Effect of Pretreatment Atmosphere on the Catalytic Performance of Cu-Fe Based Catalyst for Mixed Alcohol Synthesis from Syngas

The pretreatment process greatly influenced the textural properties of the pretreated catalysts regardless of reduction atmosphere. The loss in BET surface area is due to the sintering of crystallite particles during the reduction process. However, BET surface area of the catalysts pretreated in CO-involved atmospheres is slightly larger than that of one in H₂ and the average pore diameter shows a reverse trend, especially for pure CO. Such a change could be ascribed to the fact that the pretreatment in CO-involved atmospheres facilitated the formation of small carbide nodules with Fe₂O₃ domains and then led to a larger BET surface area than that in H₂. On the other hand, the decline in the average pore diameter over syngas or CO-treated catalysts must be due to the deposition of inactive carbon from the disproportionating reaction of CO.

Since the carbon content is only 1.6 wt.% provided that all Fe₂O₃ was transformed into ¦Ö-Fe₅C₂ (carbide detected in present catalyst),there should be some other carbonaceous species formed during the pretreatment process. This assumption is supported by the results of CHN analysis. It is proposed that depositing carbon content related to the partial pressure of CO in the CO-including atmosphere and pointed out that the rate of carbon deposition is directly proportional to the factor of P_CO/P_H2² under H₂/CO atmosphere over the commercial iron-based Fischer-Tropsch synthesis catalyst. Thus, carbon deposition can be declined by decreasing CO partial pressure, or increasing the H₂ partial pressure.

The XRD patterns of CuZn-FeMn catalyst significantly changed after the pretreatment. Over H₂-pretreated catalyst, the signals of ZnFe₂O₄ were replaced by some broad and ill-resolved peaks mainly attributed to Fe, ZnO, and Fe₃O₄. As for catalyst pretreated in syngas or CO, a new phase, ¦Ö-Fe₅C₂, emerges, suggesting that iron phase began to reduce-carburize under the present conditions. The Mössbauer spectrum over the fresh catalyst displays two doubles, being typical for the super-paramagnetic (spm) Fe³⁺ on the non-cubic sites with the crystallite diameters smaller than 13.5 nm. After pretreatment, the H₂-pretreated catalyst characterized as a sextet and two doubles, corresponding to the metal iron atom (35.6%), the spm Fe³⁺ ions (58.2%) and the spm Fe²⁺ ions (6.2%) with both crystallite diameters smaller than 13.5 nm, respectively. This is well consistent with the results of XRD. In the case of the catalyst in syngas, MES displays three sextets and two doubles, suggesting the existence of ¦Ö-Fe₅C₂ (I, II, III) (29.8%), Fe³⁺ (65.8%) and Fe²⁺ (4.4%). MES of CO-pretreated catalyst shows the similar results to that of one in syngas with nearly equal percentage of ¦Ö-Fe₅C₂ (30.1% vs. 29.8%), suggesting that the partial pressure of CO in pretreatment atmosphere had little influence on the reduction/carburization extent in the present study.

XPS spectra of the fresh catalyst are being characteristic of Cu²⁺ cation as expected. After exposure to the pretreatment conditions, the binding energy of Cu 2p_3/2 in H₂ and syngas atmospheres shifts to 932.4 and 932.8eV, respectively. Unfortunately, no visible signal can be observed over CO-pretreated catalyst due to the severe coke deposition. The fresh catalyst had a LMM line of 917.6 eV, characteristic of CuO species. After the pretreatment in H₂, the corresponding kinetic energy spectra of Cu LMM consisted of a main peak at 916.8 eV, indicating the partial reduction of Cu²⁺ cations into Cu⁺ in H₂ at 300^oC. In the case of syngas-pretreated sample, however, the kinetic energy of Cu phase is different from that of H₂-pretreated sample, being 918.4eV characteristic of metal Cu.In Fe 2p region, no clear position change of Fe 2p_3/2 peak appeared over all pretreated catalysts although ¦Á-Fe or ¦Ö-Fe₅C₂ were detected by XRD and MES. The clear discrepancies are derived from its characteristic reduction behavior as explained by so-called "inner reduction model". According to this model, the reduced iron species diffuse to neighboring precipitation point and thus result in the vacancy of iron atom sites on the surface, which will be filled in by zinc and manganese atoms in the vicinity, both of which possess strong diffusion ability. It gives rise to the enrichment of zinc and manganese atoms on the surface, supporting by the surface atom ratio of Fe/Zn and Fe/Mn.

In present catalyst system, the specific activity to alcohols of catalyst pretreated with syngas and CO appeared superior to that of catalyst by H₂. As detected in MES study, the carburization of iron species already took place in syngas and CO atmospheres while partial iron was in the form of ¦Á-Fe after H₂-pretreatment. This indicated that the carburization of iron species in the pretreatment step was indispensable for the activation of CO although a little amount of iron carbide might also be formed in the initial period of CO hydrogenation over the H₂-pretreated catalyst. Moreover, we found that ¦Á value was unequal between alcohols and hydrocarbons upon all catalysts. More importantly, the dependence of ¦Á value on the pretreatment atmospheres was also different. In terms of hydrocarbons, the change trend in ¦Á value indicated that its chain growth was directly affected by the carbon content deposited. However, ¦Á parameter went through a maximum among catalysts for the production of alcohol. The lowest ¦Á value in H₂-pretreated catalyst suggested the formation of C₂⁺ alcohol is determined not only by the carbon chain growth ability of Fe but also by the cooperation between Cu and Fe sites. It is generally accepted that the formation of higher alcohol required the synergism between two active sites. Over one active site, a CO molecule was adsorbed and activated dissociatively. After a series of reactions, a surface adsorbed metal-alkyl species was subsequently formed, which would be selectively hydrogenated into hydrocarbon. At the same time, surface metal-acyl species were formed from the insertion of a surface associative CO into the metal-alkyl species. If the CO insertion was not favored, the surface adsorbed metal-alkyl species would be hydrogenated to hydrocarbon. So any negative factor affecting CO insertion would inhibit the formation of alcohol, leading to low C₂⁺OH selectivity. In the present case, Fe sites provided dissociative CO molecule while Cu sites adsorbed CO molecule associatively. Therefore, the synergistic effect between Fe and Cu sites was requisite for the formation of C₂⁺ alcohols. Higher ¦Á value of syngas-pretreated catalyst suggested the formation of small iron carbide nodules facilitated the closer contact between two sites. However, such intimate contact was strongly destroyed by the deposition of inactive carbon, especially in the domain of the interfacial area between two sites, leading to lower ¦Á value for CO-pretreated catalyst. The deposition of inactive carbon would block the path for the surface migration of the intermediate between two active sites, weakening the CO insertion and consequently the formation of C₂⁺ alcohols. Thus, Cu species should interact with Fe species as intensive as possible to create a bi-functional center for higher alcohol synthesis.