A Collection of More than 900 Gas Mixture Adsorption Experiments in Porous Materials from Literature Meta-Analysis

Information on mixture adsorption equilibrium is vital in developing adsorption-based separation processes. Because measuring mixture adsorption is more difficult than measuring single-component adsorption, far more data of the latter kind are available. Previous efforts to compile experimental mixture adsorption data for gases have given data sets with at most a few dozen examples. Here, we report the results of systematic literature meta-analysis that produced a data set of more than 900 gas mixture adsorption experiments. This collection includes data from 125 different binary mixtures including 60 different molecular species and information from 333 different adsorbents. We refer to this data set as the Binary adsorption ISOtherm ExperimeNtal 2020 (BISON-20) Database. Because the BISON-20 data set enormously expands the number and variety of experimental results for binary gas adsorption that are readily available, it will be a useful resource for future efforts in developing new materials or processes for gas separations. As initial applications of the BISON-20 data set, we show how identifying replicate measurements can be used to assess the reliability of binary adsorption data, how the accuracy of Ideal Adsorbed Solution Theory (IAST) can be systematically tested using experimental data, and how trends in selectivity for gas separations across many materials can be examined. ■ INTRODUCTION Energy-efficient chemical separations are a critical step in many large-scale chemical processes. 1 Adsorption of gases or liquids into porous materials is an attractive mode of chemical separations that is already used commercially, for example, in air purification and hydrocarbon processing 2 and has been widely considered for CO 2 capture. 3 Enormous numbers of distinct porous adsorbents exist, 4,5 making comprehensive experimental screening of adsorbents for targeted separation applications challenging. This observation has driven efforts to use computational models to rapidly predict the adsorption equilibrium of diverse sets of chemical species in large libraries of porous materials. 6−8 Assessing the performance of an adsorbent for a chemical separation requires information on the adsorption of the relevant chemical mixtures. Despite the self-evident nature of this statement, information on single-component adsorption is far more prevalent in the scientific literature than mixture adsorption data. This situation is in large part because single-component experiments are far easier to perform than experiments with mixtures, but it is also widely thought that approximate mixing theories such as Ideal Adsorbed Solution Theory (IAST) 9,10 or heuristic performance metrics 11,12 can give useful insight using only single-component data. Multiple situations are known, however, where IAST is expected to be inaccurate, 10 and comparisons between simple performance metrics and detailed process models have pointed to the limitations of these metrics. 11 As a result, experimental data on binary or multicomponent adsorption equilibria have significant value. Several previous sources have compiled published mixture adsorption data. Wu and Sircar reviewed data of this kind in