EFFECT OF MOISTURE ON ADSORPTION OF ELEMENTAL MERCURY BY
                            ACTIVATED CARBONS
                  Y. H. Li*, S.D. Serre, C. W. Lee, and B. K. Gullett

                       U.S. Environmental Protection Agency
                        Office of Research and Development
              National Risk Management Research Laboratory (MD-65)
                         Research Triangle Park, NC 27711
Abstract
   Experiments using activated  carbon to  capture  elemental  mercury (Hg )  were
performed using a bench-scale fixed-bed reactor and a flow reactor to determine the role
of surface moisture in Hg° adsorption.  Three activated carbon samples, which have
different pore structure and ash contents, were tested for Hg° adsorption capacity. From
both fixed-bed reactor and flow reactor experimental  results, the moisture on activated
carbon surfaces  has  been found to have a  significant effect on Hg° adsorption.  A
common effect of moisture on Hg° adsorption was observed for all three samples, despite
extreme differences in their ash contents, suggesting that this effect is not associated with
ash content.  Temperature programmed desorption (TPD) experiments performed on the
carbon samples  after the Hg° adsorption experiments indicated that chemisorption of Hg°
is a dominant process over physisorption for the moisture-containing carbon samples, and
diminished for the heat-treated moisture-free samples.  X-ray absorption fine-structure
(XAFS) spectroscopy results provide evidence that  mercury bonding  on the carbon
surfaces was associated with oxygen through a mechanism likely involving electron
transfer processes.  The aromatic resonance-stabilized structures in equilibrium with the
oxygen surface  functional groups take up electrons from the mercury atoms.  The active
sites likely are  hydrated  functional groups that result from the  hydrogen bond of the
adsorbed water (t^O).  Removal of the adsorbed H2O inhibits the equilibrium conditions
that allow surface functionalities to become electron acceptors.

Introduction
   Activated carbons are widely used in the removal of pollutants from water and gases.
Direct  injection  of an  activated  carbon into the flue gas stream has been used as a
relatively simple approach for controlling mercury emissions.   Although research has
been performed to study the adsorption of  mercury by  activated  carbons,  current
knowledge of mercury  adsorption by  activated carbons is limited, presenting intriguing
scientific  questions  related  to  the  nature  of the adsorption  (physisorption  or
chemisorption) and the effects of mercury species type in the gas and solid phases.
   Adsorption of Hg°  by activated carbons at ambient temperatures  (e.g., 23 °C) has
been suggested  to be a  combination of chemisorption and physisorption,  whereas
chemisorption is prevalent at higher temperatures; e.g., 140 °C [1]. The Hg° adsorption
capacity of  activated  carbons  decreases with increasing temperature and  decreasing

                                      1-Li

-------
mercury concentration [1-3]. Many factors have been found to influence the efficiency of
mercury removal,  including carbon  characteristics,  flue  gas composition, and  the
presence of active components (e.g., fly ash) [4].  The low concentrations of Hg° in the
flue gases (~ 10 |ig/m3) and short residence times (< 3 s) of the injected sorbent require a
large  amount of activated carbon (high carbon/Hg0  ratio) in order to achieve adequate
mercury removal [5, 6].
    Studies have shown that sulfur-impregnated activated carbons exhibit significantly
greater  Hg° adsorption capacities  compared to those of the  thermally activated carbons
[1-3, 7-9].  However, heteroatoms such as sulfur (S) and chlorine (Cl), which  are reactive
for Hg° capture, generally exist only as trace elements in thermally activated carbons.  On
the other hand, appreciable amounts of oxygen (O) are almost invariably associated with
activated carbons in the form of  surface carbon-oxygen complexes produced from the
activation process.  Carbon-oxygen surface complexes are by  far the  most important
structure influencing the surface  characteristics and adsorption behaviors of activated
carbons [10]. So far, no known research has been done to understand the role of carbon-
oxygen  surface complexes in Hg° adsorption.
    Krishnan et al. [1]  showed  that Hg° capture capacities of heat-treated  [140  °C,
flowing nitrogen (N2)] activated carbon samples were far less  than those of unheated
samples.  In addition, the same heat treatment applied to a sulfur-impregnated carbon
showed no significant reduction of Hg° capture capacity.  Based on these observations,
they suggested that the active sites causing Hg° adsorption in  the thermally  activated
carbons contained O, and these weakly bonded surface oxygen complexes were effective
in capturing Hg°.  The reduction of Hg° capture capacity for the heat-treated carbons was
due to the loss of weakly bonded surface oxygen complexes.
    However, it is  generally believed  that the major effect of heat treatment  at low
temperatures (e.g.,  25-150 °C)  on carbon surfaces is the  removal of adsorbed H2O
molecules [11].  Extensive studies on  water adsorption by activated carbons found that
H2O is  adsorbed on the carbon  surfaces by means of hydrogen bonding [12-16].  The
oxygen  complexes on carbon surfaces form primary adsorption  centers, which bind the
H2O molecules at  low relative pressures.  Adsorbed H2O  molecules can then become
secondary adsorption centers as the H2O vapor pressure increases. In general, carbon-
oxygen  surface complexes are stable below 200 °C [10], but decompose to produce H2O,
carbon dioxide (€62), and carbon monoxide (CO) when heated to higher temperatures
under an inert gas atmosphere.  From the above discussions,  it appears that the reduction
of Hg° capture resulting from low-temperature heat treatment as observed by  Krishnan et
al. [1] might have been caused by the removal of moisture from the carbon surfaces.
Infrared (IR) spectroscopic studies [17] have  shown that the change of surface species
occurred during the adsorption of H2O.  The surface oxygen species or oxidized ions,
which are influenced by TC electrons from the extensive aromatic systems in carbons, may
play an important role in surface reactions, oxidation reactions in particular [18].
   The  main objective of the  work  reported in  this paper  was to understand Hg°
adsorption by activated carbons and the effect of adsorbed H2O on Hg° adsorption.  The
thermal  stability of the adsorbed mercury was determined by temperature programmed
desorption (TPD), and X-ray absorption fine structure (XAFS) spectroscopy was used to
examine the nature of the mercury bonding on the carbon surfaces.

                                       2-Li

-------
Experimental
    Three activated carbons, a lignite-based activated carbon (FGD, Norit Americas Inc.),
a bituminous-coal-based activated carbon (BPL,  Calgon Carbon  Corporation),  and  an
activated-carbon fiber (ACN, American Kynol, Inc.) were tested for their Hg° adsorption
uptake capacity in this study. The coal-based carbon samples as received (BPL, FGD)
were ground and sieved to the desired particle size range and sealed in plastic containers.
Selected characteristics of these carbons are shown in Table 1.  BET (Brunauer-Emmett-
Teller) and DR (Dubinin-Radushkevich) surface areas were measured by N2 adsorption at
77K with relative  pressure  (P/Po) up to 0.99, and carbon dioxide (CO2) adsorption at
273K with P/P0 up to about 0.03, respectively, using an adsorption apparatus  (ASAP
2400, Micromeritics).  All samples were degassed at 300 °C under vacuum for 3 h prior
to the  measurements.  The total  pore volume was evaluated from the N2 adsorption
isotherm at P/Po = 0.99, and the micropore volume was estimated from CCb adsorption at
273K using the DR equation.   A CO2 molecular density  of  1.03 g/cm3  and a cross-
sectional area of 0.187 nm2  were assumed for estimating the DR surface area [19].  The
moisture and ash contents of the samples were measured by using a thermogravimetric
analyzer (TGA-7,  Perkin Elmer). The ACN sample is derived from a phenolic resin, in
the form of needled felt with over 92% carbon and the remainder oxygen and hydrogen
[20], with no ash.   By employing the ash-free ACN sample, the effect of ash associated
with the FGD and  BPL samples on Hg° adsorption, if it is significant, can be estimated.
Table 1 Characteristics of activated carbon samples
Characteristic
BET surface area (m2/g)
CO2 surface area (m2/g)
Total pore volume (cm'Vg)
Micropore volume (cm3/g)
Particle size/fiber diameter (jim)
Moisture (% wt)
Combustible (% wt)
Ash (% wt)
FGD
540
470
0.55
0.18
16-24
4.0
68.3
27.7
BPL
1136
976
0.58
0.37
125-177
2.5
93.7
3.8
ACN
1250
1248
0.51
0.48
10a
2.0
98.0
0.0
     From manufacturer's data sheet

   Figure 1 presents the schematic diagram of the fixed-bed reactor experimental setup.
Industrial grade N2 gas was used as a purge and Hg° carrier gas.  A quartz fixed-bed
reactor (1.27 cm, I.D.) surrounded by a temperature-controlled  electrical furnace was
used. A dryer and an O2 trap were used to remove trace H2O vapor and O2, respectively,
remaining in the purge and carrier gas  stream. To determine whether mercury is in the
elemental (Hg°) or oxidized (Hg++) forms in the  reactor outlet, an additional electrical
furnace operating at a temperature of 900 °C was added downstream of the reactor to
convert Hg++ to Hg° [21].  The Hg°-laden  gas  mixture was generated using  a Hg°-
containing permeation tube in a constant temperature system (Dynacalibrator Model 190,
VICI Metronic).  An Hg° concentration of 58±2 ppb (476 |ig/Nm3)  in N2 at a total flow
rate of 340 mL/min was generated and used for  the Hg° adsorption experiments.  An

                                      3-Li

-------
ultraviolet  (UV)  mercury  analyzer  (Model  400A,  BUCK  Scientific) was used  to
continuously measure the concentration of Hg° in the outlet stream.  The experiments of
carbon  heat  treatment for  moisture  removal, Hg° adsorption, and  desorption can  be
performed consecutively in  situ, using the experimental setup shown in Figure  1, so that
moisture uptake due  to exposure  of the carbon  sample to the  atmosphere  could  be
avoided.
                   Oxygen Trap
Dryer
                                                             Fixed Bed
                                                             Reactor
                                                              Furnace
                                                         Reduction
                                                         Furnace
                             I	i
              Figure 1. Schematic diagram of fixed-bed reactor setup
   To perform a Hg° adsorption experiment on an as-received carbon sample, the Hg°-
laden gas mixture  was first sent  through the bypass to establish the baseline  Hg°
concentration prior to adsorption. After loading the pre-weighed carbon sample (about
20-40  mg) into the reactor, distributed evenly on  a fine  frit and maintained at a
temperature of 27  °C (slightly  above room temperature),  the Hg°-laden N2 flow  was
switched from the bypass to the reactor. The adsorption experiment was performed for 2
h for convenience.  The Hg° uptake for an adsorption experiment was evaluated using the
area  between the  inlet  Hg° concentration (baseline) and the  breakthrough curve.  Hg°
adsorption experiments were also performed on the moisture-free carbon samples after
heat treatment.  To remove moisture from the as-received carbon sample (loaded into the
reactor prior to Hg° adsorption),  the sample was heated at a constant rate (8 C°/min) to
110 °C under a N2 atmosphere, then held for 30 min.  The reactor was then cooled to the
Hg° adsorption temperature (27  °C) under Ni  flow, before the  adsorption  experiment,
similar to that performed for the as-received samples, was started.
   Temperature programmed desorption (TPD) experiments were performed to measure
the thermal stability of the adsorbed mercury following the Hg°  adsorption  experiment.
The reactor, containing  the Hg°-exposed carbon sample, was purged with N2 gas at 27 °C
until the concentration of Hg° measured by the UV analyzer fell to insignificant levels (<
2 ppb).   The  sample was then heated at a  constant heating rate  (8 C°/min) to the final
temperature of 420 °C.  Blank TPD experiments using carbon samples without mercury
                                      4-Li

-------
showed that there is no interference on the mercury analyzer with the loss of CC>2 and
H2O from the carbon surfaces.  The  amount of Hg° desorbed during the TPD  run  was
estimated from the area under the outlet Hg° concentration  curve.  Preliminary  TPD
experiments showed that more  than 90% of the adsorbed Hg° was desorbed when the
temperature reached 420 °C, so this was used as the final temperature.
    A schematic of the flow reactor is shown in Figure 2.  The reactor is 310.5 cm long,
has an I.D. of 4 cm, and is constructed of quartz.  Three gas sample ports are located
along the length of the reactor, labeled SP1, SP2, and SP3, which provided residence
times ranging from  5 to  18  s at a  total flow rate  of 0.014 in /inin  [STP (standard
temperature and pressure)].  The  reactor  is heated with three Lindbcrg, 3-zone electric
furnaces in series. Once the baseline Hg° concentration is established, activated carbon is
fed into the top of  the reactor  using a  fluidized-bed  feeder.  The gas-phase  Hg°
concentration is measured at one of the sample ports by sub-isokinetically pulling a gas
sample through a  1 ^irn filter, through a reducing furnace and finally to the UV analyzer.
The detailed experimental procedures can be found elsewhere [22].  Dry carbon sample
was obtained by connecting the feeder to a vacuum system and heating at 110 °C for 1  h
prior to  the test. To test the carbon sample with varied moisture contents, water was also
added to the sample by placing the carbon in an enclosed vessel with a container  of water
for several hours, thereby increasing the amount of moisture in the carbon.  The moisture
content  of the sample was tested  by  using a  Infrared  Moisture Determination  Balance
(AD-4714, A&DCo., Ltd.).
                                     Fluidized-Bed Feeder
        Lindberg
        3-Zone
        Furnaces
   Flow Reactor
                                             Hg° Permeation Oven
         Buck UV
         Analyzer
                                   Filter    Reducing
                                            Furnace
                                      L
                                     Carbon Trap
3^"Exhaust
                                                                   Carbon
                                                                   Trap

                                                                    Pump
             Figure 2. Schematic diagram of flow reactor setup
                                      5-Li

-------
    Both as-received  and  heat treated BPL and ACN carbon samples which had been
exposed  to Hg° were selected for XAFS  spectroscopy analysis, in order  to provide
information on  the nature of mercury bonding on carbon surfaces, except for the FGD
carbon due to its high ash content.  Details of the XAFS experimental procedures can be
found elsewhere [23].

Results and Discussions
    Figure 3 shows a typical Hg° concentration curve during adsorption breakthrough,
Hg° desorption  curve during N2 purge, and TPD in the fixed bed reactor.  Figure 4
presents amounts of Hg° adsorbed as a function of time during adsorption. Tests run with
two coal-based as received (-AR) samples (BPL-AR, FGD-AR) show that the Hg° uptake
of FGD-AR is slightly larger than that of BPL-AR: the latter has larger surface area and
lower ash content.  However, the  Hg° uptake of the as-received carbon  fiber sample
(ACN-AR) is much larger than those of the coal-based carbon samples, although the total
surface area of BPL is only  10 % less than that of ACN-AR sample (see Table 1). The
very similar values of BET and CC>2 surface areas for the carbon fiber sample (see Table
1) suggest that ACN is a microporous (< 2 nm) carbon with  homogeneous distribution of
microporosity.  The CC»2  surface areas of BPL and FGD samples are lower than their
BET surface area, suggesting that  the samples contain  some mesopores.  The average
micropore width evaluated from CC»2 adsorption with the  DR  method gives a similar
value of about 1.5 nm for both BPL and ACN samples, and  1.3 nm for the FGD sample.
From the above  results, it appears that there are no clear correlations between Hg° uptake
capacity and carbon pore  structure  characteristics, such  as surface area and porosity, as
well as ash content.
                  25
                        50
                               75     100    125

                                  Time (min)
                                                 150
                                                       175
       Figure 3. Mercury adsorption and desorption profile in fixed-bed reactor

                                     6-Li

-------
    Figure 4  shows that the Hg° uptake of carbon samples was  drastically reduced to
similar low values after low temperature (110 °C) heat treatment. Such heat treatment
effects were also observed by Krishnan et al. [1] for coal-based activated carbons. Heat
treatment at such low temperatures (110-140 °C) cannot change the pore structure, such
as  microporosity  and  surface area.   However,  adsorbed H2O  molecules on carbon
surfaces  are removed when heated at low temperatures; e.g., 25-150 °C [11], suggesting
that adsorbed H2O plays an important role in Hg° adsorption.
                250 -
                         20
40     60     80
Adsorption time (min)
                                                   100
120
   Figure 4. Mercury uptake of activated carbons as a function of adsorption time

   FGD carbon with  varied moisture contents (0, 4, and 16% H2O)  was tested in the
flow reactor.  The results shown in Figure 5 were  performed at  100 °C and a Hg°
concentration of 86 ppb. Mercury removal using dry carbon is very  low.  At a feed rate
of 9.5 g/hr, the removal is below 8%.  The as-received carbon (4% moisture) performed
better with a removal  of 10% at a feed rate of 3 g/hr and increased  to 30% at 6.4 g/hr.
This shows that removal of moisture also has a strong effect on Hg° capture in the flow
reactor condition.  Increasing the moisture  content of  the carbon increased the Hg°
removal significantly.  At a feed rate of 6.4 g/hr, the removal doubled  from 30% for the
as-received carbon to  60% for the 16% water carbon.  Note that this  effect might be a
result of evaporation of H2O from the carbon surface lowering the temperature of the
particle, subsequently causing higher removal.  In the flue gas, however, the direction and
driving force for H2O evaporation will depend on the moisture content of carbon, as well

                                      7-Li

-------
as the temperature and H^O concentration. It can be found from previous study [4] that
the presence of HiO in a simulated flue gas does not seem to affect Hg° uptake capacity.
Tests were conducted  to determine the moisture loss from the particle as  it falls through
the reactor with a 10 s particle residence time.  Approximately, 50% of the moisture was
lost. The extent to which moisture content of activated carbon affects Hg° removal, the
evaporation of HiO from the moisture-containing carbon, and the temperature profile of
the particle, needs further investigation.  Nevertheless,  the result from the flow reactor
shows  that the  moisture content of activated carbon is  also an important factor in Hg°
capture.
70 -,	


60


50 -
  o
  E
  £30-
20 -


10 -
    0
                                               16% Moisture
                                              4% Moisture
                                                    0% Moisture
     0
8
                                                               10
            1234567
                        Activated carbon feed rate (g/hr)
    Figure 5. Mercury removal by FGD carbon with different moisture contents in
    the How reactor at 100 °C and 86 ppb Hg°

    To better understand the mechanism of Hg° adsorption by activated carbon and how
carbon surface moisture could affect Hg capture  significantly, it would be important to
know the  nature of mercury bonding on the carbon surface.  As shown in Figure 3 from
the fixed  bed  reactor  Ni  purge   and TPD  results,  the  chemisorbed  Hg°  can  be
differentiated from physisorbed Hg°.   The  physisorbed  Hg° on the carbon surfaces is
evolved during the purge period when the reactor gas is switched from the Hg°-Iaden N2
at the end of the adsorption experiment to pure N2 (27 °C). The mercury evolved during
the subsequent  TPD run is referred to as  chemisorbed  Hg°, since it  is stable at the
adsorption temperature under the inert gas atmosphere (N2) and decomposes into  gas
species as the temperature rises. The amounts of  Hg° desorbed were calculated from the
area under  the outlet  Hg°  concentration,  and the results  are  presented in Table 2,
expressed as a percentage of the Hg° adsorbed that was measured during the adsorption
experiment  prior to the N2 purge.   Similar amounts of Hg° were measured between
duplicate runs with and without operating the reduction furnace, which suggested that the
mercury  evolved during desorption  was Hg° and not an  oxidized  form.  The mass
                                      8-Li

-------
balances between the adsorbed and desorbed Hg° in all cases were greater than 90%, and
most of the Hg° adsorbed by the carbons was recovered by heating the samples to 420 °C.
    For the as-received samples, more than 86% of the adsorbed Hg° was desorbed during
the TPD runs at higher temperatures versus > 54%  for their heat-treated counterparts.
Hg° evolved from the coal-based, as received samples (BPL-AR, FGD-AR) during the N2
purge accounts for only a small fraction of the total  Hg° adsorbed.   Adsorption and
desorption processes on the carbon surface are known to be governed by a distribution of
adsorption energies [24].  Chemical  bonds formed between the adsorbed molecules and
the adsorption sites of an activated carbon are  often  very energetic.  Since the  TPD
experiments show that Hg° is  evolved over a wide temperature range, Hg° bonds on  the
carbon  surface probably involve different  site  types, which have  varying  bonding
energies, implying that the nature of the carbon surface active  sites  for bonding Hg°
depends  on  the sample characteristics.  For the  as-received samples,  chemisorption of
Hg° appears to be the predominant process.  Note from Table 2 that  a relatively  large
amount of Hg° for the  ACN-AR sample was evolved  during N2 purge. As described,
ACN  is  a microporous carbon with homogeneous distribution of  microporosity.  The
physisorbed Hg° of ACN-AR  in this case could be explained by its microporosity.  The
reduction of Hg° capture for the heat-treated samples is caused  by  the reduction of
chemisorbed Hg°. Thus, it could be deduced that  the removal of t^O from the carbon
surfaces  by  low-temperature treatment  eliminates the  active sites  that can chemically
bond Hg°.
Table 2 Mercury uptake from adsorption and recovery from desorption
Hg adsorption at 27°C Hg desorption
Sample Hg captured (ng/g)a N2 purge (%) TPD (%)
ACN-AR
ACN- 110
BPL-AR
BPL-110
FGD-AR
FGD-110
236
34
83
20
95
22
12
42
5
20
3
42
86
54
93
71
96
56
Hg recovery
(%)
98
96
98
91
99
98
a Amount of mercury adsorbed for 100 min on dry-carbon basis

   Figure 6 shows the Mercury Lm-edge XANES (X-ray absorption near-edge structure)
spectra of the five carbon samples, and the inflection point difference (IPD, eV) derived
from the spectra. Sufficiently strong spectra have permitted the derivative analysis to be
performed in order to obtain the IPD values,  except for ACN-110, which could be due to
its low mercury concentration.  Huggins et al. [23] showed that the IPD values could be
used as  an effective parameter to identify the nature of the mercury bonding on carbon
surfaces.  It has been shown that the mercury surface compounds  with the smallest and
most ionic anions (O2~, O") have the largest values of IPD (e.g., > 9.0), whereas those
with the largest and  more covalent anions have the smallest IPD values [23].  As shown
in Figure 6, the IPD  values derived from the  XANES spectra are larger than 9.0. One of
the as-received ACN samples gives an IPD value of 9.8, close to that of mercury acetate
(10.6) [23].  Results of the XAFS analysis indicate that the Hg-O bond is significant for

                                      9-Li

-------
the as-received carbon samples, which had been exposed to Hg°.  The XAFS results are
consistent with the TPD results in that most of the adsorbed Hg° was strongly bound on
carbon surfaces and evolved at higher temperatures in all cases.
    The above XANES results suggest that carbon-bound oxygen on the surface captures
Hg° during the adsorption  experiments.  Different types of oxygen surface groups are
believed to exist on activated carbon surfaces.  Depending  on their history of formation
and activation  temperature, they could be carboxyl, lactone, phenolic,  and carbonyl
groups  [25].   Electron transfer  processes  are  likely  to  be  involved  during  the
chemisorption of Hg°.   Abundant  evidence  from the literature  shows  that different
oxygen groups can participate in the electron transfer  processes on carbon surfaces [18].
Extended n bonding in the extensive aromatic network of carbons permits electron clouds
and charges to be highly delocalized.  For example, the resonance-stabilized structures of
the aromatic network in equilibration with the surface functional groups (e.g., quinonoid
complexes) could take up electrons resulting in the functional groups' becoming anions.
                                                  80    100
                                   Energy (eV)
       Figure 6. Mercury Lm-edge XANES spectra of activated carbons

   Based on the above-proposed mechanism for Hg° oxidation, the observations that the
desorbed mercury from the TPD experiments was found to be Hg°, and that the adsorbed
H2O has a significant effect on  Hg° adsorption, could be explained as follows: when
thermal energy is applied to a chemisorbed Hg°, such as a mercury-carbonyl complex, it
would be favorable for the electron(s) to return to the mercury rather than breaking the C-
O bonds.  Although certain oxygen complexes would  also decompose to  produce CO2

                                      10-Li

-------
and  H2O at low temperatures  (e.g., <  420 °C), it is not clear if Hg° is bonded  to
complexes such as carboxyl and lactone groups. However, as described by Leon y Leon
and  Radovic [18], before the oxidation reactions can proceed they all have specific
experimental requirements that must be met such as pH, oxygen exposure, and kinds  of
surface functional groups.  The  fact that the presence of surface moisture promotes Hg°
bonding could imply that surface groups for such bonding would probably involve certain
hydrated surface functional groups, which arise  from the hydrogen bond of adsorbed H2O
[26].  Simply  removing the  adsorbed H2O at low temperature (< 150 °C) as  moisture
would inhibit the equilibrium conditions favoring the surface structures to be the electron
acceptors.  One or more types of oxygen functional groups could act as electron acceptors
if specific conditions are met. It is not known  what particular surface functional groups
are responsible for Hg° bonding from  the current investigation.   Further research  is
needed to characterize the carbon oxygen surface groups that act as electron acceptors  to
capture Hg°.

Conclusions
   The moisture on activated carbon surfaces has been found to have a significant effect
on Hg° adsorption. The results from this study show that the Hg° adsorption capacities  of
activated carbon samples were  drastically reduced after removing  their moisture. By
comparing the results of an  ash-free  activated  carbon sample (ACN) to those of coal-
based carbons, it is concluded  that mineral  matter or metal  oxides in the coal-based
carbons do not play a role in this effect.  It also appeared that no correlation could be
established between the  carbon  pore structure and Hg° adsorption capacity.  Results  of
the TPD  experiments show that  chemisorption of Hg° is a dominant process during Hg°
adsorption for the moisture-containing  carbon  samples.   XAFS analysis  provides
evidence that Hg° bonding on the carbon surfaces is associated with oxygen.
   Results from both mercury desorption (TPD) and XAFS analyses suggest that surface
oxygen complexes provide the active centers for mercury bonding.  The mechanism  of
mercury  bonding on the carbon  surfaces likely involves an electron transfer process,  in
which the aromatic, resonance-stabilized structures in equilibrium  with  the  surface
functional groups would probably take up electrons from the Hg°. The observation that
the presence of surface moisture promotes mercury bonding  suggested that the oxygen
surface groups capable of mercury bonding  might involve certain  hydrated functional
groups, which  could result from the hydrogen bond of the adsorbed water.  Removal  of
the adsorbed  H2O  at  low  temperature  (<  150  °C) as  moisture would inhibit the
equilibrium conditions that allow surface  functionalities to become electron acceptors.

Acknowledgement
   This research was supported in part by an appointment to the Postdoctoral Research
Program  at EPA's National Risk Management Research Laboratory  administered by the
Oak   Ridge Institute for  Science  and  Education   through  interagency  agreement
DW89938167  between the U.S.  Department of Energy  and the U.S. Environmental
Protection Agency. The authors appreciate the assistance of Prof. Frank Huggins at the
University of Kentucky in providing the XAFS analysis. Appreciation is also extended  to
Dr. Carlos A.  Leon y Leon for  the discussions and suggestions  on  this subject through

                                      11-Li

-------
personal communication, and Joe Hayes  of American Kynol,  Inc. for  providing the
activated carbon fibers.

References
1. Krishnan SV, Gullett BK, Jozewicz W. Environ Sci Technol 1994; 28: 1506-1512.
2. Huang HS, Wu JM, Livengood CD. Hazardous Waste & Hazardous Materials 1996;
   13:107-119.
3. Korpiel JA, Vidic RD. Environ Sci Technol 1997; 31: 2319-2325.
4. Carey TR, Hargrove OW Jr, Richardson CF, Chang R, Meserole FB. Journal of the
   Air & Waste Management Association.  1998; 48: 1166-1174.
5. Chang R, Offen GR. Power Eng 1995; 99: 51-57.
6. Rostam-Abadi M, Chen  SG, Hsi HC, Rood M, Chang  R,  Carey  T, Hargrove B,
   Richardson C, Rosenhoover W,  Meserol  F.  Presented at the EPRI-DOE-EPA
   Combined Utility Air Pollutant Control Symposium, Washington, DC, 1997.
7. Sinha RK, Walker PL Jr. Carbon 1972; 10: 754-756.
8. Otani Y, Emi H, Kanaoka C, Uchijima I, Nishino H. Environ Sci Technol 1988; 22:
   708-711.
9. Vidic DR, McLaughlin JB. Journal of the Air &  Waste Management Association.
   1996; 46: 241-250.
10. Bansal RC,  Donnet  JB,  Stoecki, F. Active Carbon. New  York, NY,  and Basel,
   Switzerland: Marcel Dekker,  1988.
11. Bacon R, Tang MM. Carbon 1964; 2: 221.
12. Dubinin MM. In Walker PL Jr, editor. Chemistry and Physics  of Carbon, Vol. 2, New
   York, NY: Marcel Dekker, 1966: 51-120.
13. Puri BR.  In Walker PL Jr, editor. Chemistry and  Physics of Carbon, Vol. 6, New
   York, NY: Marcel Dekker, 1970: 191-282.
14. Barton SS, Evans MJB, Holland J, Koresh LE. Carbon 1984; 22: 265.
15. Youssef AM, Ghazy TM,  Nabarawy TE. Carbon 1982; 20: 113.
16. Bansal RC, Dhami TL, Parkash SS. Carbon 1978; 16: 389.
17. Zawadzki J. Carbon 1987; 25 (3): 431-436, 25 (4): 495-502.
18. Leon y Leon CA, Radovic LR. In  Thrower PA, editor,  Chemistry and Physics of
   Carbon, Vol. 24, New York: Marcel Dekker, 1994: 213-310.
19. Rodriguez-Reinoso F, Garrido J, Martin-Martinez JM,  Molina-Sabio M, Torregrosa
   R. Carbon 1989; 27: 23-32.
20. Hayes JS Jr, Novoloid fibers, Kirk-Othmer: Encyclopedia of Chemical Technology.
   Vol. 16, 3rd Edition, 1981: 125-138.
21. Krishnan  SV, Gullett BK, Jozewicz W. Presented at the Solid Waste Management
   Thermal    Treatment    &    Waste-To-Energy    Technologies    Conference,
   USEPA/ORD/AEERL and A&WMA, Washington, DC, 1995.
22. Serre  SD,  Gullett  BK,  Ghorishi  SB.  In  Proceedings  of the Air and  Waste
   Management Association's 93rd Annual Meeting, Salt Lake City, UT, 2000.
23. Huggins FE, Huffman GP, Dunham GE, Senior CL. Energy & Fuels 1999; 13:  114-
   121.
24. Tremblay G, Vastola FJ, Walker PL Jr, Carbon 1978; 16: 35-39.
25. Boehm HP. Advan. Catalysis 1966; 16: 179-274.

                                    12-Li

-------
26. Zawadzki J. In Thrower PA, editor. Chemistry and Physics of Carbon, Vol. 21, New
   York, NY: Marcel Dekker, 1989: 147-380.
                                   13-Li

-------
  N RM RL- RT P- P- 543
      TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
   EPA/600/A-00/104
                            2.
                                                       3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
 Effect of Moisture on Adsorption of Elemental
  Mercury by Activated Carbons
                                                       5. REPORT DATE
                             6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
 Y.H.Li*, S. D. Serre, C. W. Lee, andB.K. Gullett
                                                       8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
 See Block 12
                                                       10. PROGRAM ELEMENT NO.
                                                       11. CONTRACT/GRANT NO.
                                                        DOE LAG DW89938167  (Li)
12. SPONSORING AGENCY NAME AND ADDRESS
 EPA, Office of Research and Development
 Air Pollution Prevention and Control Division
 Research Triangle Park, NC  27711
                             13. TYPE OF REPORT AND PERIOD COVERED
                              Published paper; 1/99 - 7/00
                             14. SPONSORING AGENCY CODE
                               EPA/600/13
15.SUPPLEMENTARY NOTES APpCD project officer is Brian K. Gullett,  Mail Drop 65, 919/
 541-1534. For presentation at Conference on Air Quality II, Tysons  Corner, McLean,
 VA,  9/19-21/00.  (*) Li is ORISE  Post-doctoral fellow.
16. ABSTRACT
               paper discusses experiments using activated carbon to capture elemen-
 tal mercury (Hgo),  and a bench-scale fixed-bed reactor and a flow reactor to deter-
 mine the role of surface moisture in Hgo adsorption.  Three activated- carbon sam-
 ples, with different  pore structure and ash content, were  tested for Hgo adsorption
 capacity.  From both fixed-bed and flow reactor experimental results, the moisture
 on activated- carbon surfaces has been found to have a significant effect  on Hgo ad-
 sorption.  A common effect of moisture on Hgo adsorption was observed for all three
 samples,  despite  extreme differences in their ash content, suggesting that this ef-
 fect  is not associated with ash content.  Temperature- programmed desorption (TPD)
 experiments performed on the carbon samples after the Hgo  adsorption  experiments
 indicated  that chemisorption of Hgo is a dominant process over  physisorption for
 the moisture- containing carbon samples,  and diminished for the heat-treated mois-
 ture-free samples.  X-ray absorption fine- structure (XAFS) spectroscopy results
 provide evidence that mercury bonding on the carbon  surfaces was associated with
 oxygen through a mechanism  likely involving electron transfer processes. The aro-
 matic resonance- stabilized structures in equilibrium  with the oxygen surface funct-
 ional groups take  up electrons from the mercury atoms.
17.
                              KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                           b.lDENTlFIERS/OPEN ENDED TERMS
                                            COSATI Field/Group
 Pollution
 Mercury (Metal)
 Activated Carbon
 Adsorption
 Moisture
                 Pollution Control
                 Stationary Sources
13 B
07B
11G
14G
07D
18. DISTRIBUTION STATEMENT
 Release to Public
                                           19. SECURITY CLASS (This Report)
                                           Unclassified
                                          21. NO. OF PAGES
                                               13
                 2O. SECURITY CLASS (This page)
                 Unclassified
                                          22. PRICE
EPA Form 2220-1 (9-73)

-------