1774 Energy & Fuels 2005, 19, 1774-1782

Desulfurization of Gaseous Fuels Using Activated

Carbons as Catalysts for the Selective Oxidation of
Hydrogen Sulfide
Xianxian Wu,* Viviane Schwartz, Steven H. Overbury, and
Timothy R. Armstrong
Oak Ridge National Laboratory (ORNL), Oak Ridge, Tennessee 37831
Received April 1, 2005. Revised Manuscript Received June 14, 2005

The removal of low concentrations of hydrogen sulfide (H2S) from hydrogen-rich gaseous fuels
by selective catalytic oxidation, using activated carbon as the catalyst, was studied. The capacities
of four activated carbons for reducing the H2S concentration down to the parts per billion (ppb)
level were determined to be strongly related to their microstructures and impurities, even though
their activity and selectivity were strongly dependent on the test conditions, such as reaction
temperature, O2:H2S ratio, space velocity, and length-to-diameter (L/D) ratio of the catalyst bed.
Because the side reactions that form COS and SO2 are sometimes unavoidable under real fuel
processing conditions, the complete and exclusive conversion of H2S to elemental sulfur (S) requires
that activated carbon has catalytic activities not only for the oxidation of H2S, but also for the
oxidation of COS and the reaction between H2S and SO2. Because one of the activated carbons
(sample W-22) had such catalytic functions and a relatively large capacity of trapping sulfur in
its micropores, this carbon showed a combination of excellent activity and selectivity on removing
H2S from both hydrogen and simulated reformate streams at a reaction temperature of ∼150 °C.

1. Introduction low concentrations of H2S from those hydrogen-rich

The presence of sulfur is a persistent problem associ-
ated with fossil fuels, and it must be removed before
the fuels can be used as energy sources or chemical
feedstocks. Through a typical desulfurization process,
the sulfur can be recovered via the Claus process1,2 and
the fuels may still contain a small amount of sulfur
compounds. For example, gasoline or diesel fuels contain
from a few parts per million (ppm) to several hundred
ppm sulfur, pipeline natural gas contains up to 20 ppm
sulfur, and the sulfur content in coal-derived gaseous
fuels may vary from several hundred ppm to more than
1.0%, depending on the type of coal, the sulfur content
in the feedstock, and the process.3 When these readily
available fuels are used to produce hydrogen for proton-
exchange membrane (PEM) fuel cells, the sulfur content
must be further reduced prior to entering the fuel cell,
because sulfur severely poisons fuel processing catalysts
and platinum electrocatalysts.4
Commonly, sulfur in fuels is present in the form of
hydrogen sulfide (H2S) or is chemically combined with
hydrocarbons and can be converted to H2S during fuel
processing. Several approaches can be applied to remove
* Author to whom correspondence should be addressed. Telephone:
1-865-576-6690. Fax: 1-865-576-8424. E-mail: wux@ornl.gov.
(1) Piéplu, A.; Saur, O.; Lavalley, J. C.; Legendre, O.; Nédez, C.
Catal. Rev.sSci. Eng. 1998, 40, 409.
(2) Smet, E.; Lens, P.; Van Langenhove, H. Crit. Rev. Environ. Sci.
Technol. 1998, 28, 89.
(3) Kohl, A. L.; Nielsen, R. Gas Purification, 5th Edition; Gulf
Professional Publishing Company: Houston, TX, 1997.
(4) Hidalgo-Vivas, A.; Cooper, B. H. Chapter 15: Sulfur Removal
Methods. In Handbook of Fuel CellssFundamentals, Technology and
Applications, Vol. 3: Fuel Cell Technology and Applications; Vielstich,
W., Gasteiger, H. A., Arnold, L., Eds.; Wiley: New York, 2003; p 177.
gaseous fuels.3 “Liquid-phase absorption” processes
absorb H2S via slurries of metal oxides or solutions of
basic compounds such as ammonia, alkanolamine, or
alkaline salts. “Dry sorption” processes scavenge small
quantities of sulfur in a nonregenerative manner, using
a dry sorbent. The main disadvantages of these methods
are their relatively high cost and the production of spent
sorbent (liquid or solid), which needs further chemical
transformation to be recycled or disposed of. In com-
parison, direct partial oxidation of H2S to sulfur via the
reaction
1 1
H2S + O2 f Sn + H2O (1)
2 n
in the presence of a catalyst is more attractive for
several reasons. First, the reaction has the thermody-
namic potential to remove H2S to the parts per billion
(ppb) level at temperatures of <200 °C.5 Second, the
gaseous fuel is cleaned and the sulfur is recovered (as
elemental sulfur) in a single-step process. Third, the
oxidation process can be easily integrated into the
current fuel processing plants,6 and it can be used with
diverse gas streams. The conversion of H2S is relatively
insensitive to the presence of steam, whereas steam
actually inhibits the capture of H2S by ZnO.7,8 Fourth,
(5) Gamson, B. W.; Elkins, R. H. Chem. Eng. Prog. 1953, 49, 203.
(6) Gardner, T. H.; Berry, D. A.; Lyons, K. D.; Beer, S. K.; Freed, A.
D. Fuel 2002, 81, 2157.
(7) Novochinskii, I. I.; Song, C. S.; Ma, X. L.; Liu, X. S.; Shore, L.;
Lampert, J.; Farrauto, R. J. Energy Fuels 2004, 18, 576.
(8) Novochinskii, I. I.; Song, C. S.; Ma, X. L.; Liu, X. S.; Shore, L.;
Lampert, J.; Farrauto, R. J. Energy Fuels 2004, 18, 584.

10.1021/ef0500890 CCC: $30.25 © 2005 American Chemical Society

Published on Web 07/16/2005
Desulfurization of Gaseous Fuels
the process does not consume H2 or other useful gas
components.9
Porous materials show different activity and selectiv-
ity as catalysts for H2S oxidation.10 Zeolites show good
activity at ∼300 °C, but they do not seem to be very
selective.10-14 Alumina and silica with a wide-pore
structure show both poor activity and selectivity.10
Metal oxidessespecially V2O5,15-21 TiO2,22 Fe2O3,23 and
ZrO210sshow high activity for H2S oxidation, whereas
many other oxides are either not very selective toward
sulfur or not very active.1,24-27 One of the main problems
associated with metal oxides as H2S oxidation catalysts
is sulfidation, which results in the loss of activity or even
the destruction of the catalyst.
Activated carbon is attractive, because carbon itself
is an effective catalyst for H2S oxidation,28 and the
impurities it contains may also be good catalysts. There
is, then, a possibility of tailoring carbon as a catalyst
as well as a catalyst support, to achieve different
catalytic behavior.29 Various activated carbons have
been studied as catalysts to remove H2S from a natural
gas stream,30 air,31,32 and other gas mixtures.6,10,33 The
catalytic behavior observed as a function of catalyst
exposure time may include three consecutive periods:
(i) the initial period, in which the catalyst has high
activity and virtually 100% conversion is observed; (ii)
the deactivation period, in which the catalyst becomes
loaded with sulfur and sulfur compounds break through
from the catalyst bed; and (iii) the “steady-state” period,
in which the so-called “stationary activity level”34 is
achieved. Many past studies have been focused on the
steady-state period, because it is the key to the develop-
(9) Lagas, J. A.; Borsboom, J.; Berben, P. H. Oil Gas J. 1988, 86,
68.
(10) Steijns, M.; Mars, P. Ind. Eng. Chem., Prod. Res. Dev. 1977,
16, 35.
(11) Ziolek, M.; Dudzik, Z. Zeolites 1981, 1, 117.
(12) Ziolek, M.; Dudzik, Z. React. Kinet. Catal. Lett. 1980, 14, 213.
(13) Dudzik, Z.; Ziolek, M. J. Catal. 1978, 51, 345.
(14) Verver, A. B.; Vanswaaij, W. P. M. Appl. Catal. 1985, 14, 185.
(15) Kalinkin, P. N.; Kovalenko, O. N.; Kundo, N. N. React. Kinet.
Catal. Lett. 2001, 74, 177.
(16) Kaliva, A. N.; Smith, J. W. Can. J. Chem. Eng. 1983, 61, 208.
(17) Song, M. W.; Kang, M.; Kim, K. L. React. Kinet. Catal. Lett.
2003, 78, 365.
(18) Li, K. T.; Shyu, N. S. Ind. Eng. Chem. Res. 1997, 36, 1480.
(19) Li, K. T.; Huang, M. Y.; Cheng, W. D. Ind. Eng. Chem. Res.
1996, 35, 621.
(20) Shin, M. Y.; Park, D. W.; Chung, J. S. Catal. Today 2000, 63,
405.
(21) Park, D. W.; Park, B. K.; Park, D. K.; Woo, H. C. Appl. Catal.,
A 2002, 223, 215.
(22) Chun, S. W.; Jang, J. Y.; Park, D. W.; Woo, H. C.; Chung, J. S.
Appl. Catal., B 1998, 16, 235.
(23) Li, K. T.; Yen, C. S.; Shyu, N. S. Appl. Catal., A 1997, 156,
117.
(24) Li, K. T.; Cheng, W. D. Appl. Catal., A 1996, 142, 315.
(25) Kim, B. G.; Park, D. W.; Kim, I.; Woo, H. C. Catal. Today 2003,
87, 11.
(26) Marshneva, V. I.; Mokrinskii, V. V. Kinet. Catal. 1988, 29, 854.
(27) Davydov, A. A.; Marshneva, V. I.; Shepotko, M. L. Appl. Catal.,
A 2003, 244, 93.
(28) Cariaso, O. C.; Walker, P. L., Jr. Carbon 1975, 13, 233.
(29) Radovic, L. R.; Rodriguez-Reinoso, F. Carbon materials in
catalysis. In Chemistry and Physics of Carbon; Thrower, P. A., Ed.;
Dekker: New York, 1997; p 243.
(30) Chowdhury, A. I.; Tollefson, E. L.; Ghosh, T. K. Process for
elimination of low concentrations of hydrogen sulfide in gas mixtures
by catalytic oxidation, U.S. Patent No. 6,652,826, November 25, 2003.
(31) Coskun, I.; Tollefson, E. L. Can. J. Chem. Eng. 1980, 58, 72.
(32) Mikhalovsky, S. V.; Zaitsev, Y. P. Carbon 1997, 35, 1367.
(33) Henning, K.-D.; Jüntgen, H.; Klein, J. Process for eliminating
hydrogen sulfide from gas mixtures, U.S. Patent No. 4,263,271, April
21, 1981.
(34) Steijns, M.; Derks, F.; Verloop, A.; Mars, P. J. Catal. 1976, 42,
87.
Energy & Fuels, Vol. 19, No. 5, 2005 1775
Table 1. Properties of Activated Carbon
property Centaur WV-B VA-507 W-22
particle size (mesh) 20 × 50 10 × 25 60 × 170 20 × 30a
apparent density (g/cm3) 0.59 0.30 0.42 0.50
BET surface area (m2/g) 815 1840 1020 860
total pore volume, Vt (cm3/g) 0.40 1.07 0.47 0.40
micropore volume, Vm (cm3/g) 0.33 0.49 0.39 0.35
a Size of the precursor.
ment of a continuous catalytic process.35 The initial and
deactivation periods were often short or even unidentifi-
able, and incomplete H2S conversion was usually ob-
served. Because a complete elimination of sulfur com-
pounds from gas streams is desired for H2 production,
the achievement of an extended initial period in which
H2S is completely and exclusively converted to elemen-
tal sulfur is preferred.
The length of the initial and deactivation periods may
be affected by the type and sorption capacity of the
catalyst, the mass of the catalyst used, the amount of
H2S fed, and the reaction conditions applied.36 In the
present study, the capacity of four selected activated
carbons to extend the initial period was investigated.
The catalysts were characterized by surface area mea-
surement, impurity analysis, and thermal analysis. The
activity and selectivity of these carbons under different
test conditions were determined. The possible explana-
tions for the catalytic effects are discussed.
2. Experimental Section
Three commercial activated carbons were selected: Centaur
20 × 50 (Calgon Carbon, bituminous coal-based, physical
activation), WV-B (Westvaco, wood-based, chemical activation),
VA-507 (PICA, coconut shell-based, physical activation). A
laboratory-synthesized activated carbon (W-22) that used a
cellulosic precursor37 was also studied. Nitrogen isotherms of
these samples were measured at 77 K using an AUTOSORB-1
system (Quantichrome Corporation). Samples were outgassed
at 300°C. Table 1 lists the BET surface area,38 total pore
volume, and micropore volume derived from the isotherms
using the density functional theory (DFT).39 The elemental
composition (C, H, N, O), ash content, and water content of
these activated carbons were analyzed according to ASTM
D5373/D5291 and D5142. Their impurities were analyzed
using inductively coupled plasma-mass spectrometry (ICP-
MS) screen and inductively coupled plasma-optical emission
spectroscopy (ICP-OES) (all by Galbraith Laboratories Inc.,
Knoxville, TN). The results are listed in Table 2. The thermo-
gravimetric analysis was performed using a TA Instruments
thermal analyzer (Q600 SDT) with an air flow of 250 cm3/min
and a heating rate of 3 °C/min.
Catalytic experiments were conducted in a laboratory-scale,
fixed-bed reactor system. The gas flow rates were controlled
by a mass flow controller system (Omega, Model FMA-700
series controllers coupled with an FMA-78P4 readout/power
supply). The glass reactor tube used was 49.5 cm long with
an inner diameter (ID) of 1.25 cm. It was vertically positioned
in an ATS single-zone split tube furnace (Series 3210) in the
down-flow mode. Activated carbons were not pretreated. The
(35) Ghosh, T. K.; Tollefson, E. L. Can. J. Chem. Eng. 1986, 64, 960.
(36) Dalai, A. K.; Majumdar, A.; Chowdhury, A.; Tollefson, E. L. Can.
J. Chem. Eng. 1993, 71, 75.
(37) Wu, X.; Kercher, A. K.; Schwartz, V.; Overbury, S. H.; Arm-
strong, T. R. Carbon 2005, 43, 1087.
(38) Brunauer, S.; Emmett, P. H.; Teller, E. J. Am. Chem. Soc. 1938,
60, 309.
(39) Lastoskie, C.; Gubbins, K. E.; Quirke, N. J. Phys. Chem. 1993,
97, 4786.
1776 Energy & Fuels, Vol. 19, No. 5, 2005 Wu et al.

Table 2. Elemental Analysis Results for Activated

Carbons
elements Centaur WV-B VA-507 W-22
moisture (wt %) 8.8 7.0 4.3 0
ash (wt %) 4.8 6.3 1.1 2.2
carbon (wt %) 88.88 83.46 94.47 93.97
hydrogen (wt %) 0.58 2.01 0.60 0.74
nitrogen (wt %) 1.07 <0.5 <0.5 0.78
oxygen (wt %)a 4.69 8.23 3.85 2.31
aluminum (ppm) 589 148 116 399
sodium (ppm) 7074 6980 495 <32
potassium (ppm) 4834 27.7 3250 754
iron (ppm) 577 4014 169 2183
calcium (ppm) 765 630 1900
phosphorus (ppm) 1568 15200 241 789
magnesium (ppm) 3760 1154 426 562
copper (ppm) 208 247 2255 1176
lithium (ppm) 61 885 1012 252
titanium (ppm) 2094 2064 231 20 Figure 1. Rate of weight loss of activated carbons in air, as
nickel (ppm) 343 242 109 80 a function of temperature (heating rate of 3 °C/min).
zinc (ppm) 31 310 32 <63
vanadium (ppm) 297 51 11 4
silicon (ppm) 1051 9710 1701 1935 tion. The activated carbons were comprised of mainly
a
carbon and ash. The inductively coupled plasma (ICP)
By difference.
analysis suggested the presence of metal impurities
such as iron, copper, potassium, vanadium, and zinc,
sample was loaded in the center region of the reactor, and the
length of the sample bed was ∼7.8 cm. The reaction temper-
whose oxide forms are generally considered as catalysts
ature was controlled by a K-type thermocouple in the furnace for H2S oxidation. However, these amounts could not
and monitored by another K-type thermocouple axially cen- be clearly correlated with the catalytic behavior ob-
tered in the reactor tube. The concentrations of the exhaust served for different carbons. Presumably, the chemical
gases from the reactor were monitored using gas chromatog- states of those impurities in various carbons may differ,
raphy (GC) and a Fourier transform infrared (FTIR) spec- because of the different activation process.
trometer (MIDAC). The GC apparatus (Model 910, Buck The thermal weight loss rates for all activated carbons
Scientific, Inc.) has a 30-m metal column (MXT-QPlot, Restek
in air are shown in Figure 1. The W-22 carbon has the
Corporation) and a flame photometric detector (FPD), which
can detect sulfur compounds present in concentrations down lowest ignition temperature, and the Centaur product
to ∼200 ppb. The conversion of H2S and sulfur selectivity were has the highest ignition temperature. The difference is
calculated using the equations almost 100 °C. The W-22 sample started losing weight
at ∼300 °C, whereas others started at temperatures at
CH2S,in - CH2S,out least 100 °C higher. Obviously, the W-22 carbon is more
Conversion of H2S (%) ) × 100 reactive than other carbons in air, which is a result of
CH2S,in
its carbon structure, surface chemistry, impurity types,
Selectivity to sulfur (%) ) contents, and chemical states. Most metals and their
CH2S,in - CH2S,out - CCOS,out - CSO2,out compounds are effective catalysts for carbon oxidation,
× 100 but their activities are also related to their chemical
CH2S,in - CH2S,out states. Elemental forms, oxides, and carbonates are
much more active than other forms of salt, such as
When no H2S or no byproducts were detected at the reactor sulfates or phosphates.40 Probably, the impurities in the
exit, the conversion or selectivity was referenced to be 100%. W-22 carbon are at those active chemical states and
Two model gas streams were used as carriers: an H2 stream contribute to the observed high oxidation reactivity.
and a simulated reformate stream (50% H2, 15% CO2, 9% CO,
2% N2, ∼23% balance H2O, all by volume). The water vapor The pore size distributions of activated carbons
contained in the reformate stream was produced by heating calculated from adsorption isotherms (using DFT) are
water that was injected by a syringe pump through the inlet presented in Figure 2. It is apparent that, although the
tubing (stainless steel). The heating temperature was moni- WV-B carbon has the highest volume of micropores
tored and controlled by a K-type thermocouple located in the (pore size of <20 Å, according to IUPAC classification),
inlet tubing. The input H2S concentration was 1000 ppm. Air the majority of them are supermicropores (7-20 Å). This
was introduced at the inlet according to a given O2:H2S ratio. carbon also has a large volume of mesopores (20-500
The experiments were conducted at atmospheric pressure with
Å). Even though the other three carbons are similar in
a space velocity of 3100 h-1 (GHSV), unless otherwise specified.
the volume of micropores, the W-22 carbon seems to
have the highest volume of ultramicropores (e7 Å).
3. Results and Discussion 3.2. Catalytic Behavior of Activated Carbons.
Tables 3 and 4 summarize the catalytic behavior of
3.1. Activated Carbon Characteristics. As shown activated carbons, using the catalyst exposure times at
in Table 2, the as-received commercial carbons contain which sulfur compounds were detected at the reactor
some adsorbed water, whereas the synthesized W-22 exit (breakthrough) and the conversion and selectivity
carbon contains no moisture, because it was sealed after
CO2 activation.37 The elemental analysis confirms that
(40) McKee, D. W. The catalyzed gasification reactions of carbon.
the majority of the heteroatoms (H, O, N) in the In Chemistry and Physics of Carbon; Walker, P. L., Jr., Thrower, P.
precursors were eliminated by carbonization and activa- A., Eds.; Marcel Dekker: New York, 1981; p 1.
Desulfurization of Gaseous Fuels Energy & Fuels, Vol. 19, No. 5, 2005 1777

Figure 2. Pore size distribution (according to density functional theory (DFT)).

Table 3. Catalytic Behaviors of Activated Carbons in H2 Table 4. Catalytic Behaviors of Activated Carbons in the
Stream Simulated Reformate
Time (h) at which Catalytic Behavior Time (h) at which Catalytic Behavior
Emission is Detected after 10 h Emission is Detected after 10 h
temp conversion selectivity temp conversion selectivity
(°C) H2S COS SO2 (%) (%) (°C) H2S COS SO2 (%) (%)
Centaur Sample Centaur Sample
110 9.5 nda 14.5 99.8 100 150 nda 0.0 5.2 100 94.6
130 nda nda 4.8 100 98.4 170 nda 0.0 1.5 100 79.4
150 nda nda 2.5 100 93.0 190 nda 0.0 1.0 100 66.5
170 nda 2.0 1.4 100 85.3 WV-B Carbon
190 nda 1.0 0.7 100 71.0 150 7.0 0.0 12.2 97.0 97.8
WV-B Carbon 190 <0.2 0.0 1.3 99.8 67.6
130 2.1 nda nda 73.4 100 VA-507 Carbon
150 3.3 nda 8.7 95.0 98.5 150 2.3 2.0 4.0 77.5 90.2
170 nda 5.3 4.9 100 96.7 190 1.0 1.0 1.5 99.8 72.2
190 nda 5.0 5.0 100 93.8
W-22 Carbon
VA-507 Carbon 130 11.3 10.5 20.5 100 100
130 1.3 nda nda 66.8 100 150 12.1 11.7 16.8 100 100
150 2.0 nd a 6.3 93.7 98.8 170 8.8 7.7 10.3 99.7 98.0
170 2.4 5.3 4.0 99.8 95.0 190 5.0 4.8 6.5 99.8 87.0
190 nda 3.2 3.0 100 90.7
a Not detected in tests up to 15 h.
W-22 Carbon
130 10.2 ndb 27.5 100 100
150 16.3 20.5 19.2 100 100 the conversion of H2S at 10 h increased as the reaction
170 ndb 13.7 13.5 100 100 temperature increased in the H2 stream, whereas the
190 ndb 9.2 9.0 100 94.8 breakthrough time of COS or SO2 and the characteristic
a Not detected in tests up to 15 h. b Not detected in tests up to selectivity each decreased as the temperature increased.
25 h. For the Centaur product, the emission of H2S was
observed at 9.5 h at 110 °C, whereas no H2S emission
after the catalyst was exposed for 10 h as characteristic was measured after 15 h when the reaction temperature
parameters. The parameters in these tables were drawn was 130 °C or higher. According to the manufacturer,
from plots of measured sulfur emissions as a function the Centaur product was not impregnated with metals
of catalyst exposure time, such as that shown in Figure or alkali, but it displayed the catalytic function of these
3. An O2:H2S ratio of 2:1 and a total gas flow rate of materials. This product, indeed, is a strong catalyst and
500 cm3/min were used. Generally, the emission of SO2 has a tendency to over-oxidize the sulfur to undesired
displayed a continuous increase after the breakthrough byproducts, as indicated by the low selectivity values
point, the emission of COS exhibited a “steady state” presented in Table 3. For the practical application, using
after a rapid breakthrough, and H2S showed a break- it at a temperature of <150 °C would be preferred if a
through, followed by a slow increase. Apparently, most combination of good activity and selectivity is desired.
of the observations were in the initial or deactivation For carbon products WV-B and VA-507, increasing the
periods and before the attainment of a “stationary temperature from 130 °C to 190 °C increased H2S
activity level”. A higher value of a parameter in the conversion significantly, but the selectivity was de-
tables would indicate better behavior of a catalyst. creased to some extent. A 100% conversion of H2S was
3.2.1. Behavior in H2. As shown in Table 3, for all reached when the reaction temperature was 170 °C or
activated carbons, the breakthrough time of H2S and higher for the WV-B carbon, but unconverted H2S (ca.
1778 Energy & Fuels, Vol. 19, No. 5, 2005
Figure 3. Typical experimental results measured for the W-22 sample in the simulated reformate.
2 ppm) was still detected for the VA-507 carbon after
2.4 h at 170 °C. When the temperature was 190 °C,
100% conversion was observed with the VA-507 carbon
in after 20 h. Overall, these two products perform better
when the reaction temperature is >150 °C.
Excellent activity and selectivity were observed for
the laboratory-made W-22 carbon. As listed in Table 3,
the conversion and selectivity were all 100% in 10 h
when it was used at 130, 150, and 170 °C. At 190 °C,
the emission of SO2 was detected after 9 h and reached
∼52 ppm at 10 h. The W-22 carbon compound can be
used in the temperature range of 130-190 °C, but a
reaction temperature of ∼150 °C may be preferred.
For all carbons, the emission of COS was also detected
in the H2 stream when the temperature was 170 °C or
higher. However, the concentrations were very low. For
example, it was ∼ 0.3 ppm at 190 °C and 0.2 ppm at
170 °C for the W-22 carbon. Probably, COS was the
product of elemental sulfur and the surface oxygen
complexes (CdO) that were formed by oxygen chemi-
sorption.41 Because the activated carbon functioned as
a reactant in this case, its reactivity might be a factor
that is affecting the COS formation. The thermal
analysis showed that the W-22 carbon started oxidation
at a lower temperature than other carbon samples. This
carbon was, indeed, the most reactive carbon sample
in the formation of COS: at 150 °C, occasional traces
of COS were detected after 20.5 h, whereas no such
phenomenon was observed for other carbons.
In a H2 stream, the “breakthrough” behavior of SO2
and COS confirms one well-known byproduct formation
mechanism, in which the sulfur source is the elemental
sulfur formed from the partial oxidation of H2S.6,42 The
formed sulfur in the initial stage of a run was all
deposited in the pores, preferentially with a diameter
of <12 Å, because of the capillary effect.42 After the
small capillaries were filled, the capillaries with the
larger diameter started to be filled, and the sulfur vapor
(41) Wu, X.; Kercher, A. K.; Schwartz, V.; Overbury, S. H.; Arm-
strong, T. R. Activated carbon as catalyst for removing hydrogen
sulfide: On the formation of byproducts. Presented at the 229th ACS
National Meeting, March 13-17, 2005, San Diego, CA.
(42) Steijns, M.; Mars, P. J. Catal. 1974, 35, 11.
Wu et al.
pressure thus increased. When the pressure reached a
certain value as the test proceeded, the SO2 or COS
formed from the reactions of sulfur with oxygen or
surface oxygen groups became detectable. Similarly, an
increase of temperature resulted in a higher sulfur
vapor pressure and, consequently, a higher emission of
SO2.
The time values in Table 3 also reflect the sulfur
loading before the “breakthrough” of sulfur emissions.
Based on the mechanism described previously, the W-22
carbon certainly has a much larger capacity than the
commercial products to keep sulfur in its small capil-
laries. In other words, the volume of pores where sulfur
is retained should be much larger in the W-22 carbon
than in the other three carbons, because its sulfur
loading reached ca. 14 wt % at 150 °C before any sulfur
emission started to be detected, whereas this value was
only ca. 2 wt % for the Centaur product. A plausible
explanation for this phenomenon may be related to pore-
size distributions and the shape of micropores. It has
been suggested that the pore size, as well as pore
geometry, may significantly affect the catalytic activity
and selectivity of activated carbon.43 Figure 2 indicates
that the W-22 carbon compound has the largest volume
of pores with a diameter of <10 Å. Most likely, the large
volume of micropores and the capacity of these pores in
trapping sulfur are among the reasons responsible for
the unique catalytic behavior of the W-22 carbon.
3.2.2. Behavior in Reformate. The catalytic behavior
of activated carbons in the reformate stream are sum-
marized in Table 4. Among the commercial catalysts,
the Centaur product again displayed a stronger catalytic
effect than the WV-B and VA-507 carbons. No H2S was
measured in tests longer than 15 h for the former
product, whereas H2S was detected in all tests for the
other two products. At the same temperature, the
Centaur product had higher SO2 emission and it came
out earlier, which resulted in a lower selectivity for this
carbon. As for COS emission, it was detected from the
beginning of the tests when the Centaur and WV-B
products were used. The concentrations were almost
(43) Klein, J.; Henning, K.-D. Fuel 1984, 63, 1064.
Desulfurization of Gaseous Fuels
constant during the course of the reaction; it was ∼20
ppm at 150 °C and ∼50 ppm at 190 °C. In the case of
the VA-507 carbon, all the H2S was converted to
elemental sulfur within the first 3 h of testing at 150
°C; however, both activity and selectivity became pro-
gressively worse afterward. At 190 °C, this carbon also
showed no breakthrough of sulfur emissions in the first
hour. After that, a breakthrough of H2S, COS, and SO2
was observed, ∼2 ppm H2S was detected, and the
selectivity decreased to 72.2% at 10 h.
The W-22 carbon again displayed a good catalytic
effect in reformate, as illustrated by the data in Table
4. Similar to its behavior in the H2 stream, an increase
in temperature resulted in a decrease of selectivity,
because the emission of COS and SO2 was detected
earlier and at a higher level. For instance, the measured
concentrations of COS and SO2 at 25 h were ∼17 and
48 ppm at 150 °C, but they were ∼35 and 250 ppm at
190 °C.
Unlike the observations in the H2 stream, an increase
of the reaction temperature is observed to result in an
earlier breakthrough of unconverted H2S in the refor-
mate stream. As shown in Figure 3, the breakthrough
of H2S was observed at 12.1 h when the reaction
temperature was 150 °C, whereas it was observed at
5.0 h when the temperature was 190 °C. Similar
phenomena were also observed for the WV-B and VA-
507 carbons, as indicated by the data in Table 4. In
addition, after the breakthrough, the concentration of
H2S was ∼45 ppm at 150 °C but was only ∼1.8 ppm at
190 °C. Overall, an increase of temperature was still
observed to increase the conversion of H2S, even though
increasing the temperature resulted in an earlier break-
through of unconverted H2S (Figure 3).
3.2.3. Explanations for the Behavior in Reformate. The
data in Tables 3 and 4 clearly show that the W-22
carbon behaved much better than the commercial
products in both the H2 stream and the simulated
reformate, especially in terms of selectivity. The break-
throughs of COS and SO2 are observed to be signifi-
cantly delayed when the W-22 carbon was the catalyst.
In reformate, the emission of COS was detected im-
mediately after the beginning of the test when the
commercial products were used, while it was detected
only after several hours when the W-22 carbon was the
catalyst. The possible reasons for such different behav-
iors have been discussed elsewhere.41 In short, the
sulfur capture by capillary effect in micropores may be
one reason accounting for the difference in SO2 emis-
sion, and the unique capacity of catalyzing a partial
oxidation of COS is one of the reasons why the W-22
carbon is superior to the commercial products.
The unusual breakthroughs of H2S observed in the
reformate stream certainly should be a result of catalyst
deactivation, rather than a equilibrium shift of the
reaction (1) to the reactant side caused by an increase
of temperature and the presence of water. The gas
stream flowing into the catalyst bed in the case of
reformate was actually a mixture that contained SO2
and COS, which are believed to be a result of H2S
oxidation and the reverse reaction of COS hydrolysis
occurring in the inlet tubing that was heated to evapo-
rate water.41 SO2 might have an important role in
catalyst deactivation, besides the deactivation caused
Energy & Fuels, Vol. 19, No. 5, 2005 1779
by sulfur deposition on the carbon surface. In the
presence of water and oxygen, SO2 can be converted to
sulfurous or sulfuric acids, and activated carbon can
catalyze such a conversion.44 The formed acids might
result in catalyst deactivation by occupying the active
sites inside the micropores or reacting with metal
impurities (maybe in the form of metal oxide or carbon-
ate) in activated carbon to form a sulfate.
Note that the reactions occurring in the heated inlet
tubing actually imply the real situation when a de-
sulfurization unit is following another gas processing
unit. The transfer tubing generally is not at room
temperature and similar conversion of H2S to COS or
SO2 may also happen in the tubing. When COS and SO2
are present as input in addition to H2S, the carbon
catalyst should be able to catalyze the conversion of
these sulfur species to elemental sulfur if no emission
of any of them is desired. The possible conversion
reactions include reaction (1), the reaction between SO2
and H2S (the Claus reaction), and the partial oxidation
of COS to elemental sulfur and CO2.41 The loss of
catalytic function on any of them would result in a
breakthrough of the corresponding gas species. There-
fore, the measured COS and SO2 emissions in the case
of reformate might contain contributions from the
breakthrough of these species formed in the inlet tubing
and the over-oxidation of sulfur that occurred in the
catalyst bed.
Therefore, a measurement of ca. 2 ppm H2S at 190
°C after a rapid breakthrough in the case of reformate
for the W-22, VA-507, and WV-B carbons should be a
result of the rapid deactivation of carbon catalysts at a
higher temperature either by occupying active sites or
by sulfation of metal impurities. The occupation of the
most active sites and the loss of active species resulted
in an earlier appearance of H2S, but a higher temper-
ature resulted in a lower level of H2S after its break-
through. (The reaction is controlled by kinetics.) In
addition, under the conditions used in the present study,
the reaction was in the regime of diffusion control.6 An
increase of temperature certainly improved the molec-
ular diffusion and, consequently, the conversion of H2S.
This might also be one of the reasons accounting for the
low concentration of H2S at 190 °C after its break-
through.
The aforementioned argument does not exclude the
possible effect of H2O on the activity and selectivity of
activated carbon when there is a presence of extra water
vapor in the gas stream. Actually, the presence of water
was reported to decrease the activity of carbon catalysts
at the “steady conversion level” and increase the pro-
duction of SO2 significantly.10,35 Because the capillary
effect may also result in water condensation in pores of
small diameter, the sorption of water may have a role
in determining the activity of activated carbons.31
Presumably, the formed water, the desorption of which
can be suppressed by the higher partial pressure of
water vapor in the gas phase, might dissolve SO2. The
formed acid then might cause deactivation of the
catalyst. In addition, the presence of CO2 in the feed
gas also had a detrimental effect on the activity and
(44) Derbyshire, F.; Jagtoyen, M.; Andrews, R.; Rao, A.; Martin-
Gullón, I.; Grulke, E. A. Carbon materials in environmental applica-
tions. In Chemistry and Physics of Carbon; Radovic, L. R., Ed.; Marcel
Dekker: New York, 2001; p 1.
1780 Energy & Fuels, Vol. 19, No. 5, 2005 Wu et al.

Figure 4. Effect of the O2:H2S ratio on the catalytic behavior of the Centaur product sample in a H2 stream (at 150 °C, no COS
was detected, and no H2S was detected, except at 0.5:1).

selectivity of H2S oxidation,45 because of a competition

of CO2 on H2S adsorption.
3.3. Effect of O2:H2S Ratio. Figure 4 shows the
influence of the O2:H2S ratio on the catalytic behaviors
of activated carbons in the H2 stream, using the Centaur
product at 150 °C as an example. The conversion of H2S
was 100% in the test period when the O2:H2S ratio was
1:1 or higher. It was also 100% in the first 3 h of testing
when the ratio was 0.5:1 (the stoichiometric value).
However, after that, the conversion decreased continu-
ously and reached 97% at 15 h. While a “steady state”
after the “breakthrough” is observed for SO2 emission
at a ratio of 0.5:1, the concentration of SO2 is observed
to increase continuously as the test continued when the
ratio was 1:1 or higher. A higher O2:H2S ratio results
in a higher emission of SO2, but a higher ratio does not
Figure 5. Effect of the O2:H2S ratio on the emission of COS
shift the “SO2 breakthrough” time too much. This (the Centaur product used in the simulated reformate at 150
observation, in turn, confirms the argument that SO2 °C).
was the product of over-oxidation of elemental sulfur,
because the vaporization of sulfur from small pores after value is optimal.45,46 The test results in the H2 stream
a certain amount of deposition was mainly related to also suggested that an increase of O2:H2S ratio reduced
the temperature. the selectivity and a ratio between 0.5:1 and 1:1 was
The effect of O2:H2S ratio on the catalytic behavior of sufficient for complete conversion of H2S. However, such
carbon catalysts was also tested by varying the ratio a low ratio was not sufficient for suppressing the
after a catalyst was used for some time and H2S was formation of COS in the reformate. Because the behav-
detected. For all carbons, both H2S conversion and SO2 ior in the “initial period” was of interest, and an increase
production increased as the O2:H2S ratio increased, in of O2:H2S ratio did not cause an earlier SO2 break-
agreement with the reported observations of desulfur- through, a ratio of 2:1 was selected for our subsequent
ization of natural gas using activated carbons.45 In the experiments.
case of reformate, a higher O2:H2S ratio was also 3.4. Effect of Space Velocity. The effect of space
determined to be beneficial for H2S conversion, but it velocity (SV) on the catalytic behavior of activated
increased SO2 emission. As shown in Figure 5, an carbons was tested at 150 °C and an O2:H2S ratio of
increase of O2:H2S ratio suppressed the emission of 2:1. A flow rate of 500 cm3/min of H2 mixed with 1000
COS, but such an effect became insignificant when the ppm H2S was used, and the SV was varied by changing
ratio was higher than 2:1. the mass of the catalyst in the bed. An increase of SV
To avoid the dilute effect of air on the fuel streams, essentially resulted in a decrease of contact time
it is desired that the amount of air introduced during between reagents with the catalyst and, subsequently,
the removal of H2S is minimized. Therefore, a low O2: a decrease in H2S conversion. In addition, because a
H2S ratio is generally recommended. For the continuous
(45) Yang, A. M.; Tollefson, E. L.; Dalai, A. K. Can. J. Chem. Eng.
operation of natural gas desulfurization, it has been 1998, 76, 76.
suggested that a ratio of ∼1.05 times the stoichiometric (46) Dalai, A. K.; Tollefson, E. L. Can. J. Chem. Eng. 1998, 76, 902.
Desulfurization of Gaseous Fuels
Figure 6. H2S conversion as a function of sulfur loading in the presence of the W-22 carbon when a different space velocity (SV)
was used (in the H2 stream).
Figure 7. Emission of SO2 as a function of sulfur loading in the presence of the W-22 carbon when a different SV was used (in
the H2 stream).
large portion of the sulfur produced stayed in the carbon
pores,31 as proved by a rapid decrease in BET surface
area and electron probe microanalysis,6 a higher SV
implies a larger rate of sulfur deposition per unit of
carbon and a larger sulfur loading after a given catalyst
exposure time. It is normal that a more rapid deactiva-
tion of the catalyst at a higher SV was observed. A more
meaningful comparison of SV would then be based on
sulfur loading per mass of catalyst. Figure 6 shows the
H2S conversion on the W-22 carbon at different SVs,
plotted against the sulfur loading, calculated by assum-
ing that all the sulfur stayed in the carbon pores, and
the corresponding plots for SO2 emission are shown in
Figure 7. The increase in the SV value from 3100 h-1
to 20 000 h-1 significantly reduced the H2S conversion
and resulted in the breakthrough of H2S at a much
lower sulfur loading. At an SV value of 3100 h-1, H2S
was detected when the sulfur loading reached ∼14% and
H2S conversion was still ∼99.5% when sulfur loading
was ∼21%. For an SV value of 20 000 h-1, the break-
Energy & Fuels, Vol. 19, No. 5, 2005 1781
through of H2S was observed when the sulfur loading
was only ∼2%. The test was terminated after ∼5 h,
because the exit H2S concentration reached almost 600
ppm, even though SO2 still was not detected. As shown
in Figure 7, the increase of SV did not significantly
affect the breakthrough of SO2 if the comparison is
based on the sulfur loading rather than reaction time.
An increase in the SV value from 3100 h-1 to 5000 h-1
is observed to cause the SO2 to come out at a slightly
lower sulfur loading, but a further increase to 10 000
h-1 or even 20 000 h-1 did not have the same effect as
observed on H2S conversion.
Similar results were also observed for the commercial
products: the increase in the SV value decreased the
conversion of H2S but did not affect the sulfur loading
level at which SO2 was detected. For example, for the
Centaur product, SO2 was detected at a sulfur loading
of 2%-3% when the SV value was varied from 2000 h-1
to 20 000 h-1. Such observations again confirm that SO2
was mainly formed by the over-oxidation of elemental
1782 Energy & Fuels, Vol. 19, No. 5, 2005
Figure 8. Catalytic behavior of the W-22 carbon as a function of sulfur loading in the small reactor. (T ) 150 °C, O2:H2S ) 2:1,
input H2S ) 1000 ppm.)
sulfur, and the reason the W-22 carbon emitted SO2 at
a higher sulfur loading than that of other carbons is due
to its larger loading capacity of sulfur in its micropores.
Because the target gaseous fuels under the real condi-
tions may only contain a much lower concentration of
H2S than 1000 ppm, it takes a much longer time to reach
the same sulfur loading in the carbon catalyst when it
is used in real gaseous fuels. Subsequently, the opera-
tion time of activated carbon before the breakthrough
of sulfur species should be extended proportionally.
The aforementioned experiments indicated that the
conversion of H2S could be improved by increasing
temperature, increasing the O2:H2S ratio (within limits),
or decreasing the SV. However, these approaches did
not show a minimizing effect on the formation of SO2.
Most of them actually increased the production of SO2.
Under certain conditions, there are other approaches
that might improve H2S conversion and minimize SO2
formation at the same time. These approaches include
increasing the pressure, gas velocity, and length/
diameter (L/D) ratio of the catalyst bed. It has been
found that if the reaction is performed under elevated
pressure, H2S conversion is enhanced and SO2 produc-
tion is suppressed,35,45 and the effect is more significant
at a lower temperature. When activated carbons are
used as catalysts, the reaction is generally in the
diffusional control region, and the adsorption of H2S is
critical for the reaction. An increase of gas velocity and
L/D ratio of the catalyst bed (to some extent) might
improve the conversion of H2S and possibly suppress
the formation of SO2. Using a reactor with a small
diameter (0.7 cm), the catalytic performance of the W-22
carbon was tested at SV values of 10 000 and 20 000
h-1, and the results are shown in Figure 8. Because the
same gas flow rate and catalyst volume were used, a
higher gas velocity (3.2 times) and a larger L/D ratio
(5.7 times) were attained in this reactor. Compared to
the data in Figures 6 and 7, it is determined that the
H2S conversion was enhanced and the emission of SO2
was delayed at the two SVs. The results indicate that
the use of a reactor with a higher L/D ratio was
beneficial for both conversion and selectivity. However,
Wu et al.
there are two limitations for taking advantage of this
approach: (i) a higher pressure drop results from using
the same amount of catalysts, and (ii) it is only helpful
before reaching turbulent flow conditions.
4. Conclusions
The experimental study of four activated carbons as
desulfurization catalysts showed that controlling the
reaction conditions only has a limited capacity for
achieving a complete and exclusive conversion of hy-
drogen sulfide (H2S) to elemental sulfur, even though
the activity and selectivity of activated carbons showed
a strong dependence on the test conditions, such as gas
components, temperature, O2:H2S ratio, space velocity
(SV), and length-to-diameter (L/D) ratio of the catalyst
bed. The preferred approach should be the development
of carbon-based catalysts with the desired micro-
structure, impurity type, and contents. Because the
laboratory-synthesized W-22 carbon possesses a rela-
tively large narrow micropore volume for trapping the
formed sulfur, and a unique capacity of catalyzing COS
partial oxidation, it showed a superior ability to remove
sulfur down to the parts per billion (ppb) level in both
the H2 stream and the simulated reformate stream.
Acknowledgment. This research was sponsored by
the U.S. Department of Energy, Office of Hydrogen, Fuel
Cells and Infrastructure Technologies (under Contract
No. DE-AC05-00OR22725 with UT-Battelle, LLC, at
ORNL). The research was supported in part by an
appointment to the ORNL Postdoctoral Research As-
sociates Program administered jointly by ORISE and
ORNL. X. Wu thanks F. C. Montgomery for the access
of the TGA instrument and N. C. Gallego and A. K.
Kercher for reviewing the manuscript.
Note Added after ASAP Publication. In the ver-
sion published on the Web 7/16/2005, a production error
was made regarding the table title and first column
heading in Table 2. The correct version was published
7/27/2005.
EF0500890