The global edible oils market is estimated at a value around $100 billion and is expected to grow substantially in coming years.  Vegetable oils are a large part of the market and there are many reasons why this is the case. Health conscious consumers and growing awareness about food contents, especially in today’s social media environment, have pushed demand away from unhealthy food contents like trans fats, partially hydrogenated oils, and cholesterol and more towards healthy oils like olive and canola.  Analysis of raw materials used for oil manufacturing and processing is important.  Fatty acid content is the most essential parameter in differentiating between healthy and unhealthy edible oil products.  Oxidation is a major issue after edible oil production. Oils rich in polyunsaturated fatty acids are susceptible to the formation of peroxides and hydroperoxides after exposure to oxygen, heat, and light. Monitoring the potential for oxidation during production and after product formation is important for quality control.  Product authenticity is another parameter that requires monitoring but is often difficult to determine in practice. Two examples of this are determining if an oil is transgenic or non-transgenic and if an expensive edible oil is adulterated with a cheaper product. Health conscious consumers have also created a demand for new products, such as combinations of butterfat and vegetable oil as an alternative to butter that contains less saturated fat. This process is done using a catalyzed reaction and real-time reactor monitoring would be useful for optimizing the reaction.  The demand for such products has created a need for fast, cheap, real-time monitoring of parameters in vegetable oils at all stages of the production and storage process that can replace expensive, laborious, and time-consuming wet chemistry methods. One such method that has been examined is NIR spectroscopy.  


  • Total Oil 
  • Oleic Acid 
  • Linoleic Acid 
  • Linolenic Acid 
  • Palmitic Acid 
  • Stearic Acid 
  • Essential Oil Content (EOC) 
  • Saturated Fatty Acids (SFA) 
  • Monounsaturated Fatty Acids (MUFA) 
  • Polyunsaturated Fatty Acids (PUFA) 
  • Trans Fatty Acids (Trans FA) 
  • Natural α -Tocopherol 
  • Transgenic and Non-Transgenic Classification 
  • Lard Adulteration of Palm Oil 
  • Peroxide Value (PV) 
  • Conversion Degree/Peak Ratio 
  • Solid Fat Content (SFC) 

Summary of Published Papers, Articles, and Reference Materials 

Measurement of chemical parameters of vegetable and other edible oils for quality control purposes has been studied using NIR spectroscopy.  The results of most studies have been promising.  Peanut oil is an important parameter in determining the use of peanuts after harvesting and two breeds of peanuts were scanned through the shell for the purpose of determining total oil as well as individual fatty acids.  Results showed that the total oil content could be predicted with accurate enough results for quality control analysis, while the individual fatty acid models could work for initial screening purposes.  Determining parameters in the raw material used to create and process vegetable oil is critical. One study compared using a handheld spectrometer and FT-NIR spectrometer for determining essential oil content (EOC) in oregano.  Results using the FT-NIR were accurate enough for real-time monitoring while the handheld results were not as accurate. More calibration work should improve the results of using a handheld spectrometer for this purpose.  Fatty acid content is an important parameter in any edible oil and one study examined measuring SFA, MUFA, PUFA, and Trans FA in various off-the-shelf edible oils. There was some difficulty in measuring Trans FA but this is likely because a “Zero Declared Trans FA” oil can in reality have up to 2% Trans FA. α-Tocopherol is an active form of Vitamin E and an important natural antioxidant of lipids that is found in vegetable oils. This compound has been successfully measured using NIR spectroscopy.  Generic engineering has become an important part of food manufacturing and the feasibility of separating transgenic and non-transgenic soybean oils was studied and found to be good enough for screening purposes. Adulteration is an issue with many types of oil, especially palm oil, which is one of the edible oils most in demand as well as having the highest projected growth according to market forecasts. Lard adulteration in palm oil was studied and the calibration model proved the feasibility of detecting and quantifying the lard adulterant.  Oxidation of vegetable oils is a major problem during storage and one parameter for determining oxidation is peroxide value (PV).  Studies were conducted measuring PV in olive, sunflower, rapeseed, corn, and soybean oils, all showing excellent results. The demand from health conscious consumers has led to new products in the market, such as mixtures of butterfat and edible oil, which are catalyzed in a reactor using different enzymes.  The feasibility of monitoring such a reaction between butterfat and oil blends has been examined.  Triglyceride profile analysis and solid fat content (SFC) were modeled using reference data and NIR spectra, with the Conversion Degree/Peak Ratio showing good correlation and the SFC only showing discernible change at a certain temperature. Nevertheless, the SFC model could be used but the Conversion Degree/Peak Ratio model would be the model of choice for real-time analysis of the reaction. Overall, the papers discussed below all show promise for implementing NIR spectroscopy throughout the production and storage process for vegetable and other edible oils.   

Scientific References and Statistics 

Determination of In-Shell Peanut Oil and Fatty Acid Composition Using Near-Infrared Reflectance Spectroscopy – Sundaram, Kandala, Holser, et al., J Am Oil Chem Soc (2010) 87:1103-1114 

NIR spectroscopy was used to analyze two different varieties of in-shell peanuts for total oil and fatty acid concentration.  50 kg of Virginia and Valencia peanut pods obtained over two consecutive harvests were procured for the study.  For each group, sample sets were divided out for calibration and validation. Within each set, thirty spectra were collected from 400 nm -2500 nm at 0.5 nm intervals. The thirty spectra were then averaged into one spectrum for one data point per set. Reference tests were performed for total oil and fatty acid concentration. 


Total OilR2= 0.9772 
Oleic AcidR2= 0.9838 
Linoleic AcidR2= 0.977 
Linolenic AcidR2= 0.981 
Palmitic AcidR2= 0.9838 
Stearic AcidR2= 0.9507 


Total OilR2= 0.9601 
Oleic AcidR2= 0.5734
Linoleic AcidR2= 0.8289 
Linolenic AcidR2= 0.8475 
Palmitic AcidR2= 0.9532 
Stearic AcidR2= 0.7681 

Correlation coefficients were all above 0.95 for the Virginia data set but lower for the Valencia set.  Predictions on the validation set for both groups indicated that the total oil content could be predicted with results good enough for quality control and analysis. In the case of the individual fatty acids, the results were not as accurate but good enough to work for initial screening purposes. Oleic acid showed especially low correlation for the Valencia data set. While it is hard to pinpoint the exact reason for this, model data showed numerous outlier samples that were not fit into the model with good correlation.  It should be noted that oleic acid has been correlated using NIR spectra in other studies in different types of oils (including the Virginia set in this study), so the reason clearly is due to something either in the samples themselves or in the reference method.  More calibration work will be required to validate the fatty acid measurements for use in a real-time setting.

Prediction of Essential Oil Content of Oregano by Hand-held and Fourier Transform NIR Spectroscopy – Camps, Gerard, Quennox, et al., J Sci Food Agric 2014; 94: 1397 – 1402 

Several species of oregano over two separate harvests encompassing a wide range of essential oil content (EOC) were procured for the study. Samples were scanned using both a MEMS-based handheld spectrometer and an FT-NIR spectrometer.  The MEMS instrument scanned through a glass vial from 1000 nm -1800 nm using a resolution of 8 nm and thirty scans per average.  Multiple spectra were collected per sample and the vial was slightly rotated between each spectrum collection.  The FT-NIR instrument scanned from 1000 nm -2500 nm using a resolution of 12 cm-1. Spectra were collected in reflectance mode and the samples were rotated in a glass dish. Six spectra were collected per sample. After all the data was collected, the spectra for both instruments were split into calibration and validation sets.  


EOCR2= 0.58


EOCR2= 0.91

Correlation was much better for the FT-NIR model than the handheld model.  The FT-NIR instrument gave validated results proving an accurate model for prediction and can be used in a practical real-time setting. The handheld instrument did not show accurate enough results for real-time use, especially when considering that a bias correction was necessary to achieve decent prediction results. One likely reason for the poorer correlation when using the handheld instrument is the limited spectral range.  C-H combination bands in oil are prevalent in the wavelength range from 2300 nm -2500 nm.  The FT-NIR model contained this range while the handheld model did not.  However, the handheld approach is promising and more calibration work should improve the calibration model.   

Rapid FT-NIR Analysis of Edible Oils for Total SFA, MUFA, PUFA, and Trans FA with Comparison to GC – Mossoba, Azizian, Tyburczy, et al., J Am Oil Chem Soc (2013) 90:757-770 

Thirty commercial brands of oils and fats were procured for the study. Samples included olive, canola, vegetable, corn, walnut, grapeseed, peanut, flax, coconut, sunflower, and safflower oils as well as one shortening.  Two different FT-NIR spectrometers were used, both fit with an adjustable pathlength transflectance probe. Five spectra were collected per sample and averaged into one spectrum using 8cm-1 resolution.  Pre-developed PLS calibration models were used for the SFA, MUFA, PUFA, and Trans FA predictions and the results were compared to a reference GC method.   

SFA (FT-NIR vs. GCR2= 0.9993 
MUFA (FT-NIR vs. GC)R2= 0.9957 
PUFA (FT-NIR vs. GC)R2= 0.9974 
Trans FA (FT-NIR vs. GC)Poor Correlation for quantifying declared zero Trans FA samples 

The FT-NIR spectroscopic results were successful for SFA, MUFA, and PUFA and showed agreement with the GC results for these three parameters. There was a marked difference between values declared on the product labels and the two reference methods. Possible reasons for this are new high-oleic oil varieties and oxidation of the products during storage. In the case of Trans FA, the actual Trans FA for a declared value of zero is defined as having a concentration of less than 2% Trans FA as a percentage of total fat. While the results obtained by both reference methods did successfully measure the samples to meet the zero criteria, actual predicted values were different when the Trans FA was less than two percent. The inconsistency likely occurs due to differences in the spectroscopic and chromatographic techniques as well as from the addition of minor components that affect the NIR spectra. Results from this study are very promising when considering that GC is very laborious and expensive to implement in a process setting. While more calibration work would be required to implement measuring edible oil FA on-line using NIR spectroscopy, the benefits from a financial standpoint would be substantial.

NIR Spectroscopy and Partial Least-Squares Regression for Determination of Natural α- Tocopherol in Vegetable Oil – Szlyk, Szydlowska-Czerniak, Kowalczyk-Marzec, Journal of Agricultural and Food Chemistry, Vol. 53, No. 18, 2005 

α -Tocopherol is the most active form of Vitamin E and an important natural antioxidant of lipids that are present in vegetable oils.  Multiple types of commercially available edible oil were used for this study, including sunflower, soybean, corn, rapeseed, a mix of rapeseed, soybean, & corn, grapeseed, extra virgin olive oil, and a mixture of virgin and refined olive oil.  Samples of α -Tocopherol were extracted from each oil with ethanol ranging from 0.54 – 53.54 mg/ml. Extracted samples were scanned from 10,000 cm-1 – 4000 cm-1 with fifty scans per average and three averaged spectra used per reading. Resolution was 8 cm-1. A calibration model was created using the spectra and HPLC references values. Selective wavelength analysis and mathematical algorithms were applied to the calibration model. 

Calibration Model for αTocopherol in extracted samplesR2= 0.9931
Validation samplesNIR vs. HPLC methodR2= 0.9515

Correlation for the calibration model was high and samples of all eight types of edible oils were chosen for validation samples. After α -Tocopherol extraction, the samples were scanned and the correlation between the NIR and HPLC was above 0.95. The results prove that NIR can be used as a quality control tool for monitoring the oxidative stability of edible oils.

Rapid Characterization of Transgenic and Non-Transgenic Soybean Oils by Chemometric Methods Using NIR Spectroscopy – Luna, da Silva, Pinho, et al., Spectrochimica Acta Part A 100 (2013) 115-119 

Genetic engineering of food has become advanced in recent years, but in many parts of the world such food is considered undesirable, creating the need for rapid screening methods for proper labeling.  Forty transgenic and forty non-transgenic soybean oil retail samples were procured for the study. Transmission spectra were collected from 1100 nm – 2500 nm at 1 nm intervals. Five spectra were collected per sample and averaged into one spectrum. Various post-processing of the spectra and classification methods were performed to determine the best method for separating the two types of samples. Best results were obtained using Support Vector Machine-Discriminant Analysis (SVM-DA), a technique used for binary classification.  This technique corrected predicted 100% of the non-transgenic samples and 90% of the transgenic samples in the validation set. Since there were no reference tests conducted on the samples and the classification was based strictly on what the retail label showed, it is possible that some of the samples were misclassified to start. However, the results shown in this study are good enough for screening purposes to classify transgenic and non-transgenic soybean oils.

Classification and Quantification of Palm Oil Adulteration Via Portable NIR Spectroscopy – Basri, Hussain, Bakar, et al., Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 173 (2017) 335-342 

Lard can produce cheap oil and can be mixed with other vegetable oils as an adulterant. A NIR spectrometer was set up to operate in both transmission and transflectance mode for comparative purposes.  Samples were prepared of pure palm oil, pure lard, and mixtures at various increments of added lard from 0.5% to 50%. Absorbance spectra of all samples were collected in both modes from 950 nm -1650 nm and exported for chemometric analysis. Classification analysis was performed on pure vs adulterated palm oil and regression models for quantifying lard in palm oil was created from both sets of spectra. 


Classification Accuracy0.95 
LardR2 = 0.9998


Classification Accuracy0.93
LardR2= 0.9996 

Results for transmission were slightly better in both classifying pure palm oil and adulterated samples and in measuring the % lard adulteration in palm oil. A Cumulative Adaptive Reweight Sampling (CARS) algorithm was applied to determine wavelengths critical to the model while eliminating noisy wavelengths. Recalculation of the model after applying CARS showed improved results and this contributed to the high correlation coefficients using both scanning modes.

Near and Mid Infrared Spectroscopy and Multivariate Data Analysis in Studies of Oxidation of Edible Oils – Wojcicki, Khmelinskii, Sikorski, Sikorska, Food Chemistry 187 (2015) 416-423 

Both FT-NIR and FT-MIR with an ATR crystal were used to study oxidation of retail samples of olive, sunflower, and rapeseed oils. Accelerated oxidative degradation of oils at 60°C was monitored for fifteen days using peroxide values and the samples were scanned in transmittance mode with both spectrometers.  NIR absorption spectra were collected from 12,500 cm-1 to 4000 cm-1 with thirty-two scans per average at 16 cm-1 resolution. MIR spectra were collected from 4000 cm-1 to 650 cm-1 with sixteen scans per average at 4 cm-1 resolution.  Three spectra were recorded for each sample using both instruments and peroxide values were recorded using the standard iodometric method. 


Peroxide Value (PV)R2= 0.970


Peroxide Value (PV)R2= 0.710


Peroxide Value (PV)R2= 0.954

Calibration models were created using the individual sets of NIR and MIR spectra as well as combining them both. Correlation demonstrated the ability to evaluate the oxidative stability of oils, particularly using the NIR method. One likely reason for the lower correlation using MIR is the smaller pathlength and less penetration into the sample. It is known that implementing NIR in a process setting is simpler and easier to do than using MIR, especially when it comes to sample preparation. The recommended next step is continued calibration work using different oils and the NIR spectrometer.

Monitoring PV in Corn and Soybean Oils by NIR Spectroscopy – Yildiz, Wehling, Cuppett, JAOCS, Vol. 79, no. 11 (2002) 

Samples of corn and soybean oils were scanned with an NIR spectrometer for the purpose of measuring oxidation levels by peroxide value. Absorbance spectra were collected from 400 nm to 2500 nm at 2 nm intervals. PV oxidation levels were recorded as reference values using a standard method. All chemical analyses were collected in duplicate and the mean was used for the final value.  

Corn Oil: 

Peroxide Value (PV)R2= 0.987

Soybean Oil: 

Peroxide Value (PV)R2= 0.996 

Corn and Soybean Oils 

Peroxide Value (PV)R2= 0.993 

Individual calibration models were created for both types of oil and for both types combined into one model. Correlation was excellent for all three models and the results show that the combined model is robust enough to incorporate both types of oil without introducing error. While the correlation coefficient is slightly lower for the combined model, prediction results using a validation set showed a statistically negligible difference between the individual and combined models.

Monitoring Lipase-Catalyzed Butterfat Interesterfication with Rapeseed Oil by Fourier Transform Near-Infrared Spectroscopy – Zhang, Mu, Xu, Anal Bioanal Chem (2006) 386:1889-1897 

This study used FT-NIR spectroscopy to monitor the enzymatic interesterfication process for butterfat modification.  Health conscious consumers have created a demand for alternative food products contained less saturated fat. Such demand has led to new products like a combination of butterfat and oil blends, made in a reactor catalyzed by different enyzmes.  For this study, a blend of butterfat and rapeseed oil were placed in a flatbed reactor and catalyzed by lipase.  Spectra were collected in transmission mode from 12000 cm-1 to 4000 cm-1 at 70°C. Reference values for conversion degree (evaluated from the triglyceride profile and obtained by the triglyceride peak ratio) and solid fat content (SFC) were collected. Triglyceride profiles were determined using reversed-phase HPLC and peaks were identified by triglyceride standards with a known equivalent carbon number (ECN). Peak ratio is defined as (ECN48/ECN46).  All reference values were determined in triplicate and averaged for one value.  Calibration models for these two parameters were created using the spectral data and reference values. 

Peak RatiR2= 0.935 
SFCR2= 0.939 

The best correlation for both parameters was obtained used the reduced wavenumber range from 5269 cm-1 – 4513 cm-1. It must be noted that the SFC only changed at 5°C and the change was only slightly greater than 5%. Because of this, peak ratio is much better suited for online monitoring of this reaction. If it were required to measure SFC, reasonable precision could be expected using the calibration model created here.