Classification of Starch Derivatives
Significance of the
Barium Hydroxide Method
G. V. CAESAR AND M. L. CUSHING
Stein, Hall & Company, Inc.
T HE use of the hydroxides of the alkaline earths to purify
sucrose has long been known, and barium hydroxide has
been employed commercially for this purpose in Italy.
Browne (2) states that “the structural constitution of these
and other saccharates is not as yet understood, the place and
manner of attachment of the base to the sucrose molecule
not being established.” This was written in 1912, prior to
the present acceptance of other forms of valence bonds than
direct linkages.
Sidgwick’s postulation (6)of the covalent coordinate bond
involving a transfer of electron pairs from a donor to an
acceptor goes far to explain the formation of a barium sac-
charate. Some such structural arrangement as the following,
in which barium may hang from one or more of the hydroxyl
units like globes on a chandelier, is indicated:
CHzOH
H I
0 The results of the tests reported here tend
H to prove that the barium hydroxide method
CHzOH
I
CHzOH
L&* . O i .coordinate link ages
\/
Ba
It is reasonable to postulate that the settling or precipitat-
ing action of barium hydrate on sucrose is caused by a weight-
ing down of the sucrose molecule by the heavy barium atom,
attached by coordinate bonds.
I n 1918 Babington, Tingle, and Watson proposed a method
of evaluating starch products ( I ) ; it was based upon the
precipitation of “starch” with a specified amount of saturated
barium hydroxide. A transcript of the original method
follows :
DETERMINATION OF DEXTRIN GUM. The sample (1 gram) is
warmed in a 100-cc. graduated flask with 30 cc. of water until
just gelatinized, and cooled quickly. Fifty cc. of a cold saturated
barium hydroxide solution is next added (the flask being mean-
while shaken), followed by enough water to bring the total volume
to 100 cc. The solution is filtered through a dry 15-cm. paper
and an aliquot portion (50 cc.) of the filtrate is pipetted into a
platinum dish. After the addition of two drops of a 1 per cent
phenolphthalein solution, 0.1 N hydrochloric acid is added cau-
tiously till the neutral oint is just passed. A faint ink color
is then restored by the agdition of two or three drops of t\e barium
hydroxide solution. A weighed quantity (about 10 grams) of
sand is added, and the dish heated on a water bath. The sand
should be stirred when almost dry, to expose the maximum sur-
face t6 the air, after which the dish is transferred to a well-
regula&?oven and dried t o constant weight at 120” C., below
which temperature crystallized barium chloride is dehydrated.
Dextrin gum being hygroscopic, precautions must be taken
921
accordingly. The dish is now heated strongly, but not beyond
the temperature necessary for the complete combustion of the
organic matter. During ignition the sand should be well stirred
at intervals t o allow full exposure to the air. The dish and
contents are then cooled and weighed. The difference in weight
before and after ignition represents the dextrin gum in 50 cc. of
filtrate-i. e., half the weight of sample taken.
This method of evaluating starch products is ingenious.
In application, however, to commercial starch products, this
designation of a “dextrin” is arbitrary and has led a t times
to erroneous conclusions. The purpose of this paper is to
show the true significance of the method.
Theoretical Considerations
Certain present conceptions of the physical and chemical
character of starch and its derivatives (also celluloses, syn-
thetic resins, proteins, etc.) envisage aggregates very much
larger than the individual molecules. Meyer and Mark (6),
of “dextrin” evaluation is arbitrary; the
results within certain limits and with cer-
tain products vary linearly with the con-
centration of barium hydroxide. It has
been used for arbitrary classifications in
tariff matters; and for this purpose it is
suitable, provided its theory is understood
and its limitations are considered. The
method appears to constitute primarily an
index of the relative size of amylaceous
micelles.
Sidgwick (6),Hirst, Plant, and Wilkinson (4), and others (8)
have given expression to this idea of a sheaf or micelle, for which
there is very strong experimental evidence. For cellulose this is
defined as an oriented sheaf of “main-valence chains”, the
chains being comprised of P-glucopyranose units (3). The
picture for starches is similar, but the chains are much
shorter in length and have an alpha linkage (S), and the
starch organization contains noncarbohydrate residues.
Thus we have main or direct-valence chains of length n as-
sociated into micelles of width m through secondary valence
forces. This explanation goes far to explain the complex
behavior of starch and its derivatives. In undegenerated
starches, where the degree of disorganization from hydrolytic
scission or mechanical action is a t a minimum, TZ and m are of
maximum value; that is, the micelles are of maximum size.
922 INDUSTRIAL AND ENGINEERING CHEMISTRY
We may conceive of these as bundles of amylaceous matter
bristling with hydroxyl groups and, when in aqueous dis-
persion associated more or less loosely with mater groups,
probably through hydrogen bonds (6). Under normal meth-
ods of dispersing starch in water, the granular organization
more or less persists; what we actually have, then, in an aque-
ous dispersion are superniicelles. Treatment with hydrolyzing
agents, such as dilute acids, shortens the direct valence chains
90 M canary type of tapioca dextrin-i. e., a light colored, highly
CONVERSION
FIGURE1. TYPICAL
CHART OF A CANARYDEXTRIN
SHOWINGVISCOSITYAND SOLU-
BILITY RELATIONS
(TAPIOCA)
through hydrolytic scission. This in turn weakens the sum
strength of the coordinate associative bonds and permits of a
greater degree of lateral dispersion under the mechanical
stresses of heat and agitation, and the modified starch is said
to be “thin-boiling.” If the initial disorganization process is
carried far enough, the chain shortening and lateral dis-
persion are such as to cause an approach to complete solubility,
since the micelles may have become so small that in their as-
sociation with water they have approached the index of refrac-
tion of water. A recent conception of a clear and stable dextrin
solution is that of single chains associated with water, where
the hydrogens of the original starch hydroxyls have largely
disappeared through dehydration and oxidation (8). Such a
product can donate (through its oxygens) but it cannot
accept. Consequently, associative forces between the chains
are a t a minimum, and the viscosity is also at a minimum.
Owing in part to the peculiarities of granular organization
and to methods of conversion, a starch derivative, even a
highly converted dextrin, is naturally heterogeneous. I n a
starch the granular packages may vary greatly in size, and
there is evidence that the larger granules are less dense and
more readily degenerated than the smaller ones. For this
and other reasons a conversion process always yields more or
less heterogeneity; its degree depends upon various factors,
some of which are beyond control. Thus it is difficult to
define precisely what constitutes starch or dextrin, except
perhaps in those extremes in which color, microscopic char-
acter in aqueous dispersion, viscosity, reaction to aqueous
iodine, and solubility may present an unmistakable classifica-
tion to one familiar with starch derivatives.
Conversion Tests
I n order to shed more light on precisely what the “dextrin
gum” signifies, as determined by the barium hydroxide
process, a series of conversions by roasting was made both on
tapioca flour and on cornstarch. The conversion processes
VOL. 31, NO. 7
were similar to those employed commercially, and in one
instance (Figure 2 ) represent a commercial conversion.
Samples were withdrawn a t recorded times and temperatures
and tested for solubility (and viscosity), alkali lability ( 7 ) ,
and ‘(dextrin” (by the barium hydroxide method). These
values were plotted against time of conversion and recorded
in Figures 2 to 5 .
The relation of “dextrin” values to concentration of
barium hydroxide was also studied (Figure 5).
Figure 1 represents a typical conversion chart of a modified
soluble dextrin of moderately heavy body produced by heat-
ing t o rather high temperatures a tapioca flour treated with
a moderate amount of a suitable acid.
Between the period of 30 t o 60 minutes the solubility
curve (at 25” C.) rises rapidly with a correspondingly rapid
fall in viscosity, as measured by an Ostwald viscometer; for
the next 30 minutes the slopes of both of these curves become
progressively less and approach a constant. The zone
from 30 to 60 minutes for this particular conversion is there-
fore one of rapid disorganization of the original amylose,
in which the chemical and physical properties are changing
from those of a starch to what is accepted by the art as a
dextrin. Specific rotation would also change during this
-
TIM€ M/NUT€S
FIGURE 2 . CANARYTYPEOF CONVERSION,
SHOWING RELATIONS OF “DEXTRIN” ASSAY
TO SOLUBILITY
AND ALKALI LABILITY
(TAPIOCA)
interval; but the rate of change of specific rotation is far
less than for viscosity or solubility, and so this old method of
evaluation is not a t present favored over the far more sensitive
indications of viscosity and/or solubility. The values of
alkali lability (7) and “dextrin” by the barium hydroxide
method are not recorded.
Figure 2 represents the progressive changes in the full
conversion of a 100 per cent soluble canary-type tapioca
dextrin, in which a relatively large amount of acid was used
in the pretreatment and in which the viscosity of the final
product (not recorded) is of a low order. Temperature of
conversion, solubility, “dextrin” (by the barium hydroxide
method), and alkali lability (7) are plotted against time.
For this particular type of high-acid conversion the “dex-
trin” curve is well above the solubility curve in the early
stages of conversion; it has a value of 47.5 per cent “dextrin”
at the low solubility of 16.1 per cent and a conversion time of
30 minutes. The two curves then cross each other, and a t a
time of 45 minutes the “dextrin” content is 87.4 per cent and
the solubility 96.2 per cent. At 60 minutes and more, the
“dextrin” and solubility values coincide a t 100 per cent.
JULY, 1939 INDUSTRIAL AND ENGINEERING CHEMISTRY 923

(7). The values represent milligrams of iodine absorbed per

100 milligrams of sample, after pretreatment with aqueous
alkali. Our researches have proved it to be a sensitive
and suggestive index of the chemical and physical mechanics
of starch degradation. The curve of alkali labile values in
conversions by roasting is invariably of the sine type, the
peak of the curve apparently representing the zone of most
pronounced chain scission, with consequent disorganization.
This is shown by the rate of change of viscosity and solubility.
The relative height of the peak indicates the degree of scission.
This has a significant effect upon the “dextrin” content as
determined by the barium hydroxide method for conversions
of starches of similar botanical origin. The further sig-
nificance of the alkali labile curve is discussed under Figure 3.
Figure 3 A represents a tapioca dextrin conversion of what
may be termed a (‘modified British gum type.” Although
the solubility of the end product is 98.9 per cent, it was pro-
duced with relatively low acid pretreatment, and the final
viscosity is relatively heavy as compared to the end product
of Figure 2. The iodine test shows a red coloration, as in the
case of the end product of Figure 2 . The conversion rate is
slower than in Figure 2.

FIGURE3. MODIFIEDBRITISHGUM TYPECON-

VERSION, SHOWINGRELATIONSOF “DEXTRIN”
ASSAYTO SOLUBILITY
AND ALKALI LABILITY

Except, therefore, at relatively low temperatures and times 50 55 60 65 70

c c. 5 A /on)z
of conversion, the barium hydroxide method of evaluating
“dextrin” is in reasonably close accord with the solubility 5 . RELATION
FIGURE OF “DEXTRIN”
Assay
values for high-acid type of converted starch products. This TO VOLUME
OF BARIUM
HYDROXIDE
conclusion is further supported by numerous tests made from
time to time on products similarly produced both from The relation of the solubility and “dextrin” curve is here
tapioca and corn. very different ; the form of the alkali labile curve supplies the
The curve of alkali labile values is interesting and sig- explanation. Up to a conversion time of 105 minutes the
nificant. It represents a strictly chemical method of in- “dextrin” and solubility curves are roughly parallel; the
terpretation of starch disorganization developed by Taylor former is slightly higher than the latter a t times beyond an
hour. Beyond 105 minutes, however, the solubility curve
begins to rise with characteristic rapidity, crossing the
“dextrin” curve a t 135 minutes. The “dextrin” curve con-
tinues its gradual upward slope and at the end of the con-
version shows a value of only 52.3 per cent.
The height of the alkali labile curve is much less than in
Figure 2 for the high acid conversion, and the generally low
level of the “dextrin” curve signifies a pronounced precipita-
tion of so-called starch by barium hydroxide. Since the
height of the alkali labile curve signifies a degree of disorgani-
zation much less than in Figure 2 , the micellar organization
in Figure 3A is much larger. In accordance with the con-
ception of coordinate valence postulated by Sidgwick and our
postulation of sucrose separations, the heavy barium hy-
droxide molecules may be expected to act as sinkers weighting
down the micelles. The larger the micelles, the more readily
they may be precipitated (in general accordance with Stokes’
law), and this appears to be the case. The larger micellar
I I I I l l I
20 40 60 80 /oo structure (proved by viscosity) in Figure 3A suffers the greater
g. SOfUB/L/TY precipitation of “starch.” The larger micelles are clearly
FIGURE
4. “DEXTRIN”ASSAYSas FUNCTIOXS
OF indicated by the lower order of alkali lability, a fact which
SOLUBILITY
FOR DIFFERENT
TYPESOF CONVERSION may be further confirmed by sensitive viscosity measure-
924 INDUSTRIAL AND ENGINEERING CHEMISTRY
ments. A sample of ground tapioca flour, for which viscosity
tests indicated considerable disorganization, yielded a so-
called dextrin content of 28 per cent by the barium hydroxide
method.
The alkali labile curve affords perhaps the most sensitive
and rational approach to that most intangible of classifica-
tions-a starch and a dextrin. If a curve is plotted of the
alkali labile variations as a function of time during the con-
version of a starch by roasting, then it may be said that the
integration of the rising segment of the curve will represent
the starch zone, and the integration of the descending segment
will represent the dextrin zone. Thus, in Figure 3A samples
taken a t less than about 105 minutes of conversion will be
definitely starchy in character, as defined by color, by iodine
coloration, by hydration characteristics under microscopic
observation, by solubility, and by viscosity. At greater than
105 minutes conversion, these same characteristics will par-
take of a dextrinous nature. Our investigations and those
of Taylor (8) have indicated a low order of association, m,
on the descending side of the alkali labile curve; this fact is
in accord with Taylor’s conception (8) of a true dextrin,
which’appears to be a polymer of a more or less oxidized
glucosan (7). Further investigations as to the physical and
chemical nature of dextrins are in progress.
The previous conversions discussed were of tapioca. Fig-
ure 3B shows a similar analysis of a corn conversion of a type
of dextrin of high commercial importance. The forms of the
curves are, in general, similar to those of 3A except that the
slope of the “dextrin” curve near the end of the conversion is
steeper than for the tapioca of a somewhat similar character.
The end product yields a “dextrin” value of 70.0 per cent and
a solubility of 90.6 per cent as against 52.3 per cent ((dextrin”
and a solubility of 98.9 per cent for the tapioca. Experience
here has shown that for British gum types of dextrin of
similar conversion treatment, the corn gums tend to yield
higher ((dextrin”contents by the barium hydroxide method
than are yielded by tapioca gums. In Figure 3B the alkali
labile curve shows relatively low values throughout. The
apparent anomaly of a relatively low alkali labile curve and a
relatively high “dextrin” curve may be caused by the known
presence in corn of relatively large proportions of fatty acids
which may act to block off in part the coordinate valence
linkages with barium hydroxide and thereby reduce the
weighting-down or precipitating action. The probable
shorter chain lengths of commercial samples of corn would
VOL. 31, NO. 7
also contribute to this effect through reducing the linear
dimension of the micelles. The quantity of acid used in the
pretreatment of the starch of Figure 3B was also less than for
the tapioca of 3A, and the lower alkali labile values of the
former reflect this reduced acid pretreatment.
In Figure 4 the relation between solubility and “dextrin”
by the barium hydroxide method is plotted; the values are
taken from the curves of Figures 2 and 3. The barium
hydroxide method of “dextrin” evaluation tends to give
values which, with respect to solubility, are very different
under different conditions of starch degradation. The curve
of the tapioca canary type of dextrin shows that its “dextrin”
content would assay as high as about 40 per cent a t a solu-
bility of only 10 per cent and show a blue iodine reaction;
at a solubility of about 95 per cent a British gum type of
tapioca dextrin might still assay only 40 per cent “dextrin”,
so much does the character of the conversion treatment
affect the barium hydroxide test.
The effect of increasing the concentration of saturated
barium hydroxide in the solution tested is revealed in Figure
5. Two samples were selected from the conversion of Figure
2-one of 99.5 per cent and the other of 16.1 per cent solu-
bility. The standard amount of saturated barium hydroxide
(50 cc.) was increased to 60 and 70 cc., and determinations
were run. The result within these limits appears to be a
linear function of the concentration of barium hydroxide in
the solution tested, the curves being parallel. The postula-
tion of a weighting-down action is strongly supported, for
the precipitation is proportional to the concentration of
barium hydroxide. Thus it is possible to obtain, within cer-
tain limits and from certain products, a variable “dextrin”
content by arbitrarily varying the quantity of the precipitat-
ing agent.
Literature Cited
(1) Babington, Tingle, and Watson, official method of Analytical
Lab., Dept. of Customs, Ottawa, Canada, July, 1918.
(2) Browne, C. A., Handbook of Sugar Analysis, New York, John
Wiley & Sons, 1922.
(3) Haworth et al., J.Chem. SOC.,1926-37.
(4) Hirst, Plant, and Wilkinson, Ibid., 1932,2375.
( 5 ) Meyer and Mark, “Der Aufbau der hochpolymeren organischen
Naturstoffe”, Leipzig, 1930.
(6) Sidgwick, N . V., “Physical Properties of the Covalent Link in
Chemistry”, Ithaca, Cornel1 University Press, 1933.
(7) Taylor, T. C., J . Am. Chem. SOC.,55, 264 (1933).
(8) Taylor, T. C., private communications.