Professional Documents
Culture Documents
Journal of Catalysis
journal homepage: www.elsevier.com/locate/jcat
a r t i c l e i n f o a b s t r a c t
Article history: The active sites and kinetics of dry reforming of methane (DRM) on lanthanum zirconate (LZ) pyrochlore
Received 9 January 2014 catalysts are studied as a function of Rh substitution, temperature, and partial pressures of CH4 and CO2.
Revised 8 April 2014 In this work, we focus specifically on determining the catalytic active sites for CH4 and CO2 activation and
Accepted 27 April 2014
their role in the DRM mechanism over the two Rh-substituted pyrochlores, i.e., 2 wt% Rh, designated
Available online 2 June 2014
L2RhZ and 5 wt% Rh, designated L5RhZ. Kinetic rate modeling suggests dual-site mechanism, where
CH4 and CO2 are activated on different sites (dual-site mechanism). Eleven different rate models were
Keywords:
tested against the kinetic rate data obtained over these two Rh pyrochlores. Statistical analysis shows
Dry reforming
La-oxycarbonates
valid and similar fits for only two of eleven models: one in which activation of CH4 is rate-determining
Pyrochlores and one in which CO2 activation is rate-determining. This dual-site mechanism is studied further in detail
Lattice oxygen to test the validity of the intermediate steps predicted by the kinetic model. CH4/CD4 isotope switching
Kinetic isotope effect shows a strong deuterium kinetic isotope effect on CH4 and CO2 conversion, suggesting that CH4 disso-
Transient pulsing ciation is the rate-determining step. Higher apparent activation energies of CH4 (Ea, CH4) versus CO2 (Ea, CO2)
on both catalysts also confirm that CH4 activation is rate-determining. Basic nature of La–O sites activates
mildly acidic CO2 to form La2O2CO3 complexes as confirmed by FTIR. From different polymorphs of La2-
O2CO3, the spectator and reactive species were distinguished by pulsing CH4 over La2O2CO3. Alternating
pulses (CH4/Ar ? CO2/Ar ? CH4/Ar) at 550 °C showed simultaneous formation of CO and H2, suggesting
that surface carbon, formed by CH4 decomposition, is oxidized by H2O; a crucial step in understanding
the catalytic behavior of pyrochlores in the reaction mechanism. These experiments were used to identify
the two types of sites taking part in the dual-site DRM mechanism. The work reported here helps in
determining a single set of kinetically significant steps that most closely represent the mechanistic
scheme of DRM over L2RhZ and L5RhZ.
Ó 2014 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.jcat.2014.04.023
0021-9517/Ó 2014 Elsevier Inc. All rights reserved.
D. Pakhare et al. / Journal of Catalysis 316 (2014) 78–92 79
CO2 activation and decrease carbon formation [12–14]. Addition of Aesar, 99.9%), and rhodium nitrate [Rh(NO3)32H2O] (Alfa Aesar,
basic promoters such as La2O3, CeO2, and alkaline earth metal oxi- 99.9%) were used. The dissolved metal salts were combined in a
des such as MgO, CaO, and BaO help in CO2 activation, thus increas- beaker with continuous stirring. Complexing agent citric acid
ing reactive oxygen for surface carbon oxidation [2,15–21]. Basic (CA) solution with molar CA:metal ion ratio of 1.2:1 was added
promoters also help increase the active metal dispersion [13]. A to the metal solution beaker and heated to 70 °C to which the poly-
comprehensive review on DRM kinetics and mechanism by Pak- merizing agent ethylene glycol (EG) was added with 1:1 M ratio of
hare and Spivey [22] suggests various ways to control deactivation EG:CA. The solution was stirred until a clear viscous gel remained
of DRM catalyst. in the beaker. The gel was then heated to 130 °C to promote the
Activation of CH4 is a crucial step in DRM mechanism [17,22–24]. polyesterification reaction. The resulting amorphous polyester-
However, there is no consensus in the literature on the mechanism type resin was calcined at 1000 °C for 8 h to form pyrochlore
of CH4 activation on catalytic sites, even on the same metal. For catalyst.
example, Luntz and Winters [25] and Wei and Iglesia [26] proposed
that on Pt and Pd [27] clusters, the activation of CH4 occurs by direct 2.2. Activation energies
decomposition where all CAH bonds simultaneously dissociate
resulting in C(s) and evolution of H2(g). However, theoretical calcula- The activation energy was obtained in the AMI-200 Altamira
tions by Seets et al. [28] suggest activation of CH4 by the formation of Instruments reactor using the catalyst:alumina of 1:10. Because
reactive intermediates such as formates or CHx species. Yamaguchi the activity of L5RhZ was higher than L2RhZ, the temperature
and Iglesia [27] observed that CAH bond activation in CH4 is struc- range chosen for L2RhZ (585–605 °C) was higher than that for
ture sensitive and its rate increases with decreasing Pd clusters size. L5RhZ (520–560 °C) to limit the conversion in both cases and to
For DRM, the activity decreases in the order Ru, Rh > Ir > Ni, Pt, minimize any effects of concentration/temperature gradients
Pd > Co > Fe, Cu [29], with noble metals showing higher activity across the catalyst bed. Gas composition and flow was 10%
and greater resistance to carbon deposition [30–33]. DRM has been CH4/He and 10% CO2/He at 20 mL/min each.
studied on various catalysts; however, we are aware of no papers
where pyrochlores have been studied for DRM other than those 2.3. Kinetic rate modeling
reported by our group [34–36]. Pyrochlores are crystalline ther-
mally stable ternary metal oxides with a cubic unit cell and general The equipment details and catalyst dilution ratio were the same
formula of A2B2O7. The A-site is usually a large cation (typically as reported previously (Section 2.2). The partial pressure depen-
rare earth elements) and the B-site cation has a smaller radius dence of CH4/CO2 reaction rate was determined by keeping partial
(usually transition metal) [22,37]. These materials have two impor- pressure of one reactant (CH4 or CO2) constant at 0.05 atm and
tant properties that make them interesting for this reaction. The varying that of the other (CO2 or CH4) as 0.03, 0.035, 0.04, and
first is oxygen ion conductivity; vacancies at the O sites, created 0.045 atm (balance He). These set of experiments were performed
during synthesis, facilitate oxygen ion migration, which minimizes at 570 °C, 590 °C, and 610 °C for L2RhZ and at 520 °C, 540 °C, and
carbon deposition formed by CH4 dissociation [38]. The second is 560 °C for L5RhZ. The conversion was kept below 15%, so as to
the ability to isomorphically substitute catalytically active transi- approach differential conditions.
tion metals into the B-site, where they do not sinter at high DRM
temperatures. Candidate catalytically active transition metals for 2.4. CH4/CD4 kinetic isotope effect
B-site substitution include Ru, Rh, and Pt [34,35]. The most closely
related paper for DRM over pyrochlores is by Ashcroft et al. [39] Deuterium kinetic isotope effect was studied by using 10% of
who studied Eu2Ir2O7 for DRM. However, the pyrochlore structure CD4/He, 10% CH4/He, 10% CO2/He (all 99.99% pure). The reactor is
in that study decomposed to Eu2O3 and metallic Ir under DRM same as mentioned previously (Section 2.2). Before testing for
gases above 340 °C. DRM, the catalysts were reduced in 4% H2/He by ramping from
Here, we report a systematic study of the active sites and kinet- 25 °C to 600 °C and maintaining isothermal conditions at 600 °C
ics of DRM on two Rh-substituted lanthanum zirconate [LZ] for 30 min then purged in He for 15 min. For testing the change
pyrochlores: 2 wt% [L2RhZ] and 5 wt% [L5RhZ]. A detailed charac- in the rate of consumption of CH4 and CD4, alternate cycles of
terization and activity study of these materials has been reported CH4/CO2/He and CD4/CO2/He of 30 min each were performed on
in our earlier work [36] where we confirm the formation of the L2RhZ and L5RhZ.
pyrochlore phase and substitution of Rh at the B-site. Here, the
process of CH4 and CO2 activation as a function of temperature 2.5. In situ FTIR and transient pulsing
and Rh substitution in the pyrochlore structure was studied by
Fourier transform infrared (FTIR) spectroscopy. The role of CH4 in For in situ FTIR studies, a Thermo Electron Nexus 670 spectrom-
rate-limiting step in DRM mechanism is examined by studying eter with a liquid N2 cooled MCT detector and a diffuse reflectance
the CH4/CD4 kinetic isotope effect. The elementary steps involved infrared Fourier transform spectroscopy (DRIFTS) cell from Pike
in oxidation of surface carbon were studied by transient pulsing Technologies was used. KBr beamsplitter provided spectra in the
of CH4 and CO2 with Ar tracer. Kinetic rate models based on postu- wavenumber range of 4000–500 cm1. Prior to spectrum collec-
lated mechanisms were validated by statistically fitting the kinetic tion, catalysts were reduced in flowing 4% H2/He (20 mL/min) by
data obtained over the L2RhZ and L5RhZ pyrochlores. heating from 25 °C to 550 °C and maintaining at 550 °C for
30 min. The reduced catalysts were then flushed by He for
15 min for background collection. The DRIFTS cell was connected
2. Experimental section to an 8-port valve which allowed rapid switching of gases without
interrupting the continuous He flow, thus avoiding pressure drop
2.1. Catalyst synthesis across the cell. The valve was fitted with 0.5 cm3 loop, allowing
injection of known amount of reactant gas pulses in the continuous
The LZ, L2RhZ, and L5RhZ pyrochlores studied in this work were He flow.
prepared by the modified Pechini method [40]. The metal precur- After background collection at 550 °C, the catalysts were treated
sors of lanthanum nitrate hexahydrate [La(NO3)36H2O] (GFS in flowing 10% CO2/He (20 mL/min). This resulted in the formation
Chemicals, 99.9%), zirconium oxynitrate [ZrO(NO3)2xH2O] (Alfa of carbonate on the catalysts and the gas-phase CO2 was later
80 D. Pakhare et al. / Journal of Catalysis 316 (2014) 78–92
Table 2
Summary of key results used in formulating reaction mechanism.
i.e., r = k. PCH4, and is independent of the co-reactant partial pulsing experiments were performed. These experiments will also
pressure. However, studies by Munera et al. [42,48], Sierra Gallego help confirm the validity of important intermediate steps such as
et al. [54], Verykios [55], and Carrara et al. [16] suggest fractional oxidation of C(s) by H2O and dissociative adsorption of CH4.
order dependence on CO2 for La2O3-based catalysts.
Based on the mechanistic work reported in the literature, a series 3.3. CH4/CD4 kinetic isotope effect
of kinetically significant steps, with one of the steps being rate lim-
iting, is postulated here. This mechanistic sequence was then used to Activation energies (Table 1 and Fig. 1) show that CH4 dissocia-
derive the mathematical kinetic rate expressions. The assumptions tion is the rds during DRM on pyrochlores. In order to verify the
of kinetically significant steps were based on theoretical kinetic expected rate-limiting CAH bond cleavage in DRM, CH4/CD4
models such as Eley–Rideal and Langmuir–Hinshelwood (single kinetic isotope effect was studied. In this experiment, CH4/CD4
and dual site) mechanisms. Various possibilities of rds, including and CO2 consumption rates were measured before and after
CH4 and CO2 dissociation, oxidation of C formed from CH4 dissocia- switching of CH4 to CD4. Figs. 2 and 3 show the changes in CH4,
tion were assumed. The combinations of the theoretical models,
kinetically significant steps, and possible rds resulted in a matrix
of eleven kinetic rate models (details in Supplemental information),
out of which only two were statistically valid for the pyrochlores
studied in this work. The fitting parameters obtained from these
two models had (a) positive values, (b) 95% confidence intervals
did not include zero, (c) adsorption constants decreased with
increasing temperature, (d) and the rate constants increased with
increasing temperature. The reaction sequence for the two kinetic
models that resulted in a statistically valid fit to the kinetic data is
as follows:
H2 þ O S2 ¡H2 O þ S2 ð5Þ
C S1 þ H2 O¡CO þ H2 þ S1 ð6Þ
The proposed mechanism is a dual-site mechanism, where CH4
activates on S1 and CO2 on S2. This mechanism predicts dissociative
adsorption of CH4 to form surface carbon (C(s)) and H2. The C(s) is
then oxidized by H2O formed from H2 and CO2–S2 reaction. This
is a very important step and will help in limiting carbon formation
of catalyst surface. The two rate models were derived using this
mechanism sequence, but differed in the rds. In the first model
(designated – dual site [1]), the irreversible CH4 dissociation (Eq.
(2)) was assumed to be rds, and in the second model (designated
– dual site [2]), CO2 dissociation (Eq. (4)) was assumed to be irre-
versible and rate determining. Of the eleven models tested, only
two resulted in statistically valid fit meeting criteria (a–d) men-
tioned above. Since multiple rate models resulted in a good statis-
tical fit, it is important to employ other experimental techniques to
distinguish between the 2 models and confirm the likelihood of
occurrence of the intermediate steps proposed in the mechanism.
To determine the active sites S1 and S2 further in detail and to con- Fig. 2. CH4/CD4 kinetic isotope effect on (a) CH4/CD4 conversion and (b) CO2
firm the rds, following CH4/CD4 kinetic isotope, FTIR, and transient conversion, observed in alternating cycles of CH4/CO2 and CD4/CO2 for L2RhZ.
82 D. Pakhare et al. / Journal of Catalysis 316 (2014) 78–92
The conversion of CH4 (or CD4) and CO2 over L5RhZ was greater
than L2RhZ in all cycles, and no conversion was observed over non-
substituted LZ pyrochlores. The XPS results reported earlier [36]
suggest that the surface concentration of Rh on L5RhZ is twice of
that on L2RhZ. Linic et al. [56] performed DFT calculations and
found that CH4 activation barrier on terrace sites is higher from
that on stepped sites and that uncoordinated metal sites assist in
CAH bond activation. Since the active site for CH4 activation is
Rh metal, the difference in CH4 conversion is attributed to greater
Rh surface concentration and presumably greater number of
stepped sites on L5RhZ pyrochlores compared to L2RhZ. When
CH4 was switched with CD4 in cycle 2, the drop in CD4 conversion
is consistent with the hypothesis that breaking the CAH bond in
methane is rate limiting suggesting occurrence of dual-site [1]
mechanism (Section 3.2). In the early part of the CD4 + CO2 cycle,
CD4 adsorbs on the Rh sites causing a high apparent CD4 conver-
sion even though it is not reacting. As time progresses, the slow
dissociation rate of CD4 prevents the turnover of CD4 and active
sites gradually get saturated with CD4. This gradual saturation of
Rh sites prevents further CD4 adsorption and results in a gradual
drop in the apparent CD4 conversion. When the CD4 is switched
back with CH4, a significant amount of CD4 from the previous cycle
desorbs and leaves the reactor along with the products of the suc-
ceeding CH4 cycle.
Higher activity of L5RhZ than L2RhZ could also be due to the dif-
ference in the mechanism of activation or the kinetics of CH4/CD4
activation over the two pyrochlores. Breaking of the CAH or CAD
bond in rds should result in an isotope effect factor (kCH4/CD4) value
of >1 [57]. In this case, it is evident that CD4 conversion is consis-
tently lower than CH4 conversion and kCH4/CD4 is >1 at all times.
Due to the lower vibrational frequency of CD4, it is less reactive
than CH4 [56,57] and the clear isotope effect shows that the rds
involves CAH (D) bond dissociation [56–58]. Wang and Au [57]
observed that over Ni/SiO2 deuterium (CD4) isotope had no effect
on CO2 conversion and the effect was less prominent on CO forma-
tion; thus, they proposed that the activation of CO2 is independent
of the activation rate of CH4. However, in this work, deuterium iso-
tope effect is observed on CO2 conversion, suggesting that the acti-
vation of CH4 and formation of H2 is essential for the conversion/
reduction of CO2. This dependence of CO2 reduction on H2 will be
Fig. 3. CH4/CD4 kinetic isotope effect on (a) CH4/CD4 conversion and (b) CO2 studied in detail in CH4 TPR (Section 3.6) and transient (Section 3.7)
conversion, observed in alternating cycles of CH4/CO2 and CD4/CO2 for L5RhZ.
results reported later in this paper.
CD4, and CO2 conversion over L2RhZ and L5RhZ, respectively. Dur- 3.4. Theory of La-oxycarbonates (La2O2CO3)
ing the first 30-min cycle (i.e., t = 0–30 min) of CH4/CO2 at 600 °C,
significant CH4 and CO2 conversion (Figs. 2 and 3) was observed. La-oxycarbonates are polymeric complexes built up of slabs
When CH4 was switched with CD4 in the next cycle, a strong deu- forming layer-type structures [59] of La2O2+ 2
2 and CO3 ions formed
terium isotope effect was apparent, where the CD4 conversion at from the chemisorption of mildly acidic CO2 by basic La2O3 sites;
the time of switch (t = 30 min) was about 15% for L2RhZ and 30% La2O3 + CO2 ? La2O2CO3 [44]. Depending on the positioning of
for L5RhZ and dropped to 0% and 10%, respectively, by the end of the La2O2+ 2 ions in the complex, three different crystalline poly-
30-min (i.e., t = 60 min) cycle. The CO2 conversion during this cycle morph types I, Ia, and II are defined (Fig. 4) [60]. Characterization
(t = 30–60 min) was lower as compared to that during the first of these structures shows that there are no weight changes
cycle (t = 0–30 min) for both catalysts, suggesting that deuterium recorded during type-I ? type-Ia ? type-II transformations, con-
had a strong isotope effect on CO2 conversion as well. When the firming that these types are polymorphs to each other [59].
reaction mixture was switched back to CH4/CO2 (t = 60–90 min), Type-I form of La-oxycarbonates are square layers (Fig. 4) of
the catalysts regained their first-cycle activity. In the next cycle (La2O2+ 2
2 )n in a tetragonal crystalline form separated by CO3 ions.
(t = 90–120 min) when CH4 was switched again with CD4, the exact Type-Ia form is the monoclinic distortion of type-I [61]. Type-II
same decay of CD4 conversion with time observed in the first carbonates are hexagonal unit cell crystallites found in A-form
switch was repeated. sesquioxides with more than one local coordination of La [59,61].
Increasing the temperature from 600 °C to 620 °C resulted in Formation of these three polymorphs is a function of carbonation
an increase in the CH4 and CO2 conversion (t = 120–150 min), temperature, the presence of H2O as an impurity, and heating/
and the same isotope effect was observed for both the catalysts. cooling rate [62]. A detailed study of La-oxycarbonate formation
Finally, the conditions of the first cycle were repeated (600 °C) as a function of different pre-treatments is reported by Irusta
and showed no deactivation of the catalyst over the 180 min of et al. [60]. These are the types of La-oxycarbonates that have been
the previous cycles. reported to be formed on La2O3 supports; hence, it is important to
D. Pakhare et al. / Journal of Catalysis 316 (2014) 78–92 83
acknowledge the possibility of minor differences in the La-oxycar- inactive for free ions but is active for crystalline complexes [59].
bonates formed on lanthanum zirconate pyrochlores. The distinct appearance of 1066 cm1 bands suggests that carbon-
ate is a part of the crystalline monoclinic tetragonal type-Ia struc-
3.5. In situ FTIR and transient pulsing ture and is closely associated with La ion. The peak at 1511 cm1
corresponds to m3 vibration of CO2 3 associated with type-II form
3.5.1. Adsorbed La-oxycarbonates of oxycarbonate [63]. Taylor and Schrader [62] reported the pres-
In situ FTIR bands in the range 1000–1600 cm1 corresponding ence of type-Ia at 1363 cm1 and type-II form of oxycarbonate at
to the formation of La-oxycarbonates observed on CO2 treated LZ, 1499 cm1 for La2O3. The shift in the peaks could be assigned to
L2RhZ, and L5RhZ (Fig. 5), suggesting strong CO2 chemisorption the difference in the surface sites on LZ compared to that on
[11]. Spectra in Fig. 5 are obtained after 15 min of CO2 flow fol- La2O3. Taylor and Schrader [62] also report the presence of bands
lowed by 15 min of He flush, leaving only the strongly adsorbed in the range of 500–1000 cm1 for La2O3 supports; however, these
species on the surface. For determining the likelihood of adsorption bands are infrared inactive for the pyrochlores reported here. This
of CO2 on La2O3 and ZrO2, FTIR spectra for CO2 adsorption on La2O3 suggests a difference in the coordination of La ions on LZ (and other
and ZrO2 standards were studied. No adsorbed CO2/carbonate reported pyrochlores) as compared to that over La2O3. A shoulder
species were observed on ZrO2, but strong adsorption bands were at 1573 cm1 marks the presence of formates which is confirmed
observed for La2O3, suggesting that La–O is the active site for the by CAH stretching bands at 2834 cm1 [42,43,48,66]. A small peak
activation of CO2 on the lanthanum zirconate pyrochlores. How- at 2163 cm1 is due to the presence of weakly linearly adsorbed
ever, it is important to acknowledge that there could be differences ionic CO as the band is at a higher wavenumber as reported in
in the surface sites over La2O3 and LZ, L2RhZ, and L5RhZ pyrochlor- the literature [43,55,67,68]. There are negative bands at 3631
es. Although ZrO2 did not show any activation of CO2, the presence and 3652 cm1, which correspond to stretching frequencies of
of Zr in the pyrochlore structure could contribute to the formation hydroxyl (OAH) species [29,43,59,60,66]. The negative absorbance
of carbonates due to the difference in the structure of oxides such peaks suggest that there were chemisorbed hydroxyl species pres-
as La2O3 (or ZrO2) and pyrochlores. ent on LZ after reduction with H2 and were accounted for in the
The carbonate bands (1100–1600 cm1) for all three catalysts background spectra. However, after reaction with CO2, these
show a twofold splitting, suggesting more than one polymorph of hydroxyls are consumed to form formate species, resulting in a
carbonate [63–65]. Carbonate (CO2 3 ) is a non-linear molecule,
negative peak at these wavenumbers [43,69].
and thus, it should have 3n 6 (i.e., 6) modes of vibration; how- The FTIR spectra for L2RhZ (Fig. 5b) show type-Ia oxycarbonate
ever, due to the degeneracy of the modes and overlap with absorp- peaks at 1068 cm1 and 1367 cm1 [48]. Presence of Rh in the
tion bands of other surface species, not all modes of carbonate structure changes the electronic structure by adding an ionic char-
vibrations appear in the spectra [61]. For LZ (Fig. 5a), peaks at acter to type-Ia oxycarbonates, thus causing it to shift to a higher
1066 and 1349 cm1 correspond to the presence of type-Ia poly- wavenumber compared to LZ. Type-II oxycarbonate appears on
morph of La-oxycarbonate [59,60,62]. The band at 1066 cm1 L2RhZ at wavenumber 1509 cm1, which corresponds to that
corresponds to symmetric stretching, and this mode is infrared formed over LZ at 1511 cm1. Small shoulder at 1745 cm1
84 D. Pakhare et al. / Journal of Catalysis 316 (2014) 78–92
Table 3
Summary of the change in the ratio of the absorbance of type-Ia peak to that of type-II
peak as a result of CH4 pulses and CO2 readsorption for L2RhZ and L5RhZ.
Fig. 8. Response for pulse scheme CH4/Ar ? CO2/Ar ? CH4/Ar for L2RhZ, (a) MS signal response and (b) normalized pulse response.
Fig. 9. Pulse response for scheme CH4/Ar ? CO2/Ar ? CH4/Ar for L5RhZ, (a) mass spec signal response and (b) normalized pulse response.
88 D. Pakhare et al. / Journal of Catalysis 316 (2014) 78–92
Table 5
Summary of the quantification of the MS signal pulse during transient experiments.
to produce a surface carbon, and precursor-mediated activation later) results suggests that carbon oxidation step is not rate deter-
[28], in which CH4 forms adsorbed intermediates such as CHx or mining, but is slow enough to be detected in the transient pulse
formates. During CO2 pulse 2, the MS signal showed no traces of experiments. Oxidation of surface carbon is a crucial step in the
CH4 and H2 (Fig. 8a) suggesting no CHx, adsorbed formates, or DRM mechanism and is greatly influenced by the activation of CH4.
adsorbed H2 intermediates were present on the catalyst from
previous CH4/Ar pulse 1. This would mean that the activation of 3.7.4. L5RhZ CH4 pulse 1
CH4 is by direct dissociation over pyrochlores, where it decom- The process of methane activation appears to be occurring to a
poses completely to C(s) and H2. About 33.7% of CO2 is converted greater extent on the L5RhZ compared to the L2RhZ due to greater
during the CO2 pulse 2, but only 19.1% of the CO2 converted Rh content. The MS signal for CH4 pulse 1 over L5RhZ (Fig. 9a) sug-
resulted in CO formation (Table 5). The rest of the CO2 likely forms gests significant amounts H2 and CO formation. In addition, CO2
surface carbonates. The normalized response for CO2 in pulse 2 formation is an order of magnitude greater than that over L2RhZ.
suggests simultaneous exit of CO and CO2, which is slower than During this pulse, 64.3% of CH4 was converted, which is greater
diffusion of Ar (Fig. 8b). This suggests that CO is a product of either than that (55%) observed over L2RhZ. The experimental H2
the reduction of CO2 or the oxidation of surface carbon by CO2 or produced is about 82.5% of theoretical value calculated from CH4
both. The next CH4 pulse (pulse 3 below) provides insight into conversion (Table 5). The carbon balance computed from CO and
the mechanism of the formation of CO and oxidation of the surface CO2 formation accounts for about 61.8% of carbon reacted in CH4
carbon. pulse 1. The remaining carbon likely remains on the catalyst as
solid adsorbed carbon, C(s). The carbon balance is greater for
L5RhZ (61.8%) compared to L2RhZ (53.1%). The normalized
3.7.3. L2RhZ CH4 pulse 3
response suggests that the CH4 response is faster than Ar response
The MS signal intensity for H2 in CH4 pulse 3 is smaller than in
(Fig. 9b). H2 and CO formation are simultaneous, similar to what
CH4 pulse 1 (Fig. 8a). Quantification suggests that only 48.1% of CH4
was observed in L2RhZ CH4 pulse 1. The response for CO2 is slower
was converted in CH4 pulse 3 as opposed to 55% in CH4 pulse 1
than CO and H2; also there is a shift in the peak of CO2. The result-
(Table 2), possibly due to carbon (formed during CH4 pulse 1)
ing CO2 response curve has a higher residence time over the cata-
blocking the active sites. The H2 selectivity, however, remains
lyst than any other component. This could be either due to lower
essentially the same at 81.3% in CH4 pulse 3 (compared to 83.1%
diffusivity or increased interaction with the catalyst surface. The
in CH4 pulse 1). Carbon balance suggests that there was an increase
response of CO2 during the blank experiment was observed to be
in CO formed during CH4 pulse 3 compared to CH4 pulse 1. About
much faster than the CO2 response over L5RhZ; thus, the delay in
64% of carbon converted in CH4 could be accounted for by CO and
the CO2 response curve is due to interaction/slow surface reaction
CO2 formation. The carbon in CO and CO2 could be from the CH4
[84] and not due to lower diffusivity value of CO2. The fact that the
activated in CH4 pulse 3, reduction of the carbonates formed during
peak for CO2 appears later than the peak for CO and H2 suggests
CO2 pulse 2, and oxidation of the carbon formed during CH4 pulse
that CO2 might be formed from further oxidation of CO. A delay
1. It is important to note that carbon yield and formation of CO in
in the response curve would also mean that the catalyst has
the CH4 pulses is significantly greater than the CO2 pulse 2. This
strongly adsorbing CO2 sites, which increases its residence time
suggests that the presence of H2 has a positive effect on the forma-
over the catalyst [76].
tion of CO.
Consider the normalized CH4 pulse 3 to understand the role of
H2 in CO formation (Fig. 8b). H2 and CO exit the reactor simulta- 3.7.5. L5RhZ CO2 pulse 2
neously in spite of large differences in their molecular diffusivities, The CO2 pulse 2 for L5RhZ (Fig. 9) is qualitatively and quantita-
suggesting that both H2 and CO are produced in a same slow surface tively very similar to that observed for L2RhZ (Fig. 8). Conversion of
reaction [81]. Based on this, it can be postulated that H2 formed CO2 was about 32.7%, and no H2 and CH4 were observed in the MS
from dissociation of CH4 does not desorb immediately but reacts signal suggesting the absence of any CHx or H2 species. Thus, CH4
with oxygen associated with either surface carbonates or with the activates on L5RhZ by direct dissociation mechanism forming C(s)
pyrochlore lattice to form H2O. If this H2 bonds with carbonate oxy- and H2. About 22% of the CO2 converted resulted in CO formation;
gen, it would reduce the carbonates to CO and if it reacts with lattice the remaining 78% of the converted CO2 presumably results in car-
oxygen, it would reduce the pyrochlore (see Section 3.6). This H2O bonate formation and could not be completely reduced to CO due
then oxidizes the carbon formed from CH4 activation (C(s) + H2- to the absence of H2. The normalized response of CO is accompa-
O ? CO + H2), liberating CO and H2 simultaneously. Osaki et al. nied by that of CO2, suggesting that this limited amount of CO is
[82] observed the same trend of delayed H2 and CO response over formed either from the reduction of CO2 or from the oxidation of
Ni/SiO2 catalyst and proposed that oxidation of carbon by H2O is C(s) formed in the CH4 pulse 1.
the rate-limiting step in the mechanism. This step of carbon oxida-
tion by H2O over pyrochlores here must be a slow step in order to 3.7.6. L5RhZ CH4 pulse 3
cause a delay in the response of H2 and CO with respect to Ar During the CH4 pulse 3, the CH4 conversion dropped to 58.8%
[83]. The presence of strong CH4/CD4 isotope effect (as will be seen compared to 64.3% in CH4 pulse 1. As the carbon balance during
D. Pakhare et al. / Journal of Catalysis 316 (2014) 78–92 89
CH4 pulse 1 suggested substantial carbon formation, this presum- activation of mildly acidic CO2 and accelerates the rate of carbon-
ably blocked the active sites and resulted in lower CH4 activation. ate formation, consistent with the literature [2,15,43,60,86]. The
The H2 yield was 82.5% which is nearly the same as that in CH4 presence of La–O on the L2RhZ and L5RhZ surface increases the
pulse 1 (Table 5). The carbon balance improved in CH4 pulse 3 to rate of CO2 activation to form carbonates (Section 3.5). These
about 68.8% as compared to 61.8% for CH4 pulse 1. Some of the car- results suggest that the rds is activation or dissociation of CAH
bonates formed during CO2 pulse 2 were reduced to CO during CH4 bond in CH4.
pulse 3, which resulted in an improved carbon balance. The nor- Oxidation of carbon formed from CH4 activation is also an
malized response for CH4, H2, and CO during CH4 pulse 3 is very important step. From CH4 TPR (Section 3.6) and transient pulsing
similar to that during CH4 pulse 1. Simultaneous exit of H2 and experiments (Section 3.7), it was observed that surface carbon is
CO suggests oxidation of the surface carbon by H2O. However, oxidized by carbonates via H2O formation. This is an intermediate
there is a difference in the normalized CO2 response during the step assumed in the postulated mechanism. Thus, the most proba-
two CH4 pulses. The CO2 normalized response during CH4 pulse 3 ble DRM mechanism over L2RhZ and L5RhZ would be the dual-site
is delayed compared to CH4 pulse 1, likely due to increased catalyst [1] mechanism i.e., steps (1–6) and step (2) as an irreversible
interaction of CO2. Also, there are multiple peaks observed in pulse rate-determining step.
3. The multiple peaks suggest multiple reactions with varying rates The statistical fitting parameters K1, k2, K3–K6 determined by fit-
of CO2 formation [76]. The peak broadness suggests the presence of ting the kinetic data for L2RhZ and L5RhZ to the above rate expres-
different types of CO2 adsorbing sites on the catalyst surface, as sion are shown in Table 6. The parameters were estimated by using
reported by others [9,84]. For L5RhZ, the mechanism of CH4 activa- the function minimization Marquardt’s algorithm [47].
tion and product formation during CH4 pulse 1 differs from that Here, K1 and K3 are adsorption constants for CH4 and CO2
during CH4 pulse 3 as seen by the difference in the CO2 response adsorption, respectively. The rate constant for the rate-limiting
curve. This difference could also be present for L2RhZ but was dissociation of CH4 is given by k2, other constants K4–K6 are equi-
not clearly apparent due to limited CO2 formation. librium constants for the Steps 4–6. Note that for large values of K6
Increased CH4 conversion over L5RhZ compared to L2RhZ sug- (reaction of steam with surface carbon) and small values of K1
gests that the active site for CH4 is Rh. Also, L5RhZ showed better (adsorption of CH4, Step 1), the rate becomes first order in meth-
carbon balance numbers suggesting that Rh plays an important ane, as observed on other catalysts reported in the literature
role in the surface carbon oxidation. Considering that CO2 activa- [26,27,52,53].
tion is primarily by La sites on pyrochlores [11,60,85], the oxida- The parity plots (Fig. 10) for L2RhZ and L5RhZ show reasonable
tion of the surface carbon could be occurring at Rh–La interface. agreement of the predicted reaction rate of CH4 consumption and
This is in agreement with the transient experiments where the experimental reaction rate, over both L2RhZ compared to
L5RhZ having higher surface Rh concentration had higher carbon L5RhZ based on the adjusted R2 values. It can be concluded that
oxidation rate as seen by higher carbon balance. This greater num- a Langmuir–Hinshelwood dual-site mechanism with dissociative
ber of Rh–La interfacial sites on L5RhZ compared to L2RhZ leads to adsorption of CH4 as rds is consistent with the kinetic measure-
greater rates of surface carbon oxidation. ments, and also with the transient pulsing results. Differences in
the rates of the DRM reaction on the two catalysts are due to
differences in the nature of the surface Rh sites. Specifically, the
3.8. Kinetic rate modeling
L5RhZ contains surface Rh sites that activate CH4 via a lower
energy transition state (Eapp = 21.8 kcal/mol) than L2RhZ
The rate expression derived for the proposed mechanistic
(Eapp = 34.2 kcal/mol).
sequence is given in Eq. (7). Using this expression and kinetic data,
the adsorption and rate constants giving the best statistical fit were
3.9. Summary of the DRM reaction mechanism
determined.
A detailed characterization and activity study of L2RhZ and
k2 PCO PH2
r CH4 ¼ PCH4 1 þ K 1 PCH4 þ L5RhZ has been reported in our previous work [36]. The character-
K1 K 6 P H2 O
ization results suggested higher surface coverage of Rh on L5RhZ
K3
P H2 O ¼ K 5 P H2 PCO2 PCO ð7Þ compared to L2RhZ. Also, CH4 and CO2 conversion was observed
K4
to be greater for L5RhZ than L2RhZ as suggested by the activity
The two models giving a valid statistical fit had the same mech- results. To study the catalytic active sites responsible for dissocia-
anistic sequence but differed in the assumed rate-limiting step, i.e., tion of CH4/CO2 and formation of H2 and CO, a detailed study of the
either dissociation of CO2 or dissociation of CH4. The kinetics activation of CH4 and CO2 is reported here. This study showed
isotope effect results (Section 3.3 above) clearly showed a strong strong deuterium (CH4/CD4) isotope effect suggesting that break-
deuterium isotope effect when CH4 was replaced with CD4. Pres- ing of the CAH bond in CH4 is the rds. CH4 is activated on Rh sites
ence of alkaline and alkaline earth metals such as La, Mg, Ce, Ca, on both catalysts, although these sites appear to be different on
and Sr adds to the basicity of the catalysts, which helps in the L2RhZ compared to L5RhZ.
Table 6
Summary of the statistical fitting parameters for L2RhZ and L5RhZ obtained by fitting the kinetic data to the derived rate expression.
This H2O later reacts with surface carbon at the La–Rh interface
evolving CO and H2 simultaneously (Steps 6 and 7 in Fig. 11). Since
H2 and CO are observed to evolve much later than Ar, either the
surface diffusion of H2 or the oxidation of surface carbon is a slow
step. This delay in the formation of H2 and CO compared to Ar
could also be due to slow activation of CH4, which is rds. Since
CH4 and Ar are pulsed simultaneously, a slow CH4 dissociation
would result in slower formation of H2 and thus a further delay
in the H2O formation and subsequent oxidation of carbon. Oxida-
tion of surface carbon via H2O formation is an important step in
understanding DRM mechanism over these pyrochlores. In the pre-
vious rate modeling Section 3.8, a dual-site mechanism was consis-
tent with our kinetic data. These two types of sites that are
responsible for CH4 and CO2 activation have been clearly identified
as Rh and La respectively. Hence the previously reported reaction
mechanism can be restated as follows:
CH4 þ Rh¡CH4 Rh
C Rh þ H2 O¡CO þ H2 þ Rh
4. Conclusion
Fig. 11. Pictorial representation of the mechanistic steps of DRM over Rh-substituted pyrochlores. (This cycle is for graphical presentation purpose and does not necessarily
suggest the exact location of the surface atoms or the exact sequence of reaction steps.)
[27] A. Yamaguchi, E. Iglesia, J. Catal. 274 (2010) 52. [59] R.P. Turcotte, J.O. Sawyer, L. Eyring, Inorg. Chem. 8 (1969) 238.
[28] D.C. Seets, M.C. Wheeler, C.B. Mullins, Chem. Phys. Lett. 266 (1997) 431. [60] S. Irusta, L.M. Cornaglia, E.A. Lombardo, Mater. Chem. Phys. 86 (2004) 440.
[29] M.C.J. Bradford, M.A. Vannice, Catal. Rev.: Sci. Eng. 41 (1999) 1. [61] A. Olafsen, A.K. Larsson, H. Fjellvag, B.C. Hauback, J. Solid State Chem. 158
[30] N.D. Subramanian, J. Gao, X.H. Mo, J.G. Goodwin, W. Torres, J.J. Spivey, J. Catal. (2001) 14.
272 (2010) 204. [62] R.P. Taylor, G.L. Schrader, Ind. Eng. Chem. Res. 30 (1991) 1016.
[31] Y.J.O. Asencios, E.M. Assaf, Fuel Process. Technol. 106 (2013) 247. [63] T. Levan, M. Che, J.M. Tatibouet, M. Kermarec, J. Catal. 142 (1993) 18.
[32] B.C. Enger, R. Lodeng, A. Holmen, Int. J. Hydrogen Energy 37 (2012) 10418. [64] G. Vanhoyland, R. Nouwen, M.K. Van Bael, J. Yperman, J. Mullens, L.C. Van
[33] T. Johansson, D. Pakhare, D. Haynes, V. Abdelsayed, D. Shekhawat, J. Spivey, Poucke, Thermochim. Acta 354 (2000) 145.
Chem. Pap. (2014) 1. [65] B. Bachiller-Baeza, C. Mateos-Pedrero, M.A. Soria, A. Guerrero-Ruiz, U.
[34] D. Pakhare, D. Haynes, D. Shekhawat, J. Spivey, Appl. Petrochem. Res. 2 (2012) Rodemerck, I. Rodriguez-Ramos, Appl. Catal. B 129 (2013) 450.
27. [66] J. Guo, H. Lou, L. Mo, X. Zheng, J. Mol. Catal. A: Chem. 316 (2010) 1.
[35] D. Pakhare, C. Shaw, D. Haynes, D. Shekhawat, J. Spivey, J. CO2 Util. 1 (2013) 37. [67] B. Pawelec, S. Damyanova, K. Arishtirova, J.L.G. Fierro, L. Petrov, Appl. Catal. A
[36] D. Pakhare, H. Wu, S. Narendra, V. Abdelsayed, D. Haynes, D. Shekhawat, D. 323 (2007) 188.
Berry, J. Spivey, Appl. Petrochem. Res. 3 (2013) 117. [68] F. Pompeo, N.N. Nichio, M. Souza, D.V. Cesar, O.A. Ferretti, M. Schmal, Appl.
[37] P.J. Wilde, C.R.A. Catlow, Solid State Ionics 112 (1998) 173. Catal. A 316 (2007) 175.
[38] T. Osaki, T. Horiuchi, K. Suzuki, T. Mori, Catal. Lett. 35 (1995) 39. [69] J. Spivey, D. Pakhare, V. Abdelsayed, D. Haynes, D. Shekhawat, V. Schwartz, M.
[39] A.T. Ashcroft, A.K. Cheetham, R.H. Jones, S. Natarajan, J.M. Thomas, D. Waller, Smith, Abstr. Pap. Am. Chem. Soc. 246 (2013).
S.M. Clark, J. Phys. Chem. 97 (1993) 3355. [70] N. Kumar, M.L. Smith, J.J. Spivey, J. Catal. 289 (2012) 218.
[40] D.J. Haynes, D.A. Berry, D. Shekhawat, J.J. Spivey, Catal. Today 136 (2008) 206. [71] N.D. Subramanian, C. Kumar, K. Watanabe, P. Fischer, R. Tanaka, J.J. Spivey,
[41] A. Erdohelyi, J. Cserenyi, F. Solymosi, J. Catal. 141 (1993) 287. Catal. Sci. Technol. 2 (2012) 621.
[42] J.F. Munera, L.M. Cornaglia, D.V. Cesar, M. Schmal, E.A. Lombardo, Ind. Eng. [72] A.M. Efstathiou, A. Kladi, V.A. Tsipouriari, X.E. Verykios, J. Catal. 158 (1996) 64.
Chem. Res. 46 (2007) 7543. [73] J.A. Diaz-Guillen, A.R. Diaz-Guillen, K.P. Padmasree, J.M. Almanza, A.F. Fuentes,
[43] Z.L. Zhang, X.E. Verykios, S.M. MacDonald, S. Affrossman, J. Phys. Chem. 100 J. Santamaria, C. Leon, Bol. Soc. Esp. Ceram. Vidr. 47 (2008) 159.
(1996) 744. [74] L. Lisi, G. Bagnasco, P. Ciambelli, S. De Rossi, P. Porta, G. Russo, M. Turco, J. Solid
[44] Z.L. Zhang, X.E. Verykios, Catal. Lett. 38 (1996) 175. State Chem. 146 (1999) 176.
[45] J.T. Richardson, S.A. Paripatyadar, Appl. Catal. 61 (1990) 293. [75] J. Spivey, D.S. Pakhare, V. Schwartz, V. Abdelsayed, D. Haynes, D. Shekhawat,
[46] J. Nakamura, K. Aikawa, K. Sato, T. Uchijima, Catal. Lett. 25 (1994) 265. Abstr. Pap. Am. Chem. Soc. 246 (2013).
[47] M.M.B. Quiroga, A.E.C. Luna, Ind. Eng. Chem. Res. 46 (2007) 5265. [76] J.T. Gleaves, J.R. Ebner, T.C. Kuechler, Catal. Rev.: Sci. Eng. 30 (1988) 49.
[48] J.F. Munera, S. Irusta, L.M. Cornaglia, E.A. Lombardo, D.V. Cesar, M. Schmal, J. [77] G.C. de Arauj, S.M. de Lima, J.M. Assaf, M.A. Pena, J.L. Garcia Fierro, M.d.C.
Catal. 245 (2007) 25. Rangel, Catal. Today 133 (2008) 129.
[49] M.F. Mark, F. Mark, W.F. Maier, Chem. Eng. Technol. 20 (1997) 361. [78] J. Gao, Z. Hou, H. Lou, X. Zheng, Dry (CO2) reforming, in: D. Shekhawat, J.
[50] Z.L. Zhang, V.A. Tsipouriari, A.M. Efstathiou, X.E. Verykios, J. Catal. 158 (1996) Spivey, D. Berry (Eds.), Fuel Cells: Technologies for Fuel Processing, Elsevier,
51. Amsterdam, 2011, p. 191 (Chapter 7).
[51] A.M. O’Connor, Y. Schuurman, J.R.H. Ross, C. Mirodatos, Catal. Today 115 [79] A.N.J. vanKeulen, K. Seshan, J. Hoebink, J.R.H. Ross, J. Catal. 166 (1997) 306.
(2006) 191. [80] A.C. Luntz, J. Harris, Surf. Sci. 258 (1991) 397.
[52] J.M. Wei, E. Iglesia, J. Catal. 225 (2004) 116. [81] Y. Schuurman, V.C.H. Kroll, P. FerreiraAparicio, C. Mirodatos, Catal. Today 38
[53] J.M. Wei, E. Iglesia, J. Catal. 224 (2004) 370. (1997) 129.
[54] G. Sierra Gallego, C. Batiot-Dupeyrat, J. Barrault, F. Mondragon, Ind. Eng. Chem. [82] T. Osaki, T. Horiuchi, K. Suzuki, T. Mori, Catal. Lett. 44 (1997) 19.
Res. 47 (2008) 9272. [83] Y.H. Hu, E. Ruckenstein, J. Phys. Chem. B 101 (1997) 7563.
[55] X.E. Verykios, Int. J. Hydrogen Energy 28 (2003) 1045. [84] E.V. Kondratenko, H.H. Wang, V.A. Kondratenko, J. Caro, J. Mol. Catal. A: Chem.
[56] E. Nikolla, J. Schwank, S. Linic, J. Catal. 263 (2009). 297 (2009) 142.
[57] H.Y. Wang, C.T. Au, Catal. Lett. 38 (1996) 77. [85] L.M. Cornaglia, J. Munera, S. Irusta, E.A. Lombardo, Appl. Catal. A 263 (2004) 91.
[58] H.Y. Wang, E. Ruckenstein, J. Phys. Chem. B 103 (1999) 11327. [86] X.E. Verykios, Appl. Catal. A 255 (2003) 101.