The consequences of surface heterogeneity of cobalt nanoparticles on the kinetics of CO methanation

José Castillo a, Luis E. Arteaga-Pérez bc, Alejandro Karelovic ac and Romel Jiménez *ac
aCarbon and Catalysis Laboratory (CarboCat), Department of Chemical Engineering, Universidad de Concepción, Concepcion, Chile. E-mail: romeljimenez@udec.cl
bLaboratory of Thermal and Catalytic Processes (LPTC), Department of Wood Engineering, University of Bio-Bio, Concepcion, Chile
cUniversidad de Concepción, Unidad de Desarrollo Tecnológico, UDT, Chile

Received 30th August 2019 , Accepted 7th October 2019

First published on 8th October 2019


The CO hydrogenation reaction was studied under methanation conditions (H2/CO >3, 250–300 °C) on Co/SiO2 catalysts with different mean Co nanoparticle size (dp = 4 nm, 13 nm and 33 nm). The catalysts were prepared by incipient wetness impregnation; the cobalt loading and reduction conditions were suitably selected in order to produce catalysts with different dp. Operando FTIR and kinetic measurements provided consistent data for the heat of CO adsorption at low and full coverage. The results confirm the structural sensitivity of the CO hydrogenation reaction on supported cobalt catalysts and demonstrate the energetic heterogeneity of the cobalt surface. A model derived from the Temkin formalism accurately fits the kinetic data and renders the heterogeneity of the cobalt surface by capturing the change in the heat of CO adsorption with the CO coverage. This change is normalized by the heat of CO adsorption at high coverage (ΔQCO/Q1) in order to quantify and compare the heterogeneity of cobalt surfaces. The CO hydrogenation turnover rates are lower on the catalyst with smaller nanoparticles. This surface shows higher heterogeneity and lower apparent activation energy, which is attributed to a larger amount of the more active B5-B sites in this catalyst.


1. Introduction

The increasing and worrisome climate change and the instability of fossil fuels in the international market have renewed the interest in the production of methane, this time from biomass.1,2 In fact, the conversion of biomass into synthesis gas via gasification and further conversion of this syngas into a synthetic natural gas (SNG)3,4 represents an excellent alternative for the diversification of energy resources. Thus, SNG may be generated in small and strategically distributed plants that could couple the biomass gasification with the methanation process to partially fulfill natural gas requirements. This is an attractive alternative for those energy matrices which heavily rely on external suppliers (gas importing countries) and has potential for using biomass, particularly lignocellulosic residues. Moreover, this strategy will allow taking advantage of the existing infrastructure of natural gas pipelines for injecting SNG.5,6 This context, together with the well-known use of CO methanation to remove CO from a H2-rich gas stream for ammonia and fuel cell applications, explains the increase of catalytic studies focused on COx hydrogenation in the past decades.

The interest in producing SNG has drawn renewed attention to the synthesis of new catalysts and the understanding of the catalytic phenomena. Ni-based catalysts are usually used for CO hydrogenation because of their high activity, selectivity toward CH4 and low price, despite their deactivation at high temperature due to sintering and carbon deposition.7 Cobalt-supported catalysts have demonstrated high activity for CO methanation.8–11 The structural properties of methanation catalysts (i.e. nanoclusters and nanoparticles sizes) have an effect on their activity, which means that the specific activity depends on the metal dispersion in a complex way. Efforts have been made in order to explain the previously mentioned phenomena, with the in situ and operando spectroscopy techniques being the most common and robust alternatives for gathering comprehensive data.

Specifically, the CO hydrogenation turnover frequency (TOF) on Co-based catalysts diminishes as the cobalt dispersion increases for nanoparticle sizes below 8–10 nm, while for higher particle sizes the TOF remains unaltered.12–19 However, there is still no consensus in the literature to explain this phenomenon and there are several theories behind the experimental observations. For example, some authors argue that small Co particles are easily re-oxidized, decreasing the catalyst activity;17,20,21 other theories include change in the availability of specific active sites with different nanoparticle sizes,22,23 surface reconstruction,24,25 and site blockage,16 among others. An interesting fact is that this debate on the causes of the structural sensitivity is mostly based on experimental evidence obtained under Fischer–Tropsch synthesis (FTS) conditions (H2/CO = 2, 20–30 bar, 220–260 °C),12–15,17–18 while very few studies have been focused on data obtained on cobalt nanoparticles under methanation conditions (H2/CO > 10, atmospheric pressure, T > 250 °C).16,26 Although the reduction in the activity with the cluster sizes is qualitatively the same for both conditions, the kinetically relevant reaction step for the CO hydrogenation could change. This was reported by Chen et al., who studied CO hydrogenation over Co catalysts under methanation27 and FTS28 conditions. Therefore, the global effect observed in the TOF due to the change in particle sizes could be due to different causes depending on the reaction conditions.

On the other hand, the modelling of catalyst surfaces allows correlation of theoretical physicochemical predictions with experimental data, leading to the design of more active and selective catalysts. In fact, the Langmuir–Hinshelwood model has been widely applied to represent the behavior of catalytic surfaces of the group VIII metals (i.e., cobalt (ref. 29) and noble metals (ref. 30 and 31)). This model represents a homogeneous catalytic surface by assuming that the heat of adsorption of molecules is independent of the surface coverage. Nevertheless, there are few studies on the modelling of the catalytic process considering the surface heterogeneity, as could be inferred from the existing data on Co-supported catalysts. In this sense, the Temkin model could be an option to represent the Co surfaces with different atom coordination degrees, for which heterogeneity is expected due to the combined effect of active site properties and the interaction between adsorbed species.30,32

Therefore, this work studies the effect of particle size on the kinetic behavior of cobalt surfaces during CO hydrogenation under methanation conditions and discusses the fundamental reasons for the observed structural sensitivity. Three SiO2-supported Co catalysts with significant different mean nanoparticles sizes are synthesized, characterized and kinetically evaluated. Operando FTIR measurements and the CO adsorption equilibrium are used to demonstrate the heterogeneity of cobalt surfaces. Kinetic measurements in a differential fixed-bed tubular reactor and operando spectroscopy data validate a reaction model based on the Temkin formalism, which captures the heterogeneity of cobalt surfaces and considers the hydrogen-assisted CO dissociation mechanism.

2. Materials and methods

2.1. Catalyst preparation and characterization

The Co/SiO2 catalysts were prepared using incipient wetness impregnation of Co(NO3)2·6H2O (Sigma Aldrich) onto SiO2 (Alfa Aesar). After impregnation, each sample was dried and subsequently pelletized, milled and sieved to grain sizes between 150 and 360 μm. Three catalysts with different mean particle sizes of cobalt were obtained by varying the loading of cobalt and by controlling the thermal treatment conditions (Table 1). The thermal treatment was performed for 0.5 g of catalyst under H2 flow (50 mL min−1). The resulting catalysts were characterized and named as “Co-x nm” where x refers to the mean diameter of cobalt nanoparticles calculated from the average of the mean nanoparticle sizes measured by different techniques, as shown below.
Table 1 Cobalt content and conditions for preparation of catalysts
Catalyst Co-4 nm Co-13 nm Co-33 nm
Heating rate used on reduction 1 °C min−1 1 °C min−1 5 °C min−1
Maximum temperature of reduction 500 °C 350 °C 500 °C
Time at maximum temperature 1 h 4 h 1 h
Cobalt content (nominal wt%) 4.8 15 10


X-ray diffraction (XRD) analyses were performed with D4 Endeavor (Bruker AXS) equipment using CuKα radiation. The 2θ range was scanned between 10° and 90° with 0.02° resolution. The Scherrer equation was used to calculate the size of cobalt crystallites (using the peak at 2θ = 44.3°, which is attributed to Co0).

Transmission electron microscopy analyses were performed using a JEOL JEM-1200 EX II microscope. Images were analyzed to estimate the particle size distribution and the mean cluster size of cobalt nanoparticles from image file: c9cy01753d-t1.tif. This diameter was used to calculate the fraction of cobalt atoms at the catalyst surface (Co dispersion) by assuming hemispherical particles.

The amount of surface cobalt in each catalyst was also assessed by H2 chemisorption using a homemade volumetric adsorption instrument equipped with a Dual Gauge controller (TPG 262) and an APR 260 sensor (0.1–1100 mbar, Pfeiffer Vacuum). The following procedure was used: the catalyst (already reduced) was loaded into a U-shaped quartz reactor and heated to 400 °C (heating rate 1 °C min−1) in a H2 atmosphere. Upon reaching 400 °C, the sample was evacuated, and fresh hydrogen was loaded to the reactor. This procedure was repeated three or four times to ensure the removal of surface oxygen. The APR 260 sensor was connected to a data logger in order to register the change in the H2 pressure during the reduction of the catalyst in each cycle. The reduction cycles were carried out until no change in the H2 pressure was observed and the H2 uptake was consistent with the bulk composition of cobalt in the catalyst. After evacuation and cooling to 40 °C, two H2 adsorption isotherms were collected in the 5–160 torr range. The first one takes into account the total adsorption (irreversible and reversible) and the second one (collected after evacuation) corresponds to reversible adsorption. Their subtraction allowed us to obtain the irreversibly adsorbed (chemisorbed) amount of H2. Assuming a Co/H = 1 stoichiometry, the amount of surface cobalt could be calculated.

2.2. Catalytic activity in the fixed bed reactor

Catalytic tests of each catalyst were carried out using a stainless-steel reactor with a 3/4 in. diameter. The reactor was charged with 10–40 mg of catalyst diluted with silica sand of similar grain size to avoid hot spots due to the exothermic nature of the methanation reaction. The catalyst was placed between quartz wool plugs, and a thermocouple was placed in the central part of the catalyst bed to measure and control the temperature. The Weisz–Prater and Mears criteria33 were used to rule out mass and heat transfer limitations in the catalytic bed; thus, it is assumed that the reaction took place in a fully kinetic regime.

The catalysts were in situ pretreated in pure H2 flow (50 mL min−1) using the same heating rate as for the reduction and kept at 400 °C (350 °C for the Co-13 nm) for 1 h. Then, the catalysts were cooled to the reaction temperature. The reaction was carried out at 250, 280 and 300 °C, at atmospheric pressure. In order to ensure differential reactor conditions, the feed flow was changed between 50 and 200 mL min−1 (according to the reaction temperature) in order to obtain CO conversions lower than 15%. Reactant gases were obtained from Air Liquide (11.0% CO/He, H2, N2, all with more than 99.99% purity). The partial pressure of the reactants was varied between 1 and 2 kPa for CO and 10 and 30 kPa for H2.

The concentration of CO, CO2 and CH4 was determined using a Perkin-Elmer Clarus 580 equipped with a Porapak-Q column using N2 as carrier gas. The detection is performed with a FID detector equipped with a methanizer in order to achieve high sensitivity towards CO and CO2.

2.3. Operando FTIR measurements

Transmission infrared spectra of self-supported Co/SiO2 wafers were collected in situ using a reaction cell provided by In Situ Research & Instruments. The cell was placed in a FTIR spectrometer (Nicolet iS-10, Thermo Scientific) and spectra were recorded at a resolution of 4 cm−1 and 32 scans per spectrum in the 3500–1400 cm−1 range. The IR cell is equipped with CaF2 windows, has connections for inlet and outlet reactant flows, and K-type thermocouples connected to a temperature controller. The gas flows (CO(10%)/He, He, H2 and Ar) are dosed using mass flow controllers (Kofloc 8500, Kojima Instruments). He and CO(10%)/He were purified using oxygen traps. The solid wafer of each catalyst (previously activated as shown in section 2.1) was in situ pretreated in pure H2 in the same way as shown in section 2.2. After pretreatment, the catalyst was cooled in pure H2 to the studied conditions (atmospheric pressure). The spectra were obtained in absorbance mode after subtraction of the background spectrum of the catalyst disk under a H2 atmosphere at the corresponding temperature. The gases at the outlet of the cell were analyzed using a quadrupole mass spectrometer (Omnistar GSD 320) by following the evolution of the m/z = 2(H2), 15(CH4), 18 (H2O), 28 (CO), and 44 (CO2). The inlet streams to the cell contain 10%(v/v) Ar (m/z = 40, Air Liquide, 99.999% purity) and He as balance; the former was used as the internal standard to quantify the concentrations of CH4, CO and CO2 in the inlet and outlet streams after proper calibration of 15, 28 and 44 signals, respectively.

2.4. Thermodynamic calculations

The approach to equilibrium of each reaction within the mechanism was investigated by comparing the experimental results with theoretical calculations performed with Aspen Plus 10.0 simulation software. A REquil reactor model was used to perform chemical and phase equilibria calculations for specific reaction stoichiometry and at the same inlet compositions, pressures and temperatures as those used during experiments. The properties of individual species and mixtures were estimated by the Peng–Robinson equation of state.

3. Results and discussion

3.1. Structural characterization of catalysts

The particle size distribution of cobalt clusters and the mean nanoparticle sizes were estimated from TEM micrographs (Fig. S1). These values are close to the cluster size estimated from the cobalt dispersion measured by H2 chemisorption as well as to the mean crystal size calculated with the Scherrer equation from XRD characterization (Table 2) except for the average crystal size estimated from XRD data for the Co-33 nm, which resulted in quite lower values. Consequently, for further analysis the Co nanoparticle size calculated from the average of the mean particle sizes obtained from the characterization techniques applied to each catalyst was considered.
Table 2 Mean cluster diameters of cobalt in Co/SiO2 catalysts determined from H2 chemisorption, XRD and TEM techniques
Catalyst H2 chemisorption TEM XRD
D (%) d p(Co0)H (nm) d p(Co0)T (nm) d p(Co0)X (nm)
a Metal dispersion; n.m.: Not measured.
Co-4 nm 20.0 4.4 4.7 3.7
Co-13 nm n.m. n.m. 15.0 11.2
Co-33 nm 2.2 46 34.7 19.0


3.2. Modelling of Co surface based on operando FTIR measurements

Fig. 1 shows the effect of temperature on the linearly adsorbed CO molecule under methanation conditions over Co-33 nm (largest particle size). The high H2/CO ratio is consistent with the methanation conditions and avoids the reconstruction of the Co surface, which has been reported for H2/CO <3 gas ratios.24,34 This reconstruction is a consequence of C formation on the Co° surface, leading to the transformation of Co° into Co°C sites, where CO adsorbs at a higher wavenumber (∼2060 cm−1) than on Co° sites (∼2030 cm−1).24 In fact, Fig. 1B shows that the CO peak shifts from 2041 cm−1 to 2067 cm−1 when H2 is removed from the feed at 150 °C (Fig. 1B). Consequently, the methanation condition used in this study, i.e., a feed with a H2/CO gas ratio >8, guarantees that the observed infrared CO bands correspond to linearly adsorbed CO on metallic cobalt (LCo) instead of the CO adsorbed on the reconstructed surface (LCo-C). The adsorbed CO peak increases and shifts to higher wavenumbers when the temperature decreases from 250 to 150 °C. This is attributed to the increase in CO coverage on the cobalt surface; a further decrease of temperature from 150 °C to 100 °C does not significantly affect the area of the LCo band, which means that the surface is saturated with linearly adsorbed CO and this area quantitatively represents full CO coverage (θCO = 1). A shoulder at ∼2070 cm−1 is observed at 100 °C, which disappears at higher temperatures. Small bands below 1900 cm−1 are also observed, which are associated to bridged CO species on metallic cobalt. Even though the molar adsorption coefficient of linear CO is twice as high as that of bridged CO,35 these results indicate that the adsorption of CO on the metallic cobalt is dominated by the LCo species under the studied reaction conditions, which is consistent with the previous study of Couble and Bianchi.34 However, it is important to note that the multi-bonded CO species (specifically hollow CO) have been proposed to be the main reaction intermediates in the hydrogenation of CO at 220 °C, 30% CO + 60% H2/He, while linear CO is considered a spectator that rapidly switches to multi-bonded CO species.35 The steady-state multi-bonded CO coverage observed in our work is lower as compared to the work mentioned above due to the higher H2/CO ratio used in our FTIR measurements. It is also noteworthy that the asymmetry observed for the linearly adsorbed CO band confirms the presence of different types of sites and evidences the heterogeneity of cobalt surface.
image file: c9cy01753d-f1.tif
Fig. 1 In situ FTIR signal for the linearly adsorbed CO on the Co-33 nm catalyst. (A) Effect of temperature at 2 kPa CO–18 kPa H2: (a) 100 °C, (b) 150 °C, (c) 200 °C, and (d) 250 °C. (B) Effect of H2 and CO pressures at 150 and 250 °C: (a) 2 kPa CO 150 °C, (b) 2 kPa CO–18 kPa H2 150 °C, (c) 1 kPa CO–18 kPa H2 250 °C, (d) 1 kPa CO–25 kPa H2 250 °C, and (e) 1 kPa CO–30 kPa H2 250 °C.

The intensity of the LCo peak increases with the CO pressure at 150 °C (Fig. 1B); the H2 pressure, conversely, has a negligible effect on the position and intensity of linearly adsorbed CO. These results indicate that the adsorbed hydrogen atom (H*) is not one of the most abundant surface intermediates and that the cobalt surface is dominated by the coverage of adsorbed CO species.

Similar results were observed on the Co-4 nm catalyst, although a higher CO pressure (3 kPa CO–18 kPa H2) was needed to achieve the full CO coverage at 120 °C (Fig. S2). When the surface is saturated with adsorbed CO species, the shoulder observed at ∼2060 cm−1 is more pronounced on the catalyst with a smaller mean Co cluster size. In the literature, this infrared band has been differently attributed to CO adsorption on the reconstructed Co surface,24,36 to a partially hydrogenated Co(H)–CO species,37 to the CO adsorbed on Coδ+ sites17 or to the adsorption of CO on low-coordinated Co atoms.30,36

In order to estimate the CO coverage (eqn (1)) on supported metallic cobalt catalysts from FTIR measurements, the adsorption equilibrium infrared spectroscopy (AEIR) analytical method31,32,34,38 was used at each experimental condition.

 
image file: c9cy01753d-t2.tif(1)

A(T, PCO) and AM represent the areas of the adsorbed CO bands (in absorbance mode) at each (T, PCO) condition and at full CO coverage, respectively; the latter was estimated from the infrared band of adsorbed CO under saturation conditions as described above (Fig. 1A).

The coverages of LCo species (θCO) at 0–3000 Pa (CO partial pressure) and 150–200 °C were calculated from operando FTIR measurements. A high H2 pressure (18 kPa) with a H2/CO ratio >5 was used for all the measurements in order to avoid the reconstruction of the cobalt surface; therefore ensuring the calculation of the coverage for the linearly adsorbed CO species.24 Results for the catalysts with smaller and larger mean Co cluster sizes are presented in Fig. 2A and B, respectively.


image file: c9cy01753d-f2.tif
Fig. 2 Coverages of linearly adsorbed CO calculated from operando FTIR measurements and eqn (1) at (○) 150 °C and (■) 200 °C on (A) Co-4 nm and (B) Co-33 nm. (⋯⋯): Langmuir model, (- - -): Temkin model. 18 kPa H2.

The mass spectrometry signals confirmed that the CO conversion was lower than 4% for all the operando FTIR measurements, which guarantees that the CO hydrogenation reaction does not affect the equilibrium of CO adsorption. The Langmuir model (eqn (2)) reasonably fits the CO coverage data at 150 °C, but it poorly represents the data measured at 200 °C on both catalysts (Fig. 2).

 
image file: c9cy01753d-t3.tif(2)

The equilibrium constants for CO adsorption were estimated at each temperature and the van't Hoff equation was used to calculate the heat of CO adsorption (QL) on both catalysts. The obtained values for QL were 27 and 54 kJ mol−1 for Co-4 nm and Co-33 nm, respectively (Table 3). These values are significantly below the reported range of 75–181 kJ mol−1, which has been obtained from microcalorimetric and TPD experiments as well as from theoretical calculations.29,36,39–42 This suggests that the Co surface under reaction conditions for CO methanation does not behave as ideal, and thus the approximations of the Langmuir model are not valid in this specific case. This deviation from ideality can be approached by the mathematical formalism of the Temkin model (eqn (3) and (4)), which has been previously applied to study the competitive chemisorption of CO and H2 on Pt/Al2O3 catalyst.38

Table 3 Heat of CO adsorption on cobalt clusters at low (Q0) and high (Q1) coverage calculated from Temkin and Langmuir (QL) models
Catalyst Temkin model Langmuir model
Q 0 (kJ mol−1) Q 1 (kJ mol−1) σ 2 Q L (kJ mol−1) σ 2
a image file: c9cy01753d-t4.tif, N: number of experiments, K: number of fitted parameters.
Co-4 nm 144 72 7.58 × 10−4 27 1.21 × 10−3
Co-33 nm 120 75 3.48 × 10−4 54 3.75 × 10−3


The Temkin formalism represents the heterogeneity of the cobalt surface; thus, it is used here to calculate the equilibrium constant and the heat of CO adsorption at low and high coverages from in situ FTIR data. According to the integral equation approach used to obtain the Temkin model for gas adsorption, the surface coverage with CO may be estimated by eqn (3), while the dependence of the equilibrium constant on CO adsorption as a function of CO coverage is described by eqn (4) (more details can be found in section S.3 in the ESI).

 
image file: c9cy01753d-t5.tif(3)
 
image file: c9cy01753d-t6.tif(4)

Q 0 and Q1 correspond to the heat of CO adsorption at low and high coverage, respectively (Q0 > Q1), m is the mass of the CO molecule, h is Planck's constant, k is Boltzmann's constant, T is the adsorption temperature and Q(θ) is the heat of CO adsorption at θ coverage. Also, ΔQ = Q0Q1 with image file: c9cy01753d-t7.tif, which assumes that the heat of CO adsorption changes linearly with CO coverage in consistency with the Temkin model.43

Therefore, the heat of CO adsorption at low (Q0) and high (Q1) coverage was estimated at 150 and 200 °C for Co-4 nm and Co-33 nm catalysts by combining eqn (3) and (4) and fitting the Temkin model to the CO coverage data (calculated by FTIR measurements and eqn (1)). The fitting of the CO coverage with both Langmuir and Temkin models is presented in Fig. 2 and the values for heat of CO adsorption are compiled in Table 3.

It is observed that the Temkin model represents the CO adsorption on the cobalt surface under the studied methanation conditions with a higher accuracy than the Langmuir model. The accuracy of the fitting was confirmed for the catalysts Co-4 nm and Co-33 nm. This demonstrates that the cobalt surface behaves as heterogeneous rather than homogeneous during CO adsorption under methanation conditions (high H2/CO ratio), which was also reported for Ru (ref. 44) and Pt (ref. 31) under similar experimental conditions. According to the emblematic Temkin's work,43 this heterogeneity is attributed to either different adsorption properties of surface sites (intrinsic heterogeneity) or the effect of the interaction between adsorbed species (induced heterogeneity). The greater difference between the Temkin and the Langmuir models to fit the CO adsorption data is observed at the higher temperature (200 °C). This suggests that under this condition the intrinsic heterogeneity is more important than the induced one as the coverage of CO and hydrogen species is lower at higher temperatures (Fig. 2). This is consistent with the exothermicity of the adsorption phenomena; thus, the interaction between adsorbed species should be lower at higher temperatures.

The heats of CO adsorption at high and low coverages estimated from the CO adsorption data with the Temkin model (Table 3) are in the range of those values reported in the literature.29,39–42

The heat of CO adsorption at high coverage (Q1) is practically the same regardless of the mean Co cluster size. However, at low coverage (Q0) it is significantly higher for the catalyst with smaller cobalt nanoparticles (Co-4 nm); which is explained by the higher proportion of low-coordinated atoms located at corners and edges,45 where the CO molecule adsorbs stronger,46–48 releasing higher energy.49

These results are in line with the reported values from the measurement of the isosteric heat of CO adsorption on cobalt surfaces Co(0001) and Co(10[1 with combining macron]0), where the enthalpy of CO adsorption increased from 96 kJ mol−1 at high CO coverage up to 128 kJ mol−1 at low coverage for Co(0001)42 and it changed from 120 kJ mol−1 to 145 kJ mol−1 for Co(10[1 with combining macron]0).47 Interestingly, the isosteric heat of CO adsorption of 145 kJ mol−1 measured on the Co(11[2 with combining macron]0) surface resulted to be nearly independent of the CO coverage.48

The effect of coverage on the heat of CO adsorption estimated from the Temkin model confirms its capacity to capture the heterogeneity of the Co surface. According to these results, the CO molecules tend to adsorb on low-coordinated atoms of bare surfaces before crowding the more packed surfaces. At higher coverages the average-weighted heat of adsorption decreases not only because the specific energy released by the adsorption on high-coordinated atoms is lower but also because the crowded surface restructures to balance CO surface binding and CO–CO interaction energies. This was also suggested on Pt surfaces from kinetic measurements and theoretical calculations.50

3.3. Kinetic measurements of CO methanation on cobalt catalysts

3.3.1. Thermodynamic calculation for CO hydrogenation under methanation conditions. The equilibrium values for methane (ηMET) and CO2 (ηCO2) formation were calculated for the range of reaction conditions at which the kinetic measurements were carried out (Fig. 3). The very small ηMET values indicate that this reaction is far from equilibrium; therefore, the methane formation rates obtained from the kinetic measurements in the differential reactor are effectively the forward reaction rates for CH4 formation and can be safely used for kinetic and mechanistic analysis. It is assumed that CO2 is preferentially formed from the exothermic water gas shift reaction (CO + H2O ↔ CO2 + H2). Even though the approach to equilibrium values for this reaction are significantly higher than for CH4 formation, the ηCO2 values are below the 10% for the higher temperature, which represents the condition with a lower equilibrium constant, and thus the maximum value for the CO2 formation. These results confirm that the reactions of methane and CO2 formation are far from thermodynamic equilibrium; therefore, the respective rates reported in this work represent forward turnover reaction rates.
image file: c9cy01753d-f3.tif
Fig. 3 The approach to equilibrium image file: c9cy01753d-t8.tif for the reactions of methane (filled symbols) and CO2 (empty symbols) formation as a function of temperature at CO[thin space (1/6-em)]:[thin space (1/6-em)]H2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]10 (triangles) and CO[thin space (1/6-em)]:[thin space (1/6-em)]H2 = 1[thin space (1/6-em)]:[thin space (1/6-em)]25 (squares) ratios in the feed on the Co-13 nm catalyst.
3.3.2. Effect of Co cluster size on the catalytic activity for CO methanation. The catalytic activity of three catalysts with significantly different mean cobalt cluster size was evaluated under methanation conditions, i.e., 250–300 °C, H2/CO >5. The effect of cobalt particle size on the CO consumption turnover rate at 280 °C is shown in Fig. 4. A similar trend was observed at 250 and 300 °C (Fig. S3). It is observed that the reaction rate expressed by exposed Co atoms increases sharply from Co-4 nm to Co-13 nm catalysts, but the turnover rate becomes almost independent of Co dispersion for larger clusters. A similar dependence has been reported for methanation and FTS reactions on different supported-transition metal catalysts, which have shown an increase of TOF with the particle size up to ∼10 nm.14–19 This trend has been attributed either to the blocking by adsorbed CO of low-coordinated atoms (more abundant in smaller particles) or to the diminishing of the amount of a specific type of sites (assumed as active) when the particle size decreases. Both hypotheses will be discussed later in light of our results.
image file: c9cy01753d-f4.tif
Fig. 4 Turnover frequency reaction rate of CO hydrogenation as a function of the cobalt particle size at 280 °C. (□) 1 kPa CO–10 kPa H2, (○) 1 kPa CO–18 kPa H2, (▲) 1 kPa CO–25 kPa H2, (×) 2 kPa CO–18 kPa H2, (+) 2 kPa CO–25 kPa H2.

Also, Fig. 4 evidences the effect of hydrogen and carbon monoxide pressures on the TOFCO; the turnover rate increases with PH2 regardless of the mean cluster size, while the positive effect of PCO on TOFCO is more significant on the Co-33 nm catalyst at 280 °C. These effects have been better detailed in Fig. S4 (ESI).

3.3.3. Mechanism and kinetics for CO methanation on Co catalysts. Kinetic measurements and theoretical calculations agree on the conclusion that the CO hydrogenation reaction on supported cobalt catalysts is determined by CO dissociation, which is preferably assisted by hydrogen instead of direct (unassisted) dissociation.23,26,41,51,52 Consequently, the mechanism of CO hydrogenation on Co surfaces under methanation conditions could be represented by the following sequence of elementary steps:

1. image file: c9cy01753d-u1.tif

2. image file: c9cy01753d-u2.tif

3. CO* + H* → HCO* + *

4. HCO* + H* → HCOH* + *

5. HCOH* + * → CH* + OH*

6. image file: c9cy01753d-t9.tif

7. image file: c9cy01753d-t10.tif

8. image file: c9cy01753d-t11.tif

9. OH* + H* → H2O(g) + 2*

10. OH* + CO* → CO2(g) + H* + *

It is proposed that the first H addition (step 3) is the kinetically relevant step (KRS) in consistency with the operando FTIR measurements shown in Fig. 5. The area of the infrared band associated with CH4 in the gas phase (∼3015 cm−1), which is proportional to the CH4 formation rate, was quantified as a function of the hydrogen pressure in the feed under methanation conditions on Co-4 nm and Co-33 nm catalysts. The observed apparent order on PH2 was about 0.3 for both catalysts (insets in Fig. 5), which are values closer to 0.5 (first H addition) than to 1 (second H addition). Also, the absence of variation of θCO with H2 partial pressure (Fig. 1A) demonstrates that the adsorbed hydrogen atom (H*) is not one of the most abundant surface intermediates. These results suggest that the CO bond dissociation on cobalt clusters under methanation conditions is kinetically determined by the first H addition (step 3).


image file: c9cy01753d-f5.tif
Fig. 5 Effect of H2 pressure on the infrared band associated with gas phase CH4. (A) Co-4 nm at 200 °C 0.25 kPa CO. (a) 5 kPa H2; (b) 10 kPa H2; (c) 18 kPa H2; (d) 30 kPa H2. (B) Co-33 nm at 250 °C 1 kPa CO. (a) 18 kPa H2; (b) 25 kPa H2; (c) 30 kPa H2. Total flow: 50 mL min−1.

Our results are consistent with a previous report of theoretical calculations for the CO methanation reaction on Co(0001), Co(10[1 with combining macron]2) and Co(11[2 with combining macron]0) surfaces, which demonstrate that step 3 is the KRS by comparing the energy of decomposition of the intermediate species HCO* with the energy barrier of further hydrogenations.53 Also, DFT calculations for FTS on the Co(0001) surface with 0.5 ML of CO coverage showed a higher energy barrier (138 kJ mol−1) for the first hydrogen addition (step 3) than for the second (90 kJ mol−1). However, a first order of TOFCO on hydrogen pressure was obtained from kinetic measurements, thus proposing that the first H addition (step 3) is endothermic and quasi-equilibrated and that the second H addition represents the KRS at the evaluated reaction conditions.41

Consequently, eqn (5) is proposed to represent the CO consumption turnover rate, where k3 is the rate constant for the H* + CO* reaction (step 3), θH is the coverage of adsorbed hydrogen and θCO is the coverage of adsorbed CO.

 
TOFCO = k3θCOθH(5)

Langmuir model. The Langmuir–Hinshelwood model (eqn (6)) represents the homogeneous (Langmuirian) behavior of the cobalt surface.
 
image file: c9cy01753d-t12.tif(6)

The site balance of this model considers that vacancies and adsorbed CO species dominate the active surface, which is consistent with the insensitivity of the linearly adsorbed CO band to the change in hydrogen pressure shown in Fig. 1A.

Eqn (6) was used to fit the kinetic data and the corresponding parity plot is shown in Fig. S6 of the ESI. The two parameters (KCO and β = k3KCOimage file: c9cy01753d-t13.tif) were obtained at 250, 280 and 300 °C from regression by using the least mean squares method. The effect of temperature on these parameters is shown in Fig. 6, where the values of the parameters at 300 °C for the Co-33 nm catalyst are omitted because a negative (nonsense) value for KCO was obtained at this condition.


image file: c9cy01753d-f6.tif
Fig. 6 Effect of temperature on the equilibrium constant for CO adsorption KCO (A) and on the parameter β (B), calculated from the fitting of the L–H model (eqn (6)) to kinetic data on Co-4 nm (▲), Co-13 nm (○), and Co-33 nm (□) at 250–300 °C, 1–2 kPa CO, 10–30 kPa H2.

The heat of CO adsorption (QCO) involved in the KCO was estimated from the van't Hoff plot (Fig. 6A) and the apparent activation energy (Eβ) associated to the parameter β was calculated from the Arrhenius plot (Fig. 6B); these values are compiled in Table 4. It is observed that the values for QCO (12–88 kJ mol−1) are significantly lower than those reported for the CO adsorption on cobalt surfaces (75–181 kJ mol−1).39–42 Also, the larger the Co clusters, the higher the QCO, which is consistent with the trend obtained from the fit of FTIR data with the Langmuir model (Table 3). However, this contradicts the Temkin model prediction (Table 3) as well as the trends reported from experimental and theoretical calculations in which a higher heat of CO adsorption is usually observed on smaller clusters.46–49

Table 4 Heats of CO adsorption (QCO) and apparent activation energy (Eβ), calculated from the fitting of the Langmuir model to kinetic data
Parameter Co-4 nm Co-13 nm Co-33 nm
E β [kJ mol−1] 61 58 35
Q CO [kJ mol−1] 12 62 88
E 3 − 0.5QH2 73 120 123


The apparent activation energy involved in parameter β is determined by Eβ = E3QCO − 0.5QH2; therefore, it is possible to estimate the difference between the activation energy for the first hydrogen addition (E3) and half of the dissociative adsorption (0.5QH2). However, in order to compare these activation energies for the H* + CO* elementary step on cobalt surfaces with the experimental values reported in the literature, one needs to measure the heat of H2 dissociative adsorption on each surface.

Temkin model. The Temkin formalism (eqn (7)) captures the heterogeneity of the cobalt surface by considering the change in the heat of CO adsorption with the coverage. The FTIR measurements (Fig. 1B) show that the variation in hydrogen pressure did not affect the CO coverage, which suggests that the hydrogen coverage is too low under the reaction conditions evaluated in this study. This allows the assumption that the effect of coverage on the heat of H2 adsorption and hence on its equilibrium constant is negligible, i.e., KH2 is considered independent of image file: c9cy01753d-t14.tif.
 
image file: c9cy01753d-t15.tif(7)

This means that the cobalt surface behaves as Langmuirian for H2 adsorption under these reaction conditions, and its heterogeneity is just determined by the change of the heat of CO adsorption with the coverage (eqn (8)), leading to a simpler kinetic model (eqn (10)).

 
image file: c9cy01753d-t16.tif(8)
 
image file: c9cy01753d-t17.tif(9)
 
image file: c9cy01753d-t18.tif(10)

The Temkin kinetic model (eqn (10)) depicts the CO consumption turnover rate under methanation conditions, where K0CO = KCO(Q0) and K1CO = KCO(Q1) are the equilibrium constants for CO adsorption at low and high coverages, respectively, ΔQCO = Q0Q1 and the lumped parameter image file: c9cy01753d-t19.tif contains the rate constant of the kinetically relevant hydrogen addition step and the equilibrium constant for H2 dissociative adsorption.

The kinetic data from the fixed bed reactor were fitted by eqn (10) in order to obtain the values shown in Table 5. The parity plot for the CO consumption turnover rate (Fig. 7) demonstrates that this Temkin model properly represents the data; thus, it captures the heterogeneity of the cobalt surface under methanation conditions. The heat of CO adsorption on the catalysts with the smaller and larger mean cobalt clusters at low and high coverages (Table 5) was quite similar to the values estimated from fitting FTIR measurements with the Temkin model (Table 3). It is observed that the cobalt cluster size has no effect on the heat of CO adsorption at high coverage (Q1 = 84–87 kJ mol−1); at low coverage, however, the Q0 increases as the mean cobalt cluster size decreases. This variation is more pronounced between catalysts with 4 and 13 nm Co particle size, which are the ones expected to present the most significant differences in surface morphology.45 These results evidence the consistency between data obtained from two independent measurements (kinetic in a packed bed reactor and in situ FTIR) and demonstrate the capability of the Temkin model to predict the heterogeneity of the cobalt surface during CO methanation. It should be mentioned that this Temkin model only addresses the heterogeneity of a single site surface through the change in the heat of CO adsorption with the coverage, while a more real surface would have a different type of sites (as evidenced in Fig. 1) with different heats of CO adsorption and reactivity properties. This weakness of the model, however, does not significantly affect the goodness of fit shown in Fig. 7.

Table 5 Parameters obtained from the fitting of kinetic data by the Temkin model (eqn (10)) at 250, 280 and 300 °C, 1–2 kPa CO, 10–30 kPa H2
Parameter Co-4 nm Co-13 nm Co-33 nm
a Assuming QH2 = 15 kJ mol−1 for the H2 dissociative adsorption on Co(0001) at 0.5 ML CO coverage.39
α(250 °C) [s−1] 8.30 × 10−4 1.59 × 10−3 3.25 × 10−3
α(280 °C) [s−1] 1.43 × 10−3 7.54 × 10−3 1.62 × 10−2
α(300 °C) [s−1] 2.69 × 10−3 1.88 × 10−2 5.51 × 10−2
E α = E3 − 0.5QH2 [kJ mol−1] 57 123 140
E 3 [kJ mol−1] 65 131 148
Q 0 [kJ mol−1] 141 105 94
Q 1 [kJ mol−1] 87 84 87
R 2 0.96 0.97 0.95



image file: c9cy01753d-f7.tif
Fig. 7 (left) Measured and predicted turnover rates (eqn (10)) with kinetic parameters shown in Table 5 for the CO + H2 reaction at 250–300 °C on Co-4 nm (▲), Co-13 nm (○), and Co-33 nm (□). (right) Temperature dependence of α at 250–300 °C on Co-4 nm (▲), Co-13 nm (○), and Co-33 nm (□).

The temperature dependence of the kinetic parameter α (Fig. 7) indicates that the apparent activation energy Eα increases with the mean cobalt cluster size. Since the apparent activation energy related to the alpha parameter is Eα = E3 − 0.5QH2, the energy required for the kinetically relevant first hydrogen addition (E3) can be calculated if the enthalpy of H2 adsorption (QH2 = −ΔHH2) is known. It has been reported54 that ΔHH2 changes from −116 to −64 kJ mol−1 on hollow sites of FCC Co(0001) when the CO coverage increases from 0 to 1/3 ML for 1/9 ML of hydrogen coverage, the latter is similar to that measured on cobalt catalysts under methanation conditions.16 Also, a value of −15 kJ mol−1 has been reported for the dissociative adsorption of H2 on Co(0001) with 0.5ML of CO,41 which is consistent with the 0.7 ML coverage of linearly adsorbed CO observed from operando FTIR measurements at 2 kPa CO–18 kPa H2 at 250 °C (Fig. S5). If ΔHH2 = −15 kJ mol−1 is assumed, the activation energy for the kinetically relevant first hydrogen addition is 65, 131 and 148 kJ mol−1 on Co-4 nm, Co-13 nm and Co-33 nm, respectively (Table 5). This suggests that the CO bond activation is more favored on the active sites of smaller clusters.

The kinetic data were also fitted by a Temkin model derived from the mechanism in which the first H* addition (step 3) is quasi-equilibrated and the second H* addition (step 4) is the KRS (section S.7 in the ESI). The goodness of fit (R2) resulted in lower values with this model. Moreover, the heats of CO adsorption at low (Q0) and high (Q1) coverages on Co-33 nm were too similar (Table S1), which is clearly inconsistent with the heterogeneous behavior of cobalt surfaces confirmed by operando FTIR measurements and the AEIR method (Temkin model, Table 3) on both Co-4 nm and Co-33 nm catalysts. Consequently, the CO hydrogenation kinetic data on cobalt surfaces under methanation conditions are well represented by a H*-assisted CO bond dissociation mechanism, where the activation of the CO bond occurs after the first H* addition.

According to eqn (9), the variation in the heat of CO adsorption from low to high coverage linearly correlates with the site's distribution on the surface; thus, the difference between these two opposite values (ΔQCO) is related to the heterogeneity of the catalytic surface. This heterogeneity should be determined by the proportion of surface cobalt atoms with different coordination numbers, which is well known to be related to the mean cluster size.45

The FCC cobalt nanocrystal22 may be properly modeled by the max-B5 cuboctahedral geometry.16,45 This model is used to calculate the fraction of atoms with low and high coordination numbers in the cobalt surface as a function of the mean cobalt cluster size, i.e., the fraction of atoms with low coordination number located on edges and corners (xe+c) and the fraction of atoms with high coordination number on terraces (xt). When the ΔQCO normalized by the heat of CO adsorption at high coverage is plotted as a function of the xe+c/xt ratio (Fig. 8), a straight line is observed. Since the calculation of these fractions is very sensitive to the mean cobalt cluster size obtained from different techniques (Table 2), the error bar that considers the extreme dp values are included in Fig. 8.


image file: c9cy01753d-f8.tif
Fig. 8 The change of the heat of CO adsorption from low to high coverage normalized by the heat of CO adsorption at high coverage as a function of the ratio between the fraction of low-coordinated atoms (in edges and corners) and the fraction of high-coordinated atoms (in terraces) at cobalt surfaces of Co-4 nm, Co-13 nm, and Co-33 nm.

It is observed that the heterogeneity of the cobalt surface, represented by the ordinate ΔQCO/Q1, decreases for larger clusters where the high coordination number surface atoms predominate in terraces. Conversely, the smaller nanoparticles with a higher fraction of atoms located on edges and corners show a more heterogeneous cobalt surface. This confirms that the higher value for the heat of CO adsorption observed at low coverage (Q0) on the catalysts with smaller cobalt cluster size can be attributed to the predominance of low-coordinated atoms at this cobalt surface. It is also demonstrated that the linear distribution of energy assumed in eqn (10) is able to describe the heterogeneity of cobalt surfaces; therefore, the proposed Temkin model accurately represents the behavior of the catalytic surface under CO methanation conditions.

The decrease of turnover rates observed as the cobalt particle size decreases (Fig. 4) has been eventually attributed to the irreversible CO adsorption on low-coordinated atoms located at edges and corners, which are more abundant in smaller particles.16 Thus, this strong CO adsorption could block and disable the active site where supposedly the reaction takes place. However, the change in the number of these low-coordinated atoms with the particle size cannot explain the decrease in one order of magnitude observed for the TOF on Co-4 nm as compared with the Co-33 nm catalyst. In fact, the fraction of corner- and edge-located atoms decreases about 6 times from Co-4 nm to Co-33 nm while the CO consumption turnover rate is about 15 times higher on Co-33 nm than on Co-4 nm. This hypothesis also does not explain the lower apparent activation energy observed on the catalysts with smaller cobalt clusters (Fig. 7).

On the other hand, the catalytic activity for CO hydrogenation has been attributed to a specific type of sites or ensembles of atoms named 2B3, B4, B5-A, B5-B, B6, on which the CO and H2 adsorption, activation or reactivity of intermediates is energetically favored, according to theoretical calculations.22,55,56 It has been demonstrated that the surface composition of these types of sites changes with the mean particle size of fcc Co nanocrystals.22,45 Specifically, the number of B5-B sites was correlated with the FTS catalytic activity on cobalt with particle size up to 8 nm; thus, this atom arrangement is proposed to be the active site, where the CO hydrogenation reaction requires a lower activation energy to occur.22,55

In order to explain our results, the so-called max-B5 cuboctahedron model45 was used to calculate the composition (fraction) of the different types of sites present on the Co surface for a wide range of particle sizes as evaluated in this work (Fig. 9). It is observed that the B5-B sites practically disappear for Co particles larger than 15 nm. This hardly explains the higher activities and turnover frequencies observed on the larger clusters of Co-13 nm and Co-33 nm catalysts, where the fraction of B5-B sites is almost zero, as compared with Co-4 nm, in which this fraction is near the maximum value (∼6%).


image file: c9cy01753d-f9.tif
Fig. 9 The fraction of different types of sites as a function of the cobalt particle size according to the max-B5 cuboctahedral geometric model.45 (▲): 2B3, (□): B4, (●): B5-A, (Δ) B5-B sites.

These results indicate that the observed reaction rates represent the weighted averages of the rates on different surfaces with different fractions of these types of sites. Thus, the significantly low fraction of B5-B sites in Co-13 nm and Co-33 nm catalysts suggests that the CO activation and hydrogenation also occur on the high-coordinated atoms located at the terraces of these larger clusters, although a higher activation energy is required. Conversely, the activation barrier results are lower on smaller clusters of Co-4 nm (Fig. 7 and Table 5) because the reaction occurs preferentially on the more active and abundant B5-B sites;22 however, the lower intrinsic activity of the rest of the sites in these small clusters probably cannot compensate for the amount and activity of sites located at terraces. The higher activation energies estimated for the KRS on larger clusters of Co-13 nm and Co-33 nm (Table 5) are consistent with the value of 138 kJ mol−1 reported for CO hydrogenation on large cobalt clusters at FTS conditions.41 Also, the lower activation barrier obtained on the smaller clusters of Co-4 nm is fairly close to those reported for the CO activation on B5-B sites of cobalt clusters.55,57

4. Conclusions

Kinetic measurements for CO hydrogenation under methanation conditions (H2/CO >3) on SiO2-supported cobalt catalysts with different mean Co nanoparticle sizes (4, 13 and 33 nm) show that the CO consumption turnover rate (TOFCO) is significantly lower on smaller nanoparticles with a higher proportion of low-coordinated atoms. The apparent activation energy, however, was lower on smaller nanoparticles (Co-4 nm), which is attributed to the higher presence of the more active B5-B sites in this catalyst. This structural sensitivity is closely related to the energetic heterogeneity of the cobalt surface, which is evidenced by operando FTIR measurements. A kinetic model derived from the Temkin formalism is proposed that properly represents the heterogeneity of the cobalt surface by considering the change in the heat of CO adsorption with the CO coverage. This model is congruent with a H-assisted CO bond dissociation mechanism, where the activation of the CO bond occurs after the first H addition. The heats of CO adsorption obtained from the two independent measurements, i.e., kinetic and operando FTIR, were consistent. The difference between the heat of CO adsorption at low and full coverage (ΔQCO) is normalized by the heat of CO adsorption at high coverage on each catalyst (ΔQCO/Q1) in order to represent the heterogeneity of cobalt surfaces. This ratio is lower in larger nanoparticles where the average coordination number is higher.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors gratefully acknowledge the financial support of the Chilean Government provided through the following projects: FONDECYT 1140410, 1170610, BMBF 150029 and CONICYT PIA/APOYO CCTE AFB170007. The authors also thank Carlos Navas, PhD, for the synthesis of the Co-13 nm catalyst.

References

  1. W. Wei and G. Jinlong, Methanation of carbon dioxide: an overview, Front. Chem. Sci. Eng., 2011, 5(1), 2–10 CrossRef .
  2. K. Ghaib, K. Nitz and F. Z. Ben-Fares, Chemical methanation of CO2: A review, ChemBioEng Rev., 2016, 3(6), 266–275 CrossRef CAS .
  3. A. Duret, C. Friedli and F. Maréchal, Process design of Synthetic Natural Gas (SNG) production using wood gasification, J. Cleaner Prod., 2005, 13(15), 1434–1446 CrossRef .
  4. A. Molino and G. Braccio, Synthetic natural gas SNG production from biomass gasification–Thermodynamics and processing aspects, Fuel, 2015, 139, 425–429 CrossRef CAS .
  5. I. Kiendl, et al., Dilute gas methanation of synthesis gas from biomass gasification, Fuel, 2014, 123, 211–217 CrossRef CAS .
  6. S. Rönsch, et al., Review on methanation–From fundamentals to current projects, Fuel, 2016, 166, 276–296 CrossRef .
  7. D. Hu, J. Gao, Y. Ping, L. Jia, P. Gunawan, Z. Zhong, G. Xu, F. Gu and F. Su, Enhanced investigation of CO methanation over Ni/Al2O3 catalyst for synthetic natural gas prodcution, Ind. Eng. Chem. Res., 2012, 51, 4875–4886 CrossRef CAS .
  8. M. A. Vannice, The catalytic synthesis of hydrocarbons from H2CO mixtures over the Group VIII metals: V. The catalytic behavior of silica-supported metals, J. Catal., 1977, 50(2), 228–236 CrossRef CAS .
  9. R. L. Palmer and D. A. Vroom, Mass-spectrometric measurements of enhanced methanation activity over cobalt and nickel foils, J. Catal., 1977, 50(2), 244–251 CrossRef CAS .
  10. R. C. Reuel and C. H. Bartholomew, Effects of support and dispersion on the CO hydrogenation activity/selectivity properties of cobalt, J. Catal., 1984, 85(1), 78–88 CrossRef CAS .
  11. J. K. Nørskov, et al., Towards the computational design of solid catalysts, Nat. Chem., 2009, 1(1), 37 CrossRef PubMed .
  12. E. Iglesia, Design, synthesis, and use of cobalt-based Fischer-Tropsch synthesis catalysts, Appl. Catal., A, 1997, 161(1), 59–78 CrossRef CAS .
  13. A. Barbier, et al., Characterization and Catalytic Behavior of Co/SiO2 Catalysts: Influence of Dispersion in the Fischer–Tropsch Reaction, J. Catal., 2001, 200(1), 106–116 CrossRef CAS .
  14. G. L. Bezemer, et al., Cobalt Particle Size Effects in the Fischer−Tropsch Reaction Studied with Carbon Nanofiber Supported Catalysts, J. Am. Chem. Soc., 2006, 128(12), 3956–3964 CrossRef CAS PubMed .
  15. Ø. Borg, et al., Fischer–Tropsch synthesis: Cobalt particle size and support effects on intrinsic activity and product distribution, J. Catal., 2008, 259(2), 161–164 CrossRef .
  16. J. P. den Breejen, et al., On the Origin of the Cobalt Particle Size Effects in Fischer−Tropsch Catalysis, J. Am. Chem. Soc., 2009, 131(20), 7197–7203 CrossRef CAS PubMed .
  17. G. Prieto, et al., Cobalt particle size effects in Fischer–Tropsch synthesis: structural and in situ spectroscopic characterisation on reverse micelle-synthesised Co/ITQ-2 model catalysts, J. Catal., 2009, 266(1), 129–144 CrossRef CAS .
  18. H. Xiong, et al., Correlating the preparation and performance of cobalt catalysts supported on carbon nanotubes and carbon spheres in the Fischer–Tropsch synthesis, J. Catal., 2011, 278(1), 26–40 CrossRef CAS .
  19. J.-X. Liu, et al., Particle size and crystal phase effects in Fischer-Tropsch catalysts, Engineering, 2017, 3(4), 467–476 CrossRef .
  20. C. Lancelot, et al., Direct evidence of surface oxidation of cobalt nanoparticles in alumina-supported catalysts for Fischer–Tropsch synthesis, ACS Catal., 2014, 4(12), 4510–4515 CrossRef CAS .
  21. E. van Steen, et al., Stability of Nanocrystals: Thermodynamic Analysis of Oxidation and Re-reduction of Cobalt in Water/Hydrogen Mixtures, J. Phys. Chem. B, 2005, 109(8), 3575–3577 CrossRef CAS PubMed .
  22. P. van Helden, I. M. Ciobîcă and R. L. J. Coetzer, The size-dependent site composition of FCC cobalt nanocrystals, Catal. Today, 2016, 261, 48–59 CrossRef CAS .
  23. W. T. Ralston, et al., Evidence of Structure Sensitivity in the Fischer–Tropsch Reaction on Model Cobalt Nanoparticles by Time-Resolved Chemical Transient Kinetics, Angew. Chem., Int. Ed., 2017, 56(26), 7415–7419 CrossRef CAS PubMed .
  24. J. Couble and D. Bianchi, Experimental Microkinetic Approach of the Surface Reconstruction of Cobalt Particles in Relationship with the CO/H2 Reaction on a Reduced 10% Co/Al2O3 Catalyst, J. Phys. Chem. C, 2013, 117(28), 14544–14557 CrossRef CAS .
  25. A. Banerjee, et al., CO Activation on Realistic Cobalt Surfaces: Kinetic Role of Hydrogen, ACS Catal., 2017, 7(8), 5289–5293 CrossRef CAS .
  26. T. Herranz, et al., Influence of the cobalt particle size in the CO hydrogenation reaction studied by in situ X-ray absorption spectroscopy, J. Phys. Chem. B, 2009, 113(31), 10721–10727 CrossRef CAS PubMed .
  27. W. Chen, et al., Mechanism of Cobalt-Catalyzed CO Hydrogenation: 1. Methanation, ACS Catal., 2017, 7(12), 8050–8060 CrossRef CAS PubMed .
  28. W. Chen, et al., Mechanism of Cobalt-Catalyzed CO Hydrogenation: 2. Fischer–Tropsch Synthesis, ACS Catal., 2017, 7(12), 8061–8071 CrossRef CAS PubMed .
  29. S. J. Jenkins and D. A. King, Theory of CO adsorption on Co{10[1 with combining macron]0}, Surf. Sci., 2002, 504, 138–144 CrossRef CAS .
  30. G. A. Beitel, et al., A Combined in-Situ PM-RAIRS and Kinetic Study of Single-Crystal Cobalt Catalysts under Synthesis Gas at Pressures up to 300 mbar, J. Phys. Chem. B, 1997, 101(20), 4035–4043 CrossRef CAS .
  31. J. Couble and D. Bianchi, Experimental microkinetic approach of the CO/H2 reaction on Pt/Al2O3 using the Temkin formalism. 2. Coverages of the adsorbed CO and hydrogen species during the reaction and rate of the CH4 production, J. Catal., 2017, 352, 686–698 CrossRef CAS .
  32. S. Derrouiche and D. Bianchi, Heats of adsorption of the linear and bridged CO species on a Ni/Al2O3 catalyst by using the AEIR method, Appl. Catal., A, 2006, 313(2), 208–217 CrossRef CAS .
  33. H. S. Fogler, Elementos de ingeniería de las reacciones químicas, Pearson Educación, 2001 Search PubMed .
  34. J. Couble and D. Bianchi, Heats of adsorption of linearly adsorbed CO species on Co2+ and Co° sites of reduced Co/Al2O3 catalysts in relationship with the CO/H2 reaction, Appl. Catal., A, 2012, 445-446, 1–13 CrossRef CAS .
  35. A. Paredes-Nunez, D. Lorito, L. Burel, D. Motta-Meira, G. Agostini, N. Guilhaume, Y. Schuurman and F. Meunier, CO hydrogenation on Cobalt-based catalysts. Tin poisoning unravel CO in hollow sites as main surface intermediate, Angew. Chem., Int. Ed., 2018, 57, 547–550 CrossRef CAS PubMed .
  36. C. J. Weststrate, J. van de Loosdrecht and J. W. Niemantsverdriet, Spectroscopic insights into cobalt-catalyzed Fischer-Tropsch synthesis: A review of the carbon monoxide interaction with single crystalline surfaces of cobalt, J. Catal., 2016, 342, 1–16 CrossRef CAS .
  37. M. J. Heal, E. C. Leisegang and R. G. Torrington, Infrared studies of carbon monoxide and hydrogen adsorbed on silica-supported iron and cobalt catalysts, J. Catal., 1978, 51(3), 314–325 CrossRef CAS .
  38. J. Couble and D. Bianchi, Experimental microkinetic approach of the CO/H2 reaction on Pt/Al2O3 using the Temkin formalism. 1. Competitive chemisorption between adsorbed CO and hydrogen species in the absence of reaction, J. Catal., 2017, 352, 672–685 CrossRef CAS .
  39. B. Li, et al., Vacancy-mediated diffusion of carbon in cobalt and its influence on CO activation, Phys. Chem. Chem. Phys., 2010, 12(28), 7848–7855 RSC .
  40. Š. Pick, Density-functional study of the CO adsorption on the ferromagnetic fcc Co (0 0 1) film surface, Surf. Sci., 2010, 604(3–4), 265–268 CrossRef .
  41. M. Ojeda, et al., CO activation pathways and the mechanism of Fischer–Tropsch synthesis, J. Catal., 2010, 272(2), 287–297 CrossRef CAS .
  42. H. Papp, The chemisorption of carbon monoxide on a Co (0001) single crystal surface; studied by LEED, UPS, EELS, AES and work function measurements, Surf. Sci., 1983, 129(1), 205–218 CrossRef CAS .
  43. M. I. Temkin, The kinetics of some industrial heterogeneous catalytic reaction, Adv. Catal., 1979, 28, 173–291 CAS .
  44. J. Liu, D. Hibbitts and E. Iglesia, Dense CO Adlayers as Enablers of CO Hydrogenation Turnovers on Ru Surfaces, J. Am. Chem. Soc., 2017, 139(34), 11789–11802 CrossRef CAS PubMed .
  45. R. Van Hardeveld and F. Hartog, The statistics of surface atoms and surface sites on metal crystals, Surf. Sci., 1969, 15(2), 189–230 CrossRef CAS .
  46. F. A. Reboredo and G. Galli, Size and structure dependence of carbon monoxide chemisorption on cobalt clusters, J. Phys. Chem. B, 2006, 110(15), 7979–7984 CrossRef CAS PubMed .
  47. H. Papp, The Chemisorption of Carbon Monoxide on a Co(10&1macr;0) Single Crystal Surface. Studied by LEED, UPS, EELS, AES and Work Function Measurements, Ber. Bunsen-Ges., 1982, 86(6), 555–562 CrossRef CAS .
  48. H. Papp, Chemisorption and reactivity of carbon monoxide on a Co(11[2 with combining macron]0) single crystal surface; Studied by LEED, UPS, EELS, AES and work function measurements, Surf. Sci., 1985, 149(2), 460–470 CrossRef CAS .
  49. K. Liao, et al., Single-crystal adsorption calorimetry and density functional theory of CO chemisorption on fcc Co {110}, Phys. Chem. Chem. Phys., 2013, 15(11), 4059–4065 RSC .
  50. A. D. Allian, K. Takanabe, K. L. Fujdala, X. X. Hao, T. J. Truex, J. Cai, C. Buda, M. Neurock and E. Iglesia, Chemisorption of CO and Mechanism of CO Oxidation on Supported Platinum Nanoclusters, J. Am. Chem. Soc., 2011, 133, 4498–4517 CrossRef CAS PubMed .
  51. A. Tuxen, et al., Size-dependent dissociation of carbon monoxide on cobalt nanoparticles, J. Am. Chem. Soc., 2013, 135(6), 2273–2278 CrossRef CAS PubMed .
  52. J.-X. Liu, et al., Crystallographic dependence of CO activation on cobalt catalysts: HCP versus FCC, J. Am. Chem. Soc., 2013, 135(44), 16284–16287 CrossRef CAS PubMed .
  53. J.-X. Liu, H.-Y. Su and W.-X. Li, Structure sensitivity of CO methanation on Co (0001), (1[thin space (1/6-em)]0[thin space (1/6-em)][1 with combining macron][thin space (1/6-em)]2) and (1[thin space (1/6-em)]1[thin space (1/6-em)][2 with combining macron][thin space (1/6-em)]0) surfaces: Density functional theory calculations, Catal. Today, 2013, 215, 36–42 CrossRef CAS .
  54. M. Zhuo, A. Borgna and M. Saeys, Effect of the CO coverage on the Fischer–Tropsch synthesis mechanism on cobalt catalysts, J. Catal., 2013, 297, 217–226 CrossRef CAS .
  55. P. van Helden, J.-A. van den Berg and I. M. Ciobîcă, Hydrogen-assisted CO dissociation on the Co (211) stepped surface, Catal. Sci. Technol., 2012, 2(3), 491–494 RSC .
  56. Q. Ge and M. Neurock, Adsorption and activation of CO over flat and stepped Co surfaces: a first principles analysis, J. Phys. Chem. B, 2006, 110(31), 15368–15380 CrossRef CAS .
  57. S. Shetty and R. A. van Santen, Hydrogen induced CO activation on open Ru and Co surfaces, Phys. Chem. Chem. Phys., 2010, 12(24), 6330–6332 RSC .

Footnote

Electronic supplementary information (ESI) available. See DOI: 10.1039/c9cy01753d

This journal is © The Royal Society of Chemistry 2019