ABSTRACT The catalytic conversion of biomass‐derived sugars to 5‐hydroxymethylfurfural (5‐HMF) over zeolites is challenging due to internal mass transfer limitations arising from their microporous framework. To overcome this chronic issue, we prepared a set of micro‐mesoporous zeolites (22 samples), whose properties were fully characterized in our previous study using XRD, XRF, N 2 Physisorption, IR with pyridine and tri‐tert‐butylpydirine, and 27 Al NMR. In this study, we aim to investigate how highly correlated zeolite properties, such as textural and acidic characteristics, influence fructose dehydration to 5‐HMF, and to identify the key properties that govern the catalyst efficiency for sugar conversion. Since simple correlations of catalytic activity with individual descriptors of texture and acidity could not be found, we employed a chemometric approach, by using a combination of principal component analysis (PCA) and multiple linear regression (MLR). PCA was used to visualize the relationship between properties and catalyst performance, and to reduce redundant features for the subsequent MLR model, which was developed to correlate selected catalyst properties with 5‐HMF yield. The model shows strong predictive performance, with randomly scattered residuals around zero. The property activity relationships reveal that preservation of the crystalline framework and enhancement of active sites accessibility are the key factors for improving catalytic performance for sugar conversion. However, this remains valid under the condition that excessive amorphization and related Lewis acidity are minimized. This study demonstrates that chemometrics is effective for analyzing property‐activity relationships in a set of zeolites when these relationships are complex and highly correlated.
Han et al. (Thu,) studied this question.