Chemical Engineering Journal 390 (2020) 124629

Contents lists available at ScienceDirect

Chemical Engineering Journal

journal homepage: [Link]/locate/cej

CO2 methanation and reverse water gas shift reaction. Kinetic study based T
on in situ spatially-resolved measurements
Jose A. Hernandez Lalindea, Pakpong Roongruangsreea, Jan Ilsemannb, Marcus Bäumerb,c,
⁎
Jan Kopyscinskia,
a
Department of Chemical Engineering, McGill University, 3610 University Street, Montreal, Canada
b
Institute of Applied and Physical Chemistry, University of Bremen, Leobener Str.6, Bremen 28359, Germany
c
MAPEX Center of Materials and Processes, University of Bremen, Bremen 28359, Germany

H I GH L IG H T S

• Kinetic study of CO methanation and reversed water gas shift reaction over Ni/Al O .
2 2 3

• 21,000 gas composition data were collected via spatially-resolved measurement.

• Kinetic parameters of 20 different LH rate expression were estimated.
• Best fit models include oxygenated surface intermediate HCOO* or COH*.
• HCOO* and COH* surface species were validated with DRIFTS measurements.

A R T I C LE I N FO A B S T R A C T

Keywords: The reaction kinetics for the CO2 methanation and reverse water gas shift reaction over an ordered-mesoporous
Kinetic study, CO2 methanation Ni/Al2O3 catalyst were determined. For the parameter estimation and model discrimination, the kinetic data
Reverse water gas shift reaction were obtained by means of spatially-resolved measurement in a catalytic plate reactor. In detail, ~21,000 high-
Ni/Al2O3 resolution gas composition data were gathered along the reactor axis using a movable sampling capillary con-
Spatially-resolved measurements
nected to a mass spectrometer. Additionally, the catalyst surface temperature was determined via infrared
Diffuse Reflectance Infrared Fourier Transform
thermography. The influence of reaction temperature (320–420 °C), total pressure (1.2–7.3 barabs), and GHSV, as
Spectroscopy
well as possible inhibition of products such as CH4 and H2O, were investigated.
A one-dimensional model of the reactor was developed describing the conservation of mass in the bulk gas
and catalyst phase. The Bayesian approach was used to estimate the kinetic parameters of 20 proposed
Langmuir-Hinshelwood rate expressions for the CO2 methanation that were derived based on three different
mechanisms (i.e., direct dissociation, hydrogen assisted dissociation, and hybrid mechanism). Two kinetic
models reflected the measured data very well. The most probable models suggest that the rate determining step
includes the reaction of an oxygenated complex (COH* or HCOO*) with an active site (*) or an adsorbed hy-
drogen (H*). Furthermore, water was assumed to be adsorbed as a hydroxyl species (OH*), while methane did
not influence the reaction. Temperature- and time-resolved Diffuse Reflectance Infrared Fourier Transform
Spectroscopy (DRIFTS) measurements confirmed the presence of both adsorbed surface intermediates.

1. Introduction their corresponding kinetic description is paramount in designing an

efficient reactor and process. In the methanation reactor, the carbon
A tremendous amount of research is currently done on the CO2 oxides are hydrogenated over metal-based catalysts such as Ni [1–4],
methanation within the context of the Power-to-Gas (P2G) process that Ru [5–7], and Rh [8,9]. Although catalysts based on either Ru or Rh
enables the storage of electrical energy as grid compatible renewable have been shown to be more active than Ni, the latter still remains the
natural gas. The catalyzed CO2 conversion to CH4 is the heart of the most widely studied due to its comparative abundance and low cost.
P2G technology, and the understanding of the reaction mechanism with Many demonstration units and pilot plants have been designed and

⁎
Corresponding author.
E-mail address: [Link]@[Link] (J. Kopyscinski).

[Link]
Received 27 November 2019; Received in revised form 4 February 2020; Accepted 1 March 2020
Available online 04 March 2020
1385-8947/ © 2020 Elsevier B.V. All rights reserved.
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Nomenclature rj rate of reaction R1, R2 (CO2 methanation and reversed

water gas shift) [mol kgcat–1s−1]
b width of the catalyst plate [m] Ri rate of disappearance or formation of species i [mol kgcat–1
cb,i bulk gas concentration of species i [mol m−3] s−1]
cs,i concentration on the catalyst surface of species i [mol S(θk) objective function (Eq. (28))[-]
m−3] Sh Sherwood number[-]
Ca Carberry number [-] T temperature [K]
Di,mix diffusion coefficient of species i in the gas mixture [m2 Tref reference temperature (598.15 K = 325˚C) [K]
s−1] ui. experimental condition i [-]
EA activation energy [kJ mol−1] ν(θk) covariance matrix [-]
ΔHi adsorption enthalpy of species i [kJ mol−1] wi mass fraction of species i [-]
h(xi) catalyst height in interval i [m] xi molar fraction of species i [-]
H height of the channel [m] Δxi interval width [-]
KGi mass transfer coefficient of species i [m s−1] x, y co-ordinates [m]
Keq equilibrium constant [-] Yyu experimental observation response i and event u
Ki adsorption constant of species i [differ]
Ki0 pre-exponential factor of the adsorption constant Ki Greek symbols
[differ]
kj rate constant of reaction j [differ] εi(φ) error model in posterior distribution (Eq. (27)) [-]
kj0 pre-exponential factor for rate constant kj [differ] θE,j dimensionless activation energy [-]
fk(θk,ui) k-th model with parameters θk experimental condition ui fk θH,i dimensionless heat of adsorption [-]
[-] θk, j dimensionless reaction rate constant [-]
lch characteristic length (ratio of volume to surface) [m] θK,i dimensionless adsorption constant [-]
Mi molar weight of species i [g mol−1] Θi specie i surface site concentration [-]
m catalyst mass [kgcat] Θv vacant surface site concentration [-]
ṅb, i molar flow of species i [mol s−1] θk parameters for k-th model (Eq. (27)) [-]
pi partial pressure of species i [bar] ωcat catalyst length density [kgcat m−1]
p (Mk |D , ε ) posterior probability [-] Ψ Weisz modulus [-]
R universal gas constant = 8.314 [J mol−1 K−1] νij stoichiometric factor [-]

tested, as summarized in [10–14]. Determine the reaction kinetics for processes in a chemical reactor and their interactions are represented
the strongly exothermic CO2 methanation reaction (R1) is a challenge, adequately [2]. To properly describe and quantify the methanation
especially as it is part of a reaction network comprising of the reverse kinetics, the rate of the RWGS should be included, which is often ne-
water gas shift (RWGS, R2) and the CO methanation (R3) reaction. glected [4–6,15,16]. Kopyscinski et al. [2] and Zhang et al. [17] in-
cluded the WGS reaction as part of the CO methanation network and
CO2 + 4H2 ↔ CH4 + 2H2 ΔHR0 = −165 kJ mol−1 R1 kinetic study, while Champon et al. [18] included the RWGS in CO2
methanation. Until now, none of these studies identify the adsorbed
CO2 + H2 ↔ CO + H2 O ΔHR0 = 41 kJ mol−1 R2
surface species and intermediates for the validation of their model as-
sumptions.
CO + 3H2 ↔ CH4 + H2 O ΔHR0 = −206 kJ mol−1 R3
Many kinetic studies have been conducted over Ni and Ru based
A good model prediction can only be obtained if all relevant catalysts that proposed different reaction mechanisms and elementary

Table 1
Kinetic models for CO2 methanation on Nickel-based catalyst.
Catalyst & experimental conditions Proposed kinetic model equation EA [kJ mol−1] Ref

62 wt% 160–180 °C 0.5

r CH4 = k·pCO n/a [25]
2
Ni/Cr2O3 1 bar
28 wt% 200–230 °C k·pCO2 106 [4]
r CH4 =
Ni/Al2O3 1 bar 1 + KCO2·pCO2
58 wt% 275–320 °C 0.66 0.21
r CH4 = k·pCO ·p and n/a [16]
2 H2
Ni/SiO2 17 bar k·pCO2 ·p H2
r CH4 =
1 + K H2·pH 2 + KCO2·pCO2
60 wt% 260–400 °C 4
k·pCO2 ·p H 55–58 [23]
2
Ni/SiO2 1 bar r CH4 =
( 1 + K H2·p H2 + KCO2·pCO2 )5
3 wt% 227–327 °C 0.5 ·p 0.5
k·pCO 94 [20]
2 H2
Ni/SiO2 1 bar r CH4 = 0.5 ·p 0.5 + K ·p 0.5 / p 0.5 + K ·p 2
( 1 + K1·pCO 2 H2 2 CO2 H2 3 CO )
Ni 250–350 °C 0.33 ·p
k·pCO n/a [26]
2 H2
− r CO2 =
1 + KCO2·pCO2 + K H2·pH 2 + K H2 O·p H2 O
15 wt% 240–320 °C 0.33 ·p 0.5
k·pCO 72.5 [21]
2 H2
Ni/Al2O/ 1 bar r CH4 = 0.5 + K ·p 0.5 + K 2
( 1 + KCO2·pCO 2 H2 H2 H2 O·p H2 O )
La2O3
NiAl(O)x 180–340 °C ⎛
2 ·p
pH 77.5 [5]
0.5 ·p 0.5 · 1 −
k·pCO 2 O CH4 ⎞
15 bar 2 H2 ⎜⎜ 4 ·p
pH ·K
⎟⎟
⎝ 2 CO2 eqM ⎠
r CH4 = 0.5 + K ·p 0.5 + K 0.5 2
( 1 + Kmix·pCO 2 H 2 H2 OH ·p H2 O / p H2 )

2
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

steps (i.e., direct C–O bond and hydrogen assisted C–O bond dissocia- CO2 methanation and reverse water gas shift reaction over an ordered-
tion) and thus different reaction rate expressions [4–6,15,18–23]. mesoporous Ni/Al2O3 catalyst. We demonstrate the use of an optimized
Though much effort has been made, there is no consensus on the re- optically accessible catalytic plate reactor with spatially resolved
action mechanism and its kinetic description [24]. This can be ex- measurement of the catalyst surface temperature and gas composition
plained by (1) different reaction conditions (e.g., temperature, partial profiles along the reactor axis and report one of the most comprehen-
pressure), reactor concept, and catalyst used, (2) different assumptions sive experimental kinetic study for both the CO2 methanation and re-
of the governing rate expressions (e.g., Langmuir-Hinshelwood, Power verse water gas shift reaction. Kinetic parameters of 20 different
Law, elementary steps, rate determining step) and reaction inter- Langmuir-Hinshelwood rate expressions were estimated by comparing
mediates (e.g., adsorbed surface species), and lastly (3) different in- the measured gas composition profiles with those derived from the
terpretation of the kinetic results. Two mechanisms have evolved in the models. The model discrimination was done based on the posterior
literature, which are summarized in Fig. S1. Mechanism A consists of probability share [46,47] as well as on the identification of adsorbed
the direct dissociation of CO2 to adsorbed CO* and O*, and further, to surface species during CO2 methanation via infrared spectroscopy.
surface oxygen O* and carbon C*, which are subsequently hydro-
genated to H2O and CH4, respectively. Mechanism B assumes the hy- 2. Experimental
drogen assisted dissociation via the formation of formyl and/or for-
mates species (e.g., COH*, HCOO–*). 2.1. Kinetic data acquisition
Early methanation kinetic approaches attempted to describe the
reaction without the assumption of a rate-determining step by using a An ordered mesoporous nickel-alumina catalyst, ~30 wt% NiO/
simple power law [4,16,25], while later models use complex Lang- Al2O3 (OMA-30Ni-500), was used in this study, a detailed explanation
muir–Hinshelwood (L-H) rates expressions (Table 1) that include ad- of the catalyst preparation can be found in [48]. The kinetic experi-
sorption constants for some adsorbed species such as CO2, H2, CO and ments were conducted in an optically accessible channel reactor in
H2O [20,23]. The latter was assumed to adsorb as H2O [21,26] or as which a catalyst coated plate is placed at the bottom. A description of
hydroxyl (OH) [5]. Moreover, only a few studies included an equili- the coating procedure onto the metal plates that were utilized in the
brium term for the CO2 methanation [5]. channel reactor can be found in [3]. The setup and experimental pro-
Until now, all published kinetic studies for the CO2 methanation cedure for kinetic data acquisition are published in our previous work
rely on measuring the exit gas composition (1 data point per species and [29].
experiment) using differential and/or integral reactors (e.g., small la- Briefly, the reactor channel with the catalyst coated plate was en-
boratory fixed bed reactors either with or without gas recirculation) closed from the top with a set of two quartz glass plates, enabling
[4–6,15,21,27,28]. Moreover, most traditional kinetic studies are per- catalyst surface temperature measurement via IR thermography (i.e.,
formed with highly diluted gas mixtures (> 90% inert gas) and/or short wave infrared camera, FLIR SC2500, US). From the lateral side, a
catalyst beds to maintain isothermal behavior within the reactor and to small (0.5 mm outer diameter) movable stainless-steel (SS316) capillary
avoid a huge temperature increase due to the exothermic reaction. was introduced, enabling spatially-resolved measurement of the gas
Experiments in differential reactors are usually limited to low CO2 composition above the coated catalyst via a calibrated quadrupole mass
conversions of less than 20%. Using an integral reactor, on the other spectrometer (Hiden Analytical, HPR-20, UK). Preliminary tests con-
hand, allows conducting experiments resembling industrial settings firmed that the capillary, as well as the uncoated metal plate (i.e.,
with CO2 conversions of up to 90%. However, significant hetero- FeCrAl-loy), were not active towards the CO2 methanation [3,29].
geneities such as non-uniform catalyst mass distribution, concentration, Prior to the experiments, the catalyst was reduced with 20 mol% H2
and temperature gradients along the axial and radial directions, make (99.999% MEGS Specialty Gases) in Ar (99.999% MEGS) for 8 h at the
kinetic measurements (data collection) and modeling (parameter esti- respective reduction temperature, determined via H2-TPR [48]. Before
mation and model discrimination) difficult. Spatially-resolved mea- every experiment, the reactor with the installed catalyst plate was he-
surement of the gas composition and catalyst surface temperature along ated up to the desired temperature under a constant Ar and H2 (20 mol
the reactor axis allow closing the knowledge gap as they combine dif- %) flow. The sampling capillary orifice was moved to the zero position,
ferential and integral reactor systems [2,29]. A number of techniques i.e., the beginning of the plate. Then hydrogen was set to the desired
have been designed to provide spatially-resolved profiles in different flow rate. At this point, a first IR line profile, a 2D IR emission map, and
reactor configurations such as catalytic plate (channel) reactors a qualitative emission IR picture were recorded. Then CO2 and CH4
[2,3,29,30], monolith [31–34], fluidized bed [35–37], coated metal were fed stepwise until the desired volumetric flow rates were reached.
foam [38,39] and fixed bed reactors [40–42]. Channel reactors have The moving of the sampling capillary was started after the concentra-
advantages over fixed bed, fluidized bed and coated metal foam re- tion at the zero position had reached a steady-state value. At this time, a
actors as they have a high surface-to-volume ratio and thus an excellent second IR line profile, a 2D IR emission map, and a qualitative emission
heat transfer and small pressure drop. Additionally, the catalyst mass IR picture were recorded. The line and 2D temperature differences
distribution along the channel reactor axis can be accurately de- between the first and the second temperature profile were calculated.
termined via profilometry [3], which is not possible for the other The capillary was held for 45–900 s at each position and moved in the
configurations (i.e., they require an approximation via bulk density and flow direction to the end of the reactor. In each (steady-state) experi-
bed voidage). The bed temperature is commonly measured via an in- ment, up to 25–500 QMS spectra were measured and averaged to de-
serted thermocouple (limited to the bulk phase) or pyrometer (sta- termine the molar fractions at each axial position. The spatial-resolu-
tionary or movable) [43,44], which provides only a point value per tion was set to 1 data point per millimeter over the coated area and per
space unit. A higher thermal resolution is obtained through the usage of 2 mm over the non-coated area. Additionally, 3 data points per milli-
an optically accessible reactor, which enables collecting thermal ra- meter for the IR measurements (i.e., 9 data points per 1 mm2 surface)
diation via IR - thermography [30]. Mid-wave (3000–5000 nm) [31,45] were collected. Experiments with water in the feed were carried out in a
and short-wave (900–1700 nm) infrared [2,29] cameras have been used similar way. Ar, H2 and CH4 were fed through the temperature-con-
to determine the catalyst surface temperature in various configurations. trolled water saturator, while CO2 was by-passed and then mixed with
Short-wave infrared has shown to be effective for our temperature the saturated gas right before the channel reactor.
measurement during CO2 methanation reaction conditions, where To investigate the stability of catalyst and coating, a long-term ex-
polyatomic (i.e., IR absorbing in the mid-wave range) molecules are periment (130 h) coupled with IR thermography analysis was con-
present [29]. ducted. The reactor temperature was set to 340 °C with a total flowrate
The objective of this manuscript is to derive a kinetic model for the of 100 mLN min−1, H2/CO2 = 4, and 20% Ar (reference experiment #1

3
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

and Table 2). After 75 h, the temperature was increased to 380 °C to temperatures high enough for in-situ reduction, the pellets were pre-
complete a total of 130 h on stream. Every day, the reference experi- reduced in an external oven in pure H2 at 500 °C for 8 h and stored in
ment (#1, Table 2) was repeated to evaluate catalyst deactivation and inert gas atmosphere before being placed on the temperature-controlled
verify the repeatability of the results. In addition, selected experiments sample holder in the IR cell. To remove any impurities and air in the
were repeated at least once and up to 3 times. chamber prior to the spectra collection, the sample was heated to
Conversion of carbon dioxide (XCO2) and product selectivity (Si) 400 °C in H2 for 30 min with a heating ramp of 5 °C min−1. Following
were defined per Eqs. (1) and (2). that, two types of experiments were conducted: (i) temperature-re-
solved and (ii) time-resolved at atmospheric pressure. For the former,
ṅCO2, in − ṅCO2, out
XCO2= the sample was cooled down to 100 °C in He, and the background was
ṅCO2, in (1) recorded. Next, the sample was exposed to 5 mlN min−1 CO2, 20 mlN
min−1 H2, and 75 mlN min−1 He, and the temperature was successively
ni̇ ramped up to 350 °C in intervals of 20 °C. Each temperature was held
Si =
ṅCO2, in − ṅCO2, out (2) for 5 min to ensure thermal equilibration before a spectrum was re-
corded. For the time-resolved experiment, the sample was cooled down
where ṅi is the molar flow rate of species i (i.e., CH4, CO). In all the
to 350 °C, and the background was recorded in 20 mlN min−1 H2 and
experiments, only CH4 and CO, but no C2+ nor higher hydrocarbons
80 mlN min−1 He. Next, the reactant mixture consisting of 5% CO2,
were detected within the MS signal at mass-charge ratio m/z = 1–45
20% H2, and 75% He with a total flow rate of 100 mlN min−1 was in-
(results not shown here, refer to [48]). For the kinetic modeling, nearly
troduced for 15 min, and the build-up of the surface species was tracked
110 experimental runs with a total of 21,000 high-quality data points
by collecting spectra every 2 min. After 15 min, CO2 was removed from
(gas composition) were used, see Table S1. The kinetic experiments
the feed, and only H2 and He were supplied for 40 min, during which
were conducted at six different temperatures (320–420 °C), three dif-
spectra were collected every two minutes for the first 15 min and every
ferent flow rates (50–150 mLN min−1), seven different total pressures
5 min for the remaining time.
(1.2–7.3 barabs) and seven plates with different catalyst loadings
(77–113 mg, Table S2) as summarized in Table 2.
3. Reactor modeling and parameters estimation

• Experiments 1 to 5 consider different flow rates and H /CO ratios 2 2

3.1. Catalytic plate reactor model
at various temperatures and pressures.
• Experiments 6 to 9 account for the influence of CH and H O at 4 2
The catalytic plate reactor represents a reactive flow between two
various temperatures and pressures.
• Experiments 10 and 11 investigate the direct biogas methanation at
parallel plates [29]. For the kinetic parameter estimation, a one-di-
mensional reactor model was developed similarly to [2], which consists
various temperatures and pressures.
of a bulk gas and a catalyst phase (see Eqs. (3) and (4), respectively).
∂nḃ , i Sh·Di, mix
2.2. Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) 0=− − b· ·(cb, i − cs, i )
∂x lch (3)
DRIFTS studies were performed to identify adsorbed surface inter- Sh·Di, mix
mediates during the CO2 methanation in order to validate the proposed 0 = b· ·(cb, i − cs, i ) + ωcat ·Ri
lch (4)
models. The measurements were conducted on a Varian 670 FT-IR
spectrometer with a liquid nitrogen cooled MCT detector and an IR cell 2 kg
⎡ mol ⎤ = ⎡m · 1 · m ⎛ mol ⎞⎤ + ⎡ cat · mol ⎤
in praying mantis geometry (Harrick). All spectra were recorded with a 3 ⎥ ⎢ m s·kg ⎥
⎣ s·m ⎦ ⎢ ⎣ m s ⎝ m ⎠⎦ ⎣ cat ⎦
resolution of 4 cm−1
, and the average of 128 scans is reported in ab-
sorbance mode. where ṅb, i (mol s−1) is the molar flow rate of species i in the bulk phase,
Prior to the measurements, 25 mg of the catalyst was thoroughly b (m) is the width of the channel, cb, i and cs, i (mol m−3) are the con-
mixed with 100 mg KBr powder and pelletized into four equally shaped centrations of species i in the bulk phase and on the catalyst surface, ωcat
discs (4 mm in diameter, 2 mm in height). As the cell could not achieve (kgcat m−1) the catalyst mass distribution (see Table S2) determined by

Table 2
Experimental conditions for CO2 methanation kinetic modeling in CPR.

exp. total flow Ar H2 CO2 CH4 H2O temp pressure

–1
no [mlN min ] [mol%] [°C] [barabs]
1* 100 20 64 16 - -
2 50 20 64 16 - -
3 150 20 64 16 - - 320 1.2
4 100 34 55 11 - - 340 2.2
5 100 36 48 16 - - 360 2.7
effect of CH4 and H2O 380 3.1
6 100 10 64 16 10 - 400 4.0
7 100 5 64 16 15 - 420 5.4
8 100 14.5 64 16 - 5.5 7.3
9 100 20 58 14.5 - 7.5
Biogas methanation
10 100 20 49 12.5 18.5 -
11 150 20 49 12.5 18.5 -
* base case/reference experiment

4
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

profilometry [29] and Ri (mol s−1 kgcat–1) is the rate of mol change of formates (e.g., COOH* or HCOO*), which subsequently react to CH*,
species i, which includes the proposed rate expressions. The mass CO* and/or COH*/HCO*. But also the formation of H2COO* cannot be
transfer between the two phases was expressed in terms of a Sherwood ignored. The COH*/HCO* species further dissociate to OH* and C* or
number (Sh [2,49]) that included the characteristic length lch (m) and CH*. The latter two species are subsequently hydrogenated further to
molecular diffusion coefficient in the gas mixture Di,mix (m2 s−1). The methane. Transient isotopic studies and microkinetic modeling of me-
latter was calculated via the Wilke and Fuller correlations [50]. The thane reforming over Ni/MgO-MgAl2O3 also have assumed formyl
model assumes an average bulk gas composition and average tem- (HCO*) species formed by the splitting of formates (HCOO*) as inter-
perature perpendicular to the flow direction (along the channel height). mediates, the latter resulting from the adsorbed carbon dioxide and
Moreover, isothermal conditions, as well as constant catalyst mass hydrogen [58]. Rate equations derived from the latter approach, with
distribution along the reactor axis, can be assumed, which was proven formyl (HCO*) formation taken as the rate-determining step, have been
via profilometry and high-resolution 2D temperature profile in our reported to result in identical rate equations as compared to the for-
previous work [3]. Moreover, the temperature difference between mation of carbon-hydroxyl being the rate-determining step [59].
catalyst surface and quartz glass plate was minimized to less than 5 °C, Supporting this mechanism, early approaches tried to identify cor-
as described in our previous work [29]. Therefore, the average bulk gas responding surface species using DRIFTS for studying CO2 methanation
temperature was 2.5 °C lower than the measured catalyst temperature; over Ni/Al2O3 [60], Ru/TiO2 [55], Rh/Al2O3 [61,62]. Surface bound
thus, heat transfer limitation in the bulk phase could be neglected. hydrogen-carbonate species HCO3– were observed, and its formation
The catalyst particles of 20 to 45 μm were used in creating the attributed to the reaction of CO2 with a surface hydroxyl group (OH).
coatings, which yielded a maximum thickness of 60 μm in the flat area The carbonate species are reduced to formates HCOO– and are subse-
of the coating, see Fig. S2. Thus, transport resistance (i.e., bulk and pore quently hydrogenated. On the other hand, FTIR studies for CO2 hy-
diffusion as well as heat transfer) was expected to be negligible, which drogenation over Pd/ZrO2 [63] identified the same surface compounds
was confirmed by calculating the Weisz modulus and Carberry number but did not agree with the HCO3– as a precursor for formates Similar
(see section 4.2.3). species were reported recently on the CO2 methanation on Ni/γ-Al2O3
using FTIR spectroscopy [64]. Carbonates and formates were suggested
3.2. Reaction mechanism and formulation of rate equations as the main adsorption and intermediary species, respectively. IR stu-
dies of CO2 methanation over Ni/Ce0.5Zr0.5O2 suggested carbonates
CO2 methanation can proceed through multiple pathways; based on CO32– and hydrogen carbonates HCO3– as intermediate species. These
the assumption of a specific reaction intermediate. As mentioned in the species are further reduced and hydrogenated to form formates and
introduction, two different reaction mechanisms have been proposed. then probably formaldehyde-type and methoxy species, to finally re-
lease methane [65]. Similar results were found during IR studies on the
3.2.1. Direct CO2 dissociation to adsorbed CO* and surface carbon (C*) – CO2 methanation over NiUSY zeolites [66], and provide evidence that
mechanism A CO arises from formates species when temperatures are below 200 °C.
Mechanism A has been suggested by several authors
[5,20,21,23,51–54]. This mechanism assumes the complete dissocia- 3.2.3. Hybrid – mechanism AB
tion of CO2 and subsequent hydrogenation of C* to carbenes (CH2*) and A hybrid mechanism AB, was suggested (Table 5) that includes the
methane (CH4), and the hydrogenation of adsorbed oxygen to water direct CO2 dissociation to adsorbed CO* (steps AB1 to AB3), then the
(Table 3). Hydrogen and carbon dioxide are assumed to adsorb dis- hydrogen assisted C–O bond dissociation via COH* or HCO* (AB4 to
sociative on the surface of the metal nanoparticles (such as Ni). Based AB8). It is hypothesized within this work that the first C–O bond
on this sequence, it is possible to derive multiple Langmuir-Hinshel- cleavage requires less energy and is faster than the second C–O bond
wood (L-H) rate expressions, depending on the assumption of the rate cleavage, which is activated by hydrogen. The remaining elementary
determining step (RDS). The proposal of more than one rate-de- steps are similar for mechanisms A and B; however, the rate equations
termining step is possible since, in the course of the reaction, several are formulated differently, see supplementary information.
mechanistic regimes involving different rate controlling steps might be
found, depending on the conditions at which the reaction is taking
place. It is clear that additional mechanisms could be formulated, and 3.3. Formulation of rate equations
rate equations could be determined accordingly. Therefore, the as-
sumptions made on the mechanisms based on steady-state measure- The development of rate equations follows analogously the metho-
ments in a lab-scale need to be complemented with the identification of dology adopted by Kopyscinski et al. [2,67] for the CO methanation.
the surface species, since several identical rate equations could be de- The surface coverage of the reactants and products is calculated from
rived from this mechanism. Recent attempts have been made to identify
the surface species involved in the CO2 methanation using spectro- Table 3
Direct carbon–oxygen bond dissociation – Mechanism A.
scopy. Diffuse Reflectance Infrared Fourier transformed spectroscopy
(DRIFTS) Recent experiments showed that the adsorption/dissociation Reaction Description Step
of CO2 is carried out on metallic centers to produce CO* and CO2*
H2 + 2* ↔ 2H* A1
species [8]. The latter is dissociated to CO* and O* (see A3 in Table 3). CO2 + * ↔ CO2* RDS Adsorption of CO2 A2
Subsequently, CO* is dissociated on metallic centers producing C* and CO2* + * ↔ CO* + O* RDS Dissociation of CO2 A3
O* (see A4 in Table 3). C* is then hydrogenated to CH4 (see steps CO* +* ↔ C* + O* RDS Dissociation of CO to surface C A4
A5–A8 in Table 3). In this sense, CHx/CnHx/CnHxO/ CHx - CO inter- C* + H* ↔ CH* + * RDS Hydrogenation of C A5
CH*+ H*↔ CH2* + * RDS Hydrogenation of CH A6
mediates have also been detected with Raman [66].
CH2* + H* ↔ CH3* + * RDS Hydrogenation of CH2 A7
CH3* + H* ↔ CH4* + * Hydrogenation of CH3 A8
3.2.2. Hydrogen assisted carbon-oxide bond dissociation – mechanism B CH4* ↔ CH4(g) + * Desorption of CH4 A9
This mechanism assumes that adsorbed hydrogen facilitates the O* + H* ↔ OH* + * OH formation A10
dissociation of both C–O bonds by lowering the dissociation energy [8]. OH* + H* ↔ H2O* + * H2O formation A11
CO* ↔ CO(g) + * Desorption of CO A12
Table 4 summarizes the proposed sequence of elementary steps in H2O* ↔ H2O(g) + * H2O desorption A13
analogy to the so-called hydrogen assisted pathway for CO methanation
through the intermediation of an oxygenated compound [19,55–57]. * Empty active site, x* adsorbed species, RDS rate determining step (assumed in
The adsorbed CO2* reacts with an adsorbed hydrogen to form surface literature)

5
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Table 4 (A, B, and AB).

Hydrogen assisted of carbon–oxygen bond dissociation. Mechanism B. The surface coverage and equilibrium constants of all species based
Reaction Description Step on the reaction mechanism A, see Table 2, are included in the supple-
mentary information. The reaction rates r and r’ are according to the
H2 + 2* ↔ 2H* B1 reaction mechanism A (Table 2). The surface coverage, equilibrium
CO2 + * ↔ CO2* RDS Adsorption of CO2 B2
constants of all species, and reaction rates were derived analogously for
CO2* + H* ↔ HCOO* + * RDS Formation of formates B3
HCOO* + * ↔ CO* + OH* RDS Dissociation of formates to CO B4
mechanisms A, B, and AB and are included in the supplementary in-
HCOO* + H* ↔ COH* + OH* RDS Dissociation of formates to COH B5 formation.
COH* + H* ↔ CH* + OH* RDS Dissociation of formyl B6
COH* + * ↔ C*+ OH* RDS Dissociation of formyl B7
3.3.1. CO2 methanation model 1 and 2
CH* + H* ↔ CH2* + * RDS Hydrogenation of CH B8
CH2* + H* ↔ CH3* + * RDS Hydrogenation of CH2 B9 The model development taking into consideration the RDS in Tables
CH3* + H* ↔ CH4* + * Hydrogenation of CH3 B10 3 to 5 is briefly shown as follows. CO2 adsorption is assumed to be the
CH4* ↔ CH4(g) + * Desorption of CH4 B11 rate determining step. CO2 is adsorbed as a molecule on a single site or
O* + H* ↔ OH* + * OH formation B12 two (double) adjacent sites. These model approaches are valid for
OH* + H* ↔ H2O* + * H2O formation B13
H2O* ↔ H2O(g) + * H2O desorption B14
mechanisms A, B, and AB thus, the RDS are steps A2 and B2 and AB2:
CO* ↔ CO(g) + * Desorption of CO B15
r2 = k2·pCO2 ·Θv ⎫
A2: ΘCO2 = K CO2·pCO2 ·Θv
* Empty active site, x* adsorbed species, RDS rate determining step r '2 = k '2·ΘCO2 ·Θv ⎬
⎭ (12)

Table 5 For a single site adsorption (CO2 + * ↔ CO2*), model 1:

Hybrid model. Mechanism AB.
rCH4 = kr ·pCO2 ·Θv (13)
Reaction Description Step
For a double site adsorption (CO2 + 2* ↔ CO2**), model 2:
H2 + 2* ↔ 2H* AB1
CO2 + * ↔ CO2* Adsorption of CO2 AB2 rCH4 = kr ·pCO2 ·Θ2v (14)
CO2* + * ↔ CO* + O* Dissociation of CO2 AB3
CO* + H* ↔ COH* + * RDS Formation of formyl AB4 For both cases, the catalyst surface is covered by adsorbed H, and
CO* + 2H* ↔ COH2* + 2* RDS Formation of di-hydrogenated CO AB5 OH, therefore the site concentration balance is:
COH* + * ↔ CH* + O* RDS Dissociation of formyl AB6
COH* + H* ↔ CH* + OH* RDS Dissociation of formyl AB7 Θv = 1 − ΘH − ΘOH (15)
COH2* + * ↔ CH* + OH* RDS Formation of CH species AB8
CH*+ H* ↔ CH2* + * RDS Hydrogenation of CH AB9 All CO2 methanation models presented can be summarized in one
CH2* + H* ↔ CH3* + * RDS Hydrogenation of CH2 AB10 generalized rate expression for the methanation reaction applying 20
CH3* + H* ↔ CH4* + * Hydrogenation of CH3 AB11 different sets of exponents, see Table 6.
CH4* ↔ CH4(g) + * Desorption of CH4 AB12
O* + H* ↔ OH* + * OH formation AB13 2
p H O ·pCH4
OH* + H* ↔ H2O* + * H2O formation AB14
a
k1·K Cx ·pCO 2
·K Hb2 ·pHc2 ·pHd2 O ⎜⎛1 − ⎛ 4 2
⎜
⎞ ⎟⎞
⎟

H2O* ↔ H2O(g) + * H2O desorption AB15 ⎝ ⎝ p H2 ·pCO2 ·K eqM ⎠ ⎠

rCH4 = i
CO*↔ CO(g) + * Desorption of CO AB16 p H2 O
⎛1 + K Cx ·p e ·K Hf ·p g ·p h + K H2·p H2 + K OH · ⎞
CO2 2 H2 H2 O p H2
⎝ ⎠ (16)
* Empty active site, x* adsorbed species, assumed RDS rate determining step
where k1 is the rate constant for the CO2 methanation, KCx denotes the
equilibrium adsorption of the considered species. The amounts of va- adsorption constant of the intermediate carbon species, which is a
cant sites are described by: combination of the adsorption constants of Cx, CO2, H2 and OH de-
pending on the RDS. All coefficients of the rate equation are summar-
Θv = 1 − ∑ Θi ized for the different RDS assumptions. The partial pressures pi is
1 (5)
where Θi defines the surface coverage of species i, which depends on the Table 6
adsorption constant, partial pressure, and vacant sites: Parameters and exponents of the general rate Eqs. (16) and (18) for different
RDS.
Θi = f (Ki, pi , Θv ) (6)
# Mech Kcx a b c d e f g h i
The rate of adsorption and desorption for hydrogen (H2 + 2* ↔
2 H*) are 1 A, AB, B – 1.0 – – – – – – – 1.0
2 A, AB, B – 1.0 – – – – – – – 2.0
r1 = k1·p H2 ·Θ2v (7) 3 A3, AB3 KCO2 1.0 – – – 1.0 – – – 2.0
4 A4 KCO 0.5 – – – 0.5 – – – 2.0
r '1 = k '1·Θ2H 5 A5 KC 0.33 0.5 0.5 – 0.33 – – – 2.0
(8)
6 A6 KCH 0.33 1.0 1.0 – 0.33 0.5 0.5 – 2.0
where r’ denotes the reverse reaction. In equilibrium, it is written as 7 A7 KCH2 0.33 1.5 1.5 – 0.33 1.0 1.0 – 2.0
8 AB4 KCO 0.5 0.5 0.5 – 0.5 – – – 2.0
k1 9 AB5 KCO 0.5 1.0 1.0 – 0.5 – – – 3.0
Θ2H = ·p H2 ·Θ2v
k '1 (9) 10 AB6 KCOH 0.5 0.5 0.5 – 0.5 0.5 0.5 – 2.0
11 AB7 KCOH 0.5 1.0 1.0 – 0.5 0.5 0.5 – 2.0
and with the definition of the adsorption coefficient 12 AB8 KCOH2 0.5 1.0 1.0 – 0.5 1.0 1.0 – 2.0
13 AB9 KCH 0.25 1.0 1.0 – 0.25 0.5 0.5 – 2.0
k1 14 AB10 KCH2 0.25 1.5 1.5 – 0.25 1.0 1.0 – 2.0
= K1
k '1 (10) 15 B4 KHCOO 1.0 0.5 0.5 – 1.0 0.5 0.5 – 2.0
16 B5 KHCOO 1.0 1.0 1.0 – 1.0 0.5 0.5 – 2.0
the surface coverage of one hydrogen atom is defined as: 17 B6 KCOH 1.0 1.5 2.0 –1.0 1.0 1.0 1.5 –1.0 2.0
18 B7 KCOH 1.0 1.0 1.5 –1.0 1.0 1.0 1.5 –1.0 2.0
ΘH = (K H2·p H2 )0.5·Θv (11) 19 B8 KCH 1.0 2.0 3.0 –2.0 1.0 1.5 2.5 –2.0 2.0
20 B9 KCH2 1.0 2.5 3.5 –2.0 1.0 2.0 3.0 –2.0 2.0
where K1= K H2 . Eq. (11) is, therefore, valid for all proposed mechanisms

6
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

expressed in bar and the reaction rate rj in mol s−1 kgcat–1. K eqM is the 3.4. Parameter estimation and model discrimination
equilibrium constant for CO2 methanation. The units of the reaction
and adsorption constants depend on the exponents used. Eq. (16) shows The parameters to be estimated were the pre-exponential factors of
that adsorbed hydrogen, hydroxyl, and an intermediate active carbon the rate and adsorption coefficients as well as the corresponding acti-
species (C*, CHx*, or COxHx*) are assumed to be present on the surface. vation energies and heats of adsorptions, which can be described by the
modified Arrhenius’ and van’t Hoff’s equations with a reference tem-
3.3.2. Reverse water gas shift reaction (RWGS) model perature of Tref = 613.15 K (340 °C), respectively:
The rate equation for the reverse water gas shift also includes the Tref ⎫
water gas shift reaction, due to the small equilibrium constant kj = exp {θk, j}·exp ⎧θE , j ⎛1 −
⎜
⎞⎟

⎨
⎩ ⎝ T ⎠⎬ ⎭ (25)
(Keq,RWGS = 0.03 at 320 °C and Keq,RWGS = 0.1 at 420 °C) in the tem-
perature range of interests. In this work, it is assumed that the RWGS
Tref ⎫
proceeds via R4 as the rate-determining step. Ki = exp {θK , i}·exp ⎧θH , i ⎛1 −
⎜
⎞ ⎟

⎨
⎩ ⎝ T ⎠⎬ ⎭ (26)
CO2* + H* ↔ CO* + OH* RDS CO formation via OH R4
The parameters were chosen to be θk, j = ln (kj, Tref ) and
OH* + H* ↔ H2O* + * H2O formation R5 θK , i = ln (Ki, Tref ) to assure non-negative values of the pre-exponential
factors. Activation energies and heat of adsorptions are estimated by
H2O* ↔ H2O + * H2O desorption R6 EA, j
introducing a dimensionless energy parameter θE , j = R·T and
ref
The rate equation for the reverse water gas shift reaction (formation ΔHi
θH , i = , respectively.
of CO) can be written as: R·Tref
For the determination of the kinetic parameters, data points at all
rCO = kRWGS ·K CO2·pCO2 ·(K H2·p H2 )0.5·Θ2v (17) temperature and pressure levels have been evaluated simultaneously by
Bayesian estimation within the software package Athena Visual Studio®
Since CO2 methanation and reverse water gas shift are assumed to
v 14.2, developed by Stewart and Caracotsios [68]. Briefly, in the
take place on the same sites, the denominator of the rate equations has
Bayesian theory, the probability of model Mk is calculated, and this
to be the same [27]. Thereby, the corresponding generalized rate
probability describes the observed data Y within the error space Σ [69].
equation for the RWGS with an adsorbed OH* species can be for-
Stewart et al. showed that the method of least square is not suitable for
mulated as:
multiresponse data analysis unless the responses have known relative
pCO ·p H O precisions and independent normal distribution of their errors [47,68].
k2·pCO2 ·K H0.52 ·pH0.52 ·⎛1 − ⎛kβ· p ·p ·K 2 ⎞⎞
The term response refers to the number of measured and simulated gas
⎝ H CO2 eqRWGS ⎠
rRWGS = ⎝ 2 ⎠
p H2 O
2 compositions per spatial position (i.e., 6 in the current case: CO2, CH4,
⎛1 + K C ·p e ·KHf ·p g ·p h + K H2·p H2 + K OH · ⎞ CO, H2O, H2, and Ar).
CO2 H2 H2 O p H2
⎝ ⎠ (18)
For the general case, the postulated models are related to n ex-
where the rate constant is k2 = kCO ∙K CO2 , and K eqRWGS is the equilibrium perimental data points data by:
constant for the RWGS. The adsorption constants of the involved spe-
Mk : ri = fk (θk , uk ) + εi (φ) (27)
cies depend on the catalyst and experimental conditions. Due to the
described approach, the integration of kβ in the RWGS, Eq. (18), con- Where ri is the rate of reaction for the i-th experimental condition, fk
siders the theoretical gas phase equilibrium compared to the equili- (θk,ui) is the k-th model with the parameters θk and experimental
brium on the catalyst surface. Preliminary results (not shown) demon- condition ui. The experimental error is described by the error model εi
strated kβ was necessary to represent measured gas species composition, (φ). In the present case, the error space is determined by replicate
especially CO, adequately. The same approach has been used in [2]. experiments and thus not assumed to be normally distributed as com-
Eqs. (16) and (18) consider equilibrium for the CO2 methanation and monly assumed in many studies. For the multiresponse parameter es-
reverse water gas shift, respectively. timation, the GREGPLUS solver within the Athena Visual Studio soft-
The first two models assume that CO2 adsorption is the rate-limiting ware [68] minimizes the objective function S(θk) (eq. (28)) and
step over 1 or 2 active sites, which is valid for all three mechanisms. calculates the maximum posterior probability density of the parameters
Models #3 and #4 consider the CO2* or CO* dissociation as rate-lim- θk and the posterior distribution of the proposed model k [69].
iting, respectively. Models #5 to #7 consider the hydrogenation of C*,
S (θk ) = (n + m + 1)·ln |ν (θk )| (28)
CH*, or CH2* are rate-limiting for mechanism A, respectively. Models
#8 to #14 refer to the hybrid mechanism AB in which the formation of In most kinetic studies, the objective function is equal to the sum of
COH* (#8), COH2* (#9) or the dissociation of COH* (#10, 11), COH2* the squares of the residuals, whereas in this study the objective function
(#12) or the stepwise hydrogenation of CH* (#13) and CH2* (#14) are takes into account the number of responses m, the number of events per
rate-limiting. Models #15 to #20 are for mechanism B and assume that response n, and the determinant of the covariance matrix of the re-
either the dissociation of HCOO* (#15, 16) or COH* (#17,18) or the sponses |ν(θk)|. Each element in the covariance matrix ν(θk) is defined
hydrogenation of CH* (#19) and CH2* (#20) is rate-limiting. as:
The reaction rates for the species H2, CO, CH4, CO2, H2O, and Ar n
were calculated using the following relations: νyj (θk ) = ∑ [Yyu − fyu (εu, θk )]·[Yju − f ju (εu, θk )]
R H2 = −4·r1 + r2 y−1 (29)
(19)
where Yyu is the experimental observation and fju (ξu,θk) is the pre-
RCO = r2 (20)
dicted value for response y and event u. Each element is the sum of the
RCO2 = −r1 − r2 product of the deviation of observed y and predicted j responses. The
(21)
size of the square νij(θk) matrix is determined by the number of re-
RCH4 = r1 (22) sponses. The model discrimination is based on the posterior probability;
once it is known, the preference for the different models is quantified.
R H2 O = 2·r1 + r2 (23) In multiresponse parameter estimation, the model with the smallest
sum of the squares of the residuals may not be the best model [69]. The
RAr = 0 (24) model discrimination is based on the posterior probability share of

7
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

model Mk: to 1.6 mol% and then decreased again towards the end of the plate. As
seen in the base case experiment, Fig. S3, the increase of CO was due to
p (Mk |D , ε )
π (Mk |D , ε ) = the reverse water gas shift reaction accompanied by H2O formation.
∑k p (Mk |D , ε ) (30) Whereas the decrease of CO from millimeter 15 onwards was most
where p(Mk|D,ε) is the posterior probability, a detailed description can likely due to the direct CO methanation or a combination of water gas
be found in [46,70]. shift and CO2 methanation. The reverse water gas shift reaction is en-
dothermic (41 kJ mol−1), while the CO and CO2 methanation are
highly exothermic with –206 and –165 kJ mol−1, respectively. Thus,
4. Results and discussions while CO was formed, the temperature of the catalyst surface did not
change much. However, once CO and CO2 were converted to CH4, a
4.1. Spatially resolved kinetic data slight temperature increase of 2 K was observed at around 18–26 mm.
With decreasing CO and CO2 concentration, the temperature difference
4.1.1. Catalyst coating stability and data reproducibility declined as well due to lower reaction rates. This kind of behavior
The results of the long-term experiment (130 h) coupled with IR cannot be observed in standard end-of-pipe measurements with packed
thermography analysis was conducted are depicted in Fig. 1. At 340 °C, bed reactors. Therefore, the use of the catalytic plate reactor with
the catalyst did not exhibit any sign of deactivation. Within the first spatial profiling is of great advantage.
75 h, a steady CO2 total conversion of 33% and CH4 selectivity of 92% Fig. 3 depicts the combined results of the effect of temperature and
were achieved, as illustrated in Fig. 1A. At 380 °C, the CO2 conversion total mass flow rate at 1.2 barabs on the CO2 conversion, and CH4 se-
and CH4 selectivity increased to 53% and 94%, respectively, and were lectivity (experiments #1–3, Table 2) determined at the end of the
steady until the end of the run (130 h). Fig. 1B depicts an IR thermo- coated plate. As expected, the CO2 conversion increased with higher
graphy image of the catalyst coating at t = 130 h on stream at 380 °C reaction temperature and a lower total flow rate. The CH4 selectivity,
(IR thermography results for 340 °C are shown in Fig. S4). At 380 °C, however, showed a slightly different behavior. For experiment #2 (i.e.,
the 3-D temperature plot shows nicely the homogeneous and isothermal 50 mLN min−1) with the lowest gas hourly space velocity, the CH4
catalyst surface temperature distribution, as well as a homogeneous selectivity increased with increasing temperature from 95.0 to 98.2%.
catalyst mass distribution. At the side of the plate, the surface tem- With the increasing total mass flow rate, the selectivity decreased.
perature was a bit lower (up to 8 °C) than in the center due to con- Moreover, for experiments #1 and #3 (i.e., 100 and 150 mLN min−1),
ductive heat losses at the reactor wall. However, the surface tempera-
ture around the centerline ( ± 10 mm of sampling capillary) was
always within 2 °C or better.
The difference between the temperature profiles obtained at
t = 75 h and t = 130 h, at 380 °C, was negligible with
around ± 0.2 °C, Fig. 1C, indicating a stable catalyst coating perfor-
mance. CO2 conversion and CH4 selectivity calculations, in combination
with IR thermography analysis, revealed that the catalyst was still ac-
tive at 130 h time on stream, thus confirming the results obtained in our
previous work with fixed bed reactor experiments [48].
Fig. S3 depicts the results of the base case experiment. The error
bars confirm the low variability of the catalyst surface temperature, and
gas composition profiles result for the first base case experiment
(t = 0 h), after 2 days of the experiment (t = 48 h). and after yet
another 3 days of experiments (ttotal = 120 h). In total, 25 runs (#1–5,
Table 2) at 1.2 barabs with temperatures of 320–420 °C and total flow
rates from 50 to 150 mLN min−1 were carried out in between the first
and the last base case experiment. By comparing the concentration and
temperature profiles of the base cases, it was shown the spatially-re-
solved experiment was reproducible, and that the catalyst deactivation
was negligible for experimental conditions applied. The atomic bal-
ances along the reactor axis of C, O, and H for the base case experiments
were within 100 ± 2% (results not shown). For all experiments con-
ducted, the C and O balances were always with 100 ± 5% range,
whereas in some cases (especially at high pressure and temperatures),
the H balances were a bit larger, but always within 100 ± 7%.

4.1.2. Effect of temperature

At lower temperatures, the rates of disappearance of CO2 and H2
and the rates of formation of H2O and CH4 decreased compared to the
reference case at 340 °C, see Fig. 2A vs. Fig. S3. At 320 °C, the total CO2
conversion and CH4 selectivity were XCO2 = 22.2% and SCH4 = 93.5%,
respectively. At higher temperatures (400 °C), the rates of dis-
appearance of CO2 and H2, and the rates of formation of H2O and CH4
increased, Fig. 2B. The CO2 conversion was XCO2 = 65.4%, and CH4 Fig. 1. Long-term results for experiment #1: H2/CO2 = 4 with 100 mLN min−1
selectivity was SCH4 = 96.2%. The temperature profiles for the ex- at 1.2 barabs. (A) CO2 conversion and CH4 selectivity as a function of time
periment conducted at 320 °C exhibited only a very small temperature stream for plate 2. (B) 3D surface temperatures obtained during CO2 metha-
increase of ~ 0.5 °C in the first 35 mm of plate, which corresponds nation at 130 h with and T = 380 °C and 1.2 barabs; and (C) Temperature
nicely with the CO2 conversion. Interestingly, at higher temperature, difference between t = 75 h and t = 130 h measured along the plate in the
the CO concentration increased in the first 10 to 15 mm of the coating direction of the dashed lines for 380 °C.

8
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Fig. 2. Measured axial gas composition profiles and

catalyst surface temperature difference during CO2
methanation for exp. #1 with H2/CO2 = 4, 100 mLN
min−1 and 1.2 barabs over plate 2 (A) 320 °C, and
(C) 400 °C. The dashed lines indicate the beginning
and the end of the coated area. Top: temperature
difference, bottom: concentration of (red) H2,
(green) Ar, (orange) CH4, (black) CO2, (blue) H2O,
and (dark green) CO. (For interpretation of the re-
ferences to colour in this figure legend, the reader is
referred to the web version of this article.)

the end of the plates were close to equilibrium CO2 conversion, espe-
cially at high temperatures. At 1.2 barabs and 420 °C the equilibrium
CO2 conversion would be XCO2 = 80.6%, while experimentally
XCO2 = 73.4% was achieved.

4.1.3. Effect of pressure

The influence of pressure on the CO2 methanation was investigated
by increasing the total pressure up to 7.3 barabs. As an example, Fig. 4 in
its upper part of the diagram illustrates the temperature difference
(Twith CO2 - Twithout CO2) along the catalyst plate, and in the lower part,
the gas composition profiles for the reference experiment (#1 in
Table 2) conducted at 1.2 and 5.4 barabs at 380 °C. As expected, at
5.4 barabs, the concentrations of H2 and CO2 declined faster than for the
low-pressure experiment. In detail, H2 decreased from 64 mol% at the
inlet, to 15.0 mol% for 5.4 barabs and to 35.9 mol% for 1.2 barabs at the
outlet; while CO2 decreased from 16 mol% to 4.7 mol% and 8.5 mol%,
respectively. The product gas concentrations of CH4 and H2O increased
steadily along the plate axis and reached 17.7 and 35.2 mol% at 5.4
barabs, and 10.2 and 20.9 mol% at 1.2 barabs, respectively. The CO
concentration profiles for both experiments exhibited similar behavior,
as observed and discussed in the last section. Briefly, CO increased at
the beginning, reached a maximum of 1.2 and 0.6 mol%, at 1.2 and
5.4 barabs respectively, at 10 to 15 mm of the coating, and then de-
creased towards the end of the plate to 0.68 and 0.02 mol% at 1.2 and
5.4 barabs respectively. At higher pressures, the CO formation was in-
hibited due to the fast CO methanation. The final gas compositions at
the end of the plates for both experiments were far from the chemical
equilibrium. For example, at 5.4 barabs and 380 °C the equilibrium
concentrations would be Ar = 28.4 mol%; H2 = 6.7 mol%;
CO2 = 1.66 mol%; CH4 = 21.1 mol%, H2O = 42.1 mol% and
0.018 mol% of CO. Based on the results, only CO was close to the
equilibrium.
Fig. 3. (A) CO2 conversion, solid orange line denotes chemical equilibrium, and
(B) CH4 selectivity for experiments #1, #2, and #3 (100, 50, and 150 mLN
min−1) as a function of reaction temperature at 1.2 barabs over plate 2. Dashed 4.1.4. Effect of CH4 and H2O addition
line for guidance only. (For interpretation of the references to colour in this Reaction products such as CH4 and H2O can inhibit the CO2 me-
figure legend, the reader is referred to the web version of this article.) thanation as they may occupy active catalytic sites and have possible
slow desorption rates. Thus, experiments with co-feeding of these spe-
cies have been conducted to investigate possible inhibition effects. Fig.
the smallest CH4 selectivity was determined at 340 and 360 °C, whereas
S5 and S6 illustrates the evolution of the observed rates for CO2 con-
at 320 °C and 380 to 420 °C the selectivities were higher. The reason for
version, CH4, and H2O formation, as well as the rate of CO mol change
that was the faster reverse water gas shift reaction at higher tempera-
over the catalyst coated area for experiments #1, #6, #7 and #8
tures forming CO (decreasing CH4 selectivity). With a further tem-
(Table 2) at 1.2 barabs and 360 °C. It is clearly shown that CH4 did not
perature increase, the subsequent CO methanation becomes dominant,
inhibit the CO2 methanation reaction; while H2O had a slight negative
which again increased the CH4 selectivity. The final gas compositions at
impact.

9
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Fig. 4. Effect of total pressure on the CO2 metha-

nation at (A) 1.2 barabs and (B) 5.4 barabs at 380 °C
(exp. #1). Top: temperature difference, bottom:
concentration of (red) H2, (green) Ar, (orange) CH4,
(black) CO2, (blue) H2O, and (dark green) CO. (For
interpretation of the references to colour in this
figure legend, the reader is referred to the web
version of this article.)

4.1.5. Biogas methanation are summarized in Table 7. The results indicate that models 11 and 15
CO2 from biogas upgrading facilities is an excellent source of carbon gave the best fit to the experimental data (i.e., valid for 320–420 °C and
oxides that can be used in the P2G process. In current processes, the 1.2–7.3 barabs), with the highest posterior probability share of 61.8%
biogas (i.e., composition of 60 vol% CH4 and 40 vol% CO2) is upgraded and 37.4% respectively. All other models had a share of less than 0.5%.
via CO2 separation, whereas the CH4 is injected into the gas grid. It is of On physical grounds, model 11 follows the novel hybrid mechanism AB
great interest to convert the biogas directly to renewable natural gas, in which the first C–O bond cleavage of the adsorbed CO2* to CO* is
which would allow avoiding the CO2 separation step. Therefore, ex- rather fast, while the second C–O bond cleavage, which is hydrogen
periment #11 at 1.2 barabs with H2: CO2: CH4 molar flow rates of 49 assisted (i.e., generic form of COH*), is assumed to be the slowest step
: 12 : 18 (mLN min−1) was carried out. Fig. 5A and B illustrate the (AB7). Model 15, on the other hand, assumes the hydrogen assisted C–O
corresponding gas species concentration and catalyst temperature bond cleavage from the adsorbed HCOO* to be rate-limiting (B4). The
profiles for the direct biogas methanation at 360 and 380 °C, respec- model discrimination indicates that the direct CO2 dissociation
tively. The total CO2 conversion increased with temperature from pathway (mechanism A) was not favored on our Ni/Al2O3 catalysts. It is
XCO2 = 45.4% to 55.4%. Whereas the CH4 selectivity was about the very likely that hydrogen indeed is needed to lower C–O bond dis-
same, SCH4 98.0% and SCH4 = 98.1%. These are very promising results, sociation energy, which has been suggested by other researchers as well
indicating that direct biogas methanation is technically attainable. [8,71]. The GREGPlus solver did not find a solution for models 3, 12,
Therefore, the subsequent kinetic parameter estimation should include and 19; therefore, the model discrimination values are not presented in
these conditions. Table 7. Interestingly, model 17 was favored based on the values for the
sum of the squares of residuals. However, their objective function was
4.2. Modeling larger than for models 11 and 15. Koschany et al. found that model 8
with the formation of COH* (see step AB4 in Table 5) as rate de-
4.2.1. Parameter estimation and model discrimination. termining step for their Ni-alumina catalyst [5]. In our case, model 8
A total of 21,000 kinetic data points (gas compositions) were ob- had a zero percent probability share and in addition the RSS value was
tained and used in the kinetic study, comprising of parameter estima- much larger than for other models.
tion and model discrimination based on the Bayesian approach [47,70]. The posterior probability share of model 17 was only 0.47%. The
The outcome of the model discrimination along with the objective rate determining step of model 17 is equal to model 11 with one main
function, the sum of the square of residuals, posterior distribution, difference. The former follows mechanism B in which both C–O bond
posterior probability share, and the number of parameters estimated cleavages are hydrogen assisted, while for model 11, only the second

Fig. 5. Measured axial gas composition profiles and

catalyst surface temperature difference during direct
biogas methanation for exp. #10 at (A) 360 °C, and
(C) 380 °C at 1.2 barabs. The dashed lines indicate
the beginning and the end of the coated area. Top:
temperature difference, bottom: concentration of
(red) H2, (green) Ar, (orange) CH4, (black) CO2,
(blue) H2O, and (dark green) CO. (For interpretation
of the references to colour in this figure legend, the
reader is referred to the web version of this article.)

10
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Table 7
Summary of model discrimination results.
Model # Assumed carbon Parameters estimated/ Posterior probability, Objective function, Sum of square of Posterior probability share,
species KCx* maximum log10 p (Mk |D, ε ) S (θk ) residuals, RSS π (Mk |D, ε )

1 – 8/10 −667.5 58,275 26,026 0.08%

2 – 10/10 −674.0 58,817 25,175 0.00%
3 KCO2 – – – – –
4 KCO 8/12 −938.8 82,028 30,634 0.00%
5 KC 8/12 −693.5 60,536 33,403 0.00%
6 KCH 8/12 −735.6 64,228 27,842 0.00%
7 KCH2 11/12 −666.2 58,125 23,974 0.30%
8 KCO 8/12 −730.3 63,766 25,833 0.00%
9 KCO 10/12 −672.6 58,694 22,991 0.00%
10 KCOH 11/12 −671.1 58,546 23,182 0.00%
11 KCOH 12/12 −660.9 57,776 21,928 61.76%
12 KCOH2 – – – – –
13 KCH 10/12 −675.3 58,934 25,817 0.00%
14 KCH2 10/12 −674.3 58,846 22,289 0.00%
15 KHCOO 12/12 −661.4 57,689 22,125 37.38%
16 KHCOO 8/12 −684.0 59,715 40,566 0.00%
17 KCOH 12/12 −665.8 58,071 21,722 0.47%
18 KCOH 12/12 −671.5 58,575 25,160 0.00%
19 KCH – – – – –
20 KCH2 10/12 −691.2 72,798 60,538 0.00%

Bold values indicate the best two models based on the objective function and probability share.
* Assumed surfaces species in the rate-limiting step

C–O bond dissociation is hydrogen assisted. The rate equations for the correlation) between parameter i and j. The HDP interval defines the
CO2 methanation and reverse water gas shift (RWGS) for model 11 region in which 95% of the potential values of the parameters are found
(RDS hybrid step AB7: COH* + H* ↔ CH* + OH*) and the estimated [69].
parameters are shown in Eqs. (31) and (32) and Table 8. The CO2 methanation and reverse water gas shift (RWGS) rate
equations for model 15 (RDS step B4: HCOO* + * ↔ CO* + OH*) are
2
p H O ·pCH4
0.5
k1·K COH ·pCO ·K H2·p H2 ·⎜⎛1 − ⎛ 4 2
⎜
⎞ ⎟⎞
⎟
shown in Eqs. (33) and (34), while the corresponding HPD interval and
2 p H ·pCO2 ·K eqM
⎝ ⎝ 2 ⎠⎠ covariance matrix are given in Table 9.
rCH4 = 2
p H2 O
⎛1 + K COH ·p 0.5 ·K H0.5·p 0.5 + K H2·p H2 + K OH · ⎞ 2
p H O ·pCH4
k1·KHCOO·pCO2 ·K H0.52 ·pH0.52 ⎜⎛1 − ⎛ 4 2 ⎞ ⎟⎞
CO2 2 H2 p H2
⎝ ⎠ (31) ⎜ ⎟
p H ·pCO2 ·K eqM
⎝ ⎝ 2 ⎠⎠
rCH4 = 2
pCO ·p H O p H2 O
k2·pCO2 ·K H0.52 ·pH0.52 ·⎛1 − ⎛kβ· p ·p ·K 2 ⎞⎞ ⎛1 + KHCOO·pCO · KH0.5·p 0.5 + K H2·p H2 + K OH · ⎞
2 H2 p H2
⎝ ⎝ H2 CO2 eqRWGS ⎠
⎠ ⎝ ⎠ (33)
rRWGS = 2
p H2 O
⎛1 + K COH ·p 0.5 ·K H0.5·p 0.5 + K H2·p H2 + K OH · ⎞
CO2 2 H2 p H2 pCO ·p H O
⎝ ⎠ (32) k2·pCO2 ·K H0.52 ·pH0.52 ·⎛1 − ⎛kβ· p ·p ·K2 ⎞⎞
⎝ H2 CO2 eqRWGS ⎠
rRWGS = ⎝ ⎠
In addition to the best-fit value, the highest posterior density (HDP) p H2 O
2
⎛1 + KHCOO·pCO · KH0.5·p 0.5 + K H2·p H2 + K OH · ⎞
interval, and the normalized covariance matrix are included in Table 8. 2 H2 p H2 (34)
⎝ ⎠
The latter has not been published in multi-parameter estimation for the
CO2 methanation and many other reactions, but it is an important Based on the posterior probability share model, 11 is favored over
criterion to evaluate the quality of the parameters and the model [69]. model 15. In addition, the HPD interval for model 15 was rather large
The elements in the normalized covariance matrix Eij are within [−1, for the activation energy (θ EA,1) of the CO2 methanation and the heat of
1], where a value of + 1 (–1) indicates a strong correlation (anti- hydrogen adsorption (θΔH , OH ) . Moreover, the estimated parameters in

Table 8
Estimated kinetic parameters, HPD interval, and normalized covariance matrix for model 11.
Model 11 95% HPD interval Normalized covariance matrix

θk,1 −0.22 ± 0.06 1.00

θ EA,1 15.64 ± 0.91 −0.75 1.00
θk,2 –1.64 ± 0.06 −0.50 0.42 1.00
θ EA,2 38.19 ± 1.35 0.38 −0.18 −0.63 1.00
θK , COH 1.02 ± 0.09 −0.90 0.73 0.63 −0.42 1.00
θΔH , COH 11.86 ± 1.55 0.75 −0.74 −0.52 0.63 −0.81 1.00
θK , H –1.43 ± 0.02 0.00 0.00 0.00 0.00 0.00 0.00 1.00
θΔH , H –2.45 ± 0.39 0.00 0.00 0.00 0.00 0.00 0.00 −0.82 1.00
θK , OH 0.55 ± 0.09 0.63 −0.24 0.01 0.28 −0.34 0.36 0.00 0.00 1.00
θΔH , OH 12.62 ± 0.96 −0.57 0.60 −0.03 0.19 0.33 −0.22 0.00 0.00 −0.66 1.00
θkβ 1.51 ± 0.03 0.13 −0.06 0.12 −0.04 −0.05 0.00 0.00 0.00 0.30 −0.33 1.00
θΔH , β −4.19 ± 0.42 −0.09 0.02 −0.09 −0.08 0.02 0.03 0.00 0.00 −0.30 0.34 −0.86 1.00
RSS 21,928
S (θk ) 57,776

11
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Table 9
Estimated kinetic parameters, HPD interval, and normalized covariance matrix for model 15.
Model 15 95% HPD interval Normalized covariance matrix

θk,1 0.60 ± 0.05 1.00

θ EA,1 4.48 ± 4.54 −0.28 1.00
θk,2 –2.23 ± 0.07 0.66 −0.39 1.00
θ EA,2 35.13 ± 1.45 −0.46 0.54 −0.76 1.00
θK , HCOO −0.36 ± 0.30 −0.99 0.22 −0.61 0.42 1.00
θΔH , HCOO 18.90 ± 3.85 0.16 −0.97 0.27 −0.39 −0.11 1.00
θK , H 1.18 ± 0.71 0.96 −0.34 0.73 −0.55 −0.93 0.20 1.00
θΔH , H –3.71 ± 13.13 0.02 0.93 −0.25 0.44 −0.09 −0.93 −0.10 1.00
θK , OH 0.98 ± 0.24 0.89 −0.55 0.74 −0.52 −0.84 0.44 0.89 −0.31 1.00
θΔH , OH 10.11 ± 4.57 −0.11 0.97 −0.34 0.52 0.03 −0.94 −0.19 0.97 −0.44 1.00
θkβ 1.49 ± 0.03 −0.04 0.04 0.06 0.02 0.05 −0.04 −0.06 0.03 0.04 −0.03 1.00
θΔH , β −4.00 ± 0.41 0.04 −0.09 −0.01 −0.16 −0.05 0.09 0.07 −0.08 −0.02 −0.02 −0.89 1.00
RSS 22,125
S (θk ) 57,689

model 15 have a stronger correlation and anti-correlation between each value for the activation energy for the reverse water gas shift reaction
other than for model 11. For example, there is a strong anticorrelation was similar to model 11. Fig. 6 shows the parity plots for reactants CO2
between k1·KHCOO in the numerator of eq. (33). That means that these and H2, main products CH4 and H2O, and undesired CO gas species to
two parameters could be lumped together, as most studies have done compare the accuracy of predictions over the measured data set (~3520
[5,6,15,21,22]. The results for the lumped model 15L are summarized data points per gas species). For illustration purposes, only results for
in Table S3 and show a much tighter HPD interval, a lower correlation model 11 are shown. Parity plots for model 15 are very similar, due to a
between the parameters, and that only the value for activation energy comparable objective function S (θk ) , therefore the results for the model
was changed. The objective function and posterior probability of model with the best posterior probability are shown. In general, the measured
15L and 15 were the same, which confirms that lumping k1·KHCOO in the concentrations are in good agreement with the predicted data for model
numerator was valid. Table 10 summarizes and compares the kinetic 11 for over 21,000 kinetic data points (excluding thermal data values).
parameters with the corresponding units and HPD interval for models Values for CO2 and H2 locate outside the ε = 15% region correspond to
11, 15, and 15L. experiments performed at high pressure, 5.4 barabs, and 4 barabs at
The pre-exponential factors are based on the reference temperature 653 K (380 °C). The larger deviation at higher pressures might be re-
(Tref = 613.15 K = 340 °C). The activation energy of the CO2 metha- lated to the influence of the CO methanation, which was neglected in
nation and reverse water gas shift reaction for the hybrid model 11 are the modeling but can occur following a preceding reverse water gas
79.7 kJ mol−1 and 194.7 kJ mol−1, respectively. The value for the CO2 shift. Predicted values for CH4 concentrations are in good agreement
methanation is similar to values reported in the literature with the model; it is worth mentioning that values corresponding to
70–106 kJ mol−1 [4–6,15,21,22], whereas the activation energy for the CH4 addition (biogas) to the inlet gas concentration (measured and
RWGS differs significantly. Champon et al. estimated a value of calculated 20–30 mol% region) are located within a ε = 10% region.
97 kJ mol−1 over a commercial Ni/Al2O3 catalyst at atmospheric Most of the measured H2O values are within ε = 15% region; however,
pressure [18]. some data are within ε = 25%; thus, a small deviation in some of the
In the case of model 15, the activation energy of the methanation experiments is expected, see Fig. 6D. The data outside ε = 15% are
and reverse water gas shift reaction are 22.8 kJ mol−1 and randomly distributed over the measured value matrix; thus, it is diffi-
178.9 kJ mol−1, respectively. The value for the methanation of this cult to allocated out-of-region data to a series of specific experimental
model is difficult to compare with the ones reported in the literature runs. Undesired CO predicted gas concentrations also show a good
due to the wide HPD interval and the strong anticorrelation with KHCOO . agreement with the measured data, and only a few random kinetic data
The activation energy of model 15L (lumped kinetic parameter, points lay outside the ε = 15% region.
k1·KHCOO ) had a very tight HPD interval and yielded a value of
119 kJ mol−1 for the CO2 methanation, which is within the usually 4.2.2. Kinetic model validation
reported interval for the activation energy of 70–120 kJ mol−1. The In terms of the gas composition profile, the measured and calculated

Table 10
Kinetic parameters and HPD interval estimated for models 11 and 15.
Parameter Units Model 11 Model 15 Model 15L

–1 −1 −0.5
k CO2 meth (Tref) mol kgcat s bar 0.80 ± 0.01 1.82 ± 0.64 1.27 ± 0.08
EA,CO2 meth kJ mol−1 79.7 ± 4.7 22.9 ± 23.2 119.2 ± 6.1
k RWGS (Tref) mol kgcat–1 s−1 bar−1 0.19 ± 0.02 0.11 ± 0.01 0.11 ± 0.01
EA,RWGS kJ mol−1 194.7 ± 7.0 179.1 ± 7.4 179.1 ± 7.4
KCOH(Tref) bar−0.5 2.77 ± 0.25 – –
ΔΗCΟΗ kJ mol−1 60.5 ± 7.9 – –
KHCOO(Tref) bar−0.5 – 0.70 ± 0.21 0.70 ± 0.21
ΔΗΗCOO kJ mol−1 – 96.3 ± 19.6 96.42 ± 19.6
KH(Tref) bar 0.24 ± 0.01 3.25 ± 2.30 3.24 ± 2.30
ΔΗΗ kJ mol−1 –12.5 ± 2.0 –18.9 ± 66.9 –19.20 ± 66.7
KOH(Tref) bar−1 1.73 ± 0.16 2.65 ± 0.64 2.65 ± 0.64
ΔΗΟΗ kJ mol−1 64.4 ± 4.9 51.5 ± 23.3 51.5 ± 23.2
Kβ(Tref) – 4.52 ± 0.14 4.44 ± 0.13 4.44 ± 0.13
ΔΗβ kJ mol−1 –21.4 ± 2.1 –20.4 ± 2.1 −20.4 ± 2.1

12
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Fig. 6. Parity plot for model 11 for calculated and measured concentrations of (A) CO2, (B) H2, (C) CH4, (D) H2O, and (E) CO.

profiles for various experimental conditions are in good agreement, methanation [2], and lately for CO2 methanation [5] over supported Ni
Fig. 7. catalyst. However, none of these species were experimentally validated.
The model could correctly predict the behavior at low pressure To further discriminate between the proposed models, we checked
(1.2 bar) at both 320 °C and 420 °C (Fig. 7A and B), as well as for for the plausibility of the hypothesized reaction mechanisms by iden-
simulated biogas (Fig. 7C). At higher pressure (e.g., 7.3 bar), the model tifying the possible surface species via DRIFTS experiments. It is im-
was in good agreement with the experimental data, with some devia- portant to note that the DRIFTS experiments are not designed to pro-
tion, especially towards the end of the plate and for CO. Model 11 can pose a definite reaction mechanism but to complement the modeling
correctly predict the decrease of the CO2 conversion and CH4 formation part only.
rates as well as the rate of CO mol change along the reactor axis at 360 A common issue when evaluating DRIFTS data is that highly re-
and 380 °C at 5.4 barabs (Fig. 8A-F). The addition of the RWGS reaction active intermediates often remain undetected and are, thus, falsely
in the model and the use of the unique spatially-resolved kinetic data excluded when a reaction mechanism is proposed. In order to minimize
set were paramount for representing the mole change of CO. In the this risk, we conducted temperature-resolved DRIFTS experiments,
beginning, CO was formed via the RWGS, and then after 20 mm, CO Fig. 9A.
was converted. At 100 °C, the OH region shows two CO2 overtone signals at 3700
Similar oxygenated compounds, e.g., a COxHy complex, have been and 3730 cm−1, while the peak at 3627 cm−1 can be attributed to OH
assumed as an intermediate involved in the rate determining step in the groups on Al2O3 [72], Fig. 9A. Further, mainly hydrogen carbonates
formulation of microkinetic modeling of methane reforming [58], CO (HCO3–), with features at 1633 cm−1 and 1227 cm−1, as well as

13
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Fig. 7. Measured (symbols) and calculated (solid

line) gas species concentrations for various experi-
ments at (A) 320 °C and 1.2 barabs, (B) 420 °C and
1.2 barabs, (C) biogas composition at 380 °C and 1.2
barabs, and (D) 360 °C and 7.3 barabs. (red) H2, (or-
ange) CH4, (black) CO2, (blue) H2O, and (dark
green) CO refer to the experimental data, and solid
lines are for best fit model. (For interpretation of the
references to colour in this figure legend, the reader
is referred to the web version of this article.)

monodentate carbonates formed on the Al2O3 support (m-CO32-, 1597 cm−1, as well as a shoulder at 1390 cm−1 indicative for formates
1352 cm−1) were formed by CO2 adsorption on hydroxyl groups and (HCOO–) [74,75], occurred, whereas the peak for hydrogen carbonates
lattice oxygen, respectively, which agrees with previous DRIFTS studies became less pronounced and transformed into a small shoulder.
[73] Fig. 9C. An ill-defined peak at 1088 cm−1 in combination with a The surface coverage with carbonate species reached a maximum at
small shoulder present at 1270 cm−1 can be assigned to a carbon hy- 260 °C and decreased slightly towards 350 °C, indicating the con-
droxyl (COH) [59]. With increasing temperature, an additional peak at sumption of the species. The first signs of gaseous CH4 (3016 cm−1)

Fig. 8. Measured and calculated reaction rates for (A) CO2 conversion, (B) CH4 formation, (C) CO mol change at 360 °C and 5.4 barabs; and (D) CO2 conversion, (E)
CH4 formation, (F) CO mol change at 380 °C and 5.4 barabs. (●) refer to the experimental data, and solid lines are for model 11.

14
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Fig. 9. Temperature resolved DRIFTS from 100 to 350 °C, (A) from 1000 to 3750 cm−1, (B) zoom in from 1000 to 3100 cm−1, and (C) zoom in from 1000 to
1850 cm−1.

occur at 300 °C [74,76], Fig. 9B. Interestingly, no peaks for Ni-CO illustrates the presence of COH. Saturation of the surface with the ad-
(expected at 1800–2100 cm−1) were detected at all temperatures. sorbed species occurs fast as no increasing intensities are observed after
Time-resolved DRIFTS at 350 °C yielded similar spectra, Fig. 10A-D. 5 min. After CO2 is removed from the feed, CH4 formation stops
During CO2 methanation conditions, gaseous methane (CH4, quickly, and the surface was depleted of m-CO32- within 10 min,
3016 cm−1) [74,76], as well as surface formates (1597, 1558 , and Fig. 10C. The peak associated with formates decreased very slowly;
1392 cm−1)[74,75], and monodentate carbonates (1352 and even after 40 min, a residual peak at 1597 cm−1 was observed in
1506 cm−1) were observed as illustrated in Fig. 10A and B. Further, a Fig. 10D. Again, no adsorbed CO is detected.
small peak, which can be attributed to bidentate carbonates (b-CO32-), The DRIFTS results clearly underline the importance of the support
was present at 1304 cm−1 [77]. No clear indications for hydrogen during the reaction, as has been described by earlier studies already
carbonates were detected, while a weak feature at 1088 cm−1 [78]. The absence of any Ni-CO features indicates that the reaction

Fig. 10. Time-resolved DRIFTS from 100 to 350 °C, (A) + (C) from 2900 to 3100 cm−1, (B) zoom in from 1000 to 3100 cm−1, and (C) zoom in from 1000 to
1800 cm−1.

15
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

locus is at the metal-support interface, where H* atoms from the me- 4.2.3. Reactor model validation
tallic Ni and species adsorbed on the Al2O3 support react to methane. The reactor model assumed an average bulk gas composition per-
Here, formates have been reported to play a key role in the CO2 me- pendicular to the flow direction (along the channel height). In order to
thanation as well as in the RWGS reaction [65,66,79]. Cárdenas-Arena validate this assumption, a two-dimensional model without axial dis-
et al. [73] performed isotopic-labeled and DRIFTS experiments on a Ni/ persion was developed using model 11 with the corresponding kinetic
Al2O3 catalyst and concluded that the reaction proceeds through for- parameters to visualize the species composition profile along the height
mates, which they assume are formed upon CO2 adsorption on hydroxyl (y-axis) and the coated length (x-axis) of the channel. By doing so, the
groups accompanied by the dissociation of one oxygen. Our studies do gas composition profile in y-direction could be compared to the average
not allow to unravel the origin of the formates as temperature- and gas composition, which is used in the kinetic parameter estimation
time-resolved experiments yield no conclusive evidence. The decreasing model described in section 3.1. Details about the 2D models are re-
hydrogen carbonate coverage, while increasing formate coverage ported in the Supplementary Information. Fig. 11 depicts the contour
during the temperature-resolved measurements, indicates a pathway plot and the height concentration profiles of CO2 at four different axial
via hydrogen carbonates, as suggested by Cárdenas-Arena et al. [73], position in relation to the average CO2 compositions for two operating
while the time-resolved measurements could also hint towards a me- conditions (320 °C at 1.2 barabs and 380 °C at 5.4 barabs). The contour
chanism reported by Pan et al. via monodentate carbonates [80]. Al- plots in Fig. 11A and B show that the CO2 concentration increased from
though no further definite mechanistic conclusions can be drawn from the catalyst surface (height = 0 mm) towards the quartz glass
the DRIFTS spectra alone, the data presented here support a reaction (height = 5 mm). This difference decreased along the reactor axis and
pathway following mechanism B because it features the presence of the increased for higher temperatures (Fig. 11B and D). At 380 °C at the
proposed intermediates. The absence of any adsorbed CO on the Ni axial position of 0.5 mm and 10 mm, the catalyst surface concentration
surface in the DRIFTS spectra, even at low temperatures, indicates that of CO2 was much lower than the concentration at the glass plate (i.e.,
mechanism A is very unlikely. delta of 0.6% point). At 50 mm this difference decreased to less than
Based on the DRIFTS experiments, model 15 with the HCOO– in- 0.2% point.
termediate might be the best model, while model 11 is slightly favored Secondly, the average CO2 gas composition of the height profile was
on the basis of the model discrimination (Table 10). Model 11 suggested achieved at 1.75 to 2 mm above the catalyst at every axial position and
the formation of COH*, which was detected with DRIFTS, but showed a operating condition (Fig. 11B and D). Our sampling capillary is posi-
much smaller peak intensity than for HCOO– surface complex. Further tioned in that area as visualized in the plot as well as in our previous
transient experiments (e.g., Steady-State Isotopic Transient Kinetic work [29]. Therefore, the gas sample taken with the capillary re-
Analysis) are required to elucidate the existence of short-term surface presents the bulk average gas concentration and fulfills the pre-re-
species (active and spectators). quisite for the 1D model.
For parameter estimation, the one-dimensional model assumes that
the catalyst layer is thin enough to neglect transport resistance. In the

Fig. 11. Calculated CO2 concentration profiles for 320 °C at 1.2 barabs and 380 °C at 5.4 barabs. (A) and (C) 2D contour plot. (B) and (D) CO2 profiles over the channel
height compared to the average concentration (vertical line) at the axial position 0.5, 10, 40, and 50 mm. Note that the scale of the x-axis changes from graph to
graph.

16
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

Fig. 12. (A) Carberry number and (B) Weisz modulus for CO2 and H2 calculated at 1.2 barabs. The gray parts indicate the areas that were not used for the estimation
of the kinetic parameter.

catalytic plate reactor, possible bulk and pore diffusion as well as heat influence the work reported in this paper.
transfer limitation may occur under certain experimental conditions
(see Supplementary Information for mass and heat transfer influence Acknowledgments
calculation). Fig. 12A and B show the Carberry and Weisz modulus for
CO2 and H2 at 420 °C and 1.2 barabs, respectively, to study the influence The authors thank F. Caporuscio, L. Cousmich, A. Golsztajn, G.
of the diffusion limitation. At low temperatures (320 °C), the Carberry Lepkyj, and R Roy for technical assistance. Financial support from the
numbers were Ca = 0.004 for CO2 and Ca = 0.001 for H2 (Fig. S5A). At Natural Sciences and Engineering Research Council (NSERC RGPIN/
420 °C, the Carberry number reached maximum values of around 0.04 04685–2014) for this study is also recognized. J. Ilsemann gratefully
and 0.01 for CO2 and H2, Fig. 12A, respectively, which is much smaller acknowledges funding through the German Research Foundation (DFG)
than the threshold value of 0.1, indicating bulk diffusion influence. - graduate school 1860.
Therefore, in our case, the external effectiveness factor was close to 1
for all experimental conditions applied [81]. Pore diffusion limitation Appendix A. Supplementary data
can also be neglected as maximum values for the Weisz Ψ modulus of
0.004, and 0.001 for CO2 and H2 were determined at 420 °C, Fig. 12B. Supplementary data to this article can be found online at https://
At lower temperatures, the values were by a factor of 10 smaller (Fig. [Link]/10.1016/[Link].2020.124629.
S5B). Note, the particle size of the catalyst on the coated plates were
smaller than 45 µm with an average pore size of 20 ± 2 nm [3]. Heat References
transfer limitation can also be neglected as shown by the calculation of
the Anderson criterion [82] as illustrated in Fig S7. [1] P. Sabatier, J.B. Senderens, New methane synthesis, Acad. Sci. 314 (1902) 514–516.
[2] J. Kopyscinski, T.J. Schildhauer, F. Vogel, S.M.A. Biollaz, A. Wokaun, Applying
spatially resolved concentration and temperature measurements in a catalytic plate
5. Conclusions reactor for the kinetic study of CO methanation, J. Catal. 271 (2010) 262–279,
[Link]
[3] J.A. Hernandez Lalinde, J. Jiang, G. Jai, J. Kopyscinski, Preparation and char-
The results of nearly 110 experimental runs (21,000 high-quality acterization of Ni/Al2O3 catalyst coatings on FeCrAl-loy plates used in a catalytic
kinetic data points) were used for kinetic modeling and parameter de- channel reactor with in-situ spatial profiling to study CO2 methanation, Chem. Eng.
termination. A one-dimensional reactor model, including a generalized J. 357 (2019) 435–446, [Link]
[4] T. Van Herwijnen, Kinetics of the methanation of CO and CO2 on a nickel catalyst,
Langmuir-Hinshelwood rate expression, was developed for the kinetic
J. Catal. 28 (1973) 391–402, [Link]
parameter estimation performed following the Bayes’ theorem. The [5] F. Koschany, D. Schlereth, O. Hinrichsen, On the kinetics of the methanation of
most probable kinetic models were model 11, hybrid mechanism AB, carbon dioxide on coprecipitated NiAl(O)x, Appl. Catal. B Environ. 181 (2016)
504–516, [Link]
and model 15, mechanism B. The models RDS includes the reaction of
[6] L. Falbo, M. Martinelli, C.G. Visconti, L. Lietti, C. Bassano, P. Deiana, Kinetics of
an oxygenated complex (COH* for model 11 or HCOO* for model 15) CO2 methanation on a Ru-based catalyst at process conditions relevant for Power-
with an active site (*) or an adsorbed hydrogen (H*). Water is adsorbed to-Gas applications, Appl. Catal. B Environ. 225 (2018) 354–363, [Link]
as a hydroxyl species (OH*). The models predicted local concentrations 10.1016/[Link].2017.11.066.
[7] L. Falbo, C.G. Visconti, L. Lietti, J. Szanyi, The effect of CO on CO2 methanation
in excellent agreement with the experimental data. Temperature- and over Ru/Al2O3 catalysts: a combined steady-state reactivity and transient DRIFT
time-resolved DRIFTS measurements were used to identify the present spectroscopy study, Appl. Catal. B Environ. 256 (2019), [Link]
surface species to complement the model discrimination. The mea- apcatb.2019.117791.
[8] A. Beuls, C. Swalus, M. Jacquemin, G. Heyen, A. Karelovic, P. Ruiz, Methanation of
surements reveal the formation of formates and carbon hydroxyl spe- CO2: Further insight into the mechanism over Rh/γ-Al2O3 catalyst, Appl. Catal., B.
cies on the surface, whereas no adsorbed CO was detected, rendering 113–114 (2012) 2–10. doi:10.1016/[Link].2011.02.033.
model 15, mechanism B more likely from a spectroscopic point of view. [9] C. Swalus, M. Jacquemin, C. Poleunis, P. Bertrand, P. Ruiz, CO2 methanation on
Rh/γ-Al2O3 catalyst at low temperature: “In situ” supply of hydrogen by Ni/acti-
The discrepancy between the spectroscopic experiments and the vated carbon catalyst, Appl. Catal., B. 125 (2012) 41–50, [Link]
mathematical modeling clearly illustrates the importance and limita- [Link].2012.05.019.
tions of combining the two techniques to formulate a physically sound [10] C. Wulf, J. Linßen, P. Zapp, Review of power-to-gas projects in Europe, Energy
Procedia. 155 (2018) 367–378, [Link]
reaction mechanism. SSITKA experiments are needed to identify the
[11] M. Younas, L. Loong Kong, M.J.K. Bashir, H. Nadeem, A. Shehzad, S. Sethupathi,
occurring reaction pathway unambiguously. Recent advancements, fundamental challenges, and opportunities in catalytic me-
thanation of CO2, Energy & Fuels. 30 (2016) 8815–8831, [Link]
[Link].6b01723.
Declaration of Competing Interest [12] M.A. Bañares, M.O. Guerrero-Pérez, J.L.G. Fierro, G.G. Cortez, Raman spectroscopy
during catalytic operations with on-line activity measurement (operando spectro-
The authors declare that they have no known competing financial scopy): a method for understanding the active centres of cations supported on
porous materials, J. Mater. Chem. 12 (2002) 3337–3342, [Link]
interests or personal relationships that could have appeared to

17
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

B204494C. catalysts: Oxidation and reforming zones, transport effects, and approach to ther-
[13] M. Bailera, P. Lisbona, L.M. Romeo, S. Espatolero, Power to Gas projects review: modynamic equilibrium, J. Catal. 249 (2007) 380–393.
Lab, pilot and demo plants for storing renewable energy and CO2, Renew. Sustain. [40] D. Weber, D.J. Holland, L.F. Gladden, Spatially and chemically resolved measure-
Energy Rev. 69 (2017) 292–312, [Link] ment of intra- and inter-particle molecular diffusion in a fixed-bed reactor, Appl.
[14] M. Götz, J. Lefebvre, F. Mörs, A. McDaniel Koch, F. Graf, S. Bajohr, et al., Catal. A Gen. 392 (2011) 192–198, [Link]
Renewable power-to-gas: a technological and economic review, Renew. Energy. 85 [41] J. Touitou, K. Morgan, R. Burch, C. Hardacre, A. Goguet, An in situ spatially re-
(2016) 1371–1390, [Link] solved method to probe gas phase reactions through a fixed bed catalyst, Catal. Sci.
[15] C.V. Miguel, A. Mendes, L.M. Madeira, Intrinsic kinetics of CO2 methanation over Technol. 2 (2012) 1811, [Link]
an industrial nickel-based catalyst, J. CO2 Util. 25 (2018) 128–136, [Link] [42] J. Touitou, R. Burch, C. Hardacre, C. McManus, K. Morgan, J. Sá, et al., An in situ
org/10.1016/[Link].2018.03.011. spatially resolved analytical technique to simultaneously probe gas phase reactions
[16] J.H. Chiang, J.R. Hopper, Kinetics of the hydrogenation of carbon dioxide over and temperature within the packed bed of a plug flow reactor, Analyst. 138 (2013)
supported nickel, Ind. Eng. Chem. Prod. Res. Dev. 22 (1983) 225–228, [Link] 2858–2862, [Link]
org/10.1021/i300010a011. [43] A. Donazzi, D. Livio, A. Beretta, G. Groppi, P. Forzatti, Surface temperature profiles
[17] J. Zhang, N. Fatah, S. Capela, Y. Kara, O. Guerrini, A.Y. Khodakov, Kinetic in- in CH4 CPO over honeycomb supported Rh catalyst probed with in situ optical
vestigation of carbon monoxide hydrogenation under realistic conditions of me- pyrometer, Appl. Catal., A. 402 (2011) 41–49, [Link]
thanation of biomass derived syngas, Fuel. 111 (2013) 845–854, [Link] 2011.05.022.
10.1016/[Link].2013.04.057. [44] A. Donazzi, D. Livio, C. Diehm, A. Beretta, G. Groppi, P. Forzatti, Effect of pressure
[18] I. Champon, A. Bengaouer, A. Chaise, S. Thomas, A.-C. Roger, Carbon dioxide in the autothermal catalytic partial oxidation of CH4 and C3H8: Spatially resolved
methanation kinetic model on a commercial Ni/Al2O3 catalyst, J. CO2 Util. 34 temperature and composition profiles, Appl. Catal., A. 469 (2014) 52–64. doi:10.
(2019) 256–265, [Link] 1016/[Link].2013.09.054.
[19] Z.A. Ibraeva, N.V. Nekrasov, B.S. Gudkov, V.I. Yakerson, Z.T. Beisembaeva, [45] A. Russell, W.S. Epling, H. Hess, H.-Y. Chen, C. Henry, N. Currier, et al., Spatially-
E.Z. Golosman, et al., Kinetics of methanation of carbon dioxide on a nickel catalyst, resolved temperature and gas species changes in a lean-burn engine emissions
Theor. Exp. Chem. 26 (1991) 584–588, [Link] control catalyst, Ind. Eng. Chem. Res. 49 (2010) 10311–10322, [Link]
[20] G.D. Weatherbee, C.H. Bartholomew, Hydrogenation of CO2 on group VIII metals 1021/ie1005299.
II. Kinetics and mechanism of CO2 hydrogenation on nickel, J. Catal. 77 (1982) [46] W.E. Stewart, Y. Shon, G.E.P. Box, Discrimination and goodness of fit of multi-
460–472, [Link] response mechanistic models, AIChE J. 44 (1998) 1404–1412, [Link]
[21] T. Kai, T. Takahashi, S. Furusaki, Kinetics of the methanation of carbon dioxide over 1002/aic.690440618.
a supported Ni-La2O3 catalyst, Can. J. Chem. Eng. 66 (1988) 343–347, [Link] [47] W.E. Stewart, M. Caracotsios, J.P. Sørensen, Parameter estimation from multi-
org/10.1002/cjce.5450660226. response data, AIChE J. 38 (1992) 641–650, [Link]
[22] P. Marocco, E.A. Morosanu, E. Giglio, D. Ferrero, C. Mebrahtu, A. Lanzini, et al., 690380502.
CO2 methanation over Ni/Al hydrotalcite-derived catalyst: Experimental char- [48] A. Aljishi, G. Veilleux, J.A.H. Lalinde, J. Kopyscinski, The effect of synthesis para-
acterization and kinetic study, Fuel 225 (2018) 230–242, [Link] meters on ordered mesoporous nickel alumina catalyst for CO2 methanation, Appl.
[Link].2018.03.137. Catal. A Gen. 549 (2018) 263–272, [Link]
[23] J.N. Dew, R.R. White, C.M. Sliepcevich, Hydrogenation of Carbon Dioxide on [49] L.L. Raja, R.J. Kee, O. Deutschmann, J. Warnatz, L.D. Schmidt, A critical evaluation
Nickel-Kieselguhr Catalyst, Ind. Eng. Chem. 47 (1955) 140–146, [Link] of Navier-Stokes, boundary-layer, and plug-flow models of the flow and chemistry
10.1021/ie50541a044. in a catalytic-combustion monolith, Catal. Today. 59 (2000) 47–60, [Link]
[24] G. Garbarino, P. Riani, L. Magistri, G. Busca, A study of the methanation of carbon org/10.1016/S0920-5861(00)00271-6.
dioxide on Ni/Al2O3 catalysts at atmospheric pressure, Int. J. Hydrogen Energy. 39 [50] B.E. Poling, J.M. Prausnitz, J.P. ÓConnell, The Properties of gases and liquids, 5th
(2014) 11557–11565, [Link] ed., McGraw-Hill, New York, 2001.
[25] M. Šolc, Kinetik der hydrierung des kohlendioxyds zu methan an einem nickel- [51] G.D. Weatherbee, C.H. Bartholomew, Hydrogenation of CO2 on group VIII metals I.
chrom(III)-oxyd-katalysator, Collect. Czechoslov. Chem. Commun. 27 (1962) Specific activity of Ni/SiO2, J. Catal. 68 (1981) 67–76, [Link]
2621–2627, [Link] 0021-9517(81)90040-3.
[26] H. Inuoe, M. Funakoshi, Kinetics of methanation of carbon monoxide and carbon [52] D.W. Goodman, R.D. Kelley, T.E. Madey, J.J.T. Yates, Kinetics of the hydrogenation
dioxide, J. Chem. Eng. Japan. 17 (1984) 602–610, [Link] of CO over a single crystal nickel catalyst, J. Catal. 63 (1980) 226–234, [Link]
602. org/10.1016/0021-9517(80)90075-5.
[27] J. Xu, G.F. Froment, Methane steam reforming, methanation and water-gas shift: I, [53] S. Fujita, H. Terunuma, M. Nakamura, N. Takezawa, Mechanisms of methanation of
Intrinsic kinetics, AIChE J. 35 (1989) 88–96, [Link] carbon monoxide and carbon dioxide over nickel, Ind. Eng. Chem. Res. 30 (1991)
690350109. 1146–1151, [Link]
[28] G. Buzzi-Ferraris, Planning of experiments and kinetic analysis, Catal. Today. 52 [54] J. Falconer, Adsorption and methanation of carbon dioxide on a nickel/silica cat-
(1999) 125–132, [Link] alyst, J. Catal. 62 (1980) 280–285, [Link]
[29] J. Hernandez Lalinde, K. Kofler, X. Huang, J. Kopyscinski, Improved Kinetic Data 90456-X.
Acquisition Using An Optically Accessible Catalytic Plate Reactor with Spatially- [55] M. Marwood, R. Doepper, A. Renken, In-situ surface and gas phase analysis for
Resolved Measurement Techniques. Case of Study: CO2 Methanation, Catalysts. 8 kinetic studies under transient conditions The catalytic hydrogenation of CO2,
(2018) 86. doi: 10.3390/catal8020086. Appl. Catal. A Gen. 151 (1997) 223–246, [Link]
[30] M. Bosco, F. Vogel, Optically accessible channel reactor for the kinetic investigation 00267-0.
of hydrocarbon reforming reactions, Catal. Today. 116 (2006) 348–353. [56] S.-I. Fujita, N. Takezawa, Difference in the selectivity of CO and CO2 methanation
[31] O. Shakir, A. Yezerets, N.W. Currier, W.S. Epling, Spatially resolving concentration reactions, Chem. Eng. J. 68 (1997) 63–68, [Link]
and temperature gradients during the oxidation of propylene on Pt/Al2O3, Appl. 8947(97)00074-0.
Catal. A Gen. 365 (2009) 301–308, [Link] [57] C. Schild, A. Wokaun, R.A. Koeppel, A. Baiker, Carbon dioxide hydrogenation over
[32] M. Hettel, C. Diehm, B. Torkashvand, O. Deutschmann, Critical evaluation of in situ nickel/zirconia catalysts from amorphous precursors: on the mechanism of methane
probe techniques for catalytic honeycomb monoliths, Catal. Today. 216 (2013) formation, J. Phys. Chem. 95 (1991) 6341–6346, [Link]
2–10, [Link] j100169a049.
[33] J. Sa, D.L.A. Fernandes, F. Aiouache, A. Goguet, C. Hardacre, D. Lundie, et al., [58] L.M. Aparicio, Transient Isotopic Studies and Microkinetic Modeling of Methane
SpaciMS: spatial and temporal operando resolution of reactions within catalytic Reforming over Nickel Catalysts, J. Catal. 165 (1997) 262–274, [Link]
monoliths, Analyst. 135 (2010) 2260–2272, [Link] 1006/jcat.1997.1468.
C0AN00303D. [59] M.P. Andersson, F. Abild-Pedersen, I.N. Remediakis, T. Bligaard, G. Jones,
[34] J.-S. Choi, W.P. Partridge, C.S. Daw, Spatially resolved in situ measurements of J. Engbaek, et al., Structure sensitivity of the methanation reaction: H2-induced CO
transient species breakthrough during cyclic, low-temperature regeneration of a dissociation on nickel surfaces, J. Catal. 255 (2008) 6–19, [Link]
monolithic Pt/K/Al2O3 NOx storage-reduction catalyst, Appl. Catal., A. 293 (2005) [Link].2007.12.016.
24–40, [Link] [60] S. Fujita, M. Nakamura, T. Doi, N. Takezawa, Mechanisms of methanation of carbon
[35] J. Kopyscinski, T.J. Schildhauer, S.M.A. Biollaz, Methanation in a fluidized bed dioxide and carbon monoxide over nickel/alumina catalysts, Appl. Catal. A Gen.
reactor with high initial CO partial pressure: Part I—Experimental investigation of 104 (1993) 87–100, [Link]
hydrodynamics, mass transfer effects, and carbon deposition, Chem. Eng. Sci. 66 [61] K. Walter, O. V. Buyevskaya, D. Wolf, M. Baerns, Rhodium-catalyzed partial oxi-
(2011) 924–934, [Link] dation of methane to CO and H2. In situ DRIFTS studies on surface intermediates,
[36] J. Kopyscinski, T.J. Schildhauer, S.M.A. Biollaz, Methanation in a fluidized bed Catal. Letters. 29 (1994) 261–270. doi:10.1007/BF00814272.
reactor with high initial CO partial pressure: Part II— Modeling and sensitivity [62] F. Solymosy, Methanation of CO2 on supported rhodium catalyst, J. Catal. 68
study, Chem. Eng. Sci. 66 (2011) 1612–1621, [Link] (1981) 371–382, [Link]
12.029. [63] C. Schild, A. Wokaun, A. Baiker, On the mechanism of CO and CO2 hydrogenation
[37] J. Kopyscinski, T.J. Schildhauer, S.M.A. Biollaz, Fluidized-Bed Methanation: reactions on zirconia-supported catalysts: a diffuse reflectance FTIR study, J. Mol.
Interaction between Kinetics and Mass Transfer, Ind. Eng. Chem. Res. 50 (2011) Catal. 63 (1990) 223–242, [Link]
2781–2790, [Link] [64] Q. Pan, J. Peng, T. Sun, S. Wang, S. Wang, Insight into the reaction route of CO2
[38] R. Horn, N.J. Degenstein, K.A. Williams, L.D. Schmidt, Spatial and temporal profiles methanation: Promotion effect of medium basic sites, Catal. Commun. 45 (2014)
in millisecond partial oxidation processes, Catal. Letters. 110 (2006) 169–178, 74–78, [Link]
[Link] [65] P.A.U. Aldana, F. Ocampo, K. Kobl, B. Louis, F. Thibault-Starzyk, M. Daturi, et al.,
[39] R. Horn, K.A. Williams, N.J. Degenstein, A. Bitsch-Larsen, D. Dalle Nogare, Catalytic CO2 valorization into CH4 on Ni-based ceria-zirconia. Reaction me-
S.A. Tupy, et al., Methane catalytic partial oxidation on autothermal Rh and Pt foam chanism by operando IR spectroscopy, Catal. Today. 215 (2013) 201–207, https://

18
J.A. Hernandez Lalinde, et al. Chemical Engineering Journal 390 (2020) 124629

[Link]/10.1016/[Link].2013.02.019. methanation mechanism using Ni/CeO2 and Ni/Al2O3 catalysts, Appl. Catal. B
[66] A. Westermann, B. Azambre, M.C. Bacariza, I. Graça, M.F. Ribeiro, J.M. Lopes, Environ. 265 (2020) 118538, , [Link]
et al., Insight into CO2 methanation mechanism over NiUSY zeolites: An operando [74] X. Wang, H. Shi, J.H. Kwak, J. Szanyi, Mechanism of CO 2 Hydrogenation on Pd/Al
IR study, Appl. Catal. B Environ. 174–175 (2015) 120–125, [Link] 2 O 3 Catalysts: Kinetics and Transient DRIFTS-MS Studies, ACS Catal. 5 (2015)
1016/[Link].2015.02.026. 6337–6349, [Link]
[67] J. Kopyscinski, Production of synthetic natural gas in a fluidized bed reactor. [75] A. Erdohelyi, Catalytic hydrogenation of CO2 over supported palladium, J. Catal.
Understanding the hydrodynamic, mass transfer, and kinetic effects, Dissertation, 98 (1986) 166–177, [Link]
ETH Zurich Switzerland, Diss. No. 18800, 2010. doi:10.3929/ethz-a-006031831. [76] W.Z. Weng, M.S. Chen, Q.G. Yan, T.H. Wu, Z.S. Chao, Y.Y. Liao, et al., Mechanistic
[68] W.E. Stewart, M. Caracotsios, Computer-Aided Modeling of Reactive Systems, 1st study of partial oxidation of methane to synthesis gas over supported rhodium and
ed., Wiley \& Sons, New Jersey, 2008. ruthenium catalysts using in situ time-resolved FTIR spectroscopy, Catal. Today. 63
[69] J. Kopyscinski, J. Choi, J.M. Hill, Comprehensive kinetic study for pyridine hy- (2000) 317–326, [Link]
drodenitrogenation on (Ni)WP/SiO2 catalysts, Appl. Catal., A. 445–446 (2012) [77] M. Daturi, C. Binet, J.C. Lavalley, G. Blanchard, Surface FTIR investigations on
50–60. doi:10.1016/[Link].2012.08.027. CexZr1-xO2 system, Surf. Interface Anal. 30 (2000) 273–277.
[70] S.-H. Hsu, S.D. Stamatis, J.M. Caruthers, W.N. Delgass, V. Venkatasubramanian, [78] P. Panagiotopoulou, D.I. Kondarides, X.E. Verykios, Mechanistic aspects of the se-
G.E. Blau, et al., bayesian framework for building kinetic models of catalytic sys- lective methanation of CO over Ru/TiO2 catalyst, Catal. Today. 181 (2012)
tems, Ind. Eng. Chem. Res. 48 (2009) 4768–4790, [Link] 138–147, [Link]
ie801651y. [79] C.M. Kalamaras, G.G. Olympiou, A.M. Efstathiou, The water-gas shift reaction on
[71] M. Jacquemin, A. Beuls, P. Ruiz, Catalytic production of methane from CO2 and H2 Pt/γ-Al2O3 catalyst: Operando SSITKA-DRIFTS-mass spectroscopy studies, Catal.
at low temperature: Insight on the reaction mechanism, Catal. Today. 157 (2010) Today. 138 (2008) 228–234, [Link]
462–466, [Link] [80] Q. Pan, J. Peng, S. Wang, S. Wang, In situ FTIR spectroscopic study of the CO2
[72] S. Eckle, H.-G. Anfang, R.J. Behm, Reaction Intermediates and Side Products in the methanation mechanism on Ni/Ce0.5Zr0.5O2, Catal, Sci. Technol. 4 (2014)
Methanation of CO and CO2 over Supported Ru Catalysts in H2-Rich Reformate 502–509, [Link]
Gases, J. Phys. Chem. C. 115 (2011) 1361–1367, [Link] [81] R.A. van Santen, P.W.N.M. van Leeuwen, J.A. Moulijn, B.A. Averill, Catalysis: An
jp108106t. Integrated Approach, 2nd ed., Elsevier, Amsterdam, 1999.
[73] A. Cárdenas-Arenas, A. Quindimil, A. Davó-Quiñonero, E. Bailón-García, D. Lozano- [82] M.E. Davis, R.J. Davis, Fundamentals of Chemical Reaction Engineering, McGraw-
Castelló, U. De-La-Torre, et al., Isotopic and in situ DRIFTS study of the CO2 Hill, New York, 2003.

19