[I Li

i

IWIHJ

i
I

I

WW!

I
'
l

144
287

 

urn ~1Liar‘wsmwems .' mzv’mgmsfiwazw 'nuN'C-‘Jf: -' 4-20.?an W

9 ..-
g .3
3' '5

KL}.-

.

O
,

o

-"4~‘-'n; gilt"
"-

‘4.

‘-;;...= ‘4‘

J"
.
I

f: ,' ‘i ‘ ' 1
‘ ' I c
t . r
. ‘ . '
‘ r
I
. q
. h
h N k - '. '
' \ o - _ . ‘ .- I
‘
‘Q
_ .. . ~
I ' ' ., —. ‘ . ‘_
\ .~ v _.
l

J
p
t r
0
mun-0.... ...»:M.;flt I. .Morw: 3L3.“

"fix?" '

”at.

.. uququ-‘mtg’ -.'«g.+asamsz-.€wmmm :M35Wji'! '7': karma; mam ' ” '71- ‘254493! Ww'
,

 

mm

 

L I B R A R Y
?- {{chioan State

   

    

4.

 

 

 

 

 

 

'UW'HV Vv-I “—

 

THE STUDY AND BE FRHINATION OF

(I)

ACCURATE MICRO XE HODS OF CAFFEINE ANALYSI
IN DECAFFEINATFD COFFEE

THESIS

Submitted to the Faculty of xichigan State
College as partial fulfillment of the requirements

of the Degree of Master of Science.

BY
Winston Fields Allen

August 1929.

5.3

APPEECIeTION
The writer wished to eXpress his appreciation
to various members of the Faculty of the Chemistry
Department for their helpful advice, and also to
Mr. H. K. tilder, Chief Chemist of the W. K. KellOgg
Company for his constant co-operation and advice,

while carrying on the work of caffeine analysis.

33 “1-597"

TABLE OF COHTENTS

PAGES
Introduction --------------------------------- l - 3.
Hietorical,General ---------------------------- 4 - 7.
' , 1 Properties of ceffeine ---------- 8 - 18.
" .11 Methods of caffeine analysis ---- 19 - 42.

Experimental, I Methods,modified Fendler-Stuber-43 - 49.

' , " ' ,Nitrogen determination-~50 - b6.

“ , “ ' ,Sublimation ------------ 57 - 61.

' , " ” .Precautions ---------- L-62.

' . .II‘;~Anllwtica1 results ----------- 63 -r83.
Summary,~I-Methode ---------------------------- 84 - 85.

' .11 Analytical results,everagce- -------- 86 - 87.

Recommendations ------------------------------- 88.

a
- .—
.

,
k. l
v“.
~

.
\
.‘V
—.
‘
.
Q‘-
,-
. .1,
..
\I

- __-.4
u..--
--..
. ,
‘ _ 4
-—

-. a _— —
- I I
n
.. ”.- --.-

'1

LA

r' .-

.-——.-.—-—~-———-—

0
' r
‘ ,._

u-..—

--—.‘-

v u-
.\ ' ,' .,
.. t- . .
, .. l - .
O
..
E
" C
a "
- . I A
a
- . . ‘-

1‘.“ - o

 

 

 

 

 

1.

IETEODUCTICE

The analysis of caffeine in coffee, tea and cocoa
has been accomplished by many varied methods. This has been
due, partially at least, to the somewhat long and trouble-
some methods often employed and their comparative inaccura-
cies, resulting in the constant endeavor to perfect methods
which are comparatively short and easy to run, and at the
same time capable of giving accurate and consistent results.

In the analysis of decsffcinsted coffee for caffe-
ine by the Fendler and Etuber method, as well as by the Power
and Chesnutt method it is found that a 10 g. sample of stfee
Hag or Sanka Coffee contains on the average from 1 to 3 mgs.
of caffeine, determined either by the nitrogen content of the
final residue of caffeine, or by sublimation of this residue.
This small amount of caffeine present necessitates the employ-
ment of methods more accurate then is necessary in the case
of the caffeine analysis of ordinary coffee.

Practically all of recognized methods in use up to
the present time calculate the percentage of caffeine on the
dried weight of the final, more or less crude caffeine residue
which is left after evaporating off the solvent (usually chloro-
form) and drying in an oven at 100°C. for about 50 minutes.
Most of the methods make it optional for the analyst to deter—
mine the purity of this residue by running a nitrogen determi-

nation on it and multiplying by the actor 3.464 for caffeine.

1‘3

But this has not been made compulsory and the percentage of
caffeine may be based upon the weight of the residue alone
and be considered correct. The error involved here, due to
the oily and waxy impurities present with the caffeine, is
perhaps negligible in the case of caffeine analysis of ordi-
nery coffee, and the residue will often run 95% pure ceffeine.‘
However in the case of deceffeineted coffee the residue of
caffeine will usually run from 20 to 40% pure caffeine. Hence
basing the final results in this case upon the weight of the
residue would be entirely erroneous.

It is the object of the analytical and experimental
date given in this thesis to thoroly prove the stove state-
ment. In addition to this, an attempt was made to accomplish
the following:

1. To find possible losses of caffeine in the
various steps of the analysis. And to make suitable modifica-
tions or additions in order to make a more adaptable micr
method.

2. To further purify the final caffeine residue.

5. To accurately determine the percentage of
caffeine in this residue.

It is needless to say that to fully cover these
accomplishments in the time already spent on this work is an
impossibility, and further investigations will be made on
phases of this problem not touched upon in this thesis.

A great deal of the exyerimentel work already cov-

ered has to do with the accurate determination of the percent-

age of caffeine in the caffeine residue, by two methods:(l) by
a micro nitrOgen determination, (S) by sublimation.

In View of the fact that the F3ndler~3tuber and the
Power-Chesnut: methods are the only rec0gnized methods in
use today, all of the analytical and experimental results
‘ given in this work involve the use of these two methods, with

modifications in some cases.

Caffeine was first isolated in the pure form in 1821;
when it was obtained almost simulttneously by Bunge, Pelloter
and Caventon, and Rohiquet. Oudry discovered a similar com—
pound in tea in 1827 and called it theine, but which was later
identified by Berzelious, together with the alkaloid of guerene,
as being identical with caffeine. (Allen's Comm. Org. Analysis,
5th. edition, vol.7, p. 307).

The method of Paul and Cownley (Pherm.J.(iii),18, 417,
1887) for caffeine analysis was prominent among the earlier
methods. It consisted in drying the moistened sample with 1g;
nited magnesia and extracting with alcohol. At about the same
time B. D. Smith (Chem.Centr.l£”O-7l,1e87, from Zeit. Oster.
Aooth.Ver. 41, 353) brought out a method for caffee consieting
in boiling the sample in successive portions of water in the
presence of magnesium oxide. Both methods separated and pur—
ified the caffeine by evaporating off the solvent, taking the
residue up in hot water, filtering and shaking the aqueous
filtrate with successive portions of chloroform. The solvent
is evaporated off, the residue dried & weighed.

The determination of ce'feine in coffee has held the
attention of the Association of Official figricultural Chemists
since 1908. And the various methods have been very thoroly
investigated by the Association. Prior to 1915 the Fuller
method and the Gorter method were studied (J.Assoc. Official

Agr. Chem. l,No.£,fiOe-8,ldlb-16) and also the hilger & Fricke

5.

method (U.S.Bur.Chem.Bul.lO7,Rev.P.153) for the purpose of
being adopted as provisional or tentative methods by the
Ass'n. The Sorter method involves Soshlet extraction with
CBCla. Purification consists simply in taking up the residue,
after evaporating off the CHCla, in hot water, filtering,
and finally shaking out the aqueous solution with successive
portions of CRCls, and evaporating off the solvent. It was
found that the caffeine was in a very impure condition, and
had to be further purified, or to has the results only upon
the nitrogen detereination. The Fuller method uses water ex-
traction, slightly acidified. Separation and purification of
the caffeine consists in precipitating the alkaloid as the
periodide, and then decomposing the iodine, filtering and
shaking out the filtrate with CHCl5 as usual. The method of
extraction and filtration is laborious and incomplete, which
is shown by the consistently lower results obtained in com-
parison with the other methods.

In 1915 the Stahlschnidt.method (J.A.0.A.C.,3,Ro.l,
21-24, 1317) was adopted as a provisional method after being
given certain modifications. Extraction is accomplished by
boiling the sample with water, and purification is made by
‘the addition of dry basic lead acetate. Excess lead is re-
moved in the filtrate by passing in 828 and refiltering.
The aqueous solution is shaken out with CHC13 as in the other
methods. The inconvenience of using large volume of water is
eliminated in the modifications by using aliquots. There is
often a tendency for the liquids to emulsify slightly when

shaking the final aqueous solution out with 03013. The final

6.

Caffeine residzm is quite sure in comparison with Fuller
method.

In 1917 and 1318 the Feridler — Etubo r metrlod (I. fi.
O.A.C.,5 Ho.?,2€ -73 ,1321) n: scriticslly studied in compari-
son eith the Stahlschnidt method. This method involves ex~
traction of the caffeine by shaking out with CHClg, the
sample being first dampened with dilute ammonia. Purifica-
tion of the dried residue consists in oxidation with 1% ”n04
at room temp. and removing the excess jerme ngenate with EE- fl_02.
After filtration, the aqueous solution is she ken out with
CBCl3, and the resulting dried residue of caffeine is of ex-
ceptional purity, in comparison with that obtained by some
other methods. It was found that there is about a 1% loss of
caffeine by oxidation by the £1th4 at room temperature. This
can be partially eliminated by keeping the flees in an ice
bath, but this lowers the purity of the final residue of
caffeine. In 1319 this method was adopted tentatively by the
Ass'n. because of its rapidity, ease of manipulation, and the
production of an exceptionally pure caffeine residue.

Also, in 1319 the Power-Chesnut method (J.Amer.Chem.
500.,14,1298,1319)end (J.A.O.£.C.5,ho.2,267-73,1321) was
thoroly studied by the Ass'n. of Official Agr. Chemists. Rx-
trection is accomplished here by the Soshlet extractor with
alcohol. The extract is purified by evaporating to dryness
in an aqueous suspension with magnesium oxide, pulverizing,
making into a pa nets and washing thoroly with we ter on a filter.

The filtrate is boiled with dilute $04, cooled, filtered

and shaken out with CHClE. Considerable coloring matter is

removed by washing the combined extracts with 1% KOH. Be-
cause of its scientific accuracy, wide apnlicability and
consistent results in close agreement with these obtained
in the Fendler-Stuber method, the Power-Chesnut method was
adopted as an official method by the A.O.A.C. in 1319.
Resulting from the thoro and critical study of
the different methods of caffeine analysis for coffee, the
Fendler~Stuber end the Power~Chesnut methods are the only
ones which, have stood the test, and are recognized today by

the Assoc. of Official Agr. Chemists.

8.

HISTORICAL‘CCont'd)

In E‘ROPEET 33:3: CFC. ‘i'i' ‘L‘Il" iii.

In order to better study the various methods of
analysis, and to enggest modifications for ca fe Mi -e analysis
of deceffeinsted coffees, it will be well to give some of
the properties of caffeine. (Allen's Connzrciel Organic
Analysis, fifth edition, Vol. 7.Veg. s.lks Ho ds, 9.608 to Sit ).
A. Physical Properties.

Caffeine is a solid at ordinary temperatures, and is a
very stable compound, It forms long, white silky needles
when crystelized from such solutions as water or chloroform.
The crystals present a characteristic appearance when magni-
fied 100 to 300 times.

The melting point of caffeine is 2:35 to 237 degrees
Centigrade, after drying. ( U.S. Phermacopeeis).

It sublimes very slowly when heated over 100 degrees,
and is gradual but continuous at 120°. Complete quantitative
sublimation takes place at 178 to 180°. Caffeine can be re-
peatedly evaporated with water at 100 degrees in an open
vessel without more than the very slightest loss of the alka~
loid, and no loss with chloroform even after repeated evegora-
tions.

Caffeine crystallizes from aqueous solutions with one
molecule of water. When heated to 100 degrees C. it loses
its water of crystallization, the loss amounting to not more
than 8. 5i, and usually not much over 7. 03% with the average

samples of commercial caffeine. The anhydrous crystals of

(
O

caffeine are opaque and quite friable, and dissolve without
turbidity in chloroform. Caffeine also becomes anhydrous by
prolonged exposure over 32804 at ordinary temperatures. And
under certain conditions anhydrous crystals are deposited

from solutions of alcohol and ether.

10.

The Solubility of caff eine in various solvents is given in the
following table:

 

Parts or solvent rec uired for one of caffeine.
(Anhydrous crfici
Solvent A. Com meille (Compt. rend.l8 $81,817) 0.3.8.

=~------—-~- .~--—-—u.-c—----—-—~------~’-——--i-m’“-c— —---—--¢.—---

Water. 74.0 8.2 96.0
Rectified spirit 44.0 86.0
Abs. alcohol 165.0 82.
Commercial ether 526.0
Pure anhydrous ether 2288.0 277.0 580.0
Chloroform 7.7 5.25 5.5
Carbon disulfide 1700.0 220.0
Petroleum spirit 4000.0

(Gookel,Chem.Centr.. 1837,2,401)
Benzene 109.8 18.9 100.0
Carbon tetrachloride 1 28.0 142.4
Acetone 50.0

-----------~---‘---—_-----‘-‘~-~----—---‘—‘---’----‘

Gms. caffeine 9 r 100 gms. of sntd, soln.
(Seidell, J.Amer. Ch m.£30x. 1907, 29,1031)

.--‘--‘--‘----‘—---‘~-~-“~--~-_~-~-*-’~“~i--—-~~--.

Temp. deg.C. Solubility per 100

““““““““““““““““““ El£:_:itg--§929;-
Eater. 25. 2.14 gms
Ether 25. 0.27
Chloroform 25. 11.0 0
Acetone 80. 2.18
Benzen 30. 1.22
Benzaldehyde. 80. 11.6%
Amyl acetate 80. ’ 0.72
ameline 80. ‘2.89
Amyl Alcohol 25. 0.43
Acetic acid 21. 5 2.44
Xylene. 82.5 1.11
Toluene 25. 0.57

 

The solubility of caffeine in water is increased by
adding sodium benzoate, sodium bromide, sodium cinniminate,

sodium silicate and entipyrine.

---.

 

 

11.

B. Chemical Properties.
Caffeine is a methylated Oxypurine compound, or 1,
8,7, tri-methyl Xanthine. It has the following structural

formula:

C33“? " ('3: 0

0 =C C - N“Cflq

3" 1' H

CH3 -n _-c - med;

It acts as a very feeble base, having a dissocia-

~14
tion constant of 4 x 10 .

HCl has no action on caffeine at temperatures be~
low 200 deg.C., but when it is heated with cone. 801. at 250
deg. under pressure for 6 to 12 hours oxidation products are
formed which include ammonia, methylamine, sarcosine, carbon
dioxide and water, and sometimes formic acid.

Dilute alkali hydroxides when warmed with caffeine
change the alkaloid to an acid. C88120584. but at first add-
ing on 820. This is easily split up into 002 and the base
caffeidine, 07812084. This is upon further heating de-
composed to 002., 80008.,Hfiz.,033882., and sarcosine (022882
C008). These are practically the same end products as when de-
composed with a cone. acid.

Caffeine is easily decomposed by heating with lime
water, or heating with soda lime at 180 deg. (forming a large
quantity of cyanide, thus distinguishing caffeine from piperine,
morthine, quinine, and cinchoninc). If ignited with soda lime
ammonia is evolved, and cyanide is formed as only an inter-
mediate product. However caffeine may be boiled with an

aqueous suspension of 250. or PbO. without the slightest loss

‘0

 

or decomposition. This fact is made use of in several meth-
ods of analysis as a means of purifying the alkaloid.

The well known murexide reaction, especially as a
qualitative test for uric acid, gives very good results with
caffeine or theobronine (8,7 dimethyl xanthene). The alka~
loid may be oxidized several ways:

1. Evaporated to dryness with Br. water.

2. Evanorate to dryness with cone. Cl. and a

crystal of K0105.

8. Evaporete to dryness with cone. 8303.

Upon adding a few drops of cone. NH4OH to the yellow
residue a bright reddish purple is produced. The addition of
EaOH will cause immediate decoloration, while with uric acid
the color turns blue. This test is a very delicate qualita-
tive one.

All methylated xenthenes give the murexide reaction
as well as uric acid. Tho the color produced is the same
the reactions for caffeine and uric acid are slightly differ-
ent, as is shown:

Caffeine-~0xidized to Amelie acid —-add ammonia, gives~

(C832(C33)408§4) Eurexoin.
(H34C8(CH3)406N5)0
Uric acid-Oxidized to Alloxantin~-- add ammonia, gives-
(08360884) hurexide.
(28408840685)

Caffeine can be distinguished from theobromine according to
F. P. Btoup (Ann.J.Pharm., 1018,91,508) by treating a sample
on a tile with one part K20r207 and twenty parts of cone.

82504. If it is caffeine the yellos color of the reagent is
turned almost immediately to a bluish green. while if it is

theobronine the color changes from yellow thru purplish green

13.

and olive green to the same bluish green of caffeine.
Potassium-iodo-bismuthate, if freshly prepared, gives
an orange colored pot. with caffeine and theobronine. Old
reagents due to traces of hydriotic acid give different re-
sults (M.Malny.,J.Pharm.Chim.,1921,:5,830). This test also
distinguishes between the two alkaloids. It is as follows:
0.05 gm. of the alkaloid is dissolved in 10 cc.of ago and
0.5 cc. of the fresh reagent and 5 drops of 13% iodine soln. co
containing about 1% hydriotic acid are added. Caffeine gives
a red ppt., while theobromine gives a brown ppt. which changes
in about 30 min. to a chocolate brown.
A very delicate test for purines in general consists

in boiling the alkaloid with cone. RN05 end K3F9(CN)5. A ppt.

of Prussian blue is formed noon dilution. CH5
fl ~C=O
Cholestroghane or dimethylparabanic acid, CQ ' is
N ~C:
formed when caffeine is boiled with an excess of '
CH

cone. HNOg or chromic acid mixture. It is a compound ”which
crystallizes in pearly laminae, having a melting point of 145.50
and A B.pt. of 275 to 277°. This compound is very easily de-
composed Iith alkalies into symmetrical dimethyl carhamide and
oxalic acid. Hence on adding H3405 and C8C12 to its aqueous
soln. and warming the liquid calcium oxalate is ppt.

Hautral iodized potassium iodide and mayors soln. do
not opt. caffeine. This property distinguishes caffeine and
theobromine from nearly all other alkaloids. However an acid
soln. of iodine and K1. ppts. caffeine quantitatively.

PhOSphonolybdic acid produces a yellow pot. with
caffeine soluble in warm Ha acetate soln., the liquid deposit-

ing free caffeine upon cooling.

l4.

'ercuric chloride (Hng ) gives a characteristic
qualitative test with all methylated purines by forming large
white rosettes of crystals in an aqueous soln. These cr3 a“t£lS
are CngQOQEéHgCIZ. They are formed immediately in a saturat-
gd soln., and after a few dsys in a solo. containing one
part of aklaloid to 4000 parts of water. This compound is
soluble in about 260 parts of cold water, hence it is not
adapted for the quantitative precipitation of caffeine. (E.B.
Davies., Pherm.J., 183 H, 5, LLL , 31.).

Gellotsnnin precipitates moderately dilute solns.
of caffeine. A variation in temperature, changes the solubil-
ity and the kind of preci: itate formed

Even tho caffeine is a feeble base and will form
salts with acids they are very easily decomposed by water,
alcohol and ether. This is due to the fact that salts of
caffeine are hydrolytically dissociated upon dilution. Hence
it is possible to completely remove the alkaloid from an acidi-
fied solution of water by shaking out M? rel times with ben-
zene or chloroform.‘

affeine hydrochloride crystallizes in colorless
prismatic needles The acid is lost at 75°C. Caffeine sul-
fate is deposited from a hot alcoholic solution in shining
needles, and is not decomposed below 103°C. Caffeine nitrate
is found as fine transparent crystals, becoming opaque in
water and changing into pseudomorphs, or microscopic needles
of the free alkaloid.

C. 0. Johns has prepared isomers of caffeine and

theobromine by alkylating with dimethylsulfete.

According to Gomberg (J.Amer.Chem.Soc., 347,1836,18)

15.

caffeine gives rise to a number of perhelidos which are note-
worthy on account of their insolubility in water. This is
accomplished by adding a solution of iodine and potassium io~
dide to an aqueous solution of the alkaloid acidified with
HCl. A pot. of Cafiloogfléfil.14 is immediately formed; and this
crystallizes into dark blue prisms when put into methyl alcohol.

Caffeine can be completely precipitated by silico
tungstic acid in the presence of 55 HCl. (C.A.Vol.16,PLBJE,
1922. & h. Azedian, Bull.Eoc.Chim.Belg.lE, 31.192? . then dried
at 30°C. the composition of the precipitate is 12 $05.8102.2320.
3(CBH10N402) plus 6820. After igniting this there is left 12
305.8102 which, multiplied by the factor 0.2646 gives the equiv—
alent of caffeine. thile this method might be used for the
quantitative determination of caffeine when considerable of the
alkaloid is present, it would not by practical or accurate
when there is present only one or two milligrams of caffeine.
Several alkaloids including caffeine can be titrated with the
above compound, using malachite green as the indicator. hale-
chite green in HCl has a reddish or brownish orange color.
@119 green color is at once restored by silico tungstic.,. acid.
As long as the alkaloid is in excess the solution remains
yellow, but as soon as the reagent is in excess the green
color appears. Volumetric silico tungstic acid.is standard-
ized against the pure caffeine to be titrated. (C.A.Vol.13.P.
792,1925. E.O.Eorth, and G.D. Beel, J.Am.Pherm.fissoc.lZ,883,
98,1001-9,1924).

According to 1.x. Kolthoff (Z.Inorg.Allgem.Chem.
112,136 to cos. 1920 and C.A.lL, 1870,1331) alkaloids and
their salts may be titrated with satisfactory accuracy by the

electrometric method in very dilute solutions. The alkaloids
themselves are titrated with acid; the conductivity curve rises
steadily from the commencement of the titration, and at the
,point where salt formation is complete the rate of increase of
conductivity augments. The elkaloidel salts are titrated with
alkali, and there is a sharp break at the neutral point.
Titration of alkaloids or their salts with indicators generally
give uncertain results because a solvent such as alcohol or
chloroform has to be used. while for eloctrometric titrations
no solvents need be used in the ceseoof alkaloidel salts, and
55% alcohol can be successfully used for the free alkaloids.
Caffeine and theobromine have a dissociation constant of
approximate l lo’léand 1 1C)"10 resoectively, and the latter
can be exactly estimated in a mixture of the two alkaloids by
dissolving in an alkali end titrating back with an acid. The

break in the conductivity curve is quite distincty.

17.

C. Physiological Properties.

While having no direct bearing upon this problem, it
may be of general interest to give a fee of the physiological
properties of caffeine.

This alkaloid is odorless, but has a bitter taste.
Eben given in excessive doses it possesses decidedly poison-
ous properties. When administered to frogs it produces
tetanas and rigor of the voluntary muscles. One fourth of 3
gm. of caffeine killed a cat in 35 minutes in one trial. In
all experiments it caused increased frequency of the heart's
action and repeated emptying of the bladder and intestines.
After death the alkaloid has been detected in the blood, bile

5 increased heart action

(0

end the urine. In men, caffeine ceus
by stimulating the cardiac muscles, and excites the nervous
system, and it has been stated to be an antidote for nicotine
poisoning. The toxicity of caffeine nn' fluctuate between
0.10 to 0.30 gm. per kilo. wt. of animal. (illen's Comm. Org.
Anslysis,5th edition, vol. 7,9.210).

A neutrrl compound of caffeine and sminoscidylphsn~
etddine is stated to be suitable for sucutnneous injections.
Stable solutions of caffeine ere prepared by dissolving in
sodium or ammonium benzoate, cinnimete, or salicylate, and are
used for hypodermic injections.

The continued use of stimulating drinks containing
caffeine results in the establishment of a tolerance for the
alkaloid in the body. The absence of this tolerance is easily
demonstrated by observing the effects of drinking tee or coffee
after having abstained from such stimulants for a considerable

length of time. However caffeine differs from most other

18.

alkaloid: in that the same dose will continue to give the same
amount of stimulation and effect upon the nervous system, after
once becoming accustomed to it. This is called the "habit
tolerance“ of caffeine. While such alkaloids as morphine and
cocaine are taken in increasingly larger doses to get the

same amount of stimulation, with the ineviteble fatal result.

Some authorities claim that the injection of caffeine
results in the increased ability of the body to take care of
foods containing a large percentage of protein. This perhaps
is partially due to the diuretic effect of the alkaloid.

It is questionable whether caffeine is changed to
uric acid to any extent in the body before being excreted.
most of it appears to be excreted in the urine as unchanged
caffeine. (E.L.Eerdell e V.C. Eeyers.?roc.50c.Expetl.Biol.

med. 23,888-30, 1326).

BIETOEICAL (cont'd)

II. E"T:£OD OF FF”IVP -3 TYSIS

\U

The caffeine found in coffee and tea, being entirely
soluble in hot water, alcohol and chloroform, has been ex-
tracted by these solvents in practically all of the earlier
.methods of caffeine analysis, as well as in the best methods
in use at the present time. Ether and carbon tetrachloride
has been used in a few of the less imnortant methods

The methods of purifying the extracted caffeine have
been quite varied. Line was need in a few of the earlier
methods, but due to the fact that it rapidly decomioses caffe—
ine when heated above 100° it was soon replaced by magnesium
oxide, which has no dthOJir7 effect up on the alkaloid at th
temp MI ture of boiling water. Some of the singler methods re-
move the fatty and cily impurities by simply adding hot water
to the dried residue after eveg~yorz ting the solvent off, filter-

qu sous solution

m

ing and reextrecting the caffeine from the
by swl king out with succesive oortions of chloroform. Further
purification is extremely necessary in the case of decsffeineted
coffee. Besic lead acetate is emgloyed successfully es s puri-
fying substance. The excess lead is precipitated as PbS. by
passing Ems thru the filtrate. Aluminum acetate is used in

one or two methods. The use of 1% exhoé solution and precip-
itating the excess with 3% 3300 is as popular way of purify-

ing the caffeine residue as the use of Ego. if not more so, be-
cause it is a simpler :nd a quic cse r procedrre of the two.

.Hosever there is a slight loss of caffeine when Kxn04 is used

‘213
L4‘ .

cepecially if the tenpereture is much over 80° 0., or if the
sample is let stand over 15-20 minutes.

Hydr lysis of some of the inrurities,

U)

uch as the
seponins, by boiling the crude extracted caffeine in dilute

acid is ernployed in com Jinntion with other steps in purifica-
tion in a few of the methods.

n all of the recognized methods the segaretion of

(0

th caffeine from the a queous solution, after purifying and
filtering, is accomplished by shaking out with from three to
six portions of chloroform, and then evaporating off the
solvent on a steam bath.

The celc zlction of the final residue of caffeine in
determining the percentage of caffeine is accomplished in one
of five hays in the v.? rious methods used up to the present time:

1. Drying the residue at 133°C. and weighing as pure

caffeine.

2. Deter: ining nitrogen in the resid one of caffeine.

3. Subliming the caffeine and weighing the sublined

crystals.

4. Precipitation of caffeine as per iodide, and

titreting the excess iodine left.

5. Precipitation of caffeine with silicotungstic

acid, igniting and weighing.

In studying the different methods of caffeine analy—
sis it can be readily observed that there are four distinct

steps in practically every method.

1. Extrcction with a suitable solvent. (Usually
setting with we ter or Er"onie if solvents
oth er than we ter or e.lcohol are used).

2. Purification of the caffeine after removal of the
solvent. (By absorption, oxidation, hydrolys s,
precipitation, or selective solubility).

3. Separation of the caffeine from the aqueous solution,
after purification & filtration, by shaking out
with chloroform.

4. Determination of the purity of the caffeine residue.

These four steps will be used as a basis for classifi-
cetion of the different methods of caffeine analysis in coffee
and tea. The majority of the methods are applicable to de—
ceffeineted ceffee, if greater care is observed in the analy-
sis, especially in the fourth step. The majority of the
methods will be described briefly, while the others will be

merely classified with the accompanying references.

’11: 0.“?
.‘q d 2 5‘:-

I. CAFFEI TEfi' “T B EITS HOT ExTEE.

A. Purification methods include the use of msgn

 

esium oxide.

1. g. D. Smith. Improved method of estimating caffeine
in COffee. (J.!2i?.-QI‘.Chem.EOC.54, £21137. 5: Chem. Centr.
1‘73- 71 ,1887., from Z “it.o tsmr f30tn.\cr. 1,3i3).

The srnple is mixes with e snail amount of Ego,
boiled 5 min. filtered. This is repeated. xede up to :03 cc.
by percolation. Combined fil es are evaiorateo to 20 cc.

Residue treated with alcohol.

r "solved in smell

.1)
u.

If)

is out with CHClg in the

1"31"+

w‘U

Car ille e T USi. XetHOQ-

f‘,
‘3.

I
In.

C)

of basic

ese arch on the determination

0

nd

5

their derivatives. (C.i.6,&473,l§lfi.

1

Ferm.Ch Bl,3:5~91).

'1

,3}?

out)

This is somewha t similar to t. e Bafli

7.
q.

1

0

pt 19

thc t the

C

a A

k.

(D

methode exce ‘1 i i d in an acid

’5!

a

(I)

U11
Jag,

neutralized wit‘ 0., cone. to 83 cc. acidified,

with CC14. ~sion with

'1' ‘5-
‘" L4-

'1

tered, treete to e no

extrecth

3. B éRQTEE &

"5 {‘5
Pt. :7...

'3

[MA

Detern.inetion of caffei tee and col

1%

'

l7,10&8,19£:0 .C;tu.,EUll. ..

1921.3231yst 4C

-(

(l) Grind sample to pg:

:2

\
MI- .4

(9) To 5 gm. te in

heavy 2:30:31; 230 cc. meter.

s

3

(E) Boil gently for

3"

01510

Giorn.

nairc*
sole.
extracted

tor, fil-

wder & re-

:.X.BE 118?.
f8 e. Cok.

7:7, .56-8

(4) Cool, dilute to volumc 3: filter thru a dry filter.
(5) To a 300 cc. aliguot in a 1T 4. JVlontoymr flask add
10 cc. of 10% H;SO4?: evao. by gentle boiling to 100cc.
(6) Filter into a sspcratory funnel, washing the flask
with small portions of 1% Hugo}.
(7) Shake out 6 times with ChC (“5,2 ,15,13,10,10,cc.
portions).
(8) Treat the combined arts. with 5 cc. of § EOE. Shsk
t? LOTOIYo
(9) raw off the CBC lglayer, and wash the KGB with 2.10 cc
portions of CHClE and add to the main bulk of the extd.
(is) Distill off the CHClgto a 2.2211 amt. & transfer to
tarcd flask. Evap. to dryness, dry at 100° to constant wt.
and weigh.
(11) Transfer residue to a digestion fla :Z'k with 3? all amt.
of H2504i Deter. 3. by Kjeldahl method. Calc. caffeine from
N. by the factor 3.464.
4. Tsssilz. (Tassily E. Es tin nation of caffeine in caffee.
Bull. Soc.Chim.(iii),17,7Bl~76 8, & J.P:or.Chem.Eoc.,
76,154).

.B. Purification of the extracted caffeine by oxidation
with KHnO4.
l. Eavini. (Savini G. The analysis of coffee and coffee
substitutes. An accurate method of determining caffeine.
Ann. Chis. Applicsts lE,?l7—EO,1925. C.A.18,lfif,1374).
In this method the aqueous extract is mcde slitalint“
wisth ammonia and extracted with CBC15 in a merino extractor.

c.

Thte solvent is evaporatsa off and the residue boiled with

f0
sh
0

sufficient 0.5% KEnOé in 50 cc. of water to give a permanent
violet color. It is reextrected with ChCls in the Karina ex-

tractor. The solvent evaporated ofi, and the residue dried at

O '1 I - o 1
98 for L5 to 23 min. and weighec as pure caffeine.

C. Purification of the extracted caffeine by precipitation of
impurities with basic lead acetate, or basic aluminum acetate.
1. codified Stablechmidt fieth d. (J.A.O.L.C. Z,No.l,2?,
1917. e Allen's Comm. Organic inalycis,Vol.5,Pt.S,P.43).

(l) Grind scmple to pass a 40 mesh sieve e weigh out

.3

C i
o
H
‘1
01

g.
(2) Add 22’ cc. of water to the sample in c 530 cc. flask,
attach a reflux condenser and boil for 3 hrs.
(3) Add 2g of dry basic lecd acetate e boil 10 min.
Cool to room temp.
(4) Transfer to a 250 cc. grccustcd flask. take up to
the mark, thoroly mix e filter thru a dry paper.

5) icesure 200 cc. of th filtrate into a 250 cc.

CD

graduated flask & pass fifS thru to remove the excess lead.
Hake up to the mark and filter.

(6) measure out 300 cc. of the filtrate, representing
2 g. of the sample, into an evaporating dish e cone. on a
steam bath to about 40 cc.

(7) bash the cone. scln. with as little water as possi-
iale into a separatory funnel e abate out 4 times with CHClg
(2E320,15 e 13 cc portions). In the case of coffee the scln.
ifs made slightly alkaline with H3403 before extraction with
EXEClg. Run the separate portions of 'HClathru e 5 cm. filter

Paper into a small Erlenmeyer flask.

?‘.’3
()1

(8) Remove the chloroform, dry the residue to a con-
stant at. at 103°, and weigh.
(3) If the miff3ino res idae is not pure ,dete rzin H.

by the Kjeldehl method & multipl" by the factor 3. 464 for

2. Hilner and Jnckcnsck method. (Hilger a. & Juchencck A.
The estimation of caffeine in tea and co“fec. J.Phcrm.
1837 (vi),6 184 -12¢. ;poth.zeit.1837,1e,145 e 433).

t

(l) 23 g. of finely ground coffee or tea are diges

1’
1+
D"
o)
B

for several hours with 333 cc. of water at room temp. E
boiled three hrs. f r green coffee & 13 hrs. for roasted
coffee, replacing the water lo;t b' evcgoretion.

(f) After cooling to 70° 7.5 to 8.5.of basic aluminum
acetate in solution is run in, end tEen 1.3 g. of fit 3C05 is
gra ually added while the mixture is well stirred.

(3) It is boiled for 5 min., cooled, enter azded to
maze the total fit. 1090 g. and filtered.

(4) 750 g. of the filtrate coresponding to 15 g. of
the samule , to which 10 g. of dried e powdered elur: invm hydro~
xide & a little filter paper he ve been added, are evaporated
to dryness.

(5 The residue is extracted for 8 hrs. with CC14.

(6) The solvent is evaporated off 3 the residue dried,
treighed 3 calculated as ca -ffe ine.

gzadien. (Bul.Soc.Chim.fielg.31,15, 1952 & alien’s

Comm.0rg.Analysis 5th. edi tion,vol.’?,9.340).

(l) fifter ex mcting the sample with water, the solo.
is: cleared With basic lead acetate.

(2) The caffeine is precio oitated in a 5» FCl soln. of

8C1 with silicotungetic acid, and let stand several hrs. or
over night.
(3) Tte liquid is filtered off and the opt. is dried

ignited and weighed. It has the coroosition of 1? 305.SIOg.,

’T'

which multiplied by the factor 0.5 6 gives the equivalent of ca

caffeine.

D. No purification or 2:33ration of impurities from the caffeine
other than shaking out the aqueous solution sittx CHClS, in an
acid or alkaline soln.
l. Iétlocfi and Thom: on. (A? .aly ,1310,Lv,l)£)
The aqueous filtrste is evaporoted to a small vol.

3303 soln. added and then extracted with CHC15 etc. in the

2. Guillot. ‘hcoid e:tima tion of caffeine in tea and coffee.
(Cham.Centr.l.8 865,1833. & Apoth.Zeit.,8,lZ$).J.Am.
Chem.30€.64,838).
This method 51.? )ly consists in boiling the sample in
successive portions of water, filtering, and shaking out with

3 portions of CfiCla.

3E. Enthods involving the use of iodine 3 KI soln. precipita-
ting the caffeine as a periodide.
l. Caffeine dctermin3d by titre ting the excess iodine in

'3‘

an) aliquot of the filtrste MI .6

H.
0

or or cioitsf in5 the. caffe

:33 the periodide.
go at wg's Proc ess. (Illen's Cons. Org. Aztslysis,3th
,dition,vil.7,p 333 & J.lner.Chem.Soc.18,3;l,1896 &
J.brer.»nem.otc.l3,~7o,l S37)

(1) 5 g. of tr.=a are boileci for 5 hr. with 400 cc. of

water, and than cisested for another 5 hr. with an excess of
freshly opt. ferric hydroxide.
) The liguid is cooled end made us to a suitable
vol. An eiiguot is take; chi filtered, scidified with dii.hCl.
(3) The csffeise is got. by a known vol. of standard-
ized iodine soln., s the opt. silosed to sett‘e after being

cede up to a convenient vol.

g4) An aliquot part of the clear supernatant liquid is

("1"

drawn off, and he excess of iodine determined as usual. The

we 1 part iodine :

fwd
f‘

amount of caffeine obtained from the Va

\

o.t834 caffeine, or ice. N/iO sodium

Cf
‘ A
H
O
"1
:1
run!
"a
1‘)
(f
m
I
0
o
C)
(3
I»\

8
caffeine. The most accurate results are obtained when the

O
J

iodine soln. is used in large xce

('9

J

r
I.

This method was devised for the estimation of
caffeine in drugs. It depends upon the fact that when a soln.
of caffeine containing 301 is treated with a sols. of iodine
and potasium iodide the .nole of the caffeine is pot. as

periodide CefllQOgfléfllrléo

(D

2. Caffeine separst d and purified by precipitating as
the periodide end then decomposing the iodine, reextrecting
the caffeine, dried e weighed es in other methods.

3. C. FUller. Fuller method for tea and coffee. (J.

'seoca Officiel Agr. Chem.l,flo.2,203—8,1315~16).

(l) A 10 g. sample is out into an Erlenmeyer flask

1330 cc. water i 10 cc. of 10% 8C1, and heated under a reflux
(condenser for 2 hrs.
(2) It is cooled s the liquid decanted thru a filter.
T;:e solid materiel is treated with 3~‘0 cc. portions of hot

waster, washing thru the filter. And then the filter is washed

with 50cc. of boiling water.

( ) The entire filtre 3 is evaporated down to 1E0 cc.
It is transferred to a Squibb type sogtratory funnel, made
alkaline with Sec. of strong 3mmoni3, and 03'“3r out with —tOcc.
portions of CHClg. A second separation may have to be EL: do if

the liquidsc misify considz3 -r3t1v. If there is still an emul—

(

sion remaining in the second separatorv funnel 3 few cc'3 of

C33 EDIE is added and -3330. Then tho separated CflClgis added
to the main portion of the solvent and 335333 out the alcohol

layer with a little CEWCl
(d) The CHC 52Xtr£ ct is evaporatod off on 3 stoam bath,
the 13st portion beirg driven off by 3 curr3 :nt of air.

(5) To the dried residuo is acded 1303. of 10% £01 3

:1;
0
Cf)
H:
"b
u
H-
F}
ii)
H-
I"
3
’1
F
H
x4
CL
H
O

50cc. of 33 er and warmed until t2
solved. It is cooled 3 50cc. of iodine soln. added (10 g.io-

dine & 20 g. of KI in 13300. of water). The 1133K is stopperod

and let 3 and over night.
(8) The 3013. is filtered, r3fii.tcrin~ if no coss any,
and the flask and ppt. mt.‘-d with iodine $013. but not neces-
cerily removing all of the pot. to the filter.
(7) The filter paper is out into tho flask containing

the rest of the ppt. & 6.5 g. of sodium acid 333 ‘phit: 2, or

sodium sulphite, Ecc. of 10% H3804; lEcc. of water 3re added,
and uarmed11ntilcli of tne iodine 1303COmJO.LQo Kore of thc

salt is added if tile amount is insufficient to decolorize.
(8) The solo. is filtered into 3 separ3tory funnel,

an excess of strong N84OH is addcd 3 than shaken out 31th 5~15

m1...»
. .LLI

combined extracts are w3shed with

(I)

cc. portions of CH

C)?

water, discarding the latter.

(9) The CHC13 is evaporated down to 10 t9 15 cc. and drv ani-

rv‘
.4

mal charcoal is aned and allowed to stand for 1 hour wit
occasional shazing.

(10) The soln. is filtered thru a small filter paper into a
tared dish, washing the flask and filter 4 times wi

'portions of CHC 3.

II. CEFFEIHE EKTRACTED WITH CHLORQFOEE.

A. Purification methods include the use of magnesium oxide.

1. J.Kats (J.C.S.86,ii,301. & Arch.?h€ m. 942,4204s,1r34).
[mount of caffeine in the coffee used as
a beverage.

(l) A 10 g. Stmyle is wet with 10 cc of ammonia, and
shaken mechanically with 200 cc. of CHC13 for $ hour.

(2) After the solid has settled 153 cc. of the chlo~
roform is filtered off thru a Sander's "cigarette filter".

The CHCls is now distilled off.

(3) The residue is mixed with 10 cc of 0.5% HCl & a
few cc's. of ether. About 0.5 g. of paraffin are added, and
the ether evaporated on a water bath.

(4) The residue is heated until the paraffin has
melted thoroly, cooled and filtered thru a wet filter. The
residue is warmed twice in succession with 10 cc. of 0.5%

HCl and filtered.

(5) The united filtrates are extracted for 2 hrs.
in a Katz percolator, with 83815. The solvent is evaporated
and the crude residue weighed.

(6) The crude caffeine is dissolves in 10 cc. of
water (plus a few drops of other) which are then boiled off)
.7 the hot soln. is heated for 10 min. with a susoension of
lead hydroxide in water (1 to £0) and mixed with about 0.2 g.
of calcined magnesia, then filtered.

(7) The residue is washed with water, extracted with
CHClg in the percolator for 2 hrs.

(8). The CH013 is evaporated off, the pure caffeine

U4».

dried & weighed.
2. C. Virchow (The determination of caffeine 3 roasted cof~

f

(D
(J

~. Ch; m Ztg.£4,10¢7. & C.A.5,54£,l$ll &
Chem.2e1t.:4,lo;7-e,1913 & J.s.:.1.23,1224).
(l) 10 g. of finely ground coffee, 2.5 g. of‘ ego. e 10
of water are allowed to stand for 2 hours.
(2) This mixture is 31:. la en with 3,103 cc. portions of
chloroform, 3-1eking each one eboxt a minute.
(5) The extract is treated with l g. of paraffin & dis-
tilled. The last portion of the chloroform being removed with

a gentle current of air.

(I)

(4) 25 cc. of hot mater are added to the flask on the st em
bath and occasionally she {en until the sex and fat is melted

(5) The soln. is poured into e beaker, washing the flask
out thoroly with hot water, and heated until the we: and fats
have completely melted.

(6) The soln. is cooled £.nd filtered, and the filtrate
evaporated down to dryness s. The residue isc ried to constant
wt. and weighed.

(7) To purify the crude caffeine u, the resid‘le is washed
into a porcelain dish and partially evesol Ited on a hot water
bath. Finely powdered Ego is added and the mixture evevoreted
to dryness.

(8) The residue is pewdered as finely as possible and ex-
tracted with CH015 in 5 portions, deczeting thru a see 11 filter
into a neighed flack.

'(3) The solvent e rsshiags ere distilled off, the residue

dried and weighed.

5. . Bonifezi. The determination of caffeine in de-

.17

 

Cuffeineted Goff .("itt.LGt Jensm Jiyg.
17,5:7-1:,152e & Ann.Chim.encl.Chim.
appi..,I.,L,1e“7) .n.21,1158,l??7).

(1) Add 5 cc. of 5% 354 H soln. to a 25 g. sample

(finely ground) in a Soxhlet engerttus, and ex-
tract 3 to 4 hrs. with CHC15 or ether.

(2) Distill off the solvent and add 80 cc. of hot
water to the residue. Boil for 5 minutes.

(3) Add 1 g. of 530 suspended in 8 little meter and
continue boiling for another 5 minutes

(4) Filter thru a moistened filter and wash 4 to 5
times with 20- 25 cc. of boiling water.

(5 ) Evzoorate to dryness, e eIt in an o’cn at 100°
for 1 hour.

(6) fidd 10 cc. of boiling seter to the residue. If
it is not entirely sole ble filter, wash and again evaporate
to dryness.

(7) TeSe up in about 50 cc. of n.' te r and add 5 cc.
of 1% K3n04 soln. dtir Ior 5 minutes end then decolorize
With 5% H302 containing 1% EHCItiC acid.

(8) Filter, if necessary, evaporate to dryness and

fle 2at for an hour at 103° in In oven.
(9) Extract the residue with boiling C8315 4 or 5

‘times, rubbing the residue with a stirring ron, and filtering
tfiiru a small filter.

(10) Fveporate off most of the CIlCl5} and then sub-
lidne the caffeine according to the directions of Vautier.

(ll) Weigh the sublimcte after drying 1 hr. at 100°

E. Purii‘icetion m thods inClJGe th 3 use of potassium

l. Fendler end Etuber. The deterninetion of caffeine

(1) After grind ng the sample to pass a 1 mm. sieve

t- so odifisd to (.5m) 10 g. are Dit into e glass steepercd

{ >
(I)
*3

’1
.L

(’5‘

0'
(1'
(p

0 1e, and 10 cc. of 10% 33403 are added. Let stand e short
tie 3 until the sample is thoroly set.

(2) 230 g. of CHCl; ere added and shaken vigorously
for % hr. then chilled in an ice bath.

(2 ) The conte .nt: re filtered on a filter large
enOUgh to hold the entire contents of the flask, covering the
flash with a watch glass.

(4) 150 g. of the filtrete or more are collected,

hev in gthe funnel rest ing directly on the neck of tie flc ~5k,

J

I‘r‘

and the latter surrounded with ice. isssoon es the Cncl5
ceases to run from the funnel the flask is stopgered and weighed.
(5) The chloroform is distilled off on a steam bath,
removing the last of the CflCl5 with a current of air.
(5) The residue is digested sitn 60 cc. of water on
the steam bath for 10 minutes and let cool.
(7 20 cc. (i'er roasted) & 10 cc. (ior green coi‘fee)
of if Kh304 are added and the mixture let stand for 15 cinutes
shaking occasionally.
(8) 2 cc. of sfi 2502(cont'g.l cc. of glacial acetic
acid per 193 cc. free from vt:t'1t1 d) are added, and then 1 cc.

portion at a time until tile red color of he Kinog is destroyed.

It is then placed on the steam bath and 0.5 cc. portions of
add; d until the solution become 3 no lighter in color.
(3) It is cooled, filte ed thru e yoocn crucible,
weshinv sith cold water.
(10) The filtre te is transferred to e S?phrutOTY fun-
nel and she zen out with six portions (of £5 cc.oech) of CBC 3.

11) The combined wesling GV&JOT eted do.n to a

81‘

(1.
(D

smell vol. and trcnsferred to a weighed thKST, evaporated to
dryness, and dried to constant wt. in an oven at 100° for about
30? inutzs. The residue of caffeine is then .eigr ed.

(1?) To test

(1‘

he purity of the residue, the nitrogen
determination is made by the Ejeldahl method and multiplying

by the factor 3.464 for caffeine.

 

2. .Eonifszi. This sethod is described under II A.

having a double purification using both XgO & KKnO4.

C. The final caffeine residue is sublimed. Ego and Kfin04
not used.

1. J.Burmtnn. The determination of caffeine in tees

 

and roasted coffees. (C.A.5,1343,1911. & Ann. Pals.

4,39-101).

(l) 5 g. of a finely groan sample is dried to a
Constant wt. at 100°.

(2) It is snauen 10 min. with 50 cc. of petroleum

her (b.p. 60°): nd decanted thru a filte prpez. This is re—

pea ated us ing $5 cc. of ether. ‘nLI then the vhole sample is
thrown on the 9: per and sasned with 95 cc. of etime . Fetty'

matters {w sy be determined by evaporating off the other dry-

ing the residue and weighing.

(E) The residue on the filter paper is air dried and
transferred to e 233 cc glass :t039ered flask. 153 g. of C3 C13
are added end then 5 g. of a 10% NfiqOH soln. and shaken for

at least c0 minut

U}

(4) The extract is filtered thru a net pager, and the
solvent distilled off on a steam bath. Residue is dried and
weighed thus estimz ti. t.e crude caffeine.

(5) The residue is dissolved in a little CHCl3 and
transferred to a small sublimation of the following descrip~
tion: Length l8 cm., die. 18 mm. s sled at the bottom, drawn
down to a did. of 3 mm. 4 Cm from each end.

(6) The CH015 is evtoorlted o f on a steam bath, dried
at 100°.In the lower constriction an asbestos plug is placed.

(7) The tube is sent in e paraffin bath for at least

hours at 210° to 240°.

(8) The Upger bulb is nos filed off and the caffeine
crystals washed out with CHC 3 into a weighing bottle, evapor-
ated, dried and weighed. A second subiination sil give ab-

solutely pure caffeine. This method is saplic able to all sub-

stances containing caffeine.

D. do purification me thdS used other than adding hot water,

filtering and shaking the aqueous solution with chloroform.

 

1. F. Edens. The determir ation of ceff:i no in coffee.
(‘.A.5,147D,lall. e Arch.Chex. "icro:.t,215-l§

Tzis is essentially a double extraction m:: h d using
the Ks z extractor and adding paraffin to the etu sous so olut on.

2. A.C.Rottigger. A micro method for the determ m1 of

J

caffeine in coffee. (c.i. l ,747.13"7.

P?

Lo U:t9rSuCh.L:b3n m.oc,lqt- ,13 7)

.
1 O

5

t

.
01

xterwieetien

.a
).
‘~

. f.
,;
\4

‘-
.:- I.) x‘J:

r‘v—
AJ‘Z'M'
.k4—J.

'.
a‘. 1“} ‘.
—a

n
J»
.

“

itt.
t1".

0‘.

fl
‘\.'
d

invc

.00.

$‘\
b;..

-1

LA

‘H

I-
\u

 

Th1

'3

\o

t.

H

t .

¢

on

.4: O o
Ln... C \u/ . AG
L... O {a an... X n... o
. 0 An. .5 a... 5 C a...
.G o . . ...,.... .1 O. C
in g. u l u... t n... n. 3
wt m... . r4. 0 ’ .13 VHA ” .1; :.U . ‘1.
n... 0. . ..A.... 1.. .1. nu rm .5
n... m. o "In 7.... h” .2. 5... WA 9.
O r. r. x... n. "I. 9 a... C C S
.1. .3 7... ... O r. t m...“ C .
.c C a. 2 .T.. n... F. C .2 .G
p... . .. _ a O n... n... a.
C o o .1 no . .M t ”U a... n. .n.
.1 . L... ..... O .L t 0 d r...
....i n... a... . m n... O C .1
.l o .., nu. .1 C C .Q 3 O «u
I .. r 7.. 3 3. .u. .L w... +9 WV.
.3. . .. ._ .1 c C o n... m.” C
4.. 0....» . - . in... T .1. A. C
o A... in.“ w... n... ...... l. 3 t u n“
m. o W... ... ........ ....h .3 t n; T w... a... a
5 Au 1 i «I O 3. Lu” 2. A... ..C
1 .. ., .1 u t. .2. o
t ..n ,.. ~.\. . I. . AU .HA 3.... 3”» MU
F 1-. A. u... C . C a. t .3 an .5. a.” a
u,“ D .. J. 1... T D t O n... m}. t... 1L
”.1 N! a i o 0 u... f E. \H r .a. 1.” u...
.. .h .. i. R O r a... .n t r. t
.... . u... . .s U. ...- .3 C h t c... .1 n... .c.
3 no ..... .. .1... O a. ..., .L 1.. 8 w” .).
O 3 . .:.. v... . 2 t ”c I... :1 .vw . ... .d
y .l J. --. . 9.... n V n... .1 d V C
a t . o E .0. I. n... .fl A... C .1
r. F. .2 U. a. "J E I d S n t n.
R .u . . h 1... .. t a a... .1 a C 8 d.
x... 4L av In. A... .12.. fa “A .1 t an. .1
_...._ .. .. (K i S R a n... E F a...
.L LU .l t n; O t +4 t fl
.0. w.“ o r... t .0. d r .1 B. ... C .1
0 E d E 9 no t .3 o r 3 8 G
C «.1 l .i. «v u «1 5h. r f
a. “3 H“ 9 .t 1. S 1. .c .i A“ .t r.
r. H. .t d o .1 O t .1 n E 3
.l a. ... o n... S u .. . 3 n... h... a O C
9 r UL 5.0 L .l 3 B...
“I C o... n. C C n n. G
.1 .. . 1.. ’ +9 x .u .1 .3 t .5 a. . M... h.“
o... p C l r I T C i O T .. . I t
C "a,” av... .3 «I h at.
T. \I. f. \l/ x...) f ) \I/ t \II 0.
e c. O l a a g .d a A. .n 5 a .. L.
4.... .3 hi lax .I\ t .(.\ n. It...

 

I
“I
~A4
O
rt
A
&
VS"
5:
b.
fleas

i
b
O S'.
tr ctor with CflClq.
f
\
c

m...“ t d o
f... T o . u a... C .0
S «L 0. f h c e
r t C r. r
1 "0'; f tau 3 . .
P. .5 . u 0 u... n u .rw.

 

37.

III. CRFFEINE EXT? éCTED WITH L HYL £LCGEDL

A. Purification methods incluco the use of magncsium oxide.

 

1. Paul and COVEIE".(?THTE..J.(111 ,lE,4 7,1887.
Allen's Comm. Org. Anal. fit
tion. Vol.?,p L58.).

(1) 5 g. of a finely ground sample are mixed n a
mortar uith 2 g. of H30., moistened with hot water age in tri-
turated and then dried at 133°.

(C) It is extrzcted with boiling alcohol and the re-
sulting liquid 'vcifirttfiu neurl" to dryness.

(J) The residue is boiled with 53 cc. of tater and
treated with a few drop; of Hi ute E3804.

(4).!h3n cold the solo. is she en tith 05013 until
exhausted.

(5) The combined extracts are washed with a little
dil. sols. of 3303 which rem V38 some of tho coloring matters.

(6) The CHClé is distillwd off leaving the caffeine
pure and colorless or at the most a 51 ight gr—cn tinge.

2. Power and Chosnut. An improved m 3thocl for th3
Quantit.tivo c- t‘rxfl notion of
c: ffeine in veg3 3310. motoritl.
(F .B.?O‘6’CI' C *r‘ V .T':.\..h $511111; OJ.
5.313513 C5193 0. {-CO 0 ‘11-,1L-‘ L‘O-l;~3‘:,l‘3130
& Official & T 9M 3H:thods
of éany; 1:.A. J.C 13d. 3iition,
p.;;:.a c. .1: 64 ,1319).
(l). I 10 3. sample (ground to 33:3 3 0.5 mm. sieve)
is moistened with a little alcohol and is extracted for 8 hrs.

in a onhlet extractor.

(2).Tha extract 13 added to 13

J?

. of heavy 1.g0 in

t

\

lOi) cc. of mater and evaporated with frcguon stirring on a

water bath to a dry powder.

(5). It is finely pulverized and made into a pt:te
with hot water, then transferrEI with hot water to a smooth fil-
ter, cleaning the dish with a rubber tipped glass rod. It is
filtered into a l L. flask and the filter is washed until there
is 250 cc. of filtre

(4). $0 cc. of lOfi RESOé is coded and boiled gently for

minutes with a funnel in the neck of the flask.

03
C)

(5). The solo. is cooled and filtered thru a moistened
filter into a scparetory funnel, wee bin 5 the flask and filter
aith small portion: of 0.5;? H?CO4.

(6). The aqueous solo. is shaken out with 6,25 cc.

f

portions of CjC1

('4

(7). The com .bined extract of CEEC13 is sha;en in a separa—
tory funnel with 5 cc. of if 30H. Th~ CHClb 13 filterfid into
an Erlen nzeyer flask; and the K03 soln. s washed with 2,10 cc.
portions of CHCIE’ adding to the extrect.

(8). The finel C 015 extract is distilled on the steam
bath to a so all vol., and then transferreo v.ith CHC15 to a smell
“sighed beexer, carefully evaporated, dried for 30 min emu veigled.

(3). To test the piirity of the caffeine residue ,tiie N.
determination may be made, and multiply by the factor 3.464 for
caffeine.

B. Purification methods include the use of potassium

n
1. (Hrs.) S.Cobort. Dete mine tion of caffeine in tea

 

and COfffieo(CO£oEl,363,19&7l & fifin. FalSolg,
see-94,1996).

This method involves shaking the sample (previous7

set with ammonia) with 4 portions of alcohol. It differs from
the usual methods in that after each shaking with alcohol the
mixture is ccnringed. The dried residue after distilling off
the solvent is treated with hot water, filtered, treated cith

1% Enn04 etc. in the usual manner (Behdler and Stuher method)-

‘2").

IV. CSFFEIHI PVTfi’LTHD ”I AEBOE TiTE.CHLC;.

I
A. P‘s rificetion methods iuclude the use of Ecgnesiuu oxide.

 

l. I.J.c;"ill and C.P¥rtuee. The detereinetion 01
basic xenthi.es in cocoa,
tea, co;f3e and tn-ir ce—
rivetives. (C.A.6,o473,
131?. & Giorn.F:ro.Chir
61,8o7-44).

kit 1 this method the fi inely ground sample is extracted
with cc14 in a onhle extractor for sev r:l hrs. The solvent
is then shaken uith water ac idu‘i ated with 83804. This removes
the theobromine, leaving the caffeine n the C814, which is
0 is a ded

evaporated off and the residue taken up in water.

3')

H
(3.3
(P
(D
D.
G.
S
B
C+

and the mixture evago o dryness, powdered and trrns—

ferred to a flask under a reflux condenser ens extrec ed 4 to

5 times with boiling CHClI. The solvent is distilled off, and
U

the caffeine residue dri 3d to constant weir 1m and weighed.

B. Purification methods include he use of potassium

l. K.ecndrich and F.30ttbohm. 3 method for determining
ceife .e in coffee (C.£.5,
11:13. -,1.-: .3030-513 Et‘é‘tnfiy701n8to
2161 .JUIg 07.0.:(48' I..- (Jena W; 5m.
l7,£«l,-65).

(1).}. 20 g. sewnJl f coffee (fineness of 1 mm.) is
moistened with 10 cc. of water, and lrt stand with occasional
stirring for 5 hour. fifter rensferring to a paper capsule it
is extracted for 3 hrs. with CClé.

(2). 1 g. of paraffin is added to tie 0C14 extract, the
SOIVEflnt distilled off and the residue extracted with boiling
vate :3 (50, 25,25,25 cc. portions).

(3 ) The combined aqueous extract is filtered thru a

41.

moistened filter, the latter washed, and the filtrate treated
with 10 -63 cc. of 1% KKnO4.

(4). After standing 15 min. the En. is ppt. as hydrated
oxide by means of a little H203 (contg. 1i CHJCODQ). It is
heated 13 minutes on a steam bath, filtered, washed, evaporated
to dryness, and finally dried at 130°.

(5). The residue is extracted with warm CHCl3 by de-
cantation. Tue CHC13 is distilled off, the residue dried for

50 minutes at 100° and weighed.

40
bu.

V. CéEFFINE EXTRECTE.D RITE STEP...
A. Impurities separated only by treating residue with boil-
ing water and filtering.

L. E. Ventier. Determination of cai‘feine in coffee.
(fun. Calm. An :leto,:‘_-Q ' 4:07, ”’10 1318.,

 

(l). 5 g. of a finely ground sample is moistened with

extraction thistle and extracted for

p.
0‘
F.
D
'W
1.3

5 cc. of emmon
4 hours in a Soxhle a-parstus with ether

(2). The extract is evaporated, and the residue treated
with boiling water. The soln. is filtered and the insoluble
material washed s ch boiling water until 153 cc. have been
collected.

(3).T he iiitrate is evaporated, and the residue dis~
solved in a smell quaantity of boilio g water. The soln.is again
filtered, the insoluble portion washed, and the filtrate
evaporated to a smell vol. as before.

(4). The residue thus obtained is heated and the

sublimed caffeine collected on a match glass and weighed.

eve"? rwvmzt
:‘uo’L;' ..t. I“-5.l‘

I. KETEOBS
he methods of caffeine anelisis employed in this
work are, as stated in the introduction, the official 90 cr-
Chesnut method and the tentative Fendler-Ctuber method which
are given in the Official and Tentative he tb ods of Analysis
of the Assoc. of Official figr. Chemists, second edition,
p. 634-355.

The procedure for the Power-Chesnut method was
followed exactly as given, and each step was performed very
thoroly when analyzing deceffeinated coffee.

Procedure in the Fexdler-S uber method.
In the majority of the analyses made by the Fendler-

uber method a few minor modifications were edOQted, which
were thought to add to the accuracy and convenience of the
method.. Because of the preference of this method over the
Power—Chesnut method, for the caffeine anal sis of dec Hf .inet-
ed coffee,(the reasons for which will be given later), it will
-be well to give a brief discussion and description of the
various ste epso of the m At nod..fteps ho. 1,11,19 end 15 are also
applicable to the Power—Chesnut method.

1. The 55-h pie is ground to pass 8 O.E mm. sieve.
The necessity of grinding to this degree of fin neness is shown
in table No. 1. It is difficult to grind coffee to this degree
of fineness in most mills because of its oily nature.

The Wiley mill will not do satisfactorily if it is
desire.d to grind finer than the 1.0 mm. size. A oehble mill

5

works well if considerable time is available. A smell Fnter‘
prise £?4 coffee mill was used to do most of the grinding.
find it was found that by setting the plates quite closely to-
gether the most of the sample is groans to an even greater de-
gree of fineness than that which will just pass the 0.5 mm.
sieve. Hence it was found unnecessary to run the samples
thru a sieve after grinding.

2. A 10 g. sample is wet with 10 cc. of 10¢ flfl4OH.
This is a large enough sample even for the analysis of de-
ceffeineted coffee. A larger sample results in more inconven-

ience in handling with no greater accuracy in the results.

‘4
'1

‘

TEE FFFECT OF FIUEHEdS OF GFIUDIUG UPOEE THE EFFICIFUCY OF

FXTEACTIOH O

'73

AFFIIU: recs DECRFFE £1130 COUFCU.
(FUUDLUU - U- Us as U1T1:0D)

 

 

Eamjle. f oi caffeine by nitroge11determinetion.
1.0 mm. siev 0.5 mm. sieve.
Xe fee Eng 30. 1. 0.0192 s 0-0145 4
F " " " 0.0131 0.0133
" F a " 0.0106 0.0147
" " " ” 0.0130 0.0149

3
a .
:3

8. 0.0 73 0.0183 (Poser—Chesnut mstnoi)
" n " 9. 0.0186 0.0151 v 1 1
Average- 0.0135 4 0.0154 fl

 

COIPIFI MTIV FESULTE OBTEIIILD BY UEIUG Y.fi.TT“ ECTVIL Of 10% UUQOH

u
7}

V

T0 WET TUE SALPLE

CTI UJ 1ITB CULORCFOti
131" F

methdd).

 

i caffeine by nitrogen determination.

Let with mater. Bet with ammonia.

Kaffre H Ho. la 0.0147 1 0.0145 %

n " W " 0.0140 0.0 b;

” ' " 1b. 0.0106 0.0102

n " fl " 0. 0100 0.01E 1

fl " W 4. 0. 025 0.0546

” " " 5. 0.0340 0.043

U " fi 6, 0.0519 0.0130

" " fl 7. 0.0108 0.0111

" " " 8. 0.0163 0.0171

n " fi ‘ o 0001??) 00-41110

" ” " 10. 0.01S7 0.3118

" " " ll. 0. 0103 0.011:

" " ” 12. .0914 0.3: :A
rtnke Coffee 1. 0.0224 0.0203

0.0167 5 0.0184

‘1‘
4
Q
*5
Cl?
‘Q
C)
l

It is preferred to set the sample with 10% 30,00 in—

J

(stead of water, not because of obtaining higher or more accurate
results as is shown in Table Ho. 2, but because of a greater
convenience in filtering (the residue being in a more granular
state), and the securing of a more colorless aqueous solution
in the next step.

3. Approximately 150 g. of comm. redistilled CfiClg
are added to the set sample in a suitable glass steppered
bottle. As the entire 01-1013 is used, it is unnecessary to
weigh the solvent, and a mark can be made on the side of the
bottle so that in later analyses the bottles can be filled
very rapidly. A rubber band is placed around the bottle and
over the glass stopper to hold it in place while shaking.

It was found that ordinary comm. CHC15 used possessed
no caffeine destroying power as is shown on table U0. 3. But
in order to safeguard the possibility of securing 00015 of
varying composition all of this solvent used was redistilled
in the presence of caffeine.

4. The bottle is shaken in a shaking machine for at
least k hour (usually 1 hour). For absolutely thoro extraction
of the caffeine, Soxhlet extraction with 03013 is recommended.
however th s will result in a still more impure final caffeine
residue, than when shaking is used.

5. By carefully holding the StOppOr partially in place
the 08015 extract is poured into a 4 in. funnel having a small
cotton plug in the neck (instead of the usual 24 cm. filter
paper). Thus the extract is filtered into a 200 cc. Florence
flask without getting practically any of the residue into the
funnel. The residue left in the bottle is shaken out with 3

 

 

468

TaBLE no.3
COMPARISON OF ORDINARY COK‘. LOFOEOF" V-ITH fiLCOHOL FREE,
RFDIETILLED IN THE PFZCT EOr CIFBFI‘l CFL070I0“” IN

EFFICIENCY OF CAP FE I3fi EXTFACTION.
(Fendler-Stuber method)

 

% of caffeine by nitrogen determination
Commercial CHClS ‘ redistilled 03015

---~---------- -cn-C---~~‘--_--~ ----------~---‘----

 

Kaifee H?” “o. 1. 0/0101 % 0.0122 g
fl " n 2. 0.0111 0.0091
3 " " 4. 0.0289 0.0246
9 " " 5. 0.0239 0-0330
W W W 6. 00021.53 0001-30
I 3 ‘ 7. 0.0118 0.0111
Average~ 0.0183 % 0.018 2 %
TAB E HO. 4.
DIFFEREHCE IN UL“ IHG All éLIQUOT OH 1H3 EHTIFH AHOUHT (£39 RESE-

JT CEILO; 01“0:.l EXTF! CT.
(Fendler-Stuber method)
2,0f caffeine by nitrogen determination

Sample Aliquot. Entire

Kaffee Hag No. la. 0.0145 s 0. 0133 3
“ " a " 0.0133 0. 0145
" “ “ 1b. 0.0108 0.0115
' " " " 0.0100 0.0114
" ” " 1c. 0.0122 0.0131
” " " 4. 0.0946 “.'551
" " " 5. 0.0aso 0.0636
" " " 6. 0.0190 0.0012
" " " 7. 0.0111 0.0108
' " ” 2f. 0.0001 .Slsfl

0.0158 g

0.015? f

 

.i- -.

—~.

approx. 15 cc. portions of CBCl The comparative results

shown on table 00. 4 prove that the same and just as accurate
results are obtained when the entire extract is used, as when
an aliquot is taken. This singlifies and slightly shortens

the method by eliminating two seighings and the care in cooling
the extract in ice to prevent evaporation of the solvent, and
only requires the there washing of the residue. The use of a
cotton plug instead of a filter paper reduces the possibility
of a slight loss of caffeine by being retained by the filter.

6. The CHCl5 is distilled off on a steam bath. The
addition of a couple smell beads to the C0015 greatly facili—
tates uniform boiling and prevents bumping and subsequent loss
of material. The lest portion of 00015 is driven off with a
current of air.

7. 80 cc. of hot water are added, digested on the
steam bath for 10 minutes with occasional shaking, and then
cooled to as near 20° C. as possible.

8. 20 cc. of 1% KMnO4 are added and let stand for 15
minutes. Then 3 cc. of 3% 0302 (free of sceten'lide, and con-
taining 1 cc. of glacial acetic acid per 100 cc. of solution)
are added and the flask shaken. 1 cc. portions of the H202 are
added until the 33304 has been all reduced and precipitated
as dnog, and the entire disappearance of the red color to an
almost colorless solution. Th» flask is now put on a steam
bath, and after being thoroly heated 0.5 cc. portions of the
0202 solution are added until the liquid becomes no lighter in
color. Digestion on the steam bath is continued for 15 minutes.

9. After the contents of the flask have cooled it is

filtered thru a Ho. 50 moistened filter paper, into a 250 cc.

I

 

 

Fig.1. Twelve-unit shaking machine,adapted for

350-560 co.pear-shaped separatory funnels.

48.

pear shaped separatory funnel (into which has been added the
first 25 cc. portion of 6801? to shake out the agueous solu~
tion after filtration is completed). The flask and filter are

wash.d at least three times with small portions of cold dis-

1

tilled water.

10. The aqueous solution is shaken out with six 25 cc.
portions of CHClS. For this purpose a 12 unit shaking machine
is used, especially adapted for this work and which is shown in
the accompanying illustration. Each portion is shaken for about
two minutes. The portions of CECl3 are filtered thru small
filters when running out of the separatory funnels as a safeguard
to prevent any sediment and filmy material as well as water from
getting into the CflCls extract. his extract is run into the
same 200 cc.flasks which contained the aqueous solution (cleaned
and dried).

11. The ‘ECl3 is again distilled off on the steam bath,
and the last trace of the solvent is driven off with a gentle
current of air.

I 12. In samples where the weight of the residue is de--
sired for comparison, the residue is transferred into a small
weighing bottle with several small portions of CHCla, the sol-
vent is carefully evaporated, and the residue dried in an oven
at 100° C. for about 50 minutes. It is then cooled in a des-
sicator and weighed. If the caffeine is to be sublined, the
residue is transferred directly into the lower half of the sub-
limation tube, dried and weighed (with a counterpoise).

13. The amount of caffeine in the final residue is de-

termined in two ways in the analyses made thus far:

A. By the nitrogen determination.
(1). Using a micro Kjeldshl method, and apparatus.
(2). Using the ordinary macro Kjeldshl appartus,
but with micro methods.
(3). The employment of direct nesslerizstion.
B. Ry subliming the caffeine.
(1) weighing the sublimed crystals.

'I
I

(2) Determining nitrOgen on the sunlined caffeine.

The discussion of these two methods will now be

taken up in detail.

A. iethods cI eteriiuing nitrOgen in the caffeine residue.

 

fiicroa nelyc is, 1924, translated from the German edition 2,p.
94-102

This method 1

eccenti

45‘»

m

lly a refinement of the
official? NJ ldah -Cunningwérnold procedure, ‘siu; steam dis—
tillation in a special apparatus adapted ior micro enalvsis
only. How ever the method was modified and a sseciul auuaratus
was made w} ich resulted in greater simulicity and convenie ence
as well as a wider adaptrbilim for the anul'eis of the residu3

from deceffeinated coffee as eell a: from ordinary coffee the

C.»

18p sible with 9518's an it gner's modification of Pr gl's
apparatus. The procedure is as follows:
(a). 10 cc. of & digestion mixture (containing 2 cc's
of cone. $04, 1.0 g. or K? Soqcnd 0.02 g. of C3301 per 10
cc'e.) is added to the 200 cc flask containing the caffeine
esidue after evaporating off the CHClS- It takes only a few
minutes to evaporate the eater, and such a mixture is very con—
venient to use, and insures the same mix ture in ev: ry samule
thus mrking possible more con nste.nt blaik determinations. A
low flame is used in the digestion, and run a few minutes after
the mixture is entirely clear. It usually does not take over
15 minutes for complete digestion.
2. 30-50 cc. of freshly boiled distilled water are
dded to the digested mixture after it is cool. The flask is

stoppered if it is not to be distilled immediately.

 

 

Fig.2. Micro-Kieldahl distillation apparatus.

1-500 cc.KJeldahl flask used as n steam generator.
B-Funnel thru which water is added.

C-Funnel thru which NaOH is added.

D—Safoty trap made from a 25 by 200 mm. pyrex test tube.
E-made from a 25 by 200 mm. pyrex test tube.

F—Small condenser made of pyrex tubing.

G-Rubbcr stepper for Joint.

H-ZOO cc. flask containing digested sample.

1-150 cc.receiving flask containing N/lOO acid.

3. The flae:L is non placed into po 81 ’tion on the steam
distillation apparatus made esoocially for this :ork, the de~
Sign of which can best be staiied by obcervinc tnc accompanying
illustration. 1? 3 main body of this congratus is main from a

25 mm. by 203 mm. pyrex tcet tube, which is drawn out slightly
on the open end so that 3 rubber stopper can be fitted and still
be small enough to fit inside the neck of the 23. cc. lacks
used for digest on. The b“-o 0:1 the upgzr no-9-3., onion is
connected with the uaoer gortion of the condenser tubing is

also made from the some size oyrex test tube. The condenser

ffi
('3!
CD
0
O
(“a
I
'9':
L;
D:
”3
:34
H
H
H
*3
‘m‘
v
f.)
9
3
F“
L:
(i,
O
(1..
E
C‘
(+-
d'
U
N
0
Ct
D
H' H-
04
'0
I)
d
H
O
:3

flash is made of 6 mn. pyrex tubing. In fact, the ent_re cp-
paratus is made of o_"r3: glass.

10 cc. of fl/133 H3804 are nw eastlred from 10 cc.
micro burettes into a 150 cc. aide month fie.-k, 10 droos of
methyl red indicator added, and pie 03d under the rece oiv ving end
of the small condenser. A large cork is put onto the tube of
the condenser so that it Just re:. ts on the top of the receiving
flask when in position.

4. Heat is applied to the steam generator, and then
about 13 cc. of 42 i 3308 are coded to the dig Wed mixture
thru the funnel connected with the tube about half v3y between
the steam generator and where it enters he test tube.

5. Heat is applied to th3 steam generator end the di»
gestion flash in such a manner that ebout 130 cc. of liquid
distill over in 12 minutes, and ”at the volume of liquid in the

digestion f ask remains practically the cane.

C"
{'7
o

The receiving flask is now lowered so that the end of
the condenser is about 2 cm. above the licuid. The end of the
condenser is washed down with a small stream of water. Dis-

tillation is continued at an incses ed rate for about 2 minutes.

(D

The funnel (thru which the alkali is added) is re sed before
the heat is removed, to prevent suction.

6. The contents of the receiving flesh are titrated
with ‘2/100 NeOH from 10 cc. micro buretes. The end point is
taken when the m ethyl red becom es a pe z-eent c.: :ner, yellow
color.

Reerly every day, or every tie e a group of samples
are run a blank determination is made by using the same diges-
tion mixture, etc. as with semoles and a smel bit of filter

J

36

{.0

paper to digest. This serv

{‘0

e sefec heck on the apparatus
and the standard solutions. The blanks usually run from 0.30
to 0.55 cc of H/130 H2804. But t‘Cy will not vary more than
0.05 cc. for one particular time. The 330. of cc's N/l30i {303
used subtracted from the 10 00’s or N,lOO suso4 (minus the
blank) gives the ho. of cc's which neutralized the NH3 driven

over. And 103 cc. of 3/100 H2204 BQUHlS 0-000485 . of

up
A

[1)

Ti

caffeine. This multiplied by 10 (when the entire sample is
used) gives the percentage of caffeine in the 10 gm. ample.

The H/lO 3 E 804 and H“Ol are stands- dized a einst

m

k

\.

P/lOO 301. which is made from con .: met boiling point HCl.

They are checked against freshly prepared N/lOQ HCl from time
to time to insure them being correct. This is very important
when analyzing samples containing an average of less than one

milligram of nitrogen.

Q
r.
I

2. Ordinary Macro Ejeldehl procedure.
In many laboratories a micro nitrogen determination
apparstue similar to that just described will not be available.
However quite accurate and relieble results can be obtained

with the ordinary macro Kjeldahl apparatus using go or 500 cc.

Ejeldahl flasks.

5‘4 0

T’BLE 30.5.
CY OF T813 0 DT“IFY "fCVO ”J LDf ii ’THOD OF HITEOGEN
LE CC‘P'I 30 03L LILL IGELE CE L138 0? EITROGZfi.

CQEPAPF TIVE £CCUE;:
DETZEXIHATIOE In 32L 'PT

(Us ing a digestion mixture containing 3 cc. of conc..rL0 ~0.?n004
and 0. 02 pm. Cuso nu adding ap.rox. 10 cc. c0 SIC. firiore d113-

tilling. 500 cc. fiieldehl flasks are used).

 

1..Blank determinations showing the necessity of thoroly steaming out
the outfit Just bef ore using .(cc. of 3/100 3 sc4used).

 

 

 

 

 

Eot steamed Steamed 15 min . Eteeeed 30 min. Steamed 1 hr. fifter run-
at all. just before 8 hrs be W10 ju:t before ing out-
using.fused tee using. using. fit inter-
d :y bciore) mittent y
-------_- .................................................. £92-a- daz-
2.70 cc. 0. .77 cc. .03 cc. 0.55 cc. 0. 48 cc.
1 7150 0.-. 0.53 0.5
2.98 0.80 1.31 0.74 0.58
5.00 0.37 0 75 0.84 0.51
1.09 0.66 0. 63 0. 48
1.95 0.55 0.68 .35
0.60 0.71 0.76
0.50 '_ 0.65
Ave. 1.6? cc. 0.52 cc. 0.84 cc. 0.70 cc. 0.43 cc.

2. Caffeine in samples of Keffee flag and in pure caffeine determined
by both the macro and the micro cathode of nitrogen deter. and
multiplying by the factor 3.464.

Seen . D , _? ‘ ”e1 F h . By the micro-
.le . By th- macro .1610 11 m-t 00. Kjeldchl method
Steamed thoroly. Steamed 33 min. .t.0~”0 1 (Prrgyl'e modi-
just before hr.: hrs. ficution)
using. before
--- ............................... B§i§”; ...........
2.02 mg.caff. 2.02 mg. 1.86 mg. 1.31 mg. 2.0? me Blank
“ " " 2.05 1.87 2.00 I 3.03 ” deter.
" " " 2.11 1184 2.08 3.01 on micro
" " " 1.99 2.01 1.33 3.00 -aeo--_-
1.82 0.fi3 cc
Ave. 2.04 1.88 3.03 3.03 0.18
0.50
K5. .ffee [18. 0019
Ho. 163 0.0126 g 10. 0.0147 0-31
17 0.0104 17 0.0148 0-19
18 0.0113 18 0. 0130
23 0. 0373 S3 0. OYLQ
25 0. 02'80
24 0.0185 24 0.0153
25 0.0136 35 0.0104
25 0.0138
Average - 0.0167 % fiverege - 0.0166 g

‘- ._.

(n
(n

The comparative accuracy of the ordinary K331.deh1
procedure, when the same digestion mixture, N/100 standard solu-
tions etc. are used as with the micro procedure, can be ob-
served in the results obtained which are given in tsble No. 5.

It is of course taken for granted that e nitrOgen
deter. n.081stus will be steamed a short time.

These reatzlt: show the absolute necessity of thoroly
steesing the apparatus for at least one hour, or until fairly
constant black determinations are obtained which do not vary
by more then 0.30 of a cc. of N/100 acid. And this must be
done inn ediete 1y before us nz. If these precautions are not
£013.0wed the variation in blank determinations will ftimes be
greater than the acid neutralized by the ammonia given off by
the sample.

8. Direct Ne ssle rization.

Ta ble 30. 6 shows he results obtained by using di-
rect nes 81L rizetion (Koch and ficmeehin method. J.An.Chen.Soc.
46,2066,1934), in the determination of nitrogen in the caffeine
residues. tnile this method would no doubt be applicable as
a rspid method for control work, for extreme accuracy and de-

pendability it does not compare with the micro— Kjeldznl method.

T£BLE PO. 6.

COHPAFATIVE LEEULT OF fiITLOGEN DETERXIEATIQNS {.33 BY

DIRECT EELCLFDIZITIOW n} ED BY EICRO KJE .LD53L IFTX‘Jc.

 

Percentage of caffeine Percentage of crf'eine

‘lee° by direct neeoierization by micro’x3eld3‘l‘
Kaffee Bag Ho. 1 0.0175; 0.013 3
a a n g Q,QO§)6 0.0111
F N 3! 4, 0.0189 000251
9 " " 5. OoOHé? 0-0556
n n u a, 0.0148 0.0212
n n n 7 0.0095 0.0111
Average- 0.0158 % 0.0175 g

 

an.

 

Fig.3. Sublimation apparatus for the quantitative
determination of caffeine.

AvParaffin bath.

B-ZO by 200 mm.teet tube.

0-6 mm.pyrer tube with a 1 mm. Opening in the end.
D-Upper half made from a 25 by 200 mm.pyrax test tube.
E-Tight fitting ground glass Joint.

F-Lower half made from a 25 by 200 mm. pyrex test tube.

U"
Q

B. b instion ofc feine.

After making several thes of tubes for the quanta-

O

tive siblimation of caffeine in the endeavor to ot as near

0:}

100% recovery as possible and at the same time prevent the
contamination of the subl med 015stels by impurities sub-
liming along with them, the apparatus sheen in the accompany-
ing illustration was found to be the most set :.=fectory.

The sublimation tube is made from a 95 by 200 mm.
pyrex test tube, which is drawn out slightly in tee middle,
cut in two. The broken end of the upper portion is enlarged
hand fitted over the constricted end' of th e10aer portion. The
glass Joint thusz sede is ground in ur til ti ght fit is
assured. However it is made tapering enough so that tne u; pe r
portion may be lifted off without the lower part coming with it.
The outside diameter of the joint is not any larger than the
rest of the tube, and is about 30 mm. in length. The lower
tube is 100 mm. in length. The upper tube is also 100 mm. in
length, but it also has a 6 by 80 mm. tube sealed onto the
upper end after drawing the top of the test out Quite sharply
to a small diameter. The top of the small tube is hes ed until
but a very small opening remains. This was found more satis-
factory than putting a plug of cotton in the end. Both tubes
are thoroly dried before being used.

The caffeine residu o i seabed into the loser tube

01

V

with 5 to 4 about 5 cc. portions of CPC13 Tbs solvent is
carefully eveooreted off on the stzam bath.
When there is but 1 or E cc's of CHC15 lift, the

tube is rotated in an almost horizontal position so that the

residue will be deposited in a very thin layer around the in-
side of the tube. This was found to somewhat hasten the com-

‘ete sublimation of the alkaloid. The last trzmc :3 of C3013
are removed by a gentle current of air.

3. The two tubes are now out together and placed into

a 50 by 233 :3. test tube. A rather loose fitting cork is put
into the end of the outsiie tube, the small end of the upper
sublimation tube extendieg about 30 mm. above the cork.

4. The:

(I!

mbled apperat1

J
(,0

is -lsced into 3 paraffin

L

bath so that the Joint between the two tubes is at least 20 mm.
below he surface of the paraffin. The bath is kept at a tem-
p; -retuz e of 180 to 130° C. for about 10 hours. This temper-
ature was found to be tlie highest which could be safely main-
tained and prevent other materials from su liming with the
caffeine to any extent.
Practically all referee ce 5 found regarding the sub-

limation of caffeine for quantitative snelw' is (J.Burznann.
Ann. Fels., 4,33-101 r E. Phillippe, Hitt. Iebensm. Hyg.6,
177- ql,"’2$'~-47 1315. & J.Eur rmen ,Bu ll. Soc. Che:.,7,233), Speci-
fy a temtgaer mt are Hf £~O to s h'00 C. It was found if the temp.
wss allowed to remain above 190° C. for any 16n"th 0f time the
crystals of caffeine become contaminated with other me terisls
which sublimed above this temp. And tM re ange3red 3.3311
droolets of brownish material just shove the caffeine crystals.
Even at a temp. of 85° some szmales snoeefl c discoloretion

of the cryste-s, but to a much less degree, and there was an
entire absence of the drool ets of brornish substance. The com-

retive difference in at. of the sublimed caffeine at these

3

P

different temperature:3 czn be studied on table 30. 7 . Tho it

was not tried out, perhaps a still more pure product could

be secured if the temperaturewes held at 175° for at least

20 hours. To prove that all of the caffeine had sublined after
the 10 hours at about 185° C., the caffeine was washed out of
the upper portion of the apparatus, reassembled and the temper-
ature maintained at 235 to 240° for about 2 hours. No crys-
tals whatsoever were observed in the upper tube.

5. After sublimation is complete the upper tube is
immediately removed, cooled in a dessicator and weighed with a
counterpoise. The caffeine is thoroly washed out of the tube
into the 200 cc. flask,if the nitrogen determination is to be
made. The empty tube dried in an oven at 103° for 15 to 20
minutes, cooled and reneighed. In the majority of the samples
the difference in nt. 13 a little over 1 mg. Bence the weigh-
ing must be done very accurately.

6. It is realized that gravimetric methods are as a
rule hot as accurate for micro work as volumetric and color-
metric methods. However much can be learned by the sublime-
tion method, especially if a nitrogen determination is made on
the sublined caffeine, which gives results that are more near-
ly correct of any method or combination of methods yet tried
out for the caffeine analysis of deceffeinated coffee. By run~
ning the nitrogen determination on the residue left in the
lower tube after sublimation,the amount of nitrogen from non-
caffeine substances can be estimated, and gives us the error
involved when the results are based upon the nitrogen deter-
mination of the final caffeine residue.

7. In the endeavor to have the caffeine sublime in

as ure state as possible or in other words to have the wei,ht
. ,

 

60.

of the aublimod caffeine check with the nitrogen determination,
magnesium oxide, sodium carbonate and quartz sand were each so—
parotely mixed with the residue in tho subliming tube with the

idea of holding'the impurities from subliminc with the caffeine.

c:

(D
L

The results which apgeur on table $0. 7 indiCLt thtt every
one does more harm than good, as it prevents a small amount of
the caffeine from subliming and does not increase the purity of

7 offered the most possibility by

the sublimcd crystals. 553000 .
slightly increasing the rate of sublimation at a minimum temp.
with no appreciable loss of caffeine. However the final product

was no better than when it was subliméd alone.

n7 1"

F A/JJJJ

250.7.

{’21.

 

 

 

 

 

TMZLFFECT 0F £00120 SHI.LL '"OU.213 0F H£G§T11U OXIDE, 5001'32 CARHOHnTR
ORIGfiITBD QUERTZ SEE TO TdE C'““-IRE 5251003 ”EFOEE CU H1 HI {0 202 THE
222053 0F PREVEfiTI 2‘10 IEPUEITIES F502 E J2 1120 EITB T3: Ci? ’1 23 CEISTALS
3”311320 in the presence of
92919. Not sub1 imed. Eubli ed alone. fsynd
(by H.02t1r ) _ ..9 g ‘20. 3.0 QLLUJTtZ
By wt. of -y 2.22t2r. By 2t. of 2y 2. 2y wt. of 23 2.
"_---- ---------- 22222212- 22-2-222222-22222212 :2222: 2222 212- 22222.
'm--} -f\ P,410__f7 133° n f6, '3‘ t“
KCffee 4.541133. V.» .. u Lao L? U -zI‘o.
H1: C‘ 1‘0. ' ‘ ‘,I “f Ar [,4
1;; 0'317gg 0.0700% 0.01022 0.06402 0.00322
0
0 cup.) obuz’ ‘ 0U :1 0.0110 0.0107
14 0 0‘0 0 3‘20 0 “1 7
15. 0.0145 0.0125 0.0100 0.0076
16. 0.0187 0.012 0.0050
1.8. 0.0145 0.015240 0.0091:
A A 11" I"; 5' 7 ‘1’" 'L'F {‘1
twarage 0.01072 0.05:6” 0. 01152 0.02:52 0.00812

Pure caf-
feine. 0.00:0 go
I 1!

A ”a
0 JV

0.30 '19'

183 0.03i20g

0.00130

0.0”“ 0

CY
. a
L

J -1;
" 0.0050 0.00310
9 0.0010 0.00100 .0010: 0.00150
(Temp. 180-1900 c. for 10 hrs.
Pure c f-
Wflne. 0.0010 g. 0.0012 g. h.00038 g.
' " 0.0010 03103
" 0.002 0. 00f1 0.00194
" " 0.0023 0.00100
” “ 0.0022 0.00134
" ” 0.02i2 0.0020
5 n H 0.302; _ 0.00108 .
tankac.3,0.0401 0.029 2 0.0403 ;
Taf ee
52. 0.0114 2 0.011 2 0.0062 3
§6 0.0143 0.014 0.010:
.24 0 0150 0.020 0.0033
é? 0. 0125 0.012 0.0393
6‘ 0.01:9 0.015 O. 0082
Average 0.0186 2 0.0183 2 0. 0141 %

U.‘.,J18unoOu~iO 03,00 0 55.17
0.00100 0. 30270 0. H1 7
0.30242 0.232420 0. “"'rn
0.00075

 

51.13111? rd in tile

pr: »sence of 0.5 g.

t 1.0 goz§E.;:C03

135’ fits B; No
Gigttxr.

0.0013
0. 322

0.0322

C3,.0.“‘1‘r
0. 001127

0.045 2 0.0;37 2
,1. ,.F 0' .I
0.031 2 0.03:1 2

1

.l

O. DOC-8
0.0067

0.018
0.015
0.018
0.007

0.0305 0.0151 g

t“? .

.5---

-r..--... 70‘- 0-

~—_.—-. - --—
.
C

O I
D
O

O

I

—‘ “mfio‘

6E3.

Precautions observed to secure theggreate: t secure 0";

1. £11 apparatus used is made of pyrex gl5sc, and also all
glass were as far 55 pos5ible.
2. All distilled water used istniled about 15 minutes immediate-
ly before using, and it 15 tes ed often to in5ur2 its being
neutral.
3. Comm. 5:15 is riadi5tilled in the pre -ence of a 52211 amount
of caffeine, to remove any p055 ble C5ffe in e destroying
vomer it may have.

4. Heavy 230, 0.2.?. is ‘se~d at 2.11 times in the Power-Chesnut

method.

5. The 2202 used in the Fendler-Stuber method is acet mailide fr ee.

8. The samples are groung as finely as posaible. At least 95%
will go thru a 0.5 mm. sieve.

7. The sam;‘3155 are moistened thoroly before extraction in either
method.

8. 2/100 00304 and NBOH are atandarfiized agmzi st 8/100 SCI,
hich is made from constant boiling point 201. These solutions

are checked very often against fr shly orepared R/100 to in~

:1?)

sure them being correct. And also nitrogen determinations
are made often on usually 2 mg. s5mples of caffeine of
known purity as anothvr check on the procedure.

9. 10 cc. micro burete are used for measuring out the acid and
in titrating with the alkali.

10. Bl:rk determinations are 212w]; mude .uen a number of

samples are run, and the distillation apparatus is always

steamed at least 15 minutes before 2 scww 912 is distilled.

q - “r. v'.‘tq ..
t i - , ., .
Et)\pi_'tf1. 51.43.53! .. 5L

II. ANRLITICRL EEQULTS

The tabulated date, followed in some cases by brief
conclusions drawn from the results obtained, is given in the
following order:

A. Table 30. 8. Percentage of caffeine in szmoles of stfee Bag.

This includes the snal'sis of 35 5322125 of Kaffee Hag, as
well as 5 samples of Sanka Coffee. It also give the average %
of caffeine removed.

B. Table 30. 9. Determination of the oercentege of caffeine re—
moved in scroles of Eeffee Hag, finelzzod births Penile;-
Etuber method.
inelysis is made of 7 samples of coffee in 5 stages: The

green and undeceffeineted coffee, roasted but not deceffeineted,

and the roasted and dcceffe'sa ed coffee. Eoisturo determina-
tions are made and the results are based upon the moisture

free basis.

C. Table No. 10. Comgnrstige r

(0

sults of caffeine enslzsis Q; d ~

A

csffe sated coffee bx weight of the caffeine

residue, and by nitrogen determination, final-

ised by the Fondler-Ztuber and the Power-

 

Chesnut methods.
This table gives the analysis of 21 samples. 5nd the average
results obtained are very conclusive in showing the sportence

of the nitrogen determination.

  

D. Toble No. 11. gosocretive results of caffeine enel's

  

 

 

"rt... ..

 

 

6 ‘Atp

determination of the caffeine residue.
The results contained in this table and in table ho. 12
are considered by the sriter to be the most importent of any

given in this thesis. or while it ::5s3 e.¢ a: to show the neces-

w

51‘)

sity of the nitrogen determination, it w e no so easy to show
how much of the nitrogen thus determined came from caffeine.

Theser we Ml are very consistent and give the % of nitrogen

t}
m

s from th caffeine aloe Le, end not from the

(O

which really co.

impurities.

E. Table 30. 12. Distribution of nitrogenl calculgted as per-
centege of C;§§ei,e . in the sublimed crystgig
andzin theiremeinine residgg4

   

f‘pmw'vw'.‘ ‘ . s b . . 7‘“...

F. Teble Ho. 13.
tiensL calculated esApercentege of csffeinelnig
the residue from the firs t QfiClE extract! and
the residue from the final C3013 extract with
and withogt the oxidation steg of adding ;_2_

 

This table clearly shows the amount of nitrogenous sub-
stances other than caffeine which is extracted in the Fendlcr-
Etuber method.

:- 9: r1
{1 O; the ilflcg;

H.
F.)

G. Table No. 14. Fffect of furth~r onrificst

fl

 

 

 

caffeine res due by eddisgf23 cc. of water and
0.2 1.H~Q; size at: nj~, a short ti.s e on a steam

 

bath 5; filtering and shaking the Q:ueous so

IE“

tion with CHClngendier~Stuher method).
“a Table No. 15. Comgerstive results in adding 10 g. of EgQ to
the eggeous solution, digesting on the steam

bathI filtering and cooling:before oxidjging

 

 

 

_.

.o-

with EKncfi (Fendler-Stuber method).

I. Table No. 16. Congarative results obtained in shaking_out
the final chloroform extract with 1 fl K08, as
in the Poser-Chesnut method. (F.& €.method)

Tables No. 14,15 and 16 give some interesting results,

But they are of questionable value except perhaps the method

employed in table to. 18.

J. Table No. 17. toss of caffeine in the analysislfldetermined
bx adding 2 g. onnure caffeine to the residue
left after the first extraction and repeating
the analysis. (Both methods).

ts obtained when_e§;;ggting

b1 fioxhlet anbaratgs with alcohol ig the

K. Teble Ho. 18.

 

 

 

 

gggl, in the Porer-Qhesnut method.

0.;

This is a very good table showing the comparison be-
tween the two methods, and especially the two methods of
extraction.

L. Table No. 13. The comparative effectiveness of carbon tetra-
chloride as a solvent for caffeine extraction
in Kaffee Heg (Fendler-Stuber method).

This also gives the ratio of distribution of caffeine
in water and CC14o

E. Table No. 20. Comparative results obtained by using Soxhlet
chloroform extraction instead of shaking. in

the Fengler-Stuber method.

TABLE NO. 8.
PERCEJTAGE 0F CAFFEEII IN 8&2? ES OF HAFFEE HfiD.

(Average of the Fendler-Etuber and the Poxsr- Ch:.snut: ethods, and
based upon the nitrogen determination of the caffeine residue.)

 

Year 19:15" 3.5 of
5333333-??? ‘1‘?“:E‘-P 3533-53? ..... ‘3. 333151353“: __________ c armies
1. January Adams Grocery E. Larsin g, xich. 0.0109 g
2 February " fl 9 ” 0.0105
4. " “91105: Conpany Battle Creek “ 0.0 $45
5. I! H I! 1! t! I! 0.0%“8
6. " " fl " " a 0.0198
7. ' " “ ” " " 0.0034
8. April Adams Grocery E. Lansing, “ 0.0175
9. " 'A. a P. . ” “ " “ 0.0141
10. ” " ” Detroit, " 0.0139
11. " " " Clinton " 0. 0147
12. " firbau he Deot.ET.Lansing, " 0 0“2
13. key Rite Vay Store Jackson " 0.0199
14. u A. a P.0rocery ' E. Lansing, " 0.0194
15. June Delicatessen
' Grocery Larsing, " 0.0145
16. " Gooseons Groc. " " 0.0147
17. ” Kroger " " ” 0.0148
18. " C.Thomts Stores E. Lansing " 0.0125
19. ” E.R.Quigley Groc. Clinton " 0.0143
20. (19”?) " Kroger Store Flalerville, ” 0.0348
21. 9 Rita icy :‘yztem Detroit, " 0.0135
22. July Earner Stores Lansing, ” 0. 0103
:5. n For Eros.0roc. E. Lansing, " 0. 359
2:. " KrOger Store Charlotte, " 0.0155
25. " Rite say System Eellvue " 0.0170
28. " Aldrich Grocery Tecumseh " 0.0161
27. " Kroger Store ‘Howell “ 0.0117
28. R A. 0? p. Store BrightOn " 000110
29. n ” Eason " 0.0140
5 . Angust C.Thoras Etores Horth Lansing " 0.0160
51. " F.A. Rozze {tore " " a 0.0150
3?. “ Bop p's dark €;t Lansing, ” 0.0121
55. " Piggly‘ irgly W. " " 0.0143
5.. " Economy Store Garrett, I diona 0.0164
325. fl 1’” 32 P0 Store [31.90th " 000154
36. ” A»-.iCuu otcres Coldeeter, .;ich. 0.0158
Averaee~ 0.0180 %

Eercgntagg of caffeine removed, brs ed upon the undecaf—
fe nate coi ee containing an average of 1.20% caffeine ~ 98.50 5

This is well above thee company's guarantee of 97% removed.

Sank Coffee

$0.1. April,1939 Bunter'o Grocery Lansing, Xich. 0.0250 5
2. June " Fox Bros. " E.Lansing,%ich. 0.0593
5. August " ' " " ” " ' " 0.0440
Avery... Pie" 0-0554 S

Percentage of caff ine remov d, be: ed upon tne unde-
caffeinted coffee conta inine an a craze of 1.20% ceflf cine-97.05 %

‘".-_

q...

- n...

at-

15.7

IR

TJ'BLE 1‘; ’3 a 9 t

CEfiTfiGE OF CAREEIfiE HEROVED

DETEREIEATION OF THE PEE

552?:WLFS OF KRFFEE HAG.

63621233

BY THE- F'E'ZféDLER-STUBEF. 25.317300.

 

Sample % of

 

E100
Roasted
&
Decaffeinated
61. 4.77
62. 4.59
63. 4:09
64. 4.2
65. 4.32
66. 4.20
1461 3.86

Average-4.29 2

green 4 not
Decaf einated.

 

 

61. 8.38
62. 8.68
63.‘ 8.41
64. 8.14
65. 7.87
66. 7.75
1461. 8.65
Average-8.26 i
fioasted
& not
Decaffeinated.
61. 4.10
62 3.74
63 4.20
64 4.07
65 4.07
66. 3.86

AVEFSES‘ 4.00
Deca?feinated

1461.

7.09

9.590
3.620
9.580
9.506
9.593
9.614

9.593

9.291

“to or 112013"
Eoisture tare free
5513:3130

*-_----- --------¢--

‘63

cf caffeine % of caf-
feine bas.d removed.£ased
Upon meistura

ased upon
he 10 gm.wt. Upon the
moisture
free wt.

0.0123
0.0160
0.0809
0.0184
0.0134
0.0137
0.0148

0.0153 E

O
a

Q)
G)
03
19%.

1.030
1.066
1.095
1.091
1.103
1.086

1.077 S

0.0212

0.0128
0.0167
0.0217
0.0193
0.0129
0.0133
0.0155

0.0160 g

1.036
1.021
1.107
0.361
1.152
1.123
0.878

1.048 %

1.063
1.110
1.143
1.137
1.150
1.129

1.122 4

0.0228

S of caffeine

From
green

free wts.

coffee

98.84
98. 27
08.04
98.00
08.88
96.86

08.36 '

98.46

From

roasted
coffee.

98.80
08.50
38.10
98.31
98.88
98.84

 

-9-..¢,

a O
.
n
n .
u
.
o
b
.
u c
I 0
' .
O O
C 1
‘ . ‘
g, V
‘
\
a -4 ., 1
§
,_ . 7 -- -
O 0
t
I
o
v
v C
. .
wc. —.
..
I O
-- ._- - w . - - . 7 _--

COEPAEATIVE tESUL

COFFEE BY WEIGHT OF THE CAFFEIJg

TABLE NO.

T8 or CéFFEIXE ANALYSIS OF DECEFFEINLTED
11.1012, asp my r1

TEOGEH

DETEEEIRATION. AHELYZED BY THE FEHDLER & STUAER AND THE

130m

01?.“

& CHESIUT fiETHODS.

 

Sample 80.

~ Kaffeo

Hag 1‘30.

17.
18.
19.
21.
22.
25.
24.
25.
26.
87.
28.
29.
30.
51.

3?.
34.
35.

36.

Off:

Om

Had

moot-aw

0 (is?)
(D

Avc.0.045

:2
O
o

Residue.

0.064

0.060
0.068
0.064
0.02

0.056
0.064
0.058
0.048
0.040
00040
0.042
0.046
0.048
0.058
0 .045
0.054
0.044
0.040

0.075
0.084

Aveu’000795 %

.~--‘-----------------‘---------—‘-—-----------‘-‘-------~-.

Total ave.-0.0487 %

Percentage of caffeine
in the final residue.

%

0.0148 %

0.0125
0.0145
0.0106
0.0259
0.0153
0.0164
0.0146
0.0106
0.0038
0.01.2‘9
0.0157
0.0141
0.0114
0.0145
0.0150
0.0125

0.01.2:~

0.0141

0.0424
0.0461

0.0442 6
0.0170 i

64.9 %

~~--~-~-‘-

Percentage of caffeine.

Fondler & Stuber method.
By wt. of By nitrogen
determination.

--‘--~~-~‘-‘-

residue.

By wt. of Ey nitrogen
determination.

0.052 9 0.0143 g
0.058 0.0164
0.065 0.0904
0.058 0.0176
0.075 0.0518
0.057 0.0157
0.060 0.0177
0.05 0.0176
0.055 0.0129
0.048 0.0123
0.054 0.0159
0.048 0.0165
0.045 0.0158
0.039 0.0128
0.048 0.0144
0.058 0.0179
0.059 0.0145
0.043 0.0136
0.0551 5 0.0167 g
0.060 0.0363
0.077 0.0450
0.0685% 0.0531 %
0.0564 x 0.0183 g
35.: g

Ag

68.

Conclusions- These results prove cgnclusively that to base the
6 of caffeine upon the It. of the residue of caffeine would be
entirely erroneous, and would give results about three times too

high, no matter which method is used.

G)
(.1)
a

TABL

£13

30.

COXPAEATIVE FESULTS OF CAFFEIHE ERALYCIS DETAIBED B
.30 TEE EITROGEE DETL." fiATICH 05' T12 CAFFEISEE

 

Sample. Percentage of caffeine.

----““----’----‘------‘-------’------~--~-------~~‘-------’-~.

Fendler~$tnber method. Pow3r - Cheenut method.

--------—‘---‘~-~-~-- -fi---~‘--“----_~---'---

By sublimation. By nitrogen By sublimation Ey nitro-
“~I*~--~~~‘--~- :‘letrjr. u--- mmmmmmmm at...“ gizn Getter.

5:, “91 ght. By: godCtEr. By “Qightfi 9y R-

fiat-3r O

Kaffee
lib-g .310.
170 0001.4. O-Jl‘u-J. 0.01.sz (3001:. 0.0130 0.0168

18.
19.

0.012
0.014

0.005 7
0.0103

0.01?7
00 3/15..)

0.010
0.015

0.0072
0.0007

0.0120
000185'

310 0.01.2 0.0101 000}... 0. 11:2. 000114 000:09
22. 0.010 0.0080 0.0111 0.012 0.0065 0.0104
-?3. 0.030 0.0550 0.0273 0.094 0.0322 0.0270

30.

0.018

0.0113

0.0157

0 I O 00

0.0098

0.0180

El. 0.007 0.0104 0.0141 0.015 0.0086 0.0158
32. 0.011 0.0065 0.3114 0.031 0.0052 0.0128
35. 0.014 0.0102 0.0140 0.015 0.0074 0.0144
34. 0.020 0.01Sl 0.0160 0.018 0.0125 0.0l73
05. 0.012 0.0006 0.0128 0.030 0.0104 0.0145
Eanka
Coffee So.
2. 0.034 0.0360 0.0424 0.007 .0138 0.0005
3. 0.033 0.3433 0.0461 0.043 0.0310 0.0620
Average- 0. 0174? 0. 0150 % 0.0188 0.0303 % 0.012 0 0.01955

 

 

 

 

Chase & By wt. 33 *uLYIWhti: .
Santa! it. 01 0133 :13 fig 2 deter. ~.u.ter.of Total
Coffee of c yrtAls re:idue(calc'd)fl.det0r.
as S caffeine).(calc'd
as g
------- --_--_--- ............... - .......... --c;ffci;2)
.21? 1.006 1.04? 0 0.0704 § 1.197% f
{01‘ ii :2.‘
By nt. By L.D¢ter.
0.0020 gme. Pure c..feine—o.03‘lj Ems 0.00 34 5mg
" " fl 0. 03.. ” 0.00133 ”
fl 9 " v 0.00 I: " 0.03194 3
" " " 0 .03? S " 0.00200 "
V ” " " 0.0“”‘ " 0.00138 "
0.0010 ” ” " 0.0012 R .00039 ”
“ V V " 0.0010 “ 0.0310 R

H<---

,.-

7o,

 

C: EBEIPIE, I3 TEE EUEL! 1:60 CFYSTALL 9’: In TF5 pgkflIdIN‘
HBSIDUE.
Sample. Percentage of caffeine.
Fendler - Stuber method. Pow+r — Chesnut method.
In the crystals In the In the In the
---------- residue:w 95:3t91". residue
effee
flag no. .
17 0.0102 x 0.0030 % 0.0100 0.0063 %
18. 0.0097 0.0030 0.0072 0.00485
19. 0.0133 0.0080 0.0097 0.0088
21. 0.0151 0.002 0.0114 0.0035
22. 0.0080 0.0031 0.0063 0.0044
23. 0.0853 0.0020 0.0222 0.00485
24. 0.0113 0.0044 0.0036 0.0067
31. 0.0104 0.0037 0.0086 0.0072
52. 0.0063 0.0051 0.0052 0.0076
33. 0.0102 0.0041 0.0074 0.0070
34. 0.0121 0.0042 0.0125 0.0054
55. 0.0096 0.0 22 0.0104 0.0041
3 . 0.0095 0.0033 0.0085 0.0041
Sanka
Coffee No.
8. 0.0669 0.0055 0.0298 0.0065
5. 0.0403 0.0058 0.0510 0.0110
Average~ 0.0150 5 0.00:6 Z 0.0157 5 0.0066 %
Average total
% of caffeine- 0.0186 5 0.0135 %
i of non caffeine.nitrogenous
material in the unsublimed
caffeine residue- 19.3 g 34.2 %
S of nitrogen from
caffeine only ~ 80.7 % 65.8 %

 

r.-.

71.

Conclusions from tsbles 11 a 18. -

In the Fendler-Stuber method the % caffeine by wt.
of the sublimed crystals is about 0.0011% less than the %
Caffeine by nitrogen determination of the residue before sub-
limation. therees in the Power-Chesnut method it is 0.0017?
more. It was found to be a little more difficult to obtain
the pure colorless sublimed crystals from the residue obtained
by the Power-Chesnut method than it was by the Fendler- Stuber
method.

It is a striking fact that the i by wt. of the sub~
limed caffeine correSponds very closely with the nitrOgen de—
termination of the unsublimed residue. There are consistently
lower results obtained when the results are based upon the
nitregen determination of the sublined caffeine than when they
are based upon the wt. of the sublimed caffeine. 'This is
also true when pure caffeine is sublimed. There is a greater
possibility for error in weighing then in running the nitrogen
determination.

It will also be observed that while the Power~
Chesnut method usually gives higher results than the Fendler-
Etuber method when they are based upon the nitrogen determina-
tion of the caffeine residue, it gives loser results when
based upon the nitrOgen determination of the sublimsd caffeine
(Average 0.0021% less). And the nitrogen calculated as i of
caffeine left in the residue after sublimation is 0.005 x high~
er in the Power-Chesnut method than in the Fendler-Stuber

method.

Therefore these figures indicate that the Power-
Chesnut method extracts caffeine slightly less efficiently
than does the fencler-Stuber method, but extracts more non
caffeine nitrogenous materials, making it more difficult to
get the pure caffeine crystals by sublimation.

The biggest feet that these tables bring out is thnt
19.3% of the nitrogen determined and calculated as caffeine in
the final caffeine residue in the Fendler-Stuber method, and
34.2% in the Power-Chesnut method, is not from caffeine at all,
but from other nitrogenous materials.

or course, the question may be asked if it is not pos-
sible for a small amount of caffeine to be held by the residue
even after subliming a long time. This is possible, but not
very probable for the following reasons:

1. Later tables show thet non caffeine mitrOgenous
materials are extracted in the first CBC13 extraction, but are
more or less eliminated in the various steps of purification.
And it is reasonable to conclude that the impurities that es~
caped the purification process would also contain nitrogen.

2. e sublimation of pure caffeine results in nearly
a 100 % recovery, and at a much shorter period of time than
the samples are run.

3. Continued heating at a much higher temperature

(255-2450C.) does not cause even a trace of caffeine to sublime.

73.

TABLE mo. 15.

COEPARATIVE RESULTS OF THE aITeoezn DETERm NATIOHS,-CALCUL£TED
AS PEECEETAGE 0F CAFFEINE, IN TEE RESIDUR PROS THE FIRST 08313
EXTRACT AHD THE 8531002 Page TEE FieAL 03017 EXTEéCT, eITH

Ll
AED EITSOUT THE OXIDATIOH STEP OF ADDING l % KEnOé TO THE
AQUEOUS S LUTION. (Fendler-Stuber method).

 

Sample. Percentage of caffeine by nitrogen determination.
First residue. Final residue Final residue
-99£-9§i§i§eé;_ oxidized.

---‘--‘- ----‘-~vnn-----»--

Kaffee Beg

d----------‘---

I730. 89 000853 :2: 0.0431 % 0.016:- 7::
.10. 0.0704 0.0530 0.0125
11. 0.0804 0.0208 0.0129
Sanka Coffee
No. 1.0 0.0769 0.0403 0.0225
Poetum Cereal.
HO. 1. 000672 000276 0.0059
Average- 0.0774 S 0.0546 S 0.0137 %

k

Conclusions-

Over half of the non caffeine nitrogenous materials
are removed by simply putting the-residue in solution with
water, filtering and shaking the aqueous solution with 6 per—
tions of CHC13. But about five sixth'e of it is removed if
the complete Fendler-Stuber procedure is followed.

Therefore it is reasonable to conclude that the im-
purities in the final residue of caffeine, which is only about
34% pure caffeine, also contains non caffeine nitrogenous

substances.

TABLE no. 14.
EFFECT or FURTHER PURIFICATION OF 2:: FIM CLFFLILE HESIDUE BY
ALDILG 20cc. 0F LLTEL LLB 0.2 Gm. 01 L193, E‘IGEELIEG L cue"?
TILE on re: LL EATS , FILTLLILG AND SHAKIHG Tee AQUEOUS 00LU~
TIOH Lira cnciz. (Fendler-Stuber method).

Sample. Percentage of caffeine by nitrogen
determination.
OFDILLFY F. L E. LETZSD. F.L ES. L‘Tfioi,f“IIU’ DUTIIIIED

..................................................... £2:iéel-----

Kaffee Bo Bo. . , .
g 2 By wt. of resioue. By nitrogen

deter.

Keffee Hag We. 2. 0.0039 % 0.0580 % 0.0079 %
" " 8. 0.0162 0.036 0.0160
5 fl “ 9. 0.0122 0.066 0.0100
” " " 10. 0.0127 0.02? 0.0039
” " " 11. 0.0123 0.030 0.0103
“ " " 12. 0.0314 0.050 0.0145
aka Coffee” 1. 0.0224 0.02 0.0158
LWO tum Cereal 1. 0.0057 0.016 0.0035
Average. 0.0141 0.051 0.0085

- .------‘-----------------*-------‘—---—-----~-------‘~-------.

COR 5051035-

The consistently lower results obtained by purifying
the caffeine residue the secor d time with magnesium oxide may
be due to two reasons.

1. The possible slight caffeine destroying power of
the ego.

2. The removal of non caffeine nitrOgenous substances.

ther ext. indicate that the caffeine destroying

power of Ego is negligible. Also, the supposedly caffeine content

of Postum Cereal, which has no caffeine in it whatsoever, shows

75.

about one half the amount of nitrOgen as it does by the
ordinary Fendler and Stuber method.

The final caffeine residue appears to be nearly
pure, as observed in the bottom of the flask, showing the
characteristic white needle shaped crystals with practically‘
no fatty or waxy material present.

Ehile this extra step would probably not be practi-
cal for commercial analysis, it shows that some such method
even better than this perhaps, should be employed if we are to
estimate all nitrogen as coming fr m caffeine in the final
results. It is probable that from 10 to 40 g of the nitrogen
determined when the ordinary methods are used is from non»

caffeine nitrogenous substances.

76.

C0"PLRATIVE 82$ ULTS- IN EDDIEG 10 G. “GO TO THE AQUEOUS SOLUTIOE,
DIG ETIHG OR I? E ETELL LLTH, IILT': EIQG f.iD CSOLINGE IDEE CEI-
DIZIEG EITH POT}.E WEI 31 3“""f IGERLT 1L.

 

Pewrcent ge of caffeine.
Sample. Ordinary F.snd S. proc dure. F. and S. “ethoo Litn M :0.

------- .-----‘------~----—-~-”*-'- -‘--‘~‘----~~--------‘-~-

By wt. of By nitrogen B" at. of Ly nitrogen
residue deter. residue. deter
Keffee nég
No. 16. 0.0147 fl 0.0190 % 0. 0213 i
17. 0.034 0.0148 0.0100 0.0106
18. 0.030 0.0125 0.0160 0.0097
10. 0.062 0.0143 0.0?00 0.0127
20.(1920) 0.256 0.0948 0.0860 0.0995
21. 0.034 0.0135 0.0100 0.0100
22. 0.0150 0.0100 0.0170 0.0065
25. 0.055 0.0259 0.0420 0.0919
24. 0.064 0.0185 0.027 0.0115
25. 0.058 0.0164 0.032 0.0141
Average- 0.065 % 0.0255 % 0.027 % 0.0213 %
Pure
caffeine 0.00191 5. 0. 00121 g. 0.00155 g.
a n - 0.00162

_‘_.___ ##44 “—

TEBLF R0 0 160

00L?LELTIVE EESULTS OBTAIILED 1L 5L 'ILG OUT Te: FIL LL CHL0L0:- OF L
EXTRACT LITE 1 5 K08, AS IR THE Povnn—CEL SLUT L THOD.

 

 

Sample. Percentage of caffeine.
Ordinary F. S.method. F. L 5. method plus above step.
By st. of By nitrOgen Ey at. of by nitrogen
......... ¥§§1§221- -02Esz:----- -2221992;-- -Qszszz-----
- 19. 0.062 L 0.0143 5 0.051 i 0.0153
21. 0.054 0.010 0.028 0.0157
22. 0.024 0.0105 0.024 0.0080
25. 0.053 0.0259 0.007 0.0? 5
24. 0.064 0.0155 0.054 0.0145
25. 0.058 0.0164 0.047 0.0155
26. 0.048 0.0146 0.056 0.0142

fiverage—_ 0.049 % 0.0157 5 0.034 % 0.0154 %

 

-..o-

“a...

.-a-.-_ .~--o ~UI“

77.

Conclusions from tables 15 end 16.-

The digestion of the aqueous solution with Ego and
filtering before proceeding with the usual step of oxida-
tion with Kline4 greatly increase the purity of the final
caffeine residue. In some cases the crystals of caffeine
in the bottom of the flask had nearly the same appearance
as pure caffeine. However some samples will have over twice
the K of caffeine by wt. as by nitrOgen determination. Con-
sequently the wt. of the final residue could not be relied
Upon for calculating the % of caffeine. here is also a
slight loss of the alkaloid by being Shsorbed by the Ego.

Washing the final CHCla extract with 5 cc. of XOR,
as in the Power~Chesnut method, improves the purity of the
caffeine to quite an extent, but not as much as does the
use of Ego. It has the advantage of not having any destroy-
ing power On the caffeine. The use of this extra step on
samples that are to be sublimed considerably improves the
purity of the sublimed crystals. The result in table 15
show a decrease of 0.015 % caffeine by wt. by the use of KOH,
whereas table 15 shows a decrease of 0.038 % caffeine by the

use of Ego.

78.

TABLE RO.~17.
LOSS OF CAFFEINE N THE ANALYiIS, DETERXIEED BY fiDDIEG 2 fig
0F PURE CAFFEIHE TO TfiE EESIDUE LEFT AFTER THE FIRST EXTRAC—

TION AED REPEATIRG TEE ARéLYSIS.

 

fine of caffeine by the nitrogen determination.
sample“ Poserzghsssui_eetbsss Essélsr:§isbsr-ssieed.

' (First 03015 filtered
thru cotton plug.)

0.00200 g. 0.00160 g. 0.00202

" 0.00139 0.00187

" 0.00120 0.00132
sverege- 0.00140 g. 0.00193 g.

(First C8015 extract
filtered thru a 24 cm.
filter paper).

 

0.00200 g. (CHCIT evaporated off 0.00181 3. (Res fil~
without shaking with ter paper)
Kaffee Hag residue). 0.00194 g. (01d fil-
ter paper).
Percentege loss of caffeine- 30.0 § 0.5 %
Conclusions~

hhile it would not be Justifisble perhaps to draw con—
clusions_obtained in the above table, the results shown in some
of the other tables involving the use of Mgo strongly indicates
that the ago has a slight retentive or destroying power on the
caffeine. And there is no doubt that lower results would be
obtained, in nearly every trial, with the Power-Chesnut method
than with the Fendler-Stuber method.

The explanation might be made that perhaps the reason
the Fendler—Stuber method gave such a high recovery of caffeine
was because caffeine had not been completely extracted from

the Kaffee Hag. This is not true as the last two samples were

 

~~

-——-o—

F..-

not added to the Keffee Hag residue at all, and they gave
nearly as high results as the others. It is also interest-
ing to observe that the higzest recovery was obtained when
the CECI:5 extract was filtered thru a filter that had pre-

viously filtered a CHClZ extracts and had been thoroly washed.

80.

TABLE 30. 18.
coeennarxve RESTLTS OBTAIRED BY EXTRACTING 51TH eoxeter
APPARATUS WITH ALCOHOL IN THE ‘FHDLER-STUBER hereon, AND BY
SHAKIno WITH CHLOROFORH IR THE POhER~CEESRUT METHOD.

4

 

Sample. Percentage of caffeine by nitrogen determination.
Ordinary F.& S.method Ordinary P.&.C.
F.& S.method with alcohol ?.e C.method. method
extraction. with CHCl~
(shaking)0
----..----- ................................... 2332601190:
Keffee Hag e1. 0.0114 %
00. 1. 0.0122 % 0.0160 % 0.0091 % 0.0109 %
1. 0.0121 0.0103 0.0091 0.0116
2. 0.0111 0.0107 0.0067 0.0099
4. 0.0245 0.0591 0.0246 0.0225
5. 0.0326 0.0319 0.0550 0.0500
6. 0.0207 0.0155 0.0190 0.0177
7. 0.0112 0.0157 0.0076 0.0101
7. 0.0111 0.0181
8. 0.0162 0.0394
9. 0.0122 0.023
10. 0.0127 0.0255
11. 0.0129 0.0335
12. 0.0214 0.0501
Sanka Coffee
do. 1 0.0224 0.0275
(Sample 1 to 7)
Average 0.0178 % 0.0193
Total Average 0.0166 i 0.0251 5 0.0156 t 0.0155 %

Conclusions-

These results point toward the same conclusions as are
reached in some of the other tables, namely: The Power-Chesnut
method gives the same and uniformly lower results no wetter
which method of extraction is employed. The reason for this can
well be eXplained by the slight caffeine retentive power of the
Ego used for purification. Also, this method with alcohol ex-
traction removes more non caffeine nitrogenous materiele than
does the shaking method with CHClE. Thus the Fendler-Stuber

method gave 0.0065 fl higher results when alcohol Soxhlet extrac-

81.

TI'BLE H0. 19 o

3: co: :ELPLTIVE “”“CTIV“NTSC 0F CL? 543 ITET
FOE CLFFEILE EXTELCTION IN KLEFEE sLO. (FL 5

ASOLVEHT
‘- \ :¢:I§Tfi0D) O

('1 '11
T?"

'U

R“
”1'
F9 _
c;

{3

P‘

III,

A

 

Sample. Percentage of caffeine by nitrogen determination.
Ordinary F.& E. F.& S. rgethod, Soxhlet extrac-
method (using 200 (using 200 gm. tion with 0014.
gm. CECl. wet with C014. Let with 5 for 10 hours
cc. 10% L04OH cc. alcohol Sample wet with
shaiting 1 hour). shaking 1 hour) 10 cc Later.
’ Of .e
Koffee Lag No. 15. 0. 0179 E 0.0065 E :YCLtf§%n
" 9 u 14. 0 O? :00 0.0073 residug
" " " 15. 0.01548 0.0056
" " “ 16. 0.0147 0.0068 0.01095 0. 02$:fi
" fl " 17. 0.0148 0. 0065 0.0103 0.055
' " ' 18. 0.0125 0.0084 0.0101 0.058
" a n 19. 0.0145 0.0116 0.095
3 R I 20. 0.0948 0.0754 0.084
n ” “ 21. 0.0155 0.0087 0.024
Average of first
six samples 0.0155 g
Average of last 6
samples 0.0274 % 0.0069 5 0.0211% 0 05665
Conclusions-

The extraction of caffeine by CCl4. by the shaking method
is very incomplete. And the final residue of caffeine is only
about one half caffeine by its appearance. Hence, this modifica-
tion can be regarded as unsatisfactory.

5e find much different results when Soxhlet extraction
with C014. is used. And while the appearance of the caffeine crys-
tals shows the presence of some oily material it is less than when
the ordinary F. e S. method is used.

This is also proven by comparing the last two columns
which shows that the % caffeine by wt. is about double that deter-

mined by nitrogen. whereas, it is from three to four times as

great in the ordinary Fendlor e Stuber methodl

As undoubtedly, he foreign material present in the
final residue of caffeine contains nitrogen, the elimination
of the most of this material “ill give us more correct
results. The agreement of the above figures with other meth-
ods of purifying this residue will help greatly in the solu-
tion of this problem, as well as show that the caffeine is
not destroyed in the modifications or additional steps used.

25c ratio of distribution of caffeine in solution
in equal volumes of water and carbon tetrachloride after be-
ing shaken thoroly was found to be: 5.95 parts to 1.00 part
in water and 0014. respect. and: 1.00 part to 01.00 parts

in water and C8013 respect.

TABLE NO. 20.

COL‘ 'PkfifxTIVE RESULTS OBTAILED BY USIKG THE SOXRLET CHLOROFOEH
EXTRACTION IRE TEAD OF SiALIfiG, IN THE FEEDLEE RED STUBER
HETBOD

Samples. Percentage of caffeine by nitrogen determination.

Shaking with CHCl:5 Soxhlet 0E015 extraction.

.-----‘-----------’--“.

Kaffee Hag ho. 8. 0.0102 % 0.02 94 S
" fl 9. 0.0192 0.0231
" " ” 10. 0.0127 0.0?63
'1 R I 11' 000153 0001505
' " " 12. 0. 0014 0.0801
Senna COffea w. 10 00 0:: ’34 000273
Average. 0.0165 0.0278
Conclusions-

The consistently higher results obtained by the
Soxhlet extraction does not necessarily prove t? at the shak-
ing out method is inadequate in extracting all of the caffeine.
But because of the increased amount of meteriel carried thru
in the final residue when the onhlet extraction was used,
and also its darker color, it would seem more logical to con-
clude that the Soxhlet extraction method removed much non
caffeine nitrogenous substances than does the shaking out
method. This may be also true in the Power-Chesnut method

where alcohol Soxhlet extraction is used.

v-.— -,

   

 

 

.- *Efi-O. *-- —.

..

o- L‘-

- ‘-—

B.

84.

SUfiKAHY
I. EETHODS

The Fendler-Stuber method is preferred to the Power-
Chesnut method for the caffeine analysis of deccffeinated
coffee for the following reasons:

1. It gives fully as accurate and consistent results.

2. This method is slightly more efficient in the extrac—
tion of caffeine.

3. It extracts slightly less non caffeine nitrogenous sub-
stances, end less is carried thru into the final caffeine
residue.

4. The method is much shorter, more convenient and results
in less possibility for error, esoecially with minute
amounts of caffeine present.

5. The necessary apparatus may be found in every laboratory.

6. The method of purification employed removes a smaller per-
centage of caffeine when analyzing decnffeinated coffee.

7. The sublimed caffeine crystals are more nearly pure caf-
feine, as determined by the nitrogen content and their
appearance.

A micro Kieldehl procedure, similar to Pregyl's method us-

ing steam distillation, is one of the most accurate methods

of determining nitrogen in the caffeine residue obtained
from decaffeinated coffee.
macro Kieldahl apparatus gives fairly reliable results if

micro methods are employed.

C.

85.

If the caffeine residue from decaffeinated coffee is sub—
limed at a temperature not exceeding 185° C. for at least
10 hours, results can be obtained which agree very closely
with the nitrogen determination of the caffeine residue.
The nitrogen determinations of the sublimed caffeine gives
the most accurate and true results of any procedure yet
tried.

Sublimation proves that the final caffeine residue con—

tains non caffeine nitrogenous materiels.

{11

86.

”T2131; er

II. ngLYTICAL FFSJLT s (Averages )

35 samples of Keffee Hag analyzed gave an average of 98.5fi
caffeine remov ved, based upon the undeceffeineted coffee
containing 1.2% caffeine.

Keffee flag was 98.46% and 98.57% caffeine removed when

beeed open the green and roasted coffee respectively from

which it YES taken, in 7 enemies.

34.9% of the final residue consists of ce.ffeh 1e (By R.deter.)

by the Fondler-Stuber method, and 53.5% by the Power—

Chesnut method.

% of caffeine by weight of sublimed caffeine and by nitro~

gen determination was practically tie see e no matter which

method was used. But % of caffeine by nitrogw determina—
tion of sublimed caffeine was less in the Power-Chesnut
method than in the Fendler-Stuber method.

80. 7% of the nitrogen determined in tr e caffeine residue

comes from the caffeine only, when r in by the Fendler~

Stuber method, and 85.8% when run by tr e Doeer-“heknut

method.

There was found to be over 5 times as much nitrop en in the

residue from the first CHCl3 extract as there was in the

final caffein residue when purified and oxidized, and

over twice as much when not oxidized as when it is oxidized.

Hence there must be a small amount of non caffeine in

nitrogenous materiel in the final residue.

G.

I.

J.

L.

‘R
on.

87

The use of £30 to furt tier purify the final caffeine residue
does more harm than good, in retaining a small amount of
caffeine.

The use of LgO in purific: tion be re oxidation with Khno4
considerably incree see the purity of the -sffeine, but it
is not consistent enough so that tne h of ceffeino can be
seed upon the weight oi the residue

beehing the final CTCl '*rect with if KOH lowered the im~

r
U

purities noout 50. fl and met3rielly' increases the purity

'of sublimed ce .ffeine.

There was s 30.? loss of coffe ine (5? mg. sample) by the
Poser-Ch-snut method, and only a L.5fi losL with the Fendler-
Stuber method. This supports other results pointing to—
ward the slight c: ffeine re wt ini T? posero “f £50 esoecially

IQ.-

noticeable with minute quantitaties of ceffe in

Alcohol extraction sith the Soxhlet apparatus does renove
more non caffeine nitrOgsn sub: tcnce s than does shaking
with CHClZ. The final results are somexhat evened Jp in
the two methods (when based noon the nitrogen determination)
by the slight caffeine retaining power of the Ego.
Extraction of caffeine with C014 was very incomplete by the
shaking method, but it was practically complete when the
Soxhlet extrector has used, and resulted in a more pure
caffeine residue than nhen the regular Fendler-Stuber pro-
cedure was sod.

Soxhlet extraction with CHClz rGS‘ lted in much higher results
by the nitrogen determination than by the shaking method.
This is undoubtedly due to the extraction of considerably
more impurities which contained nitrogen, as the caffeine

residue was in a verv imoure condition.

88.

)

RgT COXXTVDATIONo

It is recommended that the procedure of caffeine analysis
for coffee by the official Power~Chesnut and the tentative Fendlor-
Stuber methods , as given in the Officis l and Tentative methods of
knelysis of the Assoc. of Official lgr. Chemists, second edition,
lsné, be given the following modifications em; lyin g Spec cificnlly
to the caffeine analysis of decaffeinsted coffee:

1. The weight of the final residue of caffeine shall be en-
tirely dier:’ gerd ed in calculating the percentage of caffeine.

2. The r roentzge of caffeine shall be based ujon the accu-
rate nitrogen determir ati on of this caffeine residue.

3. Ltlcro Ejeldehl methods shell be used for the nitrOgen de~
termination. The use of ordinary Kieloshl apparatus shall be nor-
missible, providing at least H/7O standard solutions, and an indi-
cator as sensitive as s:oth yl red srLe used, and also nosing it nec-
esse ry that blank determinations be made just before senol
run that do not very more than 0.3c cc. of N/lOO acid used.
it can be proven to be comnlcte; on” shall be esoecielly reconnenn—
ed if combined with the nitro we determination of the sublimed

caffeine.

,4.

Finally, he CHIOI e of an accurate m ore method of
caffeine analysis for deceffeinsted coffee is the Fen Lilaar-Stuber
method sith the suggested modifications, employing micro procedure
in all of the steps, and basing he gercentsge of caffeine upon

he micro nitrogen determination of the sublined caffeine.

AW .WQ

llIllllllllllllllllllllllllllllllllllllllIIIIHIHINIIHIHHll

 

31293 03037 9774