A potassium-promoted Mo carbide catalyst system for hydrocarbon synthesis

Abstract

Potassium-promoted Mo carbide catalysts prepared by temperature-programmed carburization with H₂–C₃H₈ have been evaluated for Fischer–Tropsch synthesis (FTS). Temperature-programmed carburization of MoO₃–Al₂O₃ with a mixture of H₂–C₃H₈ was a two-stage conversion involving the transformation of Mo oxide to the final carbide phase via an intermediate oxycarbide phase. Compensation effect and isokinetic relationship were observed for both oxycarbide and carbide phase formation suggesting that a similar topotactic mechanism was involved in the solid phase transformation. CO seemed to adsorb more strongly than H₂ as implicated by its higher heat of desorption and site concentration. Both acid and basic centres were detected on the Mo carbide surface. The K-promoter increased site concentration for both weak and strong basic sites. The relationship between FT activity and K loading paralleled the trend in CO₂ heat of desorption with K addition. FT activity attained a maximum at about 3%K loading. CO hydrogenation over the Mo carbide catalyst was most suitably described by an enolic intermediate mechanism. A generalized kinetic model was derived for predicting the reaction rate for individual hydrocarbons as a function of chain growth propagation and olefin-to-paraffin ratio.

---

Introduction

Following the original report by Levy and Boudart on the Pt-like characteristics of the Mo carbide catalyst,¹ it has been considered a promising catalyst for the Fischer–Tropsch synthesis (FTS) of higher hydrocarbons.^2,3 Other studies have shown that MoC_1−x (0 ≤ x < 1) exhibited high olefin selectivity, carbon and sulphur resilience.^4,5 This makes it an attractive catalyst for natural gas conversion to clean fuels since most gas fields contain sulphur compounds.

The common method for synthesizing Mo carbide with high surface area is temperature-programmed carburization between MoO₃ and H₂–CH₄ at temperature greater than 1073 K. The physicochemical properties and catalytic performance of metal carbide are however influenced by the type of carburizing agent and conditions.⁶ The utilization of higher alkanes (C₂₊) as a carbon source has been shown to reduce the carburization temperature and improve the surface area⁷ albeit with a potential for high carbon deposition. H₂–C₃H₈ was first used to synthesize an Al₂O₃-supported Mo carbide catalyst with a surface area of 92–204 m² g_cat⁻¹ from a metal sulphide precursor with an optimum of H₂ : C₃H₈ = 5 : 1 for negligible surface carbon resilience on the catalyst.⁸ Conventional FT catalysts are generally promoted by alkali addition. Thus, the possible substitution of expensive Pt noble metal with Mo carbide in FT catalysis may also benefit from similar promotion. Therefore, in this study, the effect of K loading (1–5 wt%) on supported Mo carbide prepared by temperature-programmed carburization using the H₂–C₃H₈ mixture has been investigated. Specifically, the objective of this paper was to seek the correlation between K loading and the physicochemical properties as well as FT reaction metrics.

Experimental

10%MoC_1−x/Al₂O₃ catalysts promoted with different levels of K (1–5 wt%) were prepared by a co-impregnation method. Calculated amounts of aqueous (NH₄)₆Mo₇O₂₄·4H₂O and K₂CO₃ solution precursor were mixed with γ-Al₂O₃ (pre-treated at 973 K in air with 5 K min⁻¹ for 6 h to ensure thermal stability) and stirred for 3 h. The resulting slurry was subsequently dried in an oven for 16 h at 403 K. Temperature-programmed carburization between the alumina-supported MoO₃ and a mixture of H₂ : C₃H₈ = 5 : 1 at 50 mL min⁻¹ was carried out in a computer-controlled fixed-bed microreactor at 973 K using a heating rate of 10 K min⁻¹ for 2 h. FT evaluation was performed in situ in the same reactor by switching to a H₂ : CO feed mixture (7 different ratios) at 453–473 K and atmospheric pressure. A mean catalyst particle size of 100 μm and gas hourly space velocity (GHSV) of 10 L g_cat⁻¹ h⁻¹ were used for all FT runs to ensure negligible internal and external transport resistances. Reproducibility check carried out for selected runs indicated an experimental error for rate measurement of between 2.4% and 5.1% depending on the carbon chain length.

The catalyst surface area, pore volume and pore diameter were measured on a Quantachrome Autosorb-1 unit with nitrogen physisorption at 77 K. A ThermoCahn TGA 2121 unit was used to carry out temperature-programmed carburization of Mo oxide with a gas containing 5H₂ : 1C₃H₈ from 303 to 973 K with 4 different heating rates (5–20 K min⁻¹). Temperature programmed desorption runs were performed on a Micromeritics 2910 AutoChem to quantify heat of desorption, site concentration for acid (NH₃-TPD), and basic sites (CO₂-TPD) as well as reaction sites (CO-TPD, and H₂-TPD). The catalyst sample was exposed to a flow of 10%NH₃, 10%CO₂, 10%CO, and 10%H₂ diluted in an inert gas, respectively, for 1 h followed by controlled heating to 973 K with different ramping rates (5–30 K min⁻¹). X-Ray diffraction measurement was carried out on a Philips X'pert Pro MPD system utilizing Ni-filtered Cu Kα with λ = 1.542 Å operating at 45 kV and 40 mA.

Results and discussion

X-Ray diffraction measurement

Fig. 1 shows the XRD patterns of both undoped and K-promoted Mo carbide catalysts. The Joint Committee on Powder Diffraction Standards (JCPDS) database was used for analysing X-ray diffractograms.⁹ The high intensity peaks located at 45.7° and 66.7° may be attributed to the γ-Al₂O₃ support. It is evident that both face-centred cubic (FCC) α-MoC_1−x (2θ = 36.6° and 61.3° for [111] and [220] respectively) and hexagonal closed packed (HCP) β-MoC_1−x (34.0° and 39.5°) were formed in both K-free and K-modified Mo carbide catalysts. As seen in the 2%K–10%MoC_1−x/Al₂O₃ pattern (b), the peak located at 19.5° was ascribed to the formation of KAl₅O₈ (or K₂O·5Al₂O₃) while carbon deposition on the catalyst surface was observed at 2θ = 26.6°. Absence of peaks for MoO₃ (2θ = 23.40°, 25.50° and 26.75°) in the carbide catalysts indicates that Mo oxide was completely converted to the carbide phase during carburization.

Fig. 1 XRD patterns of (a) K-free and (b) K-promoted Mo carbide catalysts synthesized with H₂ : C₃H₈ = 5 : 1.

Thermogravimetric studies

The profile for the temperature-programmed carburization of Mo oxide is illustrated in Fig. 2a. Two discernible peaks, P1 (at 650–700 K) and P2 (at a high temperature of 800–850 K), were ascribed to the formation of oxycarbide and carbide phases respectively.^10,11 As seen in Fig. 2b, the peak temperature increased linearly with the heating rate. Thus, activation energy, E_a, and pre-exponential factor, A, for the production of oxycarbide and carbide may be computed from the Kissinger equation:¹²where β, T_P, and R are heating rate, peak temperature and universal gas constant respectively.

Fig. 2 Temperature-programmed carburization of 1%K–10%MoC_1−x/Al₂O₃ catalyst.

As seen in Table 1, activation energy values for transformation from MoO₃ to oxycarbide were generally lower than those for the formation of carbide from the oxycarbide phase regardless of the K-promotion or the level of K-loading. Interestingly, activation energy and pre-exponential factor of both oxycarbide and carbide forms may be fitted to the expression given by eqn (2) (with R² of 0.995) implicating the existence of a ‘compensation effect’ (cf.Fig. 3) where a and b are model parameters estimated as 1.76 × 10⁻⁴ and −1.32 (for oxycarbide), and 1.75 × 10⁻⁴ and −5.52 (for the carbide phase) respectively.

ln A_j = aE_aj + b (2)

Fig. 3 Evidence for the existence of a compensation effect.

The existence of a compensation effect and isokinetic relationship has been investigated in other studies to justify a common reaction mechanism for either a series of reactions over a specific catalyst or a particular reaction over a group of catalysts.^13–15 Although the linear relationship between ln A_j and E_aj implicates a compensation effect, it does not necessarily guarantee the occurrence of an isokinetic phenomenon. The requirement for the existence of an isokinetic relationship may be derived from the criterion recommended by Liu and Guo¹⁵ as

with h and k_B being Planck and Boltzmann constant, respectively, while ΔG^≠_j is the Gibbs free energy for the associated transition state complex during the carburization reaction. An isoequilibrium relationship exists if the plots of ΔG^≠_jvs. T for all catalysts have a common point of intersection. As seen in Fig. 4, there is a feasible isokinetic temperature, T_iso, of about 685 K for both oxycarbide and carbide phases. The existence of compensation effect and isoequilibrium relationship with similar T_iso suggests that the conversion from Mo oxide to oxycarbide and oxycarbide to carbide species was governed by a similar topotactic mechanism (substitutionary exchange of oxygen atoms in the MoO₃ lattice by carbon with negligible structural changes).

Fig. 4 Evidence for the existence of an isokinetic relationship for both oxycarbide and carbide phases.

Table 1 Estimates of E_a and A for the formation of oxycarbide and carbide phases

Catalysts	Oxycarbide (P1)		Carbide (P2)
	Activation energy, Ea/kJ mol^−1	Pre-exponential factor, A/s^−1	Activation energy, Ea/kJ mol^−1	Pre-exponential factor, A/s^−1
10%MoC1−x/Al2O3	92.39	4.75 × 10^6	176.56	1.09 × 10^11
1%K–10%MoC1−x/Al2O3	76.86	2.69 × 10^5	152.88	1.57 × 10^9
2%K–10%MoC1−x/Al2O3	111.98	7.63 × 10^7	128.55	2.49 × 10^7
3%K–10%MoC1−x/Al2O3	117.51	3.36 × 10^8	120.95	9.18 × 10^6
4%K–10%MoC1−x/Al2O3	91.61	2.19 × 10^6	108.16	4.83 × 10^5
5%K–10%MoC1−x/Al2O3	83.98	4.79 × 10^5	101.29	2.18 × 10^5

---

Fig. 5 shows that the activation energy for the carbide formation decreased rapidly with increased K loading in a hyperbolic approaching an asymptotic value at high K content probably due to the similarities between the solid properties of the high K content on Mo carbide and that of bulk K₂O. The decrease in activation energy is indicative of the more facile Mo–carbon bond formation due to the higher electron density from the K dopant. Thus, the activation energy profile shown in Fig. 5 may be expressed by

where E_K–MoC1−x and E_MoC1−x are activation energy for K-promoted and K-free Mo carbide catalysts respectively. ω is the activation energy reduction coefficient estimated as 15.53 ± 0.07 while C_K is the potassium loading.

Fig. 5 Influence of potassium loading on the carbide formation rate.

This observation is supported by the exponential rise in the carbide formation rate with K-loading seen in the same plot and appeared to hold at all heating rates used. However, the enhancement in the carbide rate cannot be indefinite since at a critical K loading, the catalyst properties would be nearly identical to the bulk of K₂O and hence the carbide formation rate will level off. The carbide formation rate may be expressed by the empirical equation

r_Carb,K = r_Carb,0e^λC_K (5)

where r_Carb,K and r_Carb,0 are the carbide formation rate of K-promoted and unpromoted MoC_1−x catalysts respectively. λ is an enhancement factor for Mo carbide formation and is a function of the heating rate. There is a linear relationship between λ and heating rate, β, given by eqn (6):

λ = λ₀ + φβ (6)

with λ₀ and φ being associated model parameters estimated as 54.49 ± 0.38 and −1.14 ± 0.01, respectively. Hence, eqn (5) may be rearranged as

r_Carb,K = r_Carb,0e^λ₀C_Ke^φβC_K (7)

Physicochemical properties

As seen in Table 2, the BET area for the promoted Mo carbide catalysts ranges between 171 and 194 m² g_cat⁻¹. Additionally, the unpromoted catalyst has a higher BET surface area, average pore volume, and average pore diameter than K-promoted Mo carbide catalysts. The surface area of Mo carbide decreased with increased K loading. However, the relatively small variation of BET area among the promoted catalysts suggests that the level of K addition is well below that for monolayer coverage and hence the K₂O–MoC_1−x crystallites are most likely well dispersed.

Table 2 Physical properties of Mo carbide catalysts

Catalyst	Average BET surface area/m^2 g^−1	Average pore volume/cm^3 g^−1	Average pore diameter/nm
Pure Al2O3	179.3	0.68	15.2
10%MoC1−x/Al2O3	194.0	0.73	15.0
1%K–10%MoC1−x/Al2O3	185.0	0.52	11.2
2%K–10%MoC1−x/Al2O3	184.3	0.64	13.8
3%K–10%MoC1−x/Al2O3	179.3	0.66	14.8
4%K–10%MoC1−x/Al2O3	175.22	0.61	14.8
5%K–10%MoC1−x/Al2O3	171.4	0.62	14.5

---

Temperature-programmed desorption runs for MoC_1−x catalysts are shown in Fig. 6. The existence of both weak (P1 at 370–400 K) and strong (P2 at 580–600 K) basic centres on the surface of Mo carbide catalysts (cf.Fig. 6a) is evident. Indeed, as seen in Fig. 6b, the weak (510–540 K) and strong (690–720 K) acid sites were also detected from NH₃-TPD profiles. However, the pure calcined Al₂O₃ support also possessed the weak acid centre suggesting that the strong acid site was formed during Mo carbide production, or in the Mo carbide phase.

Fig. 6 Temperature-programmed desorption profiles for Mo carbide catalysts. (a) CO₂-TPD, (b) NH₃-TPD, (c) CO-TPD, (d) H₂-TPD.

The nature of CO and H₂ adsorption sites was also determined as shown in Fig. 6c and d. Interestingly, the CO desorption temperature window is almost identical to that for peak P2 during CO₂ desorption (cf.Fig. 6a) suggesting that CO during the FT reaction was most likely adsorbed on a strong basic site. In particular, since the CO desorption temperature range (580–610 K) seen in Fig. 6c is higher than the FTS reaction window (453–523 K), this would suggest relatively high surface molecular CO coverage during the FT reaction. The H₂-TPD spectrum shown in Fig. 6d, however, shows that relatively little H₂ coverage (if any) would be present on the catalyst surface during CO hydrogenation.

Indeed, site characteristics are determined by these probe molecules using eqn (8):

with E_di and A_i being heat of desorption (kJ mol⁻¹) and site concentration for species i, respectively (cf.Tables 3 and 4), while C is the instrument calibration constant.

Table 3 Basic and acid character of 3%K–10%MoC_1−x/Al₂O₃ catalyst

	A NH3 (mol NH3 gcat^−1 × 10^5)	E NH3/kJ mol^−1	A CO2 (mol CO2 gcat^−1 × 10^5)	E CO2/kJ mol^−1	Temperature range/K
Weak acid site	3.76	51.22	—	—	510–540
Strong acid site	6.21	60.88	—	—	690–720
Weak basic site	—	—	6.65	85.21	370–400
Strong basic site	—	—	9.56	143.87	580–610

---

Table 4 Physicochemical properties of active sites on 3%K–10%MoC_1−x/Al₂O₃ catalyst for CO and H₂ adsorption

	A CO (mol CO gcat^−1 × 10^5)	E CO/kJ mol^−1	A H2 (mol H2 gcat^−1 × 10^5)	E H2/kJ mol^−1	Temperature range/K
CO-TPD	5.45	80.49	—	—	580–610
H2-TPD	—	—	1.52	26.22	500–610

---

Fig. 7a shows that basic site concentration, A_CO2, increased with K loading while heat of desorption for basic sites (weak, E_dW-Basic or strong, E_dS-Basic) exhibited optimal values at between 2 and 3%K loading. The enhancement of CO₂ adsorption with K loading may be ascribed to electron donation from K dopant to surface MoC_1−x thus facilitating CO₂ molecular chemisorption.¹⁶ In contrast, as seen in Fig. 7b, both acid site concentration (weak, A_W-Acid and strong, A_S-Acid acid sites) and heat of desorption for a strong acid centre, E_dS-Acid, dropped exponentially with K content. The heat of desorption and site concentration for both acid and basic centres of the pure calcined Al₂O₃ support were also estimated as seen in Table 5. NH₃ uptake on the weak acid site for the support seemed to be superior to that of doped and undoped catalysts while the difference in heat of desorption between support and catalysts was quite small. The formation of the Mo carbide phase enhanced the strength of the strong basic site (E_dS-Basic > 57.42 kJ mol⁻¹).

Fig. 7 Effect of K loading on basic, acid and reaction sites for Mo carbide catalysts.

Table 5 Physicochemical properties of pure calcined Al₂O₃ support

	Adsorbed NH3, ANH3 (mol NH3 gcat^−1 × 10^5)		Heat of desorption, Ed for NH3/kJ mol^−1		Adsorbed CO2, ACO2 (mol CO2 gcat^−1 × 10^5)		Heat of desorption, Ed for CO2/kJ mol^−1
Pure calcined	P1	P2	P1	P2	P1	P2	P1	P2
Al2O3 support	18.8	—	50.00	—	2.64	1.63	50.00	57.42

---

Fig. 7c shows that the CO uptake, A_CO, increased with K loading consistent with the results for CO₂-TPD. The higher electron density in the K-promoted catalyst means that the back donation of electrons from the CO molecule into the MoC_1−x is reduced thus increasing molecular rather than dissociative CO adsorption. Therefore, the formation of oxygenated intermediate species during FTS will be more likely. In fact, the production of alcohols over K-promoted MoC_1−x has been reported in the literature.^17,18 Furthermore, CO heat of desorption appeared to decline with increased K loading and may be captured by the empirical relation;

where E_dCO and E_dCO0 are CO heat of desorption for K-promoted and unpromoted catalysts, respectively, while σ and ψ are associated model parameters estimated as 2.02 ± 0.04 and 25.1 ± 0.51 in that order.

As seen in Fig. 7d, site concentration for H₂ adsorption reduced with promoter modification and levelled off at about 3%K loading, almost parallel to the behaviour with respect to acid character (cf.Fig. 7b). It would appear that the rich electron environment provided by the K-promoter is detrimental to H₂ chemisorption because it reduced the heat of desorption and adsorption site concentration. It may therefore be inferred that H₂ probably adsorbed on an acidic site (both weak and strong).

Interestingly, a similar trend (hyperbolic decay) for both acid site and H₂ active site characters with respect to potassium loading may be seen in Fig. 7b and d. The influence of K loading on the properties of acid site and H₂-TPD attributes may be adequately captured by;

where Ω is catalyst property (acid site, H₂ active site concentration or heat of desorption of these sites), while Ω₀ is catalytic attribute of the K-free Mo carbide catalyst and γ is an electron density-related coefficient estimated in Table 6.

Table 6 Estimated model parameters for eqn (10)

	H2 uptake, AH2 (mol H2 gcat^−1)	H2 heat of desorption, Ed-H2/kJ mol^−1	Total NH3 uptake, ANH3 (mol NH3 gcat^−1)	Weak acid site concentration, AW-Acid (mol NH3 gcat^−1)	Strong acid site concentration, AS-Acid (mol NH3 gcat^−1)	Heat of desorption for strong acid site, ES-Acid/kJ mol^−1
Ω 0	2.28 × 10^−5	61.76	3.46 × 10^−4	1.28 × 10^−4	2.19 × 10^−4	317.8
γ	0.14	0.36	1.06	0.76	0.14	1.82

---

Fischer–Tropsch performance evaluation

Influence of feed composition. The performance of the Mo carbide catalysts for FTS was evaluated at 473 K with 7 different H₂ : CO ratios as seen in Fig. 8a. The CO consumption rate increased with the H₂ mole fraction, y_H2, and attained an optimum at 0.67 ≤ y_H2 ≤ 0.75 which is somewhat lower than that for common FT catalysts (Co and Fe) where the maximum activity is often found in the region 0.8 < y_H2 < 0.9.¹⁹ It would seem that the surface active site for hydrocarbon synthesis was the oxycarbide type centre, which is formed in situ by an initial molecular CO adsorption.^20,21 In particular, since the reaction temperature window was within the H₂ desorption temperature range, there was probably little or no surface H adatoms and formation of any surface hydrocarbon precursor species would most likely be due to H₂ gas phase attack of the molecularly adsorbed CO.

Fig. 8 FT activity of alumina-supported Mo carbide catalysts.

As seen in Fig. 8b, the reaction rate profile has a maximum at 3%K loading beyond which the FTS activity dropped. It is apparent that the FT reaction curve in Fig. 8b and the CO uptake-vs.-K loading behaviour (cf.Fig. 7c) as well as the profile describing the CO₂ heat of desorption with respect to K loading (cf.Fig. 7a) are all parallel implicating a correlation between the catalyst activity and the physicochemical attributes (CO uptake and the basic site strength on the catalyst). Specially, the relation may be captured by;

where ϖ is the FT activity (or CO uptake or basic site strength) and Ψ is the enhancement factor of the promoter on the catalyst attributes while ϖ_max is the maximum value of the particular attribute at the optimal K loading, C_Kmax. Estimated model parameters are summarized in Table 7.

Table 7 Computed model parameters for eqn (11)

	Methane formation rate, rC1/mol gcat^−1 s^−1	CO reaction rate, rCO/mol gcat^−1 s^−1	Olefin formation rate, rOlefin/mol gcat^−1 s^−1	Heat of desorption for weak basic site, EW-Basic/kJ mol^−1	Heat of desorption for strong basic site, ES-Basic/kJ mol^−1	CO uptake, ACO (mol CO gcat^−1)
ϕ max	3.47 × 10^−7	7.82 × 10^−7	5.86 × 10^−8	82.67	149.60	5.40 × 10^−5
C Kmax	3.64	3.54	3.05	3.27	1.93	3.06
Ψ	1.66	1.99	1.26	2.25	6.58	4.87

---

The 3%K–10%MoC_1−x/Al₂O₃ catalyst seemed to be the optimal catalyst in terms of FT activity and olefin formation rate. However, the reaction rate decreased beyond 3%K loading probably due to weak CO adsorption and reduced surface concentration suggesting that a balance between excessive electron density and high electron-deficiency is required to obtain “best” FT activity. Fig. 9 shows the effect of CO uptake on FT activity over the Mo carbide catalyst. The hydrocarbon production rate increased with CO adsorption. However, the plot possesses a positive intercept suggesting that CO hydrogenation over Mo carbide catalysts requires an initial CO uptake, which is responsible for the formation of an active oxycarbide phase from the Mo carbide form. Generally, product distribution in FT synthesis is governed by a polymerization scheme. It is assessed from a fit of hydrocarbon formation rate data to the Anderson–Schulz–Flory model²² given as

r_n = k_ASF (1 − α)²αⁿ⁻¹ (12)

where α is the chain growth probability whilst r_n is the hydrocarbon formation rate with carbon number, n, and k_ASF is the Anderson–Schulz–Flory constant. Fig. 10 shows that the chain growth probability varied with the H₂ mole fraction and attained an optimal value at y_H2 between 0.25 and 0.5 depending on K loading. The optimum α-values for the K-doped catalyst were greater than the pure MoC_1−x catalyst but occurred at a feed composition much lower y_H2. The decrease in α at y_H2 higher than the optimum is indicative of a higher termination rate compared to the propagation rate. As seen in Fig. 11, the chain growth factor increased with K loading and exhibited a maximum at around 3–4% depending on feed composition. The chain growth probability of K-promoted Mo carbide catalysts is lower than that of conventional FT catalysts (i.e. Fe- and Co-based catalysts) with values of 0.8–0.9.²² In fact, MoC_1−x catalysts were also reported to possess a low chain growth factor of less than about 0.41 in other studies.^23,24 In comparison with the literature, K addition improved the α value of the Mo carbide catalyst by up to 52% even though FTS runs were carried out at atmospheric pressure in this study.

Fig. 9 Effect of CO uptake on the FT reaction rate over Mo carbide catalysts.

Fig. 10 Influence of H₂ mole fraction on chain growth probability.

Fig. 11 Effect of promoter loading on chain growth probability at 473 K.

Table 8 shows the FT activity and selectivity for the K-promoted 10%MoC_1−x/Al₂O₃ catalyst system at 473 K and H₂–CO = 2 : 1. CO conversion varied from 4.8% (unpromoted catalyst) to 12.7% (3%–10%MoC_1−x/Al₂O₃ catalyst). High molecular weight hydrocarbon selectivity, C₅₊, improved with K loading and peaked at 3 wt% K whilst CH₄ selectivity decreased with K addition (from 0–3 wt%) but increased up to 71% at high promoter loading. The C₅₊ and CH₄ selectivity versus K-loading profiles are in agreement with the chain growth factor vs. promoter loading behaviour (cf.Fig. 11). Although the K promoter enhanced the olefin formation rate as seen in Fig. 8b, the unpromoted Mo carbide catalyst appeared to be the optimal catalyst in terms of olefin selectivity. As may be seen in Table 9, CO consumption rates over K-promoted Mo carbide catalysts seemed to be comparable with the performance using conventional FT catalysts suggesting that it could be a replacement for the expensive noble metal catalysts (Pt- and Ru-based systems).

Table 8 Fischer–Tropsch synthesis performance over K-promoted Mo carbide catalysts at 473 K and H₂–CO = 2 : 1

Catalyst	CO conversion (%)	FT product selectivity (%)
		CH4	Olefin (C2+)	Paraffin (C2+)	C5+
10%MoC1−x/Al2O3	4.76	69.52	18.55	9.67	2.35
1%K–10%MoC1−x/Al2O3	5.44	69.33	9.21	16.21	2.97
2%K–10%MoC1−x/Al2O3	8.54	67.14	11.81	16.12	3.48
3%K–10%MoC1−x/Al2O3	12.66	66.73	13.27	15.91	3.61
4%K–10%MoC1−x/Al2O3	10.73	70.76	17.68	9.38	3.53
5%K–10%MoC1−x/Al2O3	8.74	71.22	7.50	16.53	3.43

---

Table 9 Summary of Fischer–Tropsch activity over conventional catalysts at H₂–CO = 2 : 1

Catalyst	Reaction conditions		CO consumption rate × 10^7 mol^−1 gcat^−1 s^−1	Ref.
	Temperature/K	Pressure/atm
0.2%Ru–33.5%Co/ZrO2–SiO2	463 K	8.8 atm	7.17	25
1.5%K–3.5%Cu–69%Fe/Al2O3	523 K	31 atm	6.67	26
20%Co/Al2O3	493 K	20 atm	9.30	27
10%Co–10%Mo/Al2O3	493 K	20 atm	0.96	27
Ru-promoted 33.5%Co/ZrO2–Aerosil	463 K	6 atm	3.72	28
4%K–6%Co–1%Mo/SiO2	523 K	1 atm	0.78	4
4%K–6%Co/SiO2	523 K	1 atm	2.39	4
1%K–Fe2O3	523 K	8 atm	3.00	29
2%K–10%MoC1−x/Al2O3	473 K	1 atm	4.93	This work
3%K–10%MoC1−x/Al2O3	473 K	1 atm	7.32	This work
4%K–10%MoC1−x/Al2O3	473 K	1 atm	6.20	This work

---

Effect of reaction temperature. Since 3%K–10%MoC_1−x/Al₂O₃ exhibited highest CO and olefin reaction rate, it was utilized as a catalyst for additional FT runs over the range 453–473 K. Fig. 12 shows the rate profiles at three temperatures. It is evident that the optimal feed composition at y_H2 = 0.67 was unchanged with reaction temperature and CO consumption rate improved with increasing reaction temperature. Interestingly, the chain growth probability also increased with temperature from 1.6% to 18.3% depending on feed composition. However, the α value may decrease at critical temperature due to an increase in the termination rate of surface olefinic species to paraffins and decreasing re-adsorptivity of α-olefins for chain propagation.^30,31

Fig. 12 Effect of reaction temperature on the CO reaction rate over 3%K–10%MoC_1−x/Al₂O₃ catalyst.

From CO-TPD and H₂-TPD experiments, it is apparent that CO has a higher heat of desorption than H₂ and indeed, greater site concentration. Thus, CO will have a higher surface coverage than H₂ during FTS and most likely molecularly adsorbed. Moreover, the desorption peak for H₂ was located well within the temperature range for FTS (>453 K) suggesting that there would be little if any H₂ on the catalyst surface during the reaction. This implies that gas phase H₂ attack of the adsorbed CO molecule would be the preferred route for the formation of surface oxygenated species. In fact, St. Clair et al.³² have observed that H₂ did not chemisorb on Mo carbide in the presence of CO. Furthermore, alcohols have been produced over the Mo carbide catalyst confirming the presence of precursor surface oxygenated species. Even so, Vo and Adesina²¹ have developed several models derived from both enolic and carbide mechanisms in which either associatively or dissociatively adsorbed CO reacts with (molecularly or dissociatively chemisorbed) H₂ or in the gas phase for capturing the formation rate of individual hydrocarbons. They found that hydrocarbon synthesis over the Mo carbide system may be best represented as;

or

HCOH_X + H₂ → CH₃OH + X (for alcohol formation) (17)

Chain growth propagation step;

where R and R′ are alkyl groups with R possessing one carbon atom more than R′. X is the active site formed in situ by CO chemisorption on the surface of MoC_1−x catalyst. From eqn (13), (14) and (15), we obtain

θ = K₁P_COθ₀ (22)

θ_CO = K₂P_COθ (23)

θ_HCOH = K₃P_H2θ_CO (24)

where K₁ = k₁/k₋₁; K₂ = k₂/k₋₂; and K₃ = k₃/k₋₃. θ_i is the fractional surface coverage for species i whilst θ₀ is the initial fractional vacancy on the MoC_1−x and θ is the fractional active site formed in situ by coordinative bonding of CO to a vacant Mo site.

For the methane production, eqn (16) may be assumed as an irreversible rate-controlling step, thus;

which after the formal procedure yields;with K_a = K₁K₂, K_b = K₁K₂K₃, and k_rxn = k₄K_b. The experimental data were then fitted to the methanation model. The estimated adsorption equilibrium constants, K_i, and the methane reaction rate constant, k₄, are summarised in Table 10. As seen in the parity plot (cf.Fig. 13), there is a good correlation between experimental data at all temperatures and the proposed Eley–Rideal model.

Table 10 Computed model parameters for methanation reaction over 3%K–MoC_1−x/Al₂O₃ catalyst

Model parameters	453 K	463 K	473 K	BMV criteria
K 1 (atm^−1)	0.83	0.62	0.50
K 2 (atm^−1)	10.52	5.14	4.01
K 3 (atm^−1)	1.27	1.90	2.18
k 4 (×10^7 mol gcat^−1 s^−1 atm^−2)	10.9	26.5	32.6
ΔH1 (cal mol^−1)	−10 814.90			Satisfied
ΔS1 (cal mol^−1 K^−1)	−24.26
ΔH2 (cal mol^−1)	−20 579.50			Satisfied
ΔS2 (cal mol^−1 K^−1)	−40.90

---

Fig. 13 Parity plot for the methanation reaction model over 3%K–10%MoC_1−x/Al₂O₃ catalyst.

The thermodynamic consistency of parameter estimates was also further checked against the Boudart–Mears–Vannice (BMV) criteria³³ provided by eqn (27) and (28).

0 < −ΔS_i < ΔS_ig (27)

10 ≤ −ΔS_i ≤ 12.2 − 0.0014 ΔH_i (28)

where ΔS_ig is the entropy of CO in the gas phase, 47.3 (cal mol⁻¹ K⁻¹), whilst ΔH_i and ΔS_i are experimental enthalpy (cal mol⁻¹) and entropy changes (cal mol⁻¹ K⁻¹), respectively, estimated from;

As seen in Table 10, the computed ΔH_i and ΔS_i values satisfied the BMV criteria. The activation energy (cf.Fig. 14) for production of CH₄ and the HCOH building unit were also estimated as 97.81 and 48.26 (kJ mol⁻¹), respectively, with the associated pre-exponential factor, A, of 2.31 × 10⁵ and 4.86 × 10⁵ (mol g_cat⁻¹ s⁻¹).

Fig. 14 Estimated activation energy for formation of CH₄ and HCOH species.

From eqn (20) and (21), the production rate for olefin, r_Cnolefin, and paraffin, r_Cnparaffin, may be given as

r_Cnolefin = k_toP_H2θ_RCH2COH (30)

where k_to and k_tp are termination rate constants to olefin and paraffin respectively. Thus, the termination rate, r_t, for a hydrocarbon with chain length, n, may be obtained as;

r_t = r_Cn = r_Cnolefin + r_Cnparaffin = (k_to + k_tpP_H2)P_H2θ_RCH2COH (32)

Additionally, combining eqn (30) and (31), the olefin-to-paraffin ratio, ROP, may be derived as

As seen in Fig. 15, the olefin-to-paraffin ratio for each carbon number increased linearly with decreasing H₂ partial pressure in agreement with eqn (33).

Fig. 15 Effect of H₂ partial pressure on olefin-to-paraffin ratio for each hydrocarbon at 463 K.

Since HCOH species are the monomeric units for chain growth propagation, the relationship between θ_RCH2COH and θ_HCOH may be written as;

θ_RCH2COH = αⁿ⁻¹θ_HCOH (34)

Substituting eqn (33) and (34) into eqn (32), the reaction rate for individual hydrocarbon, r_Cn, may be given as

From eqn (25) and (35), the hydrocarbon synthesis rate for a given chain length, n, may be expressed as a function of olefin-to-paraffin ratio and chain growth probability;

r_Cn = (ROP + 1)αⁿ⁻¹k_oxyr_C1 (36)

where k_oxy = k_tpk₄⁻¹. Fig. 16 shows that the proposed reaction model gave a reasonable fit to the experimental data. Moreover, combining eqn (25), (30) and (34), the olefin reaction rate, r_Cnolefin, may be obtained as

Experimental data were fitted to eqn (37) to estimate k_tok₄⁻¹ at different reaction temperatures using Polymath 6.0 software for nonlinear regression. Since the activation energy for methane has been previously determined to be 97.8 (kJ mol⁻¹), the intrinsic activation energy values for termination to olefins, E_to, and paraffins, E_tp, may be computed individually as 141.4 and 166.2 (kJ mol⁻¹), respectively, with the corresponding pre-exponential factor of 9.01 × 10⁹ and 18.92 × 10¹⁰ (mol g_cat⁻¹ s⁻¹). Given that the value of E_to is less than E_tp, it is apparent that olefin is the primary hydrocarbon product for FTS over the Mo carbide catalyst. Furthermore, the lower value for CH₄ activation energy compared to olefins and paraffins explains why the higher production rate of CH₄ during FTS is due to kinetic control although thermodynamically higher hydrocarbons (olefins and paraffins) are more favourable.

Fig. 16 Estimated k_oxy parameter for eqn (36) over 3%K–10%MoC_1−x/Al₂O₃ catalyst.

Thus, by combining eqn (26) and (36), the FT reaction rate over the Mo carbide catalyst for individual hydrocarbon species with different feed compositions may be given as

This equation may therefore be used for the kinetic evaluation of any hydrocarbons during FTS.

Conclusions

The promotional effect of K-addition to an alumina-supported MoC_1−x catalyst during CO hydrogenation has been investigated. The Mo carbide catalyst was prepared via thermocarburization of MoO₃ using a 5H₂ : 1C₃H₈ feed mixture with heating rates of 5–20 K min⁻¹. Solid-state TGA indicated that the Mo oxide conversion proceeded via a 2-stage process involving the intermediate oxycarbide and a final carbide (MoC_1−x) phase. K-addition influenced both the physicochemical properties (BET area, acid–base character, H₂ and CO-TPD, and XRD measurements). FTS performance evaluation at various temperatures and feed composition (H₂ : CO ratio) revealed that the optimum effect for nearly all reaction metrics (activity and olefin selectivity) occurred at about 3 wt% K loading. Reaction rate data supported the proposition of an Eley–Rideal type mechanism in which the hydrocarbon rate-controlling step was the reaction between surface CO and gas phase H₂.

Acknowledgements

Thanks are due to Australian Research Council (ARC) for supporting the research programme in the Reactor Engineering & Technology Group, UNSW, through a Discovery Grant.

References

1. R. L. Levy and M. Boudart, Science, 1973, 181, 547–549.
2. G. S. Ranhotra, A. T. Bell and J. A. Reimer, J. Catal., 1987, 108, 40–49.
3. L. Zhao, F. Sotoodeh and K. J. Smith, Catal. Commun., 2010, 11, 391–395.
4. H. Chen and A. A. Adesina, Appl. Catal., A, 1994, 112, 87–103.
5. E. Furimsky, Appl. Catal., A, 2003, 240, 1–28.
6. T. Xiao, A. P. E. York, K. S. Coleman, J. B. Claridge, J. Sloan, J. Charnock and M. L. H. Green, J. Mater. Chem., 2001, 11, 3094–3098.
7. T. Xiao, A. P. E. York, V. C. Williams, H. Al-Megren, A. Hanif, X. Zhou and M. L. H. Green, Chem. Mater., 2000, 12, 3896–3905.
8. T. H. Nguyen, T. V. Nguyen, Y. J. Lee, T. Safinski and A. A. Adesina, Mater. Res. Bull., 2005, 40, 149–157.
9. JCPDS Powder Diffraction File, International Centre for Diffraction Data, Swarthmore, PA, 2000.
10. D.-V. N. Vo and A. A. Adesina, Fuel Process. Technol., 2011, 92, 1249–1260.
11. D.-V. N. Vo and A. A. Adesina, in Synthetic liquids production and refining, ACS Symposium Series, ed. A. de Klerk and D. L. King, American Chemical Society, 2011, ch. 7, pp. 155–184.
12. H. E. Kissinger, Anal. Chem., 1957, 29, 1702–1706.
13. G. C. Bond, Catal. Today, 1999, 49, 41–48.
14. J. Zhang, H. Xu and W. Li, Appl. Catal., A, 2005, 296, 257–267.
15. L. Liu and Q.-X. Guo, Chem. Rev., 2001, 101, 673–696.
16. F. Solymosi and L. Bugyi, Catal. Lett., 2000, 66, 227–230.
17. J. S. Lee, S. Kim and Y. G. Kim, Top. Catal., 1995, 2, 127–140.
18. M. Xiang, D. Li, H. Qi, W. Li, B. Zhong and Y. Sun, Fuel, 2007, 86, 1298–1303.
19. A. A. Adesina, R. R. Hudgins and P. L. Silveston, Can. J. Chem. Eng., 1986, 64, 447–454.
20. C. Phamhuu, M. J. Ledoux and J. Guille, J. Catal., 1993, 143, 249–261.
21. D.-V. N. Vo and A. A. Adesina, Appl. Catal., A, 2011, 399, 221–232.
22. R. B. Anderson, The Fischer–Tropsch Synthesis, Academic Press, New York, 1984.
23. H. Shou, D. Ferrari, D. G. Barton, C. W. Jones and R. J. Davis, ACS Catal., 2012, 2, 1408–1416.
24. A. Griboval-Constant, J.-M. Giraudon, G. Leclercq and L. Leclercq, Appl. Catal., A, 2004, 260, 35–45.
25. H. Schulz and M. Claeys, Appl. Catal., A, 1999, 186, 91–107.
26. E. V. Steen and H. Schulz, Appl. Catal., A, 1999, 186, 309–320.
27. C. G. Cooper, T. H. Nguyen, Y. J. Lee, K. M. Hardiman, T. Safinski, F. P. Lucien and A. A. Adesina, Catal. Today, 2008, 131, 255–261.
28. H. Schulz and M. Claeys, Appl. Catal., A, 1999, 186, 71–90.
29. R. A. Dictor and A. T. Bell, J. Catal., 1986, 97, 121–136.
30. G. P. Van der Laan and A. A. C. M. Beenackers, Catal. Rev. Sci. Eng., 1999, 41, 255–318.
31. D.-V. N. Vo, T.-H. Nguyen, E. M. Kennedy, B. Z. Dlugogorski and A. A. Adesina, Catal. Today, 2011, 175, 450–459.
32. T. P. St. Clair, S. T. Oyama and D. F. Cox, Surf. Sci., 2000, 468, 62–76.
33. M. Boudart, D. E. Mears and M. A. Vannice, Ind. Chim. Belge., 1967, 32, 281.

---