Honey origin authentication using cyclic voltammetry and nanostructured carbon electrodes

doi:10.21203/rs.3.rs-1965060/v1

Download PDF

Research Article

Honey origin authentication using cyclic voltammetry and nanostructured carbon electrodes

https://doi.org/10.21203/rs.3.rs-1965060/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Honey is a sweet substance that bees produce from the nectar of flowers. The geographic origin is an intrinsic factor for the characteristics of honey, which are evaluated through physicochemical analyzes that demand high analysis time and cost. Cyclic voltammetry can be used as an alternative analytical tool for the discrimination of honey samples, being a fast, simple, and inexpensive technique. This study aimed to apply cyclic voltammetry for honey origin authentication from Ortigueira (Paraná – Brazil) using graphite electrodes modified with nanoparticles of nickel oxide, iron oxide, copper oxide, and carbon nanotubes. The voltammograms collected were treated using chemometric methods for exploratory analysis and classification. The principal component analysis (PCA) demonstrated that cyclic voltammetry had better discrimination performance than the physicochemical analysis. The best classification model (PLS-DA) was also obtained with data from cyclic voltammetry with a percentage of correct classification for the prediction set of 94.44%. On the other hand, the PLS-DA model with physicochemical data achieved 88.57% of correct classification in the prediction set. The PLS-DA model using the voltammograms obtained with the developed electrodes showed more accuracy and greater selectivity when compared to the PLS-DA model built with the physicochemical data. Furthermore, the electrochemical system developed has low cost and allows quick analyzes that can be applied to verify the authenticity of the geographical origin of honey samples.

chemometrics

metallic nanoparticles

electrochemistry

Honey is a natural substance produced by various species of bees worldwide. The product resulting from the collection and processing of nectar by bees from the vegetation around them is a dense and viscous solution consisting of sugars, proteins, enzymes, and amino acids (Minden-Birkenmaier and Bowlin 2018).

The world’s consumption of honey from the earliest times is based on recognizing its beneficial effects on human health. This consumption is associated with honey’s physicochemical, sensory, and nutritional characteristics. The predominant floral source in the production area and geographic location will result in different attributes, mainly in honey’s color, flavor, and aroma. Therefore, the regional characterization and the establishment of authentication methodologies are of great importance, considering each region’s botanical diversity and climatic variation (Castro-Vázquez et al. 2008; Welke et al. 2008; Baroni et al. 2009).

Brazil has a high potential to supply the honey market. However, many challenges still need to be overcome, involving improvements in the sector’s technology, the definition of quality standards, healthy control, and private labels that add value to the product, thus increasing consumption, the domestic market, and the expansion of the foreign market (Buainain and Batalha 2007). Furthermore, in Brazil, honey production stands out in agrarian reform settlements and small landowners who produce this food to increase family income (Ferreira Dias et al. 2013).

In Brazil, the region of Ortigueira in the state of Paraná is widely recognized as a producer of high-quality honey that can be attributed to the characteristics of the flora of this territory, the most prevalent being wild assapeixe (Vernonia polysphaera) and capixingui (Croton floribundus) (Scholz et al. 2020). In the productive context, Paraná occupies the third position in Brazil among the producing states of honey and derivatives, with the city of Ortigueira as the leading producer. In 2013, Ortigueira started the certification of a geographical indication for honey produced locally, proving the potential and recognition of its production. Documents proving its authenticity are required through certification. Therefore, in the case of honey production, its origin must be confirmed through physicochemical and sensory analysis (Woodcock et al. 2007; Viuda-Martos et al. 2010). Due to the characteristics and history of the local beekeeping industry, the city of Ortigueira received certification of origin, in May 2015, by the Instituto Nacional da Propriedade Industrial (INPI) with the number BR412013000002-0.

Various instrumental methods, such as high-performance liquid chromatography (HPLC) (Zhang et al. 2019), gas chromatography with mass spectrometry (GC-MS) (Rissato et al. 2007), and rheological properties (Wei et al. 2010), are used to determine botanical origins and to verify the authenticity of honey samples. Most of these analytical methods are time-consuming and involve sample preparation steps. For this reason, some attempts have already been made to identify the type of honey by electrochemical analysis methods such as electronic tongues (Wei et al. 2009; Major et al. 2011).

Voltammetric systems, on the other hand, require little sample preparation, are fast, have relatively simple instrumentation, are selective and sensitive, and have a low signal-to-noise ratio (Bagnasco et al. 2014). Generally, an electrochemical cell consisting of three electrodes is used; the working electrode, the counter electrode (typically platinum), and the reference electrode (generally Ag/AgCl). The cell coupled to a potentiostat allows recording voltammograms using different voltammetric techniques (e.g., cyclic voltammetry, square wave voltammetry, and differential pulse voltammetry). Voltammetric systems are limited in their applicability to redox-active substances. However, the range of applications can be increased by modifying the surface of working electrodes with various materials, allowing for increased sensitivity and selectivity for multiple compounds (Baig et al. 2019). There is a wide variety of working electrodes and, therefore, several applications, such as the analysis of white wines made with different grape varieties using square wave voltammetry (Rodríguez-Méndez et al. 2008); predicting bisulfite concentrations in wine samples using pulse voltammetry (Labrador et al. 2009); evaluation of water quality parameters using cyclic voltammetry (Campos et al. 2012); wine discrimination based on barrel maturation and prediction of global scores assigned by a sensory panel (Cetó et al. 2017). No studies were found applying cyclic voltammetry to authenticate honey, including those from the Ortigueira region. Therefore, in this work, a cyclic voltammetry system developed by the authors (potentiostat and nanostructured working electrodes) was applied to authenticate the origin of honey from Ortigueira-PR.

2.1. Samples

The Instituto de Desenvolvimento Rural do Paraná (IDR – Paraná) provided one hundred and seven samples of honey from twenty-three cities distributed in the north, central-east, and south regions of the state of Paraná and from two cities in the northeast of the State of Santa Catarina, collected in 2010 and 2011.

The samples were divided into those collected in the Ortigueira region (O) and those collected outside the Ortigueira region (E). The samples were stored in polyethylene bottles and kept at -18°C until the analysis was performed. The samples were kept at this temperature to reduce degradation reactions (da Silva et al. 2016).

In the Ortigueira region, the humid subtropical mesothermal climate prevails. In the summer, the temperature is moderate; in the winter, there are frosts of varying intensities. Based on the climatic adaptation of each species in these microregions, different combinations of botanical families can be found: assapeixe, capixingui, and eucalyptus are the main plants mentioned by beekeepers. The beekeepers indicated the botanical origin based on the beekeepers’ experience and the honey color. In the set of samples evaluated, samples belonging to species of the botanical families Asteraceae, Mimosaceae, Poaceae, and Euphorbiaceae were identified by pollen analyses (melissopalynology analysis). Regarding the colors, the kinds of honey were classified as extra light amber, extra white amber, light amber, white, extra light, amber, and dark (Scholz et al. 2020).

2.2. Physicochemical analysis

The physicochemical analyses performed on the honey samples include moisture, diastase activity, hydroxymethylfurfural, electrical conductivity, reducing sugar, total sugar, free acidity, lactonic acidity, total acidity, pH, proline, and color using Pfund and CIELab scale. All analyses are described in a previously published work (Scholz et al. 2020).

2.2 Nanostructured working electrodes

A working electrode C60 was made with graphite (60.00%) and epoxy resin (40.00%). Graphite electrodes (59.75%) and epoxy resin (39.75%) were also made with metallic nanoparticles (0.5%) of nickel oxide (Ni), iron oxide (Fe), copper oxide (Cu), and carbon nanotubes (NT). A commercial glassy carbon (CV) electrode (Analion) was used for comparison with the built electrodes. The nanoparticle synthesis, their morphological characterization, and the preparation of the electrodes are described by Santos (2022) and Possebon (2019). Multi-walled carbon nanotubes (Sigma-Aldrich, reference 65258), graphite (Dinâmica), and epoxy resin (Araldite) were used.

2.3 Potentiostat and electrochemical cell

For the voltammetric analysis, an electrochemical cell consisting of three electrodes, a working electrode, a platinum counter electrode (5 cm x 0.5 mm – Dinâmica), and an Ag/AgCl reference electrode (Analion, Brazil) was used.

The electrochemical cell was connected to a low-cost potentiostat designed to be coupled to a RaspberryPi single-board computer. The PotenciosPi software, developed in Python, was used to control the voltammetry parameters and for data acquisition. More details about the potentiostat and its construction are available in Santos (2022).

2.4 Sample preparation and voltammetry

Samples were prepared by dissolving 1 g of honey in 25 mL of electrolyte solution, H₂SO₄ (Sigma) 0.1 mol L^− 1, and KCl (Vetec) 0.1 mol L^− 1 (Possebon 2019) prepared with type 1 water (Millipore, Synergy). The diluted samples were evaluated by cyclic voltammetry in the range between − 1.0V to + 1.0V, with a resolution of 10 mV, a scan speed of 100 mV s^− 1, and three consecutive scans. After the analysis of each sample, the working electrode was polished with movements like an eight on sandpaper and then rinsed with type 1 water and dried on absorbent paper (Elgrishi et al. 2018).

The voltammetry parameters and the amount of honey used were defined in preliminary tests. The signal repeatability after the polishing step of the working electrode was also tested in these preliminary experiments.

2.5 Multivariate data analysis

MATLAB R2022a software was used to apply exploratory and classification methods to analyze the collected data. The analyses were carried out for the cyclic voltammetry data and for the physicochemical data for honey samples.

First, the signals obtained for each electrode were transferred to MATLAB and plotted for general visualization of the voltammogram’s behavior. For multivariate analysis, the average of the last two scans of each sample was used.

For exploratory analysis, principal component analysis (PCA) was applied using the autoscaled matrix for physicochemical data (because the variables have different scales) and the mean-centered matrix of voltammograms (variables on the same scale). The criterion of eigenvalue greater than unity for autoscaled data and the criterion of accumulated variance greater than 95% for mean-centered data were used to define the number of principal components (Bona et al. 2018).

As a supervised classification method, the partial least squares method with discriminant analysis (PLS-DA) was used (Brereton 2018). The data from the physicochemical analyzes of the honey samples were used in a PLS-DA model to discriminate the honey samples from the Ortigueira region (O) from the honey samples from outside places (E). For physicochemical data, autoscaling and pareto were used as preprocessing. For cyclic voltammetry, mean-centering, autoscaling, and Pareto were tested as preprocessing.

The Kennard-Stone algorithm (Westad and Marini 2015) separated the samples into calibration and prediction sets for cyclic voltammetry and physicochemical data. The calibration set consisted of 2/3 of the samples (71) and the prediction set with the remaining 1/3 (36). A 5-fold cross-validation was used to estimate each model’s most appropriate number of latent variables (LVs). The number of LVs was set to obtain the smallest root-mean-square error of cross-validation (RMSECV) and the largest percentage of correct classification of cross-validation (PCCCV).

Outlier evaluations were also performed with the Q (square residuals of the X matrix) and T² (hotelling) statistics in the sample set. Samples presenting Q and T² values higher than the critical values are considered outliers with 95% accuracy (Santana et al. 2020).

Figure 1 shows the variation in the shape of the voltammograms as a function of the working electrode used to analyze the honey samples. The mean voltammogram for each honey class, Ortigueira (O) and outside places (E), is shown in Figure 2. It is possible to observe that the working electrodes C60, Fe, and Ni presented more differences in the mean signal intensity associated with each honey class. For all used electrodes, the Ortigueira samples present lower current intensities.

Table 1 shows the variance represented at each PC for PCA of physicochemical data and PCA of voltammetry data for each working electrode. The number of PCs that accumulated a variance greater than 95% ranged from 4 (graphite electrode) to 6 (iron and nanotube electrodes). Two PCs were sufficient to result in an accumulated variance greater than 80% for all data sets. For the physicochemical data, five PCs accumulated 77.19% of the variance, but from the sixth PC onwards, the eigenvalues were lower than unity. Therefore, only the first five PCs were kept considering the eigenvalue criterion for autoscaled data.

Figure 3a shows the score plot obtained for the physicochemical analyses; it is impossible to observe a clear separation between Ortigueira samples (O) and outside places (E). In PC1 and PC4, it is possible to observe a predominance of O samples in the negative quadrant and E samples in the positive quadrant. Figure 3b shows the biplot of PC1 and PC4; the outside samples present higher values of hydroxymethylfurfural (HMF), absorbance at 635 nm (C635), total acidity (AcT), lactonic acidity (AcLa), and a*. The samples from Ortigueira showed higher luminosity (L*) values, total sugar (AT), reducing sugar (AR), diastase activity (AD), b*, electrical conductivity (CE), and proline. A previously published work offers a more in-depth discussion of the physicochemical data (Scholz et al. 2020).

Figure 4 shows bidimensional score plots for the voltammograms obtained for working electrodes used in this work. For the graphite electrode, it was possible to identify a separation of Ortigueira samples in the positive quadrant of PC1 and the negative quadrant of PC2 (Figure 4a). The iron electrode voltammograms (Figure 4b) separate the samples mainly in PC2, with the Ortigueira samples in the negative quadrant and the outside samples in the positive quadrant. The nickel electrode voltammograms (Figure 4c) show a separation of the samples in PC2, with the Ortigueira samples in the negative quadrant and the outside samples in the positive quadrant.

In the other score plots (Supplementary Material), it was impossible to visualize a clear separation of the samples. It should be highlighted that, according to the PCA results, the commercial glassy carbon electrode presented lower discrimination of the samples concerning graphite, iron, and nickel electrodes. In addition, a greater separation of Ortigueira samples from outside samples can be seen in the voltammograms of these electrodes compared to the PCA of the physicochemical data.

The loading plots for PC1 and PC2 of the graphite, iron, and nickel electrodes (Figure 5) confirmed that the external samples present higher current intensity, mainly in the region of 0.6 – 1.0 V. This result is in agreement with the plots of the average signal of the electrodes for each honey class (Figure 2). It has been reported that electrodes containing copper nanoparticles respond to the presence of carbohydrates by exhibiting characteristic peaks due to irreversible oxidation at approximately 0.70 V. It has been suggested that copper nanoparticles play a prominent role in glucose and fructose oxidation due to their electrocatalytic property (Santos et al. 2016). However, in our results, the copper electrode did not discriminate the samples with the same performance observed for graphite, iron, and nickel electrodes.

After exploratory analysis, classification models were built using the PLS-DA method. The figures of merit for the best model obtained for each electrode and physicochemical analysis are shown in Table 1. The best PLS-DA model for physicochemical analyses was obtained with Pareto normalization, three latent variables, a correct classification ratio of 88.57%, and an RMSEP of 0.3081 for the prediction set. Voltammograms obtained with graphite, iron, and nickel electrodes resulted in PLS-DA models with similar or superior performance to the model obtained using the physicochemical data. On the other hand, the other electrodes did not present good classification results. A hypothesis to be confirmed in future works is that the concentration of nanoparticles was insufficient to increase the electrode’s sensitivity. The best PLS-DA model using cyclic voltammetry data was obtained with data from the graphite electrode with two latent variables, obtaining a correct classification ratio of 94.44% and RMSEP of 0.2854 for the prediction set.

Table 3 shows other figures of merit for the PLS-DA model using voltammograms obtained with graphite electrode and the PLS-DA model using the physicochemical data. The PLS-DA model using the voltammograms obtained with the graphite electrode showed greater accuracy and selectivity. The physicochemical data resulted in a model of high sensitivity but low selectivity resulting in high false-positive rates.

The significant variability of the samples in terms of color and flora leads to the significant variability of chemical composition. Even in this scenario, the classification performance for the graphite electrode was good. Also noteworthy is the speed (about 5 minutes per sample) of the analysis using cyclic voltammetry and the low cost for producing graphite electrodes compared to the analysis time and cost for executing the physicochemical analysis. Therefore, the proposed methodology proved to be adequate for the origin authentication of honey from Ortigueira.

It was possible to authenticate the origin of Ortigueira honey by applying cyclic voltammetry with nanostructured carbon electrodes. Using graphite, iron, and nickel electrodes, PCA demonstrated the highest discriminatory capacity for voltammetry data. The PLS-DA models showed that the correct classification ratio obtained with the glassy carbon electrode was very close to that obtained with the physicochemical analysis. This positive result proved the cyclic voltammetry’s discrimination potential. The best results were obtained with the graphite electrode. Furthermore, the correct classification ratio for the prediction set obtained with the graphite electrode was higher than that obtained with the physicochemical analysis. It is important to emphasize the short preparation time required for constructing the graphite electrode and its low cost. Together with the cyclic voltammetry, the built system performs very fast analyzes. In addition, the proposed methodology can be tested for other food products whose market value is associated with the cultivation and production region.

Funding:

This work was supported by Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPQ, grant ID 312595/2021-2).

Acknowledgments:

The authors wish to thank Central Analítica Multiusuário at UTFPR Campus Campo Mourão (CAMulti-CM).

Bagnasco L, Cosulich ME, Speranza G, et al (2014) Application of a voltammetric electronic tongue and near infrared spectroscopy for a rapid umami taste assessment. Food Chemistry 157:421–428. https://doi.org/10.1016/j.foodchem.2014.02.044
Baig N, Sajid M, Saleh TA (2019) Recent trends in nanomaterial-modified electrodes for electroanalytical applications. TrAC Trends in Analytical Chemistry 111:47–61. https://doi.org/10.1016/j.trac.2018.11.044
Baroni M, Arrua C, Nores M, et al (2009) Composition of honey from Córdoba (Argentina): Assessment of North/South provenance by chemometrics. Food Chemistry 114:727–733. https://doi.org/10.1016/j.foodchem.2008.10.018
Bona E, Março PH, Valderrama P (2018) Chemometrics Applied to Food Control. In: Holban AM, Grumezescu AM (eds) Food Control and Biosecurity, 1st edn. Elsevier, London, pp 105–133
Brereton RG (2018) Chemometrics: Data Driven Extraction for Science, 2nd edn. John Wiley & Sons, Hoboken, NJ, USA
Buainain AM, Batalha MO (2007) Cadeias produtivas de flores e mel. IICA: MAPA/SPA, Brasília
Campos I, Alcañiz M, Aguado D, et al (2012) A voltammetric electronic tongue as tool for water quality monitoring in wastewater treatment plants. Water Research 46:2605–2614. https://doi.org/10.1016/j.watres.2012.02.029
Castro-Vázquez L, Díaz-Maroto MC, González-Viñas MA, et al (2008) Influence of Storage Conditions on Chemical Composition and Sensory Properties of Citrus Honey. Journal of Agricultural and Food Chemistry 56:1999–2006. https://doi.org/10.1021/jf072227k
Cetó X, González-Calabuig A, Crespo N, et al (2017) Electronic tongues to assess wine sensory descriptors. Talanta 162:218–224. https://doi.org/10.1016/j.talanta.2016.09.055
da Silva PM, Gauche C, Gonzaga LV, et al (2016) Honey: Chemical composition, stability and authenticity. Food Chemistry 196:309–323. https://doi.org/10.1016/j.foodchem.2015.09.051
Elgrishi N, Rountree KJ, McCarthy BD, et al (2018) A Practical Beginner’s Guide to Cyclic Voltammetry. Journal of Chemical Education 95:197–206. https://doi.org/10.1021/acs.jchemed.7b00361
Ferreira Dias T, Nunes EM, de Lima Torres F, et al (2013) O Programa de Aquisição de Alimentos da Agricultura Familiar (PAA) como estratégia de inserção socioeconômica: o caso do Território da Cidadania Sertão do Apodi (RN). Revista Brasileira de Gestão e Desenvolvimento Regional 9:100–129
Labrador RH, Olsson J, Winquist F, et al (2009) Determination of Bisulfites in Wines with an Electronic Tongue Based on Pulse Voltammetry. Electroanalysis 21:612–617. https://doi.org/10.1002/elan.200804457
Major N, Marković K, Krpan M, et al (2011) Rapid honey characterization and botanical classification by an electronic tongue. Talanta 85:569–574. https://doi.org/10.1016/j.talanta.2011.04.025
Minden-Birkenmaier B, Bowlin G (2018) Honey-Based Templates in Wound Healing and Tissue Engineering. Bioengineering 5:46. https://doi.org/10.3390/bioengineering5020046
Possebon G (2019) Análise Voltamétrica Cíclica: Uma ferramenta para a análise qualitativa do mel. Instituto Politécnico de Bragança
Rissato SR, Galhiane MS, de Almeida M v., et al (2007) Multiresidue determination of pesticides in honey samples by gas chromatography-mass spectrometry and application in environmental contamination. Food Chemistry 101:. https://doi.org/10.1016/j.foodchem.2005.10.034
Rodríguez-Méndez ML, Parra V, Apetrei C, et al (2008) Electronic tongue based on voltammetric electrodes modified with materials showing complementary electroactive properties. Applications. Microchimica Acta 163:23–31. https://doi.org/10.1007/s00604-007-0907-8
Santana F, Souza A, Almeida M, et al (2020) Experimento didático de quimiometria para classificação de óleos vegetais comestíveis por espectroscopia no infravermelho médio combinado com análise discriminante por mínimos quadrados parciais: um tutorial, parte V. Química Nova 43:371–381. https://doi.org/10.21577/0100-4042.20170480
Santos FCU, Paim LL, Luiz Da Silva J, Stradiotto NR (2016) Electrochemical determination of total reducing sugars from bioethanol production using glassy carbon electrode modified with graphene oxide containing copper nanoparticles. Fuel 163:. https://doi.org/10.1016/j.fuel.2015.09.046
Santos LD (2022) Desenvolvimento de eletrodos de grafite modificados com nanopartículas e emprego na quantificação de sacarose. Universidade Tecnológica Federal do Paraná
Scholz MB dos S, Quinhone Júnior A, Delamuta BH, et al (2020) Indication of the geographical origin of honey using its physicochemical characteristics and multivariate analysis. Journal of Food Science and Technology 57:1896–1903. https://doi.org/10.1007/s13197-019-04225-3
Viuda-Martos M, Ruiz-Navajas Y, Zaldivar-Cruz JM, et al (2010) Aroma profile and physico-chemical properties of artisanal honey from Tabasco, Mexico. International Journal of Food Science & Technology 45:1111–1118. https://doi.org/10.1111/j.1365-2621.2010.02243.x
Wei Z, Wang J, Liao W (2009) Technique potential for classification of honey by electronic tongue. Journal of Food Engineering 94:260–266. https://doi.org/10.1016/j.jfoodeng.2009.03.016
Wei Z, Wang J, Wang Y (2010) Classification of monofloral honeys from different floral origins and geographical origins based on rheometer. Journal of Food Engineering 96:469–479. https://doi.org/10.1016/j.jfoodeng.2009.08.028
Welke JE, Reginatto S, Ferreira D, et al (2008) Caracterização físico-química de méis de Apis mellifera L. da região noroeste do Estado do Rio Grande do Sul. Ciência Rural 38:1737–1741. https://doi.org/10.1590/S0103-84782008000600038
Westad F, Marini F (2015) Validation of chemometric models – A tutorial. Analytica Chimica Acta 893:14–24. https://doi.org/10.1016/j.aca.2015.06.056
Woodcock T, Downey G, Kelly JD, O’Donnell C (2007) Geographical Classification of Honey Samples by Near-Infrared Spectroscopy: A Feasibility Study. Journal of Agricultural and Food Chemistry 55:9128–9134. https://doi.org/10.1021/jf072010q
Zhang Y, Li XQ, Li HM, et al (2019) Antibiotic residues in honey: A review on analytical methods by liquid chromatography tandem mass spectrometry. TrAC Trends in Analytical Chemistry 110:344–356. https://doi.org/10.1016/j.trac.2018.11.015

Table 1. Captured variance for each PC and each data set.

Data Set	PC1	PC2	PC3	PC4	PC5	PC6	Sum
Physicochemical analysis	31.14	15.69	12.28	10.20	7.87	-	77.19
Glassy carbon electrode (CV)	72.62	9.81	5.64	5.03	2.82	-	95.92
Graphite electrode (C60)	78.38	12.87	3.04	1.62	-	-	95.91
Copper nanoparticles electrode (Cu)	75.78	10.74	4.43	2.48	1.61	-	95.04
Iron nanoparticles electrode (Fe)	65.26	12.67	8.64	5.73	1.82	1.30	95.42
Carbon nanotubes electrode (NT)	71.73	11.60	6.83	2.56	1.86	1.15	95.73
Nickel nanoparticles electrode (Ni)	56.63	27.61	5.78	4.13	2.26	-	96.41

Table 2. PLS-DA models for classification of honey using physicochemical data and data obtained by cyclic voltammetry.

Electrode	Normalization	LV	RMSEC	PCCC	AUCC	RMSEP	PCCP	AUCP	VARX	VARY	AC	AP
*Physical-Chemical Data*
-	Pareto	3	0.3855	81.43	0.8059	0.3081	88.57	0.9394	0.7223	0.3103	70	35
*Cyclic Voltammetry*
C60	Mean centered	2	0.3300	83.10	0.9257	0.2854	94.44	0.9722	0.9015	0.5026	71	36
Cu		2	0.4127	76.05	0.7717	0.4243	69.44	0.8229	0.8084	0.2222	71	36
CV		6	0.3824	80.28	0.8641	0.3930	83.33	0.8542	0.9467	0.3324	71	36
Fe		3	0.3029	88.73	0.9746	0.3602	88.89	0.9028	0.9365	0.3742	71	36
NT		6	0.3598	83.09	0.8913	0.4057	75.00	0.8542	0.9518	0.4089	71	36
Ni	Pareto	8	0.2740	90.91	0.9783	0.3096	90.91	0.9130	0.9576	0.6445	66	33

*LV – Latent variables; RMSEC – Square root of the mean squared error of calibration; PCCC – Percentage of correct calibration classification; AUCC – Area under the ROC curve for calibration; RMSEP – Square root of the mean squared error of forecast; PCCP – Percentage of correct prediction classification; AUCP – Area under the ROC curve for the prediction set; VARX – Accumulated variance for matrix X; VARY – Accumulated variance for matrix Y; AC – Calibration samples; AP – Prediction samples.

Table 3. Figures of merit for the PLS-DA models using the voltammograms obtained with the graphite electrode and physicochemical data.

Class	Confusion matrix		Accuracy (%)	Sensibility (%)	Selectivity (%)	False positive rate (%)
Class	O	E	Accuracy (%)	Sensibility (%)	Selectivity (%)	False positive rate (%)
Calibration set – Electrode C60
O	39	2	84.51	81.25	91.30	8.70
E	9	21	84.51	91.30	81.25	18.75
Prediction set – Electrode C60
O	23	1	94.44	95.83	91.67	8.33
E	1	11	94.44	91.67	95.83	4.17
Calibration set – Physicochemical analysis
O	45	10	81.43	93.75	54.55	45.45
E	3	12	81.43	54.55	93.75	6.25
Prediction set – Physicochemical analysis
O	24	4	88.57	100.00	63.64	36.36
E	0	7	88.57	63.64	100.00	0.00

No competing interests reported.

Materialsuplementar.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Honey origin authentication using cyclic voltammetry and nanostructured carbon electrodes

Status:

Version 1

Abstract

Figures

1. Introduction

2. Material And Methods

2.1. Samples

2.2. Physicochemical analysis

2.2 Nanostructured working electrodes

2.3 Potentiostat and electrochemical cell

2.4 Sample preparation and voltammetry

2.5 Multivariate data analysis

3. Results And Discussion

4. Conclusion

Declarations

Funding:

Acknowledgments:

References

Tables

Additional Declarations

Supplementary Files

Status:

Version 1