V O L U M E 26, NO. 9, S E P T E M B E R 1 9 5 4
1451
( 2 ) Cady, W. E., and Seelig, H. S., Ind. Eng. Chem., 44, 2636 (1952).
(3) Furby, N.W., gaper presented before the Division of Petroleum .~~ Chemistry at the 116th Meeting of the A Y E R I CCHEMICAL SOCIETY, Atlantic City, K.J., September, 1949. (4) Hibbard, R. R., Ind. Eng. Chem., 41, 197 (1499). ( 5 ) Kuhn, R., and Winterstein, A , , Helr. Chim.Bcta.. 11, 87, 116, 123, 144 (1928). (6) Lillard, J. G., Jones, W.C., Jr., and .4nderson, .J. .I.,Jr., 2nd. Eng. Chem., 44, 2623 (1952). (7) Lipkin, hl. R., Hoffecker. W. A., Martin, C. C., and Ledley, R. E., A s . 4 ~CHEM., . 20, 130 (1948). (8) Lipkin, 11. R., Martin, C. C., and Hoffecker, W..I.,gaper presented before the Division of Petroleum Chemistry at the CHEMICAL SOCIETY, Chicago, 113th hleeting of the AMERICAX Ill., April 1948. (9) 1Iair, B. J., Ind. Eng. Chem.. 42, 1355 (1950).
(10) Jlair, B. J., Gaboriault. .4.L., and Rossini, F. D., Ibid., 39, 1072
(1947). (11) Mair, B. J., Sweetman, .4. J., and Rossini, F. D., Ibid., 41, 2224 (1949). (12) Mair, B. J., Willingham, C. B., and Streiff, A. J., J . Rrscurch .Vatl. Bur. Standards, 21, 581 (1938). (13) Xes, K. van, and Westen, H. A. r a n , “Aspects of the C‘onstitution of Mineral Oils,” iYew York, Elsevier, 1951. (14) Smit, W. AI., Anal. Chim. Acta., 2, 671 (1948). (15) Willingham, C. B., J . Research 9 a t l . Bur. Standard.?, 22, 321 (1939). (16) Winterstein, A . , and Schon. K., 2.physiol. Chem., 230, 146 (1934). (17) Winterstein, -4.. Schon, K., and Velter, H., I b i d . , p. 158. RECEIVED f o r review August 17, 1953. Accepted June 14, 1954. Presentrd before the Division of Petroleum Chemistry at the 124th Meeting of the AMERICAN CHEMICAL SOCIETY, Chicago, Ill.. September 1953.
Adsorption Chromatography and liquid Partition of High Polymers Silicones D. W. BANNISTER, C. S. G. PHILLIPS,
and
R. J. P. WILLIAMS
lnorganic Chemistry Laboratory, O x f o r d University, Oxford, England
The molecular w-eightdistribution in high polymers has never been satisfactorily investigated. Two new countercurrent methods were therefore examined as fractionating techniques. The initial experiments have been carried out on silicone polymers which are known to be linear. Experiments using gradient elution, in which carbon was the adsorbent and the eluting solvent was ether in methanol, gave a series of fractions of the silicones, the molecular weight of the fraction increasing with effluent volume. A second method, liquidliquid partition in a vortex column, w-as successfully applied to fractionation of low molecular weight material. The methods allow rapid determination of the molecular weight distribution in silicones. The fractions obtained are sharp and will permit a fresh examination of much of the theory of high polymers.
T
HE problem of the fractionation of high polymers obviously
presents greater difficulties than those usually encountered in the separation of other substances because of the very small differences bet\vern the properties of the components. Present methods available for the fractionation of polymers are almost invariably batch methods, and although some attempts have been made t o utilize continuous multistage processes, none of them have been developed into routine techniques. The only detailed work on the chromatography of high polymers (3) used the frontal analysis method and involved an intricate and uncertain mathematical examinat,ion of the polymer front irhich emerged from the column. This method does not, frnctionate the polymrr. T h e fractionation procedure which is described here can he used either as an analytical method or preparatively. It is a n application of gradient elution ( 1 , 6). I n this technique the zones of the substances, in this case t,he different molecular weight fractions, are adsorbed onto the column material from n poor solvent (or if the substances are liquids they can be put onto the column directly). T h e development of the zones commences with the poor solvent but the elution is carried on Ivith increasingly more powerful eluting solutions by introducing continuously a good solvent for the substances. The good solvent must be so chosen that i t will elute the substances under examination from the column quantitatively and it-ith very small reten-
tion. The apparatus used to obtain the solvent gradient is illustrated in Figure 1 ( I , 6). APPLICATION TO THE SILICONE POLYMERS
Several adsorbent-solvent combinations were tried before a suitable system was found for these polymers. Alumina and silica gel were found to be unsuitable adsorbents as none of the solvents which were examined would elute the higher molecular weight polymer*. However, if the silicones were adsorbed onto charcoal (animal charcoal, Harrington Bros., Ltd., London. England), they could be quantitatively eluted from this adsorbent by diethyl ether. The recovered material was examined by the determinaVESSEL tion oi“ its viscosity. In this way it was proved that none of the material, even of the highest molecular weight,, had been irreversibly adsorbed, as the viscosities obtained in these tests viere always identical x i t h those of the starting materials. F r a c t i o n a t i o n could now be attempted. METHOD O F FRACTIOX.ATIOS
F$gii ducing
zp{iz:
Gradient
T h e carbon column, 35 cm. in length and 1.5 cm. in diameter, \vas prepared in methanol. To the top of the column \vas added 3.5 grams of a silicone fluid-e.g., a Dow Corning 200 (straight-chain polymethylsiloxanes) of bulk viscosity 1000 centistokes of formula:
I
I n
CH,
where n, t h e number of repeating units, is 140 on average. The polymer wa8 then washed onto the column with methanol. The E t o a 200-mI. mixing vessel which was column W ~ connected
ANALYTICAL CHEMISTRY
1452 filled with methanol, the poor solvent. The second solvent container from which good solvent can be introduced into the methanol, Figure 1, was first filled with ethanol, a better solvent than methanol for silicones, and the elution of polymer commenced. The effluent was collected on automatic fraction collector, working on a time basis, and the flow rate was maintained a t from 5 t o 10 ml. per hour. The solvent in the top container was changed after 230 ml. had flowed from the column, to a 50-50 mixture of diethyl ether and ethanol. It was changed again after a further 60 ml. of effluent had flowed from the column to a 25 to 75 ethanol-diethyl ether mixture. It was finally changed to pure ether after a further 60-ml. effluent had been collected.
I
x >’
50
900
1
400 EFFLUENT VOLUME, MI.
500
300
Figure 4. Analysis of a Mixture of Dow Corning 200 Fluids (50, 200, a n d 1000 Cs.) Together w i t h Dow Corning 702 and 703
i:
Viscosity, flow t i m e (in seconds) through a n Ostwald viscometer
300 400 500 EFFLUENT VOLUME, MI.
600
Figure 2. Analysis of Dow Corning 200 1000 Cs. Fluid Viscosity, flow t i m e (in seconds) through an Ostwald viscometer
As the fractions of the polymer are eluted in different solvents the viscosity of the fractions could not be determined directly. Instead, the solvent was distilled off, the last trace of the solvent being removed by placing the polymers on watch glasses and blowing a current of hot air over them.
1000 cs. fluid. The figure is a double plot of the concentration of the polymer and the viscosity of that polymer which leaves the column a t a given position in the gradient. This position was found to be reproducible within 20 ml., providing that the dimensions of the column, the nature of the solvent gradient, and the load on the column ryere not changed (Figure 7). The viscosity curve of Figure 2 shows that in the analysis of this polymer the fractions were eluted in order of increasing molecular weight. I n all the experiments with the straight-chain polymethylsiloxanes, the viscosity of the silicone increased steadily with improving solvent. The concentration curve in Figure 2 shows a single maximum but the analysis of certain other polymers, such as that of the 50 cs. Dow Corning 200 fluid, show two maxima, Figure 3.
DISCUS SIOS
Although the silicones are fluids, their volatility is very low and i t has been demonstrated that no loss of material occurs through evaporation, except in the case of the very lowest molecular weight members. Each fraction (0.1 gram) of the polymer was then dissolved in 1.0 ml. of toluene and the viscosity of the 10% solution measured in an Ostwald viscometer a t 15’ C. The weight of polymer in each fraction was determined by direct weighing.
E8
60
t
t
n
105 lP5
n 100 300 EFFLUENT VOLUME, MI.
Figure 3.
100 PO0 300 EFFLUENT VOLUME, MI.
400
Figure 3. Analysis of Dow Corning 200 50-Cs. Fluid Viscosity, flow t i m e (in seconds) through a n Ostwald viscometer
The total weight of polymer recovered frequently exceeded 95% of the original material despite the detailed handling of the fractions involving many transfer operations. The deviation from 100% recovery was shown to be insignificant by experiments in which it had been proved that the fluids could be completely recovered. Figure 2 shows an analysis of the fractions obtained from the
500
Analysis of Figure 4 Mixture with S o m e Dow Corning 550 Fluid Added
Viscosity, flow t i m e (in seconds) through an Ostwald viscometer
It must not be assumed that the double peak necessarily represents two molecular weight distribution patterns for it can also arise from a simple distribution of molecular weights in the polymer. A concentration peak early in the elution diagram often implies that the low molecular weight material has been but slightly adsorbed under the chromatogram conditions. The later peak, which is found between 300 to 400 ml. of effluent in the experiments illustrated in this paper (Figures, 2, 3, 4, and 5 ) ,indicates that a t higher molecular weights there is relatively little difference in property between different members of the polymer series, so that there is a crowding together of the material. This
V O L U M E 2 6 , NO. 9, S E P T E M B E R 1 9 5 4 is not a reflection on the fractionating efficiency of the method but only upon the choice of solvent gradient. The gradient elution method described here is best suited for the fractionation of silicone polymers of between 100 and 1000 CP. bulk viscosity. Gradients of many other solvent pairs (Figure 7 ) can be selected and adjusted (by the use of a miying vessel of the required size) so as to extend the range of the separation to lower and higher polymers and to increase the fractionating efficiency. Figure 4 represents the analysis of a comples mixture of silih cones made u p from all the polymers listed in Table I, ~ i t the exception of the Dow Corning 550 fluid. T h e gradient in this experiment was commenced a t a mixture of the mentioned solvents equivalent to 200 ml. of effluent in Figures 2 and 3.
Table I.
Siloxanes ilnalyzed by Chromatography on Charcoal
Silicone FluidD
Bulk Viscosity,
Spec. Yiscosityb (10% Soln in Toluene) 200 50 0 . 4 9 0 (39.6) 200 200 1.090 ( 5 5 . 8 ) 200 1000 2.260 (86.8) 550 100 0 . 3 3 8 (35.4) 20 702 0.163 (31.0) 25 0.177 (31.4) 703 a T h e molecular weights of these materials are available in reference (e) b T h e values in parentheses give t h e times in seconds for t h e flow of t h e polymer solution in a n Ostwald viscometer. These are t h e values used in plotting the figures
(D.C.)
cs.
1453 fluid, which has a bulk viscosity of 100 to 150 cs., is not eluted until 400 ml. of effluent has been collected from the column. The concentration plot against effluent volume is unlike that found for any other polymer and although the viscosity plot rises continuously it does so very slowly. The Doiv Corning 550 fluid includes material containing phenyl side groups instead of the methyl groups of the Dow Corning 200 fluids. Phenyl groupings are invariably much more strongly adsorbed onto carbon than methyl groups and have undoubtedly enhanced the adsorption of the Dow Corning 550 fluid. This polymer appears to have a different molecular M eight distribution pattern from the Dow Corning 200 fluids also. These and other points relating to the properties of the fractionated polymers will be described and discussed in a further paper. The separation process discussed here might not be thought as one of selective adsorption but of simple extrusion. T h e behavior of the Dow Corning 550 fluid argues against this suggestion. However, the point can be demonstrated more conclusively. Figure 7 shows tvio plots of polymer viscosity times against fraction number using two different columns, the righthand curve referring to the larger column. Polymer of a given viscosity is eluted considerably earlier in the smaller column although the dead volumes of the columns differed by less than two fractions. This result is to be expected if adsorption plavs a part in the fractionation. The figure illustrates a second point. The analyses of five different polymer mixtures on the same column, the larger, are represented by the five sets of points. This method can be used for a rough determination of the molecular weight distribution of the polymer without determining the viscosities. The factors which limit the accuracy of this determination are being studied. 150
PO0
400
t
I
600
EFFLUENT VOLUME, MI.
Figure 6. Analysis of Dow Corning 500 Fluid Viscosity, flow time (in seconds) through an Ostwald viscometer
T h e resulting analysis showed a very high concentration of
low molecular weight polymer in the first few fractions. This was expected as the experiment was commenced with a relatively good solvent. This peak includes most of the low molecular weight material, for example, much of the Doiv Corning 200 50 cs. fluid (Figure 3)) mhile the later peak includes the higher molecular weight fractions-for example, much of the Dow Corning 200 1000 cs. fluid (Figure 2). I n the experiments so far described the concentration falls very rapidly a t the end of the run and the viscosity rises steadily t o a maximum value a t the last polymer fraction. Marked contrast with this behavior is shown in Figure 5 , which gives the analysis of a mixture of all the polymers listed in Table I. The gradient in this experiment was the same as that used in the experiments described b y Figures 2 and 3 so that the low molecular weight material is better fractionated than that in Figure 4 and the early concentration peak is not so marked. The inclusion of the Dow Corning 550 fluid has introduced an anomalous tail in the concentration and viscosity plots, contrast between Figures Iand 5 , T h e analysis of the DONCorning 550 fluid alone reveals the cause (Figure 6). It shows t h a t material of low viscosity from this
8
13
18
23
28
FRACTION NUMBER
Figure 7.
Viscosity t's. Effluent Volume, Given as Fraction Number
Each fraction contained 11.0 ml. Gradient used in the series of experiments was: good solvent, 15%methanol-8570diethyl ether; poor solvent, 2570 methanol-75%ethanol Mixing vessel, 150-ml. volume
The method can also be used for the preparation of considerable amounts of the fractions of the polymer. I n single experiments with the silicone fluids, the column has been loaded with 10 grams of material. However, the very lowest members of the polymer series are but very slightly adsorbed on the charcoal used. Such polymers can be fractionated by a technique to be described. COUNTERCURRENT LIQUID PARTITION EXPERIhRIENTS
An attempt has also been made to fractionate the low molecular weight silicones (which pass straight through the gradient elution columns) by means of a vortex countercurrent extractor. Similar extractors have been described (4, 5 ) . The extractor consists of a rod which rotates inside a close-fitting cylinder. Two immiscible liquids are made to pass in opposite directions
ANALYTICAL CHEMISTRY
1454 through the annular space, the heavier down and the lighter u p the extractor. At suitably chosen speeds of rotation of the rod, the two flowing phases are broken up into vortices. Any material in either of the two phases is now efficiently partitioned in the extractor. The flow of both solvents can be made continuous by leading one solvent, the heavier, in a t the top of the column and out a t the bottom, while the lighter phase flows in a reverse manner. Such extractors have been used mostly for stripping one component out of a large volume of solvent containing a mixture ( 5 ) . Fractionation of a material in this type of extractor can be brought about by cycling one of the phases-e.g., the heavier allowing it t,o pass down t'he extractor under gravity and then raising it by an external air-lift so that it is returned to the top of the extractor and recycled. The lighter phase can be passed up and through the extractor in the normal manner. After extraction has proceeded for some time the light phase can be modified so that new extracting conditions obtain, and another fraction of the mixture is separated. Such an apparatus has been used in this lahoratory in the fractionation of silicone polymers, the method being part.icularly useful for removing the lowest molecular weight polymers from the bulk of a fluid. The solvents which were convenient for this purpose are equilibrated phases from the threecomponent system methanol-carbon tetrachloride-cyclohexane. Methanol and cyclohexane are only partially miscible, the partit#ion coefficient of t,he silicone st,rongly favoring cyclohexane. Carbon t,etrachloride modifies the two phases so that they are made more similar. By dissolving the silicones initially in the heavier phase of the three-component system-that is, the phase which contains most of the cyclohexane and least, of the methanol-placing this phase in the extract,or, and then extracting \\\-it,hthe lighter phase, the extracted polymer was found to have a much loiwr viscosity than that remaining in the extractor. Procedure. A mixture (specific viscosity 0.28) of 3.8 ml. of Dow Corning TO2 and 3.2 ml. of Dow Corning 550 was dissolved
in the heavier phase of the solvent composed of 30 ml. of carbon tetrachloride, 50 nil. of methanol, and 100 ml. of cyclohexane. This gives two liquid layers of almost equal volume. This phase was continuously cycled in the rotary extractor as described and polymer was extracted from it by the flowing of the lighter phase through the extractor. After 100 ml. of the lighter phase had been collected it was analyzed and found to contain 4.1 ml. of polymer having a specific viscosity of 0.19, considerably lower than that of the initial mixture. A further extract of the remaining polymer with 100 ml. of the light phase obtained from equilibrating 50 ml. of carbon tetrachloride, 50 ml. of methanol, and 150 ml. of cyclohexane was shown to contain 0.5 ml. of a polymer having a specific viscosity of 0.26. The residual polymer in the extractor had a specific viscositv of 0.41. Thus, most of thp low molecular weight material had been removed from the mixture. The method is as simple to run as a chromatographic column, all phase separations being carried out automatically. The fractions can be collected on an automatic fraction collector. AS illustrated the mrthod can he used for large quantities of polymer. ACKhOWLEDGMENT
The authors would like to thank Midland Silicones Ltd., 49 Park Lane, London, for the supply of polymers. D. W, Bannister wishes to acknoxyledge his indebtedness to Courtauld's Scientific and Educational Trust for the a v a r d of a research scholarship. LITERATURE CITED
(1) A h , 11.
S.,Williams, R. J. P., and Tiselius, d..Acta Chem.
Scand., 6, 826 (1952). (2) Barry, A. J., J . A p p l . Phys., 17, 1020 (1946). (3) Claesson, S.,A r k i z Kemi, 1, KO.24 (1949). (4) Short, J. F., J . Chern. SOC.,1952, 1278. (5) Spence, R., and Streeton, R. J. W., Analyst, 77, 579 (1952). (6) Williams, R. J. P., Zbid.,77, 905 (1952). RECEIVED for review December 23, 1'253.
Accepted .June 15. 1954.
Maleic and Fumaric Acids Origin of Split Polarographic Waves and Analytical Significance PHILIP J. ELVING'
and
ISADORE ROSENTHALZ
The Pennsylvania State University, State College, f a .
Certain unexplained apparently anomalous effects in the polarographic behavior of maleic and fumaric acids constitute a potential source of error in analysis. These acids have been systematically investigated over the pH range of 0.7 to 12 at different levels of ionic strength and with different buffers. At about pH 5 , the original w-ave of each acid begins to decrease; simultaneously, a second more negative wave appears, whose height increases with pH until it alone remains. The total current for both waves is essentially equal to that for the single wave in the low pH region. Above pH 8, the second maleic acid wave begins to disappear and is practically nonexistent at pH 10. The relation of these wave-splitting phenomena to the kinetics of the acid-anion equilibrium and to the sigmoid relation between pH and Eli2 for each wave is discussed. A possible procedure is indicated for separating pH erects due to bond reduction from those due to acid-anion kinetics. The application of Herasymenko's equation to the data is examined; the indecisiveness of the agreement is discussed. The analytical implications and importance of environmental control in the case of
polarographic wave-splitting are discussed and recommendations are made for the modification and improvement of existing analytical procedures far maleic and fumaric acids.
A
LTHOUGH fumaric and maleic acids have been repeatedly studied polarographically, numerous questions on their electrochemical behavior still remain unanswered. The literature on their polarographic determination has been summarized by Warshowsky, Elving, and Mandel ( 2 1 ) , with the exception of a subsequent paper by Silverman (19). General summaries are given by Kolthoff and Lingane ( I S ) and by Elving and Teitelbaum ( 7 ) . The latter, using conditions similar to those of the present study, investigated the acids and their diethyl esters over the pH range 2 to 10. They ascribe many of the discrepancies between the work of different investigators to the use of noncomparable background solution compositions and to the lack of adequate buffering. They obtained a modified S-shaped curve when E1 2 was plotted against pH and observed an apparently 1 2
Present addiess, University of .\Iichigan. Ann Irbor, ,Mich Present address. Rohni & Haas Co , Philadelphia. Pa