Food Sci. Technol. Res., +- (.), ,3+ῌ,3/, ,**1
Technical paper Visible and Near Infrared Spectroscopy for Rapid Analysis of the Sugar Composition of Raw Ume Juice Jie Yu CHENῌ, Han ZHANG and Ryuji MATSUNAGA Faculty of Bioresource Sciences, Akita Prefectural University, Akita *+*ῌ*+3/, Japan Received June 0, ,**0 ; Accepted May +2, ,**1 Visible and near infrared spectroscopy was investigated as a method for rapid analysis of the sugar compositions of raw ume juice. In total, -+. raw ume juice samples were collected over a long growth period and visible and near infrared transmittance spectra between .** and +2/* nm were acquired using a spectrophotometer with a quartz cuvette with +-mm thickness. The partial least squares (PLS) calibration models for the total sugar content of ume fruit juice were developed using original spectra and pretreated spectra (normalization, first derivative, second derivative and multiplicative scatter correction) in five di#erent wavelength ranges. The best models used the second derivative spectral data in the wavelength ranges of ++**ῌ+2/* nm with the lowest standard error of prediction (*.+--ῌ). Moreover, good calibration models for sugar compositions (fructose, glucose, sorbitol, sucrose) of ume juice were obtained using the same second derivative spectral data in the wavelength range of ++**ῌ+2/* nm, and gave relatively good predictions with high value of R, and low value of SEP : fructose, *.30 and *.*0,ῌ ; glucose, *.3. and *.*02ῌ, sucrose, *.3, and *.*0/ῌ, sorbitol, *.21 and *.*.0, respectively. Results indicated that near infrared spectroscopy provided useful method to rapidly analyze the total sugar content and the sugar composition of raw Ume fruit juice. Keywords : Visible and near infrared spectroscopy, Ume juice, Partial least squares regression, Sugar composition, Fructose, Glucose, Sorbitol, Sucrose
Introduction Ume (Japanese apricot, also known as the Japanese plum) is an important fruit in Japan. Although ume is unsuitable for consumption as a fresh food, consumption of its processed goods, such as ume juice, ume sake (Japanese alcoholic beverage) and various pickled ume products has increased over recent years, due to its sweet and sour taste, which is particularly appreciated in Japan. The taste of the processed goods is influenced by variations in the sugar composition of ume, which varies according to the harvesting season, cultivars and production area. Therefore, determination of the sugar composition of ume is very important for the production process in order to provide good quality products. High performance liquid chromatography (HPLC) is commonly used for analysis of sugar components in biomaterials (Bouzas et al., +33+ ; Yuan et al., +332 ; Gomis et al., ,**.). The HPLC method is very accurate, but the method is rather time-consuming, requires heavy sample preparation and cost-intensive. Therefore, there is a demand for rapid analytical techniques for determining the sugar compositions of raw biomaterials in the production field. Recently, near infrared (NIR) spectroscopy has become ῌ
To whom correspondence should be addressed. E-mail :
[email protected]
a well-accepted method for the analysis of food constituents, since it is fast, requires little or no sample preparation, and does not cause any environmental chemistry pollution. The NIR technique has been used for analyzing the total sugar content of Satsuma mandarin juice (Fujiwara and Honjo, +33/), strawberry juice (Fujiwara and Honjo, +330), mango juice (Duarte et al., ,**,), tomato juice (Goula et al., ,**-) and in beet juice (Roggo et al., ,**.). Moreover, there has been some application of the NIR technique to the analysis of sugar composition of Japanese pear juice (Tanaka and Kijima, +330) and apple juice (Sinnaeve et al., +331). However, the NIR technique has never been applied to ume juice. In the present study, we tried to apply the NIR technique to analyze the main sugar compositions of raw ume juice. Specifically, the aim of our study was to investigate the potential of the NIR technique as a means for analyzing the sugar composition of raw ume juice.
Materials and Methods Samples Preparation The fruit of three cultivars, namely Bungo-Ume, Koshino-Ume and Togoro-Ume, were cultivated in Kotonooka-machi, Akita Prefecture, Japan, in ,**.. In order to obtain a wide range of sugar compositions, the fresh ume were sampled twice a week for approximately two months during the growth period, and
292
J.Y. CHEN et al.
a total of -+. fruits were collected. After harvesting, the flesh of the fruit samples was removed and broken small by a mixer, and then it was wrapped in gauze and the juice was squeezed out for use as the study samples. Each raw juice sample was individually separated to two parts. The one part was used for the reference chemical analysis, while the other part was used for the spectral measurement. Reference Analyses The sugar composition (fructose, glucose, sorbitol, sucrose) of the ume juice samples were analyzed by the HPLC method. Separations were performed in a column (KS-2*+, Shodex Ionpak) with an eluate (acetonitrile solution of 1/ῌ). The operating condition was as follows : flow rate of + mL/min ; column temperate of .*῍C ; injection sample volume of +* mL ; refractive index detector (Shimadzu, RID-+*A). The raw Ume juice samples were centrifuged at +*** RPM for +* min, diluted + : - by adding distilled water then passed through a *../-mm porosity filter. NIR Spectra The transmittance (T) spectra of ume juice samples were measured using a spectrophotometer (NIRSystems 0/** ; Foss NIRSystems, Inc., Silver Spring, MD) and a quartz cuvette with +-mm thickness. The spectra were measured in the wavelength range from .** to +2/* nm at ,-nm intervals and the absorbances were recorded on a linked computer as log (+/T). The NIR spectrum of each sample was obtained by taking the average of -, scans. All operations were performed at room temperature (,*῍C). Statistical Analysis The -+. raw ume juice samples were separated into two groups : ,++ samples for calibration and +*- samples for validation. The significant sugar (fructose, glucose, sorbitol, sucrose) components of the ume juice samples selected for the calibration and validation sample sets are shown in Table +. The calibration and validation sample sets showed similar means and standard deviations for all sugar components. Ume juice samples in the calibration set were used to establish the calibration models (multivariate equations) between the spectral data and laboratory reference values, while samples in the validation set were used to evaluate calibrations. The partial least square (PLS) regression was used to develop calibrations for analyzing the sugar compositions
of the raw ume juice samples. The PLS calibrations were performed with the Unscrambler software (Version 1.0 ; CAMO AS, Trondheim, Norway), and used both the spectral response and the respective reference data to determine the latent variables (PLS factors) in the calibrations data set. The validation set (which did not include any samples from the calibration set) was used to check the performed calibration and to determine the optimum number of factors by the standard error of validation (SEP) based on the validation sample set. The standard error of calibration (SEC) and the standard error of validation (SEP) and the coe$cient of determination between the reference values and the NIR values (R,) were calculated. Spectral pretreatment methods such as normalization, first derivative, second derivative and multiplicative scatter correction (MSC) were used to try to remove the variation in spectra caused by unknown sources that tend to increase errors in the calibration models. These pretreatment approaches were performed with the Unscrambler software too.
Results and Discussion Development of PLS Calibrations for the Total Sugar Content A total of ,/ PLS calibration models were developed for the analysis of total sugar contents in raw Ume juices using the calibration sample sets. These involved five wavelength ranges and five forms of spectral date (original, treated by normalization, first derivation, second derivation and multiplicative scatter correction). The summary results of this work are presented in Table ,. The most accurate model involved second derivative spectral data in the wavelength range ++**ῌ+2/* nm, used nine PLS factors, and produced low SEC and SEP values
Table ,. Results of calibration and validation of total sugar content for various preprocessing methods and spectra wavelength ranges.
Table +. Mean, range and standard deviation (SD) of the sugar (fructose, glucose, sorbitol, sucrose and total sugar) components of the ume juice samples in the calibration and validation sample sets.
῍
nῌ0- for calibration, nῌ-+ for validation.
R,, determination coe$cient of calibration ; SEC, standard error of calibration ; SEP, standard error of validation.
Visible and Near Infrared Spectroscopy for Rapid Analysis of the Sugar Composition of Raw Ume Juice
Fig. +. Plot of total sugar content versus NIR predicted value for second-derivative spectra (++**ῌ+2/* nm).
Table -. Calibration and validation results for each sugar component in raw ume juice for second-derivative spectra (++**ῌ+2/* nm).
῍
nῌ0- for calibration, nῌ-+ for validation ; R,, determination coe$cient of calibration ; SEC, standard error of calibration ; SEP, standard error of validation.
equal to *.+.ῌ and *.+-ῌ, respectively. The calibration and validation results were represented graphically by plotting the reference total sugar contents versus the predicted values as shown in Figure +. The plot showed strong linearity and relationship between actual the total sugar contents and the NIR predicted values. Development of PLS Calibrations for the Sugar Compositions Based on above discussions, the second derivative spectral data in the wavelength range ++**ῌ+2/* nm were also used to develop the calibration models for the analysis of sugar composition of raw ume juice. PLS regressions were performed based on sugar composition (fructose, glucose, sorbitol and sucrose) and the spectral data of the calibration sample sets. The summary results as statistical parameters of the calibration models are presented in Table -. In the cases of fructose and glucose, good PLS calibration models with nine factors were obtained. The determination correlation coe$cients (R,) were *.30 for fructose, *.3. for glucose, respectively. The SEC and SEP were *.*1ῌ and *.*0ῌ for fructose, *.*1ῌ and *.*1ῌ for glucose, respectively. In the case of sucrose, good PLS calibration model used eight factors was also obtained with R, of *.3,, SEC of *.*/ῌ and SEP of *.*1ῌ. In the case of sorbitol, good PLS calibration model used +- fac-
293
tors was also obtained with R, of *.21, SEC of *.*.ῌ and SEP of *.*/ῌ. These calibration results were represented graphically by plotting the reference sugar components versus the predicted values as shown in Figure ,. In the case of all, strong linearity and high correlation were shown. As an index for judging the usefulness of the measurement accuracy of the calibration models, the ratio of the standard deviation of the reference data in the prediction sample set to the SEP (RPD) is usually utilized (Williams ,**+ ; Chen et al., ,**/). An RPD value of ,./ῌ-.* is regarded as adequate for rough screening, while a value of above -.* is regarded as satisfactory for screening. In our case, RPD values more than ,./ were obtained, and the calibration models can therefore be considered to show suitable accuracy for measuring the sugar components such as fructose, glucose and sorbitol as well as sucrose contents of raw ume juice. Discussion of NIR Calibration Models Regression coe$cients can be used to discuss the contributions of individual wavelengths to a PLS calibration model, since a regression coe$cient spectrum shows characteristic peaks and troughs that can indicate which wavelength range is important for the calibration model (Martens et al., +323 ; Chen et al., ,**- and ,**.). Figure - shows the regression coe$cients of the PLS calibration models of fructose component (A), glucose component (B), sucrose component (C) and sorbitol component (D), respectively. These regression coe$cient spectra showed many remarkable peaks. The peaks at wavelengths of +-/0, +-3., +-30 and +.+2 nm could be correlated to the combined absorption band of two C-H stretching and a C-H deformation mode corresponding to the wavelengths of +-0*, +-3/ and +.+1 nm (Osborne, +33-), respectively. The peaks at wavelengths of +.,,, +.-2, +..*, +./*, +.12, +.20, +/-0, +/2., +2+0, +2,* and +2,, nm could be correlated to the absorption band of an O-H stretching first overtone corresponding to the wavelengths of +.,*, +..*, +./*, +.2*, +.3*, +/.*, +/2* and +2,* nm, respectively. The peaks at wavelengths of +0.0, +00,, +022, +1*., +1*0, +1,0, +102, +112 and +12* nm could be correlated to the absorption band of a C-H stretching first overtone corresponding to the wavelengths of +0./, +00*, +02/, +1*/, +1,/, +10/ and +12* nm, respectively. These results suggest that the PLS calibration models for the sugar components of raw ume juice were established based on the absorptions of sugar components.
Conclusions It was concluded that near infrared spectroscopy was an acceptable technique for rapidly analyzing the total sugar content and the sugar composition such as fructose, glucose, sorbitol and sucrose components of raw ume juice.
294
J.Y. CHEN et al.
Fig. ,. Plot of each sugar component versus NIR predicted value for second derivative spectra (++**ῌ+2/* nm).
Fig. -. Regression coe$cients in the PLS calibration models for second derivative spectra (++**ῌ+2/* nm) for each sugar component in raw ume juice.
Visible and Near Infrared Spectroscopy for Rapid Analysis of the Sugar Composition of Raw Ume Juice
References Bouzas, J., Kantt, C.A., Bodyfelt, F. and Torres, A. (+33+). Simultaneous determination of sugar and organic acids in cheddar by high performance liquid chromatography. Journal of Food Science. /0, ,10ῌ,11. Chen, J.Y. ; Matsunaga, R., Ishikawa, K. and Zhang, H. (,**-). Main inorganic component measurement of seawater using nearinfrared spectroscopy. Applied Spectroscopy. /1, +-33ῌ+.*0. Chen, J.Y. ; Miao, Y. ; Zhang, H. and Matsunaga, R. (,**.). Nondestructive determination of carbohydrate content in potatoes using near infrared spectroscopy. J. Near Infrared Spectroscopy. +,, -++ῌ-+.. Chen, J.Y. ; Zhang, H. ; Miao, Y. and Matsunaga, R. (,**/). NIR measurement of specific gravity of potato. Food Sci. Technol. Res. ++, ,0ῌ-+. Duarte, I.F., Barros, A., Delgadillo, I., Almeida, C. and Gil, A.M. (,**,). Application of FTIR spectroscopy for the quantification of sugars in mango juice as a function of ripening. Journal of Agric. Food Chem. /*, -+*.ῌ-+++. Fujiwara, T. and Honjo, T. (+33/). Determination of sugar and acid contents in fruit juice of Satsuma mandarin by near infrared spectroscopy. Nippon Shokuhin Kagaku Kogaku Kaishi. .,, +*3ῌ++1. Fujiwara, T. and Honjo, T. (+330). Determination of sugar and acid contents in fruit juice of Strawberry by near infrared spectroscopy. Nippon Shokuhin Kagaku Kogaku Kaishi. .-, 2.+ῌ2.2. Goula, A.M. and Adamopoulos, K.G. (,**-). Estimating the composition of tomato juice products by near infrared spectroscopy. Journal of Near Infrared Spectroscopy. ++, +,-ῌ+-0.
295
Gomis, D.B., Tamayo, D.M., Valles, B.S., and Mangas, J. J. (,**.). Detection of apple juice concentrate in the manufacture of natural and sparkling cider by means of HPLC chenmometric sugar analyses. J. Agric. Food Chem. /,, ,*+ῌ,*-. Martens, H. ; Naes, T. (+323). Multivariate calibration. John Wiley & Sons, UK, pp. ++0ῌ+0,. Osborne B.G., T. Fearn, P.H. Hindle. +33-. Practical NIR spectroscopy : with applications in food and beverage analysis. John Wiley & Sons, Inc., New York. Roggo, Y., Duponchel, L. and Huvenne, J.P. (,**.). Quality evaluation of sugar beet by near infrared spectroscopy. Journal of Agric. Food Chem. /,, +*//ῌ+*0+. Sinnaeve, G., Dardenne, P., Agneessens, R., Lateur, M. and Hallet, A. (+331). Quantitative analysis of raw apple juices using near infrared, fourier-transform near infrared and fourier-transform infrared instruments : a comparison of their analytical performances. Journal of Near Infrared Spectroscopy. /, +ῌ+1. Tanaka, M. and Kojima, T. (+330). Near-infrared monitoring of the growth period of Japanese pear fruit based on constituent sugar concentrations. Journal of Agric. Food Chem. .., ,,1,ῌ ,,11. Williams, P. (,**+). Implementation of near-infrared technology. In : Near-infrared technology in the agricultural and food industries, AACC, USA, pp. +./ῌ+03. Yuan, J.P. and Chen, F. (+322). Separation and identification of furanic compounds in fruit juices and drinks by high-performance liquid chromatography photodiode array detection. J. Agric. Food Chem. .0, +,20ῌ+,3+.
