FT-NIR spectroscopy for caffeine estimation in instant green tea powder and granules

LWT - Food Science and Technology 42 (2009) 998–1002

Contents lists available at ScienceDirect

LWT - Food Science and Technology journal homepage: www.elsevier.com/locate/lwt

FT-NIR spectroscopy for caffeine estimation in instant green tea powder and granules V.R. Sinija*, H.N. Mishra Agricultural & Food Engineering Department, Indian Institute of Technology, Kharagpur, West Bengal 721 302, India

a r t i c l e i n f o

a b s t r a c t

Article history: Received 5 July 2008 Received in revised form 5 December 2008 Accepted 23 December 2008

The feasibility of measuring caffeine content in instant green tea and granules was investigated by Fourier Transform Near-Infrared (FT-NIR) spectroscopic technique. A calibration model was developed using pure caffeine standards of varying concentrations in the near-infrared region (4000–12000 cm1). The developed model was validated using test validation technique. FT-NIR spectroscopy with chemometrics, using the PLS–first derivative plus straight line subtraction method could predict the caffeine content in tea samples accurately up to an R2 value greater than 0.98 and a standard error of prediction (SEP) value less than 2.0 with 6 factors in the prediction model. The developed model was applied to predict caffeine content in tea samples within 2–5 min. The developed procedure was further validated by recovery studies by comparing with UV spectroscopic method of caffeine determination. Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Green tea Fourier transform near-infrared (FT-NIR) spectroscopy PLS regression model Caffeine content Instant tea

1. Introduction Tea is among the most popular beverage worldwide, which is of great interest due to its beneficial medicinal properties (Fujiki et al., 2001; Jian, Xie, Lee, & Binns, 2004; Nakachi, Matsuyama, Miyake, Suganuma, & Imai, 2000; Yang, Maliakal, & Meng, 2002). With the increasing consumption of the tea, quality control of tea becomes more and more important nowadays, for example, many national and international authorities are setting criteria for quality factors (Airy, 1999). In general, caffeine and total polyphenols are analyzed as the important quality factors for tea leaves. These constituents are mainly responsible for the characteristic astringent and bitter taste of tea brews (Zhang, Kuhr, & Engelhardt, 1992). Additionally, most commercially tea leaves have many varieties in the market, and these tea varieties differ not only from a botanical standpoint but also in terms of quality. These differences are recognized commercially and appreciated by the consumers also. In the past few years, many different methods of analysis had been employed to identify the tea varieties and determine some chemical compositions in tea. Some approaches were applied to identify tea varieties using modern techniques like high-performance liquid chromatography (HPLC) (Horie & Kohata, 2000; Zuo, Chen, & Deng, 2002), gas chromatography (GC) (Togari, Kobayashi,

* Corresponding author. Tel.: þ91 3222 81323. E-mail addresses: [email protected], [email protected] (V.R. Sinija). 0023-6438/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.lwt.2008.12.013

& Aishima, 1995), plasma atomic emission spectrometry etc. (Herrador & Gonzalez, 2001). Also some approaches were applied to analyze quantitatively some chemical compositions from tea leaves by HPLC (Zuo et al., 2002), capillary electrophoresis (Horie, Mukai, & Kohata, 1997), colorimetric measurements and the titration with potassium permanganate (ISO, 1994). However, all of the methods mentioned above are time-consuming. Near-infrared (NIR) spectroscopy has proved to be a powerful analytical tool used in the agricultural, nutritional, petrochemical, textile and pharmaceutical industries (Esteban-Diez, 2004; Fer˜ ez, Soldado, Martı´nez-Ferna´ndez, & de la Roza-Delna´ndez-Iban gado, 2009; McGlone, Jordan, Seelye, & Martinsen, 2002; Woodcock, Downey, & O’Donnell, 2008). Since 1990s, attempts have been made to simultaneously predict water, alkaloids and phenolic substance content in tea leaves using NIR spectroscopy (Hall, Robertson, & Scotter, 1988; Schulz, Engelhardt, & Wengent, 1999). Studies on the application of NIR spectroscopy to quantitative analysis of total antioxidant capacity in green tea were also reported (Lupaert, Zhang, & Massart, 2003; Zhang, Lupaert, Xu, & Massart, 2004). Although they gave some better results for tea using NIR spectroscopy, they had no details in discussing the prediction models even without using independent test samples to test the robustness of the models, such as Schulz et al. (1999). Caffeine may be the most popular drug in the world. Chemically, caffeine is a 3,7- dihydro-1,3,7-trimethyl-1H-purine-2,6-dione or 1,3,7-trimethylxanthine. Global consumption of caffeine has been estimated to be 120,000 tonnes per annum, approximately equivalent

V.R. Sinija, H.N. Mishra / LWT - Food Science and Technology 42 (2009) 998–1002

to one caffeine containing beverage per day by each of the planet’s 5 billion plus human inhabitants (Hopes, 1997). Hence, caffeine is almost certainly the most widely consumed psychoactive substance in the world. We consume caffeine daily in coffee, tea, cocoa, chocolate, some soft drinks and some drugs. Caffeine is also a central nervous system stimulant. In moderate doses, it can increase alertness, reduce fine motor coordination, cause insomnia, headaches, nervousness and dizziness. In massive doses, caffeine is lethal. A fatal dose of caffeine has been calculated to be more than 10 g (about 170 mg/kg body weight). Some studies have shown that caffeine causes physical dependence. Typical withdrawal symptoms associated with caffeine are headache, fatigue and muscle pain. These symptoms can occur within 24 h after the last dose of caffeine. The above reported facts have made it essential for manufacturers to monitor and assess the concentration of caffeine in their respective products (http://www.staff.washington.edu/chudler/caff.html). The objective of present study was to develop a rapid method for detection of caffeine content in instant green tea powder and granules using FT-NIR spectroscopy. In commercial production, we can use this method to analyze the amount of caffeine in the final product and also to check the same during packaging and storage with a very simple sample preparation method. The monitoring and determination of caffeine concentration in the products is essential for the manufacturers because of the fatal effects of over concentration of caffeine. The method developed in this study can be also used to determine the amount of caffeine in other products like coffee, soft drinks etc. 2. Materials and methods 2.1. Sample preparation The fresh tea leaves for preparation of tea samples were plucked from the tea gardens of Indian Institute of Technology, Kharagpur, India. Fresh tea leaves were steamed (1 kg/cm2 pressure for 1– 2 min) immediately after plucking, to arrest the fermentation process (oxidation reaction) and then subjected to crushing in a laboratory mixer grinder. From the paste thus obtained, a part of the juice was expressed out by means of hydraulic pressing in an expression unit developed at Agricultural & Food Engineering Department of IIT Kharagpur, India. The juice with a total solid content of 8–9% was used for production of instant tea powder by freeze drying and the pressed leaf residue with a moisture content of 63–65% (wb) was subjected to hot air drying (temperature 55  C, thickness 3 mm, air velocity 2.5 m/s) in a recirculatory convective air drier to produce green tea granules. FT-NIR MPAÔ spectrometer (Bruker optics, Germany) combined with Opus 5.5 software was used for analysis of tea samples by generating unique spectrum for each sample. This spectrometer with an integrated Michelson interferometer utilized the Fourier Transform and had distinct advantages compared to dispersive spectrometers. The spectra generated over a range of wave numbers from 12,000 to 4000 cm1 were interpreted based on the overtones of different functional groups in the product. The spectra were measured by keeping 5–8 ml of sample in small sample bottle. For each sample, three spectra were recorded at three different points by rotating the sample bottle by 120 . 2.2. Chemometrics: multivariate analysis Multivariate analysis was used for quantitative and qualitative analysis. Partial Least Square algorithm, which was proven to be effective in many quantitative applications, was used in the present study. The OPUS 5.5 software was used for PLS analysis. These methods with original and vector normalised spectra were used to

999

develop calibration models. The performance of the final PLS model was evaluated in terms of root mean square error of cross validation (RMSECV) for cross validation and root mean square error of prediction (RMSEP) during test validation, and the coefficient of determination (R2). For RMSECV, a leave-one-sample-out cross validation is performed: the spectrum of one sample of the training set is deleted from this set and a PLS model is built with the remaining spectra of the training set. The left-out sample is predicted with this model and the procedure is repeated with leaving out each of the samples of the training set. In test validation, a set of samples are identified as test data and with the remaining data set calibration model will develop first and then the test spectra will be used for the validation of developed model. The number of PLS vectors used is defined in the OPUS software by the size of the ‘‘rank’’. The optimum PLS rank can be calculated only if the number of calibration spectra is sufficiently high (e.g. one component and 20 calibration spectra). The PLS regression has the advantage that the PLS factors are arranged in correct sequence, according to their relevance to predict the component values. The first factor explains the most drastic changes of the spectrum. The residual (Res) is the difference between the true and fitted value. Thus the sum of squared errors (SEE) is the quadratic summation of these values (Eq. (1)).

SSE ¼

X ½Resi 2 :

(1)

The root mean square error of estimation (RMSEE) is calculated from this sum, with ‘‘n’’ being the number of samples and ‘‘r’’ the rank (Eq. (2)).

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1  SSE: RMSEE ¼ nr1

(2)

The determination coefficient, R2 (Eq. (3)) gives the percentage of variance present in the true component values, which is reproduced in the regression. R2 approaches 100% as the fitted concentration values approach the true values.

R2 ¼

SSE

1P

!

ðyi  ym Þ2

 100

(3)

where ym is the mean of the reference results for all samples. R2 can be negative (in some cases) for low ranks, when the residual are larger than the variance in the true values (yi). In case of cross validation, the RMSECV is calculated using Eq. (4).

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn 2 i ¼ 1 ðyi  yi Þ RMSECV ¼ n

(4)

where n is the number of samples in the training set, yi the reference measurement result for the sample i and yi is the estimated result for sample i when the model is constructed with the sample i removed. The number of PLS factors included in the model is chosen according to the lowest RMSECV. This procedure is repeated for each of the pre-processed spectra. For the test set, the root mean square error of prediction (RMSEP) is calculated as follows (Eq. (5)).

vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi uP  2 u n yi t i ¼ 1 yi  b RMSEP n

(5)

where, n is the number of samples in the test set, yi the reference measurement result for test set sample i and b y i is the estimated result of the model for test sample i.

1000

V.R. Sinija, H.N. Mishra / LWT - Food Science and Technology 42 (2009) 998–1002

2.3. Preparation of calibration and validation models Pure caffeine powder was dissolved in chloroform to give samples of different caffeine concentration in the range between 0 and 90 mg/100 ml. Total volume of each sample prepared was 10 ml. To avoid over fitting of the model, the rule of thumb, which states that the number of PLS factors should not be more than one sixth the number of independent specimens in the calibration set (Kemsley, 1998) was adopted. Hence approximately 36 samples of different concentrations were used to develop the calibration model, and the rest were used for validation. Each sample was analyzed in triplicate. 2.4. Estimation of caffeine by UV spectrometric method Ten milliliter of hot water extract of tea samples were taken in separating funnels, and 10 ml of chloroform was added to each sample. The separating funnel was shaken vigorously for 5 min. The solutions were then allowed to separate for 10 min at room temperature. Lower chloroform layer was collected for further analysis. One milliliter of this chloroform layer was diluted with pure chloroform appropriately to read absorbance. Absorbance of these solutions was measured at 277 nm against pure chloroform as blank using UV–VIS spectrophotometer. A standard curve was prepared for caffeine estimation in the concentration range between 0 and 0.2 mg/ml (Paradkar & Irudayaraj, 2002). Pure caffeine was obtained from Sigma Aldrich Ltd. 2.5. Recovery studies In order to check the accuracy of newly developed FT-NIR method, recovery study was conducted by artificially spiking the tea sample with caffeine and back estimating its amount through the new method. In the present study 20, 40 and 60 mg of pure caffeine was artificially introduced into water extract of tea and the caffeine content was estimated by FT-NIR spectroscopy and UV spectrophotometric methods as described above. The results were expressed as percentage recovery. 3. Results and discussion In this study, the spectra of caffeine was loaded with pure chloroform as the background (blank) for measurement. Fig. 1 shows the FT-NIR spectra of pure caffeine which has major peaks at absorbance bands (wave numbers) of 4201.4, 5285.1, 7137.4, and 8712.6 cm1. These true peaks were selected after smoothing the

Fig. 1. FT-NIR spectra of pure caffeine.

Fig. 2. Molecular structure of caffeine (1,3,7-trimethyl-1H-purine-2,6(3H,7H)-dione).

spectrum to avoid interference due to noise. The structure of caffeine is shown in Fig. 2. Major peaks at absorbance bands or wave numbers of 5285.1 and 4201.4 cm1 may be due to the stretching vibrations of >C ¼ O, C ¼ C and C ¼ N bonds of caffeine. Peaks at 7137.4 and 8712.6 cm1 may be due to –CH bonds of methyl (–CH3) groups. Caffeine molecule has 3 methyl groups on a cyclic structure; hence, the peak may play an important role in the estimation of caffeine. Some minor peaks observed in the pure caffeine spectrum may be due to unknown bond vibrations. Fig. 3 shows the spectrum of tea samples extract in the chloroform which indicates the characteristic absorption spectra of the calibration data set. Spectra are similar to those reported by Chen, Zhao, Fang, and Wang (2007). The most intensive band in the spectrum belong to the vibration of the second overtone of the carbonyl group (5285 cm1), followed by the –CH (7137 cm1), the –CH2 (5472 cm1) and the –CH3 overtone (5808 cm1). The vibration of the C ¼ O, –CH and –CH2 are caused by ingredients such as polyphenols, alkaloids, protein, volatile as well as non-volatile acids and by some aroma compounds (Paradkar & Irudayaraj, 2002). The NIR region contains bands that often overlap, making it difficult to extract spectral parameters of the individual bands. Chemometrics have provided a way of overcoming these problems through empirical models that relate the multiple spectral intensities to known analytes in the samples. As the spectra show similar basic FT-NIR spectral patterns, mathematical transformations were required to use the FT-NIR data for quantitative analysis. Despite

Fig. 3. FT-NIR spectrum of tea extract in chloroform.

V.R. Sinija, H.N. Mishra / LWT - Food Science and Technology 42 (2009) 998–1002

1001

Table 1 RMSEP and R2 values corresponding to each PLS factor for determining caffeine content with different spectral pre-processing methods. Pre-processing technique

RMSEP R2 RMSEE PLS R2 (calibration) (Validation) factor

No pre-processing 1st derivative þ straight line subtraction 1st derivative þ vector normalization Straight line subtraction 2nd derivative

95.23 98.09 84.38 90.3 98.36

4.01 1.95 10.7 8.44 3.08

96.58 99.03

2.36 1.86

7 6

91.66

9.03

4

95.89 99.78

6.62 1.81

5 8

the lack of distinct peaks, it has been shown the PLS can extract relevant information for quantitative determinations (McShane & Cote, 1998). In the application of PLS algorithm, it is generally known that the spectral pre-processing methods and the number of PLS factors are critical parameters. Here their effects on the results are discussed. The optimum number of factors is determined by the lowest RMSEP and highest value for R2. Table 1 shows values of RMSEP and R2 values corresponding to each PLS factor for determining caffeine content with different spectral pre-processing methods. Fig. 4 shows the RMSEP and R2 values plotted as a function of PLS factors for determining caffeine content with first derivative plus straight line subtraction method as the pre-processing technique. Seen from figure, RMSEP value decreases sharply with initial factors, however, gradually decreases as PLS factor increases. R2 value increased up to certain limit, reached a maximum value and thereafter decreased. The NIR calibrations and validation results for the prediction of percentage of moisture in green tea samples are presented in Fig. 5a and b. The test-validated calibration model used the entire spectral region (4000–12000 cm1). PLS regression method gave R2 values of 0.9809 for calibration and 0.9903 for validation set data and RMSEP value of 1.95 for calibration and RMSEE 1.86 for validation set. The results of this study clearly demonstrated the efficiency of FT-NIR for this application.

Fig. 5. Linear regression plot of measured versus predicted caffeine content values for (a) calibration and (b) test data set.

3.1. Analysis of tea samples Tea samples prepared were analyzed by FT-NIR spectroscopy and the previously developed chemometric method was applied to quantify caffeine content in tea samples. Results obtained from FT-NIR spectroscopy were compared with that of the conventional

UV spectroscopic method (Table 2). Results obtained from FT-NIR method was found to be slightly higher than that obtained by UV spectroscopic method. This may be due to the interference of chloroform soluble compounds other than caffeine. In tea, the major alkaloids found are methylxanthines with three distinguished compounds: caffeine, theophylline and theobromine. Due to the similarity in structure and absorbance characteristics of these methylxanthines, caffeine estimates found in tea may actually be the total methylxanthines content in the sample. Similar result was reported for tea and coffee samples using FT-IR spectroscopy by Paradkar and Irudayaraj (2002).

Table 2 Comparison of results obtained by FT-NIR method and UV spectroscopic method for caffeine determination.

Fig. 4. RMSEP and R2 plotted as a function of PLS factors (

,

).

Products

Caffeine content by UV spectroscopic method (mg/100 ml)

Caffeine content by FT-NIR method (mg/100 ml)

Mean

SD

Mean

SD

Instant green tea powder Green tea granules

40.46

1.34

43.28

2.86

46.36

2.43

49.51

3.14

1002

V.R. Sinija, H.N. Mishra / LWT - Food Science and Technology 42 (2009) 998–1002

Table 3 Results of caffeine recovery estimates from UV and FT-NIR spectroscopy. Product

Added caffeine, mg

% Caffeine recovery by UV spectroscopy method

Instant green tea powder

30 60 90

103.9 101.8 101.2

98.3 99.1 99.8

Green tea granules

30 60 90

104.2 102.4 101.8

97.9 99.6 100.1

% Caffeine recovered by FT-NIR method

3.2. Recovery studies The FT-NIR method was further validated through recovery studies by determining the caffeine content in artificially spiked tea samples. FT-NIR prediction in the recovery study was found to be successful and comparable to the result obtained by the UV spectroscopic method (Table 3). The UV spectroscopic method shows a recovery of caffeine in the range between 101.2 and 103.9% for instant tea samples, whereas recovery study predictions using FT-NIR data with PLS model developed was in the range of 98.3–99.8%. The corresponding values for green tea granules were 101.8–104.2% for UV spectroscopic method and 97.9–100.1% for FT-NIR method. The caffeine recovery by FT-NIR method was slightly lower than that of UV method showing a comparatively wider range of variation. However, the simplicity of sample presentation and applicability to a variety of food samples make this an attractive choice. ANOVA results also showed that there is no significant difference between the FT-NIR estimation and the true spiked levels of caffeine in the sample. 4. Conclusions A rapid and simple FT-NIR procedure to estimate caffeine content in green tea was developed using a single calibration model. The model was developed using the spectral region 4000– 12000 cm1. This study demonstrated the suitability of FT-NIR spectroscopy to determine the level of caffeine in tea samples. The maximum value of coefficient of determination (R2) for the prediction of moisture was 98.08 and 99.03 for calibration and validation, respectively. The developed method provides a fast, specific, simple, and easily automatable method for quantitative detection of caffeine content in tea samples. The performance of developed method was confirmed with freshly prepared samples and time taken for individual measurement is between 5 and 10 min. The FT-NIR method was successfully validated by UV spectroscopic determination of caffeine content and also by conducting a recovery study. Acknowledgement The first author is thankful to All India Council for Technical Education (AICTE), New Delhi, for the financial assistance received in the form of National Doctoral Fellowship to pursue her Doctoral programme.

References Airy, A. (1999). Sub-committee on ISO tea to meet in Calcutta next month. http:// www.expressindia.com/fe/daily/19991004/fco04012.html. Accessed 17.09.00. Chen, Q., Zhao, J., Fang, C. H., & Wang, D. (2007). Feasibility study on identification of green, black and oolong teas using near-infrared reflectance spectroscopy based on support vector machine (SVM). Spectrochimica Acta, Part A, 66, 568–574. Esteban-Diez, I., Saiz, G. J. M., & Pizarro, C. (2004). An evaluation of orthogonal signal correction methods for the characterization of Arabica and Robusta coffee varieties by NIRS. Analitica Chemica Acta, 514, 57–67. Ferna´ndez-Iban˜ez, V., Soldado, A., Martı´nez-Ferna´ndez, A., & de la Roza-Delgado, B. (2009). Application of near infrared spectroscopy for rapid detection of aflatoxin B1 in maize and barley as analytical quality assessment. Food Chemistry, 113(2), 629–634. Fujiki, H., Suganuma, M., Okabe, S., Sueoka, E., Sueoka, N., Fujimoto, N., et al. (2001). Cancer prevention with green tea and monitoring by a new biomarker hnRNP B1. Mutation Research 480–481. Hall, M. N., Robertson, A., & Scotter, C. N. G. (1998). Near-infrared reflectance prediction of quality, theaflavin content and moisture content of black tea. Food Chemistry, 27, 61–75. Hopes, M. (1997). What’s your poison, produced by Australian Broadcasting Corporation. http://www.abc.net.au/quantum/ poison/caffeine/caffeine.htm. Accessed 4.06.00. Herrador, M. A., & Gonzalez, A. G. (2001). Pattern recognition procedures for differentiation of green, black and Oolong teas according to their metal content from inductively coupled plasma atomic emission spectrometry. Talanta, 53, 1249–1257. Horie, H., & Kohata, K. (2000). Review-analysis of tea components by high-performance liquid chromatography and high-performance capillary electrophoresis. Journal of Chromatography A, 881, 425–438. Horie, H., Mukai, T., & Kohata, K. (1997). Simultaneous determination of qualitatively important components in green tea infusions using capillary electrophoresis. Journal of Chromatography A, 758, 332–335. ISO (International Standard Organization). (1994). ISO TC 34/SC 8 N 444. Jian, L., Xie, L. P., Lee, A. H., & Binns, C. W. (2004). Protective effect of green tea against prostate cancer: a case-control study in southeast China. International Journal of Cancer, 108, 130–135. Kemsley, E. K. (1998). Case studies. In E. K. Kemsley (Ed.), Discriminant analysis and class modeling of spectroscopic data (pp. 110–164). Chichester, UK: John Wiley and Sons Ltd. Lupaert, J., Zhang, M. H., & Massart, D. L. (2003). Feasibility study for the using near infrared spectroscopy in the qualitative and quantitative of green tea, Camellia sinensis (L.). Analitica Chemica Acta, 487, 303–312. McGlone, V. A., Jordan, R. B., Seelye, R., & Martinsen, P. J. (2002). Comparing density and NIR methods for measurement of kiwifruit dry matter and soluble solids content. Postharvest Biology and Technology, 26, 191–198. McShane, M. J., & Cote, G. L. (1998). Near-infrared spectroscopy for determination of glucose lactate, and ammonia in cell culture media. Applied Spectroscopy, 52, 1073–1078. Nakachi, K., Matsuyama, S., Miyake, S., Suganuma, M., & Imai, K. (2000). Preventive effects of drinking green tea on cancer and cardiovascular disease: epidemiological evidence for multiple targeting prevention. Biofactors, 13, 49–54. Paradkar, M. M., & Irudayaraj, J. (2002). A rapid FTIR spectroscopic method for estimation of caffeine in soft drinks and total methylxanthines in tea and coffee. Journal of Food Science, 67(7), 2507–2511. Schulz, H., Engelhardt, U. H., & Wengent, A. (1999). Application of NIRS to the simultaneous prediction alkaloids and phenolic substance in green tea leaves. Journal of Agricultural and Food Chemistry, 475, 5064–5067. Togari, N., Kobayashi, A., & Aishima, T. (1995). Pattern recognition applied to gas chromatographic profiles of volatile component in three tea categories. Food Research International, 28, 495–502. Woodcock, T., Downey, G., & O’Donnell, C. P. (2008). Review: better quality food and beverages: the role of near infrared spectroscopy. Journal of Near Infrared Spectroscopy, 16(1), 1–29. Yang, C. S., Maliakal, P., & Meng, X. (2002). Inhibition of carcinogenesis by tea. Annual Review of Pharmacology and Toxicology, 42, 25–54. Zhang, D., Kuhr, S., & Engelhardt, U. H. (1992). Influence of catechins and theaflavins ¨ r Lebensmittel-Untersuchung on astringent taste of black tea brews. Zeitschrift fu und -Forschung A (European Food Research and Technology), 195, 108–111. Zhang, M. H., Lupaert, J., Xu, Q. S., & Massart, D. L. (2004). Determination of total antioxidant capacity in green tea by NIRS and multivariate calibration. Talanta, 62, 25–35. Zuo, Y. G., Chen, H., & Deng, Y. W. (2002). Simultaneous determination of catechins, caffeine and gallic acids in green, oolong, black and pu-erh teas using HPLC with a photodiode array detector. Talanta, 57, 307–316.