A series of bismuth molybdate catalysts were compared using a continuous feed of isobutene in air or a cyclic/pulse or gas–gas periodic flow mode, in which the oxide is first treated with air at the reaction temperature, followed by reaction of isobutene in nitrogen or argon in the absence of oxygen. For nonpromoted catalysts, in the continuous-feed mode, relatively low selectivities for methacrolein are observed and significant amounts of byproduct CO 2 are formed. In contrast, in the pulse mode, no CO 2 and almost 100% selectivity to methacrolein can be observed with nonpromoted BiMoO x , especially for the initial reaction period. At realistic reaction temperatures, no carbon or carbon-containing product depositions are observed during the pulse mode, indicating that the very high selectivity observed is realistic for the nonpromoted catalyst under pulse conditions. The effect of Co and Fe as promoters on the catalytic performance was also investigated. The addition of Co and Fe leads to the observation of CO 2 also in the pulse mode along with some carbon deposition, although the selectivity to methacrolein remains significantly higher than that observed in the continuous-feed mode. Investigation of a more complex promoted catalyst representative of a commercial formulation [BiMo 12Fe 2NiCo 7MgSb 0.9Ti 0.1Te 0.02Cs 0.4O x ] also shows that CO 2 is observed in the pulse mode but, at much lower levels than observed with the less complex catalysts. However, in subsequent operation in the gas–gas periodic flow, the catalytic performance for this catalyst is almost identical in the continuous and pulse operations. It is concluded that, for pulse-mode operation, the best results, including almost total selectivity to methacrolein, are observed with the relatively simple nonpromoted binary oxides. In general, the first pulse-mode operation of these catalysts gives the highest yield of methacrolein and subsequent oxidation/reaction cycles gives lower yields. The origin of this effect was investigated using cyclic TPO/TPR and TPR/TPO analysis, revealing that the first TPR/TPO cycles are significantly different than subsequent cycles, likely due to the initial state of the catalyst surface.
Read full abstract