Forensics relies on the differentiation and classification of document papers, particularly in cases involving document forgery and fraud. In this study, document papers are classified by integrating Raman spectroscopy with machine learning models, namely, random forest (RF), support vector machines (SVMs), and feed-forward neural networks (FNNs). Among the machine learning models, the RF model effectively calculated the feature importance and identified the critical spectral region contributing to classification, enhancing the transparency and interpretability of the result. Spectral preprocessing with the first derivative significantly improved the classification performance. The spectral range 200–1650 cm−1 was identified as a highly informative region for differentiation, reducing the number of input variables from 756 to 360 while enhancing the model accuracy. The FNN model outperformed the RF and SVM models, with an F1 score of 0.968. The results underscore the potential of combining Raman spectroscopy with machine learning for forensic document examination, offering an interpretable, computationally efficient, and robust approach for paper classification.
In recent decades, lignocellulosic biomass has attracted increasing attention as a sustainable alternative to fossil-fuel-based resources. However, the compact and highly crystalline structure of cellulose remains a major limitation to its effective utilization. In this study, the allomorphic transformation of cellulose was induced through chemical treatments using sodium hydroxide (NaOH) and ethylenediamine (EDA), enabling the conversion of native cellulose I into cellulose II and cellulose III, respectively. The resulting changes in the crystalline structure were systematically investigated using X-ray diffraction and Raman spectroscopy. Both NaOH- and EDA-treated celluloses exhibited enhanced enzymatic digestibility compared to untreated cellulose, consistent with the observed modifications in the crystal structure. Nevertheless, some results indicate that crystalline structure is not an absolute determining factor, but rather one of several parameters, including specific surface area, particle size, and degree of polymerization.
Weave and drainage marks on paper surfaces derived from forming fabrics installed in paper machines were examined. Non-destructive differentiation of copy papers from various manufacturers was performed by a 2D Lab Formation Sensor. Individual characteristics such as angles and wavelengths formed by the fabric were selected as parameters for the analysis. These parameters were sampled from a total of 500 papers: 50 sheets of paper from each of the 10 groups by copy paper manufacturers. Using machine learning algorithms such as partial least squares discriminant analysis (PLS-DA), k-nearest neighbors (k-NN), and naive bayes (NB) models to analyze these parameters allows for the identification of copy paper groups by manufacturers. The NB classifier demonstrated superior performance, achieving an accuracy of 0.973 even with a 50 % reduction in input variables. This study shows that periodic marks on paper surfaces can act as important indicators of the forming fabric and the distinct features of a paper-making machine. In general, this paper proposes that the 2D Lab Formation Sensor along with machine learning models can be utilized for non-destructive copy paper categorization. This identification method could be broadened to determine information regarding suspicious documents for forensic identification purposes.
Lignin-based flame retardants are characterized by their environmental friendliness; however, they have several drawbacks associated with their reactivity and processability. A comprehensive understanding of the effects of structural variations induced by different extraction methods on reactivity is essential. In this study, the structure of lignin was analyzed, and phosphorylation was subsequently performed to enhance its flame-retardant properties and spinnability. The hydroxyl group distribution in lignin was quantitively determined via 31P nuclear magnetic resonance spectroscopy, providing the basis for subsequent phosphorylation. Additionally, Fourier transform infrared spectroscopy, X-ray photoelectron spectroscopy, inductively coupled plasma atomic emission spectroscopy, and thermogravimetric analysis were used to characterize the chemical structure and thermal properties. After phosphorylation, phosphorus-containing functional groups were incorporated into the lignin structure, altering the thermal decomposition behavior. The char residue increased from 26.5% for untreated lignin (L) to 53.3% in the lignin-based flame retardants. Moreover, the solution properties of lignin were altered, resulting in uniform and smooth electrospun fibers. The application of the as-developed flame retardants via electrospinning onto fully combustible polymer system, which undergoes complete combustion above a certain temperature such as PHA and PLA could successfully impart thermal stability and self-extinguishing properties. This study provides valuable insights into the design of lignin-based flame retardants.