Fresenius J Anal Chem (1996) 356 : 284–287
POSTER
Axel Meyer · Tharcisse Ngiruwonsanga ·
Günter Henze
Determination of adenine, caffeine, theophylline
and theobromine by HPLC with amperometric detection
Received: 14 December 1995 / Revised: 16 March 1996 / Accepted: 3 April 1996
Abstract A relatively simple method for quantifying caf-
feine, theobromine, theophylline and adenine by HPLC
with amperometric detection was developed. A C18-col-
umn and an isocratic elution with phosphate buffer pH
3.5/methanol (90 : 10) were employed for the chromato-
graphic separation of the investigated compounds. The
optimal detection potential was +1.4 V. The limits of de-
tection were 0.4 ng for adenine, 1 ng for theophylline and
2.5 ng for caffeine and theobromine. The method was ap-
plied to the determination of these purine alkaloids in
beverages, tea, coffee and cacao. The determination was
carried out directly or after solid-phase extraction.
Introduction
The alkaloids caffeine, theophylline, theobromine and
adenine are widely distributed in plant products and bev-
erages. They belong to the group of purines which can
cause various physiological effects, such as stimulation of
the central nervous system, gastric acid secretion and di-
uresis [1]. They have also been implicated in various dis-
orders including heart disease, carcinogenesis, kidney
malfunction and asthma [2].
Several methods have been published for the analysis of
these alkaloids in a variety of foodstuffs and stimulants.
Caffeine, theobromine and theophylline were analysed by a
Kjeldahl method [3], UV-spectrometry [4], thin-layer chro-
matography [5, 6], gas chromatography [7], potentiometric
titrations [8] and HPLC [9–11]. Adenine was determined by
adsorptive stripping voltammetry [12], gas chromatography
[13], capillary zone electrophoresis [14] and HPLC [15, 16].
However, some of these methods require tedious pre-
treatments or do not allow the separation and quantitation
of different purines present in the same sample. Reversed-
A. Meyer · T. Ngiruwonsanga1 · G. Henze (Y)
Fachbereich Chemie der Universität Kaiserslautern
und Abteilung für Anorganische und Analytische Chemie
der Universität Trier, Postfach 3825, D-54228 Trier, Germany
1 Student from Ruanda
© Springer-Verlag 1996
phase HPLC with UV-detection has been successfully ap-
plied for the separation and determination of these com-
pounds in a range of foods, beverages and biological flu-
ids [9–11, 17–19]. In some cases amperometric detection
[20–22] or cathodic stripping voltammetry [16] was pre-
ferred, because of the higher sensitivity and selectivity of
these detection techniques.
For the elimination of the matrix, different techniques
such as fluid extraction [10, 23], column chromatography
[19, 20], Soxhlet extraction [11, 24, 25] or solid-phase ex-
traction [23, 27] and other sample pretreatments are nec-
essary before the determination steps.
The aim of this study is to develop a simple, versatile
and accurate HPLC-amperometric detection method for
the direct simultaneous determination of these 4 purine al-
kaloids in food products. If necessary, solid-phase extrac-
tion is applied only for the separation of the matrix. An
enrichment step will not be used, because the sensitivity
of the amperometric detector is sufficient.
Experimental
Investigated substances
Structure R R1 R2 Compound
CH3 CH3 H Theophylline
(1,3-dimethylxanthine)
H CH3 CH3 Theobromine
(3,7-dimethylxanthine)
CH3 CH3 CH3 Caffeine
(1,3,7-trimethylxanthine)
Adenine
 285
Apparatus and reagents

The HPLC equipment consisted of an on-line degasser (CMA 260;

Axel Semrau), a Knauer pump (model 64), an injection system
(Rheodyne 8125) with a 20 µl loop and a C18 column (100 × 3.2 mm,
Bioanalytical system; 3 µm). Detection was performed using an
amperometric wall jet detector with a glassy carbon electrode as
working electrode and a Metrohm 656 electrochemical detector in
connection with an electronic controller (Metrohm 641 VA detec-
tor). As reference electrode a Ag/AgCl/3 mol/l KCl electrode was
used. The outputs from the detector were recorded by a Shimadzu
CR-6A integrator.
The isocratic chromatographic separation using phosphate
buffer pH 3.5/methanol (90:10) was performed with a flow rate of
0.8 ml/min. Stock solutions of the investigated compounds
(Merck, p.a.) were prepared by dissolving 10 mg of each in 100 ml
of water. All other high-purity reagents used were products of
Merck. Solvents and sample solutions were filtered through a 0.45 Fig. 1 Hydrodynamic voltammograms for caffeine, theobromine,
µm PTFE-filter. theophylline and adenine obtained by repeated injection at differ-
ent detector potentials (for chromatographic conditions see Exper-
imental)
Procedure
The determination of theophylline and adenine could
Solid-phase extraction. A Baker-10 SPE system was used to sepa-
rate the matrix from the investigated compounds onto Baker 6 ml
be performed using a potential range of +1.3 to +1.4 V.
octadecyl-solid-phase extraction columns. The columns were pre- The curves in Fig. 1 show that the highest sensitivities for
treated by washing with 2 × 10 ml of methanol followed by 2 × 10 ml caffeine and theobromine are obtained at a detector poten-
of high-purity water and 2 × 10 ml of phosphate buffer (pH 3.5). tial of +1.5 V. For the simultaneous determination of the 4
At this stage it is important to avoid a drying-out of the column. alkaloids a detector potential higher than +1.4 V is not
The aqueous sample solutions (8 ml of sample with 2 ml of phos-
phate buffer pH 3.5) were directly sucked through the extraction useful because of the low signal-to-noise ratio. Therefore
column by a water pump vacuum and then dried under vacuum for the optimal potential selected was +1.4 V.
10 min. The purine alkaloids were eluted with 2 × 5 ml of methanol/ The detection limits were found to be 0.4 ng for adenine,
phosphate buffer (pH 5) (8:12) into a 10 ml flask and then diluted 1 ng for theophylline and 2.5 ng for caffeine and theo-
with water up to the mark. If necessary, the solutions were diluted
with water. bromine under these conditions and per injection. In com-
parison with the UV-detection the sensitivity of the amper-
Direct determination in beverages. The fluid samples were de- ometric detection is about 2–5 fold higher [10, 11, 23].
gassed for 30 min in an ultrasonic bath. 2.5 ml of this solution
were diluted with 20 ml of phosphate buffer (pH 3.5) and 77.5 ml
of high-purity water. This solution was injected. Solid-phase extraction
Direct determination in tea. 1 g of the sample was mixed with 0.25 g For the determination of these substances in different bev-
of MgO. This mixture was extracted (45 min) in a beaker on a erages it could be necessary to eliminate the matrix. The
heating plate with 60 ml of hot water (80–90° C) [28]. After filtra-
tion of the solution through a folded filter, the beaker was washed
with 20 ml of hot water. Both filtrates were collected with 20 ml of Table 1 Recovery rates of caffeine (7.5 µg), theophylline (150
phosphate buffer (pH 3.5) in a 100 ml flask and the volume was µg), theobromine (15 µg) and adenine (150 µg) from different vol-
adjusted with high-purity water. According to the content of umes (10, 20 and 30 ml) of spiked water samples using C18-solid-
purine, the so received solution has to be diluted before injection. phase extraction columns (for chromatographic and SPE condi-
tions see Experimental)
Determination in coffee. The extraction procedure for coffee is the
same as that for tea. 10 ml aliquots of the solution were poured into Compound Sample volume Recovery rate/
the conditioned C18 cartridge. The elution step is described above. (ml) standard deviation
(%; N = 5)
Determination in cacao. Approximately 2.5 g of cacao were ex-
tracted in a Soxhlet apparatus with 150 ml of petroleum ether [20, Adenine 10 95.4 + 3.7
24]. The dried residue was processed as the coffee samples.
20 94.8 + 4.0
30 35.7 + 6.7
Results Theobromine 10 87.9 + 4.1
20 82.8 + 3.6
Hydrodynamic voltammograms 30 81.2 + 4.3
Theophylline 10 94.6 + 3.8
For selecting the optimal conditions for the amperometric 20 94.9 + 4.5
detection of the four purine alkaloids it is necessary to 30 45.1 + 7.2
record their hydrodynamic voltammograms. The peak
Caffeine 10 87.1 + 4.6
heights were measured as a function of the detector po- 20 82.9 + 3.2
tential in the range of +1.1 V to +1.5 V. The resultant hy- 30 61.3 + 8.1
drodynamic voltammograms are shown in Fig. 1.
286

most efficient technique for this purpose is the solid-phase Table 2 Determinations of caffeine, theophylline, theobromine
extraction. Therefore, recovery rates must be studied with and adenine in beverages and foodstuffs (results are the average of
triplicate determinations)
different analyte contents and sample volumes. Different
amounts of the four purine alkaloids were added to 10, 20 Sample Caffeine Theo- Theo- Adenine
or 30 ml phosphate buffer (pH 3.5) and then extracted us- (%) bromine phylline (%)
ing a octadecyl-phase sorbent. Recovery rates between (%) (%)
82% and 95% were obtained, if the sample volume was Cola beverages
smaller than 20 ml (Table 1). If it was too large, the four Coca-Cola 0.0991 – – –
purine alkaloids were not adsorbed. Using a sample vol- Pepsi-Cola 0.0102 – – –
ume of 10 ml, all compounds could be recovered with ap- Pepsi-Light 0.0105 – – –
proximately 90% efficiency over the concentration range Kinder-Cola 0.0001 – – –
investigated from 50 ng·ml–1 to 10 µg · ml–1. The de- Coffeinfreie-Cola 0.0002 – – –
scribed solid-phase extraction will be used only for the
Tea
isolation of the investigated compounds from the matrix Ceylon Assam 3.01 0.282 0.013 0.007
without an enrichment step. Dependent on the missing en- Indischer Tee 2.50 0.201 0.007 0.008
richment step, the detection limits are the same as in the Ruandischer Tee 2.58 0.197 0.010 0.005
case of the spiked water samples. Pfefferminz Tee 0.01 0.054 0.003 0.010
Coffee
Application to the analysis of beverages Jakobs-Krönung 2.93 0.009 – 0.004
Hanseaten Kaffee 2.54 0.008 – 0.001
Ruandischer Kaffee 2.32 0.008 – 0.005
The four purine alkaloids can be determined without
solid-phase extraction in cola beverages and in tea sam- Cacao
ples. Chromatograms A and B in Fig. 2 show that the am- Cebe 0.31 1.82 0.001 0.005
perometric detection is very selective. No influence of the Sarotti 0.43 1.58 0.003 0.009
matrix is observed. In cola beverages only caffeine can be

determined; in the tea samples all alkaloids are detectable

simultaneously in different contents.
In coffee the determination of adenine and caffeine is
possible without the solid-phase extraction step. The
Tb
elimination of the disturbing matrix components allows a
A Ca
Ad B sensitive determination of the low levels of theobromine
simultaneously with caffeine and adenine after solid-
phase extraction (chromatogram C in Fig. 2). Theo-
Tp phylline was not detectable in the samples investigated.
For the determination of the alkaloids in cacao samples
a Soxhlet extraction with petrol ether is necessary before
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16
the solid-phase extraction to eliminate the matrix. Chro-
t [min] t [min] matogram D in Fig. 2 demonstrates the high selectivity of
Ca
the amperometric detection and the high efficiency of the
separation. There are only a few other peaks in the chro-
matogram in contrast to the chromatogram C. In Table 2
Tb the results of the determination of the four purine alka-
Ad C D Ca loids in the different food products are summarized.

Tb Ad
References
Tp
1. Graham DM (1978) Nutr Rev 36 : 97
2. Turnball AJ (1981) Br Nutr Found-Nutr Bull 6 : 153
3. Horwitz W (1980) Methods of analysis. A.O.A.C. Association
of official analytical chemists, Washington, DC 13th edn
0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 4. Newton JM (1979) J Assoc Off Anal Chem 62 : 705
t [min] t [min] 5. Senanayke UM, Wijeskera ROB (1971) J Sci Food Agr 22 : 262
6. Harlos H (1973) Lebensm Ind 20 : 412
Fig. 2 A–D Determination of caffeine (Ca), theophylline (Tp), 7. Strahl NR, Lewis H, Fargen R (1977) J Agr Food Chem 25 :
theobromine (Tb) and adenine (Ad) in different food products. A 233
Cola-light; B Ceylon-Assam tea; C Coffee: Jakobs-Krönung; D 8. Greenhow EC, Spencer LE (1973) Analyst 98 : 98
Cacoa: Sarotti (for chromatographic and SPE conditions see Ex- 9. Gennaro MC, Abrigo C (1992) Fresenius J Anal Chem 343 :
perimental) 523

10. Abuirjeie MA, El-Din MS, Mahmoud II (1992) J Liq Chro-
matogr 15 : 101
11. Leakey TEB (1990) J Chromatogr 507 : 199–220
12. Zhao X, Jin W, Chen J, Gao Z, Wang F (1994) Anal Chim Acta
289 : 163
13. Guo Y, Ren Q, He YZ (1993) Sepu 11 : 27
14. Lee KJ, Heo GS (1991) J Korean Chem Soc 35 : 219
15. Fujimori H, Sasaki T, Hibi K, Senda M, Yoshioka M (1990) J
Chromatogr 515 : 363
16. Lu ZP, Liu JY, Ni ZM (1993) Fenxi Huaxue 21 : 527
17. Dulitzky M, De La Teja E, Lewis HF (1984) J Chromatogr 317 :
403
18. Hurst WJ, Snyder KP, Martin RA (1985) J Chromatogr 318 :
408
19. Terada H, Sakabe Y (1984) J Chromatogr 291 : 453
287
20. Sontag G, Kral K (1980) Mikrochim Acta 1980 II : 39–52
21. Greenberg MS, Mayer WJ (1979) J Chromatogr 169 : 321
22. Arai T, Kubo H, Kinoshita T, Nakazawa H, Fujita M (1986)
Bunseki Kagaku 35 : 263
23. Muhtadi FJ, El-Hawary SS, Hifnawy MS (1990) J Liq Chro-
matogr 13 : 1013
24. Mills GC (1982) J Chromatogr 242 : 103
25. Lehmann G, Neumann B (1974) Z Lebensm Unters-Forsch 155 :
348
26. Wildanger W (1976) Dtsch Lebensm Rdsch 72 : 160
27. Kiefer BA, Sheeley RM, Hurst WJ, Martin RA (1983) J Liq
Chromatogr 6 : 927
28. Hagenauer-Hener U, Hener U, Dettmar F, Mosandl A (1989)
Kontakte 24