Rapid Spectrophotometric Determination of

Caffeine
N. H. ISHLER, T. P. FINUCANE, AND EMANUEL BORKER
Central Research Laboratories, General Foods Corporation, Hoboken, N. J.

The characteristic absorption of caffeine at 272 mg the new method compare favorably with the Bailey-
is utilized to measure quantitatively its presence in Andrew procedure which is used as a reference for
coffees and crude caffeine. Interfering impurities comparison. Detailed precision studies showing
found in these samples are removed by treatment the new method to be equal to or better than the
with heavy magnesium oxide and zinc ferrocyanide, Bailey-Andrew method are reported. Decaffeinated
plus in some cases permanganate oxidation. Rapid- coffees will require a modification of the methods
ity and specificity for caffeine are outstanding herein described, which will probably involve a
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

characteristics of the method. Results obtained by solvent extraction step.

Downloaded via UNIV OF THE PHILIPPINES on November 21, 2019 at 14:27:32 (UTC).

operation of a coffee decaffeinating plant requires a great 1. =

length of light path through absorbing medium
THE
number of analytical caffeine determinations. For a number c
E
=

—
concentration
extinction coefficient
of years, the A.O.A.C. official Power-Chesnut (2) method was
used in this laboratory. This method was satisfactory, but For quantitative estimation of caffeine, an accurate deter-
excessively time-consuming for control operations. In 1943, mination of or caffeine factor (optical density per mg.
the Bailey-Andrew method (3), then official only for tea, was
per 100 ml.) was mandatory. Optical densities [log (100/%
successfully adapted for use here by increasing the specified transmittance)] may be read directly from the scale of the
amount of magnesium oxide to 25 grams and by making suitable Beckman spectrophotometer. Further, it was desired to estab-
changes in sample weight to permit its application to decaffein- lish the optimum concentrations of caffeine for reading in this
ated products and solutions as well as to green or roasted coffee. instrument.
This method has since been made official for caffeine in coffee
(1). The procedure, although more rapid than the Power- Fifty observations were made to evaluate the caffeine factor.
U.S.P. caffeine was sublimed and the sublimate dried in a vac-
Chesnut, still required an elapsed time of at least 7 hours per uum oven at 700 C. for 16 hours. From a solution of the sub-
-
determination. Even when the greatest number of these analy- limed caffeine, aliquots were taken to provide concentrations
ses was being performed, it was never possible in this labora- from 0.2 to 2.6 mg. of caffeine per 100 ml. Density values on the
tory to perform the many manipulations in much less than 1 samples were then read in the spectrophotometer at 272 mg at a
fixed slit width of 0.74 mm., which, according to the manufac-
man-hour per sample.
turers, corresponds approximately to a 2 mg band width. At this
In 1946, an investigation into the possibility of measuring narrow setting it is easy to detect any slight shifting of the ab-
concentration of caffeine by utilizing its known absorption char- sorption maximum of caffeine, an indication that impurities are
acteristics in the ultraviolet range was begun. present. Averages of the data are presented in Table I. The
factor that best fits the data is 0.510 density unit per mg. of
Absorption of light by caffeine was reported as early as 1905 caffeine per 100 ml. of solution.
by Hartley (8, 9). In 1919, Henri (10) reported that the ab-
sorption spectrum of caffeine had been studied. Castille and All measurements were made in matched quartz absorption
Ruppol (4) in 1928 reported quantitative measurements on cells of 1-cm. light path. As an additional precaution the same
caffeine indicating an absorption maximum at 271 mg. Their cell was always used as a blank while another was reserved for
work was substantially confirmed by others (7, 11, 14), who
the solution to be examined. (The same procedure was applied
reported the caffeine maximum between 271 and 275 mg at differ-
ing pH. Adherence to Beer’s law was generally indicated. Molec- in the analytical examination of unknown solutions, as even in
ular extinction coefficients ranging from 8000 to 11,530 were re- carefully matched cells observable differences can be encoun-
ported by these workers.

EXPERIMENTAL
Table I. Determination of Caffeine Factor
A model DU Beckman spectrophotometer with ultraviolet
Caffeine, Average Average
accessories was used for all the absorption measurements pre- Mg./100 Ml. Optical Density Caffeine Factor0
sented in this report. An aqueous solution of sublimed and 0.257 0.131 0.511
vacuum-dried caffeine was examined with this instrument. 0.498 0.262 0.525
0.506 0.265 0.524
The findings of earlier investigators (4, 7, 11, 14) were substan- 0.514 0.263 0.512
0.747 0.384 0.514
tially confirmed; caffeine was found to exhibit a maximum ab- 0.760 0.390 0.514
0.770 0.396 0.515
sorption at 272 and a minimum at 245 millimicrons. Variation 0.996 0.512 0.514
of pH between 5 and 9 has not been found in the authors’ work 1,013 0.518 0.511
1,027 0.523 0,509
to affect either the extinction or the absorption maximum of 1.245 0.633 0.508
1.266 0.643 0.508
aqueous caffeine solutions. 1.284 0.650 0.505
It was necessary to ascertain the conformance of caffeine 1.494 0.760 0.509
1.519 0.768 0.506
solutions to the Beer-Lambert law. 1.541 0.775 0.505
1.798 0.903 0.503
log I/Id =
Ele (1) 2.054
2.311
1.04
1.16
0.508
0.501
where 2.568 1.29 0.501
I =
intensity of light transmitted by solvent plus solute °
Density/mg./100 ml. solution.
Id =
intensity of light transmitted by solvent
1162
VOLUME 2 0, NO. 12, DECEMBER 1948 1163

ml. of 1.0 M zinc acetate solution and 6.0 ml. of 0.25 M potassium
Table IT. Spectrophotometric Recovery of Caffeine from ferrocyanide were used to prepare zinc ferrocyanide for these
Zinc Ferrocyanide Clarification clarifications. All clarifications were made at a constant tem-
(Temperature of solutions 20.5° C.)
C oncentration. Caffeine Recovered,
perature. Table II shows quantitative recovery below 2.5 mg.
Mg./100 Ml. % of caffeine per 100 ml.
1.0 101.0 The observation had been made in earlier work that the
2.5 97,8
5.0 94.1 reagents used in the clarification step exhibited some slight
10.0
25,0
92.9
92.2
absorption at 272 my, but at that time the readings were con-
sidered negligible because of the low densities. In order to
determine whether a reagent blank was a compensating factor
Table III. F,fleet of Reagent Blanks upon Recovery of for caffeine absorbed, aliquots were taken from a standard solution
Caffeine of caffeine for samples and controls to be made up volumetricallv
Caffeine, Mg./100 Ml.
Densitv Spectropho- with the filtrate from the preparation of zinc ferrocyanide and
at 272 µ tometer Theoretical with distilled water, respectively. The results in Table III
Distilled water (in sample indicate that while the zinc ferrocyanide by itself gave an optical
cell) 0.016
Filtrate from zinc acetate
0.054
density reading of 0.054 it did not contribute any error to the
ferrocyanide
Control caffeine distilled density of the caffeine solution prepared as above which was
0.518 1.02 1,03 almost identical with the density of the caffeine-distilled water
Caffeine + zinc ferro-
cyanide filtrate 0.521 1.03 1.03 control. This indicates that reagent blanks cannot be read
directly and corrected for by simple subtraction because of the
Table IV. Spectrophotometric Recovery of Caffeine from inaccuracy of very low density readings. In this case, it is
Magnesium Oxide Clarification established that light absorption by zinc ferrocyanide does not
Concentration, Caffeine Recovered, introduce an appreciable error to the density reading of a caf-
Mg./100 Ml. % feine solution containing approximately 1 mg. per 100 ml.
1 100.5 The recovery of pure caffeine at different levels of concentra-
99.9
10 98.3 tion from magnesium oxide clarification was also checked and
50 100.5
found to be quantitative spectrophotometrically in concentra-
tions from 1 to 50 mg. of caffeine per 100 ml. of solution (Table
IV). Five grams of magnesium oxide were used for each 100
tered.) The precision of the method depends in part upon the ml. of solution.
reproducibility of optical density readings, and it is felt to be of Zinc ferrocyanide followed by a magnesium oxide clarification
vital importance to utilize all the precision which the spectro- failed to remove all interfering substances in roasted coffee.
photometer affords. The absorption maxima of the resulting solutions occurred be-
Each independent investigator will need to establish his own tween 274 and 276 my instead of at 272 my. Additional clari-
factor for density in terms of caffeine concentration to allow for fication steps involving mercuric nitrate, ammonium sulfide, or
differences in instrument and particularly for differences in neutral lead acetate were also found ineffective. Attempts to
transmittance characteristics and light-path distances in sample reduce the interfering material with sodium sulfite, stannous
cells. chloride, or potassium thiosulfate were unsuccessful. Oxidation
As Table I shows, caffeine solutions conform to the Becr- by iodine, dichromate, ceric salts, perchlorate, or hydrogen per-
Lambert law, at least within the limited range covered. Thus oxide also failed.
it is feasible to determine caffeine quantitatively in the spectro- The A.O.A.C. official Fendler-Stüber method (1, 6) for caffeine
photometer if there is a means of completely removing all sub- in coffee uses potassium permanganate as one purification step
stances in coffee extracts and crude caffeine samples which adsorb in the procedure. This reagent has been used by others (13).
light at 272 my. Therefore, the first step in the solution of the When tried in conjunction with zinc ferrocyanide and magnesium
problem was the discovery of highly selective clarifying agents to oxide, encouraging results were obtained. Table V shows that
remove the interfering substances. Samples of crude caffeine and different sequences of reagents produced the characteristic caf-
green coffee extracts were used in the first studies to find these feine maximum at 272 my in the final solution. In each case
clarifying agents. the permanganate color wras discharged after 10 minutes, when
Heavy magnesium oxide, an effective agent used in several sodium sulfite and glacial acetic acid were used. No reducing
current methods, was inadequate as the sole clarifier for spectro- agent other than sodium sulfite was found satisfactory.
photometric analysis. Other commonly used clarifying agents
(15), such as zinc ferrocyanide, neutral lead acetate, ferric hy-
droxide, and alumina cream, after an initial magnesium oxide Table V. Clarification by Zinc Ferrocyanide, Magnesium
treatment of the samples, proved ineffective for complete re- Oxide, and Potassium Permanganate in Various Sequences
moval of interfering substances. (Roasted coffee extract, Bailey-Andrew, 5.59% caffeine)
Apparent Maximum
The order of clarification was reversed with encouraging re- Caffeine, Absorption,
sults. All four of the clarifiers displayed greatly improved ability Order of Sequence of Reagents %
to remove interfering substances, but only zinc ferrocyanide Zn2Fe(CN)e, KMnOifl 6.27 272
Zn2Fe(CN)e, MgO, KMnO-C 5.78 272
successfully removed all interferences from the solutions when MgO, Zn2Fe(CN)6, KMnOi“ No ultraviolet
transmission
followed by magnesium oxide treatment. MgO, KMnOi, Zn2Fe(CN)6a 5.36 272
Using solutions of some compounds known to be present in a
KMnOt treatment includes subsequent reduction with Na2S03.
coffee, the specificity of the clarifiers was determined. Mag-
nesium oxide was found to be an effective reagent for the removal
of chlorogcnic acid, while zinc ferrocyanide removed trigonelline. I.epper (13) has shown that there is a very slight loss of caffeine
In unextracted green coffee the caffeine content is about 1%, with permanganate oxidation at room temperature, and that
trigonelline 1.5%, and chlorogenic acid about 7%. the loss is minimized when the treatment is made at low tem-
The recovery from different concentrations of pure caffeine perature.
after zinc ferrocyanide clarification in 100-ml. volumetric flasks To determine the recovery of pure caffeine from permanganate
was determined. As described under Analytical Methods, 7.0 oxidation at room temperature, a standard solution was prepared.
1164 ANALYTICAL CHEMISTRY

Table VIII, recovery of caffeine from the mixtures was quantita-

Table VI. Spectrophotometric Recovery of Caffeine from tive.
1% Potassium Permanganate Oxidation and Reduction of ANALYTICAL METHODS
KMnOj with 5% Sodium Thiosulfate Reagents. Magnesium oxide; heavy MgO, U.S.P.
(Temperature 25° C.) Potassium ferrocyanide solution, 0.25 molar.
Acetic Acid Added Buffered zinc acetate solution, 1.0 molar. Dissolve 438.0 grams
After Initial Prior to of zinc acetate dihydrate in distilled water, add 60 ml. of glacial
Excess 5% Na2SOj Reduction Initial Reduction
acetic acid, and make to 2000 ml.
Solution, Ml. Caffeine recovered, neutral solution
Sulfuric acid, 0.1 N.
% % Potassium permanganate solution, 1%.
0.0 98,4 97.7 Glacial acetic acid.
0.2 97.4 96.3
0.5 97.1 96.3 Sodium sulfite solution, 5%.
1.0 97.9 94.2 Procedure for Crude Caffeine and Green Coffee. Preparation
2.0 99.5 95.3 of Sample. Crude Caffeine. Weigh 0.1 gram of sample, transfer
5.0 103.9 99.9
10.0 116.4 112.6 to a volumetric flask (500 ml.), add 400 ml. of hot water, and
shake thoroughly. Cool the volumetric flask to room tempera-
ture and make to the mark with distilled water. Pipet a con-
venient aliquot containing approximately 2 mg. of caffeine to a
One-milligram aliquots were treated in 100-ml. volumetric flasks 100-ml. volumetric flask. Add distilled water to a total volume
with 10 ml. of 1% potassium permanganate solution for 15 of approximately 50 ml.
minutes. Treatment was in neutral solution and in solutions to Green Coffee. Weigh 2.0 grams of sample and transfer to a
which 5 or 10 ml. of 0.1 N sulfuric acid had been added. In the tared 1-liter Erlenmeyer flask. Add 50 ml. of 0.1 N sulfuric acid
acid solutions, reduction to the manganous salt was made directly and 450 ml. of distilled water to the flask, heat to boiling, and boil
by approximately 5 ml. of 5% sodium sulfite solution. In the 30 minutes. Cool flask to room temperature, make to weight
neutral solutions one series was reduced directly with sodium
sulfite after the addition of 0.5 ml. of glacial acetic acid; another (tare plus 502.0 grams), and filter. Pipet a 50-ml. aliquot into a
100-ml. volumetric flask.
series was reduced by the sulfite solution to manganese dioxide
initially, and then after adding 0.5 ml. of acetic acid, reduction
was continued to the manganous salt. In the neutral solutions, Table VIII. Spectrophotometric Recovery of Caffeine
varying excesses of 5% sodium sulfite solution were added. from Artificial Mixtures
There was considerable loss of caffeine in the solutions con- Caffeine
Sample Calculated Caffeine Recovered
taining acid permanganate. Table VI indicates that satis- Mg./100 ml. Mg./100 ml. %
factory recovery of caffeine is obtained from the oxidation with Artificial soluble coffee 1.00 0.997 99.5
Semiartificial soluble coffee 1.05 1.06 100.4
1% permanganate solution in neutral solution, followed by initial
reduction in neutral solution and further reduction of manganese
dioxide in acid solution. It is also apparent that a considerable Ferrocyanide and Magnesium Oxide Clarification. To
excess of sodium sulfite solution will not affect this quantitative the volumetric flask add 7.0 ml. of zinc acetate solution and swirl
recovery. This work confirms the observations of Leppcr (IS) vigorously. Dropwise add 6.0 ml. of potassium ferrocyanide
solution with constant swirling of the flask. (An excess of zinc
concerning the loss of traces of caffeine with permanganate oxida- acetate must be present to prevent interference by the ferro-
tion. cyanide.) Make to the mark with distilled water. Mix thor-
A study was made of the effect of the length of permanganate oughly, filter through a No. 41 (15-cm.) Whatman filter paper,
oxidation time on two samples of a commercial brand of coffee. and discard first 10-ml. portion. Pipet a 50-ml. portion into a
tared 250-ml. Erlenmeyer flask containing 5.0 grams of mag-
The caffeine value obtained spectrophotometrically with oxida- nesium oxide, add 50 ml. of distilled water, and boil for 20
tion times from 0.5 to 24 hours was compared to the values minutes. Cool to room temperature and make to weight (tare
obtained from an oxidation time of 10 minutes. As shown in plus 105 grams). Filter and discard first portion (10 ml.). Read
a portion of the filtrate in the spectrophotometer at 272 µ.
Table VII, there is no appreciable loss of caffeine when the
Density reading at 272 µ divided by the factor previously deter-
permanganate oxidation is less than 1 hour. This oxidation mined equals milligrams of caffeine per 100 ml. of filtrate.
occurred in alkaline solution as a result of the initial magnesium Time for crude caffeine, 1.5 hours. Time for green coffee, 2
oxide clarification. hours.
Procedure for Green, Roasted, and Soluble Coffee. Prepara-
tion of Sample and Magnesium Oxide Clarification. Green
Table VII. Effect of Potassium Permanganate Oxidation and Roasted Coffee. Weigh 1.0 gram of roasted coffee or 1.2
Time on Caffeine Content of Roasted Coffee Solutions grams of green coffee and transfer to a tared 1-liter Erlenmeyer
flask containing 50 ml. of 0.1 N sulfuric acid. Add 250 ml. of
Caffeine Recovery Compared to 10-Minute distilled water and boil for 20 minutes. Then add 50 grams of
Oxidation Time
Oxidation Sample A. Sample B magnesium oxide and continue boiling for 20 minutes. Cool and
make to weight (tare plus 350 grams plus weight of sample).
.

Hours % %
0.5 99.2 100.8
Filter, pipet a 25-ml. aliquot into a 100-ml. volumetric flask, and
1.0 97.4 100.4
add 25 ml. of distilled water.
2.0 95.5 95.6 Soluble Coffee. Take a sample weight containing approxi-
3.0
4.0
95. o
89.3
94.8
86.9
mately 100 mg. of caffeine and transfer to a 500-ml. volumetric
24.0 56.7 53.2
flask. Make to volume, mix thoroughly, and pipet a 25-ml. ali-
quot into a tared Erlenmeyer flask containing 25 ml. of 0.1 .V
sulfuric acid. Add 200 ml. of distilled water and boil for 20
minutes. Then add 25 grams of magnesium oxide and continue
The recovery of caffeine from artificial and semiartificial mix- boiling for 20 minutes. Cool and make to weight (tare plus 275
tures of soluble coffee solutions was determined to discover any grams). Filter and pipet a 50-ml. aliquot into a 100-ml. volu-
loss of caffeine resulting from the series of purification steps. metric flask.
Permanganate and Ferrocyanide Clarification. In the
Because it is most convenient, the sequence of magnesium oxide, volumetric flask place 10 ml. of potassium permanganate solution.
potassium permanganate, and finally zinc ferrocyanide was used After 10 minutes, add 3.0 ml. of sodium sulfite solution, then add
for the analysis. The artificial mixture was prepared from a 0.5 ml. of acetic acid, and titrate with sulfite solution to the dis-
standard caffeine solution and a solution of decaffeinated soluble appearance of the manganese dioxide precipitate. Add 7.0 ml.
of zinc acetate solution and swirl vigorously. Dropwise add 6.0
coffee, which provided the necessary coffee solids. The semi- ml. of potassium ferrocyanide solution with constant swirling of
artificial mixture was prepared from the artificial mixture and a the flask. Make to volume with distilled water. Mix thoroughly
solution of soluble coffee. The theoretical caffeine values of these and filter through No. 42 Whatman filter paper, discarding the
mixtures were calculated from the spectrophotometric deter- first portion. Read a portion of the filtrate in the spectropho-
tometer. From the density reading, calculate the per cent caffeine
mination of the standard caffeine solution and the Bailey- in the sample.
Andrew caffeine values of the two soluble coffees. As shown in Time for green, roasted, or soluble coffee, 2 hours.
VOLUME 2 0, NO. 12, DECEMBER 1948 1165

able for any given sample, as a similar extraction procedure is

Table IX. Agreement of Results used in both methods.
[Sequence of reagents, spectrophotometric method, Zn2Fe(CN)e, MgO] From Tables IX and XI, it is apparent that the agreement of
No. of Per Cent Caffeine
results between the Bailey-Andrew and the spectrophotometric
Detns. per Bailey- Spectro-
Sample Method Andrew photometric method is very good. However, in about 70% of the samples
'Crudes analyzed, the spectrophotometric caffeine value is lower.
I 10 94.8 94.2
II 10 67.7 63.8 The precision data reported are intended to be only relative
III 10 28.9 26.9 and are not necessarily limiting for either method, because com-
IV 10 13.3 12.3
A 82.9 84.0 parative runs were made simultaneously by the same analyst. The
B 2 92.5 91.0
C 41.3 41.3 data do give an excellent comparison of the two methods under
D 73.8 73.0 identical conditions. Table X demonstrates a much greater
E 2 71.1 70.0
F 2 54.3 54.0 precision for the spectrophotometric procedure involving a
G 2 85.3 85.0
H 2 73.3 73.0 zinc ferrocyanide and magnesium oxide clarification. Table
J 2 68.6 69.0 XII demonstrates that when potassium permanganate is used in
K 2 69.6 71.0
the spectrophotometric method, the precision of the Bailey-
Green coffee
Columbian, coarse grind 2 1.11 1.07 Andrew method is the same as or slightly better than the spectro-
Columbian, fine grind 2 1.19 1.14
Central American 2 1.19 1.15 photometric procedure.
Aged Venezuelan 2 1.17 1.13 The greater precision obtained spectrophotometrically with
Special Santos, fine grind 2 0.95 0.93
Special Santos, flaked 0.97 0.96 only a ferrocyanide and a magnesium oxide clarification implies
Medellin 2 1.12 1.09
Santos a correspondingly greater ac-
10 1.03 0.99
_______________curacy. Even when potassium
permanganate is used in the
Table X. Precision (26) of Ten Determinations
[Sequence of reagents, spectrophotometric method, Zn2Fe(CN)e, MgO]
spectrophotometric procedure,
% Caffeine by Bailey-Andrew Method % Caffeine by Spectrophotometric Method the average value obtained
Sample Av. Range sa L.C.=5%i Av. Range 5° L.C.m%6 spectrophotometrically is prob-
'Crude caffeine ably more accurate than the
I 94.8 91.9 -97.3 2.4 6.3 94.2 93.6 -95.5 0.59 1.5
II 67.7 65.7 -68,7 1.1 2.9 63.8 63.1 -64.6 0.55 1.4 Bailey-Andrew value. This
III 28.9 27.8 -30.6 0.87 2.3 26.9 25.9 -27.7 0.47 1.2
IV 13.3 12.2 -14.6 0.83 2.2 12.3 12,1 -12.4 0.10 0.26 assumption is based on the
•Green coffee known presence of noncaffeine
Santos 1.03 1.00- 1.04 0.012 0.031 0.99c 0.98- 1,00 0.007 0.018
Santos 1.03 1.00- 1.04 0.012 0.031 1.04* nitrogen in the Bailey-Andrew
chloroform extract (5) that is
°
Standard deviation where d deviation of each observation from sample
^n
= =
mean.
__
|, calculated as caffeine in the
b
L.C.95% =
Í0.05 S ± limits from true value within which 95 out of 100 single determinations will fall.
c
Water extracted, neutral solution final result, whereas quantita-
d Acid extracted, 5 determinations. tive recovery of caffeine is
---------
obtained spectrophotometri-
cally.
These procedures involving the direct clarifications of aqueous The greatest difference in caffeine analyses made spectro-
extracts of coffee are not directly applicable to decaffeinated photometrically and by the Bailey-Andrew method is in the time
coffee. A modification of the procedure involving an initial required per analysis. Whereas 0.5 hour per analysis and 2
solvent extraction applicable to these samples is being developed.
However, there are insufficient data at this time to present a
¡finished procedure for decaffeinated coffee. Table XI. Agreement of Results
[Sequence of reagents, spectrophotometric method, MgO, KMnO,,
COMPARISON OF METHODS ZniFe(CN),]
>qn nf Per Cent Caffeine
A number of samples of each type were analyzed by the Detns. per Bailey- Spectro-
Sample Method Andrew photometric
appropriate spectrophotometric procedure and by the Bailey- Roasted coffee
Andrew method. The results obtained are given in Tables IX, Blend I 2 1.15 1.14
Blend II 2 1.12 1.14
X, XI, and XII. West Central American 2 1.32 1.33
There is some doubt as to the completeness of the extraction Standard Columbian 1.38 1.38
2 Aged Venezuelan 1.35 1.33
of caffeine from coffee by any of these methods, particularly 2 Santos 1.21 1.20
10 Brand I 1.34 1.27
green coffee. It is known, for example, that the finer the grind 10 Brand II 1.28 1.25
the greater the extraction of caffeine and that flaking further 10 Brand III 1.28 1.22
10 Brand IV 1.16 1.17
increase the yield of extractable caffeine. The increasing yield Green coffee
10 Santos 1.03 1.03
with finer grinding and flaking is clearly demonstrated in the Soluble coffee
of Columbian and 10 Brand A 5.59 5.36
analyses 10 Brand B 3.98 3.93
special Santos samples
(Table IX). It is also known
that extraction is more effi- Table XII. Precision (16) of Ten Determinations
cient in acid or alkaline rather [Sequence of reagents, spectrophotometric method, MgO, KMnO., ZmFefCNJe]
than neutral solution. This % Caffeine by Bailey-Andrew Method % Caffeine by Spectrophotometric Method
difference in extraction is dem- Sample Av. Range s L.C.95% Av. Range s L.C.96%
onstrated in the analyses of Green coffee
Santos 1.03 1.00-1.04 0.012 0.031 1 .03 1.00-1,05 0.019 0.050
Santos samples (Table X). Roasted coffee
Brand I 1.34 1.32-1.38 0.022 0.057 1.27 1.22-1.32 0.032 0.083
Acid extraction resulted in a Brand II 1.28 1.25-1.33 0.022 0.057 1.25 1.22-1.29 0.022 0.058
caffeine value slightly greater Brand III 1.28 1.26-1.30 0.012 0.031 1.22 1.20-1.24 0.016 0.042
Brand IV 1.16 1.14-1.18 0.017 0.045 1.17 1.13-1.19 0.020 0.052
than that obtained by alka- Soluble coffee
Brand A 5.59 5.50-5,68 0.047 0.123 5.36 5.21-5.47 0.069 0.178
line extraction. However, the Brand B 3.98 3.93-4.05 0.051 0.133 3.93 3.85-4.03 0.050 0.131
methods are directly compar-
1166 ANALYTICAL CHEMISTRY

hours' elapsed time are the maximum required for a spectrophoto- (2) Assoc. Official Agr. Chem., “Official and Tentative Methods of
metric analysis, the Bailey-Andrew procedure requires 1 man- Analysis," 6th ed., 18.14, p. 217, 1945.
(3) Ibid., p. 220.
hour per analysis and 7 hours’ elapsed time. Thus the spectro- (4) Castille, A., and Ruppol, E., Bull. soc. chim. biol., 10, 623 (1928).
photometric method reduces the time required to one third that (5) Clifford, P. A., J. Assoc. Official Agr. Chem., 14, 533 (1931).
previously necessary without sacrificing precision or accuracy. (6) Fendler, G., and Stuber, W., Z. Nahr. Genussm., 28, 9 (1941).
(7) Gulland, J. M., Holiday, E. R., and Macrae, T. F., J. Chem.
Soc., 1934, 1639-44.
ACKNOWLEDGMENT (8) Hartley, W. N., J. Chem. Soc., Trans., 87, 1796 (1905).
(9) Hartley, W. N., Trans. Roy. Soc. {London), A, 176, 471 (1885).
The authors wish to acknowledge the guidance and assistance (10) Henri, V., “Etudes de Photochimie,” Paris, Gauthier-Villars,
received from L. W. Elder. Thanks are due to R. G. Moores 1919.
for having provided samples of trigonelline and chlorogenic acid (11) Holiday, E. R., Biochem. J., 24, 619 (1930).
and for having given the authors the benefit of his experience with (12) Lendrich, K., and Nottbohm, F. E., Z. Nahr. Genussm., 17, 241
(1914).
similar problems. It is also desired to express appreciation for (13) Lepper, H. A., J. Assoc. Official Agr. Chem., 4, 526 (1921).
the great number of analytical determinations made for com- (14) Loofbourow, J. R., Stimson, . M., and Hart, M. J., J. Am.
parative purposes by J. J. Kelly, G. F. Lata, E. J. Sarna, and Chem. Soc., 65, 148 (1943).
R. C. Sylvester. (15) Moir, D. D., and Hinks, E., Analyst, 60, 439 (1935).
(16) Snedecor, G. W., “Statistical Methods,” 4th ed., Ames. Iowa.
Collegiate Press, 1946.
LITERATURE CITED
Received June 29, 1948. Presented before the Division of Analytical and
(1) Assoc. Official Agr. Chem., J. Assoc. Official Agr. Chem., 30, Micro Chemistry at the 113th Meeting of the Amekican Chemical Society,
70-1 (1947). Chicago, 111.

Photometric Determination of Epinephrine in

Pharmaceutical Products
J. ROY DOTY, Bureau of Chemistry, American Dental Association, Chicago, III.

A red-blue color results from the addition of a solution of ferrous sulfate and a
suitable buffer to a dilute solution of epinephrine. The transmittancy of this
solution as determined at a wave length of 530 millimicrons is used to measure
the amount of epinephrine. This method may be applied directly to many
pharmaceutical products. The procedure is extremely simple and rapid and is
capable of an accuracy of about 1% of the amount of epinephrine present. The
optimum conditions for development of color are reported in some detail.
EXPERIMENTAL
need for a reliable chemical method for the estimation of
THE epinephrine in solutions of local anesthetic agents led to the The data for the absorption curve in Figure 1 were obtained
with a Beckman Model DU spectrophotometer. Maximum
initiation of this investigation. A colorimetric procedure has
been developed and tested over a period of approximately 3 absorption occurs at a wave length of 540 millimicrons. There is
very little absorption due to the reagents alone at this same wave
years. The method is a modification of one described in 1923 by length. A Coleman Model 10S spectrophotometer with a light
Mitchell (3) for the estimation of tannins and extended by Price slit designed to give an effective spectral band width of 30 milli-
if) and Glasstone (2) to the analysis of related polyphenolic sub-
stances. The color varies with the pH of the solution. When an
alkaline buffer is added to a slightly acid solution containing
epinephrine and a ferrous salt, a blue color begins to develop at
about pH 6.5 and gradually changes to the characteristic red-blue
color which attains a maximum intensity at about pH 8 to pH
8.5.
A similar procedure was mentioned unfavorably by Barker,
Eastland, and Evers (1), who did not find the reaction sufficiently
sensitive for the estimation of epinephrine in adrenal glands. At
least part of their difficulty may have been due to an attempt to
use the blue color of lower intensity rather than the red-blue
color that develops at a higher pH. The present procedure may
be employed to best advantage when the epinephrine content is
at least 10 p.p.m. Vogeler (5) reports briefly concerning a photo-
metric study of the “ferrous iron-adrenaline complex.” Un-
fortunately his data are too meager to be of much value in formu-
lating an analytical procedure.
Yoe and Jones (6), on the other hand, have suggested the use
of disodium-1,2-dihydroxybenzene-3,5-disulfonate for the estima-
tion of ferric iron. In the early part of this investigation a solu-
tion of a ferric salt was employed to develop the colored iron-
epinephrine complex. On the basis of further experience, how-
ever, it was decided that the ferrous salt possessed some advan-
tages and it was used throughout the remainder of this work. Epinephrine Complex