper discussions regarding the source of the fly ash and its related formation pathway which is likely related Hg capture potential. Depending upon the type of coal combusted, UBC particles with varying morphologies and surface characteristics may result and drive the Hg adsorption and oxidation reactions.

An understanding of the chemical nature of the UBC within the fly ash is crucial to determining the mechanism by which elemental and oxidized forms of Hg are adsorbed. Lopez-Anton et al. [64] determined the characteristics of a variety of fly ash samples taken from the combustion of feed coal blends and have related these characteristics to Hg⁰ and HgCl₂ retention. Three of the four samples they investigated were from a pulverized coal combustion power plant with the fourth sample from a fluidized bed plant. The fuel burned in these different power plants, including (1) varying blends from a mixture of high-rank coals (CTA), (2) bituminous coals (CTSR), (3) subbituminous coals (CTES). The fly ash taken from the fluidized-bed plant where a blend of bituminous and coal wastes were mixed with limestone was termed CTP. Lopez-Anton et al. [64] provide detailed characterization of each of these fly ashes examined including anisotropic, isotropic, and inorganic components in addition to BET surface area and LOI. Highlights are presented here to provide an indication of the primary UBC components potentially responsible for Hg retention.

The CTA fly ash examined came from burning high-rank coals, that is, mainly anthracites with smaller quantities of semi-anthracites and bituminous coals. It was found that the unburned carbon in the fly ash was primarily anistropic, unfused, and dense particles derived from anthracitic vitrinite. The fused, porous, and vesiculated materials reported are derived from burning inertinite. The CTA fly ash was mainly found to be comprised of inorganic glassy material such as alumino silicates (65-70 vol. %). The CTSR fly ash was derived from burning bituminous coal and is similar to CTA with the exception that the anisotropic carbons are primarily fused, porous and vesiculated structures. CTES fly ash were formed when burning low-rank coal such as subbituminous or lignite. The UBC particles comprised within this fly ash type were reported to be mainly isotropic fused and porous structures, formed from vitrinite macerals present in the low-rank coal. In terms of their inorganic counterparts, CTES fly ash was comprised of alumino silicates in addition to quartz. Comprised mainly of undifferentiated anisotropic fragments, the CTP fly ash investigated by Lopez et al. [64] was found to be very different compared to the others from pulverized-coal-fired boilers. Additionally, a higher fraction of oxides were found in the CTP fly ash compared to the others.

Within this study it was found that the most favorable fly ash for Hg⁰ capture was the bituminous-derived CTSR, having the greatest number of anisotropic fused structures. Their study corroborated previous findings of Suarez-Ruiz et al. [71] and Suarez-Ruiz and Parra [70] in that no relationship was found between the total isotropic composition, total amount of mineral matter, and Hg retention. Lopez-Anton et al. [64] noted that HgCl₂ follows a similar trend to Hg⁰ in that it was captured mostly in the fly ash containing anisotropic components. Hassett and Eylands [178] found Hg adsorbed to fly ash containing inertinite and coked carbon. Lu et al. [64] investigated oxidation reactivity of Powder River Basin subbituminous-cyclone and Eastern US bituminous-pulverized-coal fly ashes motivated by the parallel relationship between the number of active sites available for oxidation by forming C-O groups and the same sites being available for potential Hg adsorption. They related the surface area with the internal pore area and number of active sites and found these to affect the oxidation rate of UBC in a similar manner to Hg adsorption. The surface area of the PRB-CYC fly ashes was three times higher than that of EB-PC in addition to a higher oxidation activity. The amount of Hg adsorbed in the PRB-CYC ashes was about an order of magnitude higher than the EB-PC ashes.

Although lower-rank coals are more easily activated by oxidizing gases and, hence, have higher internal surface areas which can potentially play a dominant role in capturing Hg, other aspects of coal’s chemistry such as Cl content can also play a major role (see also section 5). High levels of particulate-bound Hg are correlated with coals, primarily bituminous, containing Cl levels > 200-300 Cl μg/g (dry basis) [77]. Oxygen functional groups play a role in Hg adsorption by serving as both oxidation catalysts and binding sites [201, 218]. The primary Hg species observed on activated carbon surfaces is not elemental but oxidized Hg, implying that Hg is bound to the surface by an oxidizing functional group such as oxygen or a halogen (see Section 7). Studies supporting the oxidized nature of Hg on the carbon surface are based upon X-ray absorption fine structure spectroscopy [219] and desorption [207-208].

Low-rank coals, such as lignite, tend to have higher concentrations of Ca, promoting Hg oxidation [220]. This is supported by Gale et al.’s [221] experimental search aimed at characterizing the majority of full-scale U.S. coal-fired plants, isolating each factor that could potentially influence Hg capture. They found the greatest influences on Hg capture to be unburned carbon content, Cl concentrations, and also a synergistic enhancement of Hg by carbon and Ca in fly ash. Gale et al. [221] proposed that the presence of Ca enhances the amount of HgCl found on the UBC surface and in the absence of Ca, Hg⁰ adsorbs onto chlorinated-carbon sites and desorbs as oxidized Hg, resulting in minimal Hg capture.

9. Conclusions

• The most fundamental control on Hg in the fly ash carbons is the amount of Hg in the feed coal. Mercury in coal is most commonly in sulfide minerals, but can occur in a number of other minerals, as well as in organic association. In association with pyrite, Hg concentration can vary across several orders of magnitude.

• Aside from the amount of Hg in the feed coal to the power plant, the controls on Hg in fly ash are:

  1. the temperature at the ash collection hopper, the lower the temperature, the greater the chance of Hg adsorption;
  2. the amount of carbon in the fly ash, the higher the carbon, the greater the chance of Hg adsorption;
  3. the rank of the feed coal and the consequent type of carbon, with low-rank-coal-derived carbons having a greater Hg-capture tendency than bituminous-coal-derived carbons and, with the series of bituminous-derived carbons, adsorption and BET surface area increasing in the order inertinite to isotropic coke to anisotropic coke; and
  4. the overall chemistry of the feed coal, which determines the components of the flue gas, with fly ash carbons from high-Cl coals having a greater tendency to adsorb Hg.

• The potential of unburned carbon in fly ash to contribute to Hg capture is a function of the amount (noted above), the size distribution, surface area, and surface chemistry. All of these are a complex function of conditions in the boiler, which, in turn, can be impacted by factors such as the conversion to low-NOx combustion or the fineness of the coal feed, among many other factors.

• Mercury is in the gaseous elemental form in the boiler and, based on thermodynamic equilibrium, as HgCl₂ at the temperature of the inlet to the air pollution control device (APCD). Because oxidation of elemental Hg to HgCl₂ is kinetically limited in coal-fired boilers, a fraction of the Hg is oxidized at the inlet to the APCDs. Oxidized Hg has the potential to be adsorbed onto fly ash carbons, given favorable temperature in the APCD.

• Acid gases, including HCl and H₂SO₄, in the flue gas can enhance the oxidation of Hg. The combination of HCl and NO₂ in the gas is an effective promoter of Hg oxidation.

• Halogen adducts with carbon are thermodynamically stable at high temperatures, therefore, halogen (as HCl, Cl, or Cl₂) will react with carbon in a high temperature zone. The resulting promoted carbenium sites will be available for reaction with Hg at lower temperature zones where the Hg is able to form stable compounds.

• The presence of Ca, found more frequently in low-rank coals than in bituminous coals, can enhance Hg oxidation.