Exploring alternative plant-based “milks” – a chemometric investigation into the composition and properties of milks made from plant-based raw materials

Abstract

Plant-based milks are an increasingly important product due to many reasons such as consumer allergies to protein present in cow’s milks and both ethical and environmental concerns around the consumption of mammalian milks. While there has been a huge recent demand for plant-based milks, there is little available research on these plant-based milks made from various plant-based raw materials. In addition, some of the plant-based milks available in the market do not satisfy what consumer needs as they many do not meet the desired quality or nutritional values. The objective of this research was to make a range of alternative plant-based milks, many of which have never been made or studied before, from a range of raw materials, representing the five categories of cereal, pseudo-cereal, nuts, seeds and legumes. The properties of the resulting milks were then to be analysed, for them to be compared. Additionally, the effect of different experimental soaking conditions and their effect on these properties were also to be explored. In total, milks made from thirteen raw materials (rye, rice, soy, cowpea, black bean, hazelnut, cashew, pistachio, watermelon seed, pumpkin seed, flaxseed, buckwheat and black quinoa) using nine experimental soaking conditions (20 °C-2 hours, 20 °C-4 hours, 20 °C-8 hours, 35 °C-2 hours, 35 °C-4 hours, 35 °C-8 hours, 50 °C-2 hours, 50 °C-4 hours and 50 °C-8 hours) were made at a homebased scale in triplicate using a method developed in this work (a total of 351 milks). Once the milks were made, a series of physicochemical analyses (total soluble solids, colour and pH) and chemical analyses were performed (FT-IR and NMR) to compare differences in the properties of the resulting milks for the different types of milk and also the effect of soaking conditions. A series of measurement methods were developed to analyse all these properties of the plant-based milks and were successfully applied to all samples. Once the measurements of these properties was performed, that data was analysed using RStudio. This analysis was primarily focussed towards finding out the differences between these milks compared and their relationship to the raw materials. It was found that the total soluble solids (°Brix) of the milks all ranged between 0.3 and 7.7 with statistically significant differences between all milks and a moderate positive correlation between the protein in the raw material and total soluble solids of the milks. Furthermore, based on the statistical analysis, it was shown that the colour of the milks were very distinct from each other. Two methods were developed for colour analysis and while the more-traditional CIE method was a better tool to distinguish the colour of different milks, the alternative photographic method was found to be appropriate if the former could not be conducted. Moreover, it was found that the pH of the milks all ranged between 6.22 and 6.70 although the differences between the milks were statistically significant. Lastly, the statistical analysis showed valuable outcomes in analysing macronutrients in the different milks, using FT-IR and NMR measurements in combination with data analysis techniques, including pre-processing, in-depth PCA analysis and the use of random forest models to select the most important variables to differentiate between milk types and soaking conditions. Overall, all physicochemical analysis methodologies used in the research were non-destructive, and could be applied during the manufacture of plant-based milks in a commercial or research setting and would be particularly useful in industry to identify the ideal conditions for process design. Additionally, the chemical-composition analysis methods in combination with the statistical analysis pathways used herein would be able to be used to understand how these conditions impact on the macronutrients, which can be used to refine or optimise the designed processes.