Quantitative determination of the iodine values of unsaturated plant oils using infrared and Raman spectroscopy methods

ABSTRACT

Thirteen edible oils: sunflower, avocado, hemp, high-linolenic flax, low-linolenic flax, safflower, walnut, roasted sesame, rice, corn, rapeseed, pumpkin seed, and hazel were studied in this work. Their fatty acid composition, iodine, acidic, peroxide, and saponification values were determined. Infrared and Raman spectra of the oils were recorded in the range 400–3200 cm⁻¹. The integral intensities of the bands at about 1655 and 2852 cm⁻¹ corresponding to ν(C=C) and ν(CH₂) vibrations were evaluated and used to construct a relationship between the spectroscopic data and the iodine value. The following linear dependencies were obtained: I_ν(C=C)/I_ν(CH2) = 7.449 × 10⁻⁴ × iodine value – 0.0339 and I_ν(C=C)/I_ν(CH2) = 9.299 × 10⁻⁴ × iodine value – 0.023 for the infrared and Raman spectra with a correlation coefficient 0.988 and 0.976, respectively. These calibration lines can be used to determine the iodine value for oils with unknown unsaturation level, and may be applied for margarines and other fatty materials.

KEYWORDS:

• Plant oils
• Infrared and Raman spectra
• Lorentz deconvolution
• Iodine value
• Quantitative analysis of unsaturation degree

Introduction

Infrared (IR) and Raman (RS) studies have been widely used to characterize oils and fats for years. The first articles on this topic were published in the early 1950s.^[1,2] Several problems were solved using these spectroscopic methods. Oil authentication and genuineness was the first aim of these studies.^[3–30] Detection of adulteration of authentic oils with cheaper oils was another important problem that was solved with the use of IR and RS spectroscopy.^[31–60] These methods were also widely applied as an analytical tool in the fat industry for the determination of the unsaturation level of fats and oils.^[2,61–72] The total degree of unsaturation was evaluated from the quantitative measurement of the ν(C=C) band intensity and its relation to the intensity of the band related to the δ(CH₂) scissoring. This was achieved by controlling the ratio of the height of the bands at about 1660 and 1445 cm⁻¹ corresponding to the above vibrations. In another approach oxidation of unsaturated fatty acids was monitored using the change of the observed at about 3010 cm⁻¹ ν(C=C-H) band intensity as an indicator. In such isomers the ratio of this band intensity to the intensity of the band at about 1265 cm⁻¹ (corresponding to =C-H deformation of the unconjugated cis double bond) is correlated with the degree of unsaturation.^[64]

The IV is a parameter that characterizes the unsaturation level of an oil. The Fourier Transform Mid Infrared Attenuated Total Reflectance (FT-MIR/ATR) instrumentation was used by van de Voort et al.^[73] to determine this parameter in triglycerides. A comparison of the intensities of the bands from the ranges 2600–3200 and 1000–1600 cm⁻¹ was proposed. A strong correlation between the iodine value (IV) and the ratio of the RS band intensities from the ranges 1620–1690 cm⁻¹ [ν(C=C) mode] and 1420–1480 cm⁻¹ [δ(CH₂) mode] was found.^[62] The FT-RS technique was used by Sadeghi-Jarabchi et al.^[63] to determine the IV of oils and margarines on the basis of the intensities of the observed bands at about 1650 cm⁻¹ [ν(C=C)] and 1440 cm⁻¹ [δ(CH₂)] as well as 3010 cm⁻¹ [ν(=C-H)] and 1270 cm⁻¹ [γ(=C-H) or ν(C-C)]. Similar methods were employed by other authors in determination of the IV on the basis of the spectroscopic data.^[67,74] The review of these results, obtained by several spectroscopic methods, was presented by Lee et al.^[75]

Linear discriminant analysis (LDA) was applied a number of times to discriminate and classify various edible oils on the basis of spectral data. Yang et al.^[19] showed that, generally, Fourier transform infrared spectroscopy (FT-IR), Fourier transform near-infrared spectroscopy (FT-NIR), and Fourier Transform - Raman Spectroscopy (FT-RS) studies can be used to rapidly classify an oil, but the FT-IR method is the most effective for these purposes and it is followed by FT-RS method. FT-NIR spectroscopy was recognized as a less useful method than the former ones.^[76,77] In these studies the IV was used as one of the factors analyzed in the LDA procedure. In the present work, we propose new methods of discrimination analysis based on FT-IR and FT-RS spectra measurements. The results of chemometric studies of thirteen different oils were compared and these data were related to integral intensities of selected IR and RS bands.

Materials and methods

All the studied oils were produced by Polish Oleofarm factory located in Wrocław (Poland). The IV was determined taking 0.13 g of an oil, dissolving it in 20 mL of a cyclohexane and acetic acid mixture (mixing ratio 1:1 v/v). Next, 25 mL of Wijs solution was added and the mixture was kept in dark for 1 h. Twenty milliliters of saturated solution of KI and 150 mL distilled water was added to this solution and the whole mixture was titrated with 0.1 N sodium thiosulphate in the presence of starch until the color changed from crimson-violet to bright yellow.

Determination of the acidic value (AV) was performed using the following procedure: 10 g of an oil were dissolved in 100 mL of ethanol and ethyl ether mixture (mixing ratio 1:1 v/v). This solution was titrated with KOH ethanolic solution until reaching pink-violet color. Phenolphthalein was used as an indicator.

The peroxide value (PV) was determined according to the method proposed in the PN-ISO 3960. Two grams of an oil were placed in an iodine flask and dissolved in a blended solution of 50 mL chloroform and acetic acid (mixing ratio 2:3 v/v). A saturated solution of KI (0.5 mL) was added. The mixture was intensively shaken and kept in the dark for 5 min. After the addition of 30 mL distilled water, the mixture was titrated with sodium thiosulphate (0.01 M) until the yellow color disappeared. Then, roughly 0.5 mL of starch indicator (0.05%) solution was added. The titration was carried out until the violet color changed to yellow.

The saponification value (SV) was determined by the method in which 2 g of an oil were added to 25 mL of NaOH dissolved in ethanol and heated in a water bath for 20 min. The obtained mixture was titrated with 0.5 N HCl solution in the presence of phenolphthalein up to decolorization of the mixture. The oils composition and their acid profiles were measured using the gas chromatographic gas chromatography-mass spectrometry (GC-MS) method.

FT-IR/ATR spectra were recorded in the range 300–4000 cm⁻¹ using a Nicolet 6700 spectrometer equipped with a portable ATR set. The resolution of these measurements was 2 cm⁻¹. FT-RS spectra were recorded in the range 80–4000 cm⁻¹ using a Bruker RFS 110 spectrometer with a diode-pumped Nd:YAG laser emitting at 1064 nm. The spectra were accumulated from 150 scans measured with the resolution of 2 cm⁻¹. The background correction of the spectra was performed with the use of operating OMNIC software Suite 9 v.9.5 of the spectrometer. The spectral contours were analyzed by the commercial computer program Origin 7.5. This analysis included a background subtraction and deconvolution of the experimental bands into Lorentz components. To eliminate an accidental intensity variation in the general intensity level all the integral intensities of the observed bands were standardized using the statistical R² coefficient. For this aim, several simulations of the Lorentz deconvolution were performed using wide and variable number of components. The best fitting between the experimental and theoretical spectral course was reached when their statistical R² values were the closest to 1.0. The PCA analysis was performed using The Unscrambler program equipped with non-linear iterative partial squares (NIPALS) algorithm and cross-validation.

Results and discussion

Chemical characterization of the studied oils

Table 1 presents the chemical composition of the studied: sunflower, avocado, hemp, high-linolenic flax, low-linolenic flax, safflower, walnut, roasted sesame, rice, corn, rapeseed, pumpkin seed, hazel oils. Their fatty values are presented in Table 2.

Table 1. Chemical characterization of the studied oils.

Fatty acids composition (%)	Oil
	Sunflower	Avocado	Hemp	High-linolenic flax	Safflower	Walnut	Roasted sesame	Rice	Corn	Rapeseed	Pumpkin seed	Low-linolenic flax	Hazel
C-14:0 Lauric acid	—	—	—	—	—	—	—	—	0.02	—	0.01	—	—
C-14:0 Myristic acid	0.08	—	—	—	0.11	—	0.03	0.34	0.04	0.05	0.11	—	—
C-16:0 Palmitic acid	3.58	17.33	5.99	5.91	6.93	7.27	10.11	20.05	10.15	4.39	12.3	6.23	6.13
C-16:1 Palmitooleic acid	0.11	6.62	—	—	0.07	—	0.21	0.17	0.11	0.19	—	—	—
C-18:0 Stearic acid	2.32	0.66	2.57	2.83	2.72	2.57	4.28	1.8	2.2	1.52	5.69	3.85	2.98
C-18:1 Oleic acid	86.24	59.62	11.88	11.27	11.33	17.08	41.58	42.7	31.98	58.38	35.34	16.27	77.62
C-18:2 Linoleic acid	5.23	9.06	56.49	15.68	77.22	60.12	41.51	31.83	53.91	18.4	44.51	70.98	10.65
C-18:3 Linolenic acid	0.41	0.68	20.97	63.13	0.19	11.22	0.58	1.04	0.75	8.83	0.26	1.65	—
C-20:0 Arachidic acid	0.15	0.04	—	—	0.34	—	0.33	0.59	0.27	0.37	—	—	—
C-20:1 Gadoleic acid	0.24	—	—	—	0.13	—	0.1	—	0.16	1.05	—	—	—
C-22:0 Behenic acid	0.5	—	—	—	0.11	—	—	—	—	0.09	—	—	—

Table 2. The fatty values determined in the present work.

Oil	IV (g/100 g fat)	AV (mg KOH/g fat)	PV (mw O2/1000 g fat)	SV (mg KOH/g fat)
Sunflower	84.6	3.27	1.46	188.17
Avocado	89.47	0.22	6.32	201.65
Hemp	173.43	1.45	4.64	194.25
High-linolenic flax	212.1	2.21	1.92	191.88
Safflower	161.34	1.11	6.53	202.66
Walnut	160.53	3.12	4.64	194.25
Roasted sesame	120.76	2.01	0.92	194.68
Rice	105.96	0.22	3.87	182.54
Corn	129.8	2.02	1.9	191.44
Rapeseed	113.04	0.63	3.72	189.09
Pumpkin seed	122.2	1.02	3.67	195.65
Low-linolenic flax	188.7	0.66	0.67	187.7
Hazel	84.6	0.34	0.59	183.03

IV: iodine value; AV: acidic value; PV: peroxide value; SV: soaping value.

IR and RS studies

The FT-IR and FT-RS spectra of the studied oils are presented in Figs. 1–4. Table 3 lists the wavenumbers of the measured spectra of walnut oil, the spectra are representative for all the studied samples. The assignment of the bands to the respective normal modes was made using commonly accepted nomenclature.^[8,27]

Table 3. Wavenumbers observed in the IR and Raman spectra of the walnut oil.

Number	IR	Raman	Assignment^[Citation3,Citation22,Citation27,Citation30,Citation76]
1.	3005 m	3005 m	νas(= C-H) trans
2.	2952 msh	2954 msh	νas(CH3)
3.	2922 vs	2925 s	νas(CH2)
4.		2895 s	νas(CH2)
5.		2873 s	νs(CH3)
6.	2852 s	2853 vs	νs(CH2)
7.		2726 w	combination
8.	1742 s	1748 w	ν(C = O)
9.	1648 w	1655 m	ν(C = C)
10.	1463 m	1455 msh	δas(CH3) + δas(CH2)
11.		1438 m	δas(CH3) + δas(CH2)
12.	1418 vw		ρ(= C-H)
13.	1378 w		δs(CH3)
14.	1359 w	1362 vw	δs(CH2)
15.	1346 w		ω(CH2)
16.	1318 w		ω(CH2)
17.	1297 w	1303 m	τ(CH2)
18.	1279 w	1268 m	ω(CH2) + ν(C-O)
19.	1237 w		ν(C-O)
20.	1161 m		ν(C-O)
21.	1119 w	1117 vw	ν(C-C)
22.	1095 w		ν(C-C)
23.		1082 w	ν(C-C)
24.	1061 vw	1065 w	ν(C-C)
25.	1031 vw	1022 w	ν(C-C)
26.		971 vw	γ(= CH2)
27.	964 vw		δ(C-C = C)
28.	906 vw		γ(CH2)
29.		869 vw	γ(CH2)
30.	856 vw		γ(CH2)
31.		839 vw	γ(CH2)
32.	767 vw		γ(CH2)
33.	723 w	720 vw	γ(C-C-C)
34.	692 wsh		ρ(CH2) + γ(C-C-C)
35.	588 vw		ρ(CH2) + γ(C-C-C)

Table 4. The wavenumbers (ν) and integral intensity (A) of some selected bands from the IR and Raman spectra of the studied oils.

Oil	ν(CH2)				ν(C = C)
	IR		Raman		IR		Raman
	ν	A	ν	A	ν	A	ν	A
Hazel	2852	18.35	2853	28.67	1655	0.52	1655	1.79
Sunflower	2852	17.68	2852	20.81	1655	0.51	1655	1.32
Avocado	2852	16.95	2853	18.51	1655	0.50	1656	1.23
Rice	2852	20.77	2853	27.33	1656	0.81	1656	1.91
Rapeseed	2852	19.35	2853	20.66	1655	1.15	1655	1.58
Roasted sesame	2852	20.65	2854	23.99	1656	1.22	1655	1.62
Pumpkin seed	2852	16.13	2852	21.35	1656	0.93	1656	2.04
Corn	2853	15.80	2854	17.98	1655	0.98	1657	1.81
Walnut	2852	15.22	2854	21.00	1655	1.20	1655	2.41
Safflower	2852	17.64	2853	18.75	1655	1.56	1657	2.19
Hemp	2853	17.62	2852	18.14	1655	1.73	1657	2.35
Low-linolenic flax	2853	16.66	2852	12.46	1655	1.65	1657	2.00
High-linolenic flax	2852	18.48	2852	14.41	1655	2.37	1655	2.71

Figure 1. ATR/FT-IR spectra of oils A: Sunflower; B: Corn; C: Low-linolenic flax; D: High-linolenic flax; E: Walnut; F: Hazel; and G: Rapeseed.

Figure 2. ATR/FT-IR spectra of oils H: Roasted sesame; I: Avocado; J: Hemp; K: Safflower; L: Pumpkin seed; M: Rice.

Figure 3. Raman spectra of oils A: Sunflower; B: Corn; C: Low-linolenic flax; D: High-linolenic flax; E: Walnut; F: Hazel; G: Rapeseed.

Figure 4. Raman spectra of oils H: roasted sesame; I: avocado; J: hemp; K: safflower; L: pumpkin seed; and M: rice.

Quantification of chemical and spectroscopic data

The fatty values presented in Table 2 may be used to construct a relationship between the spectroscopic data and the IV. Similar correlations were proposed in earlier works but their authors used other procedures. Sinclair et al.^[2] discovered the existence of a linear relationship between the number of cis C=C bonds of unsaturated fatty acid methyl esters and the ratio of the 2920 cm⁻¹ band absorbance and the difference between the absorbances of the bands at 2920 and 3020 cm⁻¹. This result was confirmed by Chapman^[78] as the method allowed to determine the unsaturation degree. In a similar procedure Arnold and Hartung^[61] used the ratio of absorbances at 3030 and 2857 cm⁻¹. Afran and Newbery^[65] in such considerations used the absorbance of the bands at 3010 and 2854 cm⁻¹, but Muniategui et al.^[67] and Bernard and Sims^[63] evaluated the total degree of unsaturation from the absorption intensities of the bands at 3007 and 1658 cm⁻¹.

The correlation between the IV and some IR and RS bands intensities were described by van de Voort et al.,^[73] Bailey and Horvat,^[62] as well as by Sadeghi-Jorabchi et al.^[3] In these considerations, the bands from the ranges 2600–3200 and 1000–1600 cm^−1,[73] or 1626–1691 and 1420–1478 cm^−1,[62] as well as 1656 and 1444 cm^−1[3] were taken into account. Another pair of the bands observed at 3010 and 1270 cm⁻¹ was applied by Sadeghi-Jorabchi et al.^[64] and Baeten et al.^[74] for determination of the IV. All these were complicated, time consuming, and giving great dispersion of the results. In our work, we propose a faster and simpler procedure in which commonly used and easily accessible computer programs are employed.

The comparison of the obtained in the present work FT-IR/ATR and FT-RS spectra of 13 edible oils shows that the most characteristic bands for these samples are those at about 1650 and 2850 cm⁻¹ corresponding for ν(C=C) and ν(CH₂) vibrations, respectively. Choosing these vibrations allowed to construct a plot of integral intensities of ν(C=C) / ν(CH₂) bands versus the IV. Instead of the earlier proposed methods we developed a three step procedure. In the first step, the IR and RS spectral contours observed in the ranges 1500–1750 and 2800–3010 cm⁻¹ were deconvoluted into Lorentz components using the Origin 7.5. In the second step, the integral intensities of the Lorentz contours at about 1650 and 2850 cm⁻¹ were used to calculate the ratio I_ν(C=C)/I_ν(CH2). Finally, the relationship between this ratio and the IV was drawn. The plots obtained for 13 oils studied in the present work are shown in Figs. 5 and 6.

Figure 5. The relationship between the iodine value and integral intensity ratio of IR bands at about 1655/2852 cm⁻¹ of oils: A: hazel; B: sunflower; C: avocado; D: rice; E: rapeseed; F: roasted sesame; G: pumpkin seed; H: corn; I: walnut; J: safflower; K: hemp; L: low-linolenic flax; and M: high-linolenic flax.

Figure 6. The relationship between the iodine value and integral intensity ratio of Raman bands at about 1655/2850 cm⁻¹ of oils: A: hazel; B: sunflower; C: avocado; D: rice; E: rapeseed; F: roasted sesame; G: pumpkin seed; H: corn; I: walnut; J: safflower; K: hemp; L: low-linolenic flax; and M: high-linolenic flax.

For the relationships derived from the IR and RS studies the straight lines with the equations I_ν(C=C)/I_ν(CH2) = 7.449 × 10⁻⁴ × IV – 0.0339 and I_ν(C=C)/I_ν(CH2) = 9.299 × 10⁻⁴ × IV – 0.023 were obtained, respectively. The correlation coefficients for these relations were 0.988 and 0.976, respectively, showing a very good accordance. These calibration lines could be used to determine the IV for oils with unknown unsaturation level, and may be applied for margarines and other fatty materials. The dependences between the IV values and the IR and RS bands integral intensity ratios were approximated by straight lines drawn using Origin 7.5.

Both IR and RS spectra were analyzed using PCA in order to visualize the main trends in the data. The spectra were first- and second-order differentiated before the principal components (PCs) analysis. The first and second derivatives of the IR spectra are shown in Fig. 7a. The PLS regression was calibrated with the calibration sample set and the validation was obtained by full cross-validation. The optimum number of PCs used in the calibration model was checked by residual variance plot against the PCs, which proves how the variation in the spectra is described by different components. The explained variance PC 1 is c.a. 94% and PC 2 is 3% (Fig. 7b). It means that the pair of the bands observed at 1650 and 2950 cm⁻¹, chosen among other considered pairs of bands, gives the best fitting to the IV values of the studied oils. They were properly chosen and they differentiate the oils in regard to the IV determination.

Figure 7. The results of the PCA analysis. A: Loading from the first (PC-1) and second (PC-2) principal components of ATR-FTIR obtained for the studied oils; B: Explained variance from the PCs model.

Conclusion

A new method of the IVs determination for edible oils has been proposed in the present work. Unlike other procedures shown in earlier articles it applies the intensity ratios of IR and RS bands observed at about 1650 and 2850 cm⁻¹ corresponding to the ν(C=C) and ν(CH₂) vibrations, respectively. The good correlation between this spectral parameter and the IV allowed to construct a relationship between the spectroscopic and chemometric data of the oils. The use of this characteristic pair of bands for this purposes was not proposed earlier. Moreover, a new theoretical approach to analysis of the results has been proposed. It uses the procedure of the computer deconvolution of spectral contours into Lorentz components, evaluation of the integral intensities of the bands, and finding the correlation between the IV and the ratio of respective band intensities.

Funding

The present work was supported partly by Embassy of Libya under the doctor stipend for Abduladhim Moamer M. Albegar and by Polish Ministry of Science and Education under the statutory investigations in the Department of Bioorganic Chemistry, Institute of Chemistry and Food Technology, Wrocław University of Economics.

Additional information

Funding

The present work was supported partly by Embassy of Libya under the doctor stipend for Abduladhim Moamer M. Albegar and by Polish Ministry of Science and Education under the statutory investigations in the Department of Bioorganic Chemistry, Institute of Chemistry and Food Technology, Wrocław University of Economics.