Introduction
Coffee is one of the most heavily consumed beverages in the world. Quality is closely examined and is a key consideration in commercial trade. It is segmented based on many factors, including source type (Arabica, Robusta, and Liberica), flavored and non-flavored, product types such as whole-bean, powdered, instant, and others, and caffeinated and non-caffeinated. The principal quality parameters include moisture, blend ratio, roasting degree, and caffeine. Moisture content in green coffee beans is strictly regulated in most countries that import and export coffee as improper content can cause quality deterioration and even fungal or mycotoxin contamination that can present a danger to human health. Coffee blending is important to achieve a final product with a given flavor and aroma and this process usually occurs before roasting. Roasting initiates complex chemical changes in coffee beans that are crucial to forming the desired final product. The color of the beans is an important marker in the roasting process and is an indicator of volatile compounds that determine aroma and flavor. Caffeine is an important parameter in coffee as its physiological and psychoactive properties are a big reason why coffee is one of the most popular beverages in the world market. It is also an important compound along with similar alkaloids in determining the quality of coffee. Adulteration is considered a large problem in the food and beverage market and especially so in coffee due to its popularity as well as the range of factors that come into play when determining the desired product to fit consumer demands. Adding adulterants in coffee not only causes financial loss, it can be a threat to consumer health as well. There is a need to measure and determine these quality parameters at all stages of the coffee manufacturing process, from the initial analysis of green coffee beans all the way to determining if the final product of coffee is what is actually being labelled and marketed. Currently methods for testing these parameters such as HPLC are expensive, laborious, and time-consuming, especially when implemented in a process setting. There is a need for fast, cost-effective, and real-time monitoring of parameters at all stages of the coffee manufacturing process. One such method that has been examined is NIR spectroscopy.
Analytes
- Moisture
- Caffeine
- Theobromine
- Instant Coffee
- Plant Fat
- Sugar
- Adulterant % and Identification
- Species Classification
- Defective and Non-Defective Classification
- (Beans and Roasted & Ground Coffee)
- Sensory Properties Scores
- Weight Loss
- Density
- Antioxidant Capacity of Spent Coffee Grounds
- Antioxidant Capacity of Spent Coffee Grounds
- Ethanolic Extracts
- Total Flavonoids
- Total Phenolics
- Color
- Arabica Content (Blend Ratio)
- % Corn Adulterant
Summary of Published Papers, Articles, and Reference Materials
Measurement of chemical parameters in coffee has been studied using NIR spectroscopy and the results of most studies have demonstrated the potential of using NIR as a replacement for expensive and time-consuming wet chemistry methods. A comprehensive review paper is presented discussing various studies and applications for using NIR spectroscopy as a quality control tool for coffee. Topics include prediction of coffee composition, authentication, sample classification, defective and non-defective sample discrimination, sensory properties, roasting degree, and coffee residue analysis. The first step in coffee manufacturing is to ensure the quality of the green coffee beans before roasting and moisture is considered a crucial component in coffee beans. One study examined the feasibility of measuring moisture in green coffee beans with the best results coming from a model using a selective wavelength algorithm to determine the best wavelength ranges for correlating the spectral data with moisture. Species and origin of coffee are of vital importance as well, especially when comparing the higher quality and more expensive Arabica species to the Robusta species. Roasting color is important as well because it is directly related to the sensory properties of the final coffee product and one study examined simultaneously measuring color and blend ratio in roasted coffee. Results were considered good for an initial study, but more data and calibration work would be necessary to apply the models for quality control purposes in an on-line setting. Another study examined the effect of roasting degree on classifying Arabica and Robusta coffee samples, showing excellent results. Caffeine content is not only directly related to coffee quality, but its physiological and psychoactive properties are a big reason why coffee is one of the most popular beverages in the world market. One study examined determining not only caffeine in ground coffee, but two other significant alkaloids as well: theobromine and theophylline. Decent correlation was obtained for caffeine and theobromine, but the detection limit for the concentration of theophylline was below the detectable level of measurement for NIR spectroscopy. Another study used NIR spectra of Arabica coffee samples, reference values for caffeine, and various data pre-treatments and selective wavelength algorithms to optimize regression models for determining caffeine. Results were greatly improved using the best pre-treatment and selective wavelength algorithm, proving the feasibility of using this model to determine caffeine content in Arabica coffee. Adulteration is a huge problem in the food and beverage industry and coffee is no exception to this problem. One study examined determining the percentage of corn adulterant in Brazilian coffee samples by correlating the tocopherol profile to the percentage of corn present in the samples and then creating a regression model from the NIR spectra. Results were excellent and provide a basis for further study that would encompass using different species of coffee and various other adulterants.
Scientific References and Statistics
Application of Infrared Spectral Techniques on Quality and Compositional Attributes of Coffee: An Overview – Barbin, Felicio, Sun, et al., Food Research International 61 (2014) 23-32
This review of infrared spectral analyses and applications in the coffee industry discusses studies which reveal the potential in using these techniques to obtain information about the chemical composition and related properties of coffee. Infrared analysis not only has the ability to quantify and characterize coffee quality attributes from moisture, lipids, caffeine, quality grading, sensory properties, and other important constituents, it can do so in a rapid manner with little sample preparation and while measuring multiple constituents simultaneously. Potential benefits of widespread development of such analysis are discussed as well as the latest research and developments. Below is a breakdown of quality parameters and analysis from the research discussed in the review.
Prediction of Coffee Composition
Several studies have investigated the potential of using spectral applications to measure physical, chemical, and quality parameters of coffee. Moisture content is an important parameter in green coffee beans and raw coffee. Water above 12.5% in coffee beans causes a number of undesirable consequences, such as mycotoxin formation, microbial growth, altered sensory quality, and unstable production conditions. One study examined determining moisture in raw coffee with results acceptable enough for screening purposes.
| Moisture | R² = 0.818 | RMSEP = 0.298 g/100 g |
Caffeine is a very important component in coffee and has been studied in several investigations. Ground Arabica samples at varying roasted levels were analyzed by NIR spectroscopy using spectral data, HPLC reference values for caffeine, and various data treatments and chemometric methods. The best model proved the feasibility of using at-line application to determine caffeine content in unknown roasted coffee samples. Another study measured roasted coffee for multiple alkaloids in both Arabica and Robusta liquid coffee samples after discrimination of the green coffee beans to classify the samples. Liquid Chromatography and Mass Spectrometer reference values were used to correlate the NIR spectra with caffeine, theobromine, and theophylline. Good correlation and prediction values were found for caffeine and theobromine, but the detection limit for theophylline was too low for the NIR calibration model to be acceptable for real use. Another study used diffuse reflectance NIR spectra of liquid coffee beverages as predictors for the three main ingredients in liquid coffee: instant coffee, plant fat, and sugar. Excellent correlation was obtained for all three parameters, proving that NIR spectroscopy can be used to determine these ingredients in liquid coffee.
| Caffeine (Arabica) | R² = Unknown | RMSEP= 0.378 mg/g |
| Caffeine (Arabica & Robusta) | R² = 0.86 | RMSEP= 0.07 mg/g |
| Theobromine (Arabica & Robusta) | R² = 0.85 | RMSEP= 0.10 mg/g |
| Instant Coffee | R² = 0.9897 | RMSEP= 2.12 mg/g |
| Plant Fat | R² = 0.9994 | RMSEP= 0.72 mg/g |
| Sugar | R² = 0.9918 | RMSEP= 2.01 mg/g |
Authentication
Food authentication has become a major issue in recent years and the growth of the coffee market has made coffee a target for many different types of adulteration. Adulteration in coffee can take on multiple forms. A substance that is not coffee at all can be mixed in, such as chicory, malt, figs, cereals, caramel, starch, maltodextrins, or glucose. Coffee is often marketed as being distinct to a particular region and misrepresenting the origin of a coffee product is another form of adulteration. The Arabica coffee bean is considered superior to the Robusta bean and labeling a lower quality coffee product as a higher quality species or blend is also considered adulteration. Spectroscopic techniques have been studied as methods for identifying adulteration in coffee. One study used nine commercial roasted and ground coffee samples to identify differences in the NIR spectra as a basis for classification. Barley samples were then blended into the coffee at a range of 2% to 20% weight per weight of coffee to examine the feasibility of identifying barley adulterant in commercial coffee. Low prediction errors were obtained and the results show promise for future applications to identify and quantify adulterants in coffee. Another study used diffuse reflectance spectra of instant coffee samples and samples with various adulterants added, such as glucose, starch, and chicory. Various classification techniques were applied and an artificial neural network (ANN) model was able to classify adulterated and non-adulterated samples with a 100% success rate.
| Identifying Adulterated Samples (Instant Coffee) | 100% Correct Classification |
Classification of Samples According to Coffee Variety and Quality Features
In general, the Arabica coffee bean is considered superior to Robusta and is the more expensive and higher quality of the two beans. NIR spectroscopy has been examined in various studies as a method for discriminating and characterizing these two blends with relative success. A few studies have examined lyophilized and vacuum-dried samples for discrimination analysis. Notable spectral differences were discovered in the caffeine absorbing areas of the NIR spectrum, implying that there is a difference in caffeine between Arabica and Robusta that could be used to as a basis to classify the two blends. Classification rates were good enough to use this analysis for screening purposes. Another study used NIR spectra of various blends and Partial Least Squares (PLS) analysis to predict Robusta content, showing accurate results but as a limited study, more data encompassing the natural variety that exists in coffee (e.g. geographical origin, roasting degree) would be necessary to use this model in a practical setting. Another form of classification analysis that has been examined is classifying roasted coffee grain samples from different lots and producers in a given region using NIR spectra. One such study conducted in Brazil showed that NIR spectroscopy can be a useful tool in differentiating roasted coffee grains.
Classifying Arabica and Robusta Dried Beverage:
| Lyophilized | 87% |
| Vacuum-Dried | 95% |
Discrimination Between Defective and Non-Defective Samples
Assessing bean quality in coffee is based on discovering the relative amount of defective beans among non-defective ones. One methodology that has been studied to implement such a quality assessment using NIR spectroscopy compared two Arabica varieties and two Robusta varieties, all from different geographical regions, to determine the presence of defective beans in a batch. A Partial Least Squares (PLS) regression model relating the NIR spectra to the mass fraction of defective and non-defective beans showed correlation good enough for screening purposes. Likewise, Principal Component Analysis (PCA) was applied to spectra of roasted and ground coffee of four different groups (non-defective, black, dark sour, and light sour) to determine the feasibility of separating defective samples (both sour groups) from spectral data. Accuracy of the classification ranged from 95% to 100% depending on the particular model. A similar study used both PCA and cluster analysis to analyze the spectra, enabling separation into two distinct groups: non-defective/light sour and black/dark sour, indicating that the samples considered defective (black, immature, and dark sour) could be separated using spectral data.
| Prediction Error for Non-Defective/Defective Beans (2 Arabica and 2 Robusta varieties): | 5% |
| Classification Accuracy for Discriminating Defective and Non-Defective Roasted and Ground Coffee: | 95% to 100% (Depending on the exact classification grouping and model used) |
Prediction of Sensory Properties
Studies have been conducted to establish a relationship between sensory attributes of coffee, the chemical components of the coffee beans, and NIR spectra. Coffee beverage of roasted Arabica samples and NIR spectra were analyzed using chemometrics to establish a correlation between acidity, bitterness, flavor, cleanliness, body, and overall quality scores. Selective wavelength algorithms determined the relevant wavelength regions for each model. Good correlation was obtained for all models and confirmed the relationship between the chemical composition of the roasted grains and sensory properties is directly related to the NIR spectra of pure caffeine, trigonelline, 5-caffeeoylquinic acid, cellulose, coffee lipids, sucrose, and casein. All these components are related to the different sensory characteristics that were modelled. A similar study was conducted for espresso quality assurance using scores for perceived acidity, mouthfeel (body), bitterness, and aftertaste. Results of calibration models were comparable to evaluations provided by a trained sensory panel, proving the feasibility of using such calibrations as an evaluation tool for coffee sensory properties.
Degree of Roasting
Roasting color and quality parameters are attributes that have been studied using NIR spectroscopy with good results. One such study used spectral data to discriminate between medium and dark roasted commercial coffee samples, both caffeinated and decaffeinated. An external validation set was correctly predicted at a 100% rate. Chemometric analysis showed the wavelengths used for the model and predictions are related to caffeine and moisture, as decaffeinated coffee is known to have a higher moisture content. The relationship between coffee roasting variables like weight loss, density, and moisture and NIR spectra of green (raw) and coffee samples roasted at different levels was investigated in another study in order to predict roasting degree. Robust models were obtained with high correlation coefficients and prediction results were comparable to the reference analyses, proving the feasibility of this application as a tool for on-line analysis of the roasting process. A similar study focused on espresso and roasted coffee correlating NIR spectra to total acidity, caffeine content, chlorogenic acids, and roasted bean color. The regression models showed results good enough to be used for prediction of the listed quality parameters.
| Weight Loss, Density, Moisture | R² = 0.92 to 0.98 for all parameters |
Coffee Residues
Spent coffee grounds contain high levels of bioactive compounds, including flavonoids that have antioxidant properties. One study correlated NIR spectra to antioxidant capacity, total phenolics, and total flavonoids in spent coffee grounds samples. Partial Least Squares (PLS) regression modeling was used to correlate the spectral data to these parameters and results were excellent, with all parameters having a correlation coefficient well above 0.90. Another study used similar methods to measure various lignin components in coffee and banana residues. Correlation was lower than in the previously discussed study but the results did show potential for using NIR spectroscopy to measure these components in both coffee and banana residues.
| Antioxidant Capacity of Spent Coffee Grounds | R² = 0.93 |
| Antioxidant Capacity of Spent Coffee Grounds Ethanolic Extracts | R² = 0.96 |
| Total Flavonoids | R² = 0.95 |
| Total Phenolics | R² = 0.95 |
While conducted mostly on a laboratory scale, the studies documented in this review demonstrate the ability to use NIR spectroscopy for analysis of raw materials, intermediates, finished products, and as a process control tool in coffee. Increased demand for product control of coffee as well as many other liquid foods will require advanced analytical tools and NIR spectroscopy is a proven method for both on-line and at-line monitoring of coffee.
The development of new sensors has facilitated the implementation of NIR spectroscopy as a tool for monitoring the coffee process with successful results.
Rapid Prediction of Moisture Content in Intact Green Coffee Beans Using Near-Infrared Spectroscopy – Adnan, von Horsten, Pawelzik, Morlein, Foods 2017, 6, 38
Moisture is a very important quality parameter in green coffee beans and is strictly regulated by most countries that import and export coffee. The safe range for moisture is from 8% to 12.5% based on fresh matter. Moisture below 8% causes shrunken beans and an unwanted appearance. Moisture above 12.5% facilitates fungal and mycotoxin growth as well as the potential for problems during storage and the roasting process. NIR spectroscopy was examined as a method for measuring moisture content in both Arabica and Robusta green coffee beans. Twelve sets of samples were used for the study: Three Arabica species and four Robusta species of different origins for the calibration set and two Arabica species and three Robusta species of different origins for the validation set. NIR diffuse reflectance spectra were collected from all samples from 1000 nm to 2500 nm at 2 nm intervals. Each individual spectrum consisted of the average of 64 scans. Three replicates were acquired for each sample and these spectra were averaged as well, resulting in 108 total spectra of the 12 different samples. Reference values were obtained for moisture and these were used with the NIR spectra to create Partial Least Squares (PLS) calibration models for moisture content.
| Moisture (Full Wavelength Range) | R² = 0.9850 | RMSEP= 0.57% |
| Moisture (Selective Wavelengths) | R² = 0.9743 | RMSEP= 0.77% |
Two sets of PLS calibration models were created: one using the full wavelength range and the other using seven selective wavelengths that were chosen based on the correlation of the full range model. Some of these are moisture absorbing areas of the NIR spectrum and others correlate to organic compounds affected by a change in moisture: 1155 nm, 1212 nm, 1340 nm, 1409 nm, 1724 nm, 1908 nm, and 2249 nm. Prediction results on the validation set using both models proved the feasibility of the measurement. Results were comparable for both models and either could be applied in an on-line setting to determine moisture in green coffee beans.
Simultaneous Determination by NIR Spectroscopy of the Roasting Degree and Arabica/Robusta Ratio in Roasted and Ground Coffee – Bertone, Venturello, Giraudo, et. Al, Food Control 59 (2016) 683-689
The roasting color of coffee beans and the varietal composition of blends are two crucial factors in sensory properties of brewed coffee. Color is a critical control parameter and is used to verify the performance of the roasting, as there is a direct relationship between color and the desired sensory characteristics of the final product. Blend composition is important because in general, the Arabica species shows better sensory characteristics than the Robusta species, resulting in a marked difference in the market price of the two species. NIR spectroscopy was examined as a method for simultaneously determining both of these important parameters in blended roasted and ground coffee. 130 commercial blends of roasted and ground coffee belonging to both Arabica and Robusta species were used for the study. The samples were of varying worldwide geographical origin and all harvested in the same season. They showed ten different levels of Arabica content ranging from 0% to 100% w/w. One hundred samples were used for the calibration set and thirty were used for a validation set. After roasting and milling, the samples were scanned using an FT-NIR spectrometer from 12500 cm-1 to 3500 cm-1. Spectral resolution was 16 cm-1 and 64 scans were averaged for each individual spectrum. Reference tests were conducted to determine color values and these were used along with the blend ratio values and NIR spectra to create Partial Least Squares (PLS) calibration models.
| Color | R² = 0.87 | RMSEP= 1.28 A.U. |
| Arabica Content | R² = 0.97 | RMSEP= 4.34% w/w |
The results obtained here are considered good enough to use this method as a quality control tool and for fraud identification, but a larger data set and more accurate prediction values would be necessary for application in an industrial setting. A wider data set incorporating more varieties of the two blends as well blends that are individually mixed to create more data points should improve model performance. It is also important to consider the variability between annual blends of coffee and incorporate multiple harvests into the calibration models before implementing such an application in an industrial setting.
Characterization of the Effects of Different Roasting Conditions on Coffee Samples of Different Geographical Origins by HPLC-DAD, NIR, and Chemometrics – De Luca, De Filippis, Bucci, et al., Microchemical Journal 129 (2016) 348-361
The effect of roasting conditions on both the NIR and HPLC profiles of coffee samples was evaluated using various classification algorithms to determine if the roasting degree had a marked effect on determining whether the samples were of Arabica or Robusta origin. Thirty-six samples of green coffee beans (twenty-three Arabica and thirteen Robusta) of different geographical origins were used for the study. Six were analyzed by HPLC while thirty were used for NIR spectroscopic analysis. Each sample was roasted in the laboratory under different conditions trying to reproduce the industrial roasting process. For each sample, NIR spectra were collected at the following roasting times: 0 minutes (green), twenty-five minutes, fifty minutes, and seventy-five minutes. Scan parameters were from 10000 cm-1 to 4000 cm-1 at a nominal resolution of 4 cm-1 and eighty-two scans per average. After data collection, both Partial Least Squares-Discriminant Analysis (PLS-DA) and Soft Independent Modeling of Class Analogies (SIMCA) classification algorithms were used to build models for determining the varietal origin of the coffee beans.
PLS-DA
| Arabica Classification | 100% |
| Robusta Classification | 95% |
SIMCA Specificity:
| Arabica Classification | 96% |
| Robusta Classification | 96% |
The results shown above prove that NIR spectra can be used to classify the species of coffee irrespective of the roasting degree at an accuracy level of 95% or higher. Similar analysis was conducted using the HPLC fingerprints of those samples and results were comparable. The NIR approach allows for authenticating the species of coffee beans with a rapid, low-cost, non-invasive technique and could be implemented as an application for the quality control of coffee beans at all levels of the roasting and manufacturing process.
Analysis of Caffeine, Theobromine, and Theophylline in Coffee by Near Infrared Spectroscopy (NIRS) Compared to High-Performance Liquid Chromatography (HPLC) Coupled to Mass Spectrometry – Huck, Guggenbichler, Bonn, Analytica Chimica Acta 538 (2005) 195-203
NIR spectroscopy was examined as a method for quantifying the three main alkaloids found in coffee: caffeine, theobromine, and theophylline. Eighty-three samples of roasted Arabica and Robusta coffee from different geographical origins were provided for the study and ground before NIR spectra collection. Scan parameters were from 9996 cm-1 to 4500 cm-1 using 12 cm-1 resolution and ten scans per average. Three separate spectra were collected for each sample for a total of two hundred forty-nine spectra. A portion of each sample was used for HPLC analysis to determine the reference values for caffeine, theobromine, and theophylline. Two separate LC analyses were performed: LC-UV (Liquid Chromatography – UV Detection) and LC-ESI-MS (Liquid Chromatography – Electrospray Ionisation Quadrupole Ion Trap Mass Spectrometry). LC-UV was chosen as the reference method for regression models using the NIR spectra to correlate to the three alkaloids of interest.
| Caffeine | R² = 0.86 | Range = 0.95-4.13 g/100 g | RMSEP= 0.40 g/100 g |
| Theobromine | R² = 0.85 | Range = 0.10-0.67 g/100 g | RMSEP= 0.10 g/100 g |
| Theophylline | Concentration below the limit detectable by NIR |
Calibration models for caffeine and theobromine showed correlation and prediction results comparable to the LC-UV reference method that can be considered suitable for screening purposes. In the case of theophylline, the lower limit of detection (LOC) for LC-UV is 0.244-0.60 ng/100 g while the LOD for the NIR method is 0.05 g/100 g, making the analysis of theophylline using NIR spectra impossible. However, the results for the other two alkaloids provide a potential alternative to the more expensive and time-consuming GC method.
Improvement of Near Infrared Spectroscopic (NIRS) Analysis of Caffeine in Roasted Arabica Coffee by Variable Selection Method of Stability Competitive Adaptive Reweighted Sampling (SCARS) – Zhang, Li, Yin, et al., Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 114 (2013) 350-356
NIR spectroscopy was examined as a method for quantitatively determining caffeine content in roasted samples of Arabica coffee. Seventy-two ground coffee samples were procured for the study. NIR spectra were collected from 10000 cm-1 to 4000 cm-1 using 8 cm-1 spectral resolution and thirty-two scans averaged per spectrum. This process was repeated three times for each sample with changing position in the sample holder for each run and all three spectra were then averaged to create one spectral data point. Likewise, the entire process was repeated for each sample, making a total of one hundred forty-four spectra that were used in the study. Sixty-two samples (one hundred twenty-four spectra) were used for the calibration set and ten samples (twenty spectra) were used for the validation set. HPLC-UV analysis was used to determine reference values for caffeine. Various data pre-treatments and selective wavelength analysis were performed on the NIR spectra in order to determine the best data set for Partial Least Squares (PLS) regression analysis.
Stability Competitive Adaptive Reweighted Sampling (SCARS) – PLS Model:
| Caffeine | R² = 0.918 | RMSEP= 0.375 mg/g |
Multiple PLS models were created and the best results came using the SCARS selective wavelength algorithm with a second derivative pre-treatment of the NIR spectra. Eighty-three total wavelengths were chosen for the caffeine correlation. All were concentrated in the following four regions: 4196 cm-1– 4018 cm-1, 5046 cm-1– 4412 cm-1, 6105 cm-1-5577 cm-1, 7706 cm-1– 6784 cm-1. These are all areas where the NIR spectrum of pure caffeine show distinct absorption peaks, indicating that the SCARS algorithm is choosing wavelengths which are in fact relevant to changes in the caffeine content. Validation set predictions confirmed the feasibility of the model as a method to quantitatively determine caffeine content in Arabica coffee.
Detection of Corn Adulteration in Brazilian Coffee (Coffea Arabica) by Tocopherol Profiling and Near-Infrared (NIR) Spectroscopy – Winkler-Moser, Singh, Rennick, et al., Journal of Agricultural and Food Chemistry, 2015, 63, 10662-10668
Coffee is considered a high-value commodity that is a frequent target for adulteration, as are many other food and beverage products. There is significant interest in developing and improving methods for detecting coffee adulteration and one such method that has been examined is NIR spectroscopy. Potential adulterants for coffee include corn, sticks, coffee husks, and other lower value crops. Fourteen different lots of green Arabica coffee beans from different cities and plantations in Brazil were procured for the study. All coffee samples were roasted as well as corn to be added as an adulterant. Both the coffee and corn were ground before mixing. Five different concentrations of corn adulterant were added and mixed with each sample ranging from 1% to 20% corn. In addition, the pure Arabica sample from each lot was used as well, making for a total of eighty-four samples. Samples were scanned using an NIR spectrometer from 400 nm to 2500 nm. Tocopherol content was chosen as a basis for determining the percentage of corn adulteration and reference values for tocopherol in the samples were determined using HPLC. The HPLC results and a Multiple Linear Regression (MLR) equation were used to correlate the tocopherol profile results to the percentage of corn adulteration. This correlation was then used as reference value to correlate the NIR spectra to % corn adulteration using Partial Least Squares (PLS) analysis.
| % Corn Adulterant | R² = 0.986 | RMSEP = 1.171% |
Results of the calibration model showed good correlation and prediction values proved the feasibility of the model. Both the NIR and HPLC methods showed comparable results with about a 5% sensitivity, proving the potential of NIR spectroscopy as a fast, simple, and reliable method for detecting corn adulterant in ground coffee. The suggested next step is further studies incorporating different species of coffee with other kinds of adulterants to test the feasibility of a universal model for adulteration. Such a model would have to be continuously updated with new data as different types of adulterants are emerging all the time in the world food and beverage market.