AbstractLow‐field nuclear magnetic resonance technique combined with chemometrics was employed to establish the quantitative models for rapidly monitoring the changes of indicators (hydroxyl value (OHV), epoxy value (EV), polymeric hydroxyl value (POHV, reflecting the degree of polyetherification)) during the ring opening of epoxidized soybean oils. Comprehensively considering the variations of peak retention times and peak areas of LF‐NMR spectra, OHVs of vegetable oils‐based polyols were divided into low OHV (LOHV, 1.7–91.9 mg KOH/g) group and high OHV (HOHV, 94.1–229.4 mg KOH/g) group. In both LOHV and HOHV groups, OHV presented good linearity with T2w, S23, and STotal of relaxation property with correlation coefficient of >0.90. EV and POHV also presented good linearity with relaxation property in both LOHV and HOHV groups. In LOHV group, data segment 1–1000 ms was taken with Deresolve, no preprocessing, and Deresolve as preprocessing methods for OHV, EV, and POHV, respectively, and partial least square (PLS) as the modeling method for all the three indicators, where RMSEPs of OHV, EV, and POHV models were 4.282, 0.114, and 4.061, respectively. In HOHV group, data segment 1–1000 ms was taken with no preprocessing as the preprocessing method and PLS as the modeling method for all the three indicators, where RMSEPs of OHV, EV, and POHV models were 3.117, 0.088, and 4.964, respectively. The established method will benefit polyol industry to online adjust the production parameters according to the monitoring results and achieve high‐quality of vegetable oils‐based polyols.
Read full abstract