Simultaneous optimization of preparation conditions and composition of the methanol synthesis catalyst by an all-encompassing calculation on an artificial neural network

A combinatorial approach comprised of a high-pressure high-throughput screening (HTS) reactor, an artificial neural network (ANN), and an all-encompassing calculation was applied to the development of catalyst for methanol synthesis at 1 MPa. Because the optimum catalyst composition is usually dependent on both preparation parameters and reaction conditions, the composition of the catalyst (Cu-Zn-Al-Sc-B-Zr), calcination temperature (300-360 degreesC), and amount of precipitant (1.0-2.5 times that equivalent to total cations) were optimized simultaneously. In the HTS reactor using 96-well microplates, activities of 190 catalysts with random composition, prepared using the random amount of precipitant and calcined at random temperature, were measured under pressure (1 MPa). The results with an additional 44 datasets were used for the training of ANN. After training, the ANN can map the catalyst activities as a function of the catalyst composition, amount of precipitant, and calcination temperature. An all-encompassing calculation of the 2.6 million activities of all possible combinations of the parameters was conducted to find that the space-time yield Of Cu0.43Zn0.17Al0.23SC0.11B0.00Zr0.06O1.22 precipitated by 2.2 equiv of oxalic acid and calcined at 334 degreesC was the global optimum activity.