Wood is a complex, chemically heterogeneous material. Its components can be divided in two groups: structural components (cellulose, hemicelluloses, lignin) and non-structural ones (extractives and inorganic compounds). It is known that during delignification both content of components and their structure can change. Development of new technology for production of half-finished fiber materials needs new knowledge about transformation of basic wood components in cooking processes.

At studying structure of the components two problems occur. The one is searching isolation methods which do not change strongly structure of component. The second problem is connected with selecting experimental method of analysis. A number of chemical and physical methods used for lignin structure characterization are destructive.

We develop approach for studying variation of wood components in the process of kraft cooking of different wood species without their extraction from fibers by using FTIR-spectroscopy. When cellulose, lignins, hemicelluloses are studied in situ by the spectral method there is a risk of interference from wood constituents. The structural complexity of major wood polymers and their possible transformation contribute to the uncertainty of spectroscopic data. To decide this problem different mathematical tools are used.

In this report the factor analysis was applied to study the spectra of four wood species and kraft pulps from soft- (spruce and fir) and hardwood (birch and aspen) obtained at different degrees of delignification (Klason lignin: 2–30%) by laboratory conventional kraft cooks. Each sample was characterized by lignin content and yield value. These two characteristics are important parameters for this processing. After carrying out some treatment of IR-spectra the calibration models for definition of Klason lignin and yield value in investigated woods and pulps with using PLS1 method for each series and for all samples have been constructed. Calibrations were built by using all the spectral data or selected regions. For lignin content, the calibrations present good performances in terms of R² (> 0.96). The yield parameter (range of 45–100 %) was more difficult to predict.

The factor analysis allowed to reveal the number of principal spectra, see correlations of bands and do some interpretations. For each group of pulps from four wood species there was evaluated a number of factors to describe experimental spectra with good precision. The main factors were critically analyzed by comparing with spectra of "pure" wood components: isolated cellulose, hemicellulose and hard- and softwood lignins.

Thus, the use of chemometrics allowed to create calibration models for a fast estimation of the important chemical and economical parameters, and also to reveal the some information to interpret spectra of different origin pulps.