The grape juice used for fermentation contains glucose and fructose. Under the action of yeast, glucose and fructose are converted into ethanol, which ultimately forms wine. Therefore, during the fermentation process, the main components of the fermentation broth are glucose, fructose, and ethanol.

In addition, red wine fermentation broth contains pigments, which have strong fluorescence under short wavelength laser irradiation, and the fluorescence will drown out Raman signals. Therefore, red wine testing requires the use of a 1064nm long wavelength laser. The left figure a shows the Raman spectrum of the red wine fermentation mixture under 1064nm excitation. The Raman characteristic peaks of glucose are located at 450 cm ⁻¹,520 cm ⁻¹ and 1063 cm ⁻¹, with the first two peaks being stronger. The Raman characteristic peaks of fructose are mainly located at 629 cm ^-1,710 cm ⁻¹, etc. The most typical Raman characteristic peak of ethanol is located at 880 cm⁻¹.

The characteristic peaks corresponding to these components increase with the increase of this component, so real-time detection of the Raman signal intensity of the fermentation broth can reflect changes in the composition of the fermentation broth. By observing the changes in Raman signals, only qualitative observations can be made of the increase or decrease in composition. In the actual product development process, quantitative analysis of the specific concentrations of each component is crucial for process control. To achieve quantitative analysis, it is necessary to establish a calibration model for the concentration of each component corresponding to the Raman intensity of the sample in advance. We configured a series of samples (more than 20) using commercial red wine as the background liquid and referring to the concentration of glucose and fructose during the fermentation process of red wine, tested their Raman signals, and established a calibration model shown in Figure b on the left using statistical methods (this calibration model has been proven to have an accuracy of up to 99% by predicting a small number of standard samples). Using this calibration model, we predicted the concentrations of glucose and fructose in samples during the actual fermentation process of red wine, and compared the glucose concentration with the concentration detected using biosensors, and the results were highly consistent. Due to the use of commercial red wine as the background solution, which already contains ethanol, this experimental protocol did not conduct ethanol testing. In addition, biosensors lack corresponding enzymes to test fructose. The next step of Raman fructose testing results will be compared with chromatographic results, which also reflects the advantages of Raman technology. A Raman probe can analyze multiple components simultaneously.