ADSORPTION OF POLYVINYL ACETATE
May, 1958
+
So(g) = S’(trans1.) 9.2N - So(.) = 3/2 R In ilf 26.00 9.2N
+
+
- So(a)
(6)
where M is the molecular mass, N the number of “skeletal” bonded atoms in the compound and is the entropy loss because of the presence of doubly-bonded oxygen in the molecule. Thus, = 205.8 e.u. while SO(f) 126.1 R In ( a o / a e ) = 134.1 e.u. so that an entropy loss of 71.7 e.u. occurs in transferring palmitic acid from the gaseous to the film standard states. A decrease of 42.7 e.u. would have occurred if only the translational entropy of the gas were lost. The larger value found accordingly suggests that appreciable losses of internal entropy must also occur in going from the ’=
+
541
gas to the film, and that the molecules in the fiIm do not possess translational freedom of movement. On the other hand, the entropy increase on spreading from the crystal is sufficientIy large to allow for the rotation of the molecules about their long axes in the film. Measurements of the heat of rotational pre-melting in hexadecano12’ give an entropy increase of about 13 e.u. A similar increase might be expected in the spreading of palmitic acid if axial rotation of the film molecules occurs. The (Ah’,), of spreading for the acid is 18.0 B.U. so that other internal degrees of freedom in addition may also be released. (27) G. 5. Parks and R. D. Rowe, ibid., 14, 507 (1846).
THE ADSORPTION OF POLYVINYL ACETATE’ BY J. KORAL,~ ROBERT ULLMAN AND F. R. EIRICH institute for Polymer Research, Polytechnic Institute of Brooklyn, Brooklyn, N . Y . Received M a y 16, IS67
The purpose of this work was to investigate the adsorption of polyvinyl acetate on to solid surfaces under a variety of conditions. Among the variables which were considered were the adsorbent, the moIecular weight of the polymer, the distribution of molecular weights in the polymer, the presence of small numbers of active groups on the polymer, the solvent from which adsorption takes place, the temperature and the regularity of the adsorbent surface. The rates of adsorption and desorption were also determined. The results of these experiments were analyzed in terms of a simple model of polymer molecules in solution and of present theories of polymer adsorption. A reasonable picture of the polymer molecule adsorbed on a solid surface was created which explains the known facts for the case where the adsorbent is completely or nearly completely covered by the polymer. The results of this icture are in good agreement with estimates of the thickness of the adsorbed layer which were determined from the flow ofdilute polymer solution through a fine capillary.
A. Introduction The adsorption of polymers from solution on to surfaces differs from the adsorption of small molecules in several respects. Firstly the polymer consists of repeated identical chemical groups along the molecule, any one or all of which may be bound to the adsorbent surface, in contrast to smaller molecules which usually adsorb at a single adsorbent site. Secondly, by virtue of its great mass, a polymer molecule diffuses very slowly and further, because of its size and low mobility, it does not penetrate into pores on a solid surface which are easily accessible t o smaller molecules. As a result, it is to be expected that polymer adsorption would be very sensitive to geometric irregularities 011 an adsorbent surface. Thirdly, most polymer molecules are flexible to some degree and possess many internal degrees of freedom some of which are lost or restricted when adsorption takes place. The extent to which the intraniolecular configurations of the polymer molecule are limited is an important factor in the analysis of polymer adsorption. The purpose of the work reported here was to investigate systematically a single polymer, polyvinyl acetate (and to a lesser extent, a simple derivative, a partially hydrolyzed polyvinyl acetate) under a great variety of experimental conditions, and to analyze the results of the experiments in terms of current models of the polymer adsorption process. It was intended that the study be (1) This work was supported in part by the Office of Naval Reaearoh. (2) Submitted in partial fulfillment of the requirements for the Doctor of Philosophy degree a t the Polytechnic Institute of Brooklyn.
sufficiently complete so that a fairly clear picture of the important factors in this particular case of polymer adsorption would emerge. Of the many variables in polymer adsorption which influence or may influence the quantity of adsorbed polymer and the structure of the adsorbed layer, the following were considered: (1) the chemical structure of the polymer molecule, especially the presence of chemical groups (perhaps only a very few) on the polymer molecule which are particularly effective in promoting adsorption: (2) the molecular weight of the polymer; (3) the distribution of molecular weights in the polymer; (4) the solvent from which adsorption takes place; (5) the temperature; (6) the adsorbent; (7) the presence of other chemical species in the solvent or on the adsorbent surface; (8) the geometric smoothness or roughness of the adsorbed surface. I n addition, it is important to know the degree of reversibility of the adsorption process, the rate at which adsorption takes place, and the dependence of these quantities on other parameters such as solvent, temperature and chemical structure.
B. Experimental 1. Materials. a. Adsorbents. (1) Iron Powder.Iron powder prepared from iron carbonyl by Antara Chemicals Division of the General Dyestuff Corporation was used. The sample was an “SF” type, has a chemical analysis of 98.2-98.8% iron, 0.50-0.70% carbon, 0.10-0.13% oxygen and 0.55-0.75% nitrogen. The company states that the average particle diameter is 3 f i . The powder was kept in a vacuum desiccator which was opened only to weigh out samples. (2) Tin Powder.-Finely divided, highly purified tin powder was purchased from the Fisher Scientific Company.
J. KORAL,ROBERT ULLMAN AND F. R. EIRICH
542
It was a 200-mesh powder. Like the iron it was stored in a
vacuum desiccator. (3) Activated Alumina.-Activated Alumina, Grade F-20, manufactured by the Aluminum Company of America. It was sized a t 80 to 200-mesh and was stored in a vacuum desiccator. b. Solvents.-Analytical reagent grade acetone, methyl ethyl ketone, benzene, chloroform and carbon tetrachloride were obtained from the Fisher Scientific Company. Spectrogrades of acetonitrile and 1,2-dichloroethane were obtained from the Eastman Kodak Company. c. Monomer.-Eastman Kodak vinyl acetate was freed of hydroquinone and trace inhibitors by distillation under one atmosphere of pre-purified nitrogen through a column of sixty theoretical plates. This column was constructed using Podbielniak Heli-Pak random type fractionating column packing. The distilled monomer was stored in the dark a t 4'. 2. Preparation of Polymers. a. Polymerization.-The distilled vinyl acetate was photopolymerized in bulk at a temperature of 4' by a procedure similar to that of A. R. Schultz.a Azo-bis-isobutyronitrile was used as an initiator, and a medium pressure mercury arc lamp was used as a source of illumination. The quantity of initiator required was varied in order to obtain polymers of different molecular weight. The polymerization mixture containing 10-15% polymer was diluted by approximately twice its volume of acetone, and precipitated by adding to a large excess of hexane a t 20'. The precipitate was washed with hexane, dissolved in benzene, frozen, and the benzene removed by sublimation. b. Two samples of polyvinyl acetate, the precursors of XYHL and XYSG polyvinyl butyral polymers, were supplied by the Bakelite Company. These were dissolved in benzene, and dried by sublimation of the solvent from the frozen solution. These polymers will be called XYHL and XYSG in what is to follow. c. Two partially hydrolyzed polyvinyl acetate samples in solution were also received from the Bakelite Company. One, designated 24-9 was a 24% solution in toluene and the other designated 28-14 was a 28% solution in methyl acetate. To 20 ml. of each solution was added 80 d.of acetone. The solutions were then precipitated in an excess of hexane. The partially hydrolyzed polymers were not soluble in .benzene and were therefore dissolved in dioxane, frozen, most of the solvent removed by sublimation, and then dried to constant weight in a vacuum oven a t 50". 3. Characterization of Polymers. a. Viscosity.-Viscosity measurements of polymer solutions were made in an Ubbelohde viscometer at 24.7' and intrinsic viscosities ([VI = (n - n o ) / n ~ c were ) determined. c is concentration in g./ml. Methyl ethyl ketone was used as the solvent in all cases and, in addition, for some of the polymer samples measurements were made in benzene, carbon tetrachloride, chloroform, acetonitrile and 1,2-dichloroethane. The flow time through the capillary was sufficiently great so that kinetic energy corrections were negligible. weights and radii of b. Light Scattering.-Molecular gyration were obtained by measuring the light scattered at 45, 90 and 135" to the incident beam.4 Corrections for dissymmetry were made by using the Doty-Steiner tables,s and the depolarization factor was also taken into account. The refractive index increment was determined in a Rayleigh interferometer .6 Some light scattering and viscosity results are listed in Table I. c. Determination of the Per cent. Hydrolysis of the Partially Hydrolyzed Polymers.-An 0.5 N solution of pot.assium hydroxide in methanol was added dropwise to a methanol solution of partially hydrolyzed polymer and refluxed for 2.5 hours and cooled. The excess base was titrated with standard acid and the per cent. acetate calculated. The method was checked using unhydrolyzed polyvinyl acetate, and in several determinations the percentage of acetylated hydroxyl groups on the polymer was calculated to be 100 f 270.
:z0
(3) A. R.. Schulta, J. Am. Chem. S o c . , 7 6 , 3422 (1954). (4) The photometer used was a Brice-Speiser instrument manu-
factured by the Phoenix Precision Instrument Company. (5) P. Doty and R. F. Pteiner, J . Chem. P h y s . , 18, 1211 (1950). (6) Manufactured by the Baird Instrument Co., Cambridge, Mass.
Vol. 62
TABLE I CHARACTERIZATION OF POLYVINYL ACETATESAMPLESI N , METHYLETHYL KETONEAT 24.7' Polymer [?I, ml./g. Mol. wt. X lo-*@ R(A.1.b 2 270 905 460 4 1GO 480 386 108 307 264 6 861 450 7 . 260 XYSG 80.5 250 204 XYHL 55.5 140 188 a The weight average molecular weights have been corrected for dissymmetry and depolarization. R represents the radius of gyration of the molecules and is defined by
where R 2 0 h is the square of the distance from the center of mass of the polymer molecule to the kth segment of the molecule. There are N segments in a molecule. The intensity of the carbonyl peak a t 5.73 p in the infrared spectrum7of the partially hydrolyzed polymer also was used to calculate the percentage of free hydroxyl groups. The results derived from both titration and infrared experiments are in good agreement. 4. Surface Area of Adsorbenk-The surface areas of the iron and tin powders were determined by adsorbing palmitic acid from carbon tetrachloride solution. The quantity of acid adsorbed was determined by measuring the difference between the palmitic acid concentration in the solution before and after adsorption. This concentration was determined by measuring the infrared peak of the carboxyl group a t 5.83 p . Electron micrographs were taken of the iron powder at 1OOOX linear magnification.* The particles showed up as spheres when properly dispersed, and were of variable size, some particles less than 0.1 p and others as large as 10 p . By measuring and counting particles, an average size was determined. The specific surfaces of the iron and tin adsorbents are 2180 cm.Z/g. and 255 cm.Z/g., respectively. 5 . Fractionation of Polymers.-A 0.5% solution of a high molecular weight sample of polyvinyl acetate in acetone was fractionated by successive additions of n-hexane a t 25 f 0.05'. The addition of hexane caused phase separation, and the fractions were obtained by separating the polymer-rich phase, evaporating most of the solvent, dlssolving in benzene, freezing and drying by sublimation. Viscosity and light scattering results on the fractions were obtained. 6 . Determination of the Quantity of Polymer Adsorbed. -The determination of the amount of polymer. adsorbed was conducted by analysis of the polymer solution before and after adsorption. The analyses in all experiments were performed by measurement of the height of the carbonyl peak a t 5.73 p , calibrated with polyvinyl acetate solutions of known concentration. A plot of optical density versus concentration was linear in the low concentration range for all solutions, but showed slight deviations from linearity as polymer concentration increased. The slope of the optical density-polymer concentration curve was found t o be a function of the solvent in some cases, and separate calibration curves were constructed for each solvent. The different polyvinyl acetate samples all fit the same curve in a given solvent except, of course, for the partially hydrolyzed polymers for which separate Calibration curves were constructed. The amount of polymer adsorbed was determined from the difference in the concentration of polymer solution before and after shaking with adsorbent. The ratio of adsorbent to polymer solution did not affect the adsorption-concentration relationship, and was fixed a t 20 ml. of polymer solution to 15 g. of iron and 5 ml. of polymer solution to 20 g. of tin for most experiments. (7) All infrared spectra were determined on a double beam PerkinElmer recording infrared spectrophotometer. The authors are grateful t o Mr. Herbert Talts who carried out the infrared analyses. ( 8 ) The authors wish t o thank Mr. Thomas Turnbull of the North American Phillips Co. for making the electron micrographs.
ADSORPTION OF POLYVINYL ACETATE
May, 1958
7. Analysis for Polymer in Very Dilute Solution.-In order to gather information on adsorption of polymer a t low surface coverage of the adsorbent, it was necessary to conduct the adsorption experiments in the very low concentration range. The analysis of very dilute solutions (0.003%) would be very inaccurate if carried out directly, and the following procedure was adopted. A large sample of a dilute polymer solution (60 ml.) in benzene was adsorbed on 60 g. of iron powder. Forty ml. of the supernatant liquid was frozen and sublimed under vacuum. The remaining polymer was dissolved in 2 ml. of benzene which solution was used for analysis. The analyses were carried out by the infrared technique previously described.
C. Experimental Results 1. Rates of Adsorption.-The rates of adsorption of several polyvinyl acetate solutions were determined on the iron, tin and alumina powders described earlier in this paper. The results given in Fig. 1 show that adsorption on both iron and tin appears to be complete within an hour, but the amount of polymer adsorbed on alumina continued to change with time after seven hours. It should be mentioned that the iron and tin powders are smooth and apparently non-porous while the alumina surface is very irregular. 2. General Characteristics of the Adsorption Isotherms.-While the adsorption isotherms of polyvinyl acetate show characteristic differences when certain parameters, such as solvent, temperature and molecular weight are changed, there are certain characteristics of all isotherms which are similar and these will be discussed first. The amount of polymer adsorbed per unit area of adsorbent is a very rapidly increasing function of the equilibrium solution concentration a t low concentrations, and reaches a plateau which shows little or no change as the polymer solution concentration is further increased. The maximum quantity of polymer adsorbed is several times greater than that which is calculated by assuming that a monolayer of polymer is formed with the polymer molecule extended so that all segments of the polymer molecule are in intimate contact with the adsorbent surface.
6.0 0.5
Temp.,
Solvent
3. Adsorption as a Function of Solvent and Solvent Power.-The effect of the solvent on polymer adsorption has been studied on both iron and tin powders. Good, poor, polar and non-polar solvents have been used. The results are listed in Tables I1 and I11 and appear in Figs. 2 and 3. Also given in these tables are intrinsic viscosities and dielectric constants. The former are good relative measures of solvent power, the better the sol-
4
6
t.
Fig. 1.-The rate of adsorption of polyvinyl acetate, mg. of polymer adsorbed per g. of adsorbent ( A ) versus time in hours. The scale on the left applies to the metal adsorbents, the scale on the right to the alumina. 0, XYSG, benzene, 30.4' onto iron; 0 , 2, 1,2-dichloroethane, 89.5' on to iron; a, XYHL, 1,2-dichloroethane, 30.4' on to tin; c), XYHL, carbon tetrachloride, 30.4' on to alumina.
0.04
0.12
0.20
C.
Fig. 2.-Equilibrium adsorption of XYSG on to iron at 30.4' as a function of solvent, mg. of polymer adsorbed per g. of adsorbent ( A ) versus equilibrium concentration ( e ) in g./100 ml. 0, carbon tetrachloride; 0 , benzene; (3, 1,2-dichloroethane; 0, chloroform.
0.25
Am
In1 (mg./g.P 2 30.4 CzH4Cla 368.0 0.088 2 30.4 Benzene 314.0 ,121 4 51.5 Benzene 186.0 .I72 4 51.5 CCla 65.5 .291 XYHL 51.5 Benzene 64.5 .149 XYHL 51.5 CCl, 22.6 .228 XYHL 30.4 CzHiClz 73.0 ,073 XYHL 30.4 cC14 22.6 ,221 a A is equal to the mg. of polymer adsorbed at maximum adsorption per g. of adsorbent under the given conditions. C.
5.0
2
4 0.15 TABLE I1 ADSORPTION O N TIN POWDER AS A FUNCTION OF SOLVENT Polymer
543
0.05
1
7
1
I'
K
L
t
0.02
0.06
0.10
I
0.14
C.
Fig. 3.-Equilibrium adsorption of polyvinyl acetate on tin as a function of both solvent and molecular weight. Mg. of polymer adsorbed per g. of adsorbent ( A ) versus eqtilibrium concentration ( e ) . e, 4, carbon tetrachloride, 51 ; a, XTHL, carbon tetrachlyide, 51'; 0,XYHL, benzene, 51 ; 0 , 2, benzene, 51 ; a, 4, benzene, 51'; 0, 2, 1,2-dichloroethane, 30.4' ; 8 , XYHL, 1,2-dichloroethane, 30.4'.
vent, the higher the intrinsic viscosity; the latter yields information on the polarity and polarizability of the solvent. Adsorption is generally much higher from a poor solvent than a good solvent, the amount of polymer adsorbed being roughly inversely proportional to the intrinsic viscosity. The case of acetonitrile is exceptional. Acetonitrile is a good solvent for
544
J. KORAL, ROBERT ULLMAN AND F. R. EIRICH
Vol. 62
that the dependence on molecular weight is more pronounced in a poor solvent than in a good solvent. ADSORPTIONAS 0.4
0.02
0.10
0.06
Fig. $.-Equilibrium adsorption of polyvinyl acetate from carbon tetrachloride onto iron a t 30.4' as a function of molecular weight. Mg. of polymer adsorbed per g. of adsorbent ( A ) versus equilibrium concentration in g. poly, 2; 0, 4; @, 6; @, mer per 100 ml. of solution (c). . XYSG; 8, XYHL. 0.8
r
I
M W X 10-6
Am
cc1,
XYHL XYSG
CCL CCI4 CCln Benzene Benzene CzHdClz CzHrClz
4 2 XYSG 2 XYSG 2
1.40 2.50 3.07 4.80 9.05 2.50 9.05 2.50 9.05
1.35 1.54 1.80 2.18 2.74 0.680 .698 .635 .607
CCla
ADSORPTIONAS Solvent
0.4 0.04
0.20
0.12 C.
Fig. 5.-Equilibrium adsorption of polyvinyl acetate onto iron a t 30.4" as a function of molecular weight a d solvent. Mg. of polymer adsorbed per g. of adsorbent ( A ) uersus concentration in g. polymer per 100 ml. of solution. 0, 2, benzene; 0 , XYSG, benzene; Q, 2, 1,2-dichloroethane; @, XYSG, 1,2-dichloroethane.
30.4"
Polymer
Solvent
0.14
C.
TABLE IV FUNCTION OF MOLECULAR WEIGHTO N I R O N POWDER A T
I
1
A
CCL cc1, Benzene Benzene Benzene CsH4C12 CZH,C12
6
A
TABLE V FUNCTION OF MOLECULAR WEIGHTON TINPOWDER
Temp., "C.
Polymer
51.5 51.5 51.5 51.5 51.5 30.4 30.4
XYHL 4 XYHL 4 2 XYHL 2
Mpr
X 10-8
1.40 4.80 1.40 4.80 9.05 1.40 9.05
Am
0.228 .291 .149 .172 .218 .073 ,088
When alumina was used as an adsorbent, much more low molecular weight polymer was adsorbed than high molecular weight material, as is shown TABLE I11 in Table VI. This might well be expected since ADSORPTION AS A FUNCTION OF SOLVENT O N IRONPOWDER part of the irregular alumina surface is accessible to low molecular weight polymer which may not AT 30.4' be available to a higher molecular weight sample. Dielectric Am Polymer [VI Sol5 e n t constant (mg./g.) This difference in adsorption between the two polyXYSG CHaCN 81.0 38.8 0 mer samples probably was accentuated by the fact XYSG CHCla 138.0 5.48 0.348 that the contact time between adsorbent and poly10.0 ,535 CzH4Clz 110.0 XYSG mer solution was only three hours, which is much 93.5 XYSG Benzene 2.32 .680 less than the time required for equilibrium as shown 33.0 2.24 1.54 XYSG CClr by our rate studies. 10.0 0.607 CzHaClz 368.0 2 TABLE VI 2 Benzene 314.0 2.32 0.698 ,
2
111.0 2.24 2.74 cc14 The dielectric constant of the polymer is 3.3-4.1.
polyvinyl acetate and yet no adsorption of the polymer on to iron or tin was detected. At the same time acetonitrile differs from the other solvents used in that its dielectric constant is much higher. The possibility exists that a specific interaction between the nitrile group and the oxidized iron surface may also play a part in precluding adsorption. 4. Adsorption as a Function of Molecular Weight.-Adsorption as a function of molecular weight of the polymer has been studied in carbon tetrachloride, benzene and ethylene dichloride on iron and tin powder. Some measurements were made also using activated alumina as an adsorbent for solutions of PVAc No. 2 and PVAc-XYSG in ethylene chloride at 30.4'. Experimental results on iron and tin are listed in Tables .IV and V and are shown in Figs. 3, 4 and 5. The general conclusion is that the amount of polymer adsorbed increases with molecular weight and
ADSORPTION OF POLYVINYL ACETATE ON TO ACTIVE ALUMINA DICHLOROETHANE AT 30.4" AFTER 3 HOURS AGITATION
FROM
Polymer
M w X 10-1
Am
2 XYSG
9.05 2.75
0.64 2.37
5. Adsorption as a Function of Molecular Weight Distribution.-The polyvinyl acetate fractions, prepared as described earlier in the paper, were studied in adsorption experiments, and the relevant data are given in Table VII. The difference between unfractionated and fractionated polymer is small, but the adsorption of fractionated polymer was uniformly greater for the cases studied. Because of the small differences between the adsorption of fractionated and unfractionated polyvinyl acetate, it was considered unnecessary to use fractionated polymer for most of the other experiments reported here. 6. Adsorption as a Function of Temperature.The experimental results on adsorption versus temperature are listed in Table VIII. Some of
ADSORPTION OF POLYVINYL ACETATE
May, 1958
545
TABLE VI1 ADSORPTION OF FRACTIONATED AND UNFRACTIONATED SAMPLES BY IRON POWDER M w X 10-6 Solvent
Polymer
Temp., 'C.
Am
Fraction No. 2" from 7 8.03 CzHrCll 50.0 0.650 2-(unfractionated) 9.05 CzHdC12 51.5 0.618 Fraction No. 1" from 6 4.70 CClr 30.4 2.30 P(unfracti0nated) 4.80 CC4 30.4 2.18 2.75 CCh 30.4 1.83 Fraction No. 2" from 6 6- (unfractionated) 3.07 CC14 30.4 1.80 a Polyvinyl acetate 7 was separated into 6 fractions. Fraction No. 2 was the second highest in molecular weight. Fractions 1 and 2 of polyvinyl acetate 6 were the top two molecular weight fractions, respectively.
these are shown in detail in Fig. 7. I n these systems the amount of polymer adsorbed increases with temperature, Results from other laboratories also have, under certain circumstances, shown positive changes in adsorption with temperature; in other cases zero or negative changes were recorded. The magnitude of the temperature coefficient is small in all cases reported.
0.04
.
TABLE VI11 hdsorbent
Solvent
Temp., OC.
Am (mdg.)
XYSG XYSG XYSG 2 2 2 XYHL XYHL 2 2
Fe Fe Fe Fe Fe Fe Sn Sn Sn Sn
CCl* CC4 cC14 CzHiClz CaHrClz C2HrC4 CCIr CCla Benzene Benzene
30.4 48.5 69.5 30.4 51.5 69.5 30.4 51.5 30.4 51.5
1.54 1.62 1.66 0.607 .618 ,632 .121 .218 ,221 .228
0.16
Fig. 6.-Adsorption of polyvinyl acetate and partially hydrolyzed polyvinyl acetates from 1,2-dichloroethane onto iron a t 30.4 Mg. of polymer adsorbed per g. of adsorbent versus e uilibrium concentration in g. of polymer per 100 ml. of sjution. 0 , 28-14; 0,24-9; 0 , XYHL.
ADSORPTION AS A FUNCTION OF TEMPERATURE Polymer
0.12
0.08 C.
0.8
0.04
0.12
C.
Fig. 7.-Adsorption of XYSG from carbon tetrachloride onto iron as a function of temperature. Mg. of polymer adsorbed per g. of adsorbent versus equilibrium concentr: tion of polymer in g. per 100 ml. of solution. 0 , 69.5 ; @, 48.5"; 0,30.4'.
used for the polymer solutions. The maximum amount of ethyl acetate adsorbed ( A m ) was taken 7. The Adsorption of Partially Hydrolyzed from the flat part of the isotherm and was calcuPolyvinyl Acetate.-Adsorption isotherms were lated to be 0.065 mg. per g. of adsorbent. Polyconstructed for two samples of partially hydrolyzed vinyl acetate adsorption under the same conditions polyvinyl acetate in l,Bdichloroethane, and the was 20 to 40 times as great. 9. Desorption and Reversibi1ity.-A set of exresults compared with a sample of unhydrolyzed periments were designed to see if the polymer polymer (see Table IX and Fig. 6). The polymer containing the highest percentage of hydroxyl adsorption process was truly reversible. These groups is adsorbed the most, though the relative consisted of adsorbing XYSG from carbon tetraincrease in adsorption caused by a small percentage chloride onto iron or tin a t 30.4", decanting as of hydroxyl groups on the polymer is greater than much of the excess solution as possible, adding some that obtained by further increasing the hydroxyl pure solvent, and desorbing by agitating the iron and solvent slurry. The desorbing solvents used concentration on the polymer. were carbon tetrachloride, 1,2-dichloroethane and TABLE IX acetonitrile. Some polymer was desorbed in all ADSORPTIONRESULTSOF PARTIALLY HYDROLYZED POLY- cases, though in carbon tetrachloride desorption VINYL ACETATES I N CzH4C12AT 30.4 O N IRON POWDERwas very slow and not complete in 120 hours. (BATCH 2) Desorption with acetonitrile was a t least 85-90% Polymer M w X 10-6 [VI" % Hydrolysisb Am(mg./g.) complete in the first washing, and entirely com73.0 0 0.352 XYHL 1.40 plete after two or three washings. The desorption 75.0 13 f 2 0.960 24-9 1.73 using ethylene chloride as desorbent was a t least 28-14 1.30 53.0 26 f 2 1.065 80% complete in a single washing. The words "at least 80% complete" are used since in the first a Intrinsic viscosities were determined in 1,a-dichlorethane. bThis is the average of the infrared and titration mashing the solvent contains both desorbing solvalues. vent (acetonitrile or ethylene chloride) and some 8. The Adsorption of Ethyl Acetate.-Ethyl carbon tetrachloride which was not removed by acetate was used as a monomeric analog of the decantation. Consequently the equilibrium isorepeating unit in polyvinyl acetate. The ad- therm is not known in the mixed solvent and the sorption of ethyl acetate from a solution in carbon total amount that should be desorbed a t equilibtetrachloride was determined by the same procedure rium cannot be calculated exactly.
546
J. KORAL, ROBERT ULLMAN AND E’. R..EIRICH
D. Discussion The rate of diffusion of polymer molecules is much less than the corresponding rate for low molecular weight species. As a result, it is necessary that adsorption experiments with polymer solutions be carried out with vigorous agitation so that the adsorbent surface is accessible to the slowly moving polymer. If the surface is irregular and contains pores into which the polymer molecule can penetrate only slowly, much time must pass before equilibrium is attained. This was particularly evident in the polyvinyl acetate adsorption experiments in which alumina was used as an adsorbent. First of all, the amount of polymer adsorbed on the alumina was still increasing with time even after seven hours of vigorous agitation of solution and adsorbent together. Secondly, the amount of low molecular weight polymer adsorbed on the alumina after three hours was more than three times as great as the quantity of a high molecular weight sample. Some of this difference may be attributed to the decrease in diffusion constant with increasing molecular weight, but it is also probable that there are tiny pores in the alumina surface which are accessible to low but not at all to high molecular weight polymers. It is significant that the molecular weight dependence of adsorption was reversed completely when this porous adsorbent is used. In some other investigations, 9-11 it had been reported that the amount of polymer adsorbed decreases with increasing molecular weight. This is opposite to that which we found on the smooth adsorbents, but is in accord with our results on alumina. It would appear that the decrease in adsorption with increasing molecular weight is a phenomenon which is not characteristic of the polymer, but rather a result of the relative inaccessibility of irregular adsorbent surfaces to the larger polymer molecules. Results similar to ours were obtained by Heller and Tanaka12in their investigation of the adsorption of polyglycols. The area occupied on a liquid-air interface by a monomeric unit of polyvinyl acetate was determiiied1a~14 to be 26 Using this result, and assuming that the polymer molecules lie flat on the surface of a solid adsorbent, the maximum adsorption values (A,) of polymer in all cases corresponds to 10-20 molecular layers of material. If, instead, the effective surface was considered to be that determined by ethyl acetate adsorption and by setting the area of a polyvinyl acetate monomeric unit equal to that of an ethyl acetate/molecule, the polymer film would be 20-40 molecules thick. Whichever standard is used, it is clear that the surface cannot accommodate all the polymer segments in the first molecular layer. In the subsequent discussion it shall be generally assumed that the area of the polymer segment is 26 (9) I. Claesson and S. Claesson, Arkiu Kemi Mineral. Geol., 19A, No. 5 (1945). (10) I. Jullander, ibid., ala, No. 8 (1945). (11) J. F. Hobden and H. H. G. Jellinek, J . Polymer Sci., 11, 305 (1953). (12) W. Heller and W. Tanaka, P h y s . R e v . , 83, 302 (1951). (13) D.J. Crisp, J . Coll. Sci., 1, 49, 161 (1946). (14) M.J. Schick, J . Polymev S c i . , 2S, 465 (1957).
Vol. 62
It should be remembered that under the circumstances of these adsorption experiments, the value 52 A.2 might be considered a better choice. This does not change the trend of the results to be discussed in any way though the number of effective monolayers is doubled by use of the latter value. The general shape of the adsorption isotherms of polyvinyl acetate shows a steep rise and then a flat top in which the amount of polymer adsorbed per gram of adsorbent does not change appreciably with increasing solution concentration, is characteristic of monolayer adsorption for non-polymeric materials, but because of the quantity of polymer adsorbed, an ordinary monolayer cannot describe the polymer configuration. I n a more reasonable picture of polymer adsorption one would envisage the polymer molecule attached to the adsorbent a t a number of sites with most parts of molecular chain free to move in the region of solvent above the surface to which the chain is bound. From the ethyl acetate adsorption data, one would expect that a polyvinyl acetate molecule would be attached to an adsorbent at anywhere from 1 to 10% of the possible bonding groups depending on experimental conditions. The above model may be further specialized by assuming that the space occupied by a polymer molecule on the adsorbent may be calculated from the radius of gyration, R, of the polymer molecule in solution. If both radius of gyration and molecular weight are known, theoretical A values may be computed and compared with experimental results. A m is the maximum amount of polymer adsorbed from a given system. It is expfessed as mg. polymer per g. of adsorbent. In making this calculation, it is assumed that n molecules of radius of gyration, R, will exactly cover the surface of the adsorbent S ( S = nrR2). On this basis the calculated maximum adsorption A m is given by p)
lOOOMX A m = ___
(1) rR2N where M is the molecular weight of the polymer, and N is Avogadro’s number. From the light scattering measurenieiits in methyl ethyl ketone, both M and R 2 are known. However, R2 is a function of the solvent in which the polymer is dissolved, larger values of R corresponding to better solvents. R has been calculated in solvent systems other than methyl ethyl ketone by using the (at least approximately correct) relation that the intrinsic viscosity is proportional to R3. From viscosity measurements in the various solvents and the light scattering results in methyl ethyl ketone, R may be obtained from the equation R = RMEK([71/[71MEK)1/S (2) Using this relation in eq. 1, theoretical A values have been calculated. These are listed in Table X. The experimental values are 7 to 40 times greater than the theoretical, which shows that the model of the adsorption of the undistorted polymer coil is a gross oversimplification. A modification of this model could be formulated by assuming that the internal configurations of the polymer molecules in solution are distorted so as to occupy a smaller area when adsorbed on the sur-
ADSORPTION OF POLYVINYL ACETATE
May, 1958
face. Each polymer molecule is assumed to be ellipsoidal (or cylindrical) with its long axis perpendicular to the plane of the adsorbent. Because of the lateral compression of the coil, there is a loss in entropy of the polymer molecule the magnitude of which may be calculated from a formula developed by Wa11.16 Presumably, the energetic gain would be sufficient under usual experimental conditions so that the laterally compressed polymer on the surface plus the surrounding solution would be in a state of minimum free energy. The modification suggested in the previous paragraph is not the only possibility. It may be that entanglement and interpenetration of the coiled polymer molecules at a high local concentration a t the surface takes place to a considerable extent, and that the lateral compression is not a significant factor, This interpenetration would increase the number of polymer-polymer nearest neighbors at the expense of solvent-polymer nearest neighbor in the vicinity of the surface, and this, in a good solvent at least, corresponds to an energetically unfavorable arrangement. A small amount of direct evidence from precision viscosity measurements is available on this subject. Ohrnl6 determined the thickness of an adsorbed layer of polystyrene ( M w= 500,000) on the walls of the glass capillary of an Ostwald viscometer by measuring the relative increase in flow time of a polystyrene solution in toluene for two capillaries of different raGius. He found this thickness to be about 1500 A. A number of authors have studied the dimensions of polystyrene in toluene. In one investigation it was found that the end-to-end distance of the polymer molecule was related to the molecular weight by the equation17 L = k'M" (in A.), k' = 0.196, a = 0.61 The linear dimension which is characteristic of flow properties of the polymer is an effective hydrodynamic radius's given by Rh2
=
5 18 L2
For 9 polystyrene polymer of M = 50O,OQO, Rh e 300 A., the hydrodynamic diameter is 600 A., which is 1/2.5 less than the actual thickness of the adsorbed layer. This is strong evidence that some biaxial compression of the polystyrene molecuIe takes place on adsorption. Some similar experiments have been publishedlg in which the thickness of the layer of polyvinyl acetate adsorbed from a toluene solution on to a glaFs capillary was found to be of the order of 5000 A. This polymer had a viscosity molecular weight of 800,000which in toluene would have a hydrodynamic diameter of about 1000 A. A prolate ellipsoid with a long axis of 5000 A. could serve as model of a polyvinyl acetate molecule on a surface. If this model is correct, the A , (theor.) values in Table X would be increased by a (15) F.T. Wall, J . Chem. Phys., 10, 132, 485 (1942). (16) 0. E. o h m , J . Polymer Sci., 17, 137 (1955). (17) N. T. Nutley and P. J. W. Debye, zbzd., 17,99 (1955). (18) J. J. Hermans and J. Th. G. Overbeek, Rec. trav. chzm., 67, 761
(1948). (19) C. A. F. TuiJnman, Thesls, Lelden, 1956; C. A. F Tuijnrnan and J. J. Hermans, J. Polymer Scz., 26, 385 (1957).
547
factor of five thus corresponding much more closely to the experimental results. Even this is not sufficient to account for the total amount of adsorbed polymer, and the probable explanation is that not only are the polymer molecules laterally compressed into a cylindrical or ellipsoidal form but also a certain amount of intermolecular entanglement does occur. The relative importance of molecular distortion and interpenetration in the surface layer cannot be resolved a t the present time but depends on more direct studies of the structure of the surface layer. The expected dependence of polymer adsorption on molecular weight may be examined by setting R 2 = KMB in eq. 1 where 1 S ,6 5 1.4. jl = 1 if the polymer is dissolved in a poor solvent, and increases toward 1.4 as the solvent becomes better.20 The result obtained is that (3)
which would require that polymer adsorption be independent of molecular weight in a poor solvent, and decrease with molecular weight in a good solvent. The experimental result is that A m increases markedly with molecular weight in carbon tetrachloride, a poor solvent, and increases only slightly with molecular weight in benzene and l12-dichloroethane which are good solvents. This might be understandable if the ease with which the polymer coil is deformed upon adsorption were a function of molecular weight so that (4)
where 4 ( M ) is a deformation function which is independent of solvent. Taking 4 ( M )to be of the order of Mo.2would provide a reasonably good fit with the experimental adsorption results. The dependence of adsorption on solvent power according to eq. 1 should lead to the result that A , [ q ] ' / ~= const. as the solvent is varied, and the polymer kept the same. This does not fit the experimental results too well, though A m is certainly inversely related to the intrinsic viscosity. One factor which complicates the relation between polymer adsorbed and intrinsic viscosity (or molecular size) is that the solvent, to a greater or lesser extent, competes with the polymer for adsorption sites, and that the number of sites on the surface occupied by a single polyvinyl acetate molecule depends not only on its size but also on the relative affinities of the solvent molecule and a polymer segment for the adsorbing surface. A particularly striking case is that of acetonitrile from which no adsorption of the polymer onto an iron or tin surface could be detected. Adsorption does take place from the other solvents, but it is not clear to what degree solvent-surface interaction controls the ultimate quantity of adsorbed polymer. The influence of temperature on polymer adsorption is shown in Figs. 7 and 8, and Table XI. (20) (a) P. J. Flory, J . Chem. Phys., 17, 303 (1949); (b) T. G . Fox, Jr., and P. J. Flory, THISJOURNAL, 63, 197 (1949); ( 0 ) P. J. Flory and W. R. Krigbaum, J . Chem. Phys., 17, 1347 (1949); (d) iaia., is, 1 0 8 ~(1950).
'
J. KORAL, ROBERT ULLMANAND F. R. EIRICH
548
Vol. 62
TABLE X Polymer
Benzene Am (theorJa
[?]
Am (exp.) b
[SI
Chloroform Am Ao. (theor.)' (exp.) b
[.I]
Dichloroethane Am Am (theor.)' (exp.) b
Carbon tetrachloride ACO Am (theor.) (exp.)
[?]
2 314 0.044 0.698 368 0.044 0.607 111 0.090 2.74 .068' 2.18 4 65.5 XYSG 93.5 0.062 0.680 138 0.048 0.348 110 0.056 0.535 33.0 .126 1.54 XYHL 22.6 .081 1.35 A m (theor.) is calculated from eq. 1. [77] is obtained from viscositieR measured at 24.7'. A m (exp.) is obtained froni adsorption isotherms on iron powder. Similar results were obtained using tin as an adsorbent.
TABLE XI HEATSOF ADSORPTION OF XYSG CHLORIDE ON TO
0.1 h
m
e
P
-2 .,. fi g l
v
u 0.01
2.8
3.0 3.4 1/T X 10s. Fig. &--The concentration of polymer a t a given surface coverage ver.sus the reciprocal of the absolute temperature. Data taken from the adsorption of XYSG from carbon tetrachloride onto iron. See Fig. 7: 0, A = 1.5; A = 1.4; 0, A = 1.3; a,A = 1.2.
*,
One may approximate the "isosteric" heat of adsorption from these data in the following way. The chemical potential of the polymer in the liquid phase (11) is equal to the chemical potential of the polymer a t the surface ( p 3 ) . Therefore one may write
UP
+ RT In a1 = uso+ RT In a.
(5)
where a1 and a, are activities of the polymer in the liquid and on the surface, respectively. Differentiating this relationship leads to
where Hos - H J are the partial molal enthalpies of the polymer on the surface and in the liquid. The polymer solutions in these systems are very dilute, and therefore a1 is proportional to concentration GI. If it is assumed that the activity of the polymer at the surface does not change if the amount adsorbed is kept constant, eq. 5a may be written
The assumption that a, is constant if cs is unchanged is the weakest point in the argument, since the arrangement of polymer molecules in the surface phase may change drastically at constant surface concentration. This possible limitation should be kept in mind in examining AH data. Data from the adsorption of polymer XYSG on iron powder from carbon tetrachloride solution at
FROM
CARBON TETRA-
IRON POWDER
Amt. of polymer on surface, mg./g. adsorbent
AH X lo-', oal./rnole
1.5 1.4 1.3 1.2
37 41 49 62
30.4,48.5 and 69.5' were used to calculate A H . The AH values have been calculated from the data in Figs. 7 and 8. They are all positive and vary with surface coverage as shown in Table X. All these data are obtained from systems in which the surface already was mostly covered by adsorbed polymer. It was not possible to determine AH at low surface coverage on this system because of the difficulty in analyzing for polyvinyl acetate with accuracy at very low solution concentrations. From the graphs in Fig. 8, it appears that AH tends to become more positive a t low surface coverage, though this tendency is not large. I n another system (polyvinyl acetate No. 2 in ethylene chloride), AH appears to be negative a t low coverage and is positive at higher coverage. The data in the low concentration range are somewhat erratic, and therefore we regard this last result as tentative. These experimental values of AH reflect several factors on a molecular scale, among which are the heat of adsorption of polymer segments, the heat of adsorption of solvent molecules, and the heat or energy associated with the rearrangement of those polymer segments of adsorbed molecules, which segments are not directly attached to the surface. The free energy of adsorption must be negative, however, for adsorption to take place, and since AH is positive, TAX must be positive and greater than AH(AF = AH - T A S ) . This is quite understandable in this system since the adsorption of one polymer molecule on several adsorption sites releases several solvent molecules into solution. The number of translational degrees of freedom is substantially increased, and the entropy change is therefore positive and relatively large. The adsorption of partially hydrolyzed polyvinyl acetate on an iron surface is several times greater than the adsorption of a sample of unhydrolyzed polymer as may be seen from Fig. 6 and Table IX. This might well be expected since the iron is coated with a thin film of oxide on which a certain small amount of water is adsorbed, and the free hydroxyl groups possess a relatively high affinity for such a surface. An additional factor of interest is that the relative effect on ultimate adsorption of increasing the percentage of free hydroxyl groups is most pronounced for the first few hydroxyl groups,
ADSORPTION OF POLYVINYL ACETATE
May, 1958
on the molecule. Sample 24-9 containing 13% of hydroxyl groups adsorbs over three times as much as XYHL (containing no hydroxyls), while sample 28-13 containing 26% hydroxyl groups adsorbs only 25y0 more. If it is imagined that a loosely packed polyvinyl acetate layer on the adsorbent surface does not saturate the surface sites, a few hydroxyl groups on the polymer will greatly enhance the amount of polymer adsorbed. However, this partially hydrolyzed polyvinyl acetate forms quite a dense layer on the surface, and the introduction of further hydroxyl groups has a lesser influence on further adsorption than the first few per cent. did. It also should be pointed out if the energy of the bond between the hydroxyl group on the polymer and the adsorbing surface were sufficiently great, and the loss in entropy arising from uncoiling the polymer chain not too excessive, then single polymer molecules containing hydroxyl groups would uncoil on the adsorbent, fewer polymer molecules would be required to cover the surface, and less rather than more polymer would be adsorbed. This does not occur with our polymers, and no examples of such behavior on other polymer adsorbent systems have been reported to the best of our knowledge. One of the curious results of these polyvinyl acetate adsorption experiments (also reported in other polymer adsorption studies) is that many of the results fit the Langmuir isotherm A=--1
kC
+ Ic'c
(6)
where A is the amount of polymer adsorbed per unit of adsorbent, and c is the concentration of polymer solution in equilibrium with the adsorbed layer. The Langmuir isotherm is derived for an assembly of rigid molecules, and should not be expected to be valid where the molecules can change shape and occupy varying areas on the surface as the solution concentration is varied. I n this connection it should be mentioned that experimental plots of c/A vs. c, and 1/A vs. l/c should be linear if the Langmuir equation is valid. These two plots tend to emphasize experimental results in different concentration ranges. If c is the abscissa, low concentration experiments are compressed on the curve near the c/A axis, while if l/c is the abscissa, the low concentration data are spread out and the results at high concentration all fall together near the vertical axis. When c / A is plotted versus A , the Langmuir isotherm seems to apply quite well. This is shown in Fig. 9 which was drawn from the adsorption of XYSG from benzene onto iron at 30.4'. Other results which conform to the Langmuir isotherm are also shown in Fig. 10. Note that these curves deviate systematically from linearity in the low concentration range. The adsorption of XYSG from benzene onto iron a t 30.4" has been investigated in the very low concentration range. If a graph of 1/A versus l/c is constructed, the resulting line is not straight. This may be in part due to the inapplicability of the Langmuir isotherm to polymer adsorption, and in part due to inaccuracy in the measurements at low concentration.
549
0.1
0.05 C.
Fig. 9.-A Langmuir plot c / A versus c drawn from the adsorption of XYSG from benzene onto iron at 30.4'.
0.05
0.10 C.
Fig. 10.-Langmuir plots c / A versus c drawn from the adsorption of several samples from carbon tetrachloride at30.4'. 0, XYSG; 0 , 6 ; a J 4 ; @, 2. I
I n a recent set of paperslZ1a statistical theory of polymer adsorption was developed. This theory is based on the model of a Gaussian chain which is deposited on a surface containing adsorbing sites. I n this model some segments of the chain adsorb on the surface while others penetrate into the solution forming loops above the surface. The theory is applied primarily to the case where the adsorbing surface is only slightly covered, though a section of the second paper is devoted to the case where the polymer concentration a t the surface is high. The great majority of the data which have been presented in this paper apply to the nearly completely covered surface, and for this reason agreement or lack of agreement between theory and experiment is by no means a convincing test of the theory. Nevertheless a brief comparison between theory and experiment will be attempted in what is to follow. 1. The theory predicts an adsorption isotherm which initially rises more steeply than the Langmuir isotherm but reaches its asymptotic limit very slowly. I n the case where the polymer is adsorbed on a single segment, the Langmuir isotherm becomes valid. Also, when the surface is largely covered and the polymer segments interfere considerably with each other, the Simha-Frisch-Eirich (SFE) isothermz1approaches the Langmuir isotherm. The adsorption isotherms determined herein seem to obey the Langmuir isotherm fairly well in (21) (a) R. Simha, H. L. Frisch and F. R.Eirioh. THISJOURNAL. 57, 584 (1953): (b) H. L. Frisch and R. Simha, ibid., 18, 507 (1954); (c) H. L. Frisch, ibid., 59, 633 (1955).
A. K. GRZYBOWSKI
550
the high concentration range. This is in qualitative agreement with the theory since this corresponds to high surface coverage, and consequently there is much interference between polymer segments. The experimental isotherms appear to be quite flat, though in some cases the amount of polymer adsorbed continues to increase as the solution concentration goes up. Further adsorption experiments a t much higher solution concentrations would be required to determine whether the asymptotic limit of adsorption is approached slowly. 2. The theory predicts that adsorption from a poor solvent is favored over adsorption from a good solvent which is in agreement with the fact that adsorption of polyvinyl acetate from carbon tetrachloride solution is much greater than adsorption of the polymer from solut,ions in better solvents. 3. The theory predicts enhanced adsorption
Vol. 62
with increasing molecular weight, but that chain interference near the surface would tend to decrease the molecular weight dependence perhaps in an extreme case causing the over-all adsorption to decrease with increasing molecular weight. The results of the experiments indicate that polymer adsorption increases with increasing molecular weight, but that this effect is most pronounced in a poor solvent where chain interference near the surface is greatest. This latter is in contradiction with the SFE theory, but probably could be brought into agreement if the known modifications of the random coil statistics of the polymer chain in a good solvent were specifically introduced into the theoretical calculation. 4. The theory predicts that adsorption may either iiicrease or decrease upon elevating $he temperature. The experiments show an increase in adsorption at higher temperatures.
THE STANDARD POTENTIAL OF THE CALOMEL ELECTRODE AND ITS APPLICATION I N ACCURATE PHYSICOCHEMICAL MEASUREMENTS I. THE STANDARD POTENTIAL BY A. K. GRZYBOWSKI Department of Biochemistry, University College, London, England Received August $8,I967
Details are given of the preparation of a reproducible calomel electrode suitable for use in cells without liquid junction. The standard potential of the electrode has been determined a t 5" intervals from 0 to 60". The cubic equation relating the EO to absolute temperature is Eo = 0.266469 (3.46466 X 10-4)(t - 30) - (2.86482 X 10-6)(t - 30)2 (8.5384 X 10-9) (t - 30)a. The thermodynamic quantities associated with t'he cell reaction and the activity coefficients of some HCI solutions (0.001, 0.002, 0.005, 0.01 and 0.02 mole of HC1 per 1000 g. of water) have been calculated at 25" and compared with the values obtained in other investigations.
-
I. Introduction The very accurate value of the standard potential (*lo p . ) obtained by Hills and Ives1V2at 25" for an improved calomel electrode suggested the possibility that it might prove superior to the Ag ;AgC1 electrode in accurate physicochemical measurements involving cells without liquid junction. This investigation was undertaken to study the behavior of the calomel electrode and to measure its standard potential over a wide temperature range in the hope of applying it in accurate determinations of acid dissociation and metal complex stability constants, together with the associated thermodynamic quantities, of compounds important in biological systems. The method of preparation of the electrode itself is based on that of Hills and Ives;2 the cells and electrode vessels, however, have been redesigned to yield reasonably rapid and accurate results with not unduly large amounts of test solution. 11. Experimental Procedures The cells were essentially similar to those described by Ashby, Crook and Dattaa for use with Ag;AgCl electrodes ( 1 ) G . J. Hills and D. J. G . Ives, Nature, 166,530 (1950). (2) G . J Hills and D. J. G. Ivsa. J . Chem. SOC.,305 (1951). ( 3 ) J. H.Ashby, E. M. Crook and 6 . P. Datta, Biochem. J . (Lond o n ) , 66, 190, I98 (1954).
+
with two modifications, namely, an additional set of saturators for the calomel electrode compartment and a tap between the two compartments. The whole cell could thus be kept thoroughly deoxygenated by bubbling .with hydrogen (British Oxygen Co.) freed from oxygen by "deoxo" platinum catalyst purifiers. This is particularly important as the calomel electrode is known to be very sensitive to dissolved ~ x y g e n . ~The ! ~ tap was used to connect the two electrodes only when measurements were made in order to minimize any reduct,ion of mercurous ions a t the hydrogen electrode, although no considerable alteration of the e.m.f. was in fact observed (HCI solutions) when the two compartments were allowed to remain in contact for a period of time. The cells and electrode vessels were rendered hydrophobic by treatment with a 1% CC14 solution of silicone fluid (Midland Silicones Ltd.) MS 200/1000 C.S. and baking a t 180" for several hours, the excess silicone being subsequently removed by washing with pure CCla. This was done in the hope of reducing ion exchange between the solution and the glass and to prevent the formation of a film of solution between the mercury and calomel of the electrode and electrode vessel. a (i) The Calomel Electrode.-A.R. grade mercury was purified by distillation under a reduced pressure of oxygen and washing with dil. .HNO3-Hg2(N03)2 mixture. Just before use final purification was carried out by distillation in vacuo in an all-glass still; oxygen was excluded by displacing the air in the still before evacuation with nitrogen or hydrogen. The calomel was prepared by the addition, with continuous stirring, of an excesa of bromine free 2 N HCl to a solution of purified Hg2(N03)z(about 20 g. in approximately (4) R. H. Oerke, J . Am. Chsm. Soc., 44, 1684 (1922). (5) M . Randall and L. E. Young, ibid., 60, e89 (1928).
