ADSORPTION ON HETEROPOLAR SURFACES' ALFRED CLARK* AND B. D. THOMAS' Battelle Mentorial Institute, Columbue, Ohio Received March $6, 1858 3NTRODJJCTION
The properties of the solid-liquid interface have Peceived much attention recently because of their direct application to problems suoh as flotation and lubrication in many industrial fields. Although the phenomena encountered in these fields are explainable in terms of the theory of this interface, theoretical investigations of the interface have lagged far behind the practical applications. This is because the complexity of the solid surface introduces numerous experimental difficulties in any study of the solid-liquid interface. One of the most obvious phenomena connected with the solid liquid interface is the liberation of heat when such an interface is formed. This is usually called the heat of wetting or the total energy of immersion. The literature contains much work on the heata of wetting, but a great deal has been done on silica gel and charcoal. Since these substances possess niihute capillaries in which adsorption takes place, they present phenomena entirely different from and more complex than those obtained when impenetrable crystalline particles $re used. Thie paper has been restricted to the latter materials, since a study of such systems will probably throw more light on the nature of the solid-liquid interface. In particular, the heats of wetting of silica, calcium fluoride, lead sulfate, and barium sulfate powders by water and by methyl, ethyl, propyl, isopropyl, and butyl alcohols have been determined. BRIEF R ~ S U MOF ~ EARLIER WORK
In 1822 Pouillet (16) observed that heat is evolved when finely powdered solids are wet by liquids. Four liquids-alcohol, oil, ether, and waterwere used with various solids,-metals, oxides, glass, porcelain, silica, and sulfur. H e eonnected the heat effects with physical phenomena, such as capillarity. Prrseeted at the Ohio-Michigan Regional Meeting of the American Chemical Society, held at Columbus, Ohio, November 19, 1937. * Research Associate. 8 Physical Chemist. 679
580
ALFRED CLARK AND B. D. THOMAS
Thirty years later Tate (18), on the basis of his experiments, concluded that chemical action a t the interface was the explanation for liberation of heat. Maschke (13) attempted to account for it on the basis of friction between the liquid and solid. Rose (17) and Jungke (9) agreed that solids condense a layer of liquid upon them under compression and that a heat of compression was liberated. Lamb and Coolidge (11) have had much success with the compression theory in explaining heats of wetting of charcoal by various vapors. Martini (12) and Cantoni (3) agreed on the solid solution theory, reasoning that, just as solids are dissolved by liquids and thereby become liquid, so water may be dissolved by silica and thereby become solid. Gaudechon (4) believes the heat of wetting to be due to association or polymerization of molecules of liquid in contact with a solid. Patrick and Grimm (15) have considered the heat of wetting of silica gel from the standpoint of surface energy changes involved. They distinguish between capillary adsorption and adsorption on a plane surface. The initial film of adsorbed liquid is admitted to be under compression, but is held to be negligible in amount as compared with’the total condensation in a solid having capillary struoture. Harkins and Ewing (7) have obtained direct experimental evidence of the compression of the layer of organic liquids adsorbed on activated charcoal. Bartel and coworkers (1) have conducted extensive researches on solidliquid interfaces, measuring the adhesion tension (free energy of adsorption) by a very ingenious method. de Boer and coworkers (2) have done considerable work on the adsorption of dipoles on calcium fluoride and have used electrostatic equations to interpret their results. Illiin (8) and coworkers have had a measure of success in calculating the heat of wetting barium sulfate by water by use of the formula
E = -N p Z e r: where E is the potential of adsorption, N the number of lattice ions per square centimeter, Ze the charge on the lattice ion, T O the equilibrium distance from the center of the lattice ion to the center of the dipole, and p the dipole moment. THEORETICAL
The adsorbed dipole on a heteropolar surface has been pictured as an oscillating molecule, oscillating about an equilibrium position, T O , the distance of the center of the lattice ion from the center of the dipole. This equilibrium position is mttintained by two opposing forces, one attractive, the other repulsive. I n figure 1, C, D, and B represent the centers of lattice ions on a portion
ADSORPTION ON HETEROPOLAR SURFACES
58 1
of a heteropolar crystal surface, and A represents the center of a dipole adsorbed on the positive ion whose center is C. In general then, letting ad = T , where T is the distance of the center of the lattice ion from the center of the dipole a t any point of its oscillation and d is the lattice constant, the attractive energy is
where p is the dipole moment, 2 the valence of the lattice ion, and e the electrostatic charge. This may be summed up over the surface. For the first three charges the energy terms will be
FIG.1. Forces acting on an adsorbed dipole Summing over the entire surface we obtain as the attractive energy of an oscillating dipole adsorbed on a heteropolar crystalline surface
+
. . . E,,= E
I
- rZe9 dZ
(-l)*(n+m)[a2+ n2 + m’]”’
wherein m = c/2 and c takes on all integer values positive and negative and 0. For the energy of adsorption at the equilibrium position this becomes
where a& = TO. The repulsive energy will be in the form:
where b is a constant, and p i s greater than 2 and in all probability is of the order of 10. The repulsive terms arising from charges other than the
582
ALFRED CLARK AND B.
D. THOMAB
primary one upon which the dipole is adsorbed may be neglected becaucle of this large exponent. Therefore we may write as the complete expression for a dipole on a crystal surface:
This expression neglects such secondary effects as induction. interaction between dipoles, and lattice layers other than the first. If we wish to take account of layers other than the first, the expression becomes
where q is 0 for the first layer, 1 for the second, etc. This expression may be used for calculating the e.,ergy of adsorption of powders of known area of the sodium chloride type, wherein the naturally occurring faces are all identical. EXPERIMENTAL
A . Preparation of powders (1) Silica. Silica sand was put through a disc mill several times. The resulting product was grey in color, 8s it contained considerable quantities of iron from the grinder plates. It was treated several times with a mixture of strong nitric and hydrochloric acids to dissolve this iron. The silica was then washed by decantation with distilled water at frequent intervals for a period of 3 weeks. It was then dr'ed and ignited to drive off residual water and carbon, which remained afte the solution of the iron. A pure white material was obtained which, however, was not fine enough for determination of heats of wetting; therefore it was ground wet in a porcelah ball mill for a period of 24 hr. The resulting product was dried atld ignited. Accurately weighed quantities of this material were placed in thin glass bulbs, sealed to a vacuum apparatus, and heated to 400°C. for 24 hr. in vacuo. (2) Fluorite. Very pure powdered fluorite was obtained from the RosiClare Lead and Fluorspar Mining Company and was ground in the ball mill similarly to the silica, dried at llO°C., and heated in uacw, just below
4
400°C. (3) Lead sulfate. Pure crystalline lead sulfate was prepared by adding 6 N sulfuric acid to a saturated solution of lead nitrate, followed by washing by decantation until free from acid. It was heated in thin glass buibs
ADBORFTION ON EETEROPOLAR BURFACEB
583
in uacw) to 200°C. Above this point sintering occurred. This was the method used by Koehler and Matthews (10). (4) Barium sulfate. Barium sulfate was prepared by the method of von Weimarn (19), which gave a pure crystalline product. To a 0.2 N solution of barium thiocyanate a saturated solution of manganous sulfate was added. The precipitate was washed by decantation, dried at llO°C., and heated in uacw) to 200°C.
B. Purifiation of alcohols The c. P. alcohols were refluxed and distilled over lime, the first and last portions being rejected.
C. Apparatus The apparatus for the evacuation of the powders consisted of a vacuum oil pump, a mercury ditTusion pump, and a small electrical resistance furnace for heating the bulbs of powder. The calorimeter itsplf consisted of a Dewar flask with an especially designed brass top to which the flask was fastened. The temperature rise crtused by the heat of wetting was determined with the aid of a 36-junction copper-Constantan thermopile. Each junction was insulated by dipping it into a dilute solution of Duco cement in acetone and drying it. The thirty-six hot junctions were inserted into an 8-mm. Pyrex tube and just brought into contact with a thin glam membrane at one end of the tube, the tube then being filled with paraffi. The cold junction was constructed similarIy. This thermopile could be used in any type of solution without fear of damage. The sensitivity was such that 148 microvolts was equivalent to 0.1OC. The rrtlorimeter was electrically calibrated by a nichrome heating coil. The E.M.F. tmoss the coil was measured and the current in the coil measured by determining the drop across a standard 1-ohm resistance in series with the circuit. The constant-temperature bath in which the calorimeter was immersed was held at 25OC. The cold junction of the thermopile was also immersed in a Dewar flask and maintained at 25°C. in the bath.
D. Procedure The procedure for the determination of the heats of wetting was as follows: The sample of the powder in the thin glass bulb was placed in a holder in the calorimeter, and the calorimeter, filled with 300 cc. of liquid, was damped into position. The constant-temperature bath was regulated to 25°C. and the Dewar flask, containing the cold junction of the thermopile, filled with water a t this temperature. Twenty-four hours was allowed to reach equilibrium, The calorimeter stirrer was started, and
584
ALFRED CLARK AND B. D. THOMAS
the electrical heater was used to bring the temperature of the calorimeter to that of the cold junction. Ten minutes before the bulb was broken temperature readings were taken at 1-min. intervals in order to determine the rise in temperature due to stirring. The bulb of powder was then broken by means of a rod inserted through the hollow stirrer shaft, and temperature readings were taken every minute. The heat of wetting wa8 usually completely liberated in 3 min. After temperature readings were taken for 5 min., the heating coil was connected for exactly 1 min., the current and voltage being measured. The temperature rise waa again measured by the thermopile. By comparison with this temperature rise, produced by a known quantity of heat from the heating coil, the heat evolved during the wetting process was calculated. At least five determinations of the heats of wetting of each powder by each liquid were made; they agreed within approximately f 0.01 cal. per gram of powder. RESULTS AND DISCUSSION
i
As shown in table 1, the heats of wetting of calcium fluoride and lead sulfate by water and by methyl, ethyl, propyl, isopropyl, and butyl alcohols are zero. The result on lead sulfate is in agreement with the work of Koehler and Matthews (lo), who have previously found that the heat of wetting of lead sulfate by water is zero. On the basis of any electrostatic theory of adsorption it is difficult to explain the wide variance between barium sulfate and lead sulfate, since they have the same crystal structure, their lattice constants are very close to each other, and their negative ions are identical. The greatest difference is in the radii of the metallic ions,lead ion = 0.85 A. and barium ion = 1.43 A. This, if anything, should make dipole adsorption of somewhat greater energy on the lead ion. This is an anomaly which needs further work for explanation. It is believed, however, that it is probable that the surface has been altered in some manner, possibly by the reaction with adsorbed water during drying. de Boer and Dippel (2) have presented evidence to this effect in the case of calcium fluoride, which reacted with adsorbed water while drying in vacuo above 400°C. to form Ca(0H)F and HF. I n other words, calcium fluoride and lead sulfate may not adsorb water in the ordinary manner, but may form some sort of definite hydrate which is not removed by ordinary treatment, and which, consequently, prevents further appreciable adsorption when these powders are immersed in liquid. The reason for using the homologous series of alcohols for determinations of heats of wetting is because they all have the same dipole moment within the limits of experimental error, namely 1.6 X lo-" E.S.U. (5). The experimental values obtained give some information as to the orientation
585
ADSORPTION ON HETEROPOLAR BURFACE8
of the adsorbed molecules when considered in the light of the dipole theory of adsorption, wherein the potential of dipole is given by the formula:
E = -PZe r2 If all the hydroxyl groups were oriented towards the crystal surface, one would expect all the alcohols from methyl through butyl to have practically the same energy of adsorption. Since this is not the case, we make the assumption that the hydroxyl group is oriented towards the crystal surface when it is adsorbed on the positive ions, and reverses itself on the negative ions. Adsorption then on negative ions is weak and orientation is probably much less pronounced than on the positive ions. The decrease in heats of wetting of silica gel as one ascends the homologous series of alcohols has been explained by increasing steric hindrance in the fine capillaries of the gel. Although there are probably many surface cracks on the silica that we have prepared by grinding, the hindrance TABLE 1 Average values of heats of wetting in calories per gram CaFr
Methyl alcohol.. . . . . . . . . . . . . . . . . . . . . . . . . . . . Ethyl alcohol.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . n-Propyl alcohol. . . . . . . . . . . . . . . . . . . . . . . . . . . n-Butyl alcohol. . . . . . . . . . . . . . . . . . . . . . . . . . . .
0.69 0.65 0.60 0.50
"&
1
PbSO,
0
0 0
1
0 0
0
1
0
Bas06
0.49 0.24 0.22 0.19 0.16 0.21
due to these is small in comparison with capillary hindrance, and the barium sulfate is no doubt more free from surface cracks than the silica. I n the case of isopropyl alcohol the hydroxyl group is attached to the center carbon atom; such a molecule adsorbed with the hydroxyl group oriented towards the crystal surface covers more surface area than a primary alcohol. One might be led to suspect that there would be hindrance to the placing of such molecules, one on each lattice ion. The amount of hindrance would, of course, depend upon the magnitude of the lattice constant, d. No sharp decrease in the heat of wetting of isopropyl alcohol over those of the primary alcohols was observed, which seems to indicate that no such hindrance took place in the surfaces of those solids investigated. Approximate curves (figures 2 and 3) have been calculated and drawn for the energy of adsorption of water and of methyl, ethyl, propyl, and butyl alcohols on barium sulfate, using the formula: b E = -p Z- e?
r10
586
ALFRED CLARK AND B. D. THOMAS
R-
iffism
FIQ.2. Energy of adsorption of water on barium sulfate
R -hGslRaMs
FIQ.3. Energy of adsorption of alcohols on barium sulfate
b is determined in the usual manner by setting the derivative of the above exprwsion equal to 0 and giving t its equilibrium value.
587
ADSORPTION ON HETEROPOLAR SURFACES
It has been assumed that the energy of adsorption on the barium ion is the same regardless‘ of the alcohol, since they all have the same dipole moment. Energy of adsorption on the sulfate ion decreases as one ascends the series, because the dipole is farther removed from the ion; also, the displacements from the equilibrium position become greater. The energy of binding on the sulfate ion is so small that actually it is quite probable that orientation is much less pronounced there than on the barium ion and that the average life period of an adsorbed molecule is very short. These calculations are intended to be only approximate, but it is believed that they represent the general trend of conditions on heteropolar surfaces. TABLE 2 Comparison OJ ezperimental and theoretical values of heals of wetting i n calories ’ per gram SiOt LIQUID
Theoret-
1.00 0.69 0.65 0.60 0.50
1.00 0.61 0.55 0.52
ical
~
Water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methylalcohol.. . . . . . . . . . . . . . . . . . . . . . . . . . . . Ethyl alcohol... . . . . . . . . . . . . . . . . . . . . . . . . . . n-Propyl alcohol... . . . . . . . . . . . . . . . . . . . . . . . . n-Butyl alcohol.,. . . . . . . . . . . . . . . . . . . . . . . . . .
1
Exwrimental
Theoret-
---___-
Water . . . . . . . . . . . . . . . . . . . . . . . . . 1.00 Butyl alcohol... . . . . . . . . . . . . . . Amyl alcohol.. . . . . . . . . . . . . . . . . 0.50
~
0.50
ZnO
Mi30 LIQUID
Bas04
Ex&mental
ical
1.00 0.46
Experimental
Theoret ical
0.49 0.24 0.22 0.19 0.16
0.49 0.28 0.25 0.23 0.21
~
_
TiOr
Experimental
Theoretical
1.00 0.58
1.00 0.51
Exwrimentsl
j Theoret ical
%p-
With water as a standard, table 2 compares the experimental heats of wetting and those obtained by use of the formula:
This formula was found to be an excellent first approximation to the more complete formula
588
ALFRED CLARK AND B . D. THOMAS
derived on page 581. The sum of the two energies of adsorption (on the positive and negative ions) is taken as proportional to the total energy of adsorption. Table 3 gives similar results, using the data of Harkins and Dahlstrom (6) for the heats of wetting of titanium dioxide and zinc oxide by water and butyl alcohol, and the data of Meissner (14) for the heats of wetting of magnesium oxide by water and amyl alcohol, all on the basis of water = 1.00. SUMMARY
1. The heats of wetting of silica, calcium fluoride, lead sulfate, and barium sulfate by water and by methyl, ethyl, propyl, isopropyl, and butyl alcohols have been determined. 2. An explanation for the zero heats of wetting of calcium fluoride and lead sulfate has been offered. 3. A general theory for the adsorption of dipoles on heteropolar surfaces has been presented. 4. Simple theoretical calculations of the heats of wetting of barium sulfate and silica have been shown to be in fair agreement with experimentally determined values.
This work was carried out under the Research Associate Plan of Battelle Memorial Institute. Grateful acknowledgement is made to Dr. 0. E. Harder for helpful suggestions and criticisms, and to Mr. Clyde E. Williams, Director, for permission to publish this material. REFERENCES
(1) BARTELAND Fu: Colloid Symposium Monograph 7, 135 (1936). BARTELAND OSTERHOF:Colloid Symposium Monograph 6, 113 (1927). (2) DE BOERAND DIPPEL:Z. physik. Chem. B26,399 (1934). Rend. ist. lombardo sci. 8, 135 (1866). (3) CANTONI: (4) GAUDECHON: Compt. rend. 167, 209 (1913). (5) GROVESAND LUGDEN:J. Chem. SOC.1937, 158. (6) HARKINSAND DAHLSTROM: Ind. Eng. Chem. 22, 892 (1930). (7) HARKINSAND EWING: J. Am. Chem. SOC. 43, 1795 (1921). (8) ILLIINAND COWORKERS: Phil. Mag. 163, 294 (1937). (9) JUNGKE: Ann. Physik u. Chemie 126, 292 (1865). (10) KOEHLERAND MATTREWS: J. Am. Chem. SOC.46, 1158 (1924). (11) LAMBAND COOLIDGE: J. Am. Chem. SOC.42, 1146 (1920). (12) MARTINI:Atti ist. Veneto sci. 8, 102 (1896). Ann. Physik u. Chemie 146, 431 (1872). (13) MASCHKE: (14) MEISSNFIR: Ann. Physik u. Chemie 29, 114 (1886). AND GRIMM:J. Am. Chem. SOC.43, 2144 (1921). (15) PATRICK (16) POUILLET: Ann. chim. phys. 20, 141 (1822). (17) ROSE:Ann. Physik 73, 1 (1847). (18) TATE: Phil. Mag. 20, 508 (1860). (19) VON WEIMARN: J. Russ. Phys. Chem. SOC.40,125 (190R).
