Journal of Power Sources 193 (2009) 359–366

Contents lists available at ScienceDirect

Journal of Power Sources

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

Catalytically upgraded landfill gas as a cost-effective alternative for fuel cells

W. Urban ∗ , H. Lohmann, J.I. Salazar Gómez
Fraunhofer Institute for Environmental, Safety, and Energy Technology (UMSICHT), Osterfelder Strasse 3, 46047 Oberhausen, Germany

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

Article history: The potential use of landfill gas as feeding fuel for the so-called molten carbonate fuel cells (MCFC) imposes
Received 9 October 2008 the need for new upgrading technologies in order to meet the much tougher feed gas specifications
Received in revised form of this type of fuel cells in comparison to gas engines. Nevertheless, MCFC has slightly lower purity
28 November 2008
demands than low temperature fuel cells. This paper outlines the idea of a new catalytic purification
Accepted 5 December 2008
process for landfill gas conditioning, which may be supposed to be more competitive than state-of-the-art
Available online 24 December 2008
technologies and summarises some lab-scale results. This catalytic process transforms harmful landfill
gas minor compounds into products that can be easily removed from the gas stream by a subsequent
Keywords:
Landfill gas
adsorption step. The optimal process temperature was found to be in the range 250–400 ◦ C. After a catalyst
Catalytic purification screening, two materials were identified, which have the ability to remove all harmful minor compounds
Fuel cells from landfill gas. The first material was a commercial alumina that showed a high activity towards the
Volatile organic compounds removal of organic silicon compounds. The alumina protects both a subsequent catalyst for the removal of
Siloxane removal other organic minor compounds and the fuel cell. Due to gradual deactivation caused by silica deposition,
the activated alumina needs to be periodically replaced. The second material was a commercial V2 O5 /TiO2 -
based catalyst that exhibited a high activity for the total oxidation of a broad spectrum of other harmful
organic minor compounds into a simpler compound class “acid gases (HCl, HF and SO2 )”, which can be
easily removed by absorption with, e.g. alkalised alumina. The encouraging results obtained allow the
scale-up of this LFG conditioning process to test it under real LFG conditions.
© 2008 Elsevier B.V. All rights reserved.

1. Introduction [1] and to mitigate the dependency to non-renewable fuels (e.g.

oil and coal), respectively. This has driven companies and govern-
The main driving force for the utilisation of landfill gas (LFG) is ments to look at renewable sources, such as biogases as an attractive
to avoid greenhouse gas emissions and to lower gas emissions with energy source. Among the various types of biogases, landfill gas
extremely high ozone depletion potential (ODP). Due to the fact that (LFG) appears as a promising but equally challenging alternative for
most landfills are found outside urban areas, the utilisation of heat fuel cells [2]. Its utilisation may not only reduce uncontrolled emis-
from gas engines is barely possible. That is why fuel cells (FC) with sions of greenhouse gases and help to mitigate the dependency to
a better power to heat ratio in comparison to gas engines seem to fossil fuels but also to eliminate the wide range of pollutants found
be very attractive in a prospective view. Additionally, in the recent in this type of gas, which pose a threat to human health. According
years there has been an increasing need for a decentralised power to the German Ministry of Environment (BMU), in the year 2007
supply and the search for alternative sources of energy in order around 1.143 million tonnes of CO2 were saved in Germany due to
to improve the current efficiencies (heat and power utilisation) the utilisation of landfill gas [3].
The main constituents of LFG are methane (40–60%) and carbon
dioxide (35–50%) [4]. Additionally, LFG contains small amounts of
oxygen (0–3%), nitrogen and is saturated with water vapour. The
Abbreviations: MCFC, molten carbonate fuel cell; FC, fuel cell; LFG, landfill high energy content makes it an interesting energy source [5]. Due
gas; ODP, ozone depletion potential; VOCs, volatile organic compounds; CI-MS,
chemical ionisation mass spectrometer; GC-TCD, gas chromatography-thermal con-
to its characteristics, LFG is already widely utilised in gas engines
ductivity detector; GC-ECD, gas chromatography-electron capture detector; GC-FID, and turbines to produce electricity and heat [6].
gas chromatography-flame ionisation detector; TCE, tetrachloroethylene; CFC-113, The application of LFG into molten carbonate fuel cells (MCFC)
1,1,2-trichlortrifluorethane; MPV, multi-position valve; EDX, energy dispersive X-ray would enhance the overall energy conversion efficiency. Compared
analysis; L2, hexamethyldisiloxane; D4, octamethylcyclotetrasiloxane.
∗ Corresponding author. Tel.: +49 208 8598 1124; fax: +49 208 8598 1423.
to other fuel cell types, the MCFC is the most adequate fuel cell
E-mail addresses: [email protected] (W. Urban),
technology for LFG utilisation due to its high operation temper-
[email protected] (H. Lohmann), atures (600–700 ◦ C) and higher flexibility regarding feeding fuels
[email protected] (J.I.S. Gómez). and it also has lower purity demands. MCFCs are uniquely suited

0378-7753/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jpowsour.2008.12.029
360 W. Urban et al. / Journal of Power Sources 193 (2009) 359–366

Fig. 1. Damaged blade wheel of a turbine caused by acid corrosion (left), SiO2 deposition on a cylinder head caused by siloxane decomposition (right), both after more than
3500 h of operation.

for biogas applications, because no significant derating of power evaluated in a field-test plant to oxidise LFG minor compounds to
output is expected due to the methane dilution by CO2 in LFG. In reactions products such as HCl, HF, and SO2 [15] showed encourag-
addition, the presence of CO2 at the cathode raises the electrical ing results. The acidic reaction products were removed by using a
efficiency [7]. MCFC technology is near to commercialisation, and water scrubber device.
the elevated temperatures make the use of noble metal catalysts In general, V2 O5 /TiO2 and V2 O5 –WO3 /TiO2 catalysts are active
unnecessary [8]. for the combustion of VOCs [16] and the most largely used catalytic
Apart from the main components, LFG also contains relatively systems for the oxidation of chlorinated VOCs [17]. Therefore, two
high amounts of hydrogen sulphide and a broad spectrum of volatile V2 O5 /TiO2 -based catalysts were chosen among a large variety of
organic compounds (VOC): organic-sulphur compounds (e.g. car- commercial oxidation catalysts for testing at the lab scale.
bonyl sulphide, mercaptans), silicon-containing compounds (e.g. An alternative way to upgrade LFG to fuel cell quality is the com-
siloxanes), halogenated compounds, aromatics and aliphatic hydro- bination of catalytic hydro-processing and the use of disposable
carbons [4,9]. The utilisation of biogases like LFG for power adsorbents for the removal of the acidic reactions products H2 S
generation may result in damages for the downstream power gen- and HCl [18]. However, this process needs additionally hydrogen
erating units, since during combustion the halogenated compounds as reactant, whereas a catalytic oxidation/hydrolysis process can
and sulphur-containing compounds are transform into acid gases use for the degradation of VOCs, oxygen and water vapour which
like H2 SO4 , HCl and HF, which cause corrosion problems. In addi- are normally contained in LFG. The main problems of using cat-
tion, siloxanes are transformed into micro-crystalline silica, which alytic processes for LFG purification are the irreversible and rapid
deposits on the engine parts contributing to abrasion and poorer deactivation of the catalysts by deposition of siloxane degradation
combustion efficiency. Fig. 1 shows typical damages caused by acid products on the catalyst surface (microcrystalline silica). Thus, a
corrosion and the deposition of siloxane decomposition products previous abatement process for siloxane removal is necessary.
(microcrystalline SiO2 ) on turbine and gas engine parts after 3500 h For this reason Fraunhofer UMSICHT is developing a combined
of operation. This type of damages results in shorter lifetimes for process that consists of three steps: (1) siloxane abatement at
the machinery, more frequent maintenance and subsequently, in around 300 ◦ C, (2) catalytic oxidation/hydrolysis of the remaining
higher operating costs. Due to the lower tolerance to impurities, the trace compounds at 300 ◦ C, and (3) removal of the acidic reaction
utilisation of LFG in conjunction with MCFCs with the present state- products by adsorption/absorption on alkalised media.
of-the-art gas qualities would result in poisoning of the reformer
catalyst in only a few hours of operation. 2. State-of-the-art technologies
For this reason, the harmful trace contaminants have to be
removed by a suitable purification process, before landfill gas can Current state-of-the-art gas treatments fail to upgrade landfill
be used in MCFCs. These trace compounds have been shown to be gas up to a fuel cell quality. There are many reasons for it, but
very harmful for MCFCs [10,11]. Present upgrading processes for the economical factors are perhaps the most decisive, since up
landfill gas such as cooling, adsorption or absorption with inorganic to date no purification system is able to eliminate all the minor
and organic solvents are designed for the conditioning of LFG to a components present in LFG, it resulting in the implementation
combustion engine gas quality. Furthermore, these state-of-the-art of sequential steps, which target specific substances, for exam-
technologies often fail because of technical or economic reasons, ple dehumidification, desulphurisation, siloxane removal and VOCs
for example, low removal efficiencies, or high operational costs removal. Moreover, a purification system must be able to cope with
[12]. They appear to be unsuitable to fulfil both the technical and concentration peaks, which are usually observed in landfill gas for
economic demands of landfill gas utilisation in MCFCs. this kind of pollutants.
Catalytic removal of minor compounds from LFG appears as a
cost-effective and attractive alternative solution for the cleanup of 2.1. Adsorption methods
LFG, since such catalytic process would reduce the complex LFG
matrix of minor compounds into one compound class (acidic gases), Among all the conventional purification processes, the applica-
which can be easily removed by state-of-the-art technologies like tion of active carbon alone or in combination with chilling systems
adsorption or absorption. is perhaps the most common methodology for the removal of pol-
Even though noble metals and metal oxides are widely used to lutants from landfill gas. Although commercial active carbons are
eliminate VOCs by catalytic oxidation [13,14], to our knowledge only fairly cheap, one of the drawbacks of its application is the relatively
a few studies regarding catalytic LFG clean-up have been published. low capacity towards the most damaging substances, such as silox-
The applicability of several commercial oxidation catalysts has been anes, halogenated compounds and sulphur-containing compounds,
 W. Urban et al. / Journal of Power Sources 193 (2009) 359–366 361

although impregnated active carbons are known to have high selec- 2.4. Catalytic process
tivity towards H2 S [19]. It is due to the competing adsorption effects
that take place in the carbon matrix, since the aliphatic hydrocar- Although VOCs are usually destroyed by catalytic methods, a
bons and the aromatics present in landfill gas, which are found in complex matrix such as landfill gas makes their removal more
relatively high concentrations, are favourably adsorbed, thus result- challenging, since many minor components present in LFG act as
ing in shorter breakthrough times for the more harmful substances poisons even for non-noble metal oxides. That may explain why
and therefore higher operational costs. For conventional active car- catalytic systems are relatively unusual at landfill sites. Fraunhofer
bons, the presence of water reduces significantly the adsorption UMSICHT has been designing a catalytic system, which may remove
capacities towards aromatics [20]. all minor components present in landfill gas without compromis-
It is worth mentioning that active carbon is not a definite solu- ing the methane contained in it. Due to the process characteristics
tion from an environmental point of view because the pollutants and the final gas quality and composition it is thought that a cou-
adsorbed have to be disposed after saturation of the material. The pling with a molten carbonate fuel cell would enhance the overall
most common way of treating used active carbon is by burning it, energetic utilisation.
which means that the pollutants are oxidised producing new dan- The process consists of three main stages:
gerous substances, and thus, not really solving the problem, since
it is shifted to the waste incineration plants. 1. Siloxane removal: a cheap catalyst material dedicated to the cat-
Other materials like silica gel have shown comparable siloxane alytic decomposition of siloxanes has been identified, which has
removal capacities (100 mg g−1 ) to active carbons with the advan- shown very good performance and higher capacities than con-
tage of being regenerable by thermal treatment [21]. However, due ventional active carbons at working temperatures around 300 ◦ C.
to their hydrophilicity, the adsorption capacities are remarkably Additionally, this material exhibits catalytic activity towards the
reduced. removal of other minor components such as H2 S, C2 Cl4 , terpenes
and aromatics. Due to the SiO2 deposition, the material has to be
periodically exchanged. Therefore, regeneration of the activated
2.2. Absorption methods
alumina is not possible since the SiO2 microcrystals deposit on
the microporosity damaging irreversibly the active sites of the
Water scrubbing processes are reasonably applicable to gases
material. However, the very low price of this material makes
containing high concentrations of H2 S, recovering sulphur through
regeneration unnecessary.
oxidation. For less concentrated gases this technology seems to
2. In the second stage, the siloxane-free gas, which still con-
be less convenient. Water has a relatively good efficiency for H2 S
tains VOCs and sulphur-containing compounds is oxidised at
removal, but CO2 is also absorbed and hence, for the case of LFG
around 300 ◦ C to form HCl, HF and SO2 over a vanadium-based
it is unsuitable. Moreover, large water volumes are needed and
catalyst without compromising the methane content. Regener-
reboiling is necessary for regeneration. For other LFG impurities
ation of the vanadium-based catalyst is in principle possible
very poor removal efficiencies are achieved because of their low
as long as oxygen is present in the system in excess in order
solubilities in water and high vapour pressures. Absorption meth-
to oxidise the VOCs. Thus, in situ regeneration may take place
ods have been applied for the removal of target substances like
enhancing the lifetime of the catalyst. However, a regenera-
siloxanes, where concentrated solutions of acids and bases have
tion step at the second stage has been considered as a future
shown to have moderate removal efficiencies or the application of
prospect.
organic solvents at high temperatures. Nevertheless, in all cases
3. As last stage, an alkalised material can be used to selectively
their utilisation complicates plant design due to safety concerns
remove the acid gases produced during the oxidation process.
and hence incurring in higher operational costs. Moreover, for the
As a result, the previously complex LFG matrix has been reduced
case of alkaline solutions, the presence of carbon dioxide in land-
to the simpler gas mixture CH4 /CO2 . A simplified sketch of the
fill gas makes its implementation unpractical. The application of gas
catalytic process is depicted in Fig. 2. The acid gas removal is
scrubbing with organic solvents have shown to be disadvantageous
a state-of-the-art technology and therefore, not investigated in
since the more volatile siloxanes, such as hexamethyldisiloxane (L2)
detail since all the market available products exhibit very good
or trimethylsilan may be stripped into the gas phase at high gas flow
absorbing capacities.
rates [12].
The reaction temperature was so chosen since a shift in
2.3. Chilling processes behaviour in a higher range of temperatures is expected since the
catalyst stability may be affected by thermal effects like sintering.
The application of cooling systems is usually carried out in order Moreover, deactivation by high temperature reactions may occur.
to remove water and thus avoiding corrosion damages in engine Additionally, the methane – which is the fuel – may be oxidised and
parts. Parallel to water removal other substances, such as heavy depleted considerably at higher temperatures. Furthermore, due to
aromatics and siloxanes can be at least partially removed. It has the oxygen excess (in relation to the minor compounds), a change in
been shown [12] that at −25 ◦ C a removal of 25.9% for volatile temperature plays a relatively small role for the oxidation of minor
methyl siloxanes (VMS) is possible and that a removal of around compounds.
99% can only be achieved by freezing temperatures as low as −70 ◦ C. By changing the space velocity only conversion of the minor
According to Schweigkofler [22] 89% of the silicon compounds in a compounds was lower but no by-products were observed. Besides,
dried gas can be removed by a cooling temperature of just 5 ◦ C. it has been observed that as long as oxygen is present in excess
Higher removal efficiencies are achieved at lower temperatures. no by-products are formed and therefore, other temperatures are
Previous studies carried by Siloxa Engineering AG and Fraunhofer of no interest due to methane oxidation. Large space velocities
Institute UMSICHT [23] have shown that freeze-cooling systems for are no profitable and low space velocities result in long residence
the removal of siloxanes are only cost-effective in combination with times.
active carbon in the scenario of high flow rates with relatively high In addition, this catalytic purification process was developed
loads of siloxanes. In general, the energetic requirements for the because the waste heat produced by the fuel cell (∼480 ◦ C) can be
application of deep chilling systems are so high that such process, used to pre-condition the raw landfill gas (20 ◦ C), thus enhancing
as experience for the case of sewage gas [24] is not profitable. the overall efficiency of the process. A description of the gain in effi-
362 W. Urban et al. / Journal of Power Sources 193 (2009) 359–366

were chosen as representative of the LFG minor compounds, was

to choose those minor compounds that are either chemically stable
(more challenging to convert) or found in relatively high concen-
trations in LFG from every compound class. Availability of the
chemicals was another important criterion since for the case of
many chlorofluorocarbons their purchase is not longer possible
due environmental regulations. The authors considered the chosen
model compounds as good representatives of the different com-
pound classes found in LFG, for example, benzene as stable aromatic
compound, R113a as a stable and still available FCC, H2 S as the
sulphur-containing compound present in higher concentration in
LFG and L2 and D4 as typical siloxanes (linear and cyclic) found in
LFG.
All model compounds were purchased from Sigma–Aldrich with
reagent grade. Hydrogen sulphide (1000 ppm) supplied by Air Liq-
uide was added directly from a gas cylinder into the feed gas flow.
All the other gases were purchased from Linde Gas with the follow-
ing technical qualities: N2 (5.0), CH4 (2.5), CO2 (n.a.), O2 (3.5). With
this setup it was possible to generate dry or humid simulated LFG.
The gas composition of the simulated LFG at the inlet and outlet
was monitored with a chemical-ionisation mass spectrometer (CI-
MS) model Airsense 500 from the company MS4-Analysentechnik,
Rockenberg, Germany. Three ionisation gases (Xe, Kr and Hg) were
used as ionisation agents depending on the ionisation energies of
the different compounds. The advantage of the CI-MS is that one
measures directly the molecular mass and no fragments as with
conventional electron ionisation MS. This allows a simultaneous
Fig. 2. Simplified sketch of the three-step catalytic process for LFG conditioning. online measurement of reactants and products.
Controlling software (HP data acquisition in lab, Siemens SPS S7
in the field test) was used to generate the desired gas mixtures by
ciency through waste heat utilisation has been shown elsewhere
controlling the mass flow controllers (Type MFC-D-5111 from M&W
[25,26].
Instruments) and to set the temperatures at the reactor, pipe lines
and humidifier. In the field-test facility, the real LFG was obtained
3. Experimental from the landfill site provider through a pipeline installed before the
dehumidification system. The LFG was then compressed and guided
3.1. Materials and methods to the reactors in the container. The test facility could be operated in
series or parallel. To monitor the changes in concentration before
The lab-experiments were carried out in a custom-made setup and after every reactor a multi-position valve (MPV) was used to
(see Fig. 3). With this setup it was possible to monitor online the select the desired gas for the analysis. In order to accelerate the
oxidation of the model compounds of a simulated landfill gas. The breakthrough of the selected siloxanes a similar evaporation system
model substances used for the tests to represent the typical minor as that one used in the laboratory was utilised for the enrichment
compounds were hydrogen sulphide (H2 S), benzene, toluene, tetra- of the LFG before reactor 1. The gas flow rate at the inlet of reac-
chloroethylene (TCE) and the chlorofluorocarbon CFC-113, which tor 1 was around 18.5 Nl min−1 and the inlet concentrations of the
were enriched into the system by conventional evaporation with enriched siloxanes L2 and D4 were ca. 230 ppmv and ca. 110 ppmv,
a similar setup as that one used by Schweigkofler [22]. Among respectively. To determine the trace compound concentrations at
different factors, one of the reasons why the selected substances the reactor inlet and outlet the CI-MS was used alone or in combina-
tion with two gas chromatographers, one equipped with a thermal
conductivity detector (Micro-GC-TCD, “Quad Series” from Agilent)
for the measurement of the LFG main components and an Agilent
6890 GC-FID/ECD (with a flame ionisation detector and a electron
capture detector) for the analysis of the trace compounds.
The surface area of the different catalysts tested was charac-
terised by nitrogen adsorption at 77 K following the BET model after
degassing for 4 h at 200 ◦ C in order to eliminate water and any other
substance adsorbed at the surface. Energy dispersion X-ray charac-
terisation was used to determine the capacities of the exhausted
materials.

4. Results and discussion

4.1. Lab-scale experiments

4.1.1. Catalytic oxidation of VOCs

Fig. 3. Sketch of a custom-made setup used for monitoring the catalytic oxidation Two commercial V2 O5 –TiO2 catalyst were tested in a lab-scale
of model LFG minor components. tubular reactor (see Fig. 3) of dimensions: length = 500 mm and
 W. Urban et al. / Journal of Power Sources 193 (2009) 359–366 363

Table 1 For the oxidation of aromatic compounds such as benzene, it

Characteristics of the catalysts chosen for VOC oxidation.
is essential the presence of coordinatively unsaturated cations and
V2 O5 [wt.%] TiO2 [wt.%] Diameter [mm] BET surface [m2 g−1 ] anions at the catalyst surface. To break an aromatic nucleus a V5+
Catalyst A 12.0 75.0 4.0 200
cation should be surrounded by several oxygens which make an
Catalyst B 20.5 Balance 1.6 47 abstraction of hydrogens with a rupture of an aromatic ring pos-
sible [29]. It is to be expected that hydroxylation of coordinatively
unsaturated cations and anions by water suppress the oxidation
internal diameter = 17 mm. Table 1 summarises important prop- of aromatic compounds as indicated by the conversion rates for
erties of the catalysts. Catalyst B is composed by V2 O5 and TiO2 benzene in Table 3.
only, whereas catalyst A contained also significant amounts of WO3 , To verify the different effects of water vapour and oxygen on the
MnO2 , CuO, and Fe2 O3 . These metal oxides may act as promoters to conversion of LFG minor compounds, further experiments have to
enhance the activity of the catalyst. be conducted in a field-test facility. Interestingly, the conversion
The simulated LFG (45% methane, 25% carbon dioxide, 1% oxy- rate of TCE seems to be almost independent of the experimen-
gen, balance nitrogen) was obtained by mixing different volume tal conditions. This could indicate that there is no competitive
flow rates of methane, carbon dioxide, nitrogen and oxygen via adsorption between water and TCE for the active centres or TCE
mass flow controllers. Experiments were carried out with and with- can use active centres both before and after hydroxylation by
out water vapour. The reaction temperature was always 300 ◦ C to water.
ensure both a sufficient activity of the catalysts and avoid any degra- The selectivity of the catalysts for the formation of complete
dation of methane. In Table 2 are listed the results for the oxidation oxidation products could only be determined in a qualitative way.
of minor compounds with catalyst A. SO2 and H2 O were the only reaction products that were detected
Catalyst B was tested with a simulated landfill gas. It showed during the oxidation of H2 S. Benzene was oxidised to CO2 , CO
comparable conversion rates for most of the used model substances and H2 O. The oxidation/hydrolysis of the halogenated model com-
in presence of oxygen (see Table 3), except for CFC-113. It was pounds TCE and CFC-113 leads to the formation of the desired
expected that testing the activity of the catalyst B in presence of reaction products CO2 , HCl, HF. However, partially oxidised reac-
water would show similar results compared to catalyst A. For this tions products such as COCl2 , COF2 and Cl2 were detected when
reason, the tests with catalyst B were not continued. the gas mixtures were at under-stoichiometric conditions. These
Depending on the reactant (oxygen or water) in the feed gas, products are both toxic and hard to remove by a subsequent adsorp-
catalyst A showed clearly for benzene and CFC-113 different conver- tion process. For this reason, one of the most important tasks in
sion rates. It is apparent that water vapour is necessary as reactant prospective catalyst tests is the quantitative determination of the
to obtain the maximum activity of catalyst A for the degradation product distribution after oxidation/hydrolysis of the LFG minor
of CFC-113. In contrast, benzene was only catalytically converted compounds. For further experiments in a field-test plant, catalyst
when an excess of oxygen was present in the water-free feed gas. A has been chosen, because it showed even for a stable compound
It is well known that the presence of water in the feed gas leads like CFC-113 significant conversion rates. The results obtained are
to a change of the active centres at the surface of a metal oxide very encouraging but it is clear that improvements have to be
catalyst. The dissociative adsorption of water on the coordinatively made in order to achieve fuel cell gas quality. Therefore, parallel
unsaturated cations and anions at the catalyst surface leads to the experiments were carried out in a field-test facility with real LFG
formation of Brønsted acid centres and causes the disappearance in order to test the performance of the selected catalyst towards
of Lewis acid centres [27]. VOC oxidation and a new chosen material for the removal of
For the decomposition of CFC-12 on a TiO2 catalyst in presence siloxanes.
of water a suggested mechanism starts with the adsorption of CFC-
12 on a Brønsted acid centre followed by a reaction of the adsorbed 4.1.2. Siloxane removal
CFC-12 with neighbouring surface hydroxyls to produce intermedi- During the lab-scale tests, a field-test facility was built in
ates such as COCl2 , which further react to form complete oxidation order to test catalysts with real LFG towards siloxane removal.
products [28]. The results in the present study seem to confirm in The main components of the test facility are two tubular reactors
a qualitative manner that CFC-113 will be decomposed via a simi- (length = 760 mm, internal diameter = 60.3 mm). The reactor inlets
lar mechanism. A detailed investigation of the reaction mechanism and outlets are connected via a multi-position valve (MPV) to a
was not the aim of this study. mass spectrometer so that the gas composition at these four points

Table 2
Catalytic activity of catalyst A towards removal of model minor compounds.

Model compound Concentration [ppmv] Carrier gas N2 [ml min−1 ] O2 [ml min−1 ] H2 Ov [ml min−1 ] MC:reactant Conversion [%]

TCE 213 504 10 1:91 97

TCE 208 504 0.8 1:73 88
CFC-113 183 504 10 1:106 78
CFC-113 185 504 0.8 1:8 40
Benzene 202 504 10 1:76 99.8
Benzene 200 504 0.8 1:6 ∼0
H2 S (with 100 ppmv O2 ) 45 504 1:2 ∼100

Table 3
Catalytic activity of catalyst B towards removal of model minor compounds.

Model compound Concentration [ppmv] LFG total [ml min−1 ] O2 fraction [ml min−1 ] Conversion [%]

TCE 219 1000 10 93.2

CFC-113 189 1000 10 1.4
Toluene 351 1000 10 99.8
H2 S 107 1000 10 ∼100
364 W. Urban et al. / Journal of Power Sources 193 (2009) 359–366

Fig. 5. Conversion rates of the siloxanes L2 and D4 on activated alumina.

Fig. 4. Simplified diagram of the field-test facility.

can be individually measured. LFG is fed in the test facility by a with the siloxanes D4 (octamethylcyclotetrasiloxane) and L2 (Hex-
compressor. A simplified diagram of the field-test plant is shown in amethyldisiloxane) as model compounds to shorten the test times.
Fig. 4. A carrier gas (5 l h−1 nitrogen) flowed parallel through two tem-
The LFG used in the field-test experiments was characterised in pered gas washing bottles that contained L2 and D4 (referred to as
order to know the composition and time variation of the major and saturator unit in Fig. 4). The enriched nitrogen gas flow was mixed
minor compounds. Table 4 summarises the characterisation results. with LFG (1.11 m3 h−1 ). The resulting initial concentrations in the
From Table 4 it can be seen that the major compounds exhibit sim- feed gas were 100 ppm D4 and 200 ppm L2. The reactor temperature
ilar values as those reported in literature. For the case of minor was 300 ◦ C.
compounds, the LFG shows relatively high loadings of H2 S, aromat- Among all materials tested it was observed that the activated
ics and terpenes. The two analyses were carried out with a time alumina exhibited the highest activity towards siloxane removal.
difference of nearly a month. Fig. 5 shows the conversion rates of the degradation of the model
A test series was conducted in order to find a suitable and siloxanes L2 and D4. The initial conversion rate was 100% for both
relatively cheap material, which abates siloxanes. A screening of siloxanes, which gradually decreased due to the proceeding deac-
different materials included one alkalised bentonite, two bleaching tivation of the activated alumina. After 35 h the conversion rate for
earths, corundum, two titanium dioxide catalysts, an aluminium sil- D4 dropped to approximately 9% and for L2 around 20%.
icate and an activated alumina. In each case, 250 g of the materials The activated alumina had the best performance for the con-
were filled up into the first reactor that was operated individually. version of L2, but the decrease in reactivity was not so fast in
The materials were tested with LFG that was additionally enriched comparison with the decline of conversion activity for D4. It might
be assigned to the fact that D4 produces per molecule twice as much
SiO2 as L2, thus the number of active centres at the surface of the
Table 4
Determination of the concentrations of major and minor components in a dry LFG activated alumina that are deactivated are theoretically two times
and their time variation. higher. The initial conversion rates of L2 with the other materials
were lower than the L2 conversion rate of the activated alumina
Major components Analysis I [vol.%] Analysis II [vol.%]
and decreased in the first hours very quickly. Thereafter, the con-
Methane 54 ± 1 53.3 version rates stabilised, so that after 40 h conversion rates between
Carbon dioxide 35 ± 1 32.9
Oxygen 0.89 ± 0.05 0.7
15% (bleaching earth I) and 35% (bleaching earth II) were measured.
Nitrogen 11.0 ± 0.5 13.0 In the framework on this material screening, a detailed descrip-
tion of the possible conversion mechanisms of the siloxanes L2 and
Minor components Analysis I [mg m−3 ] Analysis II [mg m−3 ] D4 or the identification of the active centres was of no interest.
Hydrogen sulphide 230.0 280.0 The fundamental aim of this work is to find a cheap and high-
Isopropylbenzene 5.6 3.3 performance material for the removal of siloxanes with a sufficient
Limonene 25.0 15.8 lifetime.
Pinene 83.7 85.3
The silicon content of fresh and used activated alumina samples
Benzene 3.4 4.7
Toluene 21.0 18.4
was measured by energy dispersive X-ray analysis (EDX). The BET
Xylene 45.0 35.6 method was used to determine the specific surface area. The vol-
Trimethylbenzene 13.4 8.0 umes of the micro- and mesopores were obtained by analysing the
Tetrachlorethene 0.14 0.3 nitrogen desorption isotherm with the methods of Horvath Kawa-
Methanethiole 0.56
zoe (HK) and Barrett, Joyner and Hallender (BJH). The macro pores
Chlorobenzene 0.22
Trichloroethene 0.24 0.5 volume was quantified by Hg porosimetry. The results are listed in
D3 0.84 0.5 Tables 5 and 6.
D4 15.0 10.6 The EDX results clearly show that due to the deposition of silox-
D5 3.0 3.3
anes degradation products the silicon concentration of the used
L2 5.0 3.1
L4 <0.005 <0.1
sample increased. Besides the specific surface area, the micro- and
mesopores volumes of the activated alumina were also drastically
Trimethylsilanol 12 11.0
reduced, whereas the macropores volumes are nearly unaltered.
 W. Urban et al. / Journal of Power Sources 193 (2009) 359–366 365

Table 5
EDX results for fresh and used activated alumina.

Activated alumina C [wt.%] O [wt.%] Al [wt.%] Si [wt.%] S [wt.%]

Fresh sample 5.42 51.78 42.68 0.13 –

Used sample 4.56 46.34 39.08 9.61 0.22

Table 6
Pore volumes and specific surface area of fresh and used activated alumina.

Activated alumina Micropores volume Mesopores volume Macropores volume Specific surface
[mm3 g−1 ] [mm3 g−1 ] [mm3 g−1 ] [m2 g−1 ]

Fresh sample 110.3 110.8 33.3 268

Used sample 4.56 46.34 39.08 9.61

This indicates that the macropores are responsible for the trans- formed into Brønsted bases due to the adsorptive dissociation of
portation of the siloxanes into the pore system. The degradation of water.
siloxanes followed by the deposition of the reaction products takes
place in the micro- and mesopores.
5. Conclusions
Besides the solid reaction products two volatile reaction
compounds were detected at very low concentrations: D3 (hexam-
For the development of a catalytic LFG upgrading process
ethylcyclotrisiloxane) and trimethylsilanol (TMS). The formation of
two materials were identified, which showed a high potential
volatile reaction products is undesired, because they also may deac-
for the removal of all LFG minor compounds that are harm-
tivate the subsequent metal oxide catalyst. Further experiments are
ful to MCFCs. Depending on the experimental conditions a
being carried out in order to confirm these findings. The activated
commercial V2 O5 –TiO2 catalyst possessed good activity for the oxi-
alumina also exhibited moderate activity for other LFG impurities
dation/hydrolysis of organic and inorganic model compounds. A
that were monitored during the test (see Fig. 6). For example, the
commercial activated alumina was able to remove high concen-
initial conversion rate for H2 S is around 87% but decreases rapidly
trations of volatile siloxanes, which have a deleterious effect on
and after approximately 10 h, the conversion is nearly zero. Many
both, the V2 O5 –TiO2 catalyst and a downstream MCFC. In further
factors may play a role for this finding, since oxygen shortages may
experiments, the selectivity of the degradation reactions for desired
reduce the ability of the activated alumina to regenerate the active
products such as HCl, HF and SO2 must be quantitatively deter-
centres, thus resulting in poisoning by the accumulation of sul-
mined. Also the effects of the reactants water and oxygen for the
phur at the surface or competing adsorption effects may also take
complete oxidation of the different LFG minor compounds must be
place. Moreover, the parallel deposition of SiO2 reduces the num-
studied in detail in order to optimise the process. Finally, long-time
ber of active centres available at surface. For pinene and limonene,
tests must be conducted with real LFG to measure the lifetime of
which are usually found in LFG in relatively high concentrations
the materials and determine removal capacities.
(∼20–100 mg Nm−3 ), the conversions were in the range 60–70%
According to the results, the catalytic purification process intro-
and 45%, respectively. Their conversion seems to be independent of
duced could become a cost-effect alternative for LFG upgrading in
the surface area reduction caused by the SiO2 deposition. Similarly,
contrast to traditional methods, but in order to prove it, field-test
halogenated and aromatic compounds exhibit relatively constant
are underway.
conversions, it indicating that the active sites at surface responsible
for their conversion are still available, although it is assumed that
in the presence of water the Lewis-acid centres (Al3+ ) are trans- References

[1] I. Staffell, R. Green, K. Kendall, J. Power Sources 181 (2008) 339–349.

[2] R.J. Spiegel, J.C. Trocciola, J.L. Preston, Energy 22 (1997) 777–786.
[3] http://www.bmu.de/allgemein/aktuell/160.php.
[4] S. Rasi, A. Veijanen, J. Rintala, Energy 32 (2007) 1375–1380.
[5] R. Bove, P. Lunghi, J. Power Sources 145 (2005) 588–593.
[6] R. Bove, P. Lunghi, Energy Convers. Manage. 47 (2006) 1391–1401.
[7] D. Rastler, King County Carbonate Fuel Cell Demonstration Project. Case Study
of a 1-MW Fuel Cell Power Plant Fuelled by Digester Gas. Interim Report, EPRI,
Palo Alto, 2005.
[8] L. Lombardi, E. Carnevale, A. Corti, Energy 31 (2005) 3208–3219.
[9] M.R. Allen, A. Braithwaite, C.C. Hills, Inter. J. Environ. Anal. Chem. 62 (1996)
43–52.
[10] C. Tomasi, M. Baratieri, B. Bosio, E. Arato, P. Baggio, J. Power Sources 157 (2006)
765–774.
[11] C.H. Bartolomew, Appl. Catal. A 212 (2001) 17–60.
[12] R. Dewil, L. Appels, J. Baeyens, Energy Convers. Manage. 47 (2006) 1711–1722.
[13] M. Guillemot, J. Mijoin, S. Mignard, P. Magnoux, Appl. Catal. B 75 (2007)
249–255.
[14] S. Heneghan, G.J. Hutchings, S.H. Taylor, Catalysis 17 (2004) 105–151.
[15] E. Jacob, R. Richter, Final Report, MAN Technologie AG, München, 1990 (in Ger-
man).
[16] E. Finocchio, M. Baldi, G. Busca, C. Pistarino, G. Romezzano, F. Bregani, G.P.
Toledo, Catal. Today 59 (2000) 261–268.
[17] F. Bertinchamps, Total Oxidation of Chlorinated VOCs on Supported Oxide Cat-
alysts, PhD Thesis, Universite catholique de Louvain, 21, 2005.
[18] C. He, D.J. Herman, R.G. Minet, T.T. Tsotsis, Ind. Eng. Chem. Res. 36 (1997)
4100–4107.
[19] A. Bagreev, S. Katikaneni, S. Parab, T.J. Bandosz, Catal. Today 99 (2005) 329–337.
Fig. 6. Catalytic conversion of other LFG impurities measured during the siloxane [20] H.-C. Shin, J.-W. Park, K. Park, H.-C. Song, Environ. Pollut. 119 (2002) 227–236.
decomposition tests with activated alumina. [21] M. Schweigkofler, R. Niessner, J. Hazard. Mater. B83 (2001) 183–196.
366 W. Urban et al. / Journal of Power Sources 193 (2009) 359–366

[22] M. Schweigkofler, Bestimmung flüchtiger Siliziumverbindungen in Biogases [25] Company Ansaldo: http://www.ansaldofuelcells.com.
mittels Kanisterprobenahme und GC-MSD/AED Analytik, Thesis, Tehnical Uni- [26] F. Orecchini, E. Bocci, A. Di Carlo, J. Power Sources 160 (2006) 835–841.
versity Munich, 2001 (in German). [27] I. Chorkendorff, J.W. Niemantsverdriet, Concept of Modern Catalysis and Kinet-
[23] http://www.dbu.de/PDF-Files/A-14726.pdf (in German). ics, Wiley–VCH, 2003.
[24] D. Rossol, K.-G. Schmelz, Siloxane im Faulgas, Wasser Abwasser 146 (2005) [28] H.L. Greene, S. Karmakar, J. Catal. 151 (1994) 394–406.
55–61 (in German). [29] B. Grzybowska-Swierkosz, Appl. Catal. A: Gen. 157 (1997) 409–420.