Determination of Caffeine and Other Purine Compounds in Food and


Determination of Caffeine and Other Purine Compounds in Food and...

1 downloads 173 Views 88KB Size

In the Laboratory

Determination of Caffeine and Other Purine Compounds in Food and Pharmaceuticals by Micellar Electrokinetic Chromatography C. Vogt* and S. Conradi Institute of Analytical Chemistry, University of Leipzig, Linnestr. 3, 04103 Leipzig, Germany E. Rohde Department of Chemistry, University of Cincinnati, Mail Location 172, Cincinnati, OH 45221 Capillary electrophoresis (CE) is a modern separation technique, which has been applied to the separation and quantification of a wide variety of substances (1, 2). The extremely high efficiencies and minimal requirements for buffers, samples, and solvents have accelerated the development of CE and made it popular. The rapid increase in the number of applications of the technique in industrial laboratories makes it necessary to introduce it to educational laboratories at universities as well. The relatively simple construction and low price of the instrumentation, as well as low volumes of solvents and waste and the ease of maintenance, make this technique very attractive for use in undergraduate education. We developed an experiment that can be used to make students familiar with the technique of micellar electrokinetic chromatography (MEKC) as a special form of capillary electrophoresis and its separation principle. For that purpose, caffeine and other important purine compounds have been determined in various sample matrices. In this experiment students solve a relatively simple separation problem by variation of buffer pH, buffer components, and separation parameters. By doing a calibration for the analyzed purine compounds they also learn about reproducibility in CE. Caffeine and related compounds are components of the diet. They are naturally present in foodstuffs (e.g., tea, coffee, cocoa) or are added during processing. The concentration of these substances determines the stimulating or sedative effects of foods, thus having great influence on the wellbeing and health of the consumer. From the concentration of caffeine or theobromine information about the origin or processing of some foods can be derived. This is especially helpful for the quality control of decaffeinated beverages or chocolate. Many pharmaceutical formulations also contain caffeine, to compensate for the effect of the main components of the formulation on blood pressure or other body functions. The determination of caffeine is part of quality control during the pharmaceutical manufacturing process. Capillary electrophoresis has been used to determine caffeine in food samples (3) and pharmaceuticals (4), as well as in clinical samples (5, 6). The experiment we describe serves as an introduction to both CE and micellar electrokinetic chromatography. Purine compounds are separated by addition of sodium dodecylsulfate (SDS) to the separation buffer, so not only electrophoresis but also partitioning between running buffer (aqueous phase) and micelles (micellar phase) enables separation. Based upon our experiences we recommend a short discussion about the configuration and detection mode of CE instrumentation as well as separation principles of CE and MEKC—especially electrophoretic migra*Corresponding author.

1126

tion, electroosmotic flow, and partition equilibrium—before starting practical work. This paper describes the separation capabilities of MEKC for the determination of purine substances in various matrices. Optimization of experimental conditions was done by the application of a simple setup, which could be easily applied in student laboratories. The paper also discusses different sample preparation procedures developed with regard to the special properties of the sample matrices. Experimental Procedures

Instrumentation The instrument used in this work was a Beckman capillary electrophoresis model P/ACE 2100. It is equipped with a detector having a set of fixed wavelengths in the UV region. If not mentioned otherwise, absorbance was monitored at 214 nm during the experiments. The columns used were uncoated fused silica capillaries with an inner diameter of 50 µm and a length of 50 cm to the detector and 57 cm to the end. The operating voltage was 17 kV. A pressure of 3.45 kPa was used for the sample injection. The injection time was 3 s for the purine standards and all samples; an additional 2-s injection step was used to inject the internal standard phenobarbital. To obtain reproducible results, the column was flushed after each run. A 5-min rinsing step with a buffer containing 20 mM borate pH 9.00 and 90 mM SDS was followed by a 5-min rinsing step with 0.1 M NaOH. The electropherograms were recorded by an IBM/PS2 computer using the P/ACE 2000 software. Reagents Methanol, NaOH, HCl, NaH 2PO4 , Na2 HPO4 , H3 BO3, Na2 B4 O7?10 H2O, Pb(CH 3COO)2?3 H 2O, PbO, and NaHCO3, were obtained from Merck (Darmstadt, Germany). Sodium dodecylsulfate (SDS) was from Riedel–de Haën (Seelze, Germany). Caffeine, theophylline, and theobromine were from Fluka (Buchs, Switzerland), and phenobarbital (5-ethyl-5phenylbarbituric acid) from Sigma (Deisenhofen, Germany). All chemicals used were of reagent grade. Triply distilled water was taken from a laboratory water station. Darjeeling black tea and Cebe cocoa (Wilhelm Reuss GmbH) were obtained from the supermarket; the pain killer Vivimed (Dr. Mann Pharma, Berlin) was bought in a drug store.

Sample Preparation Stock solutions of 100 mM phosphate pH 4.60, 100 mM borate pH 8.25, 100 mM phosphate pH 8.25, and 300 mM SDS were prepared by dissolving the appropriate amount of the sodium salt in triply distilled water. Caffeine and

Journal of Chemical Education • Vol. 74 No. 9 September 1997

In the Laboratory theophylline standards with 1 mg/mL and a theobromine standard with 0.25 mg/mL were prepared. A 1 mM stock solution of phenobarbital was used as an internal standard. Buffers for the electrophoretic separation were prepared by diluting the stock solutions to the final concentration. For the calibration procedure, 5 mixtures of the 3 purine substances were prepared, each containing all 3 standards in concentrations between 0.1 and 0.01 mg/mL. In the standards, each of the components was contained at its minimum (0.01 mg/mL) and its maximum (0.1 mg/mL) concentration only once. Three different concentrations between these two limiting concentrations were used to prepare the remaining solutions. None of the calibration solutions contained the same concentration for 2 or more components. The stock solution of the internal standard phenobarbital was diluted to 0.2 mM before injection. To prepare the cocoa sample, a saturated lead acetate solution was prepared from 60 g of Pb(CH3COO)2?3 H2O and 20 g of PbO. The substances were mixed thoroughly and worked into a paste by adding 10 mL of water. During the heating of the paste, distilled water was added to a total volume of 200 mL. After filtration, the solution, with a density between 1.224 and 1.229 g/mL, was used to remove interfering components from the food samples by complexation or precipitation. Tablets were weighed exactly, ground thoroughly, and dissolved in 80 mL of distilled water. After filtration the solution was diluted up to the 100-mL mark of a volumetric flask. This standard solution was diluted 2-, 5-, or 10-fold before the injection to obtain a sample with a caffeine concentration of about 50–100 µg/mL. Black tea samples were prepared as follows (extraction procedure A). About 7 g of tea (exactly weighed) were covered with about 900 mL of boiling distilled water. After 4 min the extract was filtered. Then it was cooled to room temperature and diluted with distilled water to the 1.000L mark of a volumetric flask. The injection was performed with 2- or 5-fold diluted samples. Cocoa samples were prepared according the following procedure (extraction procedure B). One gram of cocoa powder or instant cocoa mix or 3 g of a cocoa-containing product (e.g., chocolate) was weighed exactly and transferred into a 300-mL Erlenmeyer flask. After adding boiling chips, the total weight of the flask was determined. Then 96 mL of cold distilled water was added. The mixture was heated and gently swirled. It was kept barely boiling for about 30 min. With constant swirling, 4 mL of the saturated lead acetate solution was added to the hot solution. The weight of the flask was determined and any evaporated liquid was replaced by addition of water to a total weight of 101.0 g (4 mL saturated lead acetate solution weighs 5 g). The wellmixed content of the flask was filtered through filter paper and the first 10 mL of the filtrate was discarded. Then solid NaHCO3 (about 2 g) was added and the filtration was repeated. Again, the first 10 mL of the filtrate was discarded. Finally the solution was filtered through a membrane filter (pore size 0.5 µm).

(a) H3C

O

O

CH3 N

N N CH3

N

(b) H

O

O N N CH3

CH3

(c)

N

H 3C

N

O

O

Results and Discussion

Method Development To establish optimal separation conditions for capillary electrophoresis, the acid/base and optical properties of the analytes must be taken into account. Good separations are obtained when the analytes are charged and their specific electrophoretic mobility gives rise to differential migration. The buffers used should therefore have pH < pKb – 1 or pH > pKa + 1. The purines investigated (Fig. 1) have very similar structures and the pKa values are in the range between 8 and 10. They are uncharged under neutral conditions. Therefore differences in charge and electrophoretic mobility could be achieved by skilled choice of buffer pH. Based on this, buffers with pH 4.60 and 8.25 were used for the separation. Optimal separation conditions were established for a mixture of caffeine, theophylline, and theobromine. Phenobarbital was used as an internal standard. In Figure 2A the separations at low pH with no additional buffer additives is shown. In a separation at pH 4.60, all four components of the standard mixture co-elute. The compounds are not charged and therefore they do not migrate. In addition, the electroosmotic flow at this pH is relatively low, causing only a slow transport of the analytes to the detector with little to no discrimination. If the pH is lowered below 4, the standards become protonated and migrate to the cathode past the detector window (not shown). The still unseparated band reaches the detector earlier than at pH 4.60. The differences in electrophoretic mobilities, however, are insufficient to achieve separation of the four components. Below pH 4 the contribution of electroosmosis is negligible. In a separation of the standards below or around their pKa values at pH 8.25, the molecules are only weakly or not at all deprotonated. Therefore, in a simple borate buffer caffeine and theobromine comigrate in one peak, theophylline (with the lowest pKa of the purines at 8.50) is partially resolved, and separation is achieved only for the internal standard, which possesses the lowest pKa of all (separation not shown). At this point, the data provided by the electropherograms should be discussed. Separation of the purine compounds solely by varying buffer pH is not possible because differences in electrophoretic mobilities are not sufficient.

CH3 N

N N

N

H

Figure 1. Structures of (a) caffeine, (b) theobromine, and (c) theophylline.

Figure 2. Separation of a standard mixture in different buffers. A: 10 mM phosphate pH 4.60; B: 10 MM borate, 10 mM phosphate pH 8.25, 45 mM SDS, 10% (v/v) methanol. 1 = theobromine; 2 = caffeine; 3 = theophylline; 4 = phenobarbital (internal standard).

Vol. 74 No. 9 September 1997 • Journal of Chemical Education

1127

In the Laboratory Therefore a second separation mechaTable 1. Calibration Parameters of the Standard Mixturea nism is introduced by adding SDS to the SDt buffer. SDS is a micelle-forming agent Compound t av rA rh Calibration Equation (%) leading to formation of a pseudoA = 17.7555*c – 0.050005 Theobromine 8.801 0.78 .9954 .9995 stationary phase in the buffer. The analytes are separated owing to differCaffeine 9.653 0.71 .9998 .9913 A = 16.63875*c + 0.097481 ences in hydrophobicity. The associated Theophylline 9.870 0.94 .9995 .9998 A = 18.85615*c – 0.026193 mass transfer process in the system P h e n o b a r b i t a l 1 3 . 4 4 7 0 . 8 4 – – – makes the separation possible (Fig. 2B). at The partition coefficients for compounds av is the average migration time of the peak; SD t is the standard deviation of the migration time; and rA and rh are the correlation coefficients of the calibration curves using peak area or peak height. between the aqueous buffer and the micellar phase are largely determined by the hydrophobic properties of the Table 2. Caffeine and Theobromine Contents of Real Samples analyte: the more hydrophobic the molecule, the higher the probability for it to Sample Concentration Concentration in be incorporated in the micellar phase. Sample Weight Compound Determined Original Sample Negatively charged SDS micelles will (g) (mg/mL) (wt %) migrate toward the inlet or site of injecBlack tea 6.8323 caffeine 0.1670 2.386 tion, retarding hydrophobic sample comtheobromine 0.0092 0.132 ponents such as phenobarbital. The exP a i n k i l l e r 1 . 5 0 c a f f e i n e 0 . 0 7 4 6 2 .406 tent of the separation depends on the Cocoa 1.00 concentration of the micelle-forming caffeine 0.0034 0.033 agent. Therefore SDS concentrations theobromine 0.1897 1.897 were varied over a range from 20 to 100 mM. The application of very high concentrations of the surfactant is not adCreatinine is transported as a neutral component under the vised, since the charge and concentration contribute to an optimized separation conditions. Since there is no interacincrease of the ionic strength; and associated with this is tion between the creatinine and the micelle-forming agent, an increase of the current in the capillary during the sepathe compound is not retarded and passes the detector first. ration. High currents cause increased Joule heating in the One disadvantage, however, is that all other neutral comcapillary and deteriorated reproducibility. For the described ponents that do not interact with the micelles travel with separation, a concentration of 45 mM SDS was found to be the same velocity as creatinine and cause peak broadening. optimal. The addition of organic solvents reduces the ionic This makes the validation of the internal standard peak strength of the buffer system and causes a decrease of the more difficult. Phenobarbital, owing to its aromatic struccurrent in the capillary. Apart from decelerating the sepature, interacts strongly with the pseudostationary phase. ration, this improves peak resolution and reproducibility. Thereby it is retarded so strongly that it is the last compoAddition of 10% methanol results in a slight increase in nent to migrate past the detector. If comigration with migration times, better resolution, and less baseline noise. sample components can be excluded for real samples, pheFor selection of a suitable detection wavelength, the nobarbital is very well suited for use as an internal stanproperties of the analytes as well as those of the buffer must dard in the separation describe above. The calibration and be taken into account. Owing to the conjugated structure of separation of the sample were performed in a 20 mM bothe purines, the highest signal intensities are expected at rate buffer pH 8.25 with 45 mM SDS and 10% methanol wavelengths of 200 nm and below. The addition of 10% added. methanol to the buffer creates a high background signal because methanol itself absorbs in this low-wavelength reCalibration gion. Best results can be expected for detection at 214 nm. Instrumental parameters affect the separation, too: the The calibration of the purines and their analysis in real longer the capillary, the better the resolution—but the samples are carried out in the same buffer system. Since longer the analysis time. Smaller diameters of the capillary the UV absorbance patterns of the three components are yield better resolution. With the lower detection volumes, very similar, their identification by comparing the UV specthe limit of detection deteriorates. High voltages provide for tra from a diode array detector is difficult. excellent separations and short analysis time. They also can By following the described procedure for preparation cause losses in reproducibility of the system. Hence the set of standards, any additional standard additions for the purof optimal parameter values always depends on the particupose of peak identification can be avoided. This is true even lar separation problem. In our case, the concentrations of if the detection is carried out at one wavelength only. For caffeine and theobromine were high enough to permit use symmetrical peaks, as in our experiment, the peak intenof 50-µm i.d. capillaries. A potential of 17 kV applied across sity of a compound is proportional to its concentration. For a 57-cm long capillary gave optimal separation with respect very exact analysis or unsymmetrical peaks, calibration must to resolution and analysis time. be done by measuring peak area, which requires special The selection of an internal standard needs careful consoftware. In our case, by preparing standard solutions with sideration. If the compositions of the standard mixture and different concentrations for each of the components, the inthe real sample differ drastically, as in our case, or separatensity allows for identification of the peaks by correlating tion conditions require excessive procedures to be mainthem with the concentration used. To simplify the preparatained constant, the use of an internal standard is recomtion of standards, the solutions contained one component mended. For an optimal separation the standard must not in steadily increasing concentrations from 0.01 to 0.1 comigrate with any compounds of the sample. It should, mg/mL, one in decreasing concentrations from 0.1 to 0.01 however, migrate closely with the analyte of interest. We mg/mL, and one in complimentary increasing and decreasused creatinine and phenobarbital in our experiments. ing concentrations. These five solutions were injected sepa-

1128

Journal of Chemical Education • Vol. 74 No. 9 September 1997

In the Laboratory

Figure 3. Separation of a black tea extract (Darjeeling) (extraction procedure A). Buffer: 20 mM borate pH 8.25, 45 mM SDS, 10% (v/v) methanol.

Figure 4. Separation of pain killer components. For separation conditions see Figure 3.

rately for 3 s each. From a separate vial, the internal standard of constant concentration was injected in an additional injection step of 2 s. The calibration results are listed in Table 1. With sufficiently long rinsing steps after each run (at least 10 min), relative standard deviations of less that 1% can be achieved. This lets the average migration time be used to identify peaks in real samples. A prerequisite for this approach is that the samples must not be too complex, to avoid peak overlaps and shifts in migration times during the separation. If migration times have shifted, it is necessary to correlate the migration time of the standard in the calibration solution and the real samples for peak identification. If there is any doubt, addition of the standard as an additional step becomes necessary. For establishing the calibration curve (table 1, column 6) the peak areas or the peak heights can be used. For both curves good correlation coefficients (table 1, columns 4 and 5) were achieved proving that the separations yielded ideal symmetrical peaks. Therefore, in contrast to chromatographic methods, correct results can be obtained form measuring peak heights only. The concentration of the standards for the calibration curve were chosen to cover the concentration range expected for all samples. Under the reported conditions caffeine, theophylline and theobromine can be determined quantitatively in the range between 0.001 and 0.5 mg/mL.

Tannins, in contrast, are only partially extracted because they are largely insoluble in water. With increasing extraction time, part of the caffeine is bound again by the tannins and the amount of extracted caffeine decreases. Under the given conditions, the weakly charged or neutral purines were detected first in the electropherogram. This greatly facilitates their identification and quantitation, since they are separated from the multicomponent mixture of tannins. The tannic compounds have single or multiple negative charges. They are transported past the detector only because of the relatively high magnitude of the electroosmotic flow. Therefore the net migration velocity for tannins is relatively small. In tea, only caffeine and theobromine were identified. No signal was observed at the site where the theophylline peak was usually detected. From the peak intensities and peak areas, the concentrations for caffeine and theobromine were calculated to be 2.386 and 0.132 wt %, respectively. These are typical values for samples of black tea. The concentration of both compounds in tea depends not only on the botanical variety of the tea plant but also on the location of the tea plantation and the processing of the tea leafs. Therefore these concentrations can vary strongly. The concentration of theophylline is normally two orders of magnitude lower than that of caffeine. The method described here is not sensitive enough to detect the low concentrations of theophylline. An analogous procedure for sample preparation can be used to determine caffeine in coffee.

Separation and Determination of Real Samples To determine purine bases in a variety of different matrices, the sample must be subjected to special preparation schemes. These procedures are designed to remove components known to interfere with the analysis or bias the quantitation. Samples were diluted so that the concentrations of the analytes fell into the calibration range. Since the migration time of the internal standard in the analysis of real samples remained unchanged, the average migration times from the calibration procedure were used to identify the analytes. The values for the identified sample components are presented in Table 2. Black Tea Purines in black tea are almost completely extracted after only a short time in the laboratory extractor (Fig. 3).

Pain Killers Most pharmaceutical products that contain caffeine or theobromine are easily dissolved in water. The fillers usually added to the preparation are less water soluble. Therefore they can be separated from the active ingredients by simple filtration. The sample is diluted so that the concentrations of the compounds of interest fall in the range of calibration (Fig. 4). Peak identification can be accomplished via either standard addition, comparison of migration times, or comparison of UV spectra from diode array scans. In our case, the average migration times of the standards were used. The concentration of caffeine was found to be 2.406 wt %, which is considerably lower than the certified concentration of 3.226 wt %.

Vol. 74 No. 9 September 1997 • Journal of Chemical Education

1129

In the Laboratory

Figure 5. Separation of cocoa components after extraction (extraction procedure B). For separation conditions see Figure 3.

Cocoa A slightly more complicated procedure for sample preparation was used to remove components such as proteins, which potentially could interfere with the analytes of interest. As a result of this the electropherograms became very simple and peak identification could be based on the average migration times of the standards (Fig. 5). In cocoa the concentration of theobromine is very high while the concentration of caffeine is low. Since both concentration levels vary little from sample to sample, the described method

1130

can be utilized to determine the cocoa content of chocolate. The values in Table 2 fall in the range expected for Dutch processed cocoa. Although the analyzed purine compounds can also be separated by HPLC (eluent: 85% sodium acetate solution [0.4%], 15% acetonitrile; flow velocity; 1.0 mL/min; column: Beckman Ultrasphere RP-C18 250 ? 4.6 mm). This experiment offers students an opportunity to compare CE and HPLC, both of which are standard instrumental techniques encountered by students during their education. It can be concluded that owing to its versatility, capillary electrophoresis is a suitable approach to analytical problems in the food and pharmaceutical industries. Variations of the separation buffer as well as optimization of the remaining separation parameters can be realized easily. Capillary electrophoresis yields reproducible results within short analysis times. Last but not least, the easy handling of this technique makes it very attractive for application in undergraduate laboratory courses. Literature Cited 1. Li, S. F. Capillary Electrophoresis: Principles, Practice and Applications (Journal of Chromatography, Library Vol. 52); Elsevier: Amsterdam, 1992. 2. Kuhn, R; Hoffstetter-Kuhn, S. Capillary Electrophoresis: Principles and Practice; Springer: Berlin, 1993. 3. Hurst, W. J.; Martin, R. A., Jr. Analysis 1993, 21, 389–391. 4. Wainright, A. J. Microcolumn Sep. 1990, 2, 166–175. 5. Atamna, I. Z.; Janini, G. M.; Muschik, G. M.; Issaq, H. J. J. Liq. Chromatogr. 1991, 14, 427–435. 6. Caslavska, J.; Hufschmid, E.; Theurillat, R.; Desiderio C.; Wolfisberg, H.; Thormann, W. J. Chromatogr. B: Biomed. Appl. 1994, 656, 219–231.

Journal of Chemical Education • Vol. 74 No. 9 September 1997