Rapid determination method of dissolving pulp properties by spectroscopic data and chemometrics

The present study was attempted to develop a suitable method for determining dissolving pulp properties such as pentosan, α-cellulose, R10, R18, viscosity, and brightness. In this study, properties are quantified first with wet chemical methods. Then, spectroscopic data of the samples were collected after running them through UV spectrophotomer and Fourier transformed-near infrared (FT-NIR) spectrophotometer. Spectroscopic data from both the instruments were pretreated with normalization, baseline correction, smoothing, Standard Normal Variate (SNV), Savitzky-Golay (S-G) smoothing with their first and second derivatives, and their combinations. Predictive models were calibrated with raw data and pretreated data from both the instruments. Results show that PLSR produce better predictive efficiency than PCR both with UV and FT-NIR data of their raw and pretreated form. In dissolving pulp, α-cellulose and R10 could be quantified with UV data treated by smoothing with moving average and S-G (first derivative) (R2 ≈ 99%) and baseline correction and smoothing with moving average (R2 ≈ 96%). Brightness could be determined with PLSR with FT-NIR raw data (R2 ≈ 92%). For R18 and viscosity, PLSR could be used FT-NIR data treated by baseline correction and smoothing with moving average (R2 ≈ 75%) and baseline correction and normalization (R2 ≈ 68%). The proposed chemometric method with pretreated spectroscopic data is a simple, rapid, and cost effective technique to determine properties of dissolving pulp.