Effect of temperature, fermentation time and yeast type on beer fermentation metabolites studied by infrared spectroscopy and ANOVA simultaneous component analysis (ASCA)

The progress of fermentation is a key point in defining the conformance of beer to its defined specifications. However, beer fermentation is characterized by a large number of biochemical transformations, mainly related to the yeast’s activity - influenced by the cell physiology, the temperature and the fermentation time. In an attempt to find rapid and reliable methods for assessing the influencing parameters in the beer making process, IR combined with ASCA, is an emerging approach that will allow studying the effect of fundamental variables on beer fermentation metabolites. In that sense, the aim of this work is to check the feasibility of IR combined with ANOVA to better understand the behaviour of beer while is fermenting.
A three-factor factorial design was constructed to understand the variability in fermentation metabolites due to yeast strains (two types), temperature (three levels) and fermentation time points (six time points). The factorial design was run in two replicates, with a total of 12 fermentations.
Samples were collected in triplicate right after pitching and then at 22, 46, 70, 142 and 216 hours of fermentation. The samples were centrifuged (15 min, 3,000 g) and analyzed by IR (ABB Bomen FT-IR spectrometer) with an ATR liquid module in the range of 4000-600 cm−1 (resolution: 4 cm-1, scans: 16 for background and samples).
In this study ASCA is applied, where PCA models of the individual effect matrices in equation 1 (Xk) are used to approximate the information in the matrices.
X=1m^T+X_time+X_yeast+X_temperatue+X_((ab))+X_((ac) )+X_((bc))+X_((abc) )+E (1)
A permutation test on the SSQk of the effect matrices has been used for testing significance of experimental factors (equation2).
〖SSQ〗_k= ‖X_k ‖^2=∑_(n=1)^N▒∑_(g=1)^G▒(〖(X_k)〗_ng )^2 ,k∈{α,β,γ,αβ,αγ,βγ,αβγ} (2)
The ASCA models presented are calculated on two separate regions. From the ASCA scores, in the region 2900-2250 cm-1, when combining the significant effect matrices of factors time (Xb, p < 0.0001) and interaction time*yeast (X(bc), p = 0.049), it is clear that the average time levels differ and that the development is different between yeast types (time*yeast interaction). The related ASCA loadings - subjected to cumulative summation to convert from first derivative to original spectral shape - clearly shows that CO2 absorbance is dominating the first ASCA component. ASCA scores, in the region 1500-980 cm-1, when combining the significant effect matrices of factors time (Xb, p < 0.0001) and interaction time*yeast (X(bc), p = 0.042) clearly show that the average time levels differ and that the development is different between yeast types (time*yeast interaction). Comparison of loadings with the pure spectra verifies that changes are related to maltose being consumed and ethanol being produced along the process.
The results of permutation testing of ASCA in variable intervals show that the fermentation time has a significant effect in all intervals, whereas other factors (time*yeast, yeast, temperature) show significant effects in smaller variable regions.
In conclusion, this work demonstrates the applicability of the combination of IR spectroscopy and ASCA to assess the influencing parameters in beer fermentation, allowing a rapid and accurate test for parameter control in beer manufacturing.