On-stream regeneration of a sulfur-poisoned ruthenium-carbon catalyst under hydrothermal gasification conditions

Article abstract of DOI:10.1002/cctc.201300791

Catalytic processes that employ Ru catalysts in supercritical water are capable of converting organics, such as wood waste or biosolids, into synthetic natural gas (CH₄) with high efficiencies at relatively moderate temperatures of around 400 °C. However, Ru catalysts are prone to S poisoning and are quickly deactivated. As S is ubiquitous in raw biomass and technologies to remove S from hydrothermal biomass feeds are lacking, regeneration protocols that efficiently reactivate S-poisoned catalysts are required to realize efficient conversion processes and long catalyst lifetimes. In this work, we developed a method to remove S from a S-poisoned Ru catalyst under hydrothermal conditions through an oxidative treatment in the aqueous phase. By using in situ X-ray absorption spectroscopy under the reaction conditions, we show that Ru is oxidized by dilute H₂O₂ at low temperatures, which leads to the removal of adsorbed S species from the catalyst surface. By optimizing the regeneration conditions, it was possible to prevent oxidation of the catalyst carbon support, as revealed by ex situ TEM. This treatment led to a reactivation of the Ru catalyst with a significant increase in carbon-to-gas conversion and methane selectivity. Don't preach, bleach! Dilute hydrogen peroxide effects the reactivation of a sulfur-poisoned ruthenium catalyst under hydrothermal conditions. This mild oxidative treatment efficiently removes adsorbed sulfur from the ruthenium surface and restores the catalytic activity without corroding the catalyst support. Copyright

Full text of DOI:10.1002/cctc.201300791

CHEMCATCHEM
FULL PAPERS
DOI: 10.1002/cctc.201300791
On-Stream Regeneration of a Sulfur-Poisoned Ruthenium–
Carbon Catalyst Under Hydrothermal Gasification
Conditions
[
a]
[a]
[a]
[a]
Marian Dreher, Matthias Steib, Maarten Nachtegaal, Jçrg Wambach, and
[a, b]
Frꢀdꢀric Vogel*
Catalytic processes that employ Ru catalysts in supercritical
water are capable of converting organics, such as wood waste
drothermal conditions through an oxidative treatment in the
aqueous phase. By using in situ X-ray absorption spectroscopy
under the reaction conditions, we show that Ru is oxidized by
dilute H O at low temperatures, which leads to the removal of
or biosolids, into synthetic natural gas (CH ) with high efficien-
4
cies at relatively moderate temperatures of around 4008C.
However, Ru catalysts are prone to S poisoning and are quickly
deactivated. As S is ubiquitous in raw biomass and technolo-
gies to remove S from hydrothermal biomass feeds are lacking,
regeneration protocols that efficiently reactivate S-poisoned
catalysts are required to realize efficient conversion processes
and long catalyst lifetimes. In this work, we developed
a method to remove S from a S-poisoned Ru catalyst under hy-
2
2
adsorbed S species from the catalyst surface. By optimizing the
regeneration conditions, it was possible to prevent oxidation
of the catalyst carbon support, as revealed by ex situ TEM. This
treatment led to a reactivation of the Ru catalyst with a signifi-
cant increase in carbon-to-gas conversion and methane
selectivity.
Introduction
The reforming of wet biomass in near- and supercritical water
forming and the methanation reactions necessary to convert
[2–4]
(NCW, SCW) to produce combustible gases like methane or hy-
complex biomolecules into methane.
However, as virtually
drogen has been suggested as a viable option to produce
all biomass contains S, usually bound in proteins, the suscepti-
bility of Ru to S poisoning is a major challenge for efficient
[
1–3]
green and sustainable energy.
In this process, water is used
[2]
as reactant and reaction medium at the same time, which
breaks down and dissolves large organic molecules to form
a uniform, supercritical phase. This process makes the drying
of biomass obsolete and thus represents a highly efficient con-
version technique. Several NCW and SCW reforming processes
are currently under investigation, which include the noncata-
and continuous catalytic biomass reforming.
Recently, we established the mechanism of S poisoning and
the structure of a S-poisoned Ru/C catalyst (S-Ru/C) by com-
bining structural analysis by in situ X-ray absorption fine struc-
[5,6]
ture (XAFS) with isotope labeling of the reaction products.
As biomass typically contains S organically bound in proteins,
DMSO was used as a water-soluble, organic S-containing
model compound. During supercritical water gasification
(SCWG) in the presence of S, the surface of the ~1.5 nm Ru
particles is covered with S, which leads to a catalyst phase
with a RuS0.33 stoichiometry (about 40% surface coverage). The
adsorbed S irreversibly blocks the active sites of the catalyst,
which decreases its ability to break CÀC and CÀO bonds and
leads to a severe decrease in the activity for the reforming and
methanation of organic compounds (both large biomolecules
lytic reforming of biomass to produce a H -rich product gas as
2
well as catalytic processes that aim to produce a product gas
[
1–3]
with a high methane content.
The catalytic processes typi-
cally employ Ni- and Ru-based catalysts, in which Ru is more
resistant to sintering and presents high activity for both the re-
+
[
a] M. Dreher, M. Steib, Dr. M. Nachtegaal, Dr. J. Wambach, Prof. Dr. F. Vogel
Paul Scherrer Institute
Laboratory for Bioenergy and Catalysis
5
232 Villigen PSI (Switzerland)
and rather small molecules such as ethanol). By using D O for
2
Fax: (+41)56-310-2199
E-mail: frederic.vogel@psi.ch
SCWG instead of H O, deuterium-labeling of the methane pro-
2
duced in the process is realized, which results in CH D spe-
x
4Àx
[
b] Prof. Dr. F. Vogel
cies in the product gas. Once desorbed from the catalyst,
methane is then hardly readsorbed on the Ru catalyst and
does not show any appreciable H–D exchange within the time-
frame of the gasification experiments (residence times in the
Fachhochschule Nordwestschweiz FHNW
Hochschule fꢀr Technik
Institut fꢀr Biomasse und Ressourceneffizienz
5
210 Windisch (Switzerland)
E-mail: frederic.vogel@fhnw.ch
[5]
reactor of less than 1 min), as shown in a previous study.
+
[
] Current address:
Technical University Munich
Department of Chemistry
Therefore, it was then possible to draw inferences from the
composition of the produced methane on the amount of hy-
8
5748 Garching (Germany)
drocarbon adsorbates (CH ) on the catalyst surface that are
x
ꢁ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 626 – 633 626

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
formed by the degradation of organic molecules on the cata-
temperatures and can therefore be considered superior to O2,
which only forms a homogeneous phase with water under
SCW conditions and is typically used in supercritical water oxi-
lyst. The gasification of organic compounds in D O over an
2
active Ru/C catalyst predominantly led to the formation of CD4
[10]
and CHD , indicative of C* and CH* adsorbates on the catalyst
dation (T>5008C). The effect of H O on the structure of the
2 2
3
[
5,6]
surface.
However, gasification over S-Ru/C led to a severe
Ru catalyst during and after oxidative treatment was probed
by in situ XAS under realistic process conditions. Structural
changes of the catalyst support at different regeneration con-
ditions were studied by ex situ electron microscopy and N2
physisorption. To conclude, we present a protocol for a mild,
liquid-phase oxidation of the poisoned catalyst that removes
adsorbed S from the active catalyst phase whilst keeping the
catalyst and catalyst support intact.
decrease in methane production, with much lower CD and in-
4
creased CHD , CH D, and CH concentrations, which shows
3
3
4
that S poisoning has a significant influence on the surface re-
action mechanisms. Hence, the formation of hydrocarbon ad-
sorbates—the precursors for methane formation—is influ-
enced by S poisoning, which results in a change in catalyst se-
lectivity and ultimately in the composition of D-labeled meth-
ane. A kinetic model for the deactivation of a Ru/C catalyst
during the gasification of algae was presented by Guan et al.,
which suggests that S poisoning is the chief cause for catalyst
Results and Discussion
[
7]
deactivation.
In situ XAS combined with isotope labeling was used to study
the on-stream oxidative regeneration of a S-poisoned Ru cata-
lyst (S-Ru/C), both from a structural and mechanistic point of
Osada et al. studied the effect of S poisoning during the hy-
drothermal gasification of lignin over Ru/TiO by soaking the
2
catalyst in H SO4 at room temperature prior to gasification.
view, during the SCWG of 7.5% ethanol in D O. The Fourier-
2
2
This H SO treatment led to a strong decrease in catalytic activ-
transformed extended X-ray absorption fine structure (EXAFS)
spectra of the catalyst and the composition of the produced
D-labeled methane collected during the four steps of a poison-
ing–regeneration cycle (catalyst activation, S poisoning, oxida-
tive regeneration, and reactivation) are presented in Figure 1.
Under the reaction conditions (4008C, 24.5 MPa), the active
2
4
ity. After washing the deactivated catalyst samples with water
at different temperatures (sub- to supercritical) prior to the
gasification experiments, they observed higher activities than
without the washing step. The authors concluded that the S-
poisoned catalyst can be partially regenerated (recovering ap-
proximately 50% of the original activity) by washing with sub-
0
catalyst is composed of fully reduced Ru particles with an
[
8]
[6]
critical water at 3008C and 25 MPa. However, it remains un-
average diameter of 1.5 nm (determined by EXAFS and TEM).
clear how relevant this H SO treatment at room temperature
In this state, the EXAFS spectrum of the catalyst only shows
metallic RuÀRu coordination shells (CS), with the first CS at
2.66 ꢂ and a much less pronounced second CS at 3.78 ꢂ
(Table 1; Figure 1, top left).
2
4
is with regards to S poisoning under hydrothermal conditions
under which irreversibly bound sulfide species are formed on
[
6]
the Ru surface.
Waldner studied the gasification of ethanol in supercritical
water over a Ru/C catalyst at low space velocities and full
The methane produced from ethanol gasification on the
active catalyst showed a very high CD content with lower,
4
[
9]
carbon-to-gas conversion. After full deactivation of the cata-
lyst by adding sodium sulfate to the ethanol feed, the S-pois-
oned catalyst was treated with dilute H O at 908C and subse-
almost equal amounts of CHD and CH D (Figure 1, center).
3
2
2
This is indicative of an active catalyst that breaks ethanol
down to C* and CH* adsorbates, as outlined in the introduc-
2
2
[6]
quently, the release of ionic species in the liquid effluent was
observed. After this treatment, almost full conversion of etha-
nol was again observed, but the conversion started to slowly
decrease to approximately 80% within 24 h. Although this oxi-
dative catalyst treatment showed promising results, it remains
unknown how the H O interacts with the catalyst. As Wald-
tion and in a previous publication. Upon S poisoning (by the
2À
addition of 250 ppm of DMSO to the feed), sulfide (S ) species
were created on the Ru surface, which resulted in the forma-
tion of RuÀS bonds that are well resolved in the EXAFS spec-
[6]
tra. The adsorption of S was irreversible and led to a perma-
nently poisoned catalyst. However, the Ru particles retained
their metallic character as the first RuÀRu CS remained the
dominant feature in the spectrum. In addition, a RuÀS CS is
visible at 2.35 ꢂ (Table 1; Figure 1, top right). S poisoning led
to a decrease in the carbon-to-gas conversion and, more intri-
2
2
ner’s experiments were run at full carbon-to-gas conversion, it
is unclear how much of the catalytic activity was regained by
this method of regeneration. As a reason for the observed re-
deactivation, the slow release of residual S species stored in
the catalyst support was suggested, which lead to a new S poi-
soning of the catalyst.
guingly, to a significant decrease in CD concentration, where-
4
as the CHD , CH D, and CH production increased (Figure 1,
3
3
4
[6]
To establish a knowledge-based approach to catalyst regen-
eration, in situ studies are necessary to probe the catalyst in its
actual state under process conditions. Following the approach
by Waldner, we studied the regeneration of a Ru/C catalyst
used in hydrothermal processing after its deactivation by S
poisoning, using dilute H O as an oxidizing agent that can
center).
The removal of the adsorbed S was realized by using an
aqueous phase, oxidative treatment with dilute H O . During
2
2
this oxidative regeneration, the reactor temperature was set to
1258C and the system pressure was maintained at 24.5 MPa.
As the S-poisoned Ru/C catalyst was exposed to 3% H O in
2
2
2
2
remove S from the catalyst surface. As a liquid-phase oxidant,
H O , although relatively expensive, allows for intimate contact
water, in situ XAS spectra were recorded. Under these condi-
tions, EXAFS measurements showed that S-Ru/C was fully oxi-
2
2
with the catalyst under aqueous conditions at low reaction
dized and converted to RuO /C (Figure 1, bottom left). After
2
ꢁ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 626 – 633 627

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
2
RuÀ1
À1. A,C) In situ FT EXAFS (c: magnitude,
Figure 1. Gasification of 7.5% EtOH in D
O at a weight hourly space velocity (WHSV) of 1244 molEtOH mol
h
c: real part) and fitted spectra (a) of a Ru/C catalyst during a full poisoning–regeneration cycle. The observed coordination shells and bond lengths are
indicated by dotted vertical lines. B) The evolution of the D-labeled methane composition during the regeneration cycle. The reactor temperature was gener-
ally at 4008C. S poisoning was initiated at a runtime of 60 min; the regeneration procedure was applied between 215 and 280 min at a reactor temperature
2 4
of 1258C (here, the evolution of O was observed; its fragment of mass 16 interfered with the mass spectrometric measurement of CH ).
flushing with water, 7.5% EtOH in D O was again fed to the re-
beled methane produced by the regenerated catalyst reached
the same composition as that produced by the active catalyst,
which suggests that adsorbed S had been stripped from the
2
actor, which led to the reduction and thus reactivation of the
catalyst. EXAFS analysis of the reactivated catalyst under the
reaction conditions (4008C, 24.5 MPa) showed metallic Ru as
the only quantifiable catalyst phase (Table 1; Figure 1, bottom
right). The absorption edge step, indicative of the total
amount of Ru in the X-ray beam, remained constant (data not
shown), which suggests that no significant leaching of Ru took
place during the regeneration process (edge step 0.192 before
regeneration and 0.189 after regeneration). Furthermore, the
comparable magnitudes of the EXAFS spectra of active and re-
activated Ru/C showed that the oxidative treatment did not
lead to a measurable particle growth (RuÀRu coordination
numbers remained stable within the analytical error). The D-la-
Ru surface and that the original distribution of the CH surface
x
adsorbates had been restored.
These results suggest that the adsorbed S can be stripped
from S-Ru/C by treatment with dilute H O2 at 1258C. The
2
EXAFS data show that Ru is readily oxidized by H O to form
2
2
RuO , whereas it is known in literature that sulfide species can
2
[11]
be oxidized to colloidal sulfur, sulfanes, and sulfate. These S
species can either dissolve in water and be washed off the cat-
alyst (sulfate and sulfane) or are not directly attached to the
Ru surface any more (colloidal sulfur). Upon contact with etha-
nol at temperatures above 1258C, the oxidized Ru catalyst is
ꢁ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 626 – 633 628

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Table 1. Structural parameters obtained by fitting the EXAFS spectra of
active Ru/C, S-poisoned S-Ru/C, oxidized Ru/C, and regenerated Ru/C
[
a]
shown in Figure 1.
Coordination shell
Active Ru/C
CN
First RuÀRu
6.7(6)
Second RuÀRu
2.2(6)
R [ꢂ]
2.66(1)
3.6(7)
0.0097(5)
3.78(1)
3.6(7)
0.0097(5)
[
b]
[b]
DE
2
0
[eV]
2
[b]
[b]
s [ꢂ ]
S-Ru/C
CN
First RuÀS
1.0(2)
First RuÀRu
5.9(3)
Second RuÀRu
1.9(4)
R [ꢂ]
2.35(2)
10.0 (2.7)
0.006(2)
2.66(1)
2.4(1.1)
0.0094(3)
3.78(1)
2.4(1.1)
0.0094(3)
[
b]
[b]
DE
2
0
[eV]
2
[b]
[b]
s [ꢂ ]
Oxidized Ru/C
CN
First RuÀO
6.0(4)
First RuÀRu
5.8(4.9)
Second RuÀO
3.7(5.6)
R [ꢂ]
1.99(2)
À2.3(9)
0.009(3)
3.11(2)
3.3(1)
2 2
Figure 3. SEM images of Ru/C samples after oxidative treatment in 3% H O
[
b]
[b]
[b]
DE
s [ꢂ ]
0
[eV]
À2.3(9)
À2.3(9)
at 24.5 MPa. A) Ru/C after treatment at 1258C for 4 h (scale bar=200 mm);
the observed cracks in the carbon support are enlarged in image b (scale
bar=100 mm). C) Ru/C after treatment at 758C for 4 h (scale bar=200 mm).
D) shows Ru/C after a two-step regeneration at 758C (20 min exposure to
2
2
[b]
[b]
0.0180(6)
0.0180(6)
Regenerated Ru/C
CN
R [ꢂ]
First RuÀRu
7.2(7)
Second RuÀRu
2.3(8)
3.77(2)
3.0(5)
0.0104(6)
H
(
2
O
2
per step) with 60 min of water rinsing at 3908C between the steps
scale bar=200 mm). In the two latter cases, no particle corrosion could be
observed.
2.65(3)
[
b]
[b]
DE
s [ꢂ ]
0
[eV]
3.0(5)
0.0104(6)
2
2
[b]
[b]
[
a] Fitting parameters: coordination number (CN), bond length (R), energy
2
shift (DE), and pseudo-Debye–Waller factor (s ). [b] Fixed to be the same
and pore structure for H O exposure times up to 2 h. After
2
2
during the fit.
4
h, a drop in the micropore area was observed. However, the
macroscopic structure of the carbon particles was severely af-
fected by H O . SEM images of catalyst grains after exposure to
0
then reduced to Ru and thus reactivated to conclude the re-
2
2
generation cycle.
H O at 1258C for 4 h are shown in Figure 3A and b. Cracks in
2 2
However, the active catalyst phase is only one part of the
picture, the other is the catalyst support. For an efficient and
reproducible regeneration procedure it is imperative to ensure
that the integrity of the catalyst support is not compromised
during regeneration.
the particles are visible that might have resulted from the me-
chanical stress induced by the evolution of O from H O de-
2
2
2
composition within the particles. These cracks cut deep into
the carbon particles and cause chunks of carbon to flake off
from the main particle surface. Over time, this could lead to
the total disintegration of the catalyst support. A reduction in
the temperature had a significant effect on the integrity of the
carbon support. After a H O exposure time of 4 h at 758C, no
Therefore, we determined the surface area (BET surface) of
Ru/C samples that were exposed to H O at 1258C for different
2
2
amounts of time (Figure 2). The micro- and mesoporous struc-
ture of the carbon support was surprisingly stable under these
conditions and showed no significant change in surface area
2
2
cracks in the catalyst support were visible (Figure 3C), which
suggests that the mechanical stress on the carbon particles
had been largely reduced.
To determine whether S-Ru/C can be oxidized sufficiently at
a lower temperature to remove the adsorbed S, the oxidation
of Ru particles during regeneration was followed by in situ X-
ray absorption near edge structure (XANES) spectroscopy at
different temperatures. A sample of S-Ru/C was heated up
stepwise in dilute H O , and XANES spectra were recorded
2
2
after a dwell time of 5 min at each temperature step (Figure 4).
Upon contact with H O , Ru was quickly oxidized, and the
2
2
spectra recorded above 708C closely resembled that of a fresh
catalyst composed of pure RuO . Linear combination fitting of
2
the XANES spectra with fully reduced and fully oxidized Ru/C
as standards showed that RuO is the predominant catalyst
2
phase above 508C, whereas for temperatures above 1108C
nearly complete oxidation to RuO was achieved (Figure 4).
2
Based on these results, the regeneration of S-Ru/C was at-
tempted at lower temperatures. Even at temperatures as low
Figure 2. BET surface area analysis of various Ru/C samples by N
tion. Treatment with H was performed at 1258C.
2
physisorp-
2
O
2
ꢁ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 626 – 633 629

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
Table 2. Carbon-to-gas conversion and selectivity of Ru/C after regenera-
tion at different temperatures. The organic feed was 7.5% EtOH in water
at 3908C and 25 MPa. Water rinsing between regeneration steps was per-
formed at 3908C for 60 min.
Sample
Regeneration Conversion Carbon selectivity
[
a]
steps
[%]
CH
4
CO
2
CO
fresh Ru/C
S-Ru/C
regenerated at 1258C 1ꢃ20 min
regenerated at 1008C 2ꢃ20 min
regenerated at 1008C 1ꢃ40 min
–
–
99.9
29.8
99.7
99.7
81.8
99.8
0.75 0.25 0.01
0.51 0.11
0.71 0.28
0.75 0.25
0.38
0
0
0.61 0.29 0.1
0.77 0.23
regenerated at 758C
2ꢃ20 min
0
[
a] Carbon-to-gas conversion.
two-step regeneration process with intermediate water rinsing
did not cause corrosion of the carbon support (Figure 3D). As
experiments at full conversion do not show the true (hence,
maximal) activity of a catalyst, the Ru/C catalyst was run at par-
tial carbon-to-gas (c-t-g) conversion to assess the efficiency
and sustainability of the regeneration procedure. This allows
us to quantify how much of the original catalytic activity can
be regained by the oxidative treatment directly. The progres-
sion of catalytic activity and c-t-g selectivity for a two-step re-
generation at 758C is shown in Figure 5. Although the active
2 2
Figure 4. A) In situ XANES spectra of S-Ru/C in 3% H O . The temperature
was increased stepwise; dwell time was 5 min, acquisition time per spec-
trum was 30 s. The spectrum of an as-received catalyst sample (RuO /C) is
2
given as a reference. The dashed vertical line indicates the absorption edge
0
of Ru . B) Linear combination fit of the XANES spectra of S-Ru/C during oxi-
0
dative treatment by using fully reduced Ru/C (Ru ) and fully oxidized Ru/C
2
(RuO ) as standards.
as 758C, the H O treatment led to a regeneration of catalytic
2
2
activity as indicated by the high conversion and methane se-
lectivity of the catalyst after the oxidative treatment (Table 2).
However, temperatures below 1258C required a two-step re-
generation procedure in which the catalyst was treated with
H O at the chosen regeneration temperature and then rinsed
Figure 5. Carbon-to-gas conversion and product gas composition during
gasification of 15% EtOH over Ru/C at 4008C with a WHSV of
À1 À1
4
975 molEtOH molRu h . After S-poisoning, a two-step regeneration at 758C
was applied; to monitor the catalytic activity, EtOH was intermittently gasi-
fied for 70 min.
2
2
with water, followed by another H O treatment. Simply dou-
2
2
bling the regeneration time at temperatures below 1258C
hence, 40 min instead of 2ꢃ20 min) did not show the same
(
catalyst showed a high methane selectivity and c-t-g conver-
sion, the conversion decreased to approximately 6% upon S
poisoning. The formation of S-Ru/C was accompanied by a de-
crease in methane selectivity and a strong increase in CO se-
lectivity. The first regeneration step restored about 85% of the
original catalytic activity and increased the methane selectivity
back to its original value. However, as gasification continued,
the conversion and methane selectivity quickly started to de-
regeneration efficiency as the two-step process. The tempera-
ture of the water rinsing step also played a role, insofar as
higher temperatures facilitated higher regeneration efficien-
cies, and SCW at 3908C showed the best results (data not
shown). Hence, our data suggest that the (oxidized) S species,
generated at low regeneration temperatures, need higher tem-
peratures to be removed from the oxidized Ru/C catalyst. The
ꢁ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 626 – 633 630

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
crease, indicative of catalyst deactivation. A second regenera-
tion step brought the conversion back to its original value and
even improved the methane selectivity of the catalyst. This
gain in catalytic activity was again followed by a decrease in c-
t-g conversion. The methane selectivity, however, remained at
a higher level than that of the S-poisoned catalyst.
after oxidative treatment also showed no quantifiable amount
of Ru (data not shown).
A repeated S poisoning of the regenerated catalyst is unlike-
ly for two reasons. Firstly, because of the very small Ru particle
size of around 1.5 nm, EXAFS is sensitive to chemisorbed S on
the Ru surface. However, no RuÀS bonds could be identified in
the EXAFS spectra even at a runtime of 2 h after regeneration
(at which point catalyst deactivation is readily observed; Fig-
ures 1 and 5). Secondly, the methane selectivity of the catalyst
remains high, whereas a typical effect of S poisoning is
a strong decrease in methane formation (Figure 5). Hence, S
poisoning does not seem to be the cause for the observed de-
crease in conversion.
To exclude sintering or leaching of the Ru particles as a po-
tential reason for catalyst deactivation, a blank experiment was
conducted in which ethanol was gasified over a fresh sample
of Ru/C, followed by catalyst treatment with H O at 758C. The
2
2
observed behavior of this catalyst sample was very different to
that of S-Ru/C. The catalytic activity and the methane selectivi-
ty of the fresh catalyst were initially increased by the oxidative
treatment and then leveled off to their original values within
Since Ru leaching could also be excluded by effluent analy-
sis, minor structural changes (such as sintering) of the Ru/C
catalyst after S poisoning and subsequent oxidative regenera-
tion remain as a potential explanation for the observed deacti-
vation. Although no deactivation (and, therefore, no significant
particle sintering) was observed for a fresh sample of Ru/C
after treatment with H O , the presence of S might have an
a few hours (Figure 6, ). In contrast, the regenerated catalyst
&
2
2
effect on the Ru particle stability. The observed increase in ac-
tivity and methane selectivity of a fresh Ru/C sample after
treatment with H O suggests that structural changes might
2
2
occur. In the given size range, small changes in particle size
and structure can have a significant effect on catalytic activi-
[12–15]
ty.
However, the coordination numbers that are extracted
from the EXAFS spectra typically have an uncertainty of
around 10%, and the error in particle size (and shape) calculat-
ed based on coordination numbers is even greater. For exam-
ple, in the particle size range of 1.5 nm, a 10% error in the co-
ordination number can result in an error in the calculated par-
[
16,17]
ticle size of up to 20%.
The particle size and structure sen-
[18–20]
sitivity of the methanation reaction is well known.
Hence,
structural changes that are too small to be detected by EXAFS
could be the cause of the observed decrease in conversion.
The elucidation of the underlying mechanisms requires further
research, in particular a detailed study of Ru particle size and
structure before, during and after oxidative catalyst treatment.
Ideally, in situ methods such as XAS should be combined with
ex situ electron microscopy to give a complete picture of the
effect of H O on the catalyst.
2
2
Conclusions
A protocol for the on-stream regeneration of a sulfur-poisoned
ruthenium–carbon catalyst under hydrothermal conditions was
presented. The regeneration procedure involved the treatment
of the catalyst with dilute hydrogen peroxide at low tempera-
ture and high pressure. The combined results from in situ ex-
tended X-ray absorption fine structure analysis and deuterium
labeling of product gases during the supercritical water gasifi-
cation (SCWG) of ethanol over ruthenium–carbon allow us to
conclude that sulfur removal from sulfur-poisoned ruthenium–
carbon is possible by using a liquid-phase, oxidative treatment.
The treatment of the catalyst with 3% hydrogen peroxide in
water at temperatures as low as 758C was sufficient to oxidize
the ruthenium catalyst and to remove adsorbed sulfur from
Figure 6. Carbon-to-gas conversion of (S-)Ru/C (*) and fresh Ru/C (
product gas composition for fresh Ru/C before and after treatment with
at 758C. H treatment of the fresh catalyst sample did not lead to
&
) and
H
2
O
2
2 2
O
catalyst deactivation, whereas a S-Ru/C sample quickly deactivated again
after the treatment. Gasification conditions were 15% EtOH, 3908C, 25 MPa,
À1 À1
WHSV 4975 molEtOH molRu
h .
(
after oxidative treatment of S-Ru/C) showed a significant de-
crease in conversion (Figure 6, *). As the oxidative treatment
of the fresh Ru/C catalyst did not lead to a decrease in conver-
sion, effects such as sintering and leaching of the active cata-
lyst phase are unlikely to be the cause of the observed deacti-
vation after H O treatment. Analysis of the reactor effluent
2
2
ꢁ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 626 – 633 631

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
the ruthenium surface. Under these mild conditions, the perox-
ide treatment did not cause corrosion of the carbon support,
which preserved its macro- and microstructure.
fitted by using scattering paths obtained from theoretical stand-
[24–26]
ards for metallic Ru, RuO , and RuS , respectively.
In the case
2
2
of the active and regenerated catalyst samples, the following fit-
ting strategy was used: the main RuÀRu CS was fitted first, over an
R range from 1.5–3 ꢂ, without constraining the fitting parameters
After reactivation of the oxidized catalyst, the catalytic activi-
ty was similar to that of a fresh catalyst before sulfur poison-
ing. After the initial gain in catalytic activity after regeneration,
a new deactivation of the catalyst was observed in the absence
of sulfur in the feed. Sulfur poisoning and leaching of the
active catalyst phase could be excluded as causes, and it
seems likely that structural changes of the catalyst were re-
sponsible for the deactivation. Further research that focuses
on ruthenium particle size and structure is necessary to shine
light on these phenomena.
2
0
2
(
S , E , Dr, s ). Then, E and Dr of the first shell were fixed at their
0
0
best fit values and the fit range was extended to 4 ꢂ. The scatter-
ing path of the second RuÀRu CS was added to the fit with its
2
0
own set of S and Dr parameters. The fitting of the first two RuÀRu
CS thus provided a single energy shift E and pseudo-Debye–
0
2
Waller factor s valid for both shells as well as amplitude reduction
2
0
factors (S ) and bond distance shifts (Dr) for each shell. The ampli-
tude reduction factors were calibrated against data from bulk sam-
ples of metallic Ru (first and second shell RuÀRu coordination num-
bers fixed at 12 and 6, respectively) to obtain the coordination
number (CN) for each shell. The same strategy was used for the S-
poisoned catalyst to determine the CN of the first and second RuÀ
Experimental Section
2
Ru CS. Then, E , Dr, and s of the RuÀRu shells were fixed at their
0
best fit values, and the RuÀS CS was fitted with its own set of fit-
Two types of tubular reactors were used for SCWG. For experi-
ments at the synchrotron, a ceramic reactor (LꢃODꢃID=200ꢃ6ꢃ
ting parameters. Again, data from a bulk sample of RuS was used
2
2
0
to calibrate the S parameter (RuÀS CN fixed at 6). All fitted spectra
3
.5 mm) that allows in situ X-ray absorption (XAS) measurements
and experimental data were plotted by using the first RuÀRu scat-
was used. The centerpiece of this setup is an aluminum nitride
tube that contains the catalyst as a fixed bed and allows for suffi-
cient X-ray transmission. For a detailed description of the setup,
tering path for phase correction.
[6,21]
the interested reader is referred to previous publications.
For
Acknowledgements
bench-top experiments in the laboratory, a tubular stainless-steel
reactor (LꢃODꢃID=300ꢃ6ꢃ4 mm) was employed. The entire
setup used to feed liquids at high pressure, and the separation of
the liquid and gaseous reactor effluents was similar for both reac-
The authors thank Erich De Boni for technical assistance and
Evalyn Alayon and Christian Kçnig for beamtime support, as well
as Alexander Wokaun for fruitful discussions. This project was
funded by the Swiss National Science Foundation, SNF Grant No.
200021_130615.
[21]
tors and has been described in detail in a previous publication.
A commercial Ru catalyst (BASF, Germany) was used in the gasifica-
tion experiments. The catalyst consisted of 2 wt% Ru supported
on activated carbon and is referred to as Ru/C. The as-received Ru/
C catalyst was crushed in a mortar and sieved to a grain size of
Keywords: heterogeneous catalysis · ruthenium · sulfur ·
supercritical fluids · sustainable chemistry
5
0–200 mm. A fixed catalyst bed (typically 200 mg of Ru/C) with
a length of about 25 mm was used in both reactor types. All chem-
icals used during gasification and regeneration experiments were
of analytical grade (Sigma–Aldrich, Alfa Aesar). Deuterated water
[1] A. Kruse, J. Supercrit. Fluids 2009, 47, 391–399.
[
[
2] D. Elliott, Biofuels Bioprod. Biorefining 2008, 2, 254–265.
3] F. Vogel, M. H. Waldner, A. Rouff, S. Rabe, Green Chem. 2007, 9, 616.
(
99.8% D) was obtained from ARMAR Chemicals, Switzerland.
The product gas composition was monitored online by using
a mass spectrometer (OmniStar 310, Pfeiffer, Germany), and precise
measurements of the gas composition were performed offline by
using GC (HP 6890 Series with HP 1, AT 5, and PLOT Q columns).
The carbon-to-gas conversion was determined by analyzing the
carbon content in the feed and the effluent with a total carbon an-
alyzer (vario TOC cube, Elementar, Germany).
[4] M. Waldner, F. Krumeich, F. Vogel, J. Supercrit. Fluids 2007, 43, 91–105.
[5] A. Peterson, M. Dreher, J. Wambach, M. Nachtegaal, S. Dahl, J. K. Nør-
skov, F. Vogel, ChemCatChem 2012, 4, 1185–1189.
[
6] M. Dreher, B. Johnson, A. Peterson, M. Nachtegaal, J. Wambach, F.
Vogel, J. Catal. 2013, 301, 38–45.
[
[
7] Q. Guan, C. Wei, P. Savage, Energy Fuels 2012, 26, 4575–4582.
8] M. Osada, N. Hiyoshi, O. Sato, K. Arai, M. Shirai, Energy Fuels 2008, 22,
845–849.
[
9] M. H. Waldner, Catalytic Hydrothermal Gasification of Biomass for the
Production of Synthetic Natural Gas, PhD Thesis 17100, ETH Zꢄrich (CH),
SEM images of catalyst samples were recorded by using a Carl
Zeiss Ultra 55.
2
007.
The linear combination fitting of XANES spectra was performed in
the range from À20 to 60 eV with respect to the absorption edge.
Reference spectra were allowed to shift in energy and were as-
signed a weight between 0 and 1, without forcing the sum of
weights to 1.
[10] P. Kritzer, E. Dinjus, Chem. Eng. J. 2001, 83, 207–214.
[
[
11] M. Hoffmann, Environ. Sci. Technol. 1977, 11, 61–66.
12] S. H. Joo, J. Y. Park, J. R. Renzas, D. R. Butcher, W. Huang, G. Somorjai,
Nano Lett. 2010, 10, 2709–2713.
[
[
[
13] H. Abrevaya, M. Cohn, W. Targos, H. Robota, Catal. Lett. 1990, 7, 183–
1
96.
14] Y. B. He, M. Knapp, E. Lundgren, H. Over, J. Phys. Chem. B 2005, 109,
1825–21830.
The in situ XAS measurements were conducted at the SuperXAS
beamline at the Swiss Light Source (SLS) at the Paul Scherrer Insti-
tute (Villigen, Switzerland). XAS data treatment was performed by
2
15] R. K. Grasselli, A. W. Sleight, Structure–Activity and Selectivity Relation-
ships in Heterogeneous Catalysis, Elsevier, Amsterdam, 1991.
16] R. Greegor, F. Lytle, J. Catal. 1980, 63, 476–486.
17] S. Calvin, M. M. Miller, R. Goswami, S.-F. Cheng, S. P. Mulvaney, L. J. Whit-
man, V. G. Harris, J. Appl. Phys. 2003, 94, 778.
[22,23]
using the IFEFFIT software suite.
All XAS data were energy-cali-
[
[
brated by measuring a Ru reference foil (EXAFS Materials, USA) si-
multaneously with the samples. Fourier transformations of the nor-
malized and background-subtracted EXAFS spectra were per-
[18] S. B. Vendelbo, M. Johansson, J. H. Nielsen, I. Chorkendorff, Phys. Chem.
Chem. Phys. 2011, 13, 4486–4493.
À1
formed over a k range from 3.5–12.5 ꢂ . The EXAFS data were
ꢁ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 626 – 633 632

CHEMCATCHEM
FULL PAPERS
www.chemcatchem.org
[
19] F. R. Garcꢅa-Garcꢅa, A. Guerrero-Ruiz, I. Rodrꢅguez-Ramos, Top. Catal.
[25] H. D. Lutz, B. Mꢄller, T. Schmidt, T. Stingl, Acta Crystallogr. Sect. C 1990,
46, 2003–2005.
2009, 52, 758–764.
[
[
20] R. van Hardeveld, A. van Montfoort, Surf. Sci. 1966, 4, 396.
21] M. Dreher, E. De Boni, M. Nachtegaal, J. Wambach, F. Vogel, Rev. Sci. Ins-
trum. 2012, 83, 054101.
[26] A. A. Bolzan, C. Fong, B. J. Kennedy, C. J. Howard, Acta Crystallogr. Sect.
B 1997, 53, 373–380.
[22] M. Newville, J. Synchrotron Radiat. 2001, 8, 322–324.
[23] B. Ravel, M. Newville, J. Synchrotron Radiat. 2005, 12, 537–541.
[24] R. Clendenen, H. Drickamer, J. Phys. Chem. Solids 1964, 25, 865–868.
Received: September 19, 2013
Revised: October 30, 2013
Published online on December 27, 2013
ꢁ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 626 – 633 633