SR-977
IDENTIFICATION OF MERCURY SPECIES IN UNBURNED CARBON
FROM PULVERIZED COAL COMBUSTION
AWMA Paper 99-72
Frank E. Huggins, Nora Yap and Gerald P. Huffman
University of Kentucky
Lexington, KY
Constance L. Senior
20 New England Business Center
Physical Sciences, Inc.
Andover, MA
Presented to the Air and Waste Management Association
1999 Annual Meeting and Exhibition
Copyright © 1999
Identification of Mercury Species in Unburned Carbon from
Pulverized Coal Combustion
99-72
Frank E. Huggins, Nora Yap and Gerald P. Huffman
University of Kentucky, Lexington, KY
Constance L. Senior
Physical Sciences, Inc., Andover, MA
ABSTRACT
Coal-fired power plants are a significant source of atmospheric mercury. In coal combustion
flue gases, mercury can be adsorbed on fly ash, reducing the emission of mercury to the
environment since fly ash is efficiently removed from the flue gas. The mechanism for
adsorption on fly ash is not known. New analytical methods are being developed to identify the
mercury species adsorbed on fly ash as well as on sorbent materials. In this paper, we present
some results of X-Ray Absorption Fine Structure Spectroscopy (XAFS) measurements of
unburned coal (char) exposed to mercury compounds in a simulated flue gas. When the gaseous
mercury species was HgCl2, the adsorbed form on char appeared to be a mercury-chlorine
compound. When elemental mercury was the gaseous form, the form of mercury adsorbed on
the char showed a dependence on the char composition. Preliminary analysis of the data
indicates that chars exposed to elemental mercury may contain a mixture of elemental mercury
and mercury chloride, depending on the amount of chlorine in the coal.
INTRODUCTION
A recent report by the U.S. Environmental Protection Agency (EPA) on emission of hazardous
air pollutants by electric utilities predicted that emissions of air toxics from coal-fired utilities
would increase by 10 to 30% by the year 2010.1 Mercury from coal-fired utilities was identified
as the hazardous air pollutant of greatest potential public health concern.
Anthropogenic emissions of mercury account for 10 to 30% of the world-wide emissions of
mercury.2 EPA has estimated that during the period 1994-1995 annual emissions of mercury
from human activities in the United States were 159 tons.1 Approximately 87% of these
emissions were from combustion sources. Coal-fired utilities in the U.S. were estimated to emit
51 tons of mercury per year into the air during this period. Based on these data, it is possible
that mercury emissions from coal-fired power plants will be regulated under the Clean Air Act
Amendments of 1990. Should regulations be imposed on mercury emissions from coal-fired
power plants, a sound understanding of the fundamental principles controlling the formation and
partitioning of mercury during coal combustion will be needed to aid in developing methods for
controlling mercury emissions.
A fraction of the mercury in flue gases is converted to adsorbed species on particulate matter.
Mercury measurements made at power plants have shown that as much as 20% of the total
1
mercury entering the plant can be removed in the ESP ash.3,4 The mechanism for mercury
capture on particulate matter in coal combustion flue gas seems to be chemical or physical
adsorption on the fly ash and depends on the oxidation state of the mercury and the composition
of the fly ash. Gas-phase oxidized mercury is readily captured by activated carbon, while
elemental mercury has a much lower affinity for carbon. The surface of the carbon is crucial to
mercury sorption; adding sulfur or iodine can dramatically increase the capacity of activated
carbon for elemental mercury. 5-8 Residual carbon from coal combustion is not the same as
activated carbon. The pore structure, surface properties, and inorganic content may be strikingly
different. There is evidence for the adsorption of mercury on coal fly ash even in the absence of
significant unburned carbon,9,10 although the specific species which are adsorbed is not known.
In a previous study,11 chars generated from three coals were used as sorbent material for both
elemental mercury and mercuric chloride in a laboratory fixed bed reactor. The temperatures of
the source as well as that of the char sorbent were carefully controlled in the range of 343 to
433 K (70-160oC). These three chars showed consistently higher mercury capture in the case of
mercury chloride (over elemental mercury) as the mercury source. Elemental mercury appeared
to react chemically with sulfur in the char, particularly with organic sulfur at lower gas-phase
mercury concentrations. For mercury chloride, a physical adsorption process seemed to be
indicated based on the correlation with char surface area. The organic sulfur content of the char
appeared to be the better predictor of the affinity of the char for elemental mercury, while the
char surface area appeared to be the better indicator of affinity of the char for mercuric chloride.
The chars that were exposed to mercury were analyzed using X-Ray Absorption Fine Structure
(XAFS) spectroscopy. XAFS spectroscopy is basically a measurement of the variation (or
fine structure) of the X-ray absorption coefficient with energy associated with one of the
characteristic absorption edges of the absorbing element. Analysis of XAFS spectra provides
information on the bonding of the element in question and can be used to distinguish different
compounds from one another. In this work, XAFS is used to determine the speciation of
mercury adsorbed on chars and various sorbent materials. An understanding of the speciation of
mercury on solids will help clarify the mechanisms for mercury adsorption.
EXPERIMENTAL METHODS
Preparation of Chars
Three coals were chosen for this study, two bituminous coals from the Pittsburgh and Illinois
No.6 seams and a sub-bituminous coal from the Powder River Basin, Wyodak seam. The ash
compositions of the bituminous coals are fairly typical of American bituminous coals,
containing primarily silica, alumina, and iron oxide. The ash from the Wyodak sub-bituminous
contains significantly less iron and more calcium than the bituminous coals. Results of the
analysis of the forms of sulfur in the coals are presented in Table 1. The data indicate that the
Pittsburgh and Illinois No. 6 coals contain significant amounts of both pyritic and organic
sulfur. The Wyodak coal contains primarily organic sulfur.
Chars were made by reacting the coals in the PSI Entrained Flow Reactor (EFR) as described in
Reference 11. The ash content of the chars and surface area of the chars are summarized in
Table 2. The forms of sulfur in the chars were determined using XAFS spectroscopy. Sulfur
2
XANES spectra were analyzed according to least-squares fitting procedure developed for sulfur
in coals and chars.12 Estimates of the percentage of sulfur in different forms in the chars are
given in Table 3. The bituminous chars have a mixture of organic sulfur, elemental sulfur, and
pyrrhotite (Fe1-x S). The latter two forms of sulfur are the result of decomposition of pyrite
during devoloatilization. The Wyodak char has primarily organic sulfur with some sulfate and
elemental sulfur.
Sorption experiments were performed using a small fixed bed reactor.11 Mercury or mercury
chloride sources were placed in an oven. A small fixed bed of char was placed in another oven.
The fixed bed was heated to either 343 K or 443 K. The temperature of the mercury source was
controlled to deliver mercury concentrations in the gas phase from 0.4 ppmw to 60 ppmw. All
the experiments were done under oxidizing conditions where the gas composition was 15% CO2,
3% O2 , 2% H2O and balance N2. This gas composition contains the major species in coal
combustion flue gas, but none of the minor species such as SO2, NO, NO2, HCl or Cl2. Recent
work has demonstrated that these species affect the interaction of gaseous mercury compounds
and activated carbon.13 Therefore, the results of this study may not be directly applicable to the
behavior of unburned coal in power plants. Future experiments will employ more realistic flue
gas compositions.
Chars were exposed to the simulated flue gas for times ranging from 5 to 25 hours. At the end
of the experiment, the concentration of mercury in the char was determined by cold vapor
atomic absorption. The data were used to determine the sorption ability and capacity of sorbent.
Char samples were analyzed by XAFS for forms of mercury.
XAFS Spectroscopy
Mercury XAFS spectroscopy was conducted principally at beam-line IV-3 at the Stanford
Synchrotron Radiation Laboratory (SSRL), Palo Alto, CA. Virtually identical experimental
practice was carried out at both synchrotrons. The mercury LIII edge at 12,284 eV was used for
the absorption experiments and all measurements on char samples were carried out in fluorescence geometry. The mercury contents of the chars varied from as much as 8.5 wt% to as
little as 8 parts per million (ppm) and a Lytle-type detector13 was used for the more concentrated
samples, whereas a 13-element germanium detector14 was used to record the spectra from less
concentrated samples. A 6µ gallium filter and Soller slits15 were used to enhance the
signal/noise ratio with both types of detector. For the most dilute samples (< 500 ppm Hg),
multiple scanning was also used to improve the signal/noise ratio. Additionally, in the course of
this investigation, the spectra of a number of mercury compounds were measured. These spectra
were measured in absorption geometry using pressed pellets of the compound in a plastic
medium or thin smear mounts on tape. A droplet of elemental mercury pressed and sealed
between two thin plastic sheets served as the calibration standard. The single inflection point in
the XAFS spectrum of elemental mercury was taken to be the zero-point of energy for
calibrating the XANES spectra shown in this investigation. Where possible, this spectrum was
obtained simultaneously in an absorption experiment after the main experiment, so that each
XAFS spectrum would be individually calibrated.
Analysis of the calibrated spectral data consisted of dividing the XAFS spectrum into separate
3
X-ray absorption near-edge structure (XANES) and extended X-ray absorption fine structure
(EXAFS) regions. Analysis of the XANES region consisted of smoothing and differentiating the
spectrum to obtain (Figure 1) the d2Abs/(dE)2 spectrum. The XANES spectrum pictured at the
top of Figure 1 exhibits two inflection points. The location of these inflection points appeared to
vary systematcially with the character of the mercury bond in the sample. In order to determine
more precisely the spacing or inflection point difference (IPD), the second derivative of the
spectrum was calculated as shown in the bottom section of Figure 1. The IPD is the spacing
between the inflection points as indicated by the arrows in the figure.
The isolated EXAFS region was analyzed in the standard manner16,17 by fitting a five-region
cubic spline to the data to obtain the chi spectrum and then converting from real space to
reciprocal (k) space. A Fourier transform was then applied to the k3-weighted chi vs. k spectrum
to obtain the radial structure function (RSF). The RSF is a one-dimensional representation of
the local structure around the absorbing mercury atom.
RESULTS AND DISCUSSION
XAFS data for various mercury compounds are summarized in Table 4. Included in this table
are values for IPD from each XANES spectrum and the position of the major peak in the corresponding RSF. As can be seen from this table the mercury compounds with the smallest and
most ionic anions (O2-, -O-) have the largest values of IPD, whereas those with the largest anions
(I-) have the smallest values of IPD. Hence, it appears that the IPD value correlates with
differences in the Hg-X bond ionicity and/or length. The value of IPD appears not to depend on
the oxidation state of the mercury because both Hg(I) and Hg(II) sulfates and chlorides have
effectively the same IPD value. It is well known that features in the XANES region can reflect
bond lengths and other structural details, and, as can be readily seen in Table 4, there appears to
be an inverse correlation between the separation of the mercury LIII edge inflection points and
the RSF peak position, as might be expected based on previous investigations of systematics in
XANES spectra.18,19 Similar trends can be seen in other compilations of mercury compound
data.20
Values for the IPD determined from the XANES spectra of various chars after exposure to
mercury, as Hg0 or as HgCl2, in simulated flue gases (SFG) are displayed in Table 5. When the
IPD values for the chars are compared with those for standards (Figure 2), we find that most of
the char samples have values comparable to chlorides and sulfides. Experiments carried out
with mercuric chloride (HgCl2) in the vapor phase give rise to higher IPD values than those
carried out with Hg0 vapor. In chars exposed to elemental mercury (Hg0), there is clearly a trend
towards lower values for the experiments carried out with the Wyodak char than with the Illinois
No.6 and Pittsburgh chars. Such differences may reflect the different chlorine contents of the
coals. The Wyodak coal is known to have a much lower content of chlorine than the other two
coals.
The data listed in Table 5 and shown in Fig. 2 make a further interesting point in that they
appear not to show a bimodal distribution around the values for HgS and HgCl2. The XANES
spectra for Hg in the chars do not resemble crystalline mercuric chloride or mercuric sulfide,
even though the IPD values for the chars are very similar to that for Hg chloride or Hg sulfide.
4
Such trends reflect either a mixture of distinct mercury species or possibly a single amorphous
mercury species of variable and mixed composition.
We have explored one possible reason for this variation in terms of a mixture of chemisorption,
involving formation of a HgCl2 or HgS species on the char surface, and physisorption of
elemental mercury. This was done by simulation of a suite of Hg XANES spectra from endmember standard spectra for metallic (liquid) mercury and either mercuric chloride or mercuric
sulfide. Examples of such simulated spectra are shown in Figure 3 for Hg-HgCl2. Also shown
are the derivative spectra for the same suite of XANES spectra. It can be seen that the peakheight ratio of the derivative spectra changes systematically with the amount of Hg in the
mixture. Hence, the combination of the IPD parameter and the derivative peak-height ratio
provides a two-dimensional plot (Figure 4) that appears to have the potential to discriminate
between chemisorption and physisorption.
Also shown in Figure 4 are the Hg-XANES data obtained for the chars and it can be seen that
these data are parallel but displaced relative to the IPD/Peak-height Ratio trend for Hg-HgCl2.
Hence, a possible explanation for the Hg XANES data trend shown by the char samples is that it
is due to the occurrence of chemisorption of Hg as HgCl2 and physisorption of metallic Hg as
competing processes. Further, the relative fraction of the chemisorption process is likely to
reflect the Cl/Hg ratio in the system, as we speculated previously. Further analysis of these
trends will be undertaken and new spectral data will be obtained for the standards as well as for
new samples to be prepared.
CONCLUSIONS
The results of this investigation have shown that the difference in energy between the two
principal inflection points (IPD) on the mercury LIII absorption edge measured in XAFS spectra
is a sensitive indicator of the structure and/or chemistry of the mercury adsorbed onto unburned
carbon and sorbent materials. Unburned coal (char) samples were exposed to mercury compounds in a simulated flue gas. When the chars were exposed to gaseous HgCl2, the adsorbed
species on char appeared to be a mercury-chlorine compound. When elemental mercury was the
gaseous form, the adsorbed mercury species on the char showed a dependence on the char
composition. Preliminary analysis of the data indicate that chars exposed to elemental mercury
may contain a mixture of elemental mercury and mercury chloride, depending on the amount of
chlorine in the coal. This investigation implies rather conclusively that there are likely to be a
number of competing mechanisms for low-temperature mercury sorption by a given material in a
real flue gas. Such mechanisms will depend on the surface chemistry, the species in the flue
gas, etc. Hence, any mechanistic interpretation of mercury sorption by carbon-based materials
in terms of a specific individual mercury form is likely to be insufficient. Rather, a more general
approach to mercury sorption must be developed.
ACKNOWLEDGEMENTS
Financial support was received from the U.S. Department of Energy through Contract No. DEAC22-95PC95101 and from the Electric Power Research Institute (EPRI). Char samples were
prepared by Mr. Joseph Morency of Physical Sciences, Inc. Mercury exposure experiments were
carried out by Dr. Baochun Wu, Prof. Thomas Peterson, and Prof. Farhang Shadman at the
5
University of Arizona. We also acknowledge the U.S. Department of Energy for its support of
the synchrotron facilities at SSRL, where the XAFS measurements were made.
REFERENCES
1. Keating, M.H.; et al. Mercury Study Report to Congress, Volume I: Executive Summary,
EPA-452/R-97-003, December 1997.
2. Stein, E.D.; Cohen, Y.; Winer, A.M. “Environmental Distribution and Transformation of
Mercury Compounds,” Critical Rev. Environ. Sci. Technol., 1996, 26, 1-43.
3. DeVito, M.S.; Rosenhoover, W.A. “Flue Gas Mercury and Speciation Studies at Coal-Fired
Utilities Equipped with Wet Scrubbers,” presented at 15th International Pittsburgh Coal
Conference, Pittsburgh, PA, September 15-17, 1998.
4. Fahlke, J; Bursik, A. “Impact of the State-of-the-Art Flue Gas Cleaning on Mercury Species
Emissions from Coal-Fired Steam Generators,” Water, Air, Soil Poll. 1995, 80, 209-215.
5. Dunham, G.E.; Miller, S.J. “Evaluation of Activated Carbon for Control of Mercury from
Coal-Fired Boilers” presented at the First Joint Power and Fuel Systems Contractors
Conference, Pittsburgh, PA, July 9-11, 1996.
6. Krishnan, S.V.; Gullett, B.K.; Jozewicz, W. “Sorption of Elemental Mercury by Activated
Carbon” Env.Sci.Tech. 1994, 28, 1506-1512.
7. Vidic, R.D.; McLaughlin, J.B.; “Uptake of Elemental Mercury Vapors by Activated
Carbons” J.Air Waste Manage.Assoc. 1996, 46, 241-250.
8. Otani, Y.; Emi, H.; Kanoaka, I.; Nishiro, H. “Removal of Mercury Vapor from Air with
Sulfur-Impregnated Adsorbents,” Env.Sci.Tech. 1988, 22, 708-711.
9. Carey, T.R.; Hargrove, O.W.; Brown, T.D.; Rhudy, R.G. “Enhanced Control of Mercury in
Wet FGD Systems” Paper 96-P64B.02 presented at Air & Waste Management Association’s
89th Annual Meeting, Nashville, TN, June, 1996.
10. Broderick, T.; Haythornthwaite, S.; Bell, W.; Selegue, T.; Perry, M. “Determination of Dry
Carbon-Based Sorbent Injection for Mercury Control in Utility ESP and Baghouses” Paper
98-WP79A.09, presented at the Air & Waste Management Association 91st Annual Meeting,
San Diego, CA, June 14-18, 1998.
11. Senior, C.L.; Morency, J.R.; Huffman, G.P.; Huggins, F.E.; Shah, N.; Peterson, T.;
Shadman, F.; Wu, B. “Interactions Between Vapor-Phase Mercury and Coal Fly Ash Under
Simulated Utility Power Plant Flue Gas Conditions,” Paper 98-RA79B.04, AWMA 91st
Annual Meeting, San Diego, CA, June, 1998.
6
12. Huffman, G.P.; Mitra, S.; Huggins, F.E.; Shah, N. Energy & Fuels, 1991, 5, p.574-581.
13. Hsi, Miller . . .
14. Lytle, F.W.; Greegor, R.B.; Sandstrom, D.R.; Marques, E.C.; Wong, J.; Spiro, C.; Huffman,
G.P.; Huggins, F.E. Nucl. Instrum. Methods 226 (1984) 542-548.
15. Cramer, S.P.; Tench, O.; Yocum, N.; George, G.N. Nucl. Instrum. Methods A266 (1988)
586-591.
16. Stern, E.A.; Heald, S.M. Rev. Sci. Instrum. 50 (1979) 1579-1582.
17. Eisenberger, P.; Kincaid, B.M. Science 200 (1978) 1441-1447.
18. Lee, P.A.; Citrin, P.H.; Eisenberger, P.A.; Kincaid, B.M. Rev. Mod. Phys. 53 (1981) 769808.
19. Bianconi, A.; Fritsch, E.; Calas, G.; Petiau, J. Phys. Rev. B 32 (1985) 4292-4295.
20. Lytle, F.W.; Greegor, R.B.; Panson, A. Phys. Rev. B 37 (1988) 1550-1562.
21. Åkesson, R.; Persson, I.; Sandström, M.; Wahlgren, U. Inorg. Chem. 33 (1994) 3715-3723.
7
Table 1. Forms of Sulfur in Coals
Sulfur
(wt%, dry
basis)
Forms of
sulfur:
Pittsburgh
Illinois No. 6
Wyodak
2.12
Wt%
Percent of
in
Total
Coal
Sulfur
3.82
Percent of
Wt% in
Total
Coal
Sulfur
0.46
Percent of
Wt% in
Total
Coal
Sulfur
Sulfate
0.01
<1
0.04
1
0.02
4
Pyritic
0.91
43
1.57
41
0.03
7
Organic
1.2
57
2.21
58
0.41
89
8
Table 2. Properties of Chars for Mercury Sorption Experiments
Pittsburgh
Illinois 6
Wyodak
Carbon content, wt%
LOI
69%
39%
83%
BET surface area, m2/g
126
94
147
2.07%
4.09%
0.96%
Total sulfur, wt%
9
Table 3. Forms of Sulfur in Chars by XAFS Analysis
Pittsburgh
Illinois 6
Wyodak
Sulfate
5%
2%
20%
Pyrrhotite
20%
35%
—
Elemental
sulfur
25%
35%
17%
Organic sulfide
50%
29%
63%
10
Table 4. Hg XAFS systematics for mercury compounds
Mercury Compound
IPD, eV
RSF, Å
HgO (yellow)
13.3
HgO (red)
13.3
1.68
Hg acetate
10.6
1.72
Hg2SO4
9.6
HgSO4
9.5
Hg2Cl2
8.4
HgCl2
8.4
2.02
HgS (metacinnabar)
7.7
2.12
HgS (cinnabar)
7.5
2.05
HgCH3I
6.9
2.49
Hg diphenyl
6.8
HgI2
6.5
KHgI4
4.6
Hg (liquid)
0.0
2.52
11
Table 5. Mercury XANES systematics for char samples
Description
Parent
Sample Identity Coal
IPD
Hg Vapor
EV
HCP-5
Pitt
HgCl2
8.4
HCY-5
Wyo
HgCl2
8.3
CIL-5
Ill 6
HgCl2
8.1
HIL-25
Ill 6
Hg
8.4
HP-25
Pitt
Hg
8.1
HIL-25
Ill 6
Hg
8.0
HY-25
Wyo
Hg
7.9
HY-25
Wyo
Hg
7.6
P8-10
Pitt
Hg
8.0
HIL-25
Ill 6
Hg
8.0
HLP-25
Pitt
Hg
8.0
HHIL-25
Ill 6
Hg
7.9
HHP-25
Pitt
Hg
7.8
HHY-25
Wyo
Hg
7.7
HLY-25
Wyo
Hg
7.3
12
Normalized Absorption
1.0
0.8
0.6
0.4
0.2
0
0.6
dAbs/dE x 10
0.5
0.4
0.3
0.2
0.1
0
1.6
1.2
d2 Abs/(dE)2 x 100
0.8
0.4
0
-0.4
-0.8
-1.2
-1.6
-20
0
20
40
Energy (eV)
Figure 1. Determination of inflection point difference (IPD) from Hg XANES spectra of a
carbon-based sorbent. Arrows indicate measurement positions.
13
10
HgS O 4
IPD Value
9
HgCl 2
8
Hg S
7
6
Hg
G as Ph ase Compound
HgC l2
Figure 2. IPD values for bituminous (s) and PRB (x) chars exposed to Hg or HgCl2 in
simulated flue gas. Range of IPD values for mercury standards are indicated on right-hand side.
14
0.6
3.2
100% HgCl2
2.8
90:10
100% HgCl 2
80:20
2.4
0.5
90:10
0.4
70:30
80:20
60:40
2
1.6
1.2
50:50
70:30
40:60
60:40
30:70
50:50
20:80
40:60
0.3
0.2
10:90
30:70
100% Hg(0)
0.1
0.8
20:80
10:90
0
0.4
100% Hg(0)
0
-0.1
-20
0
20
40
60
Energy (eV)
80
100
0
20
40
Energy (eV)
60
Figure 3. Simulated XANES spectra and derivative spectra for mercuric chloride (HgCl2) and
liquid mercury (Hg0).``
15
2.2
HgS/Hg(0)
2
HgCl2/Hg(0)
1.8
Chars
HgS
1.6
1.4
1.2
HgCl 2
1
0.8
0.6
6.5
7
7.5
8
8.5
9
9.5
IPD (eV)
Figure 4. Plot of the Inflection Point Difference (IPD) against the peak-height ratio determined
from the derivative spectra shown in Figure 3. Solid lines represent data trends from simulated
mixtures of Hg0 and either HgS or HgCl2. Individual data points are experimental points
determined for the char samples.
16