Along with coffee, cola, energy drinks, and tea are popular beverages in the market that come in a variety of flavors. One big reason why they are so popular is because of their physiological and psychoactive properties, mostly from caffeine content. In the case of cola and energy drinks, sugar or some form of artificial sweetener also provides a stimulating effect. Companies closely safeguard their recipes for manufacturing cola and energy drinks as well as their testing procedures for quality control. Tea is the second most consumed beverage in the world after coffee. It can be categorized into two types: black tea which accounts for 80% of world production and green tea which accounts for 20% of world production. Different fermentation processes produce more than three hundred types of tea worldwide. These can be classified into six main families based on the manufacturing process: full fermented black tea, non-fermented green tea, slightly fermented white tea, semi-fermented oolong tea, dark (red) tea, and post yellow tea. The main chemical constituents of tea are amino acids, polysaccharides, polyphenols, alkaloids, organic acids, volatile compounds, and proteins. Theaflavins are also an important component that contribute to the antioxidant effect in black tea. There are different quality parameters that need to be tested in these beverages, many relating to sugar, acidity, and alkaloid (such as caffeine) parameters. Current methods for testing these parameters such as HPLC are expensive, laborious, and time-consuming, especially when implemented in a process setting. There is a need for fast, cost-effective, and real-time monitoring of parameters at all stages of the cola, energy drinks, and tea manufacturing process. One such method that has been examined is NIR spectroscopy.
- Taurine, Arginine, or Neither Energy Drink Classification
- Amino Acids
- Water Extract
Summary of Published Papers, Articles, and Reference Materials
Measurement of chemical parameters in energy drinks and tea has been studied using NIR spectroscopy and the results of most studies have demonstrated the potential of using NIR as a replacement for expensive and time-consuming wet chemistry methods. It must be noted that due to the strict guarding of recipes and quality control testing procedures by soft drink companies, there are few published studies using NIR spectroscopy to measure parameters in cola. However, many of the important quality parameters in soft drinks have been measured in other drinks. Two examples of this include Soluble Solids Content and pH. Energy drinks are manufactured in a manner similar to cola and are known for their stimulating effect which mostly comes from the high caffeine and sugar (or artificial sweetener content). One study examined classifying energy drinks based on taurine, arginine or containing neither using NIR spectroscopy as well as quantifying sugar and caffeine content. Classification results showed over a 95% correct prediction rate when classifying the three groups. Correlation was high between caffeine reference values and prediction values using validation NIR spectra. In the case of sugar, two sets of reference values were used: the Schoorl method and nominal values provided on the sample containers. Results were better using the nominal values and this likely occurred because multiple tests on the same sample using the Schoorl method showed a high standard deviation, indicating a large reference error. However, the RMSEP for the caffeine model is lower than the threshold of detection for this parameter using NIR spectroscopy and more validation work would be necessary to prove the feasibility of this model. Tea is the second highest consumed beverage in the world after water and two studies measured various parameters in tea. The first study used commercial samples of tea soft drink to measure SSC using reference values and NIR spectra. Calibration models were created using both the full wavelength range and selective wavelengths, both showing good results and proving the feasibility of the measurement. The second study used black tea powdered samples to measure various parameters important for quality control: amino acids, caffeine, theaflavins, and water extract. 4 different modeling algorithms were used for the calibrations. Correlation coefficients were high and prediction results showed low error, indicating that these models can be used to measure these quality control parameters in black tea using NIR spectroscopy.
Scientific References and Statistics
Quantitative Determination and Classification of Energy Drinks Using Near-Infrared Spectroscopy – Racz, Heberger, Fodor, Analytical and Bioanalytical Chemistry, 2016, 408:6403-6411
Energy drinks are one of the most popular functional beverages among commercially available soft drinks. They are known for high caffeine concentration and stimulating properties and are marketed based on distinctive color, flavor, and unique appearances. Energy drinks can also carry dangerous side effects due to high caffeine and sugar intake. Thus, it is important to analyze these parameters in energy drinks. There are also other components, such as taurine and arginine, which are important and are strictly regulated in some countries. Current reference methods for determining these components are time-consuming and expensive. NIR spectroscopy was examined as a method for classifying energy drinks based on either the presence of taurine, arginine, or neither as well as quantitative determination of caffeine and sugar. Ninety-one commercial energy drinks were procured for the study. Some of the drinks were mixed together to cover the examined concentration ranges for caffeine and sugar as uniformly as possible. FT-NIR spectra were scanned in transmission mode from 12500 cm-1 to 4000 cm-1. Spectral resolution was 8 cm-1 and thirty-two scans were averaged for each individual spectrum. This process was repeated three times for each sample and the three spectra were averaged to create one spectrum for each sample. Reference tests were performed on the samples for caffeine using HPLC-UV and for sugar using the Schoorl method. Principle Component Analysis (PCA) and Linear Discriminant Analysis (LDA) were applied to the spectra for classification analysis. Partial Least Squares (PLS) regression models were created using the NIR spectra and reference values for caffeine and sugar. In the case of sugar, both the HPLC-UV reference values and nominal values provided on the can were used to create regression models for sugar.
|Correct Classification for Arginine, Taurine, or Neither||95.7%|
|Caffeine||R2= 96.63||RMSEP= 13.4 ppm|
|Sugar (Measured Schoorl Concentration)||R2= 94.25||RMSEP= 1.13 g/100 ml|
|Sugar (Nominal Concentration Shown on Drink Containers)||R2= 99.75||RMSEP= 0.29 g/100 ml|
For classification, PCA was conducted first and the score values from the analysis were used as a basis for LDA. Validation results from the LDA showed over 95% correct classification of the samples. This could be an important analytical test in countries that do not allow the presence of taurine in energy drinks and replace it with either arginine or nothing at all. The caffeine PLS model showed good correlation and the RMSEP is slightly greater than 5% of the total range of caffeine values in the samples used for calibration (120 ppm to 340 ppm). In the case of sugar, the nominal reference values showed better results than the reference values obtained using the Schoorl method. This likely occurred because the Schoorl method has large bias and error, a conclusion proven by the 12.4% standard deviation when performing the analysis on duplicate samples. Despite these results, it must be noted that the RMSEP for caffeine is considered to be below the detectable threshold limit for NIR spectroscopy for this parameter. It is possible that the model is correlating indirectly to a parameter affected by a change in caffeine. An indirect correlation is acceptable for a PLS regression model, but the results need to be carefully examined and validated before using such a model in a real-time setting. Therefore, more work and study would be necessary to determine the validity of the caffeine measurement from NIR spectra made in this study.
Nondestructive Measurement and Fingerprint Analysis of Soluble Solid Content of Tea Soft Drink Based on VIS/NIR Spectroscopy – Li, He, Wu, Sun, Journal of Food Engineering 82 (2007) 316-323
|SSC (PLS)||R2= 0.981||RMSEP= 0.57 °Brix|
|SSC (MLR)||R2= 0.975||RMSEP= 0.69 °Brix|
The PLS model was created over the full range from 400 nm to 1000 nm and showed excellent correlation between the spectral data and reference SSC values. Predictions on the validation set proved the feasibility of the model. Based on statistical analysis of the latent variable inputs from the wavelengths for the calibration, five statistically significant wavelengths were selected: 490 nm, 498 nm, 554 nm, 929 nm, and 970 nm. These five wavelengths were used to create the MLR model, which showed comparable results to the PLS model. The potential of using both calibration models was demonstrated and VIS/NIR spectroscopy can be used as a method to predict SSC in tea soft drink.
Prediction of Amino Acids, Caffeine, Theaflavins, and Water Extract in Black Tea Using FT-NIR Spectroscopy Coupled Algorithms – Zareef, Chen, Ouyang, Analytical Methods, Issue 25, 2018
Tea is the world’s second highest consumed beverage after water and is categorized into two types: black tea and green tea. Black tea accounts for about 80% of world tea production and green tea accounts for the other 20%. Various fermentation processes are used to produce over three hundred different types of tea worldwide. The main chemical constituents of tea are amino acids, polysaccharides, polyphenols, alkaloids, organic acids, volatile compounds, and proteins. Theaflavins are also an important component that contribute to the antioxidant effect in black tea. FT-NIR spectroscopy was examined as a method for analyzing the following four components in black tea: amino acids, caffeine, theaflavins, and water extract. Ninety-five black tea powder samples from multiple countries were procured for the study. FT-NIR spectra were collected from 10000 cm-1 to 4000 cm-1 at 3.86 cm-1 intervals. Thirty-two scans were collected and averaged into one spectrum for each data point. This process was repeated three times for each sample, with the sample cup rotated 120° two subsequent times after the first spectrum was collected. Reference values for the parameters of interest were collected using traditional methods and various pretreatments were applied to the spectral data for model optimization. Four different chemometrics algorithms were used to correlate the NIR spectra to the parameters of interest: Partial Least Squares (PLS), Synergy Interval PLS (Si-PLS), Backward Interval PLS (Bi-PLS), and Genetic Algorithm PLS (GA-PLS).
|Amino Acids||R2= 0.9396||RMSEP= 0.219 mg/g|
|Caffeine||R2= 0.9195||RMSEP= 0.192 mg/g|
|Theaflavins||R2= 0.9056||RMSEP= 0.204 mg/g|
|Water Extract||R2= 0.8886||RMSEP= 1.53 mg/g|
|Amino Acids||R2= 0.9426||RMSEP= 0.207 mg/g|
|Caffeine||R2= 0.9216||RMSEP= 0.184 mg/g|
|Theaflavins||R2= 0.9439||RMSEP= 0.156 mg/g|
|Water Extract||R2= 0.9192||RMSEP= 1.27 mg/g|
|Amino Acids||R2= 0.9446||RMSEP= 0.19 mg/g|
|Caffeine||R2= 0.9328||RMSEP= 0.171 mg/g|
|Theaflavins||R2= 0.9454||RMSEP= 0.154 mg/g|
|Water Extract||R2= 0.9172||RMSEP= 1.33 mg/g|
|Amino Acids||R2= 0.9506||RMSEP= 0.197 mg/g|
|Caffeine||R2= 0.9274||RMSEP= 0.182 mg/g|
|Theaflavins||R2= 0.9172||RMSEP= 0.19 mg/g|
|Water Extract||R2= 0.9264||RMSEP= 1.26 mg/g|
The PLS modeling algorithm uses the full wavelength range to correlate the spectral data. Both Si-PLS and Bi-PLS use interval selection analysis to select wavelength ranges to optimize the models. GA-PLS is also a variable selection method based on principles of natural and genetic selection of the data. Multiple runs are often necessary to achieve good results using this method. All four methods showed good results with correlation coefficients well above 0.9 for all parameters. GA-PLS showed the best results for caffeine and theaflavins while Bi-PLS showed the best results for caffeine and theaflavins. The results of this study prove the feasibility of measuring amino acids, caffeine, theaflavins, and water extract using NIR spectroscopy and calibration models.