Study of Atmospheric Degradation in a Semi-Arid Climate of Pure Poly(butylene adipate-co-terephthalate) Films and those Added with Orange Oil Using Conventional and Machine Learning Methods

In order to use new polymeric materials, which are less harmful to the environment, studies on natural degradation are necessary. Thus, the natural atmospheric degradation was studied in a semi-arid climate of pure poly (butylene adipate-co-terephthalate) (PBAT) films and those with 5% orange essential oil (OEL), using conventional methods (loss mass, thermal degradation, optical microscopy, carbonyl index and visual analysis) and machine learning (principal component analysis (PCA) and partial least squares regression (PLS-R)). 78 samples were exposed for 180 days. The materials showed signs of fragmentation/mass loss of 21.98% for 105 days and 14.30% for 120 days, respectively for pure PBAT and PBAT/OEL. A reduction in thermal degradation temperatures and carbonyl index was observed. PLS-R allowed evaluating the correlation between spectral bands and exposure time, obtaining coefficient of determination (R2 ) = 0.941, and a root means square error for cross-validation (RMSECV) of 8.10% for the full range and R2 = 0.950 and a RMSECV of 7.45% for the range of 600-1400 cm-1. Regarding the climate conditions, the maximum-minimum temperature, maximum-minimum humidity and global solar radiation were 38.1-18.1 ºC, 79.1-34.3%, 22.03 MJ m-2 day-1, respectively. Therefore, the samples were affected similarly and machine learning could identify a pattern between the polymer natural atmospheric degradation and the structural chemical changes in films.