Determination of the phenolic compounds profile by HPLC applied to the identification of adulteration markers of roasted and ground coffee by adding coffee husksABSTRACT
Coffee is a beverage that is widely consumed throughout the world, especially in Brazil. Its high market value makes it a prime target for fraud. To combat this fraud, different methods have been developed and improved over the years to detect potential adulterants in coffee and guarantee its authenticity. Nevertheless, these methods still pose a challenge. Therefore, this work aimed to identify adulteration markers in roasted and ground Arabica coffee using high-performance liquid chromatography (HPLC) to determine the flavonoid and phenolic acid profile. Samples of Arabica coffee beans and defective coffee beans were adulterated with coffee husks by preparing mixtures in different proportions (97:3, 90:10, 80:20 % w/w), which were then subjected to chromatographic and chemometric analyses, including principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA). Using the phenolic compounds associated with chemometric methods, protocatechuic acid was identified as a marker of adulteration in roasted and ground coffee with Arabica coffee husks, even at low adulterant concentrations (3 %).
Keywords:
Arabica coffee; chemometric methods; flavonoids; food fraud; phenolic acids
Introduction
Coffee is consumed worldwide and is especially important for Brazil, which is the largest global producer and exporter of coffee. In 2022, Brazil exported the equivalent of 39.4 million 60 kg bags of coffee. According to the National Supply Company, in 2023 Brazilian coffee production reached 55.1 million processed bags.
Due to its high market value, coffee is the target of many fraudulent acts, which use different adulterants such as coffee residues (e.g., husks). Since they are produced in large volumes during coffee processing, the husks are one of the residues identified as a major problem, as they are widely used for roast and ground coffee fraud (Moore et al., 2012; Tavares et al., 2012; Ferreira, et al., 2021; Couto et al., 2023).
The quality of coffee is a determining factor in its acceptance. Defective coffee beans are present in coffee produced in Brazil mainly due to the type of harvest. The presence of black, immature and sour beans (BIS) is the main defect found, resulting from late or premature harvest, producing a drink of inferior quality (Morais et al., 2007; Ramos et al., 2009).
In Brazil, microscopy is the official method for identifying fraud in roasted and ground coffee. This technique consists of using a stereoscopic microscope to visually analyze samples on microscopic slides to detect impurities and foreign matter (Mendes et al., 2016). The method is slow and costly, requires preliminary treatments and highly trained analysts, and its results also need to be compared with images available in the literature (Assis et al., 2020).
Over the years, different methods have been developed and improved to detect possible adulterants in coffee and guarantee its authenticity. High-performance liquid chromatography (HPLC) is a technique widely used to analyze the authenticity of foods. It is, capable of detecting and quantifying organic compounds with different polarities and molecular weights in very low concentrations, allowing for the identification of chemical markers as well as a fingerprinting approach to coffee authentication (Martins et al., 2018; Ferreira et al., 2021; Couto et al., 2023).
High-performance liquid chromatography has already been used to identify adulteration in roasted and ground coffee samples through carbohydrate analysis (Domingues et al., 2014; Pauli et al., 2014). However, studies in the literature on the use of phenolic compounds profiles for this purpose are needed. Therefore, this work aimed to identify adulteration markers of roasted and ground Arabica coffee with coffee husks by determining the phenolic profile using HPLC analysis.
Materials and Methods
The experiment was conducted at the Liquid Chromatography Laboratory of Embrapa Agroindústria de Alimentos, located in the municipality of Rio de Janeiro, in the Rio de Janeiro state, Brazil (23°00’02.9" S, 43°34’55.6" W, altitude 5 m).
Chemicals
Methanol, acetonitrile, acetone, ethyl acetate, all HPLC grade, hydrochloric acid (37 %), phosphoric acid (85 %), and sodium hydroxide (NaOH) were purchased from Tedia™. Ascorbic acid, ethylenediaminetetraacetic acid (EDTA), flavonoids and phenolic acid standards were purchased from Sigma-Aldrich. Ultrapure water (0.054 μS cm–1) was obtained using a Milli-Q system from Millipore™.
Samples
Samples of Arabica coffee (Coffea arabica L.) beans (A), Arabica defective coffee beans (A.BIS), and Arabica coffee husks (A.H) were obtained directly from producers in Cambuquira, in the southern region of the state of Minas Gerais, Brazil (21°51’00.0" S, 45°17’52.2" W, altitude 950 m).
Roasting and grinding
All samples were roasted in a Gene Café CBR 101 roaster using the drum rotation stirring method. The A and A.BIS (250 g of each sample) were roasted at 240 °C for 15 min resulting in a medium-dark roast. The A.H (250 g) samples were roasted at 240 °C for 11 min. Subsequently, the samples were ground in an IKA model A11 Basic analytical mill.
Preparation of adulterated samples
The adulterated samples (in triplicate) were prepared by mixing A and A.BIS or A.H in different proportions (97:3, 90:10, 80:20 % w/w).
Analysis of flavonoids and phenolic acids
The phenolic acids and flavonoids were extracted in two stages using the method proposed by Nascimento et al. (2017) with modifications (Figure 1) to characterize the coffee.
Flowchart of the extraction process to obtain free and hydrolyzed phenolic acids. N2 = nitrogen gas; HPLC = high-performance liquid chromatography; EDTA = ethylenediaminetetraacetic acid; HCl = chloride acid.
The samples were submitted to extraction with 4 mL of methanol:water acidified with hydrochloric acid 6 M (50:50; v/v; pH 2) followed by mechanical stirring for 1 h and centrifugation at 3,622 g for 10 min. Supernatant one was collected and reserved. Next, 4 mL of an acetone:water solution (70:30; v/v) was added to the solid residue and the stirring and centrifugation steps were repeated. Supernatant two was collected and reserved. Finally, 3 mL of both supernatants were mixed, and an aliquot was transferred to 1.5 mL vials for HPLC analysis.
To the solid residue, 5 mL of a 2 M NaOH solution containing 1 % ascorbic acid and 10 mM of EDTA were added and then alkaline hydrolysis was carried out for 60 min at 60 °C. After this, 1.5 mL of 6 M chloride acid (HCl) was added for acid hydrolysis. This solution was vortexed for 10 s, left to cool to room temperature, and then centrifuged at 3,622 g for 10 min. The supernatant obtained was collected and 6.5 mL of ethyl acetate was added. Each sample was then vortexed for 30 s, followed by ultrasound treatment for 5 min and then centrifuged at 3,622 g for 5 min. The supernatant obtained was reserved, and the procedure repeated with a new aliquot of ethyl acetate. The organic fraction was dried under nitrogen gas (N2) flow and solubilized in 80 % methanol in water for chromatographic analysis.
The two fractions (free and hydrolyzed phenolic acids) were analyzed in an Alliance Waters™ model 2690/5 high-performance liquid chromatograph coupled to a Waters™ model 2996 photodiode array detector (200 at 600 nm for scanning and 270, 290, 310, 325 and 370 nm for quantification), with a Thermo Hypersil BDS C18 column (100 × 4.6 mm × 2.4 μm), at 40 °C in gradient elution mode, composed of an aqueous solution of 0.15 % phosphoric acid (phase A) and acetonitrile (phase B), with flow rate of 1.0 mL min–1, injection volume of 5 μL and run time of 30 min. After 12 min, the acetonitrile concentration was increased to 12 %, at 18 min to 20 %, and at 20 min to 50 % acetonitrile. The acetonitrile concentration was maintained at 5 % for 25 min and then returned to the initial condition (5 %). The flavonoids (quercetin) and phenolic acids (caffeic acid, ferulic acid, feruloylquinic acid, gallic acid, p-coumaric acid, protocatechuic acid, synaptic acid, syringic acid, vanillic acid, 3-caffeoylquinic acid, 4-caffeoylquinic acid, 5-caffeoylquinic acid, 3,4-dicaffeoylquinic acid, 3,6-dicaffeoylquinic acid, 4,5-dicaffeoylquinic acid and 4-hydroxybenzoic acid) were quantified by external standardization through the construction of analytical curves according to commercial analytical standards.
Statistical analysis
Data processing was carried out using one-way ANOVA (analysis of variance) followed by the Tukey multiple mean comparison test, considering a significance level of 5 %. Next, multivariate techniques were applied, namely, principal component analysis (PCA), which is an unsupervised method and after data imputation and standardization to avoid the influence of the different units. A supervised method called partial least squares discriminant analysis (PLS-DA) was also used. Finally, heatmaps were generated, where sample grouping was carried out by hierarchical clustering of principal components. All statistical calculations were performed using R version 4.3.1.
Results
The free phenolic profiles of the A and A.BIS samples were similar (Table 1). On the other hand, significant differences were detected (p < 0.05) in the profile of the A.H, which presented protocatechuic acid and ferulic acid. In addition, chlorogenic acid, its derivatives, and caffeine were detected in all samples since they are part of the composition of coffee.
Phenolic profile in free fraction of coffee, defective coffee, coffee husks, and adulterated samples.
The highest levels of acids present were caffeoylquinic acid (CQAs), with the 5-CQA isomer, better known as chlorogenic acid, being the most abundant, followed by 4-CQA and 3-CQA. The levels found for 3-CQA, 4-CQA, caffeic acid, and 4-hydroxybenzoic acid had no mutual differences (p > 0.05), either for the A or the A.BIS samples. The levels found for 5-CQA were similar (p > 0.05) between A.BIS and A.H. The same occurred with 3,4-dicaffeoylquinic acid (3,4-diCQA) and 4,5-diCQA. In turn, feruloylquinic acids (FQAs) and 3,6-diCQA had similar levels (p > 0.05) between A and A.H samples.
The values of CQAs (3-CQA, 4-CQA, and 5-CQA), and the dicaffeoylquinic acids (diCQAs) 3,4-diCQA and 4,5-diCQA, from the roasted coffee beans in the present study are in accordance with Farah et al. (2005). Nine chlorogenic acid isomers were identified for roasted coffee samples from nine different countries, although none were from Brazil (Moon et al., 2009). The levels of the 3-CQA isomer were lower than those found in the present study, while the values of 5-CQA were higher. The values of isomers 3,4-diCQA and 4,5-diCQA were similar to those in the present study. We found two of the three main isomers of diCQAs, 3-4 and 4-5. Values for 3,6-diCQA were not found in the literature and may be a possible new isomer of diCQAs in roasted and ground coffee.
After hydrolysis, it was possible to obtain the characterization of bound phenolic compounds. The profile of each sample became more specific, with a greater difference between the A and A.BIS profiles. Furthermore, new phenolics were identified, such as p-coumaric acid, present in the three samples, and gallic, vanillic and syringic acids present in the husks (Table 2). As for CQAs, only 3-CQA and 5-CQA were also detected. The 5-CQA was detected in very low concentrations in A and A.BIS, and 3-CQA detected only in A while 3,4-diCQA was detected only in A.H and 4,5-diCQA only in A. The 3,6-diCQA isomer and FQAs were not detected.
Phenolic compounds in the hydrolyzed fraction of coffee, defective coffee, coffee husks and adulterated samples.
As regards coffee husks, Silva et al. (2021) identified 5-CQA, gallic acid, and caffeic acid in their samples, the first having higher concentrations (1.66 to 33.71 mg 100 g–1), though both had values lower than those found in our study. Gallic acid had 0.05 to 6.21 mg 100 g–1 levels, values similar to those in this study. Caffeic acid also presented lower levels than those found by ourselves.
Our caffeine values are compatible with the values found by Cangussu et al. (2021) of 618 mg 100 g–1, in samples composed by 80 % peel and 20 % pulp, obtained from Bourbon Arabica coffee. As for the levels of 5-CQA, the authors found levels of 121.55 mg 100 g–1, lower than those in the present study. As regards protocatechuic acid, Cangussu et al. (2021) found 28.24 mg 100 g–1, a higher level than we found. Ferulic acid has been identified in other studies; however, no quantitative data are available.
Discussion
Several studies have applied HPLC to identify adulteration of roasted and ground coffee, generally focusing on detecting one or more analytes in samples. Compounds such as carbohydrates (oligosaccharides and monosaccharides), trigonelline and/or nicotinic acid have been used as markers to identify coffee adulterants (Domingues et al., 2014; Pauli et al., 2014; Song et al., 2019). However, although HPLC analyses are commonly applied to detect food contamination and adulteration, the results can be uncertain since, in most adulteration cases, it is not possible to assess which adulterant was used, rendering this type of analysis ineffective (Cheah and Fang, 2020). With the identification of the phenolic profile by HPLC, associated with appropriate statistical treatment, it is possible to identify a range of distinct adulterants. In addition, many studies have evaluated phenolic compounds in coffee and even in its husks. However, in most cases, the results have been expressed in terms of total phenolic compounds rather than by presenting a complete profile of phenolics as proposed in this work, especially with the objective of identifying a marker of roasted and ground coffee adulteration with husks.
As a first approach, we carried out a principal component analysis (PCA, an unsupervised method). Principal component analysis 1 (PC1) and PC2 explained 47.7 % and 31.4 % of the total variation.
In the PCA biplot, samples with similar phenolic profiles were placed close to each other (Figure 2). Coffee and defective coffee samples had similar phenolic profiles, with greater quantities of isomers 4 and 5 CQAs, diCQAs, FQAs and caffeic acid in the first extraction stage. In the hydrolyzed fraction, the prominent phenolics were the acids 3-CQA and 4,5-diCQA.
Biplot of principal component (PC) analysis for samples of good coffee beans, defective coffee beans, husks, and their mixtures. The PC analysis was performed using the phenolic compounds profile. A = Arabica coffee; A.BIS = Arabica defective coffee; A.H = Arabica coffee husks; A.3 % = coffee plus 3 % husks; A.10 % = coffee plus 10 % husks; A.20 % = coffee plus 20 % husks; A.BIS.3 % = defective coffee plus 3 % husks; A.BIS.10 % = defective coffee plus 10 % husks; A.BIS.20 % = defective coffee plus 20 % husks; PCA = protocatechuic acid; 3-CQA = 3-caffeoylquinic acid; 4-CQA = 4-caffeoylquinic acid; 5-CQA = 5-caffeoylquinic acid; 3,4-diCQA = 3,4 dicaffeoylquinic acid; 3,6-diCQA = 3,6 dicaffeoylquinic acid; 4,5-diCQA = 4,5 dicaffeoylquinic acid; FQAs = feruloylquinic acids; FA = ferulic acid; 4-HBA = 4-hydroxybenzoic acid; CAF = Caffeine; p-CA = p-coumaric acid; GA = gallic acid; VA = vanillic acid; SA = syringic acid; QCT = quercetin; SYA = synaptic acid; CFA = caffeic acid. For phenolic acids and flavonoids, F = free fraction and B = bound or hydrolyzed fraction.
On the other hand, the coffee husk sample was very different, as indicated by its location, which was far from the others. This sample contained a higher quantity of protocatechuic acid and ferulic acid in the free fraction. In contrast, the hydrolyzed fraction had higher quantities of synaptic acid, gallic acid, vanillic acid and 3,4-diCQA. Finally, adulterated samples were also discriminated efficiently, even those with low percentages of adulterants (3 % of A.H). The PCA results enabled the detection of the phenolic profile after adulteration by distinguishing samples adulterated with husks from pure coffee samples despite having many similar compounds.
The PCA graph also indicated that protocatechuic and ferulic acids were in greater quantities in coffee husks. The PLS-DA method confirmed the results of the previous PCA. Partial least squares discriminant analysis is performed for supervised classification of the sample and is thus a robust regression technique. The sum on the axes explained 67 % of the total variability (Figure 3A). The PLS-DA confirmed the results obtained previously by PCA. There was clear discrimination of coffee samples, defective coffee, husks and adulterated samples. As can be observed in Figure 3B, the PLS-DA resulted in three components, and the receiver operating characteristic curve identified clear sample discrimination, e.g., for pure samples, their area under the curve (AUC) was one, indicating good discrimination of those samples. The lowest AUC value (0.9167) was for samples A + 10 %, A + 20 %, and A.BIS + 20 %. On the other hand, husks had AUC ≥ 0.9861 for other samples containing coffee, indicating good discrimination of the method and its possible application to fraud detection.
Partial least squares discriminant analysis (PLS-DA) for samples of coffee, defective coffee, husks and adulterated samples using a) their phenolic profiles and b) the receiver operating characteristic (ROC) curve of PLS-DA. AUC = area under the curve; A = Arabica coffee; A.BIS = Arabica defective coffee; A.H = Arabica coffee husks; A.3 % = coffee plus 3 % husks; A.10 % = coffee plus 10 % husks; A.20 % = coffee plus 20 % husks; A.BIS.3 % = defective coffee plus 3 % husks; A.BIS.10 % = defective coffee plus 10 % husks; A.BIS.20 % = defective coffee plus 20 % husks.
Finally, the heatmap (Figure 4) was generated to show variations in the levels of phenolic compounds in the samples of coffee, defective coffee, husks and their mixtures. After data normalization, 27 compounds were identified, 12 in the free fraction and 15 in the hydrolyzed fraction, which were selected as discriminants. In the heatmap, the scale in shades of purple indicates the levels of compounds detected, with the stronger tones representing greater quantities and white, the absence of phenolic compounds.
Heatmap based on the phenolic profile of samples of coffee, defective coffee and coffee husks as well as their mixtures obtained by high-performance liquid chromatography in both extraction stages (free and hydrolyzed). The samples were grouped by hierarchical clusterization of principal components. A = Arabica coffee; A.BIS = Arabica defective coffee; A.H = Arabica coffee husks; A.3 % = coffee plus 3 % husks; A.10 % = coffee plus 10 % husks; A.20 % = coffee plus 20 % husks; A.BIS.3 % = defective coffee plus 3 % husks; A.BIS.10 % = defective coffee plus 10 % husks; A.BIS.20 % = defective coffee plus 20 % husks.
The heatmap allowed for identifying markers of adulteration of roasted and ground coffee, including by coffee husks. The clearest marker was protocatechuic acid in the free fraction since this acid was present in high concentrations in coffee husks but not in coffee or defective coffee samples. Evaluation of adulterated samples at 3 %, 10 %, and 20 % revealed the presence of this acid in the free fraction in smaller concentrations proportional to the adulteration. When compared to the coffee and defective coffee samples, ferulic acid in the free fraction could be a marker of adulteration. However, in adulterated samples, in all percentages, ferulic acid was not detected. This may be attributed to low concentrations (less than the detection limit) of this acid in the other samples.
Based on the phenolic profile obtained in the hydrolysis stage in the heatmap, the coffee husks presented four acids absent in either the coffee or the defective coffee. Vanillic acid, gallic acid, syringic acid, and 3,4-diCQA had high intensity in the adulterant and could be potential markers for roasted and ground coffee adulteration. However, it was not possible to indicate these acids as markers because they were not detected in the adulterated samples, as explained above.
The analysis of phenolic compounds by HPLC associated with chemometric methods enabled the identification of protocatechuic acid as an adulteration marker of roasted and ground Arabica coffee with coffee husks in mixtures with low adulterant concentration (3 %). The main advantage of the method was the ability to obtain results with very low concentrations of the compounds. The results obtained in this work support the new approach to detecting adulteration with husks in roasted and ground coffee. Furthermore, the identification of an adulteration marker in the free phenolic fraction (first extraction step) makes the method faster and more economical, reducing the use of reagents and making it more environmentally friendly.
Acknowledgments
We thank Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro – FAPERJ (grant E-26.201.302/2022, E-26/210.306/2022, E-26/204.328/2021, and E-26/202.046/2022) and Conselho Nacional de Desenvolvimento Científico e Tecnológico – CNPq (grant 311108/2021-0) for the financial support.
References
-
Assis NM, Marques CA, Silva MC. 2020. Microscopic evaluation of coffee commercial samples. Segurança Alimentar e Nutricional 27: e020022 (in Portuguese, with abstract in English). https://doi.org/10.20396/san.v27i0.8658921
» https://doi.org/10.20396/san.v27i0.8658921 -
Cangussu LB, Melo JC, Franca AS, Oliveira LS. 2021. Chemical characterization of coffee husks, a by-product of Coffea arabica production. Foods 10: 3125. https://doi.org/10.3390/foods10123125
» https://doi.org/10.3390/foods10123125 -
Cheah WL, Fang M. 2020. HPLC-based chemometric analysis for coffee adulteration. Foods 9: 880. https://doi.org/10.3390/foods9070880
» https://doi.org/10.3390/foods9070880 -
Couto CC, Coelho CCS, Oliveira EMM, Casal S, Freitas-Silva O. 2023. Adulteration in roasted coffee: a comprehensive systematic review of analytical detection approaches. International Journal of Food Properties 26: 231-258. https://doi.org/10.1080/10942912.2022.2158865
» https://doi.org/10.1080/10942912.2022.2158865 -
Domingues DS, Pauli ED, Abreu JEM, Massura FW, Cristiano V, Santos MJ, et al. 2014. Detection of roasted and ground coffee adulteration by HPLC by amperometric and by post-column derivatization UV-Vis detection. Food Chemistry 146: 353-362. https://doi.org/10.1016/j.foodchem.2013.09.066
» https://doi.org/10.1016/j.foodchem.2013.09.066 -
Farah A, Paulis T, Trugo LC, Martin PR. 2005. Effect of roasting on the formation of chlorogenic acid lactones in coffee. Journal of Agricultural and Food Chemistry 53: 1505-1513. https://doi.org/10.1021/jf048701t
» https://doi.org/10.1021/jf048701t -
Ferreira T, Galluzzi L, Paulis T, Farah A. 2021. Three centuries on the science of coffee authenticity control. Food Research International 149: 110690. https://doi.org/10.1016/j.foodres.2021.110690
» https://doi.org/10.1016/j.foodres.2021.110690 -
Martins VC, Godoy RLO, Gouvêa ACMS, Santiago MCPA, Borguini RG, Braga ECO, et al. 2018. Fraud investigation in commercial coffee by chromatography. Food Quality and Safety 2: 121-133. https://doi.org/10.1093/fqsafe/fyy017
» https://doi.org/10.1093/fqsafe/fyy017 -
Mendes LC, Santos JCF, Corrêa JA, Alkimim Filho JF, Silva DVR, Jesus LN, et al. 2016. Validation of a method to determine the impurities husks and straw in roast ground coffee. Brazilian Journal of Food Technology 19: 1-7. https://doi.org/10.1590/1981-6723.9215
» https://doi.org/10.1590/1981-6723.9215 -
Moon JK, Yoo HS, Shibamoto T. 2009. Role of roasting conditions in the level of chlorogenic acid content in coffee beans: correlation with coffee acidity. Journal of Agricultural and Food Chemistry 57: 5365-5369. https://doi.org/10.1021/jf900012b
» https://doi.org/10.1021/jf900012b -
Moore JC, Spink J, Lipp M. 2012. Development and application of a database of food ingredient fraud and economically motivated adulteration from 1980 to 2010. Journal of Food Science 77: R118-R126. https://doi.org/10.1111/j.1750-3841.2012.02657.x
» https://doi.org/10.1111/j.1750-3841.2012.02657.x - Morais SAL, Aquino FJT, Chang R, Nascimento EA, Oliveira GS, Santos NC. 2007. Chemical analysis of Arabica coffee (Coffea arabica L.) and defective beans submitted to different degrees of roasting. Coffee Science 2: 97-111 (in Portuguese, with abstract in English).
-
Nascimento LSM, Santiago MCPA, Oliveira EMM, Borguini RG, Braga ECO, Martins VC, et al. 2017. Characterization of bioactive compounds in Eugenia brasiliensis, Lam. (Grumixama). Nutrition and Food Technology 3: 1-7. https://dx.doi.org/10.16966/2470-6086.146
» https://dx.doi.org/10.16966/2470-6086.146 -
Pauli ED, Barbieri F, Garcia PS, Madeira TB, Acquaro Junior VR, Scarminio IS, et al. 2014. Detection of ground roasted coffee adulteration with roasted soybean and wheat. Food Research International 61: 112-119. https://doi.org/10.1016/j.foodres.2014.02.032
» https://doi.org/10.1016/j.foodres.2014.02.032 -
Ramos PH, Guerreiro MC, Resende EC, Gonçalves M. 2009. Production and characterization of activated carbon prepared from PVA defect coffee. Química Nova 32: 1139-1143 (in Portuguese, with abstract in English). https://doi.org/10.1590/S0100-40422009000500011
» https://doi.org/10.1590/S0100-40422009000500011 -
Silva MO, Honfoga JNB, Medeiros LL, Madruga MS, Bezerra TKA. 2021. Obtaining bioactive compounds from the coffee husk (Coffea arabica L.) using different extraction methods. Molecules 26: 46. https://doi.org/10.3390/molecules26010046
» https://doi.org/10.3390/molecules26010046 -
Song HY, Jang HW, Debnath T, Lee KG. 2019. Analytical method to detect adulteration of ground roasted coffee. International Journal of Food Science & Technology 54: 256-262. https://doi.org/10.1111/ijfs.13942
» https://doi.org/10.1111/ijfs.13942 -
Tavares KM, Pereira RGFA, Nunes CA, Pinheiro ACM, Rodarte MP, Guerreiro MC. 2012. Mid-infrared spectroscopy and sensory analysis applied to detection of adulteration in roasted coffee by addition of coffee husks. Química Nova 35: 1164-1168 (in Portuguese, with abstract in English). https://doi.org/10.1590/S0100-40422012000600018
» https://doi.org/10.1590/S0100-40422012000600018
Edited by
-
Edited by:
Luís Guilherme de Lima Ferreira Guido
