Machine learning-assisted non-destructive plasticizer identification and quantification in historical PVC objects based on IR spectroscopy

Non-destructive detection and identification of plasticizers in PVC objects by means of machine learning-assisted Raman spectroscopy

Vibrational spectroscopic techniques, such as Raman spectroscopy, as a non-destructive method combined with machine learning (ML), were successfully tested as a quick method of plasticizer identification in poly(vinyl chloride) - PVC objects in heritage collection. ML algorithms such as Convolutional Neural Network (CNN), Random Forest (RF), Support Vector Machines (SVM), and Linear Discriminant Analysis (LDA) were applied to the classification and identification of the most common plasticizers used in the case of PVC. The CNN model was able to successfully classify the five plasticizers under study from their Raman spectra with a high accuracy of (98%), whereas the highest accuracy (100%) was observed with the RF algorithm. The finding opens doors for the development of robust and economical tools for conservators and museum professionals for fast identification of materials in heritage collections.

Development of a Measurement System Using Infrared Spectroscopy-Attenuated Total Reflectance, Principal Component Analysis and Artificial Intelligence for the Safe Quantification of the Nucleating Agent Sorbitol in Food Packaging

Sorbitol derivatives and other additives are commonly used in various products, such as packaging or food packaging, to improve their mechanical, physical, and optical properties. To accurately and precisely evaluate the efficacy of adding sorbitol-type nucleating agents to these articles, their quantitative determination is essential. This study systematically investigated the quantification of sorbitol-type nucleating agents in food packaging made from impact copolymers of polypropylene (PP) and polyethylene (PE) using attenuated total reflectance infrared spectroscopy (ATR-FTIR) together with analysis of principal components (PCA) and machine learning algorithms. The absorption spectra revealed characteristic bands corresponding to the C-O-C bond and hydroxyl groups attached to the cyclohexane ring of the molecular structure of sorbitol, providing crucial information for identifying and quantifying sorbitol derivatives. PCA analysis showed that with the selected FTIR spectrum range and only the first two components, 99.5% of the variance could be explained. The resulting score plot showed a clear pattern distinguishing different concentrations of the nucleating agent, affirming the predictability of concentrations based on an impact copolymer. The study then employed machine learning algorithms (NN, SVR) to establish prediction models, evaluating their quality using metrics such as RMSE, R2, and RMSECV. Hyperparameter optimization was performed, and SVR showed superior performance, achieving near-perfect predictions (R2 = 0.9999) with an RMSE of 0.100 for both calibration and prediction. The chosen SVR model features two hidden layers with 15 neurons each and uses the Adam algorithm, balanced precision, and computational efficiency. The innovative ATR-FTIR coupled SVR model presented a novel and rapid approach to accurately quantify sorbitol-type nucleating agents in polymer production processes for polymer research and in the analysis of nucleating agent derivatives. The analytical performance of this method surpassed traditional methods (PCR, NN).

Unpacking Phthalates from Obscurity in the Environment

Phthalates (PAEs) are a group of synthetic esters of phthalic acid compounds mostly used as plasticizers in plastic materials but are widely applied in most industries and products. As plasticizers in plastic materials, they are not chemically bound to the polymeric matrix and easily leach out. Logically, PAEs should be prevalent in the environment, but their prevalence, transport, fate, and effects have been largely unknown until recently. This has been attributed, inter alia, to a lack of standardized analytical procedures for identifying them in complex matrices. Nevertheless, current advancements in analytical techniques facilitate the understanding of PAEs in the environment. It is now known that they can potentially impact ecological and human health adversely, leading to their categorization as endocrine-disrupting chemicals, carcinogenic, and liver- and kidney-failure-causing agents, which has landed them among contaminants of emerging concern (CECs). Thus, this review article reports and discusses the developments and advancements in PAEs' standard analytical methods, facilitating their emergence from obscurity. It further explores the opportunities, challenges, and limits of their advancements.

Towards the Understanding of the Aging Behavior of p-PVC in Close Contact with Minced Meat in the Artwork POEMETRIE by Dieter Roth

This paper presents scientific investigations into the materiality and aging behavior of a copy of Dieter Roth's multiple POEMETRIE (1968), mainly made of p-PVC components and minced meat, with the aim of informing conservation-restoration strategies. The main issues were represented by plasticizer migration, fat diffusion, and leakage, which led to the formation of a sticky surface layer. Replicas of p-PVC without minced meat were prepared and artificially thermally aged, while several techniques were used to investigate both the artwork and the replicas in terms of materials and degradation state. These include UV/Vis imaging, pH measurements, FTIR-ATR, and Py-GC/MS. In addition to showing that p-PVC-based materials composed of slightly different plasticizers were affected by similar degradation pathways (i.e., plasticizer migration, yellowing, etc.), this study reports that fat components were also shown to be unstable, resulting in migration/leakage in different directions, where their degradation amplified that of the p-PVC bags. This work represents a first study of plasticizer migration and fat diffusion in the art and conservation context. Also, an ammine-wax type of lubricant was identified in the most recent p-PVC formulations as the replicas selected for this study, thus providing an important source of information in different polymer-based research areas.

Near-Infrared Spectroscopy and Machine Learning for Accurate Dating of Historical Books

Non-destructive, fast, and accurate methods of dating are highly desirable for many heritage objects. Here, we present and critically evaluate the use of near-infrared (NIR) spectroscopic data combined with three supervised machine learning methods to predict the publication year of paper books dated between 1851 and 2000. These methods provide different accuracies; however, we demonstrate that the underlying processes refer to common spectral features. Regardless of the machine learning method used, the most informative wavelength ranges can be associated with C-H and O-H stretching first overtone, typical of the cellulose structure, and N-H stretching first overtone from amide/protein structures. We find that the expected influence of degradation on the accuracy of prediction is not meaningful. The variance-bias decomposition of the reducible error reveals some differences among the three machine learning methods. Our results show that two out of the three methods allow predictions of publication dates in the period 1851-2000 from NIR spectroscopic data with an unprecedented accuracy of up to 2 years, better than any other non-destructive method applied to a real heritage collection.

Attenuated total reflection FTIR dataset for identification of type 2 diabetes using saliva

Diabetes is one of the top 5 non-communicable diseases that occur worldwide according to the World Health Organization. Despite not being a fatal disease, a late diagnosis as well as poor control can cause a fatal outcome, because of that, several studies have been carried out with the aim of proposing additional techniques to the gold standard to assist in the diagnosis and control of this disease in a non-invasive way. Considering the above, and in order to provide a solid starting point for future researches, we share a primary research dataset with 1040 saliva samples obtained by Fourier Transform Infrared Spectroscopy considering the Attenuated Total Reflectance method. Database include: gender, age, individuals (patients) with/without diabetes, the glucose value, and the result to the A1C test for the diabetic population. We believe that sharing dataset as is could increase experimentation, research, and analysis of spectra through different strategies broaden its range of applicability by chemists, doctors, physicists, computer scientists, among others, to identify the effects that the virus causes in the body and to propose possible clinical treatments as well as to develop devices that allow us to assist in the characterization of possible carriers.