FTIR 1

FTIR 1

International Food Research Journal 17: 519-526 (2010)

FTIR spectroscopy combined with chemometrics for analysis of

lard in the mixtures with body fats of lamb, cow, and chicken

 

Serdang, Selangor, Malaysia

Department of Pharmaceutical Chemistry, Faculty of Pharmacy, Gadjah Mada

University, Yogyakarta 55281, Indonesia

Abstract: Fourier transform infrared (FTIR) spectroscopy combined with chemometrics of partial least square (PLS) and discriminant analysis (DA) has been developed for simple analysis of lard in the mixtures with body fats of lamb (LBF), cow (Cow-BF), and chicken (Ch-BF). The spectral bands correlated with lard, LBF, Cow-BF and Ch-BF as well as their lard blends were scanned, interpreted, and identified. Qualitative differences among FTIR spectra are proposed as a basis for differentiating between the lard and its blends. DA with Mahalanobis distance principle in entire range of mid infrared (3300 – 650 cm-1) was successfully provide an alternate method to differentiate lard and that in the mixtures with LBF, Cow-BF, and Ch-BF. Quantitative analysis using PLS calibration model is proposed to measure the percentages of lard in LBF, Cow-BF, and C-BF at selected fingerprint region (1500 – 900 cm-1). The equation obtained between actual lard concentration in the mixture with LBF and FTIR predicted concentration in calibration model is y = 0.995 x + 0.098 with coefficient of determination (R2) was 0.995 and root mean standard error of calibration (RMSEC) of 0.98. The actual percentages of lard mixed with Cow-BF and Ch-BF were also correlated to FTIR predicted percentages at 1500 – 900 cm-1 using equations of y = 0.999x + 0.016 (R2 = 0.999, RMSEC = 0.61); and y = 1.002x + 0.034 (R2 = 0.998, RMSEC = 0.73), respectively.

Keywords: FTIR, lard, body fats, discriminant analysis, partial least square

Introduction

Lard is defined as “the fat rendered from fresh, clean, sound fatty tissues from swine (Sus scrofa)

in good health, at the time of slaughter, and fit from human consumption” (Codex Allimentarius, 1991). It is appreciated as an important ingredient for cooking. Its production has been estimated more than 5.4 million tons annually, with China, USA, and Germany are the main producing countries (De Leonardis et al., 2007).

In some countries, food producers prefer to mix vegetable oils with lard to reduce the production

cost. Currently, lard is one of the cheapest oils and is commonly available for the food industries. Lard or industrially modified lard could be effectively mixed with other vegetable oils to produce shortenings, margarines and other food oils (Marikkar et al., 2005). The presence of lard in any food systems is serious problems in view of religious concerns, because some religions like Islam, Judaism and Hinduism forbid their followers to consume any foods containing porcine and its derivatives (Regenstein et al., 2003).

Analysis of edible fats and oils are usually performed by determination of specific components present in fats or oils such as derivate of fatty acid methyl esters (FAME) using gas chromatography and triglyceride (TG) compositions with liquid chromatography (Rohman et al., 2010). Gas-liquid chromatography (GLC), combined with pancreatic lipolysis, and chemometrics of multivariate data analysis has been used for identification of lard in some vegetable oils by monitoring the changes of fatty acid composition in those vegetable oils, especially in sn-2 position. Using GLC, 2% of lard in some vegetable oils can be detected (Luddy et al., 1964; Marikkar et al., 2005). Analysis of TG composition showed that both genuine and randomized lard had six dominant TG (C46, C48, C50, C52, C54, and C56) with quite different concentrations. TG with C52 represents the major constituent of genuine and randomized lard (Rashood et al., 1996). Using HPLC, the amounts of TG containing saturated fatty acid at C-2 position of lard were larger than those of other meat fats (Saeed et al., 1989). It is reported that using this technique, a level of 5 % lard can be detected in meat products (Rashood et al., 1996).

Lard has also been analyzed using differential scanning calorimetry (DSC). Tan and Che Man (2000) showed that the DSC heating and cooling curves of edible oils can be used in qualitative and quantitative ways for identification of edible oils. This method can detect 1 % w/w lard and randomized lard in RBD palm oil (Marikkar et al., 2001). Most of methods described above are time consuming and not practical to be performed. Therefore, some efforts have been striven to develop a rapid and reliable technique for lard determination.

FTIR spectroscopy connected to multivariate calibration has been used for analysis of VCO in

binary mixtures with palm kernel oil (Manaf et al., 2007), palm oil (Rohman and Che Man, 2009a), and olive oil (Rohman et al., 2010) and to analyze extra virgin olive oil mixed with palm oil (Rohman and Che Man, 2010). A level of 3-4 % of lard can be detected in food products (Che Man et al., 2005; Syahariza et al., 2005).

Che Man and Mirghani (2001) have developed FTIR spectroscopic method to analyze the presence of lard in body fats of chicken, lamb, and cow. Furthermore, Jaswir et al. (2003) has used FTIR in combination with PLS for analysis of lard in the mixture with body fat of mutton at frequency of 3010- 3000, 1220-1095, and 968-965 cm-1 and at frequency 1419-1414 and 968-965 cm-1 for detecting lard in the mixture with body fat of cow. Both of the group researchers have used the different frequencies rather than using one frequency region for analysis of lard in certain blends with other body fats. Besides, they did not perform discriminant analysis (DA) in order to classify pure lard and that in the mixture with other body fats of lamb (LBF), chicken (Ch-BF), and cow (Cow-BF). Therefore, this study was conducted to investigate the possibility of detection of lard mixed with other animal fats such as LBF, Cow-BF, and

Ch-BF using FTIR spectroscopic techniques at one frequency regions combined with chemometrics of DA and partial least square.

In fats and oils, DA has been used to differentiate among 10 different edible oils and fats, using FTIR, FT-near infrared and FT-Raman spectroscopy (Yang et al., 2005), virgin coconut oil and other vegetable oils (Manaf et al., 2007), olive oil (Fragaki et al., 2005), lard in admixtures with animal fats by HPLC (Marikkar et al., 2005), and to differentiate four vegetable oil types (cottonseed, peanut, soybean and canola) by near infrared reflectance spectroscopy (Bewig et al., 1994).

Materials and Methods

-

Results and Discussion

The TGs are the major component of fats and oils, therefore they dominate spectra. Figure 1 shows the FTIR spectra of lard, LBF, Cow-BF, and Ch-BF. These spectra are very similarly and showed a typical characteristic of absorption bands for common TG (Safar et al., 1994). FTIR spectroscopy can be regarded as a potential analytical technique offering a differentiation between lard and LBF, Cow-BF, and Ch-BF because of its ability as “fingerprint”

technique. Peak in the region 3007 cm-1 is due to C-H stretching vinylic. The strectching vibrations of methylene (-CH2-) and methyl (-CH3) groups are observable in frequencies 2922 and 2852 cm-1, respectively. Methylene and methyl groups are also observable at regions of 1465 cm-1 and 1375 cm-1 due to their bending vibrations (Pavia et al., 2001).

The carbonyl (C=O) absorption of ester linkage is observed at frequency 1740 cm-1 with strong intensity due to the great difference of electro-negativity of carbon and hydrogen atoms. The bands at 1235, 1160, 1117, 1098 and at 721 cm-1 are results from the overlapping of the methylene rocking vibrations and the out of plane bending vibration of cis-disubstituted

olefins (Guillen and Cabo, 1997). From the point of view of spectra observations with naked eye, it is very difficult to differentiate lard and LBF, Cow- BF, and Ch-BF. However, a precise investigation in fingerprint region, especially in wavenumbers of 1500 – 1000 cm-1, revealed that there are visual differences at absorption peaks at 1162 cm-1 (a) and two adjacent peaks at 1117 (b) and 1097 cm-1 (c). An absorbance ratio at 1117 and 1097 cm-1 (A1117/A1097) was used by Alam and Hamid (2007) to differentiate among vegetable oils. The value of A1117/A1097 from

lard and LBF, Cow-BF, and Ch-BF is shown in Table 1. These values for pure animal fats are almost constant. If fats are mixed each other, these values will be changed according to type and amount of added fats. In this study, the chemometric techniques namely discriminant analysis (DA) was used for lard differentiation. Furthermore, the level of lard mixed with each other can be quantitatively determined using partial least square (PLS) technique.

Discriminant analysis

As one of the classification analysis, DA is designed to find the mathematical models capable

Table 1. Ratio value of A1117/A1097 of lard and body fats

of lamb (LBF), cow (Cow-BF), and Chicken (Ch-BF)

Animal fats        A1117/A1097

Lard             0.977

LBF             0.937

Cow-BF          0.925

Ch-BF           0.931

of recognizing the membership of each object to its proper class on the basis of a set of FTIR spectra.

Once a classification model has been obtained, the membership of unknown objects to one of the

defined classes can be predicted (Rohman and Che Man, 2009a).

Lard and lard in mixture with LBF, Cow-BF, and Ch-BF were classified into two groups, known

as pure lard and lard in mixture of body fats. DA was applied to both classes in the wavenumber of

3,300–800 cm-1. Figure 2 showed the Coomans plot for the classification of lard in mixture with LBF

(A), Cow-BF (B), and Ch-BF (c). The x-axis shows Mahalanobis distance to lard, while the y-axis shows

the distance to lard in mixture with body fats. The Mahalanobis distance is useful in assigning whether a

set of unknown value samples is similar to a collection set of known measured samples. The Coomans plot

clearly exhibits the separated group of lard and lard in mixture. In this study, the DA model revealed one

misclassified from all samples because of the close similarities in the chemical composition between lard

and one of samples misclassified.

Partial least square (PLS) model

PLS is relied on its ability to exploit FTIR spectral data from broad spectral frequencies and to correlate

spectral changes in the concentration of a component of interest while simultaneously accounting for other

spectral contributions that may perturb FTIR spectra (Syahariza et al., 2005). PLS calibration model was

developed based on the calibration standard that included the different weighted amounts of lard

blended with LBF, Cow-BF, and Ch-BF.

FTIR spectroscopy is regarded as “fingerprint” technique, especially in region of 1500 - 650 cm-1.

PLS is also called full spectrum method, therefore it can be applied for analysis of component of interest at

the whole FTIR spectral regions rather than specific regions (Faber and Rajko, 2007). For these reasons,

two FTIR spectral regions, namely using whole spectra (3300-700 cm-1) and at selected frequency

regions (1,500 – 900 cm-1) were used for developing a PLS calibration model. The selection of frequency

regions was based on the highest values of coefficient of determination (R2) and the lowest values of root mean standard error of calibration (RMSEC). Table 2 revealed the performance of PLS calibration for analysis of lard in the mixtures with LBF, Cow-BF, and Ch-BF in term of R2 and RMSEC values.

Based on the highest values of R2 and the lowest values of RMSEC as shown in Table 2,

frequency region of 1500- 900 cm-1 was selected for quantification of lard in the mixtures with LBF, Cow-BF, and Ch-BF. The relationship between actual lard concentration against the PLS FTIR predicted in mixture with LBF, Cow-BF, and Ch-BF is shown in Figure 3 A, B, C, respectively. All equations revealed a good relationship between actual lard value and

FTIR predicted values with R2 greater than 0.99.

The calibration model was further cross validated by removing one standard at a time. The R2 value of 0.993 (y = 0.982x + 0.282) for lard in mixture with LBF, R2 of 0.991 (y = 0.994x + 0.538) for lard in mixture with Cow-BF and R2 of 0.983 (y = 0.933x + 1.855) for lard in mixture with Ch-BF were obtained, respectively (Figure 4). Based on the calibration and vaidation curves, it can be stated that lard with level 1 % v/v was possible to be detected using FTIR spectroscopy in combination with multivariate calibration of PLS.

Table 3 shows the statistical results calculated from cross validation as mean difference (MD) and

standard deviation of difference (SDD) for accuracy and reproducibility methods in the determination of lard in the mixtures with LBF, Cow-BF, and Ch-BF.

Accuracy is a measure of the closeness of agreement between actual data and the predicted FTIR results. The low values of MDa (0.22, 0.24, and 0.32), and SDDa (1.56, 1.74, 1.97) for lard in the mixture with LBF, Cow-BF, and Ch-BF respectively, show that the FTIR is well suited for determining lard in the mixtures with body fats. Meanwhile, low MDr values (0.28, 0.27, 0.42) and SDDr (1.87, 1.92, 2.08), respectively, indicate that the FTIR method has appreciably high repeatability.

Conclusions

Fourier Transform Infrared (FTIR) Spectroscopy combined using attenuated total reflectance (ATR) sampling handling technique and chemometrics of DA and PLS can be used to analyze the lard contents in the mixtures with LBF, Cow-BF, and Ch-BF. The developed method was rapid, with a total analysis time less than 3 min for one measurement. Furthermore, it is environmentally friendly and the use of excessive time, chemical reagents and solvents can be avoided.