Fuel Processing Technology 140 (2015) 21–31

Contents lists available at ScienceDirect

Fuel Processing Technology

journal homepage: www.elsevier.com/locate/fuproc

Research article

Dimethyl sulfide adsorption from natural gas for solid oxide fuel
cell applications
Linda Barelli a, Gianni Bidini a, Umberto Desideri b, Gabriele Discepoli a, Elena Sisani a,⁎
a
Fuel Cell Laboratory, Department of Engineering, University of Perugia, Via Duranti 93, 06125 Perugia, Italy
b
Dipartimento di Ingegneria dell'Energia, dei Sistemi, del Territorio e delle Costruzioni (DESTEC), University of Pisa, Largo Lucio Lazzarino, 56122 Pisa, Italy

a r t i c l e i n f o a b s t r a c t

Article history: The use of solid oxide fuel cell (SOFC) systems in micro-CHP applications is of great interest because of high
Received 12 February 2015 efficiency, low emissions and absence of noise. However, SOFCs are sensitive to degradation caused by organic
Received in revised form 23 July 2015 sulfur compounds present in natural gas or added as odorants. Among them, dimethyl sulfide (DMS) is one of
Accepted 9 August 2015
the sulfur species most resistant to purification treatments and, relative to DMS removal, a lack in literature is
Available online 29 August 2015
highlighted for the investigated application.
Keywords:
Regarding adsorption technology, the present work deals with an organic sensitivity performance analysis of dif-
Adsorption ferent commercial sorbents. Virgin and impregnated activated carbons and a natural zeolite were tested, varying
Natural gas gas hourly space velocity, reactor geometry and filter assembly. Because of differences in activity towards DMS
Dimethyl sulfide exhibited by the investigated materials, to exploit their selectivity, also layered sorbents were realized and tested.
Activated carbon Starting from resulting data, for the yearly operation of 1 kWel SOFC-based micro-CHP system, an optimization of
Zeolite filter assembly (also considering multi-layered configurations) and operative conditions was performed, leading
SOFC to a strong reduction in filter volume (up to five times) and cost (more than three times), with overall pressure
drops compatible with pipeline gas distribution pressure.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
Micro-combined heat and power systems using solid oxide fuel cells
represent a very attractive opportunity for small scale heat and power
production in terms of fuel saving and lower carbon emissions, because
of high efficiency and low environmental impact. Natural gas (NG) is an
important potential feedstock for the production of hydrogen for SOFC-
based CHP systems; moreover NG is a widely applied fuel for final user
application, due to its well-developed infrastructure and relatively
abundant supply.
In all pipeline gases sulfur compounds are added intentionally as
odorants for leak detection and may be composed of thiophenes,
mercaptans, sulfides or disulfides, which are highly reactive sulfur-
containing species, well known for their disagreeable odors. Before
odorant addition, NG contains also other sulfur species naturally present
in the mixture, such as hydrogen sulfide (H2S) and dimethyl sulfide
(DMS). The first one is completely removed through absorption using
basic solutions (principally MEA and DEA) during the treatments oper-
ated immediately after the extraction to prevent pipelines corrosion,
⁎ Corresponding author.
E-mail addresses: linda.barelli@unipg.it (L. Barelli), gianni.bidini@unipg.it (G. Bidini),
umberto.desideri@unipi.it (U. Desideri), gabriele.discepoli@unipg.it (G. Discepoli),
elena.sisani@tin.it, elena.sisani@studenti.unipg.it (E. Sisani).
worsen by the high pressure (70 bar) of NG main distribution network.
On the contrary, DMS, that is one of the most resistant sulfur species, can
be able to move, almost unaltered, from the extraction point to the final
user, with concentrations in the order of several ppmv [1]. Moreover, in
many countries (not in Italy), DMS is also used as natural gas odorant,
mainly in blends with TBM [2].
Sulfur compounds strongly poison reformer and fuel cell catalysts.
Within the SOFC, all sulfur compounds are converted into nickel sul-
fides, which are chemisorbed on nickel catalysts, covering active sites
on electrodes surface [3,4]. In some cases, the presence of sulfur com-
pounds causes immediate performance deterioration; more often, a
long-term performance degradation occurs, over a long exposition
period, due to a continuous deposition on the anode catalyst surface
[5]. Therefore, to prevent detrimental effects on SOFC catalysts, sulfur
concentration has to be reduced to 1 ppmv [6] or lower [7,8].
In literature, different clean-up technologies are used to remove
both inorganic and organic sulfur compounds, but not all can reach
the ultra-low sulfur levels required for fuel cell systems [9] or can be
selective for the elimination of targeted sulfur species, as the DMS
here investigated [10]. Moreover, many desulphurization techniques
need complex integration into the plant and therefore they are not suit-
able for micro-CHP applications [1]. In this case, in fact, the desired re-
quirements for a filtering solution are: high efficiency, reliability,
system simplicity, low costs and easy maintenance [8]. Chemical

http://dx.doi.org/10.1016/j.fuproc.2015.08.012
0378-3820/© 2015 Elsevier B.V. All rights reserved.
22 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31

Nomenclature context and in order to fill the literature gap, a sensitivity analysis on
the considered functional parameters was provided. All adsorption
AB.E.T. B.E.T. specific surface area tests, aiming to DMS removal, were carried out in a fixed bed flow reac-
AC activated carbon tor, suitably designed and realized for the scopes of this work. For the
Cads adsorption capacity experimentation, three sorbent materials with different characteristics
CHP combined heat and power were used, in particular a non-impregnated activated carbon, an im-
Cin inlet concentration pregnated AC and a natural zeolite. All investigated materials were char-
Cout outlet concentration acterized before and after the sorption tests using nitrogen adsorption–
DEA diethanolamine desorption measurements (B.E.T. method), to highlight changes in spe-
DMS dimethyl sulfide cific surface area and micropore volume correlated with the outcomes
EM ethyl mercaptan of the adsorption tests.
GHSV gas hourly space velocity DMS removal experiments were realized using a pipeline natural gas
HDS hydrodesulphurization mixture collected before the odorization point, to evaluate the effect on
ID reactor inner diameter adsorption capacity of sorbent typology, gas hourly space velocity, geo-
LHV lower heating value metric reactor properties (in terms of filter height/diameter ratio) and
MEA monoethanolamine filter assembly (using a mesh or layers of rock wool). Complementary
MM methyl mercaptan runs were performed on the same materials using H2S in concentration
NG natural gas equivalent to the average amount of odorants present in natural gas
SCO selective catalytic oxidation pipelines, in order to have information about sorbents selectivity.
SOFC solid oxide fuel cell Furthermore, a layered filter solution was investigated, on the basis of
STP standard temperature and pressure sulfur removal capability of single materials.
TBM tert-butyl mercaptan Basing on all the experimental resulting data, the optimized filter as-
tbr breakthrough time sembly was determined, aiming to material requirement minimization
THT tetrahydrothiophene and pressure drop limitation compatibly to the final user (e.g. domestic)
Vμ micropore volume grid, characterized by only several mbar of pressure. As discussed in
Section 3.2.2, for the yearly operation of 1 kWel SOFC stack, a diminution
in cartridge volume up to five times, with significant decrease in related
material cost and reduced pressure drop, was obtained for the
scrubbers present drawbacks related to outlet sulfur concentration optimized filter.
higher than fuel cell tolerance limits and system complexity [11,12]. Bi-
ological systems have many disadvantages, because they require a con-
tinuous control of the bacteria growth, an accurate conservation of pH 2. Experimental
values and moisture content [13,14]. Also membrane technologies pres-
ent the difficulty to obtain high removal capacities in a single treatment 2.1. Sorbent selection and surface characterization
cycle and they often have to be coupled with other desulfurization
methods [15]. Three commercial sorbents were selected, in particular two steam
Among the dry processes, hydrodesulphurization (HDS) requires activated carbons supplied by Norit (AC RB1 and AC RGM1) and a
hydrogen supply at high temperature and pressure; moreover, system natural zeolite supplied by ATZ (Apostolico & Tanagro Zeoliti).
operation is complex and not suitable to be implemented in small Norit RB1 is a non-impregnated steam activated carbon, with a
scale applications [16]. Also selective catalytic oxidation (SCO) needs particle diameter of 1 mm, characterized by high adsorption capacity
high temperature to perform sulfur compounds oxidation [10]. and removal efficiency for contaminants present in gas flows at moder-
On the contrary, adsorption technology on porous materials can ate concentrations.
be performed using simple systems also operating at room conditions [1, Norit RGM1 is a steam activated carbon impregnated with Cu (II)
8]. Moreover, it can guarantee sub-ppm level of outlet sulfur concentra- and Cr (IV) salts, with a particle diameter of 1 mm, targeted for the
tion, with 100% efficiency [11], representing an attractive option for removal of low concentrations of sulfur compounds especially from
desulfurization finalized to micro-CHP fuel cell applications. dry gases.
Commercial sorbents, like activated carbons (ACs), zeolites, sepiolite Zeolite ATZ is a natural zeolite, mainly composed of cabasite and
or metal oxides have been extensively used for sulfur compounds phillipsite, with a particle size in the range of 0.2–1.5 mm, commonly
adsorption [17–21]. Moreover, the influence of pore size distribution, ty- used as adsorbent material for gas purification.
pology of raw materials, particles size and surface modification on filter
efficiency and selectivity has been investigated [7,10,22–25,29–32]. Re-
garding DMS uptake, specifically for natural gas purification, only some
Table 1
studies have been done [10,33–35], evidencing the difficulty in DMS re- Main characteristics of selected materials used for adsorption tests.
moval [36] and the need of more experimentation in this direction.
Sorbent removal capability strongly depends on many parameters Adsorbent Particles shape Particles size (mm) Bulk density (kg/m3)

related to the operative conditions of the system [26,27], such as gas AC RB1 Pellets 1.0 480
hourly space velocity, filter geometry and filter assembly, which have AC RGM1 Pellets 1.0 485
Zeolite ATZ Granular 0.2–1.5 875
to be optimized in order to obtain a compact filter with high removal ac-
tivity. Furthermore, sorbents display differences in adsorptive capacity
towards single sulfur compounds simultaneously present in a gas
Table 2
mixture and therefore a composite sorbent could be optimized for this B.E.T. surface area (AB.E.T.) and micropore volume (Vμ) of the selected adsorbent materials.
purpose [11,28].
A systematic and complete analysis of the effects of the cited Sample AB.E.T. (m2/g) Vμ (cm3/g)

functional parameters has not been performed yet. It is instead of funda- AC RGM1 1599 0.66
mental importance for the evaluation of filter overall performance, AC RB1 1101 0.44
Zeolite ATZ 111 0.04
needed for the design and optimization of the clean-up section. In this
 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31 23

Fig. 1. S.E.M. images of AC Norit RB1 (a) and zeolite ATZ (b) samples before adsorption runs.

Table 1 reports the main characteristics of selected materials, on the
basis of datasheets information, in terms of particles shape, size and
bulk density.
All the adsorbents were characterized through nitrogen adsorption–
desorption measurements at 77 K using a computer controlled
Micromeritics (Norcross, GA, USA) ASAP 2010 apparatus. Before the
analysis, samples were degassed at 100 °C overnight. Specific surface
area was determined by application of the Brunauer, Emmett and
Teller (B.E.T.) method [37], while micropore volume was calculated
using the t-plot method [38]. Obtained results, in terms of B.E.T. specific
surface area (AB.E.T.) and micropore volume (Vμ), are reported in Table 2
[39].
AC RB1 and AC RGM1 exhibited high B.E.T. surface area and micro-
pore volume, while lower AB.E.T. and Vμ values were calculated for zeo-
lite ATZ. This is probably due to a low content of zeolitic solid in the
used zeolitite rock. Data reported in Table 2 indicate that, for the exam-
ined samples, a high contribute to AB.E.T. value is due to nitrogen
adsorbed in micropores. The presence of a relevant microporous frac-
tion could play an important role in adsorption kinetics, especially in
the presence of low concentrations of pollutants [27,40].
S.E.M. analysis was performed on one of the two selected activated
carbon samples (Norit RB1) and on zeolite ATZ sample, to observe the
different surface structures of these two types of materials. S.E.M.
images before adsorption runs show a layered morphology and a fine
granular structure for AC Norit RB1 and zeolite ATZ respectively (Fig. 1).
2.2. Test rig description
In order to hold different filter geometries, sorption tests were per-
formed using two interchangeable fixed bed flow reactors, 290 mm
long, with inner diameters of 15 mm and 38 mm respectively. Reactor
and pipelines are made of Teflon®, while fittings and ball valves are
made of stainless steel treated with Sulfinert®, to prevent adsorption
of sulfur compounds on working surfaces of the system.
Two types of feeding gases were used: a pipeline natural gas collect-
ed before the odorization point and pressurized in cylinders (G20 —
AirLiquide), containing 5.5 mg/m3 of DMS, and a mixture of H2S
(200 ppmv) diluted in N2. Gas flow rate was regulated by two mass
flow meter controllers (Bronkhorst High Tech), with full scale values
of 5 nl/h and 250 nl/h, allowing the preparation of different gas mixtures
(accuracy 0.1% related to full scale value and 0.5% related to sensor read-
ing). Air was introduced into the system to clean pipelines and reactor
after each sorption test.
Three different electrochemical sensors were utilized for a con-
tinuous measurement of both outlet and inlet sulfur concentrations
(for the latter, using a pipeline that by-passes the reactor). One sen-
sor is able to detect DMS in the range 0–50 mg/m3 and it is placed in
series with a TBM sensor, working in the same range (both supplied
by MECCOS eTR, accuracy 1%), while the last one is specific for H2S
detection in the range 0–25 ppmv (Recom Industriale s.r.l., accuracy
1%).

Fig. 2. Lay-out of experimental set-up.

24 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31

The concentration values were also verified by gas chromatographic Table 4

analysis, using a DANI GC 1000 equipped with a Restek Rxi®-1 ms col- Test campaign related to DMS removal.

umn (30 m, 0.32 mm ID, 4.00 mm) and a Flame Photometric Detector Test Material GHSV Gas Inlet DMS h/D Filter assembly
(FPD, detection limit 10 ppb). The gas samples were collected through (h−1) matrix concentration
a sample valve (SV), using a quartz gas-tight syringe. (mg/m3)

Pressure drop across the filter was measured through a differential NG01 AC RB1 10,000 CH4 5.5 b1 Mesh
pressure sensor (dP) in the range 0–12.5 mbar (LSI-LASTEM). NG02 4174 N1
NG03 2087
Main features of the experimental set up are shown in Fig. 2.
NG04 800
NG05 AC RGM1 10,000 CH4 5.5 b1 Mesh
NG06 2109 N1
2.3. Testing procedure and methodology NG07 800
NG08 527
NG09 AC RGM1 2109 CH4 5.5 b1 Mesh
Experimental activity consisted of breakthrough testing at room NG10 AC RGM1 2109 CH4 5.5 b1 2 layers of
conditions (temperature 20 °C and pressure 1 atm), using G20 natural rock wool
gas mixture, containing 5.5 mg/m3 of DMS (about 2 ppmv) as resulting (thickness
from certified gas chromatographic analysis. G20 mixture was chosen to 15 mm)
NG11 zeolite ATZ 3804 CH4 5.5 N1 Mesh
investigate removal performance of DMS originally present in natural
NG12 AC RGM1 + zeolite 1651
gas, considering that in Italy odorization standards don't include the ATZ (1:0.5)
use of DMS as main odorant. NG13 AC RGM1 + zeolite 1357
Complementary tests with H2S as pollutant in N2 matrix were per- ATZ (1:1)
formed on the same adsorbent materials, in order to obtain additional NG14 AC RGM1 + zeolite 1000
ATZ (1:2)
data on sorbents selectivity. For these tests, an inlet H2S concentration
of 10 ppmv was used, corresponding to the equivalent total sulfur
content typical of NG pipelines (as resulting from literature and also
from data of natural gas suppliers) [1,41]. Inlet sulfur concentration was verified before and after each test to
The testing activity aimed to evaluate adsorption capacity of selected evaluate the possibility of sulfur deposition phenomena inside the cylin-
sorbents varying operating conditions, in particular gas hourly space ve- der or the need of sensors re-calibration. Moreover, after each test, pipes
locity (GHSV), filter geometry and filter assembly. GHSV variation was and sensors were cleaned using air at 150 nl/h for 30 min.
considered from about 500 h−1 (value compatible with space velocities For each test, the corresponding breakthrough curve was deter-
expected for 1 kWel SOFC micro-CHP application) up to 10,000 h− 1 mined in terms of outlet to inlet sulfur concentration ratio (Cout/Cin).
(accelerated tests in order to reduce screening test time). Using the Basing on breakthrough curves and considering 1 ppmv as maximum
two available reactors, filter geometries characterized by different sulfur tolerance limit for SOFC anode [6], sulfur sorption capacity (Cads
reactor height/inner diameter ratios (h/D) were studied: in particular, in mg/g) of selected sorbents was calculated (Eq. (1)) [10]. Specifically,
solutions with h/D b 1 (ID 38 mm) and h/D N 1 (ID 15 mm) were It was determined by measuring the breakthrough time (tbr) when the
compared, keeping constant other operating parameters. Moreover, outlet sulfur concentration was 1 ppmv H2S or 0.2 ppmv DMS, corre-
the effect of filter assembly was evaluated for two configurations: plac- sponding in both cases to Cout/Cin = 0.1.
ing the filter on a mesh or packing the filter into two rock wool layers
(thickness 15 mm). Q tot  Cin  tbr
Cads ¼ ð1Þ
Finally, different configurations of layered sorbents composed of im- m  103
pregnated activated carbon (AC RGM1) and zeolite (ATZ) were tested
under constant G20 inlet flow rate (24 nl/h), varying the mass ratio of where:
the two constituents. In particular, investigated mass ratios between
AC RGM1 and zeolite ATZ were 1:0.5, 1:1 and 1:2. The performance of Q tot = total gas flow rate (nl/h);
layered sorbents was compared to single layers of AC RGM1 and zeolite Cin = inlet sulfur concentration (mg/m3);
ATZ in similar operative conditions. The resulting filter volumes, filter tbr = breakthrough time (h);
heights (h) and GHSV operative values are indicated in Table 3 for all m = mass of adsorbent material (g).
the examined configurations.
The overall experimental activity on DMS removal was scheduled 3. Results and discussion
according to Table 4, while complementary H2S tests are detailed in
Table 5. As previously indicated, DMS adsorption capacity of selected sor-
bents was determined varying operating conditions, specifically GHSV
(Sections 3.1.1 and 3.1.2) and filter geometry (Section 3.1.3). Moreover,
Table 3
an evaluation of two different configurations of filter assembly was per-
Various configurations of layered sorbents, constituted by layers of AC RGM1 and zeolite
ATZ with different mass ratios, and corresponding values of filter volumes, filter heights formed, using a mesh or rock wool layers, to estimate the effect of gas
(h) and GHSV.

Material Mass (g) Volume (cm3) h (cm) GHSV Table 5

(h−1) Test campaign related to complementary H2S runs.
RGM1 Zeolite RGM1 Zeolite RGM1 Zeolite
ATZ ATZ ATZ Test Material GHSV Gas Inlet H2S h/D Filter assembly
(h−1) matrix concentration
Zeolite ATZ 0 5.52 0 6.31 0 3.57 3804
(ppmv)
AC RGM1 5.52 0 11.38 0 6.44 0 2109
AC RGM1 + zeolite 5.52 2.76 11.38 3.15 6.44 1.78 1651 S01 AC RB1 2087 N2 10 b1 2 layers of rock wool
ATZ (1:0.5) S02 AC RGM1 2109 (thickness 15 mm)
AC RGM1 + zeolite 5.52 5.52 11.38 6.31 6.44 3.57 1357 S03 zeolite ATZ 3804
ATZ (1:1) S04 AC RB1 4174 N2 10 b1 2 layers of rock wool
AC RGM1 + zeolite 5.52 11.04 11.38 12.62 6.44 7.14 1000 S05 1043 (thickness 15 mm)
ATZ (1:2) S06 696
 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31 25

Table 7
B.E.T. surface area (AB.E.T.) and micropore volume (Vμ) of AC RGM1 and zeolite ATZ after
H2S adsorption tests (H2S 10 ppmv; h/D b 1).

Sample AB.E.T. [m2/g] Vμ [cm3/g]

AC RGM1_H2S 991 0.42

Zeolite ATZ_H2S 85 0.03

to 3804 h−1, while filter geometry was characterized by h/D b 1, using

the reactor with ID 38 mm. For these runs, the filter was packed with
two layers of rock wool (thickness 15 mm).
Fig. 3 shows the resulting breakthrough curves, while Table 6 sum-
marizes the corresponding values of breakthrough time, related to Cout
equal to 1 ppmv H2S (Cout/Cin = 0.1), and H2S sorption capacity.
The impregnated AC RGM1 presented the best performance,
Fig. 3. Breakthrough curves of selected materials resulting from H2S adsorption tests (H2S exhibiting a sorption capacity nearly four times bigger than the virgin
10 ppmv; h/D b 1).
one. In fact, as highlighted in literature, impregnation methods create
more adsorption sites on the activated carbon surface, combining phys-
ical and chemical adsorption, with enhanced sulfur removal capacity
Table 6 [22,28,32].
Breakthrough time and H2S sorption capacity of selected materials (H2S 10 ppmv; h/D b 1). Zeolite ATZ showed very low activity towards H2S adsorption. Such
Test Sample Breakthrough time (h) H2S sorption capacity (mg/g) result was expected on the basis of nitrogen adsorption–desorption
measurements carried out on the virgin sample (see Table 2), resulting
S02 AC RGM1 310.5 20.48
S01 AC RB1 78.0 5.14 in low B.E.T. surface area (111 m2/g) and low micropore volume
S03 zeolite ATZ 11.0 0.73 (0.04 cm3/g): these characteristics, together with a non-functionalized
surface, don't favor adsorption mechanism.
Nitrogen adsorption–desorption measurements at 77 K were per-
distribution on filter adsorption capacity (Section 3.1.4). Then, a layered formed on the spent samples of the best and the worse performing ma-
sorbents solution, with layers of activated carbon and zeolite, was stud- terials after H2S adsorption tests, i.e. AC RGM1 and zeolite ATZ, to
ied to exploit the properties of single materials (Section 3.1.5). In some evaluate the presence of changes in their porous structure. Fig. 4 depicts
cases, complementary analysis regarding H2S adsorption was done, to the adsorption–desorption isotherms of AC RGM1 and zeolite ATZ spent
obtain information on sorbent selectivity. samples, each compared to the virgin one. Moreover, the related B.E.T.
The results of B.E.T. analysis on sorbent materials, performed before surface area and micropore volume are reported in Table 7. A decrease
and after each test, allowed the evaluation of changes in the porous tex- in B.E.T. surface area and micropore volume was found: specifically,
tures and the correlation between nitrogen adsorption–desorption around 40% and 20% diminutions were displayed for AC RGM1 and zeo-
measurements and sulfur adsorption data. lite ATZ respectively. This reduction, attributable to the obstruction due
The study of the effect of operating parameters on filter performance to non-reversible H2S adsorption in micropores, highlighted a greater
led to system optimization in terms of efficiency, compactness, costs contribution of microporous fraction in the case of AC RGM1 during dy-
and compatibility with pipeline gas distribution pressure (Section 3.2). namic adsorption tests.
Activated carbons Norit RGM1 and RB1, resulting the best
3.1. DMS adsorption results performing materials during H2S adsorption runs, were investigated in
terms of DMS adsorption, in order to select the suitable sorbent to be
3.1.1. Comparison between adsorbent materials coupled with zeolite ATZ (layered filter solution). As indicated in litera-
Before DMS adsorption tests, a first characterization of selected sor- ture [42], zeolites, characterized by low cost and high selectivity, can be
bent towards H2S was performed, using 10 ppmv of H2S in dry nitrogen used in layered configurations to improve adsorption capacity for or-
and in the absence of oxygen; the total gas flow rate was 24 nl/h and ganic sulfur compounds removal. The study of zeolite ATZ activity to-
5.52 g of sorbent was loaded. Consequently, GHSV ranged from 2087 wards DMS adsorption was presented in Section 3.1.5.

Fig. 4. Nitrogen adsorption–desorption isotherms at 77 K of AC RGM1 (a) and zeolite ATZ (b) samples before and after H2S adsorption test (H2S 10 ppmv; h/D b 1).
26 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31

Fig. 5. Breakthrough curves of RB1 and RGM1 activated carbons as result of DMS sorption tests: (a) GHSV = 10,000 h−1; (b) GHSV = 800 h−1.

Table 8
Breakthrough time and DMS sorption capacity of RB1 and RGM1 activated carbons (10,000 h−1 and 800 h−1 GHSV).

Sample GHSV = 10,000 h−1 GHSV = 800 h−1

Test Breakthrough time (h) DMS sorption capacity (mg/g) Test Breakthrough time (h) DMS sorption capacity (mg/g)

AC RB1 NG01 3.8 0.43 NG04 47.0 0.43

AC RGM1 NG05 17.3 1.96 NG07 189.0 1.71

In the case of DMS removal, Cout/Cin limit was fixed at 0.1, corre-
sponding to 0.2 ppmv of DMS outlet concentration. For both activated
carbons, two different space velocities were considered, 10,000 h− 1
and 800 h−1. The filter was placed on a mesh, using the reactor with
ID 15 mm, while h/D was kept constant for each group of runs. Related
breakthrough curves, depicted in Fig. 5, show that AC RGM1 exhibited
the best adsorption capacity for both GHSV values. In particular, the im-
pregnated AC RGM1 presented a DMS adsorption capacity about four
times higher than the virgin one (Table 8).
Comparing the breakthrough curves present in Fig. 5, a different pro-
file was shown: at 800 h−1 GHSV the curves have a bigger slope and the
outlet DMS concentration increases very rapidly after the breakthrough
point. Therefore, it is predictable that the saturation point (Cout/Cin = 1)
would be reached in a short time. Changing space velocity, the mass
transfer zone (MTZ) could assume a different profile that leads to
diverse shapes of breakthrough curves. In this case, it is reasonable
that, if space velocity is low, DMS molecules have much more possibility
to interact with the carbon surface, determining a more uniform
utilization of micropores and functional surface sites. This leads to a
consequent nearly horizontal profile for MTZ [26].
Comparing the obtained results to AC RGM1 performance towards
H2S, it can be noticed that H2S adsorption capacity of RGM1
(20.48 mg/g at about 2000 h−1 GHSV, as reported in Table 6) is an
order of magnitude bigger that DMS one, confirming, for this organic
compound, a much more difficulty in adsorption.
Nitrogen adsorption–desorption measurements at 77 K were carried
out for the spent samples after the tests described above and the
corresponding isotherms were compared to the virgin samples ones.
The curves relative to GHSV 10,000 h−1 and 800 h−1 are almost over-
lapping, for both AC RB1 and AC RGM1 (Fig. 6). Moreover, for each
space velocity, similar B.E.T. surface area and micropore volume values
were found for the spent samples (Table 9), evidencing the not signifi-
cant influence of GHSV on adsorption capacity. On this basis, it is possi-
ble to point out that the study of changes in microporous structure
occurring on the spent samples (resulting in Vμ and AB.E.T. variation)
can constitute a suitable indicator for a qualitative analysis of sorbent
performance, in terms of both H2S and DMS adsorption.
3.1.2. Effect of GHSV
Tests with different values of space velocity were carried out using
natural gas G20 for the virgin AC RB1, in the range 800–10,000 h− 1,
and for the impregnated AC RGM1, in the range 527–10,000 h− 1. In
all cases the filter was placed on a mesh. The results showed, for both
activated carbons, a reduced variation of DMS adsorption capacity,

Fig. 6. Comparison between adsorption–desorption isotherms at 77 K of virgin and spent samples of AC RB1 (a) and AC RGM1 (b) for DMS adsorption runs at different GHSV values.
 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31 27

Table 9 Table 11
B.E.T. surface area (AB.E.T.) and micropore volume (Vμ) of spent samples of AC RB1 and AC Breakthrough time and H2S sorption capacity of AC RB1 as function of GHSV (H2S 10
RGM1 after DMS adsorption runs at different GHSV values. ppmv; h/D b 1).

Sample GHSV 10,000 h−1 GHSV 800 h−1 Test GHSV Breakthrough time H2S sorption capacity
AB.E.T. [m2/g] Vμ [cm3/g] AB.E.T. [m2/g] Vμ [cm3/g] (h−1) (h) (mg/g)

AC RB1_DMS 706 0.29 743 0.30 S04 4174 19.5 2.57

AC RGM1_DMS 828 0.36 933 0.40 S01 2087 78.0 5.14
S05 1043 249.5 8.23
S06 696 487.0 10.71

Table 10
DMS sorption capacity of AC RB1 and AC RGM1 at different space velocities. The results highlight that DMS adsorption is not significantly influ-
enced by space velocity in the investigated range, while GHSV has a
Sorbent Test GHSV Breakthrough time DMS sorption capacity
(h−1) (h) (mg/g) relevant effect in the case of H2S adsorption. Considering the different
nature of the two molecules, in terms of molecular weight, shape,
AC RB1 NG01 10,000 3.8 0.43
NG02 4174 6.9 0.33
charge distribution and polarity, it is possible that micropore structure
NG03 2087 15.7 0.38 and specific functionalized surface favor H2S adsorption, with an effect
NG04 800 47.0 0.43 greater the smaller is GHSV. This behavior occurs for the higher proba-
AC RGM1 NG05 10,000 17.3 1.96 bility of interaction between H2S molecules and the surface. On the
NG06 2109 60.5 1.45
contrary, DMS probably exhibits less affinity with the surface and
NG07 800 189.0 1.71
NG08 527 233.0 1.40 therefore it is adsorbed with more difficulty, even if diminishing the
space velocity.

going from 0.33 mg/g to 0.43 mg/g in the case of AC RB1 and from
1.40 mg/g to 1.96 mg/g for AC RGM1 (Table 10).
To make a comparison between the GHSV effect on adsorption activity
towards DMS and H2S, tests for H2S removal were carried out on AC RB1,
using 10 ppmv as inlet H2S concentration, a filter configuration charac-
terized by h/D b 1 and packing the filter into two layers of rock wool
(thickness 15 mm). The tested values of GHSV ranged from 696 h− 1
to 4174 h−1.
As evident from Fig. 7 and Table 11, AC RB1 sorption capacity to-
wards H2S, contrary to the DMS case, follows a monotone decrease at
the increasing of space velocity, going from 10.71 mg/g to 2.57 mg/g.
The application of the least squares method to the adsorption data re-
vealed that the experimental points present a hyperbolic behavior,
whose explicit function is expressed by Eq. (2). This trend is similar to
the one obtained in [39] for higher H2S concentrations.
6519
Cads ¼ 1:46 þ
GHSV
3.1.3. Effect of filter geometry
Since AC RGM1 resulted the best performing material from previous
runs, it was used to study the influence of filter geometry, in terms of
h/D, on DMS adsorption capacity. Tests were carried out with 2109 h−1
GHSV (in particular, gas flow rate and sorbent mass of 24 nl/h and 5.52 g
respectively) and using a mesh as filter assembly. Fig. 8 provides a schematic
drawing of the two considered configurations.
Tests results are reported in Fig. 9 in terms of breakthrough curves,
while Table 12 summarizes breakthrough time and DMS adsorption ca-
pacity values. It is possible to observe that configuration (b), character-
ized by h/D N 1, led to an increase of more than 30% in DMS adsorption
capacity, due to a better gas distribution inside the filter.
As reported in previous works, similar results were obtained for H2S
adsorption tests performed using the virgin AC RB1 (inlet H2S concen-
tration 10 ppmv) [26] and AC RGM1 (inlet H2S concentration
200 ppmv) [39].
ð2Þ
3.1.4. Effect of filter assembly
In order to evaluate the influence of filter assembly on DMS removal,
tests were carried out with G20 natural gas using the impregnated AC
RGM1. Keeping constant operating parameters as GHSV (2109 h−1) and
h/D b 1, two different configurations of filter assembly were considered:
filter bed placed on a mesh and filter packed with two rock wool layers
15 mm thick.

Fig. 7. GHSV effect on H2S adsorption capacity for AC RB1 (H2S 10 ppmv; h/D b 1). Fig. 8. Tested filter configurations with different h/D values.
28 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31

Fig. 9. Breakthrough curves of AC RGM1 for DMS adsorption using different reactor geom-
Fig. 10. Breakthrough curves of AC RGM1 for DMS adsorption using different filter assem-
etries (GHSV = 2109 h−1).
blies (GHSV = 2109 h−1; h/D b 1).

The curves, depicted in Fig. 10, show an earlier breakthrough for the
first configuration (mesh), while, as reported in Table 13, an increase of
about 10% in DMS adsorption capacity was observed for the latter (rock
wool). Adsorption tests carried out using different layers of rock wool
without carbon resulted in immediate breakthrough for both DMS and
H2S, revealing the absence of adsorption properties for this material.
Consequently, performance enhancement can be only ascribed to filter
assembly variation.
To this regard, it is highlighted as the increase of 10% in adsorption
capacity is significant, because of the overall accuracy (1.8%) of the
measurement system (considering a flow rate of 24 nl/h supplied by a
gas flow controller with 250 nl/h full scale). The reason of the improve-
ment is attributable to the realization of a better gas distribution inside
the filter, provided by the rock wool layer placed on the filter side nearer
to inlet gas injection.
3.1.5. Layered sorbents
The selectivity of adsorbent materials can be exploited realizing
composite sorbents formed by layers of materials, each one character-
ized by different properties. In natural gas mixtures, in fact, there is
often the co-presence of various organic and inorganic sulfur com-
pounds, beside hydrocarbons with higher molecular weight and small
percentages of humidity, that could be adsorbed in the pores, determin-
ing a premature filter breakthrough. The experimental activity aimed to
study composite sorbents constituted by two layers: zeolite ATZ (placed
on the top of the filter, nearer to gas mixture injection) and AC RGM1.
The impregnated AC RGM1 is the material characterized by the best per-
formance, as resulting from previous adsorption tests, while zeolite is a
sorbent well known in literature for both capability to absorb water and
very uniform porosity, with a great selectivity for targeted organic sulfur
molecules [43–45].
As reported in Section 2.3, the AC RGM1/zeolite ATZ mass ratios con-
sidered for the layered configuration were 1:0.5, 1:1 and 1:2. In all cases,
tests were performed by using the reactor with ID 15 mm and placing
the filter on a mesh. Moreover, all the configurations were characterized
by h/D N 1, because of the filter geometry with the highest adsorption
capacity, as reported in Section 3.1.3. GHSV varied in the range
1000–3804 h− 1: in fact, basing on previous results obtained for AC
RGM1 (Section 3.1.2, Table 10), it was stated that GHSV doesn't affect
significantly filter DMS adsorption capacity for the investigated break-
through point (Cout/Cin = 0.1).
The results in terms of DMS adsorption capacity, listed in Table 14,
showed that, adding to mass m (m = 5.52 g) of AC RGM1 a layer of
zeolite ATZ equal to 0.5 m (configuration 1:0.5), the obtained sorption
capacity (1.91 mg/g) is higher than AC RGM1 one (1.45 mg/g).
Moreover, DMS sorption capacity presented an increase of about 60%,
compared to the single layer of AC RGM1, adding to mass m of AC
RGM1 an equivalent mass of zeolite ATZ (configuration 1:1). In particu-
lar, the resulting value (2.29 mg/g) is higher than the sum of the DMS
adsorption capacity of the single layers. In the third case, mass m of
AC RGM1 and mass 2 m of zeolite ATZ were considered (configuration
1:2): the corresponding DMS sorption capacity revealed a decrease
(about 40%) comparing to configuration 1:1.
The related breakthrough curves are presented in Fig. 11: an in-
crease of breakthrough time for the layered sorbents was observed for
all considered configurations, compared to single adsorbents. Moreover,
breakthrough curves of configurations 1:1 and 1:2 are practically
overlapped.
The layer of zeolite ATZ, placed on the top of the filter, is able to pre-
serve AC RGM1, maximizing its activity towards DMS removal. In partic-
ular, it allows the adsorption of humidity and other organic or inorganic
sulfur species, potentially present in the mixture and not adsorbed by
AC RGM1. Furthermore, it produces an effect similar to a rock wool
layer: zeolite, in fact, has a very uniform porosity, resulting in a better
distribution of the gas mixture on the AC RGM1 layer, thus allowing it
to better exploit its sulfur removal capability.
The calculated DMS adsorption capacity diminished after addition of
the second zeolite ATZ layer, because the increase in sorbent mass is not
compensated by an improvement of removal capability. Therefore, con-
figuration 1:1 can be considered a sort of threshold, over which a further
addition of layers of zeolite does not determine any increase of break-
through time, causing consequently a decrease of DMS adsorption ca-
pacity due to a greater amount of used sorbent.

Table 12
Effect of reactor geometry (in terms of h/D) on DMS adsorption for AC RGM1 (GHSV = Table 13
2109 h−1). Effect of filter assembly on DMS adsorption for AC RGM1 (GHSV 2109 h−1; h/D b 1).

Test Filter D filter h filter h/D Breakthrough DMS sorption Test Filter assembly Breakthrough time DMS sorption capacity
geometry (mm) (cm) time (h) capacity (mg/g) (h) (mg/g)

NG09 (a) 38 1.00 b1 45.0 1.08 NG09 Mesh 45.0 1.08

NG06 (b) 15 6.44 N1 60.5 1.45 NG10 Rock wool 49.5 1.18
 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31 29

Table 14
Breakthrough times and DMS sorption capacity of zeolite ATZ, AC RGM1 and layered
sorbents (AC RGM1 + zeolite ATZ) using different mass ratios.

Test Sample Breakthrough DMS sorption

time (h) capacity (mg/g)

NG11 Zeolite ATZ 22.5 0.54

NG06 AC RGM1 60.5 1.45
NG12 AC RGM1 + zeolite ATZ (1:0.5) 120.0 1.91
NG13 AC RGM1 + zeolite ATZ (1:1) 191.5 2.29
NG14 AC RGM1 + zeolite ATZ (1:2) 180.0 1.43

An advantage in the use of natural zeolite instead of an equal amount

of activated carbon is that zeolites have a much lower cost than activat-
ed carbons: if the comparison is made with the modified ACs, the differ-
ence in costs is about two orders of magnitude (AC RB1 6.50 €/kg, AC
RGM1 18.50 €/kg, zeolite ATZ 0.25 €/kg). Moreover, natural zeolites
can be regenerated with an almost complete preservation of their
performance [34], while regeneration of modified activated carbon
could be very difficult, because of the presence of Cu and Cr salts on
its surface [31].
3.2. Optimized filter assembly and related performance
The results of the previous tests highlighted that AC RGM1 is the best
performing material for DMS removal from G20 natural gas, with posi-
tive effects on its adsorption capacity mainly increasing h/D, packing the
filter bed with rock wool layers and, in the case of layered sorbents,
using zeolite in the filter side nearer to gas injection. In particular, the
higher value of adsorption capacity was obtained for the layered config-
uration 1:1 (2.29 mg/g).
Next Section 3.2.1 is focused on the evaluation of these solutions to
verify their compatibility with pipeline gas distribution pressure, study-
ing the variation of pressure drop across the filter. Finally, Section 3.2.2
provides the estimation of advantages, in terms of filter volume and cost
reduction, obtainable through filter optimization.
3.2.1. Evaluation of pressure drop
The pressure drop occurring through the filter was evaluated for
different tests, in order to find how it is influenced by the main studied
parameters, considering that, for an optimal system design of 1 kWel
SOFC stack fed by pipeline NG, a maximum pressure drop of few mbar
is expected [46].
As a first step, the effect on the overall filter pressure drop, due to the
possible addition of a rock wool layer, was investigated varying superfi-
cial N2 velocity. In particular the tests were carried out separately on a
single layer of rock wool 15 mm thick and on a layer of AC RGM1 with
the same thickness. For these runs, the reactor with ID 15 mm was
Fig. 12. Comparison between the pressure drop of AC RGM1 and rock wool layer
(thickness 15 mm) as a function of superficial N2 velocity.
used and therefore the geometry of the system was characterized by
h/D = 1. As visible in Fig. 12, the rock wool determines a pressure
drop more than double relative to the equivalent thickness of AC
RGM1, for all the tested values of superficial gas velocity.
Therefore, even if a 15 mm thick rock wool layer in AC RGM1 filter
determines an increase in sorption capacity (of about 10%), it causes,
at the same time, a significant increase in pressure drop. Moreover,
considering also the positive effect on gas distribution due to zeolite in
multilayered filter, the introduction of a layer of rock wool on the top
of the filter can be avoided.
Subsequently, the distributed pressure drop produced by AC RGM1,
considering a rhombohedral disposition of particles, was characterized
varying the N2 flow rate up to 240 nl/h. In particular, tests were carried
out for two configurations, with different diameter values and same
height (3.25 cm), characterized by h/D of 0.86 and 2.17 respectively.
The obtained results, in terms of pressure drop per cm (dP/cm), are
shown in Fig. 13. These data will be useful to determine the overall pressure
drop of the optimized filter, as detailed in the next Section 3.2.2.
3.2.2. Filter volume calculation and cartridge cost estimation
Using the adsorption results obtained during the experimentation,
the volume of the filter, needed for one year (8000 h) operation of
1 kWel SOFC stack, was calculated for:
– the optimal solution, i.e. layered filter configuration 1:1 (test NG13),
characterized, as obtained from the experimental activity, by an ad-
sorption capacity of 2.29 mg/g;
– two non-optimized solutions, in particular: AC RGM1 mono-layer
(test NG06) and virgin AC RB1 mono-layer (test NG04).

Fig. 11. Breakthrough curves of zeolite ATZ, AC RGM1 and layered sorbents (AC RGM1 + Fig. 13. Pressure drop occurring on AC RGM1 for different reactor geometries (h/D = 2.17
zeolite ATZ) using different mass ratios. and h/D = 0.86) as a function of N2 flow rate (filter height = 3.25 cm).
30 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31
Table 15
Filter volume and cost for different configurations.
Test Material Filter mass (g)
Optimized solution NG13 AC RGM1 + ATZ (1:1) 3342
Non-optimized solution 1 NG06 AC RGM1 5277
Non-optimized solution 2 NG04 AC RB1 17,796
Basing on the outcomes of pressure drop analysis (Section 3.2.1), for
all cases filter geometry with h/D N 1 and filter placed over a mesh were
considered.
Assuming a DMS concentration of 5.5 mg/m3 in the feeding natural
gas, the total amount of DMS to be removed during 8000 h of plant
activity is about 7650 mg. This evaluation is made considering SOFC
efficiency of 0.6, 34.5 MJ/m3 as LHV of natural gas, beyond that the
DMS adsorption capacity previously obtained. In particular, relative
to adsorption capacity, scale effect and time dependence were not con-
sidered. Moreover, to determine the filter dimensions, material
densities as reported in Table 1, void fraction of 0.26 (corresponding
to rhombohedral particles disposition), 20 cm as filter diameter
and 1.2 as factor of safety were adopted. Aiming to a comparative anal-
ysis, these assumptions were made for all the three investigated
configurations.
For the optimized solution (layered case), a compact filter character-
ized by a height of 26 cm was obtained (Table 15). The filter volume pre-
sented a reduction of 13% relative to the non-optimized solution 1
(mono-layer of AC RGM1) and it is more than five times lower consid-
ering the non-optimized solution 2 (use of non-impregnated AC RB1).
This is extremely relevant in consideration of the compactness required
by the investigated micro-CHP application.
On the basis of pressure drop characterization shown in Fig. 13, the
estimated pressure drop for the optimized configuration (ID 20 cm,
height 26 cm and, consequently, h/D 1.3) is about 8 mbar for the oper-
ative flow rate of 174 nl/h. This value is compatible with the constraints
imposed by the NG pipeline distribution.
The advantages obtainable using the optimized configuration are
much more relevant from an economic point of view. Considering, in
fact, AC RB1, AC RGM1 and zeolite ATZ costs, as indicated in
Section 3.1.5, the cartridge cost for the three investigated solutions
was determined and reported in Table 15, together with DMS adsorp-
tion capacities and calculated filter volumes. A cost reduction of more
than three times resulted for the optimal layered filter compared to
the non-optimized solution 1. Moreover, a not relevant difference in
costs between the two non-optimized solutions was found, due to the
three-times lower cost of AC RB1 in respect to AC RGM1.
This calculation wants to focus how a small difference in adsorp-
tion capacity, depending on the selection of the proper material and
the most appropriate operative conditions, can determine impor-
tant changes in terms of filter volume and cartridge cost for DMS
removal.
Moreover, the experimentation highlighted that the evaluation of
filter performance should be performed not only on the basis of adsorp-
tion datasheet information, but using operative conditions similar to the
real ones and, if possible, through preliminary lab-test campaigns.
4. Conclusions
Three selected commercial adsorbent materials, in particular
virgin AC (RB1), impregnated with Cu and Cr salts AC (RGM1) and
natural zeolite (ATZ), were studied for DMS adsorption (concentration
5.5 mg/m3) in a range of operative conditions representative of 1 kWel
SOFC based system fed by pipeline natural gas. The aim was to define
the influence of main parameters (gas hourly space velocity, filter ge-
ometry and filter assembly) on DMS adsorption capacity, in order to
perform an optimization of filter operative conditions. Subsequently, a
further experimental campaign was executed to determine the
Filter volume (cm3) Filter height (cm) DMS sorption capacity (mg/g) Cost (€)
8268 26 2.29 31.33
9313 30 1.45 97.63
44,489 142 0.43 115.67
performance of the optimized filter configuration and provide its com-
parison with non-optimized solutions in terms of reduction in cartridge
volume and cost.
The main outcomes of the sensitivity analysis:
– AC RGM1 presented the best performance for both H2S (adsorption
capacity 20.48 mg/g) and DMS removal (adsorption capacity
1.96 mg/g for GHSV 10,000 h−1 and 1.71 mg/g for GHSV 800 h−1)
compared to the other tested materials.
– Zeolite ATZ showed a low value of H2 S adsorption capacity
(0.73 mg/g), confirming the B.E.T. analysis results. Nevertheless,
it is a very inexpensive and selective sorbent that can be coupled
with AC RGM1 in layered configurations to improve filter perfor-
mance towards DMS removal.
– DMS adsorption capacity of AC RGM1 and AC RB1 is not significantly
affected by GHSV in a wide range of space velocities (about
500 h−1–10,000 h−1), differently from H2S case, where a monotone
decrease (following a hyperbolic behavior) was observed for AC
RB1.
– Filter geometry influenced AC RGM1 performance, leading to an
increase of more than 30% in DMS adsorption capacity in case of
h/D N 1, due to the possibility of a better gas distribution inside
the filter.
– Performance of AC RGM1 is also affected by filter assembly: the
presence of a rock wool layer placed on the top of the filter,
near to the injection of the gas mixture, led to an increase of
10% in DMS adsorption capacity.
The main outcomes relative to filter optimization:
– Composite sorbents made of layers of AC RGM1 and zeolite ATZ with
different mass ratios (1:0.5, 1:1 and 1:2) were realized and tested. The
best performance was obtained for the configuration 1:1 (2.29 mg/g),
showing a significant increase of DMS adsorption capacity with
respect to RGM1 mono-layer case, due to positive effects of zeolite
(capability to absorb natural gas humidity, to remove other organic
or inorganic sulfur compounds not adsorbed by AC RGM1 and to en-
hance gas distribution inside the filter due to its uniform porosity).
– The optimized solution, obtained for the layered filter (AC RGM1 and
zeolite ATZ in configuration 1:1), leads to a diminution in filter
volume (for 1 year operation of 1 kWel SOFC micro-CHP) of 13%
respect to AC RGM1 mono-layer and five times relative to AC RB1
mono-layer, using comparable operative conditions. The layered
filter, characterized by ID 20 cm and h 26 cm, fulfills compactness
requirements of micro-CHP applications, beyond that constraints
relative to pressure drop.
– Cost analysis highlighted a reduction of three times for the optimized
cartridge respect to the use of AC RGM1 mono-layer.
Acknowledgments
The authors would thank ProEng s.r.l. for financial support in the
realization of testing activity and Prof. Fabio Marmottini (Department
of Chemistry, University of Perugia) for his collaboration in sorbents
characterization.
 L. Barelli et al. / Fuel Processing Technology 140 (2015) 21–31
References
[1] D. Shekhawat, J.J. Spivey, D.A. Berry (Eds.),Fuel Cells: Technologies for Fuel Process-
ing, 2011.
[2] D. Tenkrat, T. Hlincik, O. Prokes, Natural Gas Odorization, 2006 88–104.
[3] J.R. Rostrup-Nielsen, J.B. Hansen, S. Helveg, N. Christiansen, A.-K. Jannasch, Sites for
catalysis and electrochemistry in solid oxide fuel cell (SOFC) anode, Appl. Phys. A 85
(2006) 427–430, http://dx.doi.org/10.1007/s00339-006-3702-1.
[4] J.B. Hansen, Correlating sulfur poisoning of SOFC nickel anodes by a Temkin iso-
therm, Electrochem. Solid-State Lett. 11 (2008) B178, http://dx.doi.org/10.1149/1.
2960521.
[5] J. Bao, G.N. Krishnan, P. Jayaweera, K.-H. Lau, A. Sanjurjo, Effect of various coal con-
taminants on the performance of solid oxide fuel cells: part II. ppm and sub-ppm
level testing, J. Power Sources 193 (2009) 617–624.
[6] Fuel cell handbook — seventh edition, EG &amp, Technical Services, INC, 2004.
[7] S. Hernández, L. Solarino, G. Orsello, N. Russo, D. Fino, G. Saracco, et al., Desulfuriza-
tion processes for fuel cells systems, Int. J. Hydrogen Energy 33 (2008) 3209–3214,
http://dx.doi.org/10.1016/j.ijhydene.2008.01.047.
[8] P.J. de Wild, R.G. Nyqvist, F.A. de Bruijn, E.R. Stobbe, Removal of sulphur-containing
odorants from fuel gases for fuel cell-based combined heat and power applications,
J. Power Sources 159 (2006) 995–1004, http://dx.doi.org/10.1016/j.jpowsour.2005.
11.100.
[9] E. Smet, P. Lens, H. Van Langenhove, Treatment of waste gases contaminated with
odorous sulfur compounds, Environ. Sci. Technol. 28 (1998) 89–117.
[10] H. Cui, S.Q. Turn, M.A. Reese, Removal of sulfur compounds from utility pipelined
synthetic natural gas using modified activated carbons, Catal. Today 139 (2009)
274–279, http://dx.doi.org/10.1016/j.cattod.2008.03.024.
[11] D. Mescia, S.P. Hernández, A. Conoci, N. Russo, MSW landfill biogas desulfurization,
Int. J. Hydrogen Energy 36 (2011) 7884–7890, http://dx.doi.org/10.1016/j.ijhydene.
2011.01.057.
[12] H. Su, S. Wang, H. Niu, L. Pan, A. Wang, Y. Hu, Mass transfer characteristics of H2S
absorption from gaseous mixture into methyldiethanolamine solution in a T-
junction microchannel, Sep. Purif. Technol. 72 (2010) 326–334, http://dx.doi.org/
10.1016/j.seppur.2010.02.024.
[13] M. Fernández, M. Ramírez, J.M. Gómez, D. Cantero, Biogas biodesulfurization in an
anoxic biotrickling filter packed with open-pore polyurethane foam, J. Hazard.
Mater. 264 (2014) 529–535, http://dx.doi.org/10.1016/j.jhazmat.2013.10.046.
[14] N. de Arespacochaga, C. Valderrama, C. Mesa, L. Bouchy, J.L. Cortina, Biogas deep
clean-up based on adsorption technologies for solid oxide fuel cell applications,
Chem. Eng. J. 255 (2014) 593–603, http://dx.doi.org/10.1016/j.cej.2014.06.072.
[15] D. Wang, W.K. Teo, K. Li, Removal of H2S to ultra-low concentrations using an asym-
metric hollow fibre membrane module, Sep. Purif. Technol. 27 (2002) 33–40, http://
dx.doi.org/10.1016/S1383-5866(01)00186-1.
[16] O. Van Rheinberg, K. Lucka, H. Kohne, T. Schade, J. Andersson, Selective removal of
sulphur in liquid fuels for fuel cell applications, Fuel 87 (2008) 2988–2996, http://
dx.doi.org/10.1016/j.fuel.2008.03.020.
[17] M. Al-Harahsheh, R. Shawabkeh, M. Batiha, A. Al-Harahsheh, K. Al-Zboon, Sulfur
dioxide removal using natural zeolitic tuff, Fuel Process. Technol. 126 (2014)
249–258, http://dx.doi.org/10.1016/j.fuproc.2014.04.025.
[18] A.H.M. Shahadat Hussain, B.J. Tatarchuk, Mechanism of hydrocarbon fuel desulfuri-
zation using Ag/TiO2–Al2O3 adsorbent, Fuel Process. Technol. 126 (2014) 233–242,
http://dx.doi.org/10.1016/j.fuproc.2014.05.006.
[19] P.J. De Wild, Method for desulphurisation; US 2006/0058565–2006, 2006.
[20] P.H. Ho, S.C. Lee, J. Kim, D. Lee, H.C. Woo, Properties of a manganese oxide octahedral
molecular sieve (OMS-2) for adsorptive desulfurization of fuel gas for fuel cell
applications, Fuel Process. Technol. 131 (2015) 238–246, http://dx.doi.org/10.
1016/j.fuproc.2014.11.019.
[21] R.N. Fallah, S. Azizian, Removal of thiophenic compounds from liquid fuel by
different modified activated carbon cloths, Fuel Process. Technol. 93 (2012)
45–52, http://dx.doi.org/10.1016/j.fuproc.2011.09.012.
[22] J. Wieckowska, Catalytic and adsorptive desulphurization of gases, Catal. Today 24
(1995) 405–465.
[23] P.J.M. Carrott, M.M.L. Ribeiro Carrott, R.A. Roberts, Physical adsorption of gases by
microporous carbons, Colloids Surf. 58 (1991) 385–400, http://dx.doi.org/10.1016/
0166-6622(91)80217-C.
[24] X. Zhang, Y. Zhang, Q. Liu, W. Zhou, Surface properties of activated carbon from
different raw materials, Int. J. Min. Sci. Technol. 22 (2012) 483–486, http://dx.doi.
org/10.1016/j.ijmst.2012.01.007.
31
[25] K. Sakanishi, Z. Wu, A. Matsumura, I. Saito, T. Hanaoka, T. Minowa, et al., Simulta-
neous removal of H2S and COS using activated carbons and their supported
catalysts, Catal. Today 104 (2005) 94–100, http://dx.doi.org/10.1016/j.cattod.2005.
03.060.
[26] L. Barelli, G. Bidini, F. Gallorini, H2S absorption on activated carbons NoritRB1: CFD
model development, Fuel Process. Technol. 100 (2012) 35–42, http://dx.doi.org/
10.1016/j.fuproc.2012.02.007.
[27] G. Monteleone, M. De Francesco, S. Galli, M. Marchetti, V. Naticchioni, Deep H2S re-
moval from biogas for molten carbonate fuel cell (MCFC) systems, Chem. Eng. J. 173
(2011) 407–414, http://dx.doi.org/10.1016/j.cej.2011.07.078.
[28] S.P. Hernández, F. Scarpa, D. Fino, R. Conti, Biogas purification for MCFC application,
Int. J. Hydrogen Energy 36 (2011) 8112–8118, http://dx.doi.org/10.1016/j.ijhydene.
2011.01.055.
[29] D. Lee, E.-Y. Ko, H.C. Lee, S. Kim, E.D. Park, Adsorptive removal of
tetrahydrothiophene (THT) and tert-butylmercaptan (TBM) using Na–Y and
AgNa–Y zeolites for fuel cell applications, Appl. Catal. A Gen. 334 (2008)
129–136, http://dx.doi.org/10.1016/j.apcata.2007.09.046.
[30] H. Tamai, H. Nagoya, T. Shiono, Adsorption of methyl mercaptan on surface modified
activated carbon, J. Colloid Interface Sci. 300 (2006) 814–817.
[31] P.M.F. Benoit, G. Weber, J.-P. Bellat, C. Paulin, S. Limborg-Noetinger, M. Thomas, Ad-
sorption equilibria of binary ethylmercaptan/hydrocarbon mixtures on a NaX zeo-
lite, Stud. Surf. Sci. Catal. 158 (2005) 1185–1192, http://dx.doi.org/10.1016/S0167-
2991(05)80464-1.
[32] S. Bashkova, A. Bagreev, T.J. Bandosz, Catalytic properties of activated carbon surface
in the process of adsorption/oxidation of methyl mercaptan, Catal. Today 99 (2005)
323–328, http://dx.doi.org/10.1016/j.cattod.2004.10.007.
[33] H. Wakita, Y. Tachibana, M. Hosaka, Removal of dimethyl sulfide and t-
butylmercaptan from city gas by adsorption on zeolites, Microporous Mesoporous
Mater. 46 (2001) 237–247.
[34] S. Satokawa, Y. Kobayashi, H. Fujiki, Adsorptive removal of dimethylsulfide and
-butylmercaptan from pipeline natural gas fuel on Ag zeolites under ambient condi-
tions, Appl. Catal. B Environ. 56 (2005) 51–56, http://dx.doi.org/10.1016/j.apcatb.
2004.06.022.
[35] W.K. Jo, S.H. Shin, E.S. Hwang, Removal of dimethyl sulfide utilizing activated carbon
fiber-supported photocatalyst in continuous-flow system, J. Hazard. Mater. 191
(2011) 234–239, http://dx.doi.org/10.1016/j.jhazmat.2011.04.069.
[36] H. Huang, D. Yi, Y. Lu, X. Wu, Y. Bai, X. Meng, et al., Study on the adsorption behavior
and mechanism of dimethyl sulfide on silver modified bentonite by in situ FTIR and
temperature-programmed desorption, Chem. Eng. J. 225 (2013) 447–455, http://dx.
doi.org/10.1016/j.cej.2013.04.018.
[37] S. Brunauer, P.H. Emmett, E. Teller, Adsorption of gases in multimolecular layers, J.
Am. Chem. Soc. 60 (1938) 309–319, http://dx.doi.org/10.1021/ja01269a023.
[38] B.C. Lippens, J.H. De Boer, J.H. De Boer, Studies on pore systems in catalysts: V. The t
method, J. Catal. 4 (1965) 319–323, http://dx.doi.org/10.1016/0021-9517(65)90307-6.
[39] E. Sisani, G. Cinti, G. Discepoli, D. Penchini, U. Desideri, F. Marmottini, Adsorptive re-
moval of H2S in biogas conditions for high temperature fuel cell systems, Int. J. Hydro-
gen Energy (2014) 21753–21766, http://dx.doi.org/10.1016/j.ijhydene.2014.07.173.
[40] T.J. Bandosz, On the adsorption/oxidation of hydrogen sulfide on activated carbons
at ambient temperatures, J. Colloid Interface Sci. 246 (2002) 1–20, http://dx.doi.
org/10.1006/jcis.2001.7952.
[41] G. Alptekin, S. DeVoss, M. Dubovik, J. Monroe, R. Amalfitano, G. Israelson, Regenera-
ble sorbent for natural gas desulfurization, J. Mater. Eng. Perform. 15 (2006)
433–438, http://dx.doi.org/10.1361/105994906X117251.
[42] C. Ratnasamy, J.P. Wagner, S. Spivey, E. Weston, Removal of sulfur compounds from
natural gas for fuel cell applications using a sequential bed system, Catal. Today 198
(2012) 233–238, http://dx.doi.org/10.1016/j.cattod.2012.04.069.
[43] D.M.A. Melo, J.R. de Souza, M.A.F. Melo, A.E. Martinelli, G.H.B. Cachima, J.D. Cunha,
Evaluation of the zinox and zeolite materials as adsorbents to remove H2S from nat-
ural gas, Colloids Surf. A Physicochem. Eng. Asp. 272 (2006) 32–36, http://dx.doi.
org/10.1016/j.colsurfa.2005.07.005.
[44] C.-L. Hwang, N.-H. Tai, Removal of dimethylsulfide by adsorption on ion-exchanged
zeolites, Appl. Catal. B Environ. 93 (2010) 363–367, http://dx.doi.org/10.1016/j.
apcatb.2009.10.009.
[45] J.C. Moise, J.P. Bellat, A. Menthivier, Adsorption of water vapor on X and Y zeolites
exchanged with barium, Microporous Mesoporous Mater. 43 (2001) 91–100.
[46] E. Sisani, A. Biagetti, A. Baldi, S. Angelini, Desulphurization of natural gas for fuel cell-
based micro combined heat and power applications, Proc. EFC2011 Eur. Fuel Cell —
Piero Lunghi Conf. Exhib. Enea, Rome, Italy December 14–16 2011, pp. 59–60.