Adsorption Molecular Weight Fatty Acids by An Activated Charcoal. Use of Chromatography Obtain Adsorption
Adsorption Molecular Weight Fatty Acids by An Activated Charcoal. Use of Chromatography Obtain Adsorption
Adsorption Molecular Weight Fatty Acids by An Activated Charcoal. Use of Chromatography Obtain Adsorption
MAXNESTLERAND HAROLD
G. CASSIDY Vol. 72
[CONTRIBUTION FROM THE DEPARTMENT
OF CHEWSTRY, YALE UNIVERSITY ]
The possible advantages in knowing the ad- and Virtanen,lo Miller, Wohler and Wenzel,12
sorption behavior of each component of a system Landt and Knop,13 Linner and Gortner,14 and
to be studied chromatographically have been Claesson.a The data of all of these workers3-14
pointed out in recent years by a number of in- amply illustrate the generally accepted facts of
v e s t i g a t o r ~ . ~ ,For
~ ~ ~reasons
,~ such as they have the reversibility and rapidity of adsorption, the
given, and also in order to systematically investi- usually consistent behavior of a n homologous
gate the problem of mixed adsorption in greater series as embodied in Traubes rule, the effect of
detail with a view to an analytical separation molecular structure and purity of charcoal on
method, the adsorption isotherms of the three adsorption, and most important, the over-all
saturated, aliphatic fatty acids, acetic, propionic inconsistency found when one attempts to com-
and butyric, were measured over as wide range of pare the results obtained with one charcoal with
concentration as was experimentally possible. those obtained with another. The non-uni-
Two types of isotherms were determined: (1) formity of adsorbents which gives rise to the latter
the static adsorption of each acid alone dissolved problem is of particular importance to chroma-
in water was found in the conventional manner tography, as recently pointed out by Zechmei~ter~
by shaking a quantity of adsorbent and acid and Deitz.16 The use of an adsorbent of more
solution of known concentration together until reproducible properties3>l7has been an important
equilibrium was established, whereupon the con- consideration in the present work.
centration was again determined, the difference C l a e s ~ o n using
,~ the technique and theory of
in concentration being proportional to the rela- frontal analysis to obtain adsorption isotherms for
tive amount of acid adsorbed on the charcoal; single pure solutes, and for the components of
(2) the adsorption of a n acid in the presence of the binary mixtures, has shown that the multiple
one next higher (with respect to number of C- solute Langmuir equationla may be applied
atoms) in the homologous series was measured qualitatively, and to a certain extent quantita-
using the technique and theory of frontal anal- tively, to the chromatographic adsorption analysis
ysis. of mixtures of fatty acids, esters and alcohols.
Comparatively few investigations of adsorption Unfortunately however, the greater proportion
have been made over wide concentration ranges, of data obtained for adsorption from solution can
and of these, the majority are not concerned with be expressed not by the Langmuir equation, but
the lower aliphatic fatty acids adsorbed on char- rather by that of Freundlich. And, there have
coals. Schmidt-Walter7found acetic acid in water been few systematic investigations of pure and
to be preferentially adsorbed over the range 0- mixed adsorption of an homologous series of sub-
100% acid. Though Ostwald and Schulze8 meas- stances, and especially for the case where the
ured the adsorption of acetic acid in water to only data follow only the Freundlich type of isotherm.
about 5.57, of acid, the shape of the curve is such Likewise, there has not been available an analo-
that i t would appear to predict the same results gous exponential equation for mixed adsorption.
as found by Schmidt-Walter. In the work of The results of the present investigation have
Heymann and the isotherm for butyric allowed us to compare the adsorption isotherms
acid in water was irregular and probably based for three members of an homologous series of
on too few points to support with certainty the substances, each separately obeying the Freund-
assumed slight negative adsorption of acid in the (10) G. Wiegner, J. Magasanick and A. J. Virtanen, ibid., 28, 51
region 80-100 mole per cent. (192 1).
(11) E. J. Miller, J. Phys. Chem., 36, 2967 (1932).
The adsorption of the aliphatic fatty acids a t (12) L. Wohler and W. Wenzel, Kolloid Z . , 63, 273 (1930).
low concentrations on carbon adsorbents is repre- (13) E . Landt and W. Knop, Z. p h y s i k . Chem., 8162, 331 (1932).
sented by extensive investigations, among which (14) E , R . Linner and R . A . Gortner, J. Phys. Chem., 39, 35
(1935).
may be mentioned those of Wiegner, Magasanick (15) Conference on Chromatography, N. Y. Academy of Sciences,
(1) Nutrition Foundation Pust Doctoral Research Fellow, 1946- November 29 and 30, 1946. A topic which arose during the dis-
1948. cussion following presentation of Dr. Zechmeisters paper. See
(2) Present address: Naval Research Laboratory, Chemistry H . G . Cassidy, A n n . N. Y . Acad. Sci., 49, 324 (1948).
Division, Washington 20, D. 2. (16) V. R. Deitz, ibid., 49, 315 (1948).
(3) S . Claesson, ArkJvfor K e m i , Mineral. Geol., 23A (11 (1946). (17) The choice of adsorbent charcoal for this investigation is
(4) D . DeVault, THISJOURNAL, 65, 532 (1943). based on information from Dr. W. A. Helbig of the Darco Corpora-
(5) D. Glueckauf, J. Chem. Soc., 1321 (1947). tion (by private communication in 1946), that they could furnish a t
( 6 ) H. G. Cassidy, THISJOURNAL, 62, 3076 (1940). any time reagent grade Darco G-60 of the same quality as that
(7) D . Schmidt-Walter, Kolloid Z . , 14, 242 (1914). sold in 1942.
(8) Wo. Ostwald and H . Schulze, ibid., 36, 289 (1925). (18) S. Brunauer, The Adsorption of Gases and Vapors, Prince-
(9) E. Heymann and E . Boye, ibid., 63, 154 (1933). ton University Press, Princeton, N. J., Chapter XIV, 1943.
Feb., 1950 USE OF CHROMATOGRAPEY
TO OBTAIN ISOTHERMS
ADSORPTION 681
lich equation, with their respective adsorption where @ is the specific adsorption, and c the equilib-
isotherms when mixed together as binary systems. rium concentration (in weight per cent.) of the
And, with the recent publication of Glueckaufs5 filtrate a t the break-through point.
binary Freundlich isotherm equation, the experi- When a solution containing two solutes is
mental data obtained have been used to determine analyzed by this means, two breaks (steps) are
to what extent the isotherm constants for single observed in the frontal analysis diagram, e.g.,
pure solute systems could be utilized to calculate Fig. 10. Again the filtrate obtained is pure
the adsorption isotherms for corresponding binary solvent until the adsorbent is saturated with re-
mixtures. Except for those data of Claesson3 on spect to the first, or less strongly adsorbed com-
the higher saturated fatty acids (beginning with ponent. The solution between the two steps
hexanoic) we know of no similar investigation contains only solute 1, but after saturation of the
dealing with adsorption of mixtures of two so adsorbent by component 2 also, the solution
chemically similar adsorbates as the lower satu- passes through a t the original concentrations of
rated aliphatic fatty acids. both components. The equilibrium concentra-
Adsorption isotherm data are treated by the tration of solute 1 in the first step (c1,1) is not,
empirical equation of Freundlich however, equal to its initial concentration 61.
x / m = qQ = k c l h Rather, cl,l will be greater due to partial displace-
(1) ment of solute 1 from the adsorbent by the more
where x is the weight of acid in grams adsorbed strongly adsorbed solute 2. And, because of the
by m grams of dry adsorbent, qo is the specific competition between the two adsorbates for the
adsorption (grams of acid adsorbed per gram of available active adsorption sites, they mutually
adsorbent)) c is the equilibrium concentration in diminish their adsorption and, in this case, are
weight per cent., and k and 1/12 are empirical adsorbed less than they would be from their
constants. Theoretical interpretations of this respective pure solutions a t corresponding con-
equation as regards the effect of temperature and centrations.
of quantity of adsorbent on the constants k The amounts adsorbed of 1 and 2 may be cal-
and l/n, and limitations to its use, are given by culated from the diagram. Thus
Ostwald and de Izaguirre, l9 Wohler and Wenzel12
and Ostwald.20 q; = w: x c2 (3)
The adsorption. isotherms for each component since the adsorbent is saturated with solute 2
of binary mixtures of the fatty acids studied were after the second step and the solution goes through
obtained using the technique and theory of frontal unchanged. With the concentration expressed
analysis. For a detailed discussion of this, and as weight per cent., q: may be calculated from the
the other methods of chromatographic adsorption expression
analysis, reference is made to Claesson3 and to
Weil-Malherbe.21 In a frontal analysis, as car- q; = w; x c1 - ( w ; - w ; ) x C l , l (4)
ried out during the course of this work,22the solu-
tion to be analyzed was passed through the tube where w:-cl is the weight of solute 1 originally con-
of adsorbent (filter), and the concentration of the tained in wi grams of solution. After passing
solution (filtrate) leaving the bottom of the filter through the filter, w: contains only the amount
measured and plotted against the total weight of solute 1 in the first step, namely, (w: - wy) X
of liquid which had passed through. $1.
For a system containing one solute, pure sol- It is readily seen that, for systems containing a
vent is obtained until the adsorbent is saturated, single solute, the equilibrium concentration, being
whereupon the concentration of the filtrate rises that of the original solution, can be fixed before-
sharply to the equilibrium value, and passes hand. Such is also true for the more strongly
through unchanged thereafter. The weight w of adsorbed component of a binary mixture, but not
liquid passed up to this break point is called the so for the less strongly adsorbed one due to ad-
threshold weight, and originally contained the sorption displacement. However, for a chemi-
amount of solute adsorbed in the filter. If the cally similar series of compounds such as these three
threshold weight is divided by the weight of acids, the ratios of the equilibrium concentra-
adsorbent used, the specific threshold weight, tions were found to be approximately constant
wO,is obtained. Thus per gram of adsorbent (the ratios of the initial concentrations were
po=woxc (2) held constant, but the total acid concentration
varied), so that c1,1 could be predicted roughly
(19) Wo. Ostwald and R . de Izaguirre, Kolloid Z., 30, 279
(1922). from CY after the first few frontal analyses of
(20) Wo. Ostwald, ibid., 43, 268 (1927). mixtures.
(21) H. Weil-Malherbe, J . Chem. Sac., 303 (1943). The determination of adsorption isotherms for
(22) I t is generally more common in chromatographic adsorption the components of mixtures is limited in this
analysis to first wash the column of adsorbent through with pure
solvent before the solution to be analyzed is passed in. For our
method, however, to binary solute systems only,
experiments, it was more convenient t o use the filters initially dry. since for a larger number of components there are
The several advantages are discussed later. more unknowns than equations.
682 F. H. MAXNESTLER G. CASSIDY
AND HAROLD Vol. 72
TABLE I w
d 0.20
VALUESOF THE CONSTANTS
k AND l / n FOR THE ADSORP- .*0
Y
TION EQUATION
q' =kc"" 2
Parentheses around an acid indicate that the data were w
obtained by frontal analysis of the mixture shown. Sym- 4
bols are those used to represent the curves of Figs. 3 and 4.
3 0.10
Acid Sym-
bo1 k l/n Mixture au 8
MeCOOH A 0.136 0.494 . . . . . . . . . . . . . . . . . . . 0.0178 3
(MeCOOH) A' ,041 ,350 MeCOOH +
EtCOOH ., ..
(EtCOOH) P' .190 .435 MeCOOH +
EtCOOH ., ..
EtCOOH P .232 .419 ................... 0.0309
(EtCOOH) P" .060 ,646 EtCOOH +
PrCOOH ,. ,.
(PrCOOH) B" ,293 ,228 EtCOOH +
PrCOOH .... 0.5 1.0 1.5
PrCOOH B .378 ,339 ................... 0.0667 Concentration, c, weight per cent.
k recalculated according to equation 6. Fig. 3.-Adsorption of fatty acids (curves A, P and B ,
(26) In their investigations of the adsorption of benzene on carbon resp.) at low concentrations in water: 0, static adsorption
from ethyl alcohol as solvent. Bartell and Sloan looked upon the ad-
sorption at high concentration of benzene as being equivalent
of pure acids at 25'; 0,frontal analysis of acetic +
to the adsorption of ethyl alcohol at very low concentrations in the propionic mixtures (A' and P', resp.); a, frontal analysis
other component as the solvent. They used two Freundlich equa- of propionic f butyric mixtures (P" and BY,resp.).
tions t o express the data at low concentrations for each adsorbate. Solid lines computed from equation 1.
and then b y successive approximations, obtained one composite
equation, capable of expressing the adsorption of one component over (27) V. Cheldelin and R. J. Williams, THIS JOURNAL, 64, 1513
the full concentration range in the other as the solvent. (1942).
684 F. H. MAXNESTLER G. CASSIDY
AND HAROLD Vol. 72
the purposes of this investigation, the best aver- curves of Figs. 10 and ll), WI, c ; , ~ and wz were
age straight lines were drawn. obtained by extrapolating as follows. The rate
- at which the effective concentration c;,~ changes
B- in the diffuse front portion of the first zone was
almost linear in nearly every instance; conse-
P quently it was possible to obtain w1 by extrapolat-
ing this portion of the curve to the abscissa. A
A flat, straight line was drawn for the effective
equilibrium concentration c;,~ and represented
-- O1..0I the average value for a number of points. It was
only necessary then in order to convert from
effective Concentration to weight per cent. cl,l
to multiply by the correct acid titer of the base.
The threshold weight for the second component
was obtained by drawing a straight line through
the steep portion of the second step; the value
-2.0 I/
.U
-2.0
Log c.
-1.0
1 1 1 1 1 1
0
1 1
(as read from the abscissa) of the point of inter-
section of this line with that of & extended was
w2. The specific values were computed in all
cases from the actual weight of dry charcoal, and
not from total weight of adsorbent packing.
Adsorption isotherms obtained by frontal
analysis of binary acid mixtures are given in Fig.
Fig. 4.-Static adsorption data plotted in the linear form
3, where the respective isotherms for each of the
of equation 1; symbols as in Fig. 3.
acids in the mixture acetic and propionic are repre-
sented by points along the curves A and P,
The extent to which the high concentration and for the mixture propionic and butyric by
portions of the adsorption isotherms are affected points along P and B. As previously for static
by the use of charcoal which has not been pre- adsorption data, the solid lines were calculated
viously dried is shown in Fig. 7 by data for acetic using equation 1 and constants (Table I) ob-
acid. For comparison the similar portion of the tained by plotting the experimental data in the
isotherm for acetic acid (Fig. 2) adsorbed on linear form of the Freundlich equation (as drawn
charcoal initially dry is drawn (solid line) with the in Fig. 5). The straight lines for curves P
experimental data for adsorption on wet and B are open to some question, drawn as
charcoal superimposed. It is seen that the effect shown. However, the choice of points to which
is chiefly one of dilution of the original acid solu-
tion, with, however, possibly some slight dis-
placement of the adsorption curve due pre-
sumably to a slight alteration in the surface
characteristics of the adsorbent as a result of the
adsorbed water layer. The error introduced is
particularly significant a t very low adsorptions.
That one should dehydrate the adsorbent before
use is obvious, or a t least correct the final results
for the moisture introduced by the charcoal.
DobineZ8 concluded that water so introduced
resulted in a dilution effect only. Numerous
qualitative statements may be found in the litera-
ture regarding the undesirability of having mois-
ture in the adsorbent, though quantitative data - 1.0 0
supporting the opinions are generally lacking. Log c.
Since in a frontal analysis there is no complete Fig. 5.-Adsorption data from frontal analysis plotted
separation of zones as may be possible in an in the linear form of equation 1; symbols as in Fig. 3.
elution analysis, i t was more convenient to plot
the concentration of each fraction of liquid emerg- to give greater weight was based on an evaluation
ing from the bottom of the column as a n effec- of the individual experiments. The scatter is due
tive concentration, c, defined as cc. of standard to the decreased precision with which the threshold
base/g. of liquid. The abscissa is the total weight weights could be obtained for the system pro-
of liquid collected. From the resultant frontal pionic-butyric acids as a result of a number of
analysis diagrams (examples of which are the unexplainable inconsistencies in the frontal anal-
(28) M. Dobine, Compt. vend., 212, 155 (1941). ysis diagrams.
Feb., 1950 USE OF CHROMATOGRAPHY
TO OBTAIN ISOTHERMS
ADSORPTION 685
TABLEI11
- 0.5
FRONTAL
ANALYSISDATAFOR THE MIXTUREPROPIONIC
ACID BUTYRIC ACID +
CI (C1 + cz) c1.1 4 (4- 4)
0.952 2,044 1.640 14.9 12.4
0 .
.763 1.522 1.214 17.6 14.8
-1.0
M .626 1.250 1.015 20.6 17.2
3 .483 1.003 0.788 25.4 23.2
.502 1.000 .764 23.9 18.3
.377 0.755 ,588 30.1 33.2
.296 ,598 .465 34.6 29.7
-1.5
.297 ,598 ,462 35.5 30.3
,249 ,497 ,386 37.3 48.7
-1.0 0
(log wy us. log c ~ , the
~ , equilibrium concentration,
Log c.
is included for comparison). It was interesting
Fig. 6.-comparison of experimental frontal analysis iso-
therm data (solid lines) with points calculated using equa-
tions 5a,b; symbols as in Fig. 3.
g o I I
_1
1 I I
-0.5
Log concentration.
0 4-0.5
+
PROPIONICACID 5
C1 (GI c2)+ C1,l 4 (78; - 7 4 5 M
s
0.982 1.973 1.521 9.47 8.88
.759 1.502 1.145 11.05 11.3 1.1
.752 1.500 1.153 11.79 10.96
,503 0.995 0.751 14.24 13.46
.379 ,761 ,566 16.83 16.13 -0.5 0 +0.5
.251 ,503 .367 20.75 20.6 Log concentration.
,152 ,298 .210 27.0 28.6 Fig. 9.-Frontal analysis data for mixtures of propionic
Concentration and specific threshold weights are ex- f butyric acid plotted as in Fig. 8; C,/CI approximately
pressed in weight per cent. and grams, respectively. 1.0.
686 F. H. MAXNESTLER G. CASSIDY
AND HAROLD Vol. 72
to find that the data fell on fairly good straight Le., the concentration difference, decreases with a
lines when SO plotted, and could be represented simultaneous decrease in the slope of that portion
by equations of the form y = a +
bx, where y of the curve representing the front of the second
represented the log of the specific threshold values zone. Consequently, the break in the curve where
and x the concentrations. wo is thus seen to be the more strongly adsorbed component emerges
exponentially dependent upon concentration, from the column is much less sharply defined a t
that is, for example, zu; = where log A = a. low concentrations. This loss in definition was
The constants a and b were determined from the greatest for mixtures of propionic and butyric
respective straight lines of Figs. 8 and 9, and are acids, and accounts most for the scatter of the
compiled in Table IV. The fact should be noted experimental points along the curves P" and
B" in Figs. 3 and 5 .
TABLE IV The degree to which the frontal analysis dia-
CONSTANTS a AND b FOR AN EQUATION OF THE FORM y = grams were affected by a change in the ratio of
( I + bx RELATING THE SPECIFIC THRESHOLD VALUESAND acid concentrations is shown by the three curves
EQUILIBRIUM CONCENTRATION DATAOBTAINED BY FRON- in Fig. 10 for the system acetic acid-propionic
TAL ANALYSIS acid.
U
Fig. 7
b a
Fig. 8
b
Practical limitations to the problem of a separa-
tion of fatty acids by means of frontal analysis
Log w ; us. log c1 0.985 -0.520 1.161 -0.737 are implied. From the alteration in the curves
Log w: us. log c1,1 1.082 - .549 1.320 - .674 resulting when the acid ratio, changes, it will be
Log (70," - w;, us. log
+
(c1 c2) 1.129 - .e22 1.366 -1.244
increasingly difficult to detect the presence of the
more strongly adsorbed solute as it becomes a
that, in carrying out all of the frontal analyses relatively smaller fraction of the mixture, and
represented by the above data, the ratio of the ini- similarly as the total acid concentration decreases.
tial concentrations of each acid was on the average Obviously increase of the amount of adsorbent in a
1.010 * 0.012 for the acetic-propionic system, column, in the hope of increasing separation, will
and 0.987 A 0.018 for the propionic-butyric. not be the practical solution since both threshold
The three curves shown in Fig. 10 for the system weights will be proportionately increased. At
acetic acid-propionic acid are representative of the other end of the scale, smaller amounts of the
those from which previous frontal analysis adsorp- less strongly adsorbed solute could probably be
tion data were obtained. Variations in shape detected in the presence of the second acid because
were chiefly a function of the total acid concen- of the sharpening effect on the fronts resulting
tration. Little improvement in definition is both from adsorption displacement and from the
found in the curves a t higher concentrations; higher relative concentration of the latter com-
and, a t lower values, irregularities become more ponent.
numerous. For example, the horizontal separa- Chromatograms given in Fig. 11 for the four
tion between zones increases, but the vertical, possible mixtures of the three acids show the dis-
placements which occur. It is quite evident
that such curves alone afford no means for quanti-
25 50 75 100
25 50 75 100 Total weight of filtrate, g.
Total weight of filtrate, g. Fig. 11,-Frontal analysis diagrams for mixtures of
Fig. 10.-Chromatograms showing the alteration for fatty acids. c8:c~:cl approximately 1.0, and total c ap-
acetic+ propionic acid when the ratio of initial acid con- proximately 2.07'0: 0, acetic + propionic acids; 0 ,
centrations is changed. CI/CS is: 0, 1:3; 0 , 1:l; 3, 3 : l ; propionic+ butyric acids; 0, acetic + butyric acids.
c,+ c2 = 1.0%. 6 ,acetic+ propionic + butyric acids.
Feb., 1950 TO OBTAINADSORPTION
USE OF CHROMATOGRAPHY ISOTHERMS 687
when equations 5a, b were derived. It is to be In Fig. 6 for the binary acid system acetic-
noted here that, as used by Glueckauf in his propionic, the solid lines A' and P' are those of
paper, @' = @z/@I = 1/@", and p1and @Z referred Fig. 5 and the points were calculated by means
to the surface area requirements of each of the of equations 5a,b using the average @-values,
adsorbates in the binary mixture (that is, ,B /I' = 2.24 and @" = 1.64. Comparison of the
represented the amount of adsorbent occupied two figures shows some slight redistribution of the
at saturation by 1mole of the adsorbate). points along PI and a greater change for those
Glueckauf 's binary isotherm was tested using along A' in the direction of increasing slope, which
the data of this investigation. As is seen, the is shown by the dashed line redrawn through the
form of the equation is such that i t is easier to calculated points.
compute values of the equilibrium concentration A similar treatment was given the data for the
c, using the corresponding observed values of system propionic-butyric acids, with results also
q! and qi, and the constants l / n and k (recalculated shown in Fig. 6. The solid lines P" and B"
by means of equation 6) of the single pure solute again represent the experimental isotherms. The
isotherms. When this was done, very poor redistribution of points as a result of the calcula-
agreement was obtained regardless of the value tions is much more apparent. For butyric acid
of &'@I chosen. I n particular i t was found that (B") the dashed line has an increased slope, while
when a so-called best value of @z/& was used in for propionic (P") the slope has been decreased
computing values for cl, the concentrations then slightly, though the justification for redrawing
calculated for c2 (using the inverse value of p2/P1) the latter a t all is open to some question, con-
differed very considerably from the experimental, ddering the precision of the observed data.
the deviations generally being either all positive The redrawn isotherm serves the purposes, how-
or all negative. ever, of the discussion which follows.
In order t o obtain any agreement, @ in the The question early arose in this work as to what
present discussion is interpreted only as an relationship the Freundlich isotherms for each
empirical constant calculated from the observed component of a mixture would show with respect
values of the equilibrium concentration and to the isotherms for the single pure solutes. We
specific adsorption. For each component of the have seen that the behavior predicted theoretically
two binary mixtures, values of the @-ratio (ie., by the multicomponent Langmuir equation,
P' and @") were computed from each of the namely, that under certain conditions the linear
equilibrium concentrations by substituting the form of the isotherms for the components of a
experimental values of c1 and c2, and the corre- mixture should have the same slopes as the corre-
sponding q; and qi in equations 5a,b; these are sponding isotherms for the pure solutes, has been
listed in Table V for coexisting concentrations, verified experimentally by C l a e ~ s o n . ~It ap-
with average values for @' and @" a t the foot of peared therefore to be not unreasonable to deter-
each column. mine whether the linear form of the Freundlich
TABLEV
equation could be fitted to the data in a similar
fashion, even though there was no theoretical
VALUESOF THE &RATIO(EQUATIONS 5a,b) CALCULATEDjustification for such.
FROM EXPERIMENTAL VALUESOF p0 AND c OBTAINED BY
Comparison of the values of the slopes (l/n
FRONTALANALYSISOF BINARY MIXTURESOF ACIDS in Table I) of the curves in Figs. 4 and 5 shows the
Acetic + propionic Propionic + butyric
MeCYOH EtCOOH EtCOOH PrCOOH extent to which the above reasoning applied to
8 8" B' 8" the experimental data. In only one instance
1.65 1.66 2.53 1.35 were the lines approximately parallel (the slope
1.97 1.66 1.67 1.44 of P' being about 4% greater than that of P in
2.20 1.62 1.64 1.66 the static adsorption measurements). On the
2.33 1.74 2.00 0.95" other hand, however, i t is interesting to note that
2.82 1.73 1.77 2.04 when the experimental data were calculated
2.47 1.44 2.09 1.53 according to equations 5a,b and replotted in the
3.10' 2.02" 2.31 2.30 linear form, as was shown in Fig. 6, the straight
2.38 2.43 lines (dashed) which could be redrawn through
2.87" 2.60 these points tended to be more nearly parallel to
2.24 1.64 2.05 1.92 those of the pure acids than were the lines (solid)
" Omitted in the computation of the average p-value at for the observed data. Thus, l / n for A' is 0.447, for
the bottom of each column. PI i t equals 0.587, and for B'', l / n is 0.411. These
It is seen that there is not the relationship values should be compared with those in Table I
between 0' and @" suggested by Glueckauf. for A, P and B. Considerably more data are
The above mentioned average values of @' needed for mixed adsorption in order to verify any
and p" were then used to recalculate equilibrium postulations of parallelism in Freundlich adsorp-
concentrations by means of equations 5a,b, and tion analogous to that found where the Langmuir
to compare these graphically in Fig. 6 with the equations apply.
experimental results, We conclude from the various calculations
Feb., 1950 POLYAMIDES
MULTILINKED 689
made that, though the pure acids separately follow 4. The adsorption data for each acid obtained
Freundlich isotherms, they do not, as binary by static adsorption and from frontal analysis of
mixtures, obey the binary Freundlich isotherm mixtures, when plotted according to the linear
equation proposed by Glueckauf. If, however, i t form of the Freundlich equation, gave straight
is modified as previously described (that is, use lines, corresponding ones of which appeared to be
separate @-ratio values determined from the approximately parallel to each other.
observed results, for each component of a mix- 5. Using the binary isotherm equation pro-
ture) i t gives reasonable agreement with the posed by Glueckauf t o express the adsorption
experimental data and appears t o smooth out data of mixtures, where the individual components
inconsistencies in the experimental points. But follow the Freundlich equation, .it was possible
the agreement results only by virtue of the in- to obtain fair agreement between the observed
troduction of an empirical factor for one which data, and that calculated from his equation,
previously bore a physical interpretation. but only when a modification was made in the
The authors wish to express appreciation to the interpretation attached to one of the constants in
Nutrition Foundation, Inc., for their generous the equation.
support of this work. 6. The effect, upon the static adsorption iso-
therms, of moisture adsorbed in the charcoal,
Summary appeared to be one of dilution only.
1. Measurement of the adsorption isotherms 7. Changes in adsorption, as evidenced by
of the aliphatic fatty acids acetic, propionic and alterations in frontal analysis diagrams, were not
butyric from aqueous solution and on a com- observed when various experimental factors were
mercial activated charcoal, showed each acid to varied. Thus, columns were packed differently,
be preferentially adsorbed throughout the con- flow rate was increased by doubling the external
centration range of 0-100% of acid by weight. pressure, a solid diluent was added to the charcoal
2. The experimental data could be fitted by in order to increase the rate of liquid flow and
the Freundlich isotherm equation over only a several columns were operated either initially
narrow concentration range of approximately dry, or prewetted with solvent before addition of
0.005 to 0.5%. the acid solutions.
3. The technique of frontal analysis has been 8. The threshold values for each of the com-
utilized in order to obtain data relating to the ponents of a binary mixture analyzed by frontal
adsorption of binary mixtures of low molecular analysis are related exponentially to the re-
weight aliphatic fatty acids, with a view to an spective initial or equilibrium concentrations.
analytical method. By this means, the adsorp- 9. The specific adsorption values in the static
tion isotherms for each component of the binary adsorption experiments were found to be de-
acid mixtures, acetic plus propionic, and propionic pendent upon the quantity of adsorbent used,
plus butyric, have been determine: for the equi- contrary to the findings of Burgersz9
librium concentration range of approximately
0.3 to 1.5y0. These data could be fitted also 20, D. C.
WASHINGTON RECEIVED31 JULY 11, 1949
by the Freundlich equation. (31) Original manuscript received August 2, 1948
Multilinked Polyamides
BY JOHN R. SCHAEFGEN J. FLORY~
AND PAUL
The synthesis of multichain polyamides having by treating them with a reactive diamine under
the general structure proper conditions. Using a quantity of the di-
R(-CO[-NH(CH~)sCO-]~ OH]b amine which is stoichiometrically equivalent to the
free carboxyl end-group content of the polymer,
by reaction of e-caprolactam with cyclohexanone- the interlinking of the carboxyl groups can be
tetrapropionic acid ( b = 4) or with dicyclohexa- made nearly quantitative. Polymeric networks of
noneoctapropionic acid ( b = 8) has been described definite structure may be prepared in this manner.
in a previous paperq2 In the present investigation These interlinked multichain polyamides, which
a technique has been developed for interlinking will be called multilinked polyamides, exhibit
carboxyl end-groups of these multichain polymers an interesting combination of properties. At
( l a ) Presented before the High Polymer Forum at the Atlantic ordinary temperatures they behave like linear
City Meeting of the American Chemical Society, September, 1949. polyamides. Thus they are highly crystalline
( l b ) Present address: Department of Chemistry, Cornell Univer- solids which, in the form of fibers, may be oriented
sity, Ithaca, New York.
(2) J. R . Schaefgen and P. J. Flory, THISJOURNAL, 70, 2709 by drafting to several times their initial lengths,
(1948). resulting in an increase in strength and stiffness.