Mid-infrared spectroscopy and chemometrics in corn starch classification

Mid-infrared spectroscopy and chemometrics in corn starch classification

Journal of MOLECULAR STRUCTURE Journal of Molecular Structure 410-411 (1997) 551-554 Mid-infrared spectroscopy and chemometrics in corn starch cla...

244KB Sizes 0 Downloads 102 Views

Journal of

MOLECULAR STRUCTURE Journal of Molecular

Structure 410-411

(1997) 551-554

Mid-infrared spectroscopy and chemometrics in corn starch classification N. Dupuya3*, C. Wojciechowski

b, C.D. Tab, J.P. Huvennea, P. Legrand”

‘Laboratoire de Spectroscopic Infrarouge et Raman, LASIR, CNRS, Bcit CS, Universitt?des Sciences et Technologies de Lille, 59655 Villeneuve d’Ascq Cedes France bCentre de Recherche et d’Etudes Alimentaires (CREALIS). Danone, Zone Industrielle du Teinchun’er, 19100 Brive, France Received 26 August 1996; revised 10 October

1996; accepted

11October 1996

Abstract The authentication of food is a very important issue for both the consumer and the food industry at all levels of the food chain from raw materials to finished products. Corn starch can be used in a wide variety of food preparations such as bakery cream fillings, sauces, salad dressings, frozen foods etc. Many modifications are made to corn starch in connection with its use in agrofood. The value of the product increases with the degree of modification. Some chemical and physical tests have been devised to solve the problem of identifying these modifications but all the methods are time consuming and require skilled operators. We separate corn starches into groups related to their modification on the basis of the infrared spectra. 0 1997 Elsevier Science B.V. Keywords:

FTIR; PCA; Chemometrics;

Starch

1. Introduction Corn

starches

have

always

been

employed

0.150

in the

of foods. Widely available, easily used, they provide food textures appreciated by consumers. Modified starches [l] have an improved textural stability and often will not set to a gel like conventional starches. These products exhibit superior performance in finished preparations. Modified starches answer technical needs arising from the evolution of manufacturing processes or products. New European Community legislation concerning food is strict and the food must adhere to established standards of

preparation

0.125

.$ 0.100 x: 0.075 b 4’ 0.050 0.025 0.000 1500

1400

1300

1200 Wovenumber

* Corresponding author. 0022-2860/97/$17.00 PII SOO22-2860(96)095

Fig.

0 1997 Elsevier 17-8

Science

B.V. All rights reserved

1100 (cm-’

1000

900

800

70(

1

1. Infrared spectra of four modified starches (R, H, N, P).

N. Dupuy et al./Journal of Molecular Structure 410-411

552

quality. It is important to be able to determine food composition and establish analytical controls. As modified starches show differences in their chemica1 structures, it seems possible to use FTIR spectra to classify samples in relation to their modification. The horizontal attenuated total reflectance (HATR) technique is chosen as a sampling method, with the

Scoru

8.8818.

(1997) 551-554

use of a volatile solvent to increase the reproducibility and intensity of the spectra. Principal component analysis (PCA) of infrared spectra is used as a classification method [2] in order to identify whether corn starch is modified or not and which method of modification is used.

n

6.8815.

e.eees* e.8886. e.eees*Y .. A

8. e.8883.

n

Nnrq

A P

R %

6:eee6

I -0.w

(4

P

-

3-e.

““”

.‘(’

-e.‘sei

d

“”

” e.ie1

-I 0.OCt

...

/’ 0.003

.

e:ie



Pck~pi):

e.eele.

n

e.8815. e.eelze.eees. 8.8886. 8.8883. 8. R

0.8883. *

b.8886.

R

R

3.8889. 1 -0.00ls

(W

IC,_.__I

-e.ieis .

. . . ,-..

-e.ieiz

1 -0.0049

ed.c._..

Fig. 2. (a) Scattergram of corn starches in the two directions

. -e.ies6

-e.de83

I 0

e.e’ec

PC1 and PC3. (b) Zoom of the scattergram

(PC1 -PC3).

N. Dupuy et aL/Journal

2. Materials

and methods

This work was performed classified into four groups: 1. 2. 3. 4.

of Molecular

Structure 410-411

8.25

on 32 corn starches,

non-modified, denoted N; cold water swelling, denoted P; hydroxypropylated and phosphated, denoted R; acetylated and phosphated, denoted H.

A detailed sample preparation is presented in [3]. Briefly, the powder was impregnated with 0.5 ml of dry acetone and the spectrum was recorded after total evaporation of the solvent. The addition of a non-dissolving liquid improved the refractive component of the sample by removing air present in the powder. After solvent evaporation the powder is more compact. Our FTIR measurements were performed on an IFS 48 BRUKER spectrometer. We added together 200 scans of symmetrical interferograms. For each sample we recorded three spectra and computed single-beam spectra at 4 cm-’ resolution with a triangular apodization function. PCA is a method of extracting the systematic variation in one data set. The method can be used for classification as well as description and interpretation. PCA is oriented toward modelling the variance/ covariance structure of the data matrix into a model which represents the significant variations and considers noise as an error [4].

3. Results and discussion Fingerprint regions of ATR spectra are shown in Fig. 1. The spectra seem to be similar because the different samples contain the same major constituents and a visual a priori classification is not possible. Principal component analysis of the data set constituted by 32 derived spectra explains 88% of the data matrix variance after extraction of three components. The remaining components only contributed less than 3% of the residual variance. We may consider that components extracted after the third one model the non-significant variations as noise or sampling variations. The scattergram of starch samples in the two directions (Fig. 2a) PC1 and PC3 describes a separation

‘-asdtng

L

553

(1997) 551-554

rctghts

I

E.ZB 8.15 8.18 0.85

e 4.05.

-a.ia.9.15-

-a.talCO4.

9ed.n

8li.8

715.

Fig. 3. First loading spectrum.

into two groups which can be correlated with the cold water swelling treatment. Cold water swelling samples are positively projected on the first principal component since the others are negatively projected. We can deduce from Fig. 3, which represents the first loading spectrum, that the differentiation used to classify the samples is the intensity of the vibrational band at 1014 cm-‘. A zoom on the negative part (Fig. 2b) shows a fine separation of the other groups. The non-modified ones (N) present small scores near 0. The H-modified spectra present middle scores. The R-modified spectra present high scores at the left of the scattergram. These results show the ability of FTIR spectra to distinguish between different types of corn starch, but a fine classification remains difficult since the spectra of corn starches shown in Fig. 1 are very similar. It has to be completed by new reference samples and by the application of other classification software.

4. Conclusion The use of PCA on FTIR-ATR spectra allows one to partially classify corn starches into subgroups. This method offers a new way of classifying samples very rapidly, but in the future the database must include many samples of all the types of modification

554

N. Dupuy et al./Journal of Molecular Srructure 410-411

encountered in agrofood, quality control.

so that it may be used for

References [I] R.A. De Graaf, G.A. Broekrelofs, L.P.B. Janssen and A.A Beenackers, Carbohyd. Polym., 28 (1995) 137.

(1997) 551-554

[2] N. Dupuy, J.P. Huvenne, L. Duponchel and P. Legrand, Appl. Spectr. 49 (1995) 580. [3] N. Dupuy, M. Meurens, B. Sombret, P. Legrand and J.P. Huvenne, Appl. Spectr. 47 (1993) 452. [4] O.M. Kvalheim, Leb. Int. Lab. Sys., 2 (1987) 127.