Internet of Food (IoF), Tailor-Made Metal Oxide Gas Sensors to Support Tea Supply Chain

Abstract

Tea is the second most consumed beverage, and its aroma, determined by volatile compounds (VOCs) present in leaves or developed during the processing stages, has a great influence on the final quality. The goal of this study is to determine the volatilome of different types of tea to provide a competitive tool in terms of time and costs to recognize and enhance the quality of the product in the food chain. Analyzed samples are representative of the three major types of tea: black, green, and white. VOCs were studied in parallel with different technologies and methods: gas chromatography coupled with mass spectrometer and solid phase microextraction (SPME-GC-MS) and a device called small sensor system, (S3). S3 is made up of tailor-made metal oxide gas sensors, whose operating principle is based on the variation of sensor resistance based on volatiloma exposure. The data obtained were processed through multivariate statistics, showing the full file of the pre-established aim. From the results obtained, it is understood how supportive an innovative technology can be, remotely controllable supported by machine learning (IoF), aimed in the future at increasing food safety along the entire production chain, as an early warning system for possible microbiological or chemical contamination.

1. Introduction

Tea is native to the northern hills at the foot of the Himalayas where the inhabitants chewed Camellia sinensis for medicinal purposes. The Chinese populations then invented preservation techniques to increase the shelf life and facilitate the transport of the product. Over time, the techniques have improved, and the cultivation areas have increased so as to arrive at the point today where many varieties are known [1].

It is one of the most consumed beverages worldwide, with an increase in the consumption of 23.4% in the last 7 years, reaching 297 billion liters by 2021. Generally, tea is an aromatic beverage prepared by pouring hot or steaming water over dried or fresh leaves of the Camellia sinensis (green, white, and Oolong teas), an evergreen shrub (bush) and Camellia assamica (black and Pu-Erh tea), which originate and are cultivated, respectively, in China and India [2,3].

The most produced and consumed teas worldwide are green and black teas [4].

Tea is an infusion, but not all infusions are tea, as there are infusions or herbal teas that originate from red fruits, chamomile, mint, lavender, etc.

The tea leaves undergo different treatments that determine their classification. Spices or herbs are added to the basic tea in order to spice up the flavor and taste. The first classification is based on the fermentation treatment and is also the classification used at customs; in particular, we have:

• Fermented tea: black and Pu-Erh tea
• Unfermented tea: green and white tea,
• Partially or semi-fermented tea: Oolong tea.

In addition to this classification, the teas are placed in other classes based on the following specific treatments: scented tea, flavored tea, smoked tea, blends, pressed teas, and a bouquet of tea and flowers.

There are many distinctive types of tea; some, such as Darjeeling, Ceylon, Oolong, etc., have cooling, slightly bitter, and astringent flavors, [5,6,7,8] while others have vastly different profiles that include sweet, nutty, floral, or grassy notes.

Tea leaves contain thousands of chemical compounds and release volatile compounds (VOCs), which contribute to the definition of product quality [9,10,11,12,13,14].

The aroma is one of the determining factors in the quality of tea and is due to the volatile compounds. Volatile compounds are mainly responsible for the flavor and aroma of the infusion, and many of them are not present in the fresh leaf but develop after processing.

There are more than 600 volatile compounds (VCs) in tea, resulting from the enzymatic action on odorless compounds present in the leaf that are released after rolling and fermentation [15]. The volatile aromatic compounds differ according to the types of tea, both for the “different” fresh material and for different production processes, without forgetting that the final consumer requires that the purchased product be recognized and standardized.

Recent research has shown that the volatile aromatic components of tea are influenced by several factors: cultivar, area of cultivation, cultural practices, production methods, and conservation [16,17,18,19].

Many kinds of classical analytical chemical techniques such as the gas chromatography coupled with mass spectrometry (GC-MS) have been largely used and have demonstrated their accuracy and specificity, but present important limits are normally expensive and time consuming and require appropriately trained staff to operate them.

On the other hand, an innovative tailor-made gas sensor device named small sensor systems (S3) has been applied broadly in the quality control in food the field and environmental monitoring as well exhibiting remarkable results [20,21,22].

This innovative device, S3, is fast, remotely controllable, totally user friendly, and, once trained, does not need any special skilled staff to operate it. In particular the S3 device is totally tailor-made for the specific application. That is, the sensors are grown, calibrated, and used on the basis of the class of VOCs on which they will be used, thus managing to obtain greater sensitivity and accuracy.

It can then be easily inserted into the production chain, for the evaluation of quality standards or to follow, for example, the evolution of the product over time.

The goal of this study is the characterization of the aroma of different types of tea and the definition of their volatiloma through the use of two different approaches, to support the supply chain of tea. Providing a portable device capable of monitoring a greater quantity of product at low costs and very quickly in a noninvasive way to facilitate compliance with quality standards.

2. Materials and Methods

The samples taken into consideration belong to 3 more consumed worldwide types of tea: black tea, green tea, and white tea. In this study, 20 mL chromatographic vials were used, each filled with approximately one sachet, 2 ± 0.2 g of tea. The used sample, the code name, and the number of replicas used for each technique are represented in Table 1.

Table 1. Samples description.

Kind	Code	Description	Components	Number of Replicates
				GC-MS	S3
Green tea	TGO	Green tea with orange aroma	88% Green tea + 10% natural orange aroma + 1% loto flower + 1% orange skin	3	5
	SGP	Pure green tea	100% Green tea	3	12
	SGL	Green tea with lemon aroma	Green tea 83% + lemon aroma + lemon juice concentrate 4%	3	12
	SGM	Green tea with Matcha tea	Green tea + green tea Matcha	3	12
Black tea	TBV	Black tea with vanilla aroma	91,5% tea + 8% aroma + 0,5% vanilla	3	5
	TBC	Pure black tea from Ceylon	100% Black tea Ceylon Sri Lanka	3	12
	EBL	Black tea with lemon aroma	Black tea from India and lemon aroma	3	12
	EBB	Black tea Earl Grey	Black Tea from India + bergamot juice	3	12
	EBP	Pure black tea	Biological black Tea	3	12
	SBL	Black tea with lemon aroma	Black tea + aroma + 1.04% powder of lemon juice	3	12
White tea	BWL	White tea with lemongrass	White tea leaf 90%, dried lemongrass 10%	3	8
Total number of samples for each technique				33	114

During the sampling process, no chemical extraction or thermal shock was carried out on the samples in order to keep the aroma of the dried product, to evaluate its actual characteristics. The vials were closed with aluminum caps containing polytetrafluoroethylene (PTFE) and silicone septa. The operational conditions were interpreted in Section 2.2, respectively.

2.1. GC-MS Analysis Conditions

After closure, vials were placed in the autosampler HT280T (HTA s.r.l., Brescia, Italy) to proceed with vial conditioning and volatile organic compound (VOC) extraction.

Conditioning of the sample was performed as follows: filled vials were maintained for 15 min at 40 °C in order to equilibrate the headspace (HS) of the sample and to remove any variables. Afterward, VOCs extraction was performed using solid-phase microextraction (SPME) analysis, and the fiber used for the adsorption of volatiles was a divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) 50/30 µm (Supelco Co. Bellefonte, PA, USA) placed on the HT280T autosampler. The fiber was exposed to the vial HS in the HT280T oven thermostatically regulated at 40 °C for 15 min.

The GC instrument used in this work was a Shimadzu GC 2010 PLUS (Kyoto, KYT, Japan), equipped with a Shimadzu single quadrupole mass spectrometer (MS) MS-QP2010 Ultra (Kyoto, KYT, Japan). Fiber desorption took place in the GC–MS injector for 6 min at 250 °C. GC was operated in the direct mode throughout the run, while the separation was performed on a MEGA-WAX capillary column, 30 m × 0.25 mm × 0.25 μm film thickness, (Agilent Technologies, Santa Clara, CA, USA). Hydrogen was used as the carrier gas and has been produced by GENius PF500, FullTech Instruments Srl. (Rome, Italy) at a constant flow rate of 2.34 mL/min.

The GC oven temperature programming was applied as follows: at the beginning, the chromatographic column was held at 40 °C for 2 min and, subsequently, the temperature was raised from 40 to 100 °C at 2 °C/min. Next, the temperature was raised from 100 to 180 °C, with a rate of 5 °C/min; finally, the temperature was raised from 180 to 230 °C at a rate of 10 °C/min and was maintained for 5 min, for a total program time of 58 min [21,23,24].

During the analysis, the GC–MS interface was kept at 200 °C, with the mass spectrometer in the electron ionization (EI) mode (70 eV) and related to instrument tuning, and the ion source was kept at 200 °C. Mass spectra were collected over 35 to 500 m/z, in a range in the total ion current (TIC) mode, with scan intervals at 0.3 s. VOC identification was carried out using the NIST11 and the FFNSC2 libraries of mass spectra.

Chromatogram peak integration was performed using the peak area as target parameter programming an automatic integration round, using 70 as the minimum number of peak detection and 500 as the minimum area to detect. Other parameters used in the automatic peak integration were slope 100/min, width 2 s, drift 0/min, and doubling time (T.DBL) 1000 min, and no smoothing method was applied. The final round of peak integration was performed by manual peak integration for all the obtained chromatograms.

2.2. S3 Analysis Conditions

The chosen sensors for the application were installed on the S3 device, an acronym that stands for small sensor system.

The device that was designed and built by Nasys S.r.l. (www.nasys.it, accessed on 21 June 2021) is an innovative spin-off born in the University of Brescia, where tailor-made sensors were produced.

S3 is composed of three essential parts: (A) sensors chamber, (B) fluid dynamic circuit for the distribution of volatile compounds, and (C) electronics control system.

(A)

The sensors are housed inside a steel chamber isolated from the external environment, except for an inlet and an outlet path for the passage of volatile compounds. In addition to the MOX sensors, a temperature, humidity sensor, and a flow sensor are also allocated as necessary to take into account the number of variables during the analysis. The dimensions of the chamber are 11 × 6.5 × 1.3 cm.

On the 11 and 6.5 cm sides, 6 and 5 positions have been obtained, respectively: 10 positions were available for the use of metal oxide (MOX) sensors and one for the temperature and humidity sensor. The list of MOX used in this study is represented in Table 2, where the technology of production, RGTO (rheotaxial growth and thermal oxidation) or nanowire, the sensing material (SnO₂ or CuO), and the working temperature are indicated.

Table 2. Sensor array composition of S3 device and sensors characteristics.

Material	Kind	Working Temperature
SnO2 + Au	RGTO	400 °C
SnO2	RGTO	400 °C
CuO	Nanowire	350 °C
SnO2 + Au	Nanowire	350 °C
SnO2	Nanowire	350 °C

(B)

The fluid dynamic circuit consists of a pump (Knf, model: NMP05B), polyurethane pipes, a solenoid valve, and a metal cylinder containing activated carbon for filtering possible interfering odors present in the environment. The pump flow is regulated by a needle valve placed at the chamber inlet; the flow range for tea analysis was set to 100 sccm.

(C)

The electronic boards make it possible to acquire the resistances of the sensors, the correct heating of the sensors themselves to their operating temperature, and the sending of data to the Web App dedicated to the S3 device through an internet connection. In addition, it allows communication and synchronization with an autosampler. This is an autosampler of the company HTA S.r.l. (model HT2010H) which allows one to prepare batches of 42 samples per measurement session. Tea samples were conditioned for 5 min at 40 °C with 1 min in a shaking mode in order to equilibrate the headspace (HS).

The RGTO technique requires two phases of deposition: the first step is the metallic thin film by DC magnetron sputtering from a metallic target to a substrate at higher temperatures than the melting point of the metal; and the second step is the thermal oxidation period in order to produce a metal oxide coating with stable stoichiometry [25]. The surface of the thin film is rough, and this is desirable since it has a high surface-to-volume ratio and reactivity to the gaseous species [26]. In addition, the presence of this very rough surface morphology, also known as ‘spongy agglomerates’, gives rise to a highly specific area required for high-sensitivity gas sensors [27]. Nanowires display extraordinary crystalline quality and a very high length-to-width ratio, resulting in improved sensitivity and long-term material stability for extended operation [28,29]. The fabrication method consists of the evaporation of the powder (metal oxide) at high temperatures in a controlled atmosphere at pressures of less than a hundred mbar (50–200 mbar) and the subsequent mass transfer of the vapor (50–100 sccm) to substrates held at lower temperatures in relation to the evaporation source area. This growth technique is called a vapor–liquid–solid (VLS) mechanism.

For SnO₂ sensors, the powders are mounted in the middle of the furnace at 1370 °C, and the inert air flow at temperatures between 350 and 400 °C is used as a carrier from the furnace to the substrate where nanowires begin to develop [30].

S3 was previously used, with considerable success, in numerous studies applied to the field of food technology and quality control [31,32]. The output of the S3 analysis consists of the sensors’ resistance variation due to the interaction of VOCs with the sensing elements. The exposure to VOCs lasted 1 min, while 9 min passed to restore sensors’ baseline. Prior to analysis, the sensors’ responses in terms of resistance (Ω) were standardized relative to the first value of the acquisition (R0). This standardization was performed for each measure so that for all the sensors, the first point value was 1, smoothing differences in the starting value of resistances between the measures themselves. For all the sensors, the difference between the first value and the minimum value was determined during the time of analysis; thus, the value ΔR/R0 was derived. These features were used as input for principal component analysis (PCA). Here, PCA has been used to visualize the data cluster. By contrast, a hard approach to quantify the accuracy of the system was employed. In particular, the k-nearest neighbor (k-NN) algorithm was used with 5-fold crossvalidation technique. The aim of crossvalidation is to test the model’s ability to predict new data that were not used in estimating it, in order to counteract overfitting or selection bias. The accuracy provided is the mean of the accuracies of the 5 steps of prediction.

3. Results

3.1. GC-MS Results

After closure, vials were placed in the autosampler HT280T (HTA s.r.l., Brescia, Italy) to proceed with vial conditioning and volatile organic compound (VOC) extraction.

Regarding the data extracted from the GC-MS-SPME analysis, the volatile fingerprints for each tea were identified, and the tables are presented at the end of the article in Appendix A (Table A1, Table A2, Table A3, Table A4, Table A5, Table A6, Table A7, Table A8, Table A9, Table A10 and Table A11). It was possible to identify the common compounds between the green tea samples that are presented in Table 3.

Table 3. Common compounds between the analyzed green teas.

Compound	TGO	SPG	SGL	SGM
Cyanoacetic acid	1.685 × 105	1.165 × 105	5.677 × 104	6.301 × 104
Hexanal	7.22 × 105	5.751 × 105	2.889 × 104	3.529 × 105
Limonene	1.378 × 107	1.976 × 105	2.602 × 108	2.011 × 105
6-metile 5-epten-2-one	6.233 × 105	3.789 × 105	2.731 × 105	3.617 × 105
Nonanal	5.81 × 105	1.133 × 105	9.298 × 104	1.462 × 105
α-terpineol	1.098 × 106	3.145 × 106	8.024 × 106	7.628 × 105
5,6,7,7a-tetrahydro, 4,4,7a- Trimethyl 2 (4H)—benzofuranone	3.448 × 105	4.843 × 105	4.819 × 105	2.836 × 105

In total, an average of 100 volatile compounds were found for each sample analyzed, of which only seven in common to all samples.

• Cyanoacetic acid: (C₃H₃NO₂) is an organic compound that has two functional groups: COOH typical of carboxylic acids and NC with triple bond typical of nitriles. It is obtained from the treatment of chloroacetate with sodium cyanide followed by acidification or electrolysis by cathodic reduction of carbon dioxide or the anodic oxidation of acetonitrile. It is a precursor of synthetic caffeine by theophylline [33].
• Hexanal: (C₆H₁₀O) is an aldehyde. In the cell, it is contained in the cytoplasm. It has a sweet almond and honey flavor and is found in several foods including soy, cucumber, black elderberry, and black currant [33].
• Limonene: (C₁₀H₁₆) is the most widespread and most important monoterpene. It has a lemon smell and turpentine-like notes. It is obtained by steam distillation of citrus peel and pulp obtained from the production of juice [33]. It should be specified that it is present as two isomers: R-LIMONENE and D-LIMONENE. In particular, it is R-LIMONENE in the SGP, while in the others, it is D-LIMONENE, and both isomers are present in the SGM.
• 6-methyl, 5-hepten-2-one: (C₈H₁₄O) is an unsaturated ketone called sulcatone. It has a strong, greasy, green, citrus smell and tastes reminiscent of pear. It is obtained from citronella or citral oil by mixing for 12 h in aqueous solution with K2CO3 and subsequent distillation and fractionation under vacuum. It was originally identified in lemongrass; later, it was also discovered in the essential oils of lemons and geraniums. We also find this ketone in grapes, melon, peaches, avocados, cognac, mangoes, rice, olives, blueberries, and more [33].
• Nonanal: (C₉H₁₈O) is an aldehyde. It has a strong and greasy odor which develops notes of orange and rose when diluted. The fat recalls the flavor of citrus fruits. It is synthesized by the catalytic oxidation of the corresponding alcohol or by the reduction of the respective acid. In nature, we find it in orange, mandarin, lemon, and lime oils. It is also found in more than 200 foods and beverages including apples, tomatoes, rum, wine, plum, coconut, cardamom, avocado, corn oil, broccoli, milk, eggs, tea, and others [34].
• α-Terpineol: (C₁₀H₁₈O) is a monoterpenic alcohol. It has a characteristic smell of lilac with a sweet flavor reminiscent of peach. It is obtained from the hydration of the terpene or from the pentane tricarboxylic acid by cyclization or from the isoprene and methyl-vinyl-ketone. It is present in more than 150 derivatives of herbs, leaves, and flowers. Form D is found in cardamom, star anise, sage, and marjoram oil. The L form is present in lavender, lime, and cinnamon leaves. The racemic form is the eucalyptus [33,34].
• 5,6,7,7a-Tetrahydro, 4,4,7a-Trimethyl-2(4H)–benzofuranone: (C₁₁H₁₆O₂) is a heterocyclic compound. It has a coumarin and musky smell. This compound is formed from the photo-oxidation of carotene. The flavor is linked to the fruit and in particular to their point of ripeness. It is obtained from the degradation process of β-carotene in the presence of nitrogen and air. It occurs naturally in lemongrass and sweet grass oil [33].

On the other hand and regarding the GC-MS results from the black tea samples, it is possible to say that there were seven black teas subjected to GC-MS analysis, with different flavors and belonging to different origins and composition. The results obtained showed that there are four VOCs in common to all the samples (Table 4).

Table 4. Common compounds between the analyzed black teas.

Compound	TBV	EBP	EBL	EBB	SBL	TBC
Nonanal	3.739 × 105	6.164 × 104	2.278 × 105	1.129 × 105	1.041 × 105	1.198 × 105
Ammonium acetate	2.141× 106	4.641 × 104	5.945 × 105	6.868 × 105	2.162 × 105	1.023 × 105
Phenylethyl alcohol	8.167× 104	4.682 × 104	4.243 × 104	7.557 × 104	6.461 × 104	5.550 × 104
5,6,7,7a-tetrahydro, 4,4,7a- Trimethyl 2 (4H)—benzofuranone	2.784 × 105	3.536 × 104	1.473 × 105	7.543 × 104	1.073 × 105	1.053 × 105

• Nonanal: (C₉H₁₈O) is an aldehyde. It has a strong and greasy odor which develops notes of orange and rose when diluted. The fat recalls the flavor of citrus fruits. It is synthesized by the catalytic oxidation of the corresponding alcohol or by the reduction of the respective acid. In nature, we find it in orange, mandarin, lemon, and lime oils. It is also found in more than 200 foods and drinks including apples, tomatoes, rum, wine, plum, coconut, cardamom, avocado, corn oil, broccoli, milk, egg, tea, and others [34].
• Ammonium acetate: (C₂H₇NO₂) is an ammonium salt obtained from the reaction between ammonia and acetic acid. It is used to regulate acidity in food, even though the EU decided to ban its use as a food additive [33].
• Phenylethyl alcohol: (C₈H₁₀O) is an alcohol. It has a characteristic rose odor and initially a slight bitter taste. The dessert is reminiscent of peaches. It is synthesized from toluene, benzene, or styrene. It is found in esterified form in rose concentrate or distilled rose water. It is present in the essential oil of lily, narcissus, and tea leaves but not only because it has been found in more than 200 foods and drinks including peaches, grapes, coffee, tea, mushrooms, mango, kiwi, rum, whiskey, milk, butter, cheese, and more [33,34].
• 5,6,7,7a-Tetrahydro, 4,4,7a- trimethyl 2 (4H)—benzofuranone: (C₁₁H₁₆O₂) is a heterocyclic compound. It has a coumarin and musky smell. This compound is formed from the photo-oxidation of carotene. The flavor is linked to the fruit and in particular to their point of ripeness. It is obtained from the degradation process of β-carotene in the presence of nitrogen and air. It occurs naturally in lemongrass and sweet grass oil [33].

3.2. S3 Results

For each analyzed tea type (black and green), a specific matrix was created as shown in the respective PCA scores plot. Conversely, the dataset used in the first PCA was obtained by joining the three tea types considered in an initial test to check the performances of the system. In Figure 1, the results obtained from the comparative analysis of the S3 for the samples belonging to the three types of teas BWL (white), TGO (green), and TBV (black) are represented.

Figure 1. PCA analysis representing the results obtained from the comparison between black, green, and white tea.

Total explained variance reached a value of 80.11%, exhibiting a good cluster separation between the three different categories of teas. On the other hand, further exploration of the data was conducted proceeding the comparison of different green and black teas in order to explore the capacity of the instrument to distinguish the same type of tea (black or green) but with different treatments and/or aromatization. In Figure 2, the results of the analysis of SGP, SGL, and SMG for one RGTO sensor as the normalized resistance as a function of time (left) are represented; these results are more evident in the bar chart where the line on the bars is the indicated standard deviation of the mean (Figure 2b).

Figure 2. (a) Sensor response representation to the green teas RGTO sensor SnO₂ + Au and (b) mean values for the ΔR/R0 with the representation of the SD bar for the different green teas measurements.

Regarding the result obtained, applying multivariate analysis to the signals obtained from all the sensors when measuring the types of green tea is represented in Figure 3.

Figure 3. PCA analysis representing the results obtained from the comparison between green tea.

The result obtained is very satisfactory, because considering the two main components, the total explained variance enclosed in the graph reaches 87.82%. The mean accuracy achieved with k-NN (k = 5) was equal to 88.57%.

Regarding the black teas, the samples that were taken into consideration were EBP, EBL, EBB, SBL, and TBC. Figure 4a,b clearly shows the different responses of a SnO₂ RGTO sensor to the various black tea samples.

Figure 4. (a) Sensor response representation to the green teas RGTO sensor SnO₂ and (b) mean values for the ΔR/R0 with the representation of the SD bar for the different green teas measurements.

As before, PCA was then applied, and the results can be seen in Figure 5. The expected results were confirmed using PCA analysis, obtaining a good cluster separation, apart from the EBL, EBB, and EBP. The mean accuracy achieved with k-NN (k = 5) was equal to 83.45%.

Figure 5. PCA analysis representing the results obtained from the comparison between black tea.

4. Discussion

Regarding the GC-MS results from the treated green teas, an average of 100 volatile compounds were found for each sample analyzed, of which only seven in common to all samples (Table 3). As can be seen from the description, all are naturally occurring compounds in various foods and drinks, especially fruit and vegetables, except for cyanoacetic acid which is a precursor of synthetic caffeine. Volatile compounds give the tea floral, fruity, vegetable, spicy, and aromatic notes. On the other hand, regarding the GC-MS results from the black tea samples, seven teas were subjected to GC-MS analysis, with different characteristics. The results obtained showed that there were four VOCs in common to all the samples (Table 4). In general, EBP, SBL, and TBC were present in more than 100 VOCs, but in the others, about 90 have been found. The few compounds common to black teas can be explained by vanilla flavored black tea (TBV), which has fewer compounds in common with the other samples.

The four citrus-flavored samples (TBL, EBB, EBL, and SBL) have five compounds in common over the previous four: β-myrcene, D-limonene, Benzaldehyde, α-terpineol, and Carvone, which are present in citrus fruits both in essential oils and in peels. In TBV, there is a maximum peak in the direction of vanillin, which is the aromatic aldehyde that gives the vanilla aroma, which records an abundance of 8.287 × 10⁶. In general, even in black teas, as already found in green teas, volatile compounds are present in nature and impart aromas that have floral, vegetable, spicy, fruity, and aromatic notes.

Nonetheless, the results obtained throughout the use of the sensor device show a good rate of identification for the teas, since more than the 80% of the explained variance was enclosed between PC1 and PC2, and good clustering capacity can be seen in Figure 2.

Taking into consideration the response of one of the RGTO sensors for the green teas’ analysis, in particular SGP, SGL, and SMG, it is evident, in the graphs of the normalized resistance as a function of time (Figure 2a), how the samples are well separated according to the different aromatization, even if some have similar values. In fact, as can be seen in Figure 2a, the violet and blue values are more similar to each other. It looks even better in bar charts where the line on the bars is the indicated standard deviation of the mean, highlighting its reproducibility (Figure 2b).

On Figure 3, the results of the PCA analysis performed on green teas measurements is shown, and it is evident the separation between the green tea samples, in particular SGP, is far from SGL and SGM. This result is important not only because the instrument is able to recognize the aromatization tea from the pure one, separating the 2 different ones which actually slightly overlap. The result obtained is very satisfactory, because considering the two main components, the total explained variance enclosed in the graph reaches 87.82%.

On the other hand, EBP, EBL, EBB, SBL, and TBC samples were taken into consideration for the analysis of the sensors’ response. Figure 4a,b clearly shows the different responses of a SnO2 RGTO sensor to the various black tea samples. This aspect is even more evident in the bar graphs (Figure 4b). It can be observed that the samples of EBP, EBB, and EBL, all of the same brand, have a variation that is very similar to each other, while the other two (SBL and TBC) have a very different variation from the first three and are also quite different between themselves. In this case, two samples of black tea aromatized with lemon were taken into consideration but of different brands, and as can be seen in the figures, the two are very different from each other, as the variation associated with SLB tea is much greater than that of ELB tea.

PCA analysis multivariate analysis for black teas (Figure 5) confirms the previous discussed data, obtaining a good cluster separation, apart from the EBL, EBB, and EBP. These last three samples belong to the same brand and have been reported previously from the sensor response to have a lower sensor response than the other two samples of black tea and consequently are all close and partly overlapping each other, in particular EBB and EBL. This last consideration can be explained by observing the chromatograms where it can be seen how EBB and EBL are mainly characterized by limonene and linalyl acetate, while in EBP, the spectrum has a net peak in correspondence with linalyl acetate and has in general a chromatogram that appears to be richer than the previous two. In this case, the explained variability enclosed by the main components greater than the one obtained for the green teas, in fact, is 98.8%, and also in this case, the greatest variation explained is always along the PC1.

5. Conclusions

The work was based on the analysis of 11 different samples of green, black, and white tea. By applying the two techniques GC-MS and S3, the complete determination of the aromatic profile of the teas was achieved, evaluating and highlighting the similarities and differences between them so as to arrive at a discrimination based on the VOCs profile. Nearly, 100 different volatiles were identified by the mean in each tea sample. The GC-MS requires a longer and, in some ways, more elaborate analysis, while S3 allows one to obtain the result in a shorter time and more easily and user-friendly way. It can be concluded that an innovative technology such as S3 has the potential to be used from farm to fork, in food companies and in the production chain, so as to report anomalous products and prevent them from reaching the market or in any case arriving at a final product that must be excluded because it is not safe. In fact, once the anomaly is reported, it would be possible to implement a correction or exclusion of the batch from the distribution chain. It should also be emphasized that the use of this sensor technology once trained would have a positive economic impact for the production business, is also easy to use, and, thanks to the connection to the network, allows for remote data processing. In a short-term future, this technology could be applied in industrial reality, so that food safety and the producer can benefit from it. This kind of technology could be also implemented in household environments, to control not only the dried product but also the final stage of the beverage. In fact, its use is not limited only to tea but can concern any foodstuff, as already demonstrated in other studies.