1258 JOURNAL OF THE AOAC (Vol. 59, N o .

6, 1976)

C O F F E E AND TEA
High-Pressure Liquid Chromatography of Caffeine in Coflfee
BRYAN L. MADISON, WILLIAM J. KOZAREK, and CECILIA P. DAMO
Procter & Gamble Co., Winton Hill Technical Center, Cincinnati, OH

A new method is described for the deter- signal was plotted by Du Pont Instruments Model

Downloaded from https://academic.oup.com/jaoac/article/59/6/1258/5710592 by guest on 04 November 2020

mination of caffeine in coffee, based on high- 846 recorder operated at 10 mv full scale.
pressure liquid chromatography. The caffeine (b) HPLC column packing.—Zipax® SCX
is extracted from the sample with water and/or (strong cation exchange resin) (Du Pont).
methylene chloride, and then separated from (c) Mechanical shaker.—Burrell Model 75.
interfering materials by passing an aliquot of
the extract through a high-pressure column Reagents
containing sulfonated cation exchange resin, (a) Nitric acid.—0.01M. Prepare with boiled
using O.OlAf nitric acid as the mobile phase. and cooled water.
An ultraviolet detector measures the absorption (b) Caffeine standard solution.—2.0 mg/ml.
of the solution directly. The method is rapid Weigh 1 g anhydrous caffeine (Matheson Coleman
and eliminates the lengthy separations common & Bell No. CX 80-5013), dissolve in ca 250 ml
to other methods. The procedure was applied hot water, cool, and dilute to 500 ml with water.
successfully to decaffeinated and non-decaf-
feinated green, roasted, and instant coffees. Preparation of Sample
This method gives a more accurate measure of (a) Non-decaffeinated coffee.—Accurately weigh
the caffeine content in decaffeinated coffee 2 g either regular coffee or green coffee or 1 g
samples than the micro Bailey-Andrew and regular instant coffee into 125 ml Erlenmeyer
modified Levine methods, with equal or better flask. Add ca 50 ml water, heat to boiling, and
precision. This method gives results equal to boil > 5 min, shaking occasionally. Cool to room
those obtained by the official methods for non- temperature. Filter directly into 100 ml volumetric
decaffeinated samples. flask. Wash filter paper thoroughly with water.
Dilute to volume with water and mix.
The need for a rapid and precise general (b) Decaffeinated coffee.—Accurately weigh 2
purpose method for the determination of caffeine g regular coffee or 2 g green or 1 g regular instant
coffee into 125 ml Erlenmeyer flask. Add 5 ml
in coffee has been recognized. Generally ac-
NH 4 0H solution (1+2). Swirl flask until sample
cepted methods are shown in Table 1 along is wetted or dissolved (regular instant coffee) after
with an indication of the quality of results mixing. Add 50 ml reagent grade CH2C12 from
usually obtained. In addition, these methods are volumetric pipet, and stopper flask. Place on me-
lengthy and complex, require large quantities of chanical shaker and extract regular roasted and
solvents, and are not general purpose. Recently, green coffee samples 15 min or regular instant
high-pressure liquid chromatographic (HPLC) coffee 2 min. Transfer contents to 125 ml sepa-
methods (5-9) have demonstrated the potential ratory funnel and let 2 layers separate. Drain
lower layer (CH2C12) through powder funnel, with
of determining caffeine in analgesic drugs and
aqueous beverages. This paper reports the ap-
plication of an HPLC system to the separation Table 1. Generally accepted methods for the
determination of caffeine"
and determination of caffeine in coffee.
Non-
METHOD decaffein- Decaffein-
Method ated coffee ated coffee
Apparatus
(a) High-pressure liquid chromatograph. — Micro Bailey-Andrew (1) = + +
Levine (2) ++ =
Laboratory-assembled from Du Pont Instruments Modified Levine (3) + =
Model 480 connected sequentially to Valco Model ISO (4)
CV-6-HPax high-pressure injection valve, 4 mm X HPLC
30 cm stainless steel column, and Du Pont In-
struments Model 835 ultraviolet monitor fitted ° =, Accepted as yielding correct (true) value; + + ,
tends to give high results; +, tends to give variable
with 254 nm photocell/filter assembly. Detector results (less precise).
MADISON ET Ah.: HPLC OF CAFFEINE IN COFFEE 1259

stem loosely packed with cotton, and collect in (b) Decaffeinated coffee samples.—
50 ml beaker. Transfer 30 ml aliquot of caffeine-
Caffeine, % = (fig caffeine in 10 id X 105 (d
containing extract to 100 ml beaker. Evaporate
X 10"6 g/pg X 100)/[g sample weight
to dryness on steam bath with nitrogen. Add 5 ml
water and warm on steam bath to dissolve caf- X 30/50 (aliquot of CH2C12)]
feine. Transfer to 10 ml volumetric flask and dilute Results and Discussion
to volume with water.
Reverse phase, high-pressure liquid chromato-
Determination graphic separations of caffeine were investigated
with a /iBondapak/Cls column and by cation

Downloaded from https://academic.oup.com/jaoac/article/59/6/1258/5710592 by guest on 04 November 2020

Using loop injector, place 10 /tl sample onto
column with 0.01M H N 0 3 eluant flowing at 4.0 exchange with a Zipax SCX column. Zipax SCX
ml/min. Measure peak height for quantitation was preferable because of chromatographic time
and compare against standard calibration curve. and reproducibility.
The reverse phase separations we examined
Standardization used mobile phases of acetic acid-water, meth-
Weigh 1.0 g regular instant decaffeinated coffee anol-water, acetonitrile-water, and nitric acid-
into 4 separate 100 ml volumetric flasks. Add 50 water. Dry-packed Zipax SCX columns were
ml water and heat on hot plate to completely conditioned by pumping the mobile phase
dissolve sample. Pipet into 3 of the volumetric through the column for 2 hr prior to use. Pre-
flasks, respectively, 5.0, 15.0, and 25.0 ml caffeine dictably, the extent of separation of caffeine on
standard solution to give 0.1, 0.3, and 0.5 pig caf-
these ion exchange packings was determined
feine/jul solution. Remaining solution serves as a
blank. Cool, dilute to volume with water, and mix more by the pH of the mobile phase than by
thoroughly. Inject 10 /d standard onto column as any other parameter. By decreasing the pH, the
in Determination. Measure peak height of each time of chromatography and resolution of the
standard. Plot ng caffeine in 10 /A vs. peak height caffeine increased. The optimum pH was about
to prepare calibration curve. 2, and the mobile phase chosen was O.OlAf
HNO3. The flow rate was adjusted to 4 ml/min,
Calculations which eluted the caffeine in less than 6 min.
(a) Non-decaffeinated coffee samples.— Integration by peak height gave more repro-
Caffeine, % = (/ig caffeine in 10 /d X 105 /d ducible answers than electronic integration be-
X 10-6 g//«g X 100)/(g sample weight) cause it is not affected by the inherent small

0% 1% 2% 3% 4%

J LJ I L J I
0 2 4 6 0 2 4 0 2 4 6 0 2 4 6 0 2 4 6
Retention Times, minutes
FIG. 1—Chromatograms of varying caffeine levels added to soluble decaffeinated coffee.
1260 JOURNAL OF THE AOAC (Vol. 59, N o . 6, 1976)

0.136 %
Caffeine

Downloaded from https://academic.oup.com/jaoac/article/59/6/1258/5710592 by guest on 04 November 2020

0.047%
Caffeine

without with without with

preconcentration preconcentration preconcentration preconcentration
Decaffeinated Decaffeinated
Roasted Coffee Instant Coffee
FIG. 2 - Chromatograms showing effects of preconcentrating caffeine with methyiene chloride.

variations in flow rate which affect the residence obtained by HPLC agree with these accepted
time in the detector and thus the area inte- methods.
grated. The coefficients of variation reported in Table
The chromatograms obtained of caffeine-free 3 show that the HPLC method is more precise
coffee fortified with known amounts of caffeine than the current methods now used for the
are shown in Fig. 1. A caffeine-free coffee was determination of caffeine in coffee products.
used to approximate the sample matrix. Non-
decaffeinated materials are easily analyzed. Caf- Table 2. Comparison of 3 methods for
feine is extracted from both roasted coffee and per cent caffeine
green coffee beans with boiling water, whereas Bailey-
instant coffee samples are dissolved directly in Andrew
hot water before a small portion of the sample Samples micro" HPLC6 ISO"
is introduced onto the HPLC column. Decaf- Non-decaffeinated Samples
feinated coffee samples require a preconcentra-
Green coffee
tion step before HPLC analysis. Caffeine is Robusta 1.91 1.91 1.95
extracted from an NH 4 0H (1+2) solution, the Arabica 1.27 1.27 1.24
methyiene chloride extract is evaporated to dry- Roasted coffee
Robusta 1.98 2.02
ness, and the residue containing caffeine is dis- Arabica 1.30 1.31
solved in water and introduced onto the HPLC Instant or soluble
3.52 3.53
column. Typical chromatograms for decaffein- 3.32 3.26
ated coffee and the increased sensitivity which
resulted from preconcentration are shown in Decaffeinated Samples
Fig. 2. The peak height is more accurately meas- Green coffee 0.057 0.066 0.055
ured, and the slope of the baseline decreased. 0.034 0.028 0.028
The HPLC method, the official AOAC meth- Roasted coffee 0.047 0.048
0.031 0.034
od (1), and the proposed ISO method (4) were Instant coffee 0.179 0.186
compared for 12 different coffee samples. The 0.140 0.136
data obtained are shown in Table 2. All values " Average result for 4 laboratories.
6
Average result for 1 laboratory, 5 determinations.
Received February 9, 1976. ° Average result for 8 laboratories.
MADISON ET AL.: HPLC OF CAFFEINE IN COFFEE 1261

Table 3. Comparison of 4 methods for per cent caffeine and precision of results
Bailey-Andrew micro Mod. Levine ISO HPLC

Coeff. of Coeff. of Coeff. of Coeff. of

Year of study Av. var., % Av. var.,% Av. var., % Av. var., %

Roasted and Ground Coffee

1959 1.32 2.3

1964 1.43 9.1
1975 1.31 5.1 1.30 1.8

Downloaded from https://academic.oup.com/jaoac/article/59/6/1258/5710592 by guest on 04 November 2020

Regular Soluble Coffee

1960 3.10 2.6

1964 3.17 7.6
1975 3.26 5.7 3.32 2.5

Decaffeinated Soluble Coffee

1960 0.12 12.5

1964 0.080 6.9
1975 0.186 3.2 0.179 1.7

The linearity of the detector response was REFERENCES

established from the analytical calibration (1) Official Methods of Analysis (1975) 12th Ed.,
curves and begins to deviate from linearity at AOAC, Washington, DC, sees. 15.022-15.025
about 10 jxg caffeine injected onto the HPLC (2) Levine, J. (1962) JAOAC 45, 254-255
(3) Borker, E. (1963) JAOAC 46, 319-320
column. The lower limit of caffeine concentra-
(4) Vitzthum, O. (1974) International Standards
tion that can be quantitatively analyzed by Organization, Document ISO/TC 34/SC 8/WG
this method is 0.01% or about 0.001 fig caffeine 2 (Ser 91) 233E of 1974-06-20
injected. (5) Murgia, E., Richards, P., & Walton, H. F.
The proposed HPLC method permits com- (1973) J. Chromatogr. 87, 523-533
plete elimination of organic solvents in non-de- (6) Du Pont Liquid Chromatography Methods
caffeinated coffee samples and only a modest Bulletin (1970) 820M5, E. I. du Pont de
use of a relatively innocuous organic solvent for Nemours & Co., Wilmington, DE
(7) Ascione, P. P., & Chrekian, G. P. (1975) J.
decaffeinated coffee samples. The method is
Pharm. Sci. 64, 1029-1033
more rapid than existing methods, 0.3 hr vs. (8) Nelson, J. J. (1973) J. Chromatogr. Sci. 11,
1.5-2.5 hr for the current or proposed methods. 28-35
As many as 35 samples have been analyzed in a (9) Smyly, D. S, Woodward, B. B., & Conrad,
single working day. E. C. (1976) JAOAC 59, 14-19