Available online at www.sciencedirect.

com

Journal of Chromatography A, 1170 (2007) 914

Rapid determination of caffeine in one drop of beverages and foods using drop-to-drop solvent microextraction with gas chromatography/mass spectrometry
Kamlesh Shrivas, Hui-Fen Wu
Department of Chemistry, National Sun Yat-Sen University, Kaohsiung 804, Taiwan Received 24 April 2007; received in revised form 4 September 2007; accepted 11 September 2007 Available online 14 September 2007

Abstract A simple and rapid sample cleanup and preconcentration method for the quantitative determination of caffeine in one drop of beverages and foods by gas chromatography/mass spectrometry (GC/MS) has been proposed using drop-to-drop solvent microextraction (DDSME). The best optimum experimental conditions for DDSME were: chloroform as the extraction solvent, 5 min extraction time, 0.5 L exposure volume of the extraction phase and no salt addition at room temperature. The optimized methodology exhibited good linearity between 0.05 and 5.0 g/mL with correlation coefcient of 0.980. The relative standard deviation (RSD) and limits of detection (LOD) of the DDSME/GC/MS method were 4.4% and 4.0 ng/mL, respectively. Relative recovery of caffeine in beverages and foods were found to be 96.6101%, which showing good reliability of this method. This DDSME excludes the major disadvantages of conventional method of caffeine extraction, like large amount of organic solvent and sample consumption and long sample pre-treatment process. So, this approach proves that the DDSME/GC/MS technique can be applied as a simple, fast and feasible diagnosis tool for environmental, food and biological application for extremely small amount of real sample analysis. 2007 Elsevier B.V. All rights reserved.
Keywords: Caffeine; Beverages and foods; Drop-to-drop solvent microextraction; Gas chromatography/mass spectrometry

1. Introduction Caffeine (1,3,7-trimethylxanthine or 3,7-dihydro-1,3,7trimethyl-1H-purine-2,6-dione) is an alkaloid occurs naturally in tea leaves, coffee beans and cocoa beans/leaves and is traditionally used for its stimulatory effects. People consume caffeine in coffee, tea, cocoa, chocolate, many soft drinks, and some drugs. Pure caffeine is odorless and has a bitter taste [1,2]. Caffeine is one of the legally permitted additives in beverages, with a tolerated upper limit of 0.015% (w/w) [3]. It stimulates central nervous system, heart and acts as a diuretic and intensies brain activity. However, high amount of caffeine exists in human beings may cause many undesired symptoms such as tremor, tachycardia, gastrointestinal problem, seizures and even death [4,5]. Therefore, it is important to monitor caffeine in beverages and foods for healthy purposes of human beings.

Corresponding author. Tel.: +886 7 5252000x3955; fax: +886 7 5253908. E-mail address: hwu@mail.nsysu.edu.tw (H.-F. Wu).

Several analytical methods have been proposed for the determination of caffeine in beverages and foods based on UVvis spectrophotometry [6,7], high-performance liquid chromatography (HPLC) [810], gas chromatography (GC) [11], Fourier transform infrared spectrometry (FTIR) [12,13], ion chromatography [14,15], thin-layer chromatography/mass spectrometry (TLC/MS) [16] and capillary electrophoresis (CE) [17]. Among these methods, spectrophotometric methods involve tedious, laborious and time-consuming sample pre-treatment steps and also chance of interferences such as theobromine and theophylline. Chromatographic determination of analytes requires sample preparations through liquidliquid extraction (LLE) [18] or solid-phase extraction (SPE) [1921]. These methods are also time consuming and more amount of reagents and solvents is required. Recent microextraction of analytes through solid-phase microextraction (SPME) [22,23] and liquid-phase microextraction (LPME) including single-drop microextraction (SDME) [24,25], hollow-ber liquid-phase microextraction (HF-LPME) [2628] and drop-to-drop solvent microextraction (DDSME) [29] have been successfully applied for analysis

0021-9673/$ see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2007.09.020

10

K. Shrivas, H.-F. Wu / J. Chromatogr. A 1170 (2007) 914

of pharmaceutical and environmental samples. From these methods, SPME and HF-LPME techniques require bers for extraction of analytes. As to the SDME method, only a single microdrop of solvent is used as acceptor phase, which is hanging at angle cut tip of a microsyringe and immersed in the aqueous solution for the extraction of analytes. This method is very simple, convenient and almost cost free compared with SPME and HF-LPME. The SDME typically used several to 20 mL of aqueous solution to extract analytes into organic phase. From commercial and economic point of view, sample volume should be reduced to a small amount could useful for consumer and production industry because this may denitely reduce cost of the samples. Thus, DDSME is a best technique to t these purposes since only one drop (7 L) of sample solution was used for extraction of analyte. In our knowledge, this is the rst application of the DDSME/GC/MS method applied on real sample analysis for food and drink where 7 L (one drop) of sample solution was used for extraction of caffeine for 5 min at room temperature and subsequently determined by GC/MS. Extraction parameters including solvent selection, extraction time, and exposure volume of extraction phase, addition of salt were optimized in order to obtain the best extraction efciency for the DDSME/GC/MS method. 2. Experimental 2.1. Chemicals and reagents The puried water was prepared on a Milli-Q reagent water system (Millipore, Milford, MA, USA) for spiked water sample of caffeine. All organic solvents were of HPLC grade. Methanol was purchased from Mallinckrodt (Phillipsburg, NJ, USA). Chloroform and dichloromethane were obtained from Mallinckrodt (Paris, KY, USA). Toluene was purchased from J.T. Baker (NJ, USA). Octane was purchased from SigmaAldrich (Steinheim, Germany). Caffeine was purchased from Aldrich (St. Louis, MO, USA). Stock standard solution was prepared by dissolving 10 mg of caffeine in 10 mL of methanol and was stored in glass-stopper bottles at 20 C. Working Standard solutions were prepared by appropriate dilution of stock standard solution in deionized water. 2.2. Sample collection and preparation Six beverages including Coca Cola, Spirit, Coke Light (Coca Cola Co., Taiwan), Pepsi (Pepsi Co., Taiwan), Colombia Coffee (Uni-President Co., Taiwan), Tea Drink (Yes Water Co., Taiwan) and six foods including Black, Green and Oolong Tea (TenRen Tea, Co. Ltd., Taiwan), Nescafe Coffee (Nestle, Singhai Ltd., China) Kaisers Milk Chocolate (Taiwan Kaisers Food Industry, Taiwan), Raisin Choco Ball (I-Mei Food Co., Taiwan) branded samples were collected from local markets of Tamsui, Taiwan. A 25 mL of beverage samples were taken in 100 mL of beaker and kept in ultrasonic bath for 10 min to degas carbonate present in the samples. Then, after it was ltered through Whatman No. 44 lter paper and appropriate amount ltrate was diluted with deionized water according to caffeine present

in samples. A 0.5 g of Black, Green, Oolong tea and Nescafe coffee samples were weighed separately into a 100-mL beaker, and 45 mL of boiling water was added to each. The samples were kept on boiling water for 5 min, with constant temperature at 80 C. The solution was ltered through Whatman No. 44 lter paper and ltrate was diluted according to concentration of caffeine present in sample. A 4.0 g of chocolate sample was dissolved in 45 mL of boiling water and mixing with a magnetic stirrer with maintaining temperature at 80 C for 30 min. The solution was ltered through Whatman No. 44 lter paper and dilution was made according to concentration of caffeine present. 2.3. Extraction procedures of the drop-to-drop solvent microextraction A sketch of DDSME apparatus for the extraction of caffeine from one drop of sample solution is illustrated in Fig. 1. One drop (7 L) of spiked pure water containing 10 g/mL of caffeine was taken in 100 L of sample vial. Then, 1 L of acceptor organic solvent was taken in 10 L of microsyringe (with an angledcut needle tip) for DDSME extraction and injection to GC/MS. Before extraction of caffeine, the microsyringe was washed thoroughly with the extraction solvent to prevent air bubbles, which can lead to errors in the reproducibility of the results. The needle was passed through the sample vial septum and immersed in the aqueous sample. The syringe was hold tightly by clamp in xed position in order to consistently place the needle tip in the 7 L of aqueous phase. Then syringe plunger was depressed down to expose a 0.5 L of the extraction solvent into the aqueous phase. After 5 min of extraction, only organic drop was taken back into the syringe as original mark of 1 L. Finally, withdraw syringe from sample vial and extracted solvent was directly injected into GC/MS for caffeine analysis. All experiments were performed at room temperature.

Fig. 1. Sketch of DDSME apparatus for the extraction of caffeine from one drop of sample solution.

K. Shrivas, H.-F. Wu / J. Chromatogr. A 1170 (2007) 914

11

2.4. Gas chromatography/mass spectrometry analysis Analysis of caffeine was performed on a Varian 3800 model GC system combined with an internal source ion trap mass spectrometer (Saturn 2000, Varian, Walnut Creek, CA) under the electronic ionization (EI) mode. Helium (99.999%) was used as carrier gas and ow rate of 1.0 ml/min. The GC separation of caffeine was performed using a DB-5 capillary column (30 mm 0.25 mm I.D., 1 m lm thickness) obtained from Supelco, Bellefonte, PA, USA. The following temperature program was maintained during separation of caffeine. One hundred and fty degree Celsius for 1 min; 20 C/min to 230 C for 2.0 min and total time of analysis was 7 min. The injections were performed in the splitless mode at 250 C. The trap, transfer line, and manifold temperature were set at 200, 280, and 50 C, respectively. The mass range of scan spectra was 50250 Da. Selective ion monitoring mode (SIM) was used to perform all quantication experiments for caffeine. The utilized ions used for conrmation of m/z values of caffeine were 194, 109, 82 and 52. 3. Results and discussion 3.1. Optimization of the DDSME method In this study, the DDSME method was applied for all the extractions of caffeine from aqueous phase to organic phase. The best extraction efciency of caffeine was achieved by optimizing the effects of solvent selection, extraction time, exposure volume of extraction phase and also addition of salt to the aqueous phase. Extraction efciency of the method was evaluated as their average peak area of three successive injections (n = 3). 3.1.1. Selection of organic solvent The different chemical properties of various organic solvents were preferred based on their polarity, solubility of analytes, and immiscibility with aqueous phase. Also it should exhibit an excellent GC behavior [28]. The organic solvents used for the extraction of spiked caffeine from aqueous phase were chloroform, dichloromethane, toluene, and n-octane for 5 min of extraction time at room temperature in the absence of salt addition. Chloroform was found to be the most suitable solvent to extract caffeine from aqueous solution. The reason is due to its high solubility of analytes with chloroform. Also the solubility of caffeine in chloroform is nine times more than water at room temperature. Thus, chloroform was chosen for further experimental work. The results of solvent effect are given in Table 1.
Table 1 Effect of solvent on extraction efciency of caffeine using DDSME/GC/MS Solvents Chloroform Dichloroform Toluene Octane Peak area 316089 57234 19651 14251 RSD, n = 3 (%) 2.4 3.4 3.1 3.5

Fig. 2. Effect of time proles on extraction efciency of caffeine obtained from DDSME (concentration level at 10 g/mL, chloroform as extraction solvent; exposure volume of acceptor phase, 0.5 L; and no salt addition at room temperature).

3.1.2. Extraction time Extraction time is another factor that may be effects extraction efciency of analytes from aqueous phase to organic phase. As with LLE and SPME, it is not necessary for LPME to reach equilibrium, since this is not an exhaustive extraction but a partitioning between the organic and aqueous phase [24,27,28]. The effect of extraction time on extraction efciency of caffeine from aqueous phase was studied by monitoring the variation of peak area with extraction time of 3, 5, 7, 10 and 15 min at room temperature. A series of spiked water samples (10 g/mL of caffeine) were prepared and the variation of the peak area was investigated as shown in Fig. 2. The extraction efciency was found maximum at 5 min and after it rapidly decreased which could be assessed in the curve. In DDSME, after 5 min of extraction time the problem of solvent depletion started and this caused the decrease in the extraction efciency of solvent. The results obtained were consisted with the previous work of Wu et al. [29]. Therefore, 5 min of extraction time was used for the extraction of caffeine for further experiments. 3.1.3. Exposure volume of extraction phase The exposure volume of organic solvent into aqueous phase is a very important factor that affects the extraction efciency of DDSME. Thus, different volume of extraction phase from 0.2 to 1.0 L was exposed to the aqueous phase. The DDSME extraction of caffeine from aqueous phase was set at 5 min of extraction time in the absence of salt under room temperature. Fig. 3 shows the increase of extraction efciency of caffeine when the exposure volume of extraction phase was increased to 1.0 L. But, there was chance of miscibility of aqueous phase as the volume of exposure acceptor phase increased more than 0.5 L and this also caused the poor precision (RSD, >12%) of the results shown in Fig. 3. An analysis of variance (ANOVA) test was performed in order to check the variability of the results at the different exposure volume. The variance value obtained by the ANOVA test was not statistically signicantly different. However, for further experiments, 0.5 L exposure volume of extraction phase was used in order to maintain better precision (RSD, <5%) of the results and easy manipulation of organic solvent from extremely small volume of the sample solution (7 L).

12

K. Shrivas, H.-F. Wu / J. Chromatogr. A 1170 (2007) 914 Table 2 Determination of caffeine in one drop of beverages by the DDSME/GC/MS method Sample Coca Colaa Coke Lighta Pepsia Spirita Gourmet coffee Colombia coffee Tea drink Caffeine (g/mL) 74.8 80.9 146.6 ND 108.7 389.6 100.0 RSD, n = 3 (%) 5.6 7.3 7.1 ND 3.0 4.2 5.5

Fig. 3. Effect of exposure volume of acceptor phase on extraction efciency of caffeine obtained from DDSME (concentration level at 10 g/mL, chloroform as extraction solvent; extraction time, 5 min; and no salt addition at room temperature).

Carbonate containing sample, ND = not detected.

3.1.4. Addition of salt The addition of salt in aqueous solution can reduce the solubility of analytes in the aqueous phase and enhance their migration towards the organic phase in SDME and LLE [28,30]. The effect of salt concentration on extraction of the caffeine was evaluated by adding potassium chloride (030%) into spiked aqueous solution (10 g/mL of caffeine) at 5 min of extraction time. Fig. 4 shows that the peak area was decreased with the increment of KCl concentration in the studied range. This is due to free migration of analyte were prohibited by increasing salt ion concentration and viscosity of the aqueous phase. Wu et al. and Liu et al. obtained similar contradictory results in the LPME studies [26,29,30]. Therefore, no salt was used for further experiments. 3.2. Enrichment factor and evaluation of the method The enrichment factor (EF) is dened as the ratio of concentration of analyte between the organic acceptor phase (Co ) and the initial concentration of analyte (Ci ) in aqueous solution as below. EF = Co Ci

time at room temperature was 16. A linear calibration graph was obtained over the concentration range of 0.055.0 g/mL of caffeine for spiked standard solutions. The value of correlation coefcient was 0.980, showing good correlation with peak area as a function of concentration of caffeine. In addition, the linear response of SIM peaks obtained with authentic stock solution was compared to the DDSME in Fig. 5. The detection limit (LOD) of the DDSME method was calculated as three times of standard deviation (3 ). The LOD obtained in deionized water was 4.0 ng/mL. The reproducibility of the method was assessed by carrying out six replicate analysis (n = 6) of standard solution of 10 g/mL of caffeine. The relative standard deviation (RSD) of the method was found to be 4.4%. 3.3. DDSME method to determine caffeine in one drop of beverages and foods The DDSME method has been successfully applied for the determination of caffeine in various beverages and food products. A 7 L (one drop) of diluted sample of beverages, tea, coffee and chocolate were taken in sample vial for extraction of caffeine by DDSME and then analyzed by GC/MS. The quantitative amount of caffeine in beverages and foods were calculated based on the calibration curve. The results obtained for beverages and foods of different variety are summarized in Tables 2 and 3, respectively. The concentration of caffeine in beverages and foods were ranging from 74.8 to 389.6 g/mL and 0.1530.51 mg/g with RSD of 3.07.3 and 2.37.5%, respectively. To verify the validity of the proposed method, known amounts of caffeine were added to sample of beverages and foods in order to evaluate the relative recovery (determined as ratio of the concentration of analyte found in sample to spiked (added) amount of analyte) as shown in Table 4. The relative
Table 3 Determination of caffeine in one drop of foods by the DDSME/GC/MS method Sample Black tea Green tea Oolong tea Nescafe coffee Raisin choco ball Kaisers milk chocolate Caffeine (mg/g) 29.98 24.80 30.51 0.41 0.25 0.15 RSD, n = 3 (%) 3.7 3.4 6.0 2.3 7.5 6.7

The optimal extraction conditions were used to investigate enrichment factor by spiking 0.5 g/mL of caffeine in deionized water. The enrichment factor of caffeine for 5 min of extraction

Fig. 4. Effect of salt addition on extraction efciency of caffeine obtained from DDSME (concentration level at 10 g/mL, chloroform as extraction solvent; extraction time, 5 min; exposure volume of acceptor phase, 0.5 L; at room temperature).

K. Shrivas, H.-F. Wu / J. Chromatogr. A 1170 (2007) 914

13

Fig. 5. SIM peaks obtained with DDSME (eh) in comparison to an authentic stock solution of caffeine (ad).

14

K. Shrivas, H.-F. Wu / J. Chromatogr. A 1170 (2007) 914

Table 4 Results of relative recovery in one drop of beverages and foods Sample Pepsi Spirit Black tea Oolong tea Caffeine addeda 200 50.0 30.0 30.0 Caffeine founda 198.3 48.3 29.62 30.3 Relative recovery (%) 99.1 96.6 98.7 101 RSD, n = 3 (%)

NSYSU) is acknowledged for the assistance in interpreting statistical results. References

4.0 6.3 3.9 5.2 [1] M.J. ONeil, A. Smith, P.E. Heckelman, J.R. Obenchain, J.A.R. Gallipeau, M.A. DArecca, S. Budavari, Merck IndexAn Encyclopedia of Chemicals, Drugs and Biologicals, 13th ed., Merck & Co., Rahway, NJ, 2001, p. 275. [2] J.A.T. Pennington, Food Values of Portions Commonly Used, 18th ed., J.B. Lippincott, Philadelphia, PA, 2005. [3] Ministerio de Agricultura, Pescay Alimentacion. Recopilacion Legislativa Alimentaria, Capitulo 29, bebidas no alcoholicas, Madrid, 1982. [4] A.H. Varnam, J.P. Sutherland, Beverages: Technology, Chemistry and Microbiology, Chapman & Hall, London, 1994, p. 234. [5] B. Stavric, Food Chem. Toxicol. 26 (1988) 541. [6] A. Cepeda, P. Paseiro, J. Simal, J.L. Rodrguez, Alimentaria 211 (1990) 23. [7] O. Lau, S. Luk, O. Cheng, T.P.Y. Chiu, Analyst 117 (1992) 777. [8] E. Nishitani, Y.M. Sagesaka, J. Food Compos. Anal. 17 (2004) 675. [9] H. Horie, A. Nesumi, T. Ujihara, K. Kohata, J. Chromatogr. A 942 (2002) 271. [10] J.P. Naik, J. Agric. Food Chem. 49 (2001) 3579. [11] E.D. Conte, E.F. Barry, Microchem. J. 48 (1993) 372. [12] N.M. Naja, A.S. Hamid, R.K. Afshin, Microchem. J. 75 (2003) 151. [13] Y. Daghbouche, S. Garrigues, M.T. Vidal, M. de la Guardia, Anal. Chem. 69 (1997) 1086. [14] Q.C. Chen, J. Wang, J. Chromatogr. A 937 (2001) 57. [15] Q.C. Chen, S.F. Mou, X.P. Hou, Z.M. Ni, Anal. Chim. Acta 371 (1998) 287. [16] M.J. Ford, M.A. Deibel, B.A. Tomkins, G.J. Van Berke, Anal. Chem. 77 (2005) 4385. [17] C.N. Chen, C.M. Liang, J.R. Lai, Y.J. Tsai, J.S. Tsay, J.K. Lin, J. Agric. Food Chem. 51 (2003) 7495. [18] A. Perva-Uzunalic, M. Skerget, Z. Knez, B. Weinreich, F. Otto, S. Gruner, Food Chem. 96 (2006) 597. [19] A. Senol, A. Aydin, J. Food Eng. 75 (2006) 565. [20] Y.R. Ku, K.C. Wen, L.K. Chang, Y.S. Ho, J. Pharm. Biomed. Anal. 20 (1999) 351. [21] J. Patsias, E.P. Mourkidou, J. Chromatogr. A 904 (2000) 171. [22] P. Vesely, L. Lusk, G. Basarova, J. Seabrooks, D. Ryder, J. Agric. Food Chem. 51 (2003) 6941. [23] M.A. Crespn, M. Gallego, M. Valcarcel, J. Chromatogr. B 773 (2002) 89. [24] B.M. Liu, P. Malik, H.F. Wu, Rapid Commun. Mass Spectrom. 18 (2004) 2059. [25] M.A. Jeannot, F.F. Cantwell, Anal. Chem. 69 (1997) 235. [26] C.H. Yan, H.F. Wu, Rapid Commun. Mass Spectrom. 18 (2004) 3015. [27] L. Zhao, H.K. Lee, Anal. Chem. 74 (2002) 2486. [28] G. Shen, H.K. Lee, Anal. Chem. 74 (2002) 648. [29] H.F. Wu, J.H. Yen, C.C. Chin, Anal. Chem. 78 (2006) 1707. [30] J. Liu, G. Jiang, Y. Chi, Y. Cai, Q. Zhou, J.T. Hu, Anal. Chem. 75 (2003) 5870.

a Amount of sample taken in g/mL and mg/g for beverages and foods, respectively.

recovery of caffeine in beverages and foods were found to be 96.6101% with precision (RSD) value of 3.96.3%. 3.4. Matrix effect The good recovery values (96.6101%) obtained in the beverage and food samples demonstrate that the matrix had no effect on DDSME analysis of caffeine. Enrichment factor was calculated in beverage sample (spirit) as similar to the deionized water. The EF found in beverage (spirit) was 14 at the optimized conditions of DDSME. This result also supported that the matrix of the analyzed beverage sample had no effect on the extraction of caffeine. 4. Conclusions For the rst time, the DDSME/GC/MS method is successfully applied for determination of caffeine in one drop of various beverages and foods. The present methodology is easy, rapid, sensitive, and requires small sample volumes (7 L) for the separation and preconcentration of caffeine. Also it is a low-cost technique due to use of minute amount of solvent and reagents for the extraction of caffeine throughout the experiments. This approach has proven that DDSME/GC/MS is a simple, effective and rapid diagnosis tool for the quantitative analysis of trace amount of drugs in ultra small sizes of biological samples such as urine and blood. Acknowledgements The authors thank the National Science Council of Taiwan for nancial support at the contract number of NSC 95-2113M110-019-MY3 issued to National Sun Yat-Sen University. Mong-Na Lo Haung (Department of applied mathematics,