Fourier Transform Infrared (FTIR) Spectroscopy as a Utilitarian Tool for

Aug 30, 2015 - A primary Fourier transform infrared (FTIR) method capable of determining acidity in ester-based oils is described and evaluated. Absol...
21 downloads 22 Views 892KB Size
Article pubs.acs.org/JAFC

Fourier Transform Infrared (FTIR) Spectroscopy as a Utilitarian Tool for the Routine Determination of Acidity in Ester-Based Oils Xianghe Meng,† Lei Li,† Qin Ye,† and Frederik van de Voort*,§ †

Ocean College, Zhejiang University of Technology, 18 Chaowang Road, Hangzhou 310014, China McGill IR Group, Department of Food Science and Agricultural Chemistry, McGill University, Montreal H9X 3V9, Canada

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on September 10, 2015 | http://pubs.acs.org Publication Date (Web): September 9, 2015 | doi: 10.1021/acs.jafc.5b02738

§

ABSTRACT: A primary Fourier transform infrared (FTIR) method capable of determining acidity in ester-based oils is described and evaluated. Absolute free fatty acid (%FFA) and acid value (AV) calibrations were devised by spiking oleic acid into a refined, acid-free oil and measuring νCOO− at ∼1569 and νphenolate− at ∼1588 cm−1, respectively, in the second-derivative differential spectra. The FTIR acidity predictions were compared to the AOCS titrimetric method using acid mixtures as well as acid containing used vendor oils of undefined makeup and provenance, using two spectroscopically divergent reference oils as AC0. Relative to the AOCS reference method, the FTIR procedure was found to be both more accurate (±0.107 vs ±0.122) and reproducible (±0.025 vs ± 0.077) in determining %FFA and similar in predicting AV. The FTIRphenolate method overcomes a variety of limitations of earlier FTIR-based methods, being particularly simple and well suited to routine, semiautomated acidity analysis of ester-based oils using a basic FTIR spectrometer. KEYWORDS: Fourier transform infrared spectroscopy, FFA, AV, oil quality, oil analysis, oil quality control, oleic acid, p-toluenesulfonic acid



INTRODUCTION Fats and oils from a wide variety of natural sources are essential for food and industrial purposes, including biodiesel production. All ester-based oils, such as natural triacylglycerols or even synthetic esters, are susceptible to ester-linkage hydrolysis by moisture, releasing free fatty acids (FFA), effectively the only source of acidity. The quality of esterbased oils is largely governed by their FFA content, indicative of its degree of refinement or, alternatively, deterioration. As a common byproduct of edible oil extraction and processing, oils are refined to reduce their FFA levels to acceptable limits, which may subsequently rise again during storage. FFAs are more susceptible to oxidation and commonly result in the development of off-flavors and rancidity in edible oils as well as a reduced smoke point. In synthetic ester-based lubricants, mechanical and thermal stresses as well as oxidative processes can produce FFAs, which contribute to polymerization, development of gums, and metal corrosion. When used for biodiesel production, FFAs hinder the efficient conversion of triacylglycerols to methyl esters. Given the importance of FFAs in ester-based oils from the standpoint of quality and serviceability, well-established standard methods1,2 are available to quantitate their presence through their acidic contribution to the oil matrix. Acidity in edible oils is commonly expressed as %FFA, usually in terms of a particular fatty acid (e.g., oleic acid). Generally speaking, acid value (AV), or AN (acid number), the latter term being commonly used in the lubricant sector, is a more versatile means of expressing acidity in hydrophobic matrices. Regardless of oil matrix, the type of acid, or its mode of expression, standard analysis for oil acidity involves titration with a strong base, usually with the oil dissolved in protic solvent with the end point determined either colorimetrically or potentiometrically. © XXXX American Chemical Society

Titrimetric procedures, even when automated, are viewed as tedious, time-consuming, and costly and are particularly problematic in terms of solvent use and its disposal costs. This is particularly true for lubricant analysis; commercial laboratories around the world routinely carry out thousands of AN analyses per month. Consequently, there is great interest in alternate instrumental approaches that could speed such analyses while reducing the amounts of environmentally problematic solvents used. Fourier transform infrared (FTIR) spectroscopy has evolved significantly in this regard, initially as a qualitative tool for general oil condition monitoring to more sophisticated quantitative stoichiometric procedures with a focus on automation culminating in a viable and efficient automated FTIR method capable of determining AV in mineral-based lubricants at rates of ∼100 samples/h.3 FTIR acidity methods have tended to take two basic approaches, either making a direct spectroscopic measurement of the νCOOH of the fatty acids present in neat or solvent-diluted oils4−9 or using indirect approaches based on stoichiometric acid−base reactions used to produce and measure the fatty acid salt (νCOO−X+) or, alternatively, that of the conjugate acid formed, if it is infrared active and spectroscopically measurable. Generally speaking, with the assistance of chemometrics, the direct method can be a workable approach for any well-defined oil type, but tends to be impractical as a universal method given the effort involved in developing a generalized calibration. In addition, because most of these methods tend to deal with neat oils, viscosity poses sample handling problems in terms of cell Received: June 2, 2015 Revised: August 25, 2015 Accepted: August 29, 2015

A

DOI: 10.1021/acs.jafc.5b02738 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on September 10, 2015 | http://pubs.acs.org Publication Date (Web): September 9, 2015 | doi: 10.1021/acs.jafc.5b02738

Journal of Agricultural and Food Chemistry

selection, and baselines for calibration and validation, was carried out using Omnic (Nicolet, Madison, WI, USA) software. Analyses were carried out using a 100 μm CaF2 transmission flow cell equipped with Luer lock fittings, with the cell loaded by aspiration using a long stainless steel Leur-Lock needle connected via Tygon tubing to the cell and a three-way control valve, sample trap, and then vacuum. Sample Preparation. For spectroscopic analysis, a standardized sample preparation protocol was used, requiring the dispensing of 10 mL of 1% phenolate dissolved in 1-PrOH from a calibrated bottletopped dispenser into a 25 mL glass vial containing 5 g (±0.01) of oil. After reagent addition, the vial was capped and shaken until the solution was clear, after which it was ready for spectroscopic analysis. The sample can be taken either by opening the vial and aspirating the sample into the cell via the needle while using the valve or, if a septum cap is used, puncturing the septum with the needle and loading the cell. Calibration. Acid-free oil is required for calibration, with any refined oil serving this purpose if treated with activated silica gel to ensure that all polar constituents have been removed. Using this acidfree oil, oleic acid-spiked calibration standards (C0→n) were prepared, the base oil to which no oleic acid had been added serving as the calibration reference oil (C0). Using calibration samples prepared as above, a single-beam air spectrum was first taken, the cell loaded by aspiration, followed by a single-beam spectrum of the C0 solution to produce its absorption spectrum (AC0) and stored. The cell was then emptied by aspirating air through the cell, flushed with 1-PrOH, emptied again, and loaded with the next standard, the cell flushing step depending on whether any significant sample carry-over is encountered. Using the same air background, the single-beam spectrum of each of the subsequent spiked calibration standards was collected to produce the absorption spectrum of each standard (AC1→n). Differential spectra were then produced by subtracting AC0 from each spectrum of the standards and its 5-5 gap-segment secondderivative taken (Δ2AC1→n) and multiplied by 100 to magnify the small absolute peak height obtained to more workable values. For calibration, the peak height of the νphenolate− absorption (∼1588 cm−1) and the vCOO− of the FFA salt (∼1569 cm−1) in the Δ2AC1→n spectra were measured relative to a zero baseline. These responses are related by linear regression to AV and %FFA, respectively, calculated from the oleic added standard to produce the calibrations. Sample Analysis. The analysis of unknowns requires that a representative acid-free reference oil (OR), which is analogous to CA0 of the calibration set, be available, which is readily prepared by passing an oil through an activated silica gel column. Sample analysis parallels the calibration procedure by collecting the spectrum of the reagentreacted reference oil, OR, to produce AOR, which was stored. The cell was then loaded with the first reagent-reacted oil sample to obtain its absorption spectrum (AOS1); its differential spectrum was calculated (ΔAS1 = AOS1 − AOR) and its 5-5 gap-segment second-derivative taken (Δ2AS1) and multiplied by 100; the second-derivative calibration equations derived for FFA and/or AV were used to predict the respective values. Analytical Spectroscopy. A set of 11 calibration standards (C0− C11) were prepared covering a range 0−4% FFAoleic (0−7.96 mg KOH/g as AV) by the gravimetric addition of pure oleic acid (±0.1 mg) to a refined, silica gel-treated peanut oil (acid-free). Using the second-derivative calibrations, a set of representative oils (soy, peanut, and sunflower), spiked with known amounts of oleic acid, were prepared and analyzed by FTIR in duplicate for their AV/FFA content. These samples were also analyzed in duplicate according to AOCS Cd 3d-63/Cd 5a-40. Both the AOCS and FTIR results were compared to the gravimetrically added AV/FFA oleic acid values in terms of accuracy and reproducibility; the gravimetric data were considered to reflect the true acidity of the samples. Three additional sets of samples were prepared in another acid-free oil, one gravimetrically spiked with p-TSA, a second with oleic acid, individually, and a third with mixtures of two acids to determine whether carboxylic and noncarboxylic acid contributions could be differentiated and quantified properly. Subsequently, FTIR and AOCS predictions were compared by analyzing used oils of unknown

loading, sample cross-contamination, and the requirement for solvent washing between samples. Because of these issues, indirect stoichiometric FTIR methods are generally a better approach to quantitation,10−15 with a variety of bases having been used to convert FFAs to their salts. In these methods, the COO−X+ salts formed can be measured spectroscopically or, alternatively, the conjugate acid (X−H+) if it is infrared active and measurable. In all cases, solvents are used to reduce viscosity and facilitate sample handling. A complicating factor in the indirect stoichiometric procedures is how the samples are handled spectroscopically, in terms of being either split-sample or single-sample analyses. The distinction here is that in the split-sample approach, paired spectroscopic analyses are required, one of the oil samples with the reagent added (SR) and the other the oil sample with only the solvent (SS) added. By spectroscopically subtracting SS from SR, the differential spectrum (ΔS = SR − SS) contains only the spectroscopic contributions related to the chemical changes taking place, the other spectroscopic contributions having been ratioed out. Although this is the more ideal approach from the standpoint of quantification, it also requires the analysis of two samples to obtain a single result. This requirement is generally considered impractical from a sample preparation and handling perspective by commercial laboratories when considering an automated method.15 Most of these issues have been overcome by using sodium phenolate in 1-propanol (phenolate/1-PrOH) as the reagent in the single-sample mode, providing for oil− solvent miscibility as well as suitably located conjugated acid (phenol) absorption. By measuring the phenol formed at ∼1569 cm−1, an accurate, single-sample, automated turnkey method has been implemented to deliver ASTM-identical AN (AV) results for the mineral-based lubricants.15 To date, the FTIRphenolate procedure has only been applied to mineral-based oils with the objective of producing ASTM-identical AN results for lubricants.11 As such, turnkey, autosampler-based FTIR instrumentation equipped with custom ancillary software designed for commercial lubricant laboratory use is required rather than a conventional, standard benchtop FTIR spectrometer. This work was designed to develop and validate a generic, semiautomatable FTIR method capable of rapidly determining the acidity of ester-based oils using any basic FTIR spectrometer to produce results as, or more than, accurate as standard titration procedures.



MATERIALS AND METHODS

Materials. All reagents were of analytical grade, with sodium phenolate (phenolate), oleic acid, and p-toluenesulfonic acid (p-TSA) obtained from Aladdin Inc. (Shanghai, China) and 1-propanol (1PrOH) obtained from Ling Feng Inc. (Shanghai, China). Vegetable oils, including soy, sunflower seed, peanut, palm, were obtained from local Chinese retailers. Freshly prepared phenolate/1-PrOH was stored in and delivered from a bottle dispenser; inlet air passed through a solution of Ca(OH)2 (Sinopharm Chemical Reagent Co., Shanghai, China) to prevent atmospheric CO2 ingression, which leads to carbonic acid formation and consumption of the phenolate. Any oil used for calibration or as reference oil (OR) was passed through activated silica gel to remove any FFA, the silica gel (60−100 mesh) being preactivated at 280 °C for 4 h before use. Instrumentation. The instrument used was a Tensor 27 FTIR spectrometer (Bruker Inc., Bremen, Germany) controlled by OPUS software (Bruker Inc.), with all spectra collected at a resolution of 4 cm−1 using 16 scans (∼30 s scanning) over a spectroscopic range of 400−4000 cm−1. General spectroscopic processing, including spectroscopic subtraction, obtaining second-derivative spectra, region B

DOI: 10.1021/acs.jafc.5b02738 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry provenance obtained from local vendors to validate the overall performance of the FTIR method. For these analyses, two spectroscopically divergent oils were used as OR to determine if oiltype and/or other matrix effects affect the FTIR predictions relative to the AOCS results. Validation Analysis. The spectra obtained were analyzed using TurboQuant (version 7.2.0) (Thermo-Electron Corp., Madison, WI, USA), a spectral analytical and chemometrics package providing for region selection, spectroscopic transformations, calibration development, and concurrent prediction of validation samples relative to known or chemically determined values as well as facilitating statistical analysis of the results.

will differ and a conversion ratio of FFA to AV (1.99 as oleic) will produce a lower result. Figure 1 illustrates the simple differential and 5-5 gapsegment second-derivative spectra obtained for phenolatereacted, oleic acid-spiked peanut oil calibration standards after subtraction of the spectrum of acid-free AC0. The changes in the differential spectra reflect the loss of phenolate and the formation of COO−Na+, but are spectroscopically difficult to interpret due to spectroscopic overlap. However, after taking the gap-segment second-derivative, these absorption changes are more clearly parsed out, with the negative phenolate band at ∼1588 cm−1 becoming smaller with acid concentration, whereas the positive oleate band at ∼1569 cm−1 grows proportionately, thus providing means of differentiating between carboxylate (FFA) acidity and total acidity contributions (AV). Figure 2 presents calibration plots devised on the basis of the second-derivative spectra by measuring the minimum and maximum, respectively, of the COO− and phenolate bands in each spectrum and relating the peak height to the %FFA and AV, respectively. The calibration equations obtained for the Δ2AC1→n spectra in terms of AV and FFA based solely on oleic acid being present are

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on September 10, 2015 | http://pubs.acs.org Publication Date (Web): September 9, 2015 | doi: 10.1021/acs.jafc.5b02738



RESULTS AND DISCUSSION Calibration. As noted, the indirect phenolate-based FTIR method (FTIRphenolate) has been applied only to mineral-based oils as a single-sample, automated procedure to determine AN with the objective of obtaining ASTM-identical results. In this work, the objective was to simplify and assess the performance of the FTIRphenolate procedure for its ability to analyze nonphosphate ester-based oils so as to make the FTIR procedure more generic and utilitarian. The phenolate stoichiometric reaction used for the analysis for AV in hydrophobic matrices can be generalized as

FTIRAV = 13.636 × Abs(1588cm−1) + 0.230

(3)

R2 = 0.9975, SD = 0.1441

FTIR FFA = − 5.454 × Abs(1569cm−1) + 0.047

In the case of ester-based oils, carboxylic acids are effectively the only source of acidity:

(4)

R2 = 0.9989, SD = 0.0469

Both of these calibrations are highly linear and their respective SD values reflect a coefficient of variation (CV) of ∼2.0 and 3.5%, respectively. Validation. With these calibration relationships in hand, one can theoretically differentiate between carboxylate and other forms of acidity using the carboxyl (%FFA) and phenolate (AV) response at 1569 and 1588 cm−1, respectively. To test this as well as the overall reproducibility and accuracy of FTIR method in determining acidity relative to the AOCS titrimetric method, soybean oil was gravimetrically spiked with oleic acid as well as with p-TSA, one set of each acid spiked into the oil alone, with another set where the two acids were present

In eqs 1 and 2, all acidity with a pKa value 1 order greater than that of phenolate should be readily measurable via the spectroscopic loss of the vphenolate absorption, whereas in eq 2, one may have the option of measuring acidity using either the vphenolate, vCOO−, or both as a measure of FFAs. By using acid-free oil to calibrate and pure oleic acid as the acid source, a calibration devised in this manner should produce an identical response regardless of the mode of measurement. However, when acidity other than FFAs is present, AV and %FFA results

Figure 1. (A) Differential and (B) 5-5 gap segment second-derivative spectra for acid-free peanut oil spiked with oleic acid using sodium phenolate in 1-PrOH as the reagent to carry out the acid/base reaction. Spectra were recorded in a 103 μm cell at 4 cm−1 resolution. C

DOI: 10.1021/acs.jafc.5b02738 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on September 10, 2015 | http://pubs.acs.org Publication Date (Web): September 9, 2015 | doi: 10.1021/acs.jafc.5b02738

Figure 2. Calibration plots of acid-free oil spiked with oleic acid and reacted with phenolate obtained using TurboQuant: (A) AV versus peak height at ∼1588 cm−1; (B) FFA% versus peak height at 1569 cm−1.

Figure 3. Collation of all second-derivative validation spectra illustrating the region over which the phenolate and COO−Na+ absorptions occur.

Figure 4. (A) Plot of AV calibration data points (●) and validation predictions (○). (B) Similar plot for %FFA, with the p-TSA containing samples (⊗) not being predictable.

together; all were also analyzed according to the AOCS method. In addition, a series of used oils of unknown provenance were analyzed according to the AOCS method and FTIR to compare their performance to gravimetric addition. The second-derivative spectra obtained for the calibration and validation samples were combined, a new calibration was derived, and the validation samples were predicted and compared to the known added acid amounts and/or chemically determined values using TurboQuant. All analyses were carried out in duplicate, including the reference titrimetric procedure, so as to obtain a comparative baseline in

terms of accuracy (a) and reproducibility (r). On the basis of these duplicate analyses the mean difference (MDr) and standard deviation of the difference (SDDr) for reproducibility were determined to be 0.032 and 0.077 mg KOH/g for AV and 0.008 and 0.022% for FFA, respectively. Both the gravimetrically prepared calibration standards and the oleic/p-TSA samples can serve as a measure of the accuracy (a) of the titration procedure, with the MDa and SDDa obtained being 0.071 and 0.122 mg KOH/g oil, respectively, for AV, using the mean value of the duplicate analyses. Figure 3 presents a D

DOI: 10.1021/acs.jafc.5b02738 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Journal of Agricultural and Food Chemistry

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on September 10, 2015 | http://pubs.acs.org Publication Date (Web): September 9, 2015 | doi: 10.1021/acs.jafc.5b02738

Figure 5. (A) Comparison plot of AV results obtained by AOCS and FTIR methods for used oil. (B) Similar plot for %FFA results.

composite grouping of all the validation second-derivative spectra collected. These spectra are somewhat similar to the calibration spectra presented in Figure 1 but demonstrate substantially more variability in the phenolate band maxima. The COO− band is more uniform, but confounded to some extent as some oils contain only p-TSA, which lacks the COO− absorption. Even so, AV, defined as a measure of all acid contributions, carboxylic acids as well as noncarboxylic acids, can still be accurately measured via the loss of the phenolate band as seen in Figure 4A, which compares the FTIR predictions against the mean of the AOCS titration results for the same samples. The ability to predict acids other than carboxylic acids correctly using the COO− absorbance is by definition not possible and is confirmed in Figure 4B, where ⊗ represents only p-TSA-containing oils. Although this is an extreme example and unlikely to occur in reality, it demonstrates that, should acids other than carboxylic acids exist, the %FFA prediction will be erroneous relative to the titrimetrically determined value in such circumstances. Included in both of these assessments are used oils of unknown provenance in terms of oil type, requiring the use of an arbitrarily selected, acid-free reference oil (OR). To assess whether the oil used as OR affects the FTIR results significantly relative to the AOCS data, two chemically and spectroscopically divergent oils (soybean and palm oils), differing significantly in their saponification number (SN) and iodine values (IV) were used as OR so as to generate their respective differential spectra. The FTIR AV and %FFA predictions of these used oils are compared to their mean AOCS AV and FFA results and illustrated in Figure 5, panels A and B, respectively, linear regression producing the following relationships: AVFTIR = 1.088 × AVAOCS − 0.280

Table 1. MD and SDD for Reproducibility (r) and Accuracy (a) for Acidity of Oleic Acid and p-Toluenesulfonic AcidSpiked Oils AVAOCS MDr SDDr MDa SDDa

AVFTIR C18:1 0.012 0.028 0.025 0.107

a

AVFTIRp‑TSA

FFAFTIR C18:1

FFAFTIRp‑TSA

0.018 0.041 0.088 0.166

0.008 0.022 0.056 0.032

NAb NA NA NA

a

Absolute AV and COOH measurements from which these statistics are derived differ by a factor of ∼2, which affects their relative magnitudes. bNA, not applicable by definition to FTIR quantitation of FFA.

as well as in terms of accuracy (MDa and SDDa) of the AOCS and FTIR results relative to the amounts of acid added. Comparison of both the AV and FFA results illustrates that the FTIRphenolate procedure consistently produces results statistically closer to the gravimetrically added acidity for COOH from the standpoint of either accuracy or reproducibility relative to the AOCS reference method. This confirms that the method is on firm ground in being considered a primary method in its own right. In the case of the p-TSA measurement, the accuracy is slightly lower, but of similar magnitude. Although one gains the impression that the COOH measure is significantly superior to that of the AV measurement on the basis of these comparative statistics, these differences are to a large degree a result of the relative magnitude of the measurements when the conversion ratio of FFA to AV is 1.99 rather than analytical differences. As structured, the FTIRphenolate method performs well in determining the acidity of ester-based oils, expressed either as %FFA or as AV. Like the standard AOCS methods, this FTIR method is a primary, rather than a secondary, method. Calibration simply requires the standard addition of oleic acid to any refined, activated silica gel treated ester-based oil to ensure no residual inherent acidity exists to bias the calibration. For the analysis of unknown samples, any reasonably representative, acid-free oil can serve as reference oil, again, readily prepared by treating it with activated silica gel. Oil-type effects on quantification are minimal and consistent, with optimal results achieved using second 5-5 gap-segment derivative differential spectra. As configured, sample handling is facile, and accurate results can be obtained for all ester-based oils, other than phosphate esters, making the method of general utility even if the oil type or provenance is undefined. Because

(5)

SD = 0.097, R2 = 0.989

% FFAFTIR = 1.099 × %FFAAOCS − 0.124

0.032 0.077 0.071 0.122

a

(6)

SD = 0.070, R2 = 0.980

On the basis of these results, there is minor oil-dependent bias introduced depending on which oil is used as OR; however, as long as one consistently uses a particular oil as OR, the results will track consistently. Table 1 presents the overall performance of the methods relative to Figure 4 in terms of mean differences (MDr) and standard deviation of the difference (SDDr) between duplicates E

DOI: 10.1021/acs.jafc.5b02738 J. Agric. Food Chem. XXXX, XXX, XXX−XXX

Article

Downloaded by UNIV OF CALIFORNIA SAN DIEGO on September 10, 2015 | http://pubs.acs.org Publication Date (Web): September 9, 2015 | doi: 10.1021/acs.jafc.5b02738

Journal of Agricultural and Food Chemistry

olein using Fourier transform infrared spectroscopy. J. Am. Oil Chem. Soc. 1999, 76, 485−490. (6) Bertran, E.; Blanco, M.; Coello, J.; Iturriaga, H.; Maspoch, S.; Montoliu, I. Determination of olive oil free fatty acid by Fourier transform infrared spectroscopy. J. Am. Oil Chem. Soc. 1999, 76, 611− 616. (7) Verleyen, T.; Verhe, R.; Cano, A.; Huyghebaert, A.; De Greyt, W. Influence of triacylglycerol characteristics on the determination of free fatty acids by Fourier transform infrared spectroscopy. J. Am. Oil Chem. Soc. 2001, 78, 981−984. (8) Sherazi, S. T. H.; Mahesar, S. A.; Bhanger, M. I.; van de Voort, F. R.; Sedman, J. Rapid determination of free fatty acids in poultry feed lipid extracts by SB-ATR FTIR spectroscopy. J. Agric. Food Chem. 2007, 55, 4928−4932. (9) Yu, X. Z.; van de Voort, F. R.; Sedman, J.; Gao, J. M. A new direct Fourier transform infrared analysis of free fatty acids in edible oils using spectral reconstitution. Anal. Bioanal. Chem. 2011, 401, 315− 324. (10) Ismail, A. A.; van de Voort, F. R.; Emo, G.; Sedman, J. Rapid quantitative determination of free fatty acids in fats and oils by Fourier transform infrared spectroscopy. J. Am. Oil Chem. Soc. 1993, 70, 335− 341. (11) van de Voort, F. R.; Sedman, J.; Yaylayan, V.; Saint-Laurent, C. The determination of acid and base number in lubricants by FTIR spectroscopy. Appl. Spectrosc. 2003, 57, 1425−1431. (12) Al-Alawi, A.; van de Voort, F. R.; Sedman, J. New FTIR method for the determination of FFA in oils. J. Am. Oil Chem. Soc. 2004, 81, 441−446. (13) Al-Alawi, A.; van de Voort, F. R.; Sedman, J. A new FTIR method for the analysis of low levels of FFA in refined edible oils. Spectrosc. Lett. 2005, 38, 389−403. (14) Aryee, A. N. A.; van de Voort, F. R.; Simpson, B. K. FTIR determination of free fatty acids in fish oils intended for biodiesel production. Process Biochem. 2009, 44, 401−405. (15) van de Voort, F. R.; Ghetler, A.; García-González, D. L.; Li, Y. D. Perspectives on quantitative Mid-FTIR spectroscopy in relation to edible oil and lubricant analysis: evolution and integration of analytical methodologies. Food Anal. Method. 2008, 1, 153−163.

the method uses an oil-miscible solvent as the phenolate carrier, sample handling issues associated with the high viscosity oils are minimized, facilitating rapid, even semiautomated, analysis. To the extent that a spectrometer and its attendant software allow one to program the spectroscopic data processing steps, it is possible to provide results directly in terms of %FFA or AV. As configured, the FTIRphenolate method differs significantly from its specialized automated counterpart used to determine AN in mineral-based lubricants, which requires more sophisticated chemometrics to ensure delivery of ASTMidentical results. As such, this method is a hybrid procedure, combining elements of both the split-sample and the singlesample approaches, capable of ratioing out (subtracting) the oil spectrum, on the one hand, while avoiding the need for paired analysis, on the other. The sensitivity of the method can be increased or decreased simply by varying the path length proportionately. Although the method as developed here is based on a fixed oil-reagent ratio, one can develop a more generalized procedure by including an oil concentration calibration using the CH overtone region in the near IR portion of the spectrum (4500−4000 cm−1). Using this approach one can spectroscopically determine the oil content of the sample prepared and dynamically normalize the differential spectra to vary the sensitivity. As designed, this simple FTIR procedure uses minimal amounts of solvent and sample, handles multiple samples readily, and provides accurate acidity results for ester-based oils, making it a utilitarian alternative to titrimetric procedures if a basic FTIR spectrometer is available.



AUTHOR INFORMATION

Corresponding Author

*(F.v.d.V.) Phone: (514) 632-7091. Fax: (514) 398-8124. Email: [email protected]. Funding

We gratefully acknowledge the financial support provided by the National Natural Science Foundation of China (No. 31271887). Notes

The authors declare no competing financial interest.



ABBREVIATIONS USED FFA, free fatty acid; AV, acid value; p-TSA, p-toluenesulfonic acid; 1-PrOH, 1-propanol; OR, acid-free reference oil; MD, mean difference; SDD, standard deviation of the difference; SN, saponification number; IV, iodine values



REFERENCES

(1) AOCS. Official Methods and Recommended Practices of the American Oil Chemists’ Society; AOCS Press: Champaign, IL, USA, 1997; Method Cd 3d-63. (2) AOCS. Official Methods and Recommended Practices of the American Oil Chemists’ Society; AOCS Press: Champaign, IL, USA, 2009; Method Cd 5a-40. (3) Winterfield, C.; van de Voort, F. R. Automated acid and base number determination of mineral-based lubricants by Fourier transform infrared spectroscopy: commercial laboratory evaluation. J. Lab. Autom. 2014, 19, 577−586. (4) Lanser, A. C.; List, G. R.; Holloway, R. K.; Mounts, T. L. FTIR estimation of free fatty acid content in crude oils extracted from damaged soybeans. J. Am. Oil Chem. Soc. 1991, 68, 448−449. (5) Man, Y. B. C.; Moh, M. H.; van de Voort, F. R. Determination of free fatty acids in crude palm oil and refined-bleached-deodorized palm F

DOI: 10.1021/acs.jafc.5b02738 J. Agric. Food Chem. XXXX, XXX, XXX−XXX