THE UNIVERSITY OF MICHIGAN INDUSTRY PROGRAM OF THE COLLEGE OF ENGINEERING THREE PAPERS ON THE FREE ENERGY IMMERSION OF COMPRESSED POWDERS F. E. Bartell Ph.D., Professor Emeritus of Chemistry R. G. Craig Ph.D., Associate Research Engineer J. J. Van Voorhis* Ph.D., University of Michigan March, 1957 IP-210 * Now with Shell Oil Company, Wood River, Illinois.

ei~2mhc Vui\ vs The first paper appears in the Journal of Physical Chemistry, Volume 60. The other two papers are scheduled for publication in the same journal. Some of the data in the second paper was taken from a thesis by John J. Van Voorhis, submitted to the School of Graduate Studies of the University of Michigan in partial fulfillment of the requirements for the degree of Doctor of Philosophy. ii

TABLE OF CONTENTS Paper Number Page I FREE ENERGY OF IMMERSION OF COMPRESSED POWDERS WITH DIFFERENT LIQUIDS Graphite Powders............... 1 II FREE ENERGY OF IMMERSION OF COMPRESSED POWDERS WITH DIFFERENT LIQUIDS Silica Powder................ 16 III FREE ENERGY OF IMMERSION OF LOOSE SILICA AND GRAPHITE POWDERS WITH ORGANIC LIQUIDS Comparison of Values from "Non-Porous" and "Porous Solid" Treatment of Adsorption Data.. 45 iii

LIST OF FIGURES Figure No. Page Graphite Powders 1 Adsorption-desorption isotherms of cyclohexane on graphite B powder and graphite B-4 plugs at 26~...................... 7 2 Adsorption-desorption isotherms of toluene on graphite A-4 and B-4 plugs at 26~......... 8 3 Adsorption-desorption isotherms of carbon tetrachloride, n-propyl alcohol and n-heptane on graphite B-4 plugs at 26'.......9..... 9 Silica Powder 1 Adsorption-desorption isotherms of benzene on Linde silica plugs-5 and Linde silica powder at 26~... 39 2 Adsorption-desorption isotherms of cyclohexane on silica gel and Linde silica plugs-5 at 26~.... 40 3 Adsorption-desorption isotherms of toluene on Linde silica plugs-5 and -10 at 26~........ 41 4 Adsorption-desorption isotherms of n-hexane on Linde silica plugs-5 and -10 at 26~........ 42 5 Adsorption-desorption isotherms of water and ethanol on Linde silica plugs-5 at 26~...... 43 6 Adsorption-desorption isotherms of chloroform and n-heptane on Linde silica plugs-5 at 260... 44 Comparison of Values from "Non-Porous and "Porous Solid" Treatment of Adsorption Data 1 Adsorption-desorption isotherms of benzene on silica powder and 1-propanol on graphite A powder at 26~...................... 53 iv

LIST OF TABLES Table No. Page Graphite Powder I Free Energy of Immersion (7 - SL) of Graphites with Various Liquids (ergs/cm.2) at 260...................... 11 II Physical Properties of Graphite Plugs...... 12 III Characteristics of Adsorbed Films on Graphite Plugs...................... 1 Silica Powder I Physical Properties of Silica Gel and of Linde Silica Powder and Plugs............. 34 II Free Energy of Immersion (YSO - YSL) of Linde Silica with Various Liquids at 26~ and Comparison with Previous Data for Siliceous Materials.... 35 III Comparison of Experimental and Calculated Initial Spreading Coefficients for Linde Silica Plugs-5 and -10 at 26~................. 36 IV Temperature Coefficients of (7S0- 7 SL) on Linde Silica................... 37 V Characteristics of Adsorbed Films on Linde Silica Plugs-10 at 26~ 38 Comparison of Values from "Non-Porous" and "Porous Solid" Treatment of Adsorption Data I Comparison of Free Energy of Immersion from "Non-Porous Solid" and "Porous Solid" Treatments of Adsorption Data for Loose Powders and Porous Plugs at 26~................... 52 V

I. FREE ENERGY OF IMMERSION OF COMPRESSED POWDERS WITH DIFFERENT LIQUIDS Graphite Powders A method is described for the determination of the energy of wetting, the free energy of immersion, of finely divided solid materialrby a liquid. The method consists of obtaining complete vapor adsorption isotherms for a liquid adsorbate with an adsorbent comprised of a finely divided solid material which has been highly compressed into the form of plugs. These plugs are porous and possess capillary systems which give adsorption effects similar to those obtained with solid porous gels. The free energy of immersion can therefore be calculated from the adsorption data by application of the Gibbs adsorption equation as used by Dobay, Fu and Bartell in their study of silica gel.1 The free energies of immersion of each of a series of different graphite powders with the liquids toluene, carbon tetrachloride, n-heptane, cyclohexane and n-propyl alcohol were determined. The values obtained show that the free energy of immersion (expressed in free energy per unit area) is independent of surface, of the particle size of the powdered material and of pressures used in compressing the plugs; it is dependent only on the nature of the surface of the powdered material. The surface areas, average pore volumes, average dry pore radii and porosities were determined for each of the graphite plugs used. The molecular cross sectional areas of the adsorbed molecules and the heats of adsorption of the first layer were determined for the various systems from the B.E.T. plots of the adsorption data. Standard free energies of adsorption were calculated in order to evaluate the relative strengths of the adsorptive bonds. -1 -

INTRODUCTION It is generally recognized that knowledge of the magnitude of free surface energy changes which occur when solids are wetted by liquids is of great importance in the field of surface chemistry. Unfortunately owing to uncertainties or difficulties attendant to known methods such data have seldom been made available. Since publication of the work of Bangham and Razouk showing that the Gibbs adsorption equation is applicable to the determination of free surface energy changes which occur on solids during adsorption of vapors upon them, investigators have made use of this method for the determination of free surface energy changes with such systems. For non-porous solids, Boyd and Livingston3 and Harkins and co-workers-'5 used a formulation for the free surface energy lowering at relative pressures from zero to one, as SO/SL = (SO - 7SL) RT P/Po 1 /m In P/Po + 7L (1) where bSO/SL is the free surface energy lowering which occurs when unit area of solid-vacuum interface is replaced by unit area of solid-liquid interface; 7 represents any free surface energy, surface tension or interfacial tension value; subscripts SO, SL and LV~ represent a solid -vacuum, a solid-liquid and a liquid-saturated vapor interface, respectively; M is the molecular weight of the adsorbate; a is the surface area of the solid; x/m is the weight adsorbed per weight of sample and p/po is the relative pressure. The expression (7So - SL) will be referred to in this paper as representing the free energy of immersion of the solid in the liquid. By making appropriate changes in the method and the theory, the above described treatment was extended by Dobay, Fu and Bartell, and by Fu and Bartell6 to include porous solids. The main difference in behavior between porous and non-porous solids is that during adsorption of a vapor on a porous solid, the solid-vacuum interface is replaced by a solid-liquid interface, and at p/pO = 1 because of the filling of the pores the liquid-vapor interface is completely destroyed; while in the case of adsorption of a vapor on a nonporous solid the area of the liquid vapor interface at p/po = 1 is not destroyed and the area is considered to be essentially equal to that of the solid-vacuum interface. The following formulation was developed for the evaluation of the free surface energy lowering at p/po = 1 on porous solids: I/0

fRT S P/Po 1 / ax/dlnp/p (2) During the adsorption of an adsorbate on a porous solid the surface area available for adsorption decreases continuously to a zero value at p/po = 1, and therefore in order to evaluate the above integral it would seem that a knowledge of the film thickness at various p/po values must'be known. This is true if the free surface energy change is desired from p/Po = 0 to p/po = b, where 0 < b < 1. Fu and Bartell have shown, however, that if the free surface energy change from p/po = 0 to p/Po = 1 only is desired, a hypothetical process can be substituted for the real process where the initial and final states of each are the same. This method makes more certain the evaluation of the integral and the subsequent calculation of the free energy of immersion (7SO - 7SL)' the work of adhesion W = (7S - SL + 7LVo) and the work of spreading or the initial spreading coefficient Ws = (YSO - 7SL.- 7LVO) In the determination of the free energy of immersion from the complete adsorption isothermofanon-porous powder several factors leading to difficulties are involved: 1) it is difficult to determine equilibrium values near p/Po = 1, 2) it is not possible to extrapolate the adsorption isotherm to a definite limiting value at p/po = 1, and 3) the "non-porous" powder functions partially as a porous solid during the process of adsorption because in a free flowing nonporous solid the particles are sufficiently close together so that large numbers of capillary spaces actually are formed. Of the three factors mentioned the last usually constitutes the greatest source of error when high area solids are used. If, however, porous solids are used these difficulties are avoided. The adsorption isotherms of organic vapors on Linde silica powders were shown by Carman and Raal7 to give type II isotherms, while if the powders were held under compression, type IV isotherms, characteristic of a silica gel, were obtained. This indicated to us that if powdered non-porous solids could be highly compressed into plugs having capillary type systems they should function as rigid porous solids, thus making possible the calculation of the free energy of immersion, the work of adhesion and the work of spreading from the complete adsorption isotherm of an adsorbate on these porous plugs. This method would avoid the difficulties which are encountered when the free flowing, -3 -

non-porous powders as such are used, but still would permit the determination of the free surface energy changes which actually occur on these finely divide solids. The present investigation consisted of taking a series of finely divided graphite powders, compressing them into rigid porous plugs, determining the complete adsorption isotherms with a series of adsorbates, and calculating the free surface energy changes which occurred during adsorption.1 EXPERIMENTAL A. Materials. Graphites. In the following description of the graphite powders a "purified graphite" will indicate that the powder was treated with hot 1-1 HC1, filtered thoroughly washed, dried, powdered, heated in an evacuated quartz tube at lOOC until a high vacuum was obtained, cooled and stored in a helium atmosphere. All graphites were furnished by the Acheson Colloids Company and have the following designations and surface areas.* Area Graphite Acheson Code m.2/g. Treatment A EC-753-96A 386 Purified B EC-753-96A 382 Unpurified B-T EC-753-96A 383 Unpurified. Evacuated and heated at 1000~ in a quartz tube until high vacuum obtained. C EC-753-72A 323 Purified D EC-603A 247 Unpurified E EC-753-23A 136 Purified F EC-753-144A 472 Purified B. Adsorbates. Toluene, reagent grade, was shaken with successive portions of concen trated sulfuric acid, washed with distilled water, washed with dilute sodium * In the determination of the B.E.T. surface areas the nitrogen area was assumed to be 16.2 X.2

carbonate solution and again with distilled water, dried over calcium chloride and activated silica gel, refluxed over mercury, and run through an alumina column. The liquid was fractionated using a thirty-six stage Snyder column, the middle fraction collected and finally dried using freshly extruded sodium ribbon. Carbon tetrachloride, reagent grade, was purified in the same manner as was toluene except that only alumina and silica gel were used for drying. Cyclohexane, reagent grade, was purified by the same procedure used for toluene except that refluxing over mercury was omitted8 n-Propyl alcohol, reagent grade, was dried by the method of Lund and Bjerrum and the liquid was fractionated directly from this mixture. After purification, all adsorbates were distilled into break-off vials, degassed, frozen, evacuated, sealed off and stored until used. C. Apparatus. The volumetric adsorption apparatus and method used to determine the surface areas of the solids were similar to those described by Emmett.9 The gravimetric adsorption isotherms were determined by using a multiple (six unit) modified McBain-Bakr1O quartz spring balance. Temperature control of the adsorption tubes was accomplished by using a water thermostated bath, the temperatures of which could be regulated to + 0.03~. Each adsorption tube was connected through a mercury stock valve to a manifold; the adsorbate tube and a mercury manometer were also connected to this manifold. When the mercury was raised in the manometer the adsorption system was completely isolated and contamination of the adsorbate by stopcock grease was eliminated. The vapor pressure of the adsorbate was controlled by the temperature of an adsorbate thermostated bath. With the multiple (six unit) adsorption balance the vapor adsorption isotherms of a single adsorbate on six different adsorbents were determined simultaneously. The techniques used to obtain these isotherms were similar to those described by Bartell and Dobay.ll Probably the most serious experimental difficulty was condensation of mercury on the quartz springs at very low pressures. Corrections were made for this by determining the zero readings of the quartz springs just prior to starting the adsorption run; after the vapor of the adsorbate had been admitted to the adsorption system this condensation ceased. The apparatus used to prepare the compressed porous plugs consisted of a steel cylinder, 1.25" thick and 2.5" in diameter with a 3/8" hole drilled through its axis. A 3/8" steel plunger, machined to give a sliding fit, and a polished steel base plate, which could be fastened to the base of the cylinder by three Allen screws, completed the apparatus. The porous plugs were prepared by placing' the graphite powder into the hole in the cylinder and then applying pressure to the plunger by an hydraulic press. Usually the pressure was applied for a period of 10 minutes. The pressure used in preparing a porous plug is given in thousands of pounds upon the ram and is shown by the -5 -

number following the graphite designation. For example, graphite A-4 indicate that graphite A powder was subjected to a pressure of 4000 psi on the ram during the plug formation. The actual pressure on the 3/8" plug was 36,216 pt (ratio, pressure on ram to pressure on plug = 1 to 9.054). RESULTS AND DISCUSSION The difference in shape of the adsorption isotherms of cyclohexane with a graphite powder and with a porous plug prepared from this powder is shown in Figure 1. The weights adsorbed in the initial portions of the isotherms are nearly identical, but at higher p/Po values, because of the greate: exposed surface area of the graphite B powder, the weight adsorbed by it becomes greater than that adsorbed by the graphite B-4 plug. With the plug, as p/pO - 1, the capillary spaces in the porous plug became filled with liqui( thus the weight adsorbed reached a limiting value. In contrast, the weight adsorbed by the graphite B powder increased rapidly as p/po - 1 and appeared not to reach a limiting value. Hysteresis was observed on desorption for both graphite B powder ana the graphite B-4 plug. The fact that graphite B powder showed hysteresis on desorption is, we believe, due to the capillary spaces formed by particle approach or contact and to irregularities in the solid particles. The hysteresis loop, as should be expected, was much larger and extended over a wider p/pO range for the graphite B-4 plug, than for graphite B powder. Figure 2 shows the difference in shape of the adsorption isotherms of toluene on a purified and on an unpurified graphite plug. This difference in shape is presumably due to the difference in the average pore radius and to the pore radii distribution. Adsorption of cyclohexane on unpurified heat treated graphite B-4T plug gives an isotherm typical and nearly identical to that with a purified graphite A-4 plug, while the isotherm of cyclohexane on graphite B-4 plug is typical of one for an unpurified graphite plug. Graphite B-T was prepared by heating graphite B in a vacuum until a high vacuum was obtained, which procedure removed surface oxides and adsorbed gases. Thus, the surface condition of the graphite powders must be one of the main factors in controlling the average pore radius and the pore radii distribution. Adsorption isotherms of carbon tetrachloride, n-propyl alcohol and n-heptane on graphite B-4 are shown in Figure 3. These isotherms of organic adsorbates on porous plugs are isotherms typical of porous solids. They show hysteresis and permit extrapolation to a limiting value for the amount adsorbs at saturation. After complete adsorption and desorption curves had been obtained, the same porous plugs were again used and second run adsorption and desorption points determined for all isotherms. Second run adsorption points agreed -6 -

0' E Iw m 0 Icw h. 1: I -q!.I 1.0 P PO0 Adsorpt i on-de s orpt i on graphite B powder and Fig. 1. — isotherms of cyclohexane on graphite B-4 plugs at 26~.

800 700 Graphite A-4 Plug Graphite B-4 Plug o Adsorption A Desorption 600 --- I/9 I I,a w Q (0 LU 400 300 200 > lip IY _ — Bw 0000. r40 ~j.Z cp-P~ 100 on 0.1.2.3.4.5.6.7.8.9 1.0 p Po Fig. 2. — Adsorption-desorption isotherms of toluene on graphite A-4 and B-4 plugs at 26~.

800 - 700 T Carbon Tetrachloride - A n-Propyl Alcohol - - n- Heptane | 600 o Adsorption /.Desorption. E 500 _ - o -V) -. 0 400 _Ig. 3,.I o 6~ "...,2"6~. I -- 0 ~ 300 PO IH O~~~~~~~ 0 2001 ~rr r 26' 100 0.1.2.3.4.5.6.7.8.9 Po. Fig. 3. - Adsorption-desorption isotherms of carbon tetrachloride, n-propyl alcohol and n-heptane on graphite B-4 plugs at 1.0

with the first run adsorption points up to p/po-. 0.1. From p/po- 0.1 to 0.5, second run adsorption points were 0-2 mg/g. higher than the first run adsorption points and from p/po %.0.5 to 1 were 2-5 mg/g.higher, with the exception of n-heptane on unpurified graphites, where at P/Po s-N 0.9 the second run adsorption points were approximately 10 mg/g. higher. Second run desorptior points always agreed with first run desorption points. This indicates that little swelling of the porous plugs occurred during adsorption and that the porous plugs could be treated as rigid solids permitting the calculation of the free energies of immersion from the complete adsorption isotherms of the various vapors on these plugs. Also, since the second run adsorption points agree with the first in the low pressure region and very nearly agree up to P/PO = 0.5, any slight swelling of the plugs would not materially change the calculated free energy of immersion values because the low pressure points are of the most importance in the energy change calculations. Thus, the integral in equation 2 can be evaluated without difficulty. The values obtained for a variety of graphites with a series of adsorbates are listed in Table I. Inspection of the table shows that within experimental error the free energy of immersion is independent of the pore size, the pore size distribution and the surface area of the solid. Also, the non-polar adsorbates, toluene, carbon tetrachloride, n-heptane and cyclohexane, show no significant difference between the purified and unpurified graphites. The probable reason for the free energy of immersion beings_ 10 ergs/cm. highe for n-propyl alcohol on the purified graphites is that the unpurified graphiteE (B-4 and D-4) still have some surface oxides present when adsorption starts. Since the n-propyl alcohol molecules tend to orient themselves with the nonpolar end toward the clean graphite surface and the polar end toward the oxide surface, the disorder and following interaction of the molecules tends to decrease the free surface energy change during adsorption on the unpurified graphite surface. This is substantiated by the calculated apparent molecular cross sectional area of n-propyl alcohol on the unpurified graphite surface, 43 A2, compared with 37 A? on the purified graphite surface. The apparent cross sectional areas of the remaining liquid adsorbates were substantially the same on both purified and unpurified graphites. Comparison of free energy of immersion values on graphites with literature values is somewhat futile since there are very few such data reported in the literature. The values that are available were determined on nonporous powdered solids and, therefore, are not strictly comparable. Heats of immersion and heats of wetting values of comparable liquids on graphites should, however, be in the same order as the free energy of immersion values. The data of Harkins and Boyd,12 of Bartell and Suggitt, 3 and of Healey, et al.,14 show this to be the case. One of the most difficult problems in comparing the free surface energy change for graphites is the lack of knowledge of the previous history and treatment of the graphites prior to adsorption. -10 -

TABLE I FRE~E ENERGY OF IMMERSION (TSO ~ TSL) OF GRAPHITES WITH VARIOUS IQUIDS (ergs/cm,) AT 262 FREE ENERGY OF IMMERSION (ySO - YSL ) OF GRAPHITES WITH VARIOUS LIQUIDS (ergs/cm.) AT 26~ SO S Carbon tetraGraphite Toluene chloride. n-Heptane Cyclohexane n-Propyl alcohol H 1 - A-2 A-4 A-6 A-10 B-4* B-4T C-4 D-4* E-4 E-6 E-10 F-4 69.2 + 0.7 71.0 71.6 70.8 70.4 72.4 60.1 + 0.5 59.8 59.2 59.7 61.1 60.9 54.5 + 1.2 50.1 55.0 52.4 55.8 + 0.6 54.1 55.7 68.9 + 0.5 57.7 + 0.9 68.3 59.6 54.0 52.3 Av. 70.9 ~ 0.3 60.1 + 0.2 53.1 + 0.5 55.2 + 0.4 68.6 + 0.2 58.6 + 0.6* * Unpurified graphite.

The average pore volumes of the purified graphite plugs A, C, E and F, listed in Table II, are in the same order as their surface areas. This is also true of the unpurified graphite plugs if they are considered separately. Porous plugs prepared from the high area purified graphites A, B-T, C and F have approximately the same average dry radius (calculated from twice the volume to surface ratio), while the low area graphite E plugs have, as expected, a larger average dry pore radius. The average dry radius of high area unpurified graphites B and D plugs are low, the average dry radius of the lowe area graphite D plug being less than that of the graphite B plug. TABLE II PHYSICAL PROPERTIES OF GRAPHITE PLUGS Surface Av. dry area, Pore vol., pore Graphite m2/g. cc/g. radius (A). Porosity A-2 381 0.66 34 A-4 383.66 34 0.70 A-6 380.59 31 A-10 380.61 32 B-4* 371.46 25.60 B-4T 362.68 38 C-4 316.58 36.70 D-4* 217.33 31.53 E-4 129.33 51.53 E-6 127.32 50 E-10 127.31 49 F-4 471.84 36.72 * Unpurified graphite. The percent porosity was calculated from the relation porosity = (p - Pa)/P (3) re p is the density of graphite and Pa the apparent density of the porous.,,g. Graphite E-4 and D-4 plugs have the same porosities and the same averag pore volumes but graphite E-4 has a.much higher average dry radius due to its lower surface area. Unpurified graphites B and D have good packing characteristics as is indicated by the relatively low porosities, average pore volumes and average dry pore radii of the-unpurified graphite B-4 and D-4 plugs. The purified graphite plugs A-4, C-4 and F-4, with areas between 316 and 471 m2/gj -12 -

have nearly the same porosities and average dry pore radii but have average pore volumes in the order of their areas. The pore radius range, calculated from the desorption isotherms, by use of the Kelvin equation, is from 15 to 1500 A. The saturation values of the adsorption isotherms of water on the porous plugs indicated that the plugs consisted of approximately 80 percent micropores and 20 percent macropores. The gravimetric adsorption data of the organic adsorbates on graphites A-4 and B-4 plugs were plotted according to the simple, linear B.E.T. equation and straight line plots were obtained in all cases in the region from P/Po 0.05 to 0.2. Values for the volumes adsorbed at monolayer coverage, Vm, obtained from the B.E.T. plots are listed in Table III. If the B.E.T. nitrogen areas are assumed to be valid, the apparent molecular cross-sectional areas of the adsorbed molecules can be calculated. The value of 37 A? listed is for n-propyl alcohol on graphite A-4, while a value of 43 A2 was obtained on graphite B-4. As had previously been observedl5,16 the cross-sectional areas obtained from the B.E.T. plots are approximately 30 percent greater than those obtained by assuming that the adsorbed molecules correspond to the plane of closest packing in the liquified gas. TABLE III CHARACTERISTICS OF ADSORBED FILMS ON GRAPHITE PLUGS Apparent mol. cross E1 sectional (B.E.T.) -AF~, m area kcal/ kcal/ Adsorbate cc,/g. 260 (A-) mole mole Toluene 29.7 47 15.4 5.5 Carbon tetrachloride 29.7 47 11.8 4.2 n-Heptane 23.1 62 15.1 5.0 Cyclohexane 26.7 52 12.5 4.6 n-Propyl alcohol* 38.9 37 15.0 4.0 * With unpurified graphite n-propyl alcohol gave a molecular crosssectional area of 43 A2, The heat of adsorption of the first layer, E1, was also calculated fr from the B.E.T. plots and the values listed in Table III show reasonable agreeme ment with experimentally determined heats of adsorption and with the generaliza zation that E1 values of various adsorbates on a single adsorbent are usually ir in the same order as the boiling points of the liquids. -13 -

Plots of the equilibrium function versus surface coverage were made for each system according to the method suggested by Graham.17 For each system the initial decrease in the equilibrium function at the lowest measured surface coverage indicated that the surface of the graphite was heterogeneous with several kinds of adsorption sites. Values for the standard free energies of adsorption calculated from the extrapolated values of equilibrium functions are listed in Table III. Although these values are only relative values they indicate the relative strength of the adsorption bond. The order of the -AF~ values can be explained in the following manner. Toluene has the highest valu since the ring is planar and the methyl group is free to rotate so that the molecule can lie flat on the surface with seven hydrogens contributing to the binding. Similarity of the molecule with the structure of the graphite could also have some effect. n-Heptane is next in order and the high -AF~ value together with its high heat of adsorption has been interpretedlO to mean that the molecule lies flat on the surface with each -CH2- contributing to the binding. Cyclohexane is next, which has been explainedl9 by the distortion of the molecule at the surface so that six hydrogens are in contact with the surface rather than three. Carbon tetrachloride is next, which can be explained since only three chlorine atoms per molecule can be in contact with the surface. n-Propyl alcohol is last which is presumably due to the effect of a polar molecule on a non-polar surface. The order of the -AF~ values is not the same as that for the (7SO - ZSL) values. This is not so surprising, however, since the former is a measure of the strength of the adsorption bond at low surface coverages, while the latter is the free surface energy change for all layers up to saturation of the solid.

LIST OF REFERENCES 1. D. G. Dobay, Y. Fu, and F. E. Bartell, J. Am. Chem. Soc., 73, 308 (1951). 2. D. H. Bangham, Trans. Faraday Soc., 33, 805 (1937); also D. H. Bangham and R. I. Razouk, ibid., 33, 1459 (1937); also Proc. Roy. Soc. (London), A166, 572 (1938). 3. G. E. Boyd, and H. K. Livingston, J. Am. Chem. Soc., 64, 2383 (1942), 4. G. Jura, and W. D. Harkins, ibid., 66, 1356 (1944). 5 E. H. Loeser, W. D. Harkins, and S. B. Twiss, THIS JOURNAL, 57, 251, 591 (1953). 6. Y. Fu, and F. E. Bartell, ibid., 55, 662 (1951). 7. P. C. Carman, and F. A. Raal, Nature, 67, 112 (1951). 8. H. Lurid, and J. Bjerrum, Ber., 64A, 210 (1931). 9. P. H. Emmett, "Advances in Colloid Science," Vol. I, Interscience Publishers, Inc., New York, N. Y., 1942, pp. 1-36. 10. J. W. McBain, and A. M. Bakr, J. Am. Chem. Soc., 48, 690 (1926). 11. F. E. Bartell, and D. G. Dobay, ibid., 72, 4388 (1950). 12. W. D. Harkins, and G. E. Boyd, ibid., 64, 1195 (1942). 13. F. E. Bartell, and R. M. Suggitt, THIS JOURNAL, 58, 36 (1954). 14. F. H. Healey, J. J. Chessick, A. C. Zettlemoyer, and G. J. Young, ibid., 58, 887 (1954). 15. H. K. Livingston, J. Colloid Sci., 4, 447 (1949). 16. J. E. Bower, Thesis, University of Michigan, 1951. 17. D. Graham, THIS JOURNAL, 7, 665 (1953). 18. R. A. Beebe, C. L. Kington, M. H. Polley, and W. R. Smith, J. Am. Chem. Soc., 72, 40 (1950). 19. R. N. Smith, C. Pierce and H. Cordes, ibid., 72, 5595 (1950). -15 -

II. FREE ENERGY OF IMMERSION OF COMPRESSED POWDERS WITH DIFFERENT LIQUIDS Silica Powder Free energy of immersion values of finely divided Linde silica powder with different liquid adsorbates were determined by obtaining complete gravimetric vapor adsorption isotherms with these adsorbates on self-supporting porous plugs prepared by compressing the loose silica powder, and subsequent application of the Gibbs adsorption equation, revised for porous solids. These plugs, formed at a given pressure, possessed a constant surface area, functioned as rigid adsorbents, and the pores formed filled up completely with liquid during adsorption, thus allowing the "porous solid" treatment. Two types of Linde silica plugs were used and they possessed noticeably different apparent densities, porosities, pore volumes, "average" pore radii and specific surface areas. The free energy of immersion of the Linde silica powder with water, methanol, ethanol, 1-propanol, benzene, toluene, chloroform, carbon tetrachloride, n-heptane, n-hexane and cyclohexane was calculated. The values of the free surface energy of immersion at 26~ for Linde silica ranged from 139.5 ergs/cm.2 with water to 27.7 ergs/cm.2 with cyclohexane. Reproducible and self-consistent values of the free energy of immersion were obtained for both types of Linde silica plugs. These values were in good agreement with similar data reported in the literature for silica gels. The experimental initial spreading coefficients compared well with the calculated values and for several adsorbates the temperature coefficient of the free energy of immersion was calculated. In addition, characteristics ofr the adsorbed films such as monolayer capacity, film thickness, cross-sectional areas of adsorbed molecules, and heats of adsorption were reported. -16 -

INTRODUCTION Since the publication of the adsorption work of Bangham and Razouk, the Gibbs adsorption equation has been used in two main (3) D. H. Bangham, Trans. Faraday Soc., 33, 805 (1937); D. H. Bangham and R. I. Razouk, ibid., 3355, 1459 (1937); Proc. Roy. Soc. (London), A 166, 572 (1938). variations to obtain free surface energy changes during adsorption. For non-porous solids, Boyd and Livingston4 and Harkins and co-workers5,6 (4) G. E. Boyd and H. K. Livingston, J. Am. Chem. Soc., 64, 2383 (1942). (5) E. H. Loeser, W. D. Harkins, and S. B. Twiss, THIS JOURNAL, 57, 251, 591 (1953). (6) G. Jura and W. D. Harkins, J: Am. Chem. Soc., 66, 1356 (1944). used a formulation for the free surface energy lowering as follows: P/Po = 1 RT x/m din P/Po + tLV~ -SO/SL = SO - SL Mas J n ) P/Po = 0 where -(SO/SL represents the free surface energy lowering when unit surface of solid-vacuum is replaced by unit surface of solid-liquid, y represents any free surface energy, surface tension or interfacial tension, the subscripts SO, SL, and LV~ represent solid-vacuum, solid-liquid, and liquid-saturated vapor interfaces, respectively, M is the adsorbate molecular weight, as is the specific surface area, x/m is the weight in mg. adsorbed per gram of adsorbent and p/Po is the relative pressure of the adsorbate. Bartell and co-workers7'8'9 made appropriate changes in the (7) Y. Fu and F. E. Bartell, THIS JOURNAL, 55, 662 (1951). (8) D. G. Dobay, Y. Fu and F. E. Bartell, J. Am. Chem. Soc., 73, 308 (1951). (9) F. E. Bartell and J. E. Bower, J. Colloid Sci., 7, 80 (1952). -17 -

method and theory and developed the following formulation for determining free surface energy changes on porous solids. P/PO = 1 RT x/m din p/p0 -SO/SL = 7S0 - SL = MaS J P/Po = 0 There was question as to whether equation 1 could justifiably be used for the determination of the free energy of immersion, 7S0 - 7SL (from P/po = 0 to P/Po = 1) of powdered non-porous solids. With such an adsorbent a limiting value of the amount adsorbed is not reached and corrections must be made for the decrease in surface area during adsorption. 10 The authors have shown that free energies of immersion may be determined (10) R. G. Craig, J. J. Van Voorhis and F. E. Bartell, THIS JOURNAL, 60, 1225 (1956). on graphite powders (generally classed as hydrophobic solids) by compressing these powders into porous plugs, which in so far as porosity is concerned are comparable to solid porous gels, thus permitting the application of equation 2. Concurrently with the research on graphite powders a typical hydrophilic solid, Linde silica powder, was used in the form of compressed plugs to evaluate free surface energy changes of eleven different liquids during adsorption. These compressed powders fulfilled the essential criteria that the pores fill up completely during adsorption, that the plugs formed at a given pressure possess a constant surface area and that the plugs function as rigid adsorbents during adsorption. -18 -

EXPERIMENTAL Materials: Silica, Linde silica powder, No. 2669-95B, was a very finely-divided, amorphous, non-porous material having a bulk density of 0.035 g./ml. and a B.E.T. nitrogen surface area of 312 m.2/g. This powder left no residue when treated with hydrofluoric acid and was used without any purification being attempted. The silica gel employed was identical with that recently used by Bartell and co-workers9 l1,. It was prepared by the hydrolysis of (11) F. E. Bartell and D. J. Donahue, THIS JOURNAL, 56, 665 (1952). (12) F. E. Bartell and Y. Fu, THIS JOURNAL, 33, 676 71929). pure silicon tetrachloride and the bulk density of this silica gel was 0.40 g./ml. and the B.E.T. nitrogen surface area was 471 m.2/g. Adsorbates: The adsorbates used were carefully purified by methods similar to those described in other work from this laboratory9,10 Special precautions were taken to remove all traces of water from the organic liquids. After purification, all liquids were placed in glass break-off vials, de-gassed, frozen, evacuated, sealed off and stored until used; chloroform samples were stored in the dark. Gases: Helium, medical grade 100 percent, was purified by passing it slowly through an activated charcoal trap immersed in liquid nitrogen. Nitrogen, 99.8 percent purity, was further purified by passing it successively through an electrically heated, four foot column of Fuller's earth impregnated with finely-divided copper metal-3; a train of drying (13) F. R. Meyer and G. Ronge, Angew. Chem., 52, 637 (1939). tubes containing silica gel, Dehydrite and Drierite; and a liquid oxygen trap. Sample Preparation: The apparatus and method used for making compressed plugs of the silica powder were similar to that previously described10. In this research the diameter of the compression cylinder was 5/8". Two pressures were chosen for compacting the Linde silica powder into plugs by application of controlled pressures between the platens of a Carver laboratory hydraulic press. A pressure of 5,000 and 10,000 p.s.i. registered on the hydraulic pressure gauge represents 12,800 and 25,600 p.s.i. on the plugs, respectively. For sake of convenience, the plugs compressed at an indicated pressure of 5,000 or 10,000 p.s.i. are designated as L.S.P.-5 or L.S.P.-10, respectively. Particle Size by Electron Microscopy: The general procedure employed in obtaining electron micrographs of the silica powder was as follows. Polystyrene latex spheres (PSL) having a diameter of 2580 + 25A. were placed on collodion films supported by copper screens of fine mesh, by evaporation of a suspension of the latex in water. The silica -19 -

powder was then carefully dusted upon this collodion film, after which the sample specimens were shadow-cast at an angle of approximately 30~ with palladium metal. Six representative exposures, each one of a different sample specimen were photographed on lantern slide contrast plates in an R.C.A. electron microscope, Type EMB. The photographic plates were reversed by contact printing on another contrast plate from which enlargements were made for study. Pore Volumes of Silica Plugs: The mercury displacement method described by Brunauer14 was used to determine apparent volume of the (14) S. Brunauer, The Adsorption of Gases and Vapors Vol. I, Phsical Adsorption, Princeton University Press, Princeton, N. J., (1945) silica plugs. Since the apparent volume is equal to the sum of the volume of the solid silica framework plus the volume of the pores, this latter volume can be calculated by assuming a density for the Linde silica. Adsorption Apparatus and Techniques: The experimental methods and techniques used to obtain the volumetric and gravimetric adsorption isotherms are described in recent work from this laboratory9,10l 1516. (15) F. E. Bartell and C. G. Dodd, THIS JOURNAL, 54, 114 (1950). (16) F. E. Bartell and D. G. Dobay, J. Am. Chem. Soc., 72, 4388 (1950). RESULTS AND DISCUSSION Physical Properties of the Adsorbents: Utilizing the PSL diameter as the measuring unit in the electron micrographs, it was found that the size of the primary silica particle varied from approximately 50 0 to 125 A. The majority of the "apparent particles", however, were secondary aggregates having considerably larger diameters than the primary particles. The average particle size diameter listed in Table I is the average of three different types of statistical diameters: (1) the mean surface diameter, (2) the mean volume diameter, and (3) the diameter defined by the ratio of the summation of the product of the number of particles and the diameter cubed to the summation of the product of the number of particles and the diameter squared. The external surface area may be calculated using the expression: -20 -

SG = 6/d x 1/P (3) where SG is the geometrical area, 6/d represents the total external area in cm.2/cm.3 of the powder for either cubical or spherical particles of diameter d, and P is the true density of the solid. It was found that the particle size and surface area were 100 A. and 270 m,./g., respectively. If the B.E.T. area of 312 m.2/g. is assumed to be the absolute area of the solid, the roughness factor is about 1.1 for the silica powder. It can be concluded that the individual silica particles have the general shape of a sphere and that they must be considered as non-porous solids, The physical properties of L.S.P.-5, L.S.P.-10 and the silica gel are also listed in Table I. The average pore volumes from the saturation values of adsorption isotherms of L.S.P.-5, L.S.P.-10 and silica gel were 0.853, 0.664, and 0.950 ml./g. respectively, and are all in good agreement with the values of the pore volumes determined by the mercury displacement method. The porosity of the three "porous solids" are quite similar, that of L.S.P.-5 comparing more closely with the silica gel. It is interesting to note that the diminution of surface area resulting from compacting the Linde silica powder amounted to only about 1 percent for L.S.P.-5 and 2 percent for L.S.P.-10. There is only a small difference in the B.E.T. surface areas of L.S.P.-5 and of L.S.P.-10 (about 1 percent), but there are considerable differences in apparent densities, pore volumes, and the porosities of these plugs (about 14 percent, 28 percent, and 10 percent, respectively). The B.E.T. surface areas also agree well with those calculated from energy changes during adsorption by the Fu and Bartell method7. -21 -

The Kelvin radius is often referred to as the wet radius of the pore and is considered to be the radius of the void space after one or more layers of molecules have been adsorbed on the pore walls. The true pore radius or dry pore radius should be larger than the Kelvin radius by 0 the thickness of the adsorbed film. The average dry pore radius of 40.4A. 0 and the average Kelvin radius of 39.4A. for the silica gel are typical de0 termined values. The average dry radius of 55.2A. for L.S.P.-5 and that o of 43.4A. for L.S.P.-10 are not in good agreement with the Kelvin radii of o o 17 149A. for L.S.P.-5 and 118A. for L.S.P.-10. Carman has determined the (17)P. C. Carman, Proc. Roy. Soc. (London), A209, 59 1951). pore-size distribution curve for silica plugs having a pore volume of 0.745 ml./g. and a porosity of 0.622 and found the average radius to be 0 110-120A., which is in reasonable agreement with those determined for the Linde silica plugs. It can be concluded that the geometry of the Linde silica plugs is too complicated to determine pore radii from volume to surface ratios alone. The values of the Kelvin radii for L.S.P.-5 and -10 can serve, however, as a rough approximation and may be used as a criterion for making a relative comparison of pore sizes of the plugs. In order to justify the treatment of Linde silica plugs as rigid adsorbents for the calculation of free energy changes during adsorption, it was necessary to demonstrate that appreciable swelling did not occur. If swelling of the plugs were a critical factor, it should produce a noticeable increase in pore dimensions of the plugs. This change in the geometry of the plugs during adsorption should be detected, if appreciable, by inconsistency of the second and third run adsorption and dew sorption points. In all cases, these points fell, within the limits of experimental errors, upon the original curves. This reproducibility of -22 -

adsorption and desorption values on Linde silica plugs justifies the conclusion that these plugs do not materially swell during adsorption and that they may be considered as rigid adsorbents in the calculation of free surface energy changes. General Characteristics of the Adsorption Isotherms: According 18 to the B.D.D.T. theory all of the adsorption isotherms obtained for (18) S. Brunauer, L. S. Deming, W. E. Deming, and E. Teller, J. Am. Chem. Soc., 62, 1723 (1940). Linde silica plugs are Type IV isotherms, and the isotherm for the adsorption of benzene on loose Linde silica powder is a sigmoid or Type II isotherm. The adsorption isotherms of cyclohexane on silica gel may be either Type IV or V from a consideration of the shape of the curve at low relative pressures, but from a B.E.T. plot of the adsorption data it is ascertained that the isotherm is Type IV. Fig. 1 illustrates the difference in the shape of the isotherms obtained with benzene on L.S.P.-5 and that obtained on Linde silica powder. The curves are identical up to a relative pressure of 0.05 and significant differences do not become apparent until a relative pressure of about 0.3 is reached. The adsorption of benzene on loose Linde silica powder remains lower than that on L.S.P.-5 until the pores of the plugs are nearly filled. The adsorption on the plugs has then reached almost a finite limit as the adsorbing surface is practically zero, whereas adsorption on the loose powder is essentially unlimited and becomes progressively larger. The increased adsorption of the plugs in the high relative pressure range, especially within the hysteresis range of the plugs, and probably also in the low relative pressure range, is in direct opposition to the expected results if the pores were filled by only adsorbed multilayers -23 -

of benzene. It seems reasonable to conclude that the increase in adsorption is produced by capillary condensation, which appears to be the main process by which the pores are filled7'19. (19) P. C. Carman and F. A. Raal, Nature, 167, 112 (1951). The prominent hysteresis on desorption with the loose Linde silica powder appears, upon first consideration, to indicate that the powder was porous. However, the adsorption isotherm saturation value could neither be determined nor even closely approached experimentally and thus the isotherm cannot be considered as a Type IV curve. This hysteresis is interpreted to be associated with capillary condensation, which, in this case is believed to have occurred at the zones of contact of the loose particles. Utsugi and Higuchi20, Pierce and Smith21 Ries 22 25 et al., and Shereshefsky and Russell have presented evidence that (20) H. Utsugi, J. Chem. Soc. Japan, 75, 631, (1954); I. Higuchi and H. Utsugi, ibid., 73, 395, 397 (1952). (21) C. Pierce and R. N. Smith, THIS JOURNAL, 52, 1111 (1948). (22) H. E. Ries, Jr., M. F. L. Johnson, and J. S. Melik, ibid., 53, 638 (1949). (23) J. L. Shereshefsky and E. R. Russell, ibid., 7, 660 (1953). capillary condensation occurs in this manner with non-porous solids. From Fig. 2 a comparison can be made of the isotherms obtained with cyclohexane on silica gel and on L.S.P.-5. The steep portions of the curves, in the hysteresis regions, are displaced toward lower relative pressures for silica gel than for L.S.P.-5. The limiting adsorption values are also approached at much lower relative pressures and the hysteresis loop is much wider with the isotherm for the silica gel than for the L.S.P.-5. These dissimilarities in the isotherms are attributed mainly to great differences in pore sizes, i.e., the silica gel has much smaller pores than the plugs.

This same comparison can be made for isotherms for L.S.P.-5 and L.S.P.-10; however, here the differences in isotherm features are not as great as those between compressed plugs and the silica gel. For the same adsorbates, it was generally observed that the hysteresis loops are slightly wider for L.S.P.-10 isotherms than for L.S.P.-5 isotherms. In the high pressure region, the limiting adsorption values occur 'at a slightly lower relative pressure for L.S.P.-10 than for L.S.P.-5. These differences in the isotherms are interpreted to mean that the average pore radii of L.S. P.-10 are smaller than those of the L.S.P.-5. The larger amounts adsorbed at saturation for L.S.P.-5 than for L.S.P.-10 are due to the higher porosity and larger pore volume of the former. A comparison of results is shown in Figures 3 and 4 for toluene and n-hexane on L.S.P.-5 and L.S.P. -10. Differences in the isotherms obtained with water and ethanol as well as with chloroform and n-heptane on L.S.P.-5 are shown in Figures 5 and 6, respectively. Since these isotherms were determined on the same types of plugs the differences in shape depend primarily on the nature and the size of the adsorbate molecules. Free Surface Energy Changes During Absorption: Employing the Gibbs adsorption equation in the form of equation 2, the quantity -as (7SO - rSL) was determined by graphical integration of plots of x/m, the amount adsorbed per gram of adsorbent, versus log P/po over the complete relative pressure range. The free surface energy change values (7SO - wSL)' were obtained by dividing -as(7ySO ' SL) by the measured B.E.T. nitrogen area, as. The free energies of immersion (7S - 7SL) at 26~ and in some cases at 40~ for a variety of adsorbates on L.S.P.-5 and L.S.P.-10 are -25 -

compiled in Table II. Values of ySO - 7SL for silica gel as well as adhesion tension values for siliceous solids are listed for comparison. A direct comparison of the calculated free energies of immersion at 26~ for L.S.P.-5 and L.S.P.-10 can be made since nine of the eleven different liquids used as adsorbates were common to both types of plugs. Except for l-propanol the results are in very good agreement. The deviations in this case are attributed mainly to experimental errors in the values of x/m for adsorption at low pressures. These errors will not affect the conclusions that may be drawn from the free energy of immersion data. It is believed that a comparison of the SO - YSL values in Table II serves as the best criterion in this research for estimating errors of this method. Although the two types of plugs did possess noticeable differences in pore volumes, porosities, pore radii, and surface areas as well as. differences in the shapes of their isotherms, the close agreement of the SO - SL values obtained for the same adsorbates on L.S.P.-5 and SO 7SL 5 L.S.P.-10 is believed to justify the conclusion that differences in physical structures of the plugs do not affect the magnitude of these values. The main requirements for the plugs are: (1) that the plugs fill up completely at or near the saturation pressure, and (2) that the plugs possess a constant surface area. Since Linde silica is a hydrophilic solid, it is to be expected that polar liquids would produce larger energy changes on interaction with this solid than would the non-polar liquids. This is substantiated by the data in Table II. The range of S0 - 7SL values was from 139.5 ergs/cm.2 for water to 27.7 ergs/cm.2 for cyclohexane on Linde silica. Since the free energy of immersion gives a measure of the degree of -26 -

wetting of a solid by a liquid, the values of the free energy of immersion of Linde silica enable one to determine the relative order in which these liquids best wet this solid. The values of YSO - 7SL obtained by Bartell and Bower9 on silica gel, listed in Table II, are in the same relative order as those obtained with same liquids on Linde silica plugs. The values are in quite good agreement for mdst of the adsorbates and it is believed that experimental errors in both researches prevented even better overall agreement in the results. 24 Harkins4 reported data for the lowering of free surface energy (24) W. D. Harkins, The Physical Chemistry of Surface Films, Reinhold Publishing Corp., New York, 1952, p. 216. of different polar solids, both porous and non-porous, which fall between the following limits in ergs/cm.2: water at 25~, 180 to 250; n-butane at 0~, 30 to 45, and n-heptane at 25~, 25 to 69. Only a rough comparison can be made, but in general, the data reported by Harkins show much higher values of gSO- 7SL than were found in this investigation. All of Harkins' values of gSO - 7SL for water are much higher whereas some of his smaller free energy of immersion values for n-heptane offer reasonable agreement with those obtained for n-heptane on Linde silica plugs. It should be pointed out that most of Harkins' data were reported for low area, non-porous solids and that no distinction was made between porous solids and non-porous solids in his treatment of adsorption data. Bartell and collaborators have obtained adhesion tension data for siliceous solids by the "pressure of displacement" method2526, and (25) F. E. Bartell and H. J. Osterhof, THIS JOURNAL, 3, 543 (1933). (26) F. E. Bartell and C. E. Whitney, ibid., 36, 3115 (1932). -27 -

by the capillary tube method27 in which a completely different approach (27) F. E. Bartell and E. J. Merrill, ibid., 36, 1178 (1932). was used to determine free surface energy changes on solids. In Table II the 7SO - YSL values obtained in this research are compared to the adhesion tension values obtained in earlier work. The values for benzene, toluene, chloroform, carbon tetrachloride and n-hexane obtained by these widely different methods are in fairly good agreement. However, the adhesion tension values for water on siliceous surfaces obtained by these other methods are much lower than those obtained by adsorption methods.* In addition to the free energy of immersion, two other free surface energy changes, the initial spreading coefficient and the work of adhesion are often used in the correlation of wetting of solids by liquids. The initial spreading coefficient, SL/S, may be calculated by subtracting YLVo from the free energy of immersion (7SO - YSL), and the work of adhesion, Wa, may be calculated by adding 7LVo to the free energy of immersion. These calculated initial spreading coefficients are compared in Table III to the experimental term, ZSO - ZSVQ for Linde silica plugs * In the "displacement pressure" method and the capillary tube method for determining adhesion tension the solids were exposed to the presence of vapor from the liquids being used in the capillary systems. Adsorption of vapor could occur at the solid-air interfaces which would alter the surface tension YSA of the solid-air interface. The adhesion tension value, 7SA - ZSL' might, therefore, be quite different from the free energy of immersion value ZSO - YSL- In the latter, the term 7SO applies to the surface tension of a solid in vacuum and not to a solid in contact with a gaseous atmosphere. Owing to the great tendency of silica to adsorb water vapor, the value of the surface tension of silica 7SA would be expected to be much lower than the surface tension value ZSO in vacuum. Accordingly, the determined adhesion tension value of water against silica in air should be considerably lower than the determined free energy of immersion value. It is somewhat surprising to find that the adhesion tension values obtained for other liquids used in this investigation give values so well in agreement with the free energy of immersion values obtained in this research. -28 -

7 and for silica gel as determined by the method of Fu and Bartell7. In general, the agreement between the values of ySo - 7SV and of So - 7SL - yLVo is reasonably good, and means that the differences between the experimentally determined values of the terms (7So - 7SL) and (7SO - 7SVo) give fairly good values for 7LVo. This agreement serves to indicate that 7SVo - 7SL = 7LVowhich is a special case of Young's equation, 7SVO - 7SL = 7LVo cos @, when G = 0~. Temperature Coefficients of the Free Energy of Immersion: The applicability of the Gibbs-Helmholtz equation to the relating of the heat of wetting and the free energy change of wetting was first realized by 28 29 Williams and by Harkins and Ewing29 and has been used by Bartell and Fu30 (28) A. M. Williams, Proc. Roy. Soc. (Edinburgh), 38, 23 (1917-1918). (29) W. D. Harkins and D. T. Ewing, J. Am. Chem. Soc., 43, 1787 (1921). (30) F. E. Bartell and Y. Fu, Colloid Symposium Annual, 1, 135 (1930). to calculate surface areas from heat of wetting and pressure of displacement data and by Bangham and Razouk5 by a modified form of the GibbsHelmholtz equation to calculate the heat of wetting for charcoal immersed in methyl alcohol. The Gibbs-Helmholtz equation in the following form: Hw = as [(sT d(so - YSL)]) ],Aw = as (7S0 - 7SL) - T 4dT in which 6SHw represents the heat of immersion, has been generally accepted, but as yet, the direct and independent measurement of the quantities, needed to experimentally check its validity, has not been reported. Since the area of the solid has already been incorporated in the heat of immersion (hSo - hSL) and the free energy of immersion (7SO - 7SL) data, equation 4 was used in the following form: -29 -

T d(y Y ) -SO SLT hso -hSL= (YSO - 7SL) (5dT ' 2 where hSo and hSL represent the surface enthalpy per cm. for a solidvacuum and a solid-liquid interface, respectively. In Table IV are shown d(yso - YSL)/dT values of benzene, cyclohexane, methanol and of water for Linde silica calculated from the unit area heat of wetting data obtained.by Bartell and Suggitt31 and the free (31) F. E. Bartell and R. M. Suggitt, THIS JOURNAL, 58, 36 (1954). energy of immersion data at 26~ of this research. Using the calculated d(yo - YSL)/dT values and the ySO - ZSL values at 26~ the free energy of immersion values at 40~, shown in Table IV, were calculated from the following relationship: (T2 - T1) d(0 - 7SL (so - SL)T2 ('SO - SLT1 + (6) 2 1 dT in which T2 = 313~K. and T1 = 299~K. Experimental values of 7yS - 7SL at 40~ for cyclohexane and for water are also listed and they show satisfactory agreement with the calculated values within the range of experimental errors. Thus, it is indicated from this investigation that the GibbsHelmholtz relationship may be used to obtain significant temperature coefficients of free energies of immersion. It is also indicated that the compatibility of the heat of wetting data and the free energy of immersion data supports the validity of this equation. Characteristics of Adsorbed Films on Linde Silica Plugs: The monolayer capacity of each vapor adsorbed on L.S.P.-5 and L.S.P.-10 was determined from the gravimetric adsorption data by application of the -30 -

linear form of the B.E.T. equation. The average values obtained for L.S.P.-5 and L.S.P.-10 are recorded in Table V. The thickness of the adsorbed monolayer was approximated for each system by the following relationship: (x/m) t = m (7) pa in which t is the thickness of the adsorbed film, (x/m)m is the capacity of the monolayer, p is the density of the bulk liquid and as is the specific surface area of the adsorbent. The thickness of the alcohol films seem to indicate that there is some degree of ordered orientation on the surface with these molecules as revealed by the regular increase in film thickness from methyl to. propyl alcohol. In these cases the alcohol molecules are, in all probability, oriented in an upright position in the monolayer with the polar group attached to the surface of the solid. For the remainder of the adsorbates no definite conclusions seem to be justified though it does appear probable that some of the remaining organic adsorbates, especially the straight-chain aliphatic molecules may be lying flat on the 52 surface of the solid. (32) C. Kemball and E. K. Rideal, Proc. Roy. Soc. (London), AL7, 53 (1946). The crossesectional area of each adsorbate molecule which must be assigned to allow agreement with the B.E.T. nitrogen areas for L.S.P.-5 and L.S.P.-10 was calculated and the results are listed in Table V. The cross-sectional areas, C.S.A., obtained for a liquified film of closest packing according to the equation: -31 -

C.S.A. (L) = 4 x 0.866 (2/3 (8) where M is the molecular weight of the adsorbate, p is the density of the bulk liquid and N is Arogadro's number, are low, varying from a factor of about 1.2 for 1-propanol to 2.2 for cyclohexane. These values need adjustment to check the C.S.A. obtained by the B.E.T. method. The real reasons for these differences are not known but cluster formation and polymerization of the molecules in the adsorbed state may have a bearing on the situation. The results are not at all unusual and are in general agreement with the findings of others3334. (33) R. T. Davis, Jr., T. W. Dewitt and P. H. Emmett, THIS JOURNAL, 51, 1232 (1947). (34) P. H. Emmett and M. Cines, ibid., 51, 1248 (1947). The linear form of the B.E.T. equation contains the parameter C, which is a dimensionless constant that has been shown through theory of multilayer adsorption to be related approximately to the heat of adsorption as: C = e (E1 - EL)/RT, (9) in which E1 is the average heat of adsorption of the first layer and EL is the heat of liquefaction of the vapor and E1 - EL has been defined as the net heat of adsorption. The net heats of adsorption, E1 - EL and heats of adsorption, El, are listed in Table V for L.S.P.-10. The values for L.S.P.-5 are not listed but were in good agreement. In every system E1 was found to be greater than EL and thus according to the B.D.D.T. theory Type II or IV isotherms should be obtained which is in agreement with the results of this research. Isosteric heats of adsorption were calculated for chloroform adsorbed on L.S.P.-5 and L.S.P.-10 from adsorption isothermal data -32 -

obtained at 26~ and 40~ by use of the following equation: RT12T (P/po)2 (Po) 2 ssei = ln + ln (10) isosteric T2-T1 (P/Po)1 (PO)i The extrapolated value for the isosteric heat at low adsorption was about 13 kcal./mole which is approximately twice the heat of condensation of chloroform. At high amounts adsorbed, 400 mg./g., the isosteric heat approached the heat of condensation of chloroform. The isosteric heat at monolayer coverage was 9.2 kcal./mole compared to 8.9 kcal./mole for El, the heat of adsorption. -33 -

TABLE I PHYSICAL PROPERTIES OF SILICA GEL AND OF LINDE SILICA POWDER AND PLUGS Linde Silica Linde Silica Linde Silica Physical Property Powder Plugs-5 Plugs-10 Silica Gel Apparent Density (D),g/mla 0.766~0.004 0.895+0.003 0.702 Pore Volume (Vp), ml/g o.850b 0.662b O.970 Av. Pore Volume(Vs), ml./g 0.853 o.664 0.950 Av. Porosity (E)a 0.651 0.593 0.681 B.E.T.Surface Area(as)m2/g. 312 306 309 471 F.B. Surface Area,m.2/ge 298 296 475 Av. Dry Radius, A. 55.2+1.6 43.4+1.7 40.4+0.8 Av.Kelvin Radius(rk),A.g 149+15 118+9 39.4+0.6 o h Av. Particle Size, A. 100 Av.Geometric Surface Area 270 /gh a. Mercury displacement method. b. Based on the density of vitreous silica (2.20g/ml). c. Calculated from adsorption isotherm saturation values. d. The surface areas calculated using a value for the area of a nitrogen molecule of 16.2.2.2 e. Calculated by the method of Fu and Bartell. f. Calculated on the basis of cylindrical geometry (2Vs/as). g. Calculated using the reduced pressures at the steepest slope of the desorption isotherm. h. Calculated from electron microscopy data.

FREE ENERGY OF IMMERSION ( qSO -SL ) OF LINDE SILICA WITH VARIOUS LIQUIDS AT FOR SILICEOUS MATERIALS 260 AND COMPARISON WITH PREVIOUS DATA Adsorbate Linde Silica Plugs-5 ( 0o - L) ergs/cm. 2 Linde Silica Plugs -10 ( ergO L) ergs/cm. 2 Silica Gelsc ( so - SL) ergs/cm. 2 Ottawa Sand g Adhesion Tension ergs/cm.2 Transparent Quart h Adhesion Tension ergs/cm.2 Water Methanol Ethanol 1-Propanol 2-Methyl-l-propanol 139.5a o09.8 100.2 90.2 125-129 76.7* 75.9* 112.1 100.1 95.4 99.9d 97.1d I I 1-Pentanol Benzene Toluene Chloroform 49.4 47.0 44.4 49.7 58.6e 44.1 73.1 45.4 46.5 47.1 43.7 43.2 47.4 Carbon Tetrachloride 35.6 35.0-38.4e 36.3 - 35.7 n - Heptane n - Hexane Cyclohexane 28.8 27.9 27.5 28.8 27.9 27.8 28.8-31.0e 27.3b 25.9 29.9 a. (7SO -7SL) for Linde Silica Plugs-5 at 40~ was 129.2 b. (SO -?SL) for Linde Silica Plugs-10 at 40~ was 25.3 c. Data from reference (9), Bartell and Bower. d. Adsorption isotherm determined at 25~. e. Adsorption isotherm determined at 26.5~. f. Data from this research. g. Data from reference (26), Bartell and Whitney. h. Data from reference (27), Bartell and Merrill. ergs/cm.. ergs/cm. 2.

TABLE III COMPARISON OF EXPERIMENTAL AND CALCULATED INITIAL SPREADING COEFFICIENTS FOR LINDE SILICA PLUGS-5 AND -10 AT 26 o Adsorbate Water Methanol Ethanol 1-Propanol Benzene Toluene Chloroform Carbon Tetrachloric n- Heptane n - Hexane Cyclohexane Linde Silica Plugs-5 Experimental Initial Calculated Initial Spreading Coefficient Spreading Coefficient (7S0 -/S ) (7So - L 7 LV~) ergs/cm ergs/cme 65.2 67.7 85.1 87.7 78.9 78.0 70.3 66.9 25.4 21.2 20.9 19.1 19.5 17.9 Linde Silica Plugs-10 Experimental Initial Calculated Initial Spreading Coefficient Sreaing Coefficient sergs/cmo SO, ergs/cm5 uergs cm L 88.2 78.4 74.5 22.3 20.6 18.9 10.3 9.8 10.7 6.3 88.9 77.9 72.1 21.5 19.2 17.2 9.6 9.1 10.1 5.2 01 le 10.0 10.1 5.3 9.1 10.1 3.2

Adsorbate Benzene Cyclohexane Methanol Water TABLE TEMPERATURE COEFFICIENTS OF (hSO - hSL)a (SO - ZSL) at 26~ at 26~ ergs/cmr. ergs/cm? 97 49.6 68 27.7 185 111.0 261 139.5 IV (7SO - SL) ON d(ySO - YSL) dT ergs/cm?/deg. -0.158 -0.135 -0.247 -0.406 LINDE SILICA alculated wso - 7SL) at 40~ ergs/cm? 47.4 25.8 107.5 133.8 Experimental YSO - YSL) at 40~ ergs/cm? 25.3 129.2 a. Data from reference (31), Bartell and Suggitt.

TABLE V CHARACTERISTICS OF Monolayer Capacitya V Adsorbate mg/g. Water 37.5 Methanol 61.8 Ethanol 75.7 1-Propanol 90.3 Benzene 77.0 Toluene 80.5 Chloroform 111.4 Carbon Tetra- 122.8 chloride n-Heptane 61.2 n-Hexane 54.6 Cyclohexane 53.9 ADSORBED FILMS ON Monolayer Thicknessa t A. 1.2 2.6 3.1 3.7 2.9 3.1 2.4 2.5 2.9 2.7 2.4 LINDE SILICA PLUGS-10 AT 26~ Apparent Net Heat of Cross-sec- Adsorptiona tional areaa E1 - EL m2 kcal/mole 24.6 26.5 31.1 34.0 51.8 58.4 54.7 63.6 83.6 80.6 76.8 1.83 2.78 3.23 3.52 1.74 1.96 1.59 1.32 1.44 1.42 1.00 Heat of Adsorption Ei kcal2mole 12.33 11.94 13.47 15.16 9.86 11.05 8.88 9.07 10.20 8.93 8.17 a. Average values for Linde Silica Plugs -5 and -10.

ILow r a Desorption c0 E 1000 -o rO 3I 0 u) 800 - I 600 p0. 400 0.1.2.3.4.5.6.7.8.9 Po' Fig. 1. -- Adsorption-desorption isotherms of benzene on Linde silica plugs-5 and Linde silica powder at 260.

900 I I E 0 w co3 cr 0 U) 0 m (D I w 0 UOL Silica Gel o Linde Silica Plugs-5- -- 700 7 " ' o Adsorption A Desorption -00 600 - / 500 / 2 0 0 - i AV /, / / 400 I" I XfX~l Cot/ I / 300 -/ / I' i 200 — /,, — l O G (frL E J Q. 0.1.2.3.4.5.6.7.8.9 1.0 P Po. Fig. 2. -- Adsorption-desorption isotherms of cyclohexane on silica gel and Linde silica plugs-5 at 26.

I - E 0 Q C) (3 0 I-, 0J CD.1.2.3.4.5.6.7.8.9 P Po. Fig. 3. -- Adsorption-desorption isotherms of toluene on Linde silica plugs-5 and -10 at 26~.

E 500 w 0 00! // f-FT o 400 0.1 2.3.4.5.6.7.8 9 1.0 ro o~:500 -" - ~0.1.2.3..4.5.6.7.8 P9 1.0 po. Fig. 4. -- Adsorption-desorption isotherms of n-hexane on Linde silica plugs-5 and -10 at 26~.

_ X E 50 -I <[ -P — 500 P 300 200 - 100...0.. — - -~ — ~ ~ 0.1.2.3.4.5.6.7.8.9 P Fig. 5. -- Adsorption-desorption isotherms of water and ethanol on Linde silica plugs-5 at 26~.

800 0 0 w 60C 400 -0.1.2.3 -4.5.6.7.8.9 POp0. Fig. 6. - Adsorption-desorption isotherms of chloroform and n-heptane on Linde silica plugs-5 at 260.

III. FREE ENERGY OF IMMERSION OF LOOSE SILICA AND GRAPHITE POWDERS WITH ORGANIC LIQUIDS Comparison of Values from "Non-Porous" and "Porous Solid" Treatment of Adsorptiori Data Previous work from this laboratory has shown that adsorption data obtained with rigid porous plugs formed by compressing non-porous powdered solids can be used for the determination of the free surface energy of immersion of the non-porous powdered solids. The method of treating adsorption data from porous solids has been termed the "porous solid" method.1 The generally used "non-porous solid" method of calculating free energy of immersion from adsorption data obtained with loose silica or graphite powders gives high values, while calculations for these systems by the formula for the "porous solid" method give low values. It is proposed that this is because neither of the formulations for the two different methods of treatment conforms to the actual adsorption process which occurs in the loose powder system. The main difficulty lies in the uncertainty of the interpretation of the free surface energy changes that actually occur at or near saturation when capillary condensation takes place to a considerable but incomplete degree. When the formulation for the "porous solid" treatment was applied to the adsorption data obtained with powdered non-porous solids the calculated free energy of immersion values were fairly close to those obtained with compressed porous plugs of this material, while calculation by the formulation for the "nonporous solid" treatment gave values which were in some cases 20 to 40 percent higher. It appears justifiable, therefore, to conclude that in the case of vapor adsorption by a mass of loose powder a definite amount of capillary condensation of vapor occurs in the spaces between the particles of powder. 1 Y. Fu and F. E. Bartell, THIS JOURNAL, 55, 662 (1951).

INTRODUCTION In previous papers2'3 it has been shown that the free energy (2) R. G. Craig, J. J. Van Voorhis and F. E. Bartell, THIS JOURNAL, 60, 1225 (1956). (3) J. J. Van Voorhis, R. G. Craig and F. E. Bartell, THIS JOURNAL, of immersion, ZSO - YSL of various liquids on graphite and silica powders can be determined from the complete adsorption isotherms of the different adsorbates obtained with porous plugs of the powder. These plugs are prepared by highly compressing the free flowing graphite or silica powders. Application is then made of the Gibbs adsorption equation in the following form: P/P = 1 RT r ~ x/m diln p/PO RT / SO/SL = 7S - SL = Ma ( P/Po 0 The adsorption isotherms obtained with the porous plugs were Type IV. These isotherms could be reproduced even when the same porous plug was used on second and third successive adsorption runs. The amount adsorbed approached a finite limiting value at or below a relative pressure of 1 and pronounced hysteresis was observed in all cases. The adsorption isotherms obtained with free flowing powders were, however, Type II isotherms. Thus, a limiting value for the amount adsorbed at saturation could not even be closely approached, and therefore, equation 1 could not be applied. In the past the following formulation has been used to determine the free energy of immersion of powdered solids when the adsorbates form a zero contact angle with the solid. P/PO = 1 RT r x/m din p/po + Y7LV SO S 7L Mas J p/S- =0 (2) -46 -

During the adsorption process it was observed that the volume occupied by the free flowing powders decreased, and on desorption the volume approached its orginal value. The isotherms obtained also indicated an unexpectedly large hysteresis loop (Fig. 1). Thus, it appeared that the powders were functioning, at least partially, as porous solids. In order to evaluate the importance of this effect on the calculated free energy of immersion values, the adsorption data were calculated both according to equation 1 (the "porous solid" treatment) and to equation 2 (the "non-porous solid" treatment). The values thus obtained were then compared. EXPERIMENTAL The materials, apparatus and methods used to obtain the ad- sorption isotherms were the same as those described in previous papers '. The evaluation of the integral in equation 2, for powders was,accomplished by extrapolation of the curve of the semilog plot of x/m versus log P/po, thus obtaining an estimated saturation adsorption value. With this exception the graphical integration was identical to that performed for the porous solids employing equation. RESULTS AND DISCUSSION The values of the free energy of immersion, ySO - SL' were calculated from the adsorption data for 1-propanol on graphite A powder, cyclohexane on graphite B powder and benzene on Linde silica powder according to equations 1 and 2. For purposes of comparison, these values obtained for the "non-porous solid" and the "porous solid" treatments of the graphite and silica powders are listed in Table I along with the values obtained for the same adsorbates on corresponding porous plugs. -47 -

It is quite apparent that all of the values of the free energy of immersion, YSO - 7SL, are much higher (20-25 ergs/cm.2) for the non-porous powders calculated according to equation 2 than for the corresponding values for the porous plugs calculated according to equation 1. It does not seem reasonable that the same solids (i.e., loose, non-porous powders and porous plugs prepared form the same powders) should produce different free energies of immersion merely on the basis of differences in physical structure of the solid mass. 4 This point is substantiated by the data of Bartell and Suggitt who re(4) F. E Bartell and R. M. Suggitt, THIS JOURNAL, a, 36 (1954). ported unit area heats of wetting (hso - hSL) for Linde silica powder and for silica gel that are in close agreement. Also two different graphites having widely different surface areas gave the same hso - hSL values within the limits of experimental error. The considerable capillary condensation which occurred during the adsorption process of the adsorbates on the loose powders indicated that integration of the adsorption data for these systems in terms of the "porous solid" treatment (equation 1) would be profitable. In as much as the actual numerical integration remains the same as before, the only difference that now prevails is the interpretation of the free surface energy change which may occur at saturation. According to the "porous solid" treatment, the interpretation of the change in free surface energy for these systems indicates that the values of S0 - 7SL 2 are as follows: benzene on Linde silica powder is 46.3 ergs/cm., cyclo2 hexane on graphite B powder is 55.7 ergs/cm. and 1-propanol on graphite A powder is 65.8 ergs/cm.. These values are in fair agreement

with the values reported for the same systems with the powders in the form of porous plugs. Equation 1, or the "porous solid" treatment is, however, not suitable for rigorous treatment of the adsorption data for the loose powders for the following reasons: (1) experimentally, the liquid-vapor interface is never completely destroyed in such systems, (2) the true saturation adsorption value is never reached, or even closely approached, experimentally and could only be estimated by extrapolation of the adsorption data, and (3) these systems do not exhibit the behavior of rigid, porous adsorbents. If the extrapolated values of the amounts adsorbed at saturation should, fortuitously, closely represent the amount of liquid required to fill up the void spaces between contiguous particles of the powders, then the values of SO - SLfor both the powders and the porous plugs should be in quite good agreement by the "porous solid" treatment of the adsorption data. Serious objections also arise in the strict application of the "non-porous solid" treatment (equation 2) to the adsorption data of the loose powders. The inability to determine with the gravimetric adsorption experiment the true saturation adsorption value, still remains as an objectionable factor; moreover, it is believed that capillary condensation does occur in the adsorption process and that this creates even greater uncertainty in this treatment of the adsorption data. In the absence of capillary condensation, the free surface energy change at saturation would be 7SO - 7SVO (or 7SO - 7SL - LV) for non-porous loose powders; but with the occurrence of capillary condensation, though not to the fullest extent as is the case with the porous plugs, the magnitude of the final change in free surface energy evaluated through -49 -

the Gibbs equation seems likely to fall between the limits of the values of - and of SO- L - LV for the porous plugs. An incomplete degree of capillary condensation would destroy only a part of the liquid-vapor interface, and the addition of the entire numerical value of the free surface energy of the liquid YLve to the value of7S0 - ZSV7On obtaining the value of ZSO - 7SL by the "nonporous solid" treatment would not be justified. If part of the liquid-vapor interface is destroyed by capillary condensation during the adsorption process, then only the undestroyed part of the liquidvapor interface, i.e., a certain fraction of the nunerical value of the surface tension of the liquid should be added to the value of the free energy change at saturation in order to determine the value of SO - 7SL No method is known, however, for the determination of that fraction of the liquid-vapor interface, which may not be destroyed. It is believed that the theories, hypotheses and assumptions used in the derivation of equations 1 and 2 do not conform to the adsorption process as it occurs on loose Linde silica and graphite powders. Neither equation 1 nor equation 2 is adequate for evaluating, unequivocally, free surface energy changes for these systems. It is further believed that the main difficulty lies in the inability to determine the free surface energy changes that actually occur in these systems at or near the saturation pressures, when capillary condensation takes place to a considerable but yet incomplete degree. Consequently, the values previously reported for SO - SL SO SL on loose Linde silica and graphite powders, as calculated both by the "porous solid" and the "non-porous solid" treatments of the adsorption data, must be regarded as questionable values. However, when the non -50 -

porous powders are compacted into rigid, porous plugs, free energies of immersion can justificably be calculated from adsorption data by application of equation 1 which covers the "porous solid" treatment. -51 -

TABLE I COMPARISON AND OF FREE ENERGY OF IMMERSION FROM "NON-POROUS SOLID" "POROUS SOLID" TREATMENTS OF ADSORPTION DATA FOR LOOSE POWDERS AND POROUS PLUGS AT 26~ Loose Powders "Non-porous Solid" "Porous Solid" Treatment (eq.2) (7SO YSL) Adsorbate ergs/cm.2 Adsorbent Treatment (eq. 1) (YSO - YSL) ergs/cm. 46.3 Porous Plugs "Porous Solid" Treatment (eq.l) (so - ySL) ergs/cm. 49.6 Linde Silica Benzene 74.5 Graphite A Graphite B 1-Propanol Cyclohexane 89.6 79.3 65.8 55.7 68.6 55.2 -52 -

E l -i00 a: II 800 I a)Ii v 600 --- 0 40D0 ----------------------— _..- e- ~-p 200 ---- ----- -—.-.'o'-'-'~ "" ---0.1.2.3.4.5.6.7.8.9 1.0 P Po' Fig. 1. -- Adsorption-desorption isotherms of benzene on silica powder and 1-propanol on graphite A powder at 26~.

UNIVERSITY OF MICHIGAN 3 90111 5 02228 98731111 I11 3 9015 02228 9873