Skip to main content


Front. Chem., 05 July 2021
Sec. Analytical Chemistry
Volume 9 - 2021 |

Analysis of Milk Using a Portable Potentiometric Electronic Tongue Based on Five Polymeric Membrane Sensors

  • 1Group UVASENS, Escuela de Ingenierías Industriales, Universidad de Valladolid, Valladolid, Spain
  • 2BioecoUVA Research Institute, Universidad de Valladolid, Valladolid, Spain
  • 3Dpt. of Materials Science, Universidad de Valladolid, Valladolid, Spain
  • 4Centro de Investigação de Montanha (CIMO), ESA, Instituto Politécnico de Bragança, Bragança, Portugal

A portable potentiometric electronic tongue (PE-tongue) was developed and applied to evaluate the quality of milk with different fat content (skimmed, semi-skimmed, and whole) and with different nutritional content (classic, calcium-enriched, lactose-free, folic acid–enriched, and enriched in sterols of vegetal origin). The system consisted of a simplified array of five sensors based on PVC membranes, coupled to a data logger. The five sensors were selected from a larger set of 20 sensors by applying the genetic algorithm (GA) to the responses to compounds usually found in milk including salts (KCl, CaCl2, and NaCl), sugars (lactose, glucose, and galactose), and organic acids (citric acid and lactic acid). Principal component analysis (PCA) and support vector machine (SVM) results indicated that the PE-tongue consisting of a five-electrode array could successfully discriminate and classify milk samples according to their nutritional content. The PE-tongue provided similar discrimination capability to that of a more complex system formed by a 20-sensor array. SVM regression models were used to predict the physicochemical parameters classically used in milk quality control (acidity, density, %proteins, %lactose, and %fat). The prediction results were excellent and similar to those obtained with a much more complex array consisting of 20 sensors. Moreover, the SVM method confirmed that spoilage of unsealed milk could be correctly identified with the simplified system and the increase in acidity could be accurately predicted. The results obtained demonstrate the possibility of using the simplified PE-tongue to predict milk quality and provide information on the chemical composition of milk using a simple and portable system.


Milk is a nutritious food containing significant components including fats, lactose, sugars, amino acids, vitamins, nucleotides, inorganic salts, and trace elements among many others. Milk composition and nutrient levels are usually assessed by classic analytical techniques such as gas chromatography/mass spectroscopy, high-performance liquid chromatography, and spectroscopy (Toldrá et al., 2021). However, there is a great interest in developing new methods for rapid detection, quantification, and evaluation of milk. Electrochemical sensors can be an alternative to classical laboratory techniques because of their high sensitivity, low cost, and inherent portability (Zeng et al., 2018). However, electrochemical sensors have a lack of selectivity, and this can be a problem when dealing with complex samples such as milk.

An interesting approach to improve the performance of non-specific sensors consists in the use of arrays of sensors. According to the IUPAC definition, an electronic tongue (ET) is a multisensor system, which consists of a number of low-selective sensors and uses advanced mathematical procedures for signal processing based on pattern recognition and/or multivariate data analysis (Vlasov et al., 2005; Rodríguez-Méndez, 2016). Several types of sensors have been used in ETs dedicated to the analysis of milk and dairy products (Poghossian et al., 2019). These include voltammetric electrodes (Cetó et al., 2014; Wei et al., 2013; Bougrini et al., 2014; Salvo-Comino et al., 2018) or impedimetric sensors (Scagiona et al., 2016). However, most of the works in ETs applied to milk have been carried out using potentiometric sensor arrays. Since the pioneering work of Toko’s group that developed an ET based on potentiometric electrodes (composed of several lipid/polymer membranes) (Hayaschi et al., 1995), the analysis of milk samples using ETs has been an active field of research. These instruments have been constantly progressing, and new improvements in sensor arrays, in data processing methods, and in applications are reported in the literature (Ciosek and Wroblewski, 2011; Ciosek and Wroblewski, 2015).

Potentiometric ETs have been used to analyze different aspects related to the quality of milk and dairy products. For example, they have been used for the discrimination and classification of milk based on fat content and brand (Ciosek and Wroblewski, 2008); for the classification of natural, fermented, and UHT milk (Tazi et al., 2018); to evaluate the organoleptic properties (Minghui et al., 2019); to detect flavor changes in bovine and goat milk (Tazi et al., 2017); in the detection of spoilage (Poghossian et al., 2019); and in the detection of adulterations of goat milk with bovine milk (Dias et al., 2009) or detection of water added to dairy products (Mabrook et al., 2006) among others. Only a few works have explored the possibility of using e-tongues to evaluate chemical parameters in milk. These works use partial least squares (PLS) models to establish correlations between the signals obtained with the sensor arrays and the chemical parameters obtained by classical techniques. For example, e-tongues have been used to determine the content of ethanol, acetaldehyde, lactic acid, acetic acid, and citric acid in probiotic fermented milk (Hruskar et al., 2010), or for the simultaneous detection of water-soluble ions (Torabi et al., 2020).

In many cases, potentiometric e-tongues consist of a large number of sensors (typically between 8 and 36) (Dias et al., 2009; Podrażka et al., 2018; Torabi et al., 2020). The idea behind these large arrays is that a higher number of sensing units with different selectivity and sensitivity can provide a larger amount of information, improving the discrimination and prediction capability of the e-tongue. However, a large number of sensors imply a considerable number of variables to manage, and in many cases on-line. In addition, a large number of variables imply the presence of features containing irrelevant or redundant information. In addition, collinearity present in the variables may affect the prediction results (Leardi, 1998). Finally, when models are built using a large number of variables, the training sample set must also be large. All these reasons make it interesting to reduce the number of variables by selecting only those that provide important information.

Several methods have been used to reduce the number of variables in e-tongues, including the wavelet transform (Holmin et al., 2002) or the kernel method (Salvo-Comino et al., 2018). Tools such as a genetic algorithm (GA) applied to PLS regression can be successfully used as a feature selection technique in voltammetric e-tongues (Prieto et al., 2013). The GA represents an efficient approach to non-linear optimization problems and has several advantages. For example, it does not require linear assumptions and is independent of the misfit criterion (Mirjalili, 2019; Reeves, 2010). The GA incorporates and exploits data collected during model space sampling, resulting in an incredibly efficient and robust optimization technique (Beg and Islam, 2016).

The main goal of this study was to develop a simplified and portable potentiometric electronic tongue (PE-tongue). To this end, the work has been carried out to 1) reduce the number of sensors forming the array using the GA; 2) investigate whether this simplified system could be used to discriminate milk with different fat content and nutritional composition using principal component analysis (PCA); 3) investigate whether acidity, density, %proteins, %lactose, and %fat could be accurately predicted using support vector machine (SVM) regression; and 4) evaluate the ability of the PE-tongue to detect spoilage of unsealed milk while exploring the possibility of predicting changes in physicochemical properties.

Materials and Methods

All the reagents used in this work were of analytical degree and used as supplied. Standard solutions of KCl, CaCl2, NaCl, lactose, glucose, galactose, citric acid, and lactic acid (Sigma-Aldrich, St. Louis, United States) were prepared in MilliQ deionized water (Merck KGaA, Darmstadt, Germany). Two sets of milk samples were used in the study. The first set was used to train the e-tongue and consisted of 13 types of milk samples (five replicas of each, total of 65 samples), including milk with different fat content (skimmed, semi-skimmed, and whole) and milk with different nutritional content (classic, calcium-enriched, lactose-free, folic acid–enriched, and enriched in sterols of vegetal origin) (Table 1). A second set of five samples was used as an external testing set to validate the results obtained with the e-tongue. The samples were stored at room temperature until used. Milk samples were analyzed by classical chemical methods: acidity (titration method ISO 22113:2012), density (hydrometer method ISO 2449:1974), fat (Röse-Gottlieb gravimetric method ISO 1211:2010), proteins (Kjeldahl method ISO 8968–1:2014), and lactose content (HPLC ISO 22662:2007). The averages of physicochemical parameters analyzed for each of the milk samples are collected in Table 1.


TABLE 1. Types of milk samples included in the study and results of the chemical analysis.

Sensors were based on polymeric membranes using high-density polyvinyl chloride (PVC) as the polymeric matrix (Sigma-Aldrich, St. Louis, United States). Additives and plasticizers were added to the polymeric matrix using tetrahydrofurane as the solvent (Sigma-Aldrich, St. Louis, United States). The final composition was 32% of PVC, 65% of the plasticizer compound, and 3% of the additive.

The additives and plasticizers are listed in Table 2. Five plasticizers (A–F) were combined with four additives (1–4) to obtain 20 different sensors that were named A1, A2, A3, A4; B1, B2, B3, B4; C1, C2 C3, C4; D1, D2, D3, D4; and F1, F2, F3, F4.


TABLE 2. List of additives and plasticizers.

The body of the e-tongue consisted of an acrylic tube in which 20 holes of 0.3 cm diameter were drilled. Each hole was filled with a conductive silver epoxy resin (EPO-TEK, Billerica, United States) and connected to a multiplexer (Agilent Data Acquisition Switch Unit 34970A) via electrical copper wires. The outer surface of each hole was covered with one of the polymeric membranes described in Table 2. An Ag/AgCl electrode was used as the reference electrode for all measurements. Figure 1 shows the schematic of the e-tongue system containing the working and reference electrodes connected to a multiplexer.


FIGURE 1. Scheme of the potentiometric e-tongue used in this work.

The sensor array was immersed in a glass cell containing 100 ml of each sample (diluted 1:1 in deionized water). Potentiometric measurements were recorded for 5 min by registering the sensor signals every 3 s. All the samples were measured in quintuplicate. Throughout the experiment, the samples were kept at room temperature and under gentle agitation.

The statistical analysis was performed using RKWard 0.7.1 and Matlab R2014b (The Mathworks Inc., Natick, United States). Data analysis included pre-processing of the potentiometric signals using the genetic algorithm and partial least squares (GA–PLS) procedure. Principal component analysis was used to assess the discrimination ability of the multisensor system. Support vector machine regression (SVMR) was used to establish correlations between the results obtained with the e-tongue and the chemical parameters given by chemical analysis. In addition, SVMR was used as a classification method to predict the quality of milk samples.


Characterization of the Individual Sensors

The performance of the 20 PVC membrane–based potentiometric sensors was evaluated using eight standard solutions of compounds commonly present in milk, including salts (KCl, CaCl2, and NaCl), sugars (lactose, glucose, and galactose), and organic acids (citric acid and lactic acid), with concentrations ranging from 1 × 10–4 to 1 × 10–1 mol/L.

After immersing the electrodes in the corresponding solution, membrane potentials were recorded for 5 min every 3 s until stabilization of the signals. Signals were considered stable when an average variation of 1.6 mV/decade was observed between each reading. Figure 2 illustrates the responses obtained when the sensors were immersed in standard solutions (figure shows the average of five replicas). This figure shows an example of salt (CaCl2) and acid (lactic acid) of monosaccharide (galactose) and of disaccharide (lactose). The figure also includes the responses obtained at four different concentrations (5 × 10−1 M, 2 × 10−1 M, 1 × 10−1 M, and 10–2 M) for ionic salts and 10−1 M, 10–2 M, 10–3 M, and 10–4 M for the rest of the compounds. As shown in Figure 2, the sensors showed a variety of responses to the different components of milk. The sensitivity values shown in Table 3 (measured as the slope of the calibration curves) were lower for sugars than for solutions containing ions and lactic acid, confirming the different reactivity of the sensors to components usually found in milk matrices. These differences confirm the cross-selectivity of the sensors and their suitability to be part of a multisensor system.


FIGURE 2. Responses of the array of sensors to standard solutions of (A) CaCl2, (B) lactic acid, (C) lactose, and (D) galactose. Measures were carried out at four concentrations: 5 × 10–1 M (blue), 2 × 10–1 M (orange), 1 × 10–1 M (black), and 10–2 M (green) for ionic salts and 10–1 M (blue), 10–2 M (orange), 10–3 M (black), and 10–4 M (green) for lactic acid, lactose, and galactose.


TABLE 3. Sensitivity values obtained from the slopes of the calibration curves.

The signal repeatability and the reproducibility of the sensors against standard solutions were analyzed. Repeatability was evaluated by analyzing the responses of the sensors immersed in 0.1 M KCl solutions. Measurements were performed in quintuplicate, and coefficients of variation between 0.1 and 1.38% were obtained for all sensors. Reproducibility was calculated by analyzing the responses of two sets of identical sensors immersed in 0.1 M KCl solution. The responses of the sensors showed coefficients of variation between 0.57 and 7.76%. The lifetime was studied by calculating the coefficients of variation of the responses of the sensors immersed in 0.1 M KCl solution for a period of thirty days. The results showed coefficients of variation between 0.52 and 8.56%. The data collected with the set of 20 sensors were used as input variables for multivariate analysis. PCA was used to assess the discrimination ability of the array. Figure 3 shows the score plot of this analysis, in which the first two principal components explained 95% of the covariance of the data (90% by PC1 and 5% by PC2). The compounds analyzed were grouped according to their chemical nature. Ionic compounds appear on the left side of the graph, sugars in the middle, and organic acids in the right part of the diagram.


FIGURE 3. PCA score plot of the standard solutions of compounds present in milk.

Analysis of Milk: Simplification of the Array of Sensors

The performance capabilities of the 20-sensor array were evaluated by registering the potentiometric signals obtained when the electrodes were immersed in milk samples of different qualities. Figure 4 illustrates the responses of the sensors to the milk samples analyzed. As can be seen, each sensor shows distinct responses toward milk with different composition. For example, classic milk showed higher potentials than calcium-enriched milk or lactose-free milk, regardless of their fat content. Moreover, all sensors showed significant differences between samples, confirming the cross-selectivity of the array.


FIGURE 4. (A) Potentiometric responses of sensor 2C toward milk. (B) Response of the array of 20 sensors to classic whole milk (C-W).

The e-tongue developed here must operate in an industrial environment where time and cost are of paramount importance. For this reason, it is relevant to simplify the system by reducing the number of sensors included in the array but without losing information. A visual inspection of the sensor array responses indicated that some sensors could provide redundant information. For example, the responses of sensor 4A to different sugars were quite similar, and sensor 1F could barely detect sugars. To reduce the number of sensors in the array, the GA procedure was chosen to select the best sensors to be included in the array and build predictive regression models (Gendreau and Potvin, 2010). With GA, it is possible to get good solutions for the optimization problems. The GA was applied separately for the five chemical parameters (acidity, density, proteins, lactose, and fat). Fitting a PLS model (GA–PLS) to the sensor array for 13 milk samples and computing the performance by a leave-one-out cross-validation procedure was the way to optimize the problem. The chosen probability of initial variable selection was 0.5, the probability of crossover was 0.5, and the probability of mutation was 0.1. The selected variables were determined to be optimal after 500 GA–PLS evaluations with changing empirical parameter values. Ten iterations per evaluation were performed to avoid overfitting.

The response variable for each GA–PLS optimization was a vector of zeros and ones, corresponding to the thirteen milk types. In this way, the GA–PLS searches for the most relevant sensors in the electronic tongue to identify differences related to milk type. The sensors that provided a greater amount of information are indicated by showing higher responses in Figure 5. These sensors provided a better differentiation between samples according to the type of milk by nutritional content.


FIGURE 5. Frequency of appearance of the sensors for 500 GA–PLS evaluations, for each of the empirical parameters: density (black), acidity (blue), fat (orange), protein (green), and lactose content (red).

As can be seen, the sensors that include oleyl alcohol in their composition showed the best responses for each of the parameters studied. Therefore, these sensors were selected to form a new simplified e-tongue consisting of five sensors (2A, 2B, 2C, 2D, and 2F). The discrimination and prediction capability of the simplified e-tongue was studied.

Evaluation of the Performance of the 20-Sensor Array and the Simplified System Based on Five Sensors

1) Assessment of the discrimination capacity using PCA

In this section, the performance of the 20-sensor–based e-tongue was compared with that of the five-sensor–based system.

A PCA was performed to evaluate the discrimination ability of the array. Figure 6A shows the score plot of this analysis, in which the first three principal components jointly explained 88% of data variability (64% by PC1, 15% by PC2, and 9% by PC3). Surprisingly, discrimination between milk samples was not dominated by fat content. Instead, milk samples were grouped according to added components (calcium or folic acid) or removed components (lactose). PC1 clearly discriminated classic milk (C: includes milk samples C-S, C-SS, and C-W) regardless of their fat content from the rest of the milk samples. PC2 discriminated calcium-enriched milk (CA: includes milk samples CA-S, CA-SS, and CA-W) from lactose-free milk (L: includes milk samples L-S, L-SS, and L-W) regardless of their fat content. Milk enriched in folic acid or in vegetal sterols appeared at the bottom of the diagram (F and P: include F-S, F-SS, F-W, and P-SS). The mixing of samples among some groups may be due to the sensitivity of the sensor device to other milk components such as fat content, which widely affect physicochemical aspects such as viscosity or density in the samples.


FIGURE 6. PCA score plot of the 13 milk samples of different fat content and nutritional characteristics analyzed using (A) a 20-sensor array and (B) a five-sensor array. C = classic milk, CA = calcium-enriched milk, L = lactose-free milk, F = folic acid–enriched milk, and P = vegetal sterol–enriched milk.

Figure 6B shows the score plot obtained from the responses to milk with different nutritional content obtained using the PE-tongue based on a five-sensor array. The first two PCs explain 89% of the total variance. As can be seen in Figure 6B, milk samples with similar nutritional content appeared in the same region of the graph. The first group contains milk with enhanced nutritional calcium content (CA: includes CA-S, CA-SS, and CA-W); the second group includes milk samples without any modification in terms of nutritional content (C: includes C-S, C-SS, and C-W); the third group contains milk that has been modified for low lactose content (L: includes L-S, L-SS, and L-W); finally, there is a mixed group with four types of milk samples with an increased amount of folic acid in their composition (F and P: include F-S, F-SS, F-W, and P-SS). The first and second groups, in the positive part of the second principal component, are separated by the first principal component (being in the positive and negative parts, respectively). The third, fourth, and fifth groups are separated by the second and third principal components (in the negative region) with a very low contribution of the first principal component.

This result indicates that the final subset of variables retained by GA–PLS presents the ability to discriminate milk classes according to their nutritional components, similar to that obtained with a complex system consisting of 20 sensors, although a complete discrimination has not been obtained.

2) Assessment of the classification capacity using SVM

The classification capability of the PE-tongue system was tested using SVM using the radial basis function (RBF) (Wu and Wang 2009) as a non-linear kernel, defined as follows:

K(xixj)=exp(γxixj2),      γ>0,

where xi and xj are the training vectors of the input data and γ is the kernel parameter.

This kernel has been chosen since the number of instances is larger than the number of features, and therefore, it is recommended to use non-linear kernels (Hsu 2016).

In this study, the SVM model was trained with data obtained from 65 samples (corresponding to the thirteen types of milk samples with different fat content and nutritional characteristics). The same scaling factors were applied for the training and testing sets.

The optimal SVM regularization parameter (C, which is the penalty parameter of the error term) was set to the highest value (C = 100), implying that classification errors were not tolerated for the set of response patterns used for SVM training. Due to the relatively small number of available measurements, the leave-one-out cross-validation method was implemented to better estimate the true success rate that could be achieved with the SVM. This assumes that, with the given n measurements, the model was trained n times using n − 1 training vectors. This vector was then used for testing.

The results obtained for the 20-sensor array showed an accuracy of 99.87% for the classification of the 13 classes in training and 98.46% in validation. The results showed that the accuracy of the five-sensor array training set was 96.92%, and those of the validation set showed an accuracy of 90.76%, for the thirteen categories. Each of the problem samples was correctly classified according to its nutritional content. These results determine that the electronic tongue developed with five sensors was able to classify the milk samples according to their nutritional content and also by their fat content.

3) Prediction of chemical parameters by means of support vector machine regression (SVMR) models

One of the main challenges in the field of e-tongues is the implementation of models that can predict chemical parameters of importance in food quality control.

In this work, SVMR was used to predict acidity, density, and percentage of protein, lactose, and fat in milk. Sixty-five samples were used as the training set, and five samples (denoted V1, V2, V3, V4, and V5) were used as the external test set. The radial basis function, which could handle the non-linear relationships between the sensor signals and the target attributes, was chosen as the core function to predict acidity, density, and percentage of protein, lactose, and fat.

Although an electronic tongue has been shown to be able to perform classifications, the challenge for the reduced PE-tongue system is the implementation of regression models that can predict physicochemical parameters with adequate correlation values. For this purpose, SVMR has been applied to the 20-sensor array data.

Two data matrices have been built: the “X” matrix (predictors) constructed from the data recorded by the electronic tongue analyzing the milk samples and the “Y” matrix (responses) containing data of chemical parameters (acidity, density, protein, lactose, and fat) of the milk samples. Regression models were created using SVM regression (epsilon SVM, kernel type: radial basis function, C value: 1, cross-validation segment size: 15, and standard deviation weighting process in all cases).

The values obtained for the correlation coefficients and errors of calibration and prediction are shown in Table 4. In the case of acidity, protein content, lactose content, and density, the developed models achieved correlation values R2 above 0.94 for both calibration and prediction, with low errors (RMSE) between 0.0239 and 0.9915. Lactose was the parameter with the lowest errors and highest correlation. In the case of fat, the correlation value only achieved 0.7789 for the prediction with a higher error of 0.6102.


TABLE 4. Results of the calibration and validation of SVMR for the 20-sensor e-tongue.

The recognition capability of the five-sensor–based PE-tongue was tested by applying an SVM classification (SVMC) model. In this study, the model was trained with the data of 65 samples for the training set (corresponding to 13 types of milk samples depending on their fat content and nutritional content), calibrated against the true type, and used to predict the classification of the problem samples (five milk samples with different nutritional content) for the test set.

The values obtained for the correlation coefficients and errors of calibration and prediction are shown in Table 5.


TABLE 5. Results of the calibration and validation of SVMR for the five-sensor e-tongue.

As can be seen, in the case of acidity, protein content, lactose content, and density, the developed models reached values above 0.85 for both calibration and prediction, with low errors between 0.0373 and 1.0509, lactose being the parameter with lowest errors (0.0373 for the calibration and 0.0434 for the prediction) and the highest correlation (0.9763 for the calibration and 0.9666 for the prediction). However, in the case of fat, it is observed that a good correlation between the data provided by an electronic tongue and the physical–chemical data was not achieved.

Once the SVMR model was built, as a verification of its applicability, regression models were used to predict the physicochemical parameters (acidity, density, proteins, lactose, and fat) of a set of five external samples that were not included in the creation of the model. The results are shown in Table 6 vs. those obtained by traditional methods.


TABLE 6. Results of the prediction of chemical parameters in a set of five external samples.

The results obtained showed that the SVM regression model was able to predict the physicochemical factors with values that showed low relative errors with respect to the values obtained by traditional analysis techniques, the lowest errors being reached for lactose content and density. However, in the case of fat content, as expected given its correlation parameters, the values obtained by prediction show a high error, with the highest value being 2.11% in the case of skim milk. These results show that the model developed is unable to predict the fat content. In spite of the number of research works dedicated to the analysis of milk with potentiometric electronic tongues, the effect of the interaction between fats and the sensor membranes has not been discussed (Ciosek, 2016; Poghossian et al., 2019; Rodríguez-Mendez et al., 2016). A possible explanation is that when sensors are immersed in milk, fats participate in the formation of the double layer that creates the membrane potential. Once the double layer is formed, the excess of lipids does not contribute to the signal.

Spoilage Monitoring

The capabilities of the simplified array were further evaluated by analyzing the capabilities to detect spoilage. For this purpose, tetrabrick packs were opened and stored at 5°C. Measurements were performed right after opening the packs and 5 days later.

PCA was used as a first approach to evaluate the ability of the PE-tongue to detect spoilage occurring in unsealed milk. As shown in Figure 7, the classic milk sample appeared clearly separated from the rest of the milk samples, and the simplified PE-tongue could detect compositional changes occurring during storage. The reason why the classic milk samples (C-S, C-SS, and C-W) have a different aging behavior may be related to the fact that the rest of the milk samples have undergone various processes that can affect their composition. When examining the rest of the milk samples, a partial overlap between fresh and aged samples could be observed. This could be due to the fact that the degradation process does not occur at the same rate in all samples, as it can be affected by multiple factors such as milk composition and sample handling.


FIGURE 7. PCA score plot of the 13 milk samples analyzed using the PE-tongue: fresh sample (blue) and unsealed sample after 5°days (orange).

SVMC was used to classify between the two milk groups: fresh milk and aged milk. The classification results using the dataset showed that the sensor array achieved classification of the fresh and aged milk samples with an accuracy of 92.31% for the training set and 90.77% for the validation set. These results determine that the electronic tongue developed with five sensors was able to classify milk samples according to their state of preservation. Using SVMR, it has been demonstrated that the PE-tongue could be used to accurately predict physicochemical properties of the unsealed milk. After predicting the acidity of the aged samples, each of the samples showed an increase of approximately 1.2ºD in acidity. These results are consistent with the expected increase in milk acidity as a result of the increased microbiological activity in the unsealed samples.


In this work, a simplified and portable electronic tongue (PE-tongue) was developed and used to predict chemical characteristics of milk samples. The system used only five potentiometric sensors that were selected from an extended array of 20 sensors using the genetic algorithm. PCA showed that the PE-tongue showed similar discrimination capabilities to the extended ET consisting of 20 sensors, but with a substantial decrease in the number of variables to be managed. The PE-tongue could be successfully used to classify milk with different nutritional characteristics and to predict acidity, density, %proteins, %lactose, and %fat, with low errors and high correlation coefficients. Potentiometric data acquired with the PE-tongue were successfully subjected to support vector machine (SVM) for classification of fresh and spoiled milk samples and to establish correlations with acidity of unsealed milk with excellent results.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Author Contributions

MR-M, CG-C, and FM-P conceptualized the idea and supervised the work. CP-G and CS-C performed the experiment, curated the data, and wrote the original draft. FM-P involved in software design and development. CP-G, CS-C, MR-P, and LD involved in formal analysis. CG-C and MR-M acquired the funding. CP-G, CS-C, FM-P, MR-M, CG-C, and LD reviewed and edited the paper. All authors provided feedback.


We appreciate the financial support of MINECO-FEDER Plan Nacional (RTI 2018-097990-B-100) and Junta de Castilla y Leon-FEDER (VA275P18) and “Infraestructuras Red de Castilla y León (INFRARED)” UVA01.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling Editor declared a past co-authorship with the authors CS, MR.


Beg, A. H., and Islam, M. Z. (2016). Advantages and Limitations of Genetic Algorithms for Clustering Records, Industrial Electronics and Applications (ICIEA), 2478–2483. doi:10.1109/iciea.2016.7604009

CrossRef Full Text | Google Scholar

Bougrini, M., Tahri, K., Haddi, Z., El Bari, N., Llobet, E., Jaffrezic-Renault, N., et al. (2014). Aging Time and Brand Determination of Pasteurized Milk Using a Multisensor E-Nose Combined with a Voltammetric E-Tongue. Mater. Sci. Eng. C 45, 348–358. doi:10.1016/j.msec.2014.09.030

CrossRef Full Text | Google Scholar

Cetó, X., Apetrei, C., Del Valle, M., and Rodríguez-Méndez, M. L. (2014). Evaluation of Red Wines Antioxidant Capacity by Means of A Voltammetric E-Tongue with an Optimized Sensor Array. Electrochimica Acta 120, 180–186. doi:10.1016/j.electacta.2013.12.079

CrossRef Full Text | Google Scholar

Ciosek, P. (2016). “Milk and Dairy Products Analysis by Means of an Electronic Tongue,” in Electronic Noses and Tongues in Food Science. Editor M. L. Rodriguez-Mendez(Academic Press), 209–223. doi:10.1016/b978-0-12-800243-8.00021-4

CrossRef Full Text | Google Scholar

Ciosek, P., and Wróblewski, W. (2008). Miniaturized Electronic Tongue with an Integrated Reference Microelectrode for the Recognition of Milk Samples. Talanta 76, 548–556. doi:10.1016/j.talanta.2008.03.051

PubMed Abstract | CrossRef Full Text | Google Scholar

Ciosek, P., and Wróblewski, W. (2015). Potentiometric and Hybrid Electronic Tongues for Bioprocess Monitoring - an Overview. Anal. Methods 7, 3958–3966. doi:10.1039/c5ay00445d

CrossRef Full Text | Google Scholar

Ciosek, P., and Wróblewski, W. (2011). Potentiometric Electronic Tongues for Foodstuff and Biosample Recognition-An Overview. Sensors 11, 4688–4701. doi:10.3390/s110504688

PubMed Abstract | CrossRef Full Text | Google Scholar

Dias, L. A., Peres, A. M., Veloso, A. C. A., Reis, F. S., Vilas-Boas, M., and Machado, A. A. S. C. (2009). An Electronic Tongue Taste Evaluation: Identification of Goat Milk Adulteration with Bovine Milk. Sensors Actuators B: Chem. 136, 209–217. doi:10.1016/j.snb.2008.09.025L

CrossRef Full Text | Google Scholar

Gendreau, M., and Potvin, J. Y. (2010). Handbook of Metaheuristics. International Series in Operations Research & Management Science, Springer, Boston, MA.

Google Scholar

Hayaschi, K., Toko, K., Yamanaka, M., Yoshihara, H., Yamafuji, K., Ikazeki, H., et al. (1995). Electric Characteristics of Lipid-Modified Monolayer Membranes for Taste Sensors. Sensors and Actuators B 23, 55–61. doi:10.1016/0925-4005(94)01522-j

CrossRef Full Text | Google Scholar

Holmin, S., Björefors, F., Eriksson, M., Krantz-Rülcker, C., and Winquist, F. (2002). Investigation of Electrode Materials as Sensors in a Voltammetric Electronic Tongue. Electroanalysis 14, 839–847. doi:10.1002/1521-4109(200206)14:12<839::aid-elan839>;2-y

CrossRef Full Text | Google Scholar

Hruškar, M., Major, N., Krpan, M., and Vahčić, N. (2010). Simultaneous Determination of Fermented Milk Aroma Compounds by a Potentiometric Sensor Array. Talanta 82, 1292–1297. doi:10.1016/j.talanta.2010.06.048

PubMed Abstract | CrossRef Full Text | Google Scholar

Hsu, C. W., Chang, C. C., and Lin, C. J. (2016). A Practical Guide to Support Vector Classification. Tech. Rep.∼cjlin/.

Google Scholar

Leardi, R., and Lupiáñez González, A. (1998). Genetic Algorithms Applied to Feature Selection in PLS Regression: How and when to Use Them. Chemometrics Intell. Lab. Syst. 41, 195–207. doi:10.1016/s0169-7439(98)00051-3

CrossRef Full Text | Google Scholar

Mabrook, M. F., Darbyshire, A. M., and Petty, M. C. (2006). Quality Control of Dairy Products Using Single Frequency Admittance Measurements. Meas. Sci. Technol. 17, 275–280. doi:10.1088/0957-0233/17/2/007

CrossRef Full Text | Google Scholar

Minghui, P., Lingjun, T., Xuelu, C., Nasi, A., Yungang, C., and Baoguo, S. (2019). Comparison of Sensory and Electronic Tongue Analysis Combined with HS-SPME-GC-MS in the Evaluation of Skim Milk Processed with Different Preheating Treatments. Molecules 24 (9), 1650. doi:10.3390/molecules24091650

CrossRef Full Text | Google Scholar

Mirjalili, S. (2019). ““Genetic Algorithm”,” in Studies in Computational Intelligence Springer, Cham, 780. Editors Evolutionary Algorithms, and Neural Networks. doi:10.1007/978-3-319-93025-1

CrossRef Full Text | Google Scholar

Podrażka, M., Bączyńska, E., Kundys, M., Jeleń, P., and Witkowska, E. (2018). Electronic Tongue—A Tool for All Tastes?. Biosensors 8-1, 3. doi:10.3390/bios8010003

CrossRef Full Text | Google Scholar

Poghossian, A., Geissler, H., and Schoning, M. J. (2019). Rapid Methods and Sensors for Milk Quality Monitoring and Spoilage Detection. Biosens. Bioelectron. 140, 18–31. doi:10.1016/j.bios.2019.04.040

CrossRef Full Text | Google Scholar

Prieto, N., Oliveri, P., Leardi, R., Gay, M., Apetrei, C., Rodriguez-Méndez, M. L., et al. (2013). Application of a GA-PLS Strategy for Variable Reduction of Electronic Tongue Signals. Sensors Actuators B: Chem. 183, 52–57. doi:10.1016/j.snb.2013.03.114

CrossRef Full Text | Google Scholar

Rodríguez-Mendez, M. L., De Saja, J. A., Gonzalez-Anton, R., García-Hernández, C., Medina-Plaza, C., and García-Cabezón, C. (2016). Electronic Noses and Tongues in Wine Industry. Front. Bioeng. Biotechnol. 4, 81. doi:10.3389/fbioe.2016.00081

PubMed Abstract | CrossRef Full Text | Google Scholar

Rodriguez-Mendez, M. L. (2016). Electronic Noses and Tongues in the Food Industry. Elsevier-academic Press.

Google Scholar

Salvo-Comino, C., García-Hernández, C., García-Cabezón, C., and Rodríguez-Méndez, M. (2018). Discrimination of Milks with a Multisensor System Based on Layer-By-Layer Films. Sensors 18 (8), 2716. doi:10.3390/s18082716

CrossRef Full Text | Google Scholar

Scagiona, V. P., Mercante, L. A., Sakamoto, K. Y., Oliveira, J. E., Fonseca, F. J., Mattoso, L. H. C., et al. (2016). An Electronic Tongue Based on Conducting Electrospun Nanofibers for Detecting Tetracycline in Milk Samples. RSC Adv. 6. doi:10.3740/103746

CrossRef Full Text | Google Scholar

Tazi, I., Choiriyah, A., Siswanta, D., and Triyana, K. (2017). Detection of Taste Change of Bovine and Goat Milk in Room Ambient Using Electronic Tongue. Indones. J. Chem. 17, 422–430. doi:10.22146/ijc.25288

CrossRef Full Text | Google Scholar

Tazi, I., Triyana, K., Siswanta, D., Veloso, A. C. A., Peres, A. M., and Dias, L. G. (2018). Dairy Products Discrimination According to the Milk Type Using an Electrochemical Multisensor Device Coupled with Chemometric Tools. Food Measure 12, 2385–2393. doi:10.1007/s11694-018-9855-8

CrossRef Full Text | Google Scholar

Toldrá, F., Leo, M. L., and Nollet, L. (2021). Handbook of Dairy Foods Analysis. CRC Press. ISBN: 9780367343132.

Google Scholar

Torabi, F., Ghasemi, J. B., Faridbod, F., and Ganjali, M. R. (2020). Simultaneous Quantification of Nine Major Water-Soluble Inorganic Ions Using a Potentiometric Electronic Tongue in Cheese Samples. IEEE Sensors J. 20, 10138–10144. doi:10.1109/jsen.2020.2990130

CrossRef Full Text | Google Scholar

Vlasov, Y., Legin, A., Rudnitskaya, A., Di Natale, C., and D'Amico, A. (2005). Nonspecific Sensor Arrays (“electronic Tongue”) for Chemical Analysis of Liquids (IUPAC Technical Report). Pure Appl. Chem. 77, 1965–1983. doi:10.1351/pac200577111965

CrossRef Full Text | Google Scholar

Wei, Z., Wang, J., and Zhang, X. (2013). Monitoring of Quality and Storage Time of Unsealed Pasteurized Milk by Voltammetric Electronic Tongue. Electrochimica Acta 88, 231–239. doi:10.1016/j.electacta.2012.10.042

CrossRef Full Text | Google Scholar

Wu, K.-P., and Wang, S.-D. (2009). Choosing the Kernel Parameters for Support Vector Machines by the Inter-cluster Distance in the Feature Space. Pattern Recognition 42, 710–717. doi:10.1016/j.patcog.2008.08.030

CrossRef Full Text | Google Scholar

Zeng, L., Wu, D., and Yang, B. (2018). “Electrochemical Methods of Analysis – Amperometry” in Electrochemical Sensors for Food Safety.

Google Scholar

Keywords: potentiometric electronic tongue, milk, dairy, fat content, lactose, acidity

Citation: Pérez-González C, Salvo-Comino C, Martin-Pedrosa F, Dias L, Rodriguez-Perez MA, Garcia-Cabezon C and Rodriguez-Mendez ML (2021) Analysis of Milk Using a Portable Potentiometric Electronic Tongue Based on Five Polymeric Membrane Sensors. Front. Chem. 9:706460. doi: 10.3389/fchem.2021.706460

Received: 07 May 2021; Accepted: 11 June 2021;
Published: 05 July 2021.

Edited by:

Nicole J. Jaffrezic-Renault, Université Claude Bernard Lyon 1, France

Reviewed by:

Zouhair Haddi, NVISION Systems and Technologies, S.L., Spain
Antonio Riul Jr, State University of Campinas, Brazil

Copyright © 2021 Pérez-González, Salvo-Comino, Martin-Pedrosa, Dias, Rodriguez-Perez, Garcia-Cabezon and Rodriguez-Mendez. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

*Correspondence: C. Garcia-Cabezon,; M. L. Rodriguez-Mendez,