ABSTRACT

Softness of cotton nonwoven for cosmetics is a key character to attract consumers. Evaluation of the cosmetic nonwoven softness is always subjective, that is, relying on human hand assessment. Development of an instrumental method for grading the cosmetic nonwoven softness is, therefore, helpful for both manufacturers and customers to avoid subjective bias and communicate with each other using numerical data. This work proposes an approach to characterizing cosmetic nonwoven softness using a set of mechanical properties. Four types of cosmetic cotton nonwoven were selected for study. The mechanical properties of these nonwovens in terms of extension, shear, and surface friction were measured using a QTest tensile tester equipped with specially designed attachments. Obtained mechanical data were input into a neural network model for softness grading. This neural network model was established on a training data set composed of a typical plain cotton shirting fabric with and without softening using the Downy® softener. The fabric softness with the Downy® treatment was defined as "1" and the fabric softness without the Downy® treatment was defined as "0." This neural network model graded softness of the four cosmetic cotton nonwovon samples between 0 (indicating the worst softness) and 1 (indicating the best softness).