Oct., 1961
ADSORPTION OF OIIAWL~XLE SULFONATES AT METAL-OIL INTERFACE
Conclusions The data continue to affirm that there is an inherent characteristic of adsorptive behavior, namely that the heah of adsorption are directly dependent on particle size. There is a uniform trend for both crystalline modifications of TiOz. The behavior with outgassing temperature is similar to that for SiOz; it is interpretable in terms of the progressive loss of physically adsorbed water and chemically bonded surface hydroxyl groups. It is fruitless to say that any one of the TiOz samples could be picked as the “representative one” without more insight into its surface structure. Acknowledgment.-The authors express appreciation to the American Petroleum Institute for their continued interest and support. Some surface areas were measured by Mr. H. D. Cole and his assistance was greatly appreciated. Dr. 11. J. C. Yates was very helpful in supplying both TiO, samples and a preprint of his paper6 which had been submitted to Journal of Physical Chemistry. DTSCUSSIOK 1). S. MACIVEF: (Gulf Research and Development Company) -Concerning Professor Wade’s observations relative to the heats of immersion of A120,and Ti02 in water, I mightmake two comments. In the first place, the heat of immersion of AI&, is dependent not only on such factors as particle size, outgassing temperature, etc., but also on the crystalline structure of the A . 1 2 0 3 . In the case of the high area aluminas such as y-Al&, several crystal modifications outgassing temperature and only becomes important a t temperatures above 200 t o 300’. Presumably, t h e A H i measurements a t higher outgassing temperatures for samples not treated with oxygen also include a heat source in thal oxygen in the colorimeter oxidizes t h e reduced Ti02 during t h e inimersional measurements. (11) C. M. Hollabaugh and J. ,J. Chessick. J . Phys. Chem.. 6S, 109 (1961).
1683
exist and attempts to compare immersional heats of a series of aluminas must take this factor into consideration. Secondly, I would think that, in the case of TiOz, plots of AH^ us. weight loss during outgassing might be more informative than plots of AHi us. outgassing temperature (Fig. 1). Have the former plots been made:? K. H. WADE.-^ would hasten to concur that gamma, chi, etc.-aluminas have not been characterized; however, the a-aluminas which are well characterized show a uniform decrease in AHi with decreasing particle sizc. Your point regarding TiO, is well taken and no plots of the type you mention have been made. Many Ti02 samples lose weight in excess of that attributable to surface losses. Presumably, decomposition occurs at higher outgassing temperatures. I,. H. REYEFLSON (University of Minnesota) .-How were your samples outgassed:) Outgassing can remove oxygen atoms from the surface so that the surface is Ti& not TiO:. This was proved in our laboratories by sorption of KO2 and magnetic susceptibility studies on TiO?. NO: gave up an oxygen at,om to fill each hole which had lost oxygen during outgassing. W. H. \?;ADE.--AI~ Ti02 samples were outgassed at the temperatures noted and a t 10-6 mm. for approximat.ely 100 hours. kV. I). Ross (E. I. du Pont Company).-The comment has been made that ‘Pi02 can lose oxygen upon heating, even if riot contaminated by organic material. A distinction should be made between anatase and rutile. Rutile is the more stable and can be heated strongly with no appreciable loss of oxygen. Anatase loses oxygen more readily: I ani not sure how strongly it can be outgassed. I t has been remarked thxt some metallic impurities affect oxygen loss: with this I agree. W. H. R’Ai)~.--Oneof the rutile samples reported in the present study showed some evidence of decomposition although it is of the highest purity attainable. I am not sure whet,her the correct explanation has yet been offered for the loss of oxygen by some Ti02 samples. A. C. ZEWLEMOYER(Lehigh University).-Our laboratory has shown that the loss of oxygen from rutile may he identified with the presence of traces of organic substances.
THE ADSORPTION OF OIL-SOLUBLE SULFONATES AT THE METAL/OIL INTERFACE BY WILLARD D. BASCOM AND C. R. SINGLETERRY U.S. Naval Research Laboratory, Washington 26, D. C. Received March 80,1961
It has been found possiblc to isolate adsorbed films of thc salts of dinoriylnaphthalenesulfonic acid on stainless steel surfaces by retraction from aromatic hydrocarbon solution. The wetting behavior of various series of liquids on the resulting film-coated surfaces suggests that the molecules adsorb to give monomolecular films by attachment of the polar sulfonate heads to the metal oxide surface with the hydrocarbon tails outward. The molecules are sufficiently close packed to yield a surface having properties similar to those of polyethylene. A study of the wettability of the soap monolayers by a series of alkylnaphthalene liquids indicates that adsorption from water-saturated solution is independent of the soap cation and is the result of dipole interactions between the hydrated sulfonate ion- air with the metal oxide surface. Adsorption from anhydrous solution does depend upon the choice of cation. Here dipore interactions between the unhydrated ion-pair and the metal-oxide surface appear to be supplemented by coordination of the cation with the oxygen of the oxide film.
Introduction Many stiidiwi have been made of thc adsorption of polar-non-po’lar solutes from non-aqiinous solution oilto oxide-coated metal surfaces. Most of these investigations, however, have been cmcerned with the carboxylic acid and amine derivatives of straight-chain hydrocarbons and fluorocarbons. The adsorptive behavior of the soaps of high molecular weight acids has received relatively minor attention despite their wide use as corrosion inhibitors and sludge dispersants in liihricating oils. A
more complete knowledge of soap adsorption at the solid-oil interface would permit a het,tcr understanding of the role of such soaps i i i thew technological areas. It has been found at this Laboratory, for instance, that the effectiveness of oil-soluble soaps in diminishing ice adhesion in lubricated syst’emsis in part. related to their adsorption at the oilmetal interface. The work reported here concerns the adsorptioii of various salts of dinonylnaphthalenesulfonic acid on stainless &el. These, compounds, n-cll char-
1681 W I L L ~ D. D BASCOM m C. R. SINGLEFERBY VOl. 65 acterized in studies concerning their micelleform- differed eufliciently in constitution to be considered sepaing properties in non-polar solvents,' were adsorbed rate membera of the aeries of alkylnaphthalenea. Surface Preparation and Film Isolation.-Stainless steel from isopropylbiphenyl solution. This solvent has cyhdem. t h r e q u u inch in diameter and in height, had a sufficiently high surface tension to be non-spread- one face ground t and then polished to a mirror 6niah with ing on the adsorbed monolayers, thus permitting h e alumina powder on a metallographio wheel. Alumina retained on the surface wy flooded off i n a isolation of the film-covered surfaces by retraction3 powder ofloosely hot tap water followed by a &fled water nnse. from the soap solutions. No previous use has been stream Microscopic examination revealed a small number of partimade of the retraction technique for the isolation cles on the polished surface that were robably alumina. of the total surand study of soap films. The few investigations They appeared to conatitute lees than area. The polishing procedure yields a metal oxide surthat have been made of soap adsorption have either face that ia completely wet by water (0" contact angle). employed the Langmub-Blodgett technique to form face Any tendency for water to wet the surface incompletely sugsoap films on solid surfacesaor have simply washed TABLE I the adsorbate solution from metal specimens with a volatile solvent to isolate film covered surfaces.4 SUEFACE. TENSIONS OF THE WEITINQLIQUIDS Surface tendon. Z i a n and co-workers have investigated the Liquids n v , dynelcm. spreading behavior of liquids on organic polymer Alky lnaphthalenea surfaces and on oleophobic monolayers obtained 1-methylnaphthalene 38.5 by retraction from non-polar solution. They have l16-dimethylnaphthalene 37.7 been able to establish5 a relationship between l-ethylnaphthalene 37.6 wettability and the chemical constitution of the Gamylnaphthalene 1 34.3 solid surface. These concepts have been utilized t-butylnaphthalene Mixed 33.6 in this study to infer the molecular configuration of nonylnaphthalene isomers 32.6 soap monolayers. 31.2 dinonylnaphthalene Experimental
18
J
Materiats.-The dinonylnaphthalendonic acid waa eepecially synthesized for reeearch ue.e. The hydrocsrbon radical of the compound waa prepared by alkylation of naphthalene with a propylene tnmer and this then waa aulfonated. AnalysisS has indicated no detectable mono- or tri-alkyl hydrocarbon. It is reaeonable to BBBume from the method of preparation that the nonyl groupa are highly branched and that the more abundant homologa have both alkyl groups in the aame ring. Also. it is rerreoned that sulfonation occurs on the ring oppoeite the alkyl substituents. The purification of the acid and the pre tion and purification of the metal sulfonate has been Gribed prev i ~ u s l y . ~The acid waa exactly neutrslieed with the a pmpriate base to form the d u m and cesium mapa an$ the divalent soaps were prepared by metatheeie of the d u m map with appropriate morgauic salta. Each preparation waa lyophiliied and stored in aealed ampoules under nitrogen. The octadecylamine waa uaed as received. Solvents used were .hpropylbiphenyl, obtained aa a mixture of neta and puna momera, and practical grade hexadecane. Each liquid waa repeatedly perchlorated through a column of Floroail adsorbent to remove polar contaminants. Their purity in thia respect wm demonstrated by the fact that they showed no tendency to spread on a freshly swept surfwe of distilled water or on the surface of water that was made alightly acid or alkaline. The wetting liquids studied are listed in Table I along with their sdace tensions aa determined using the ring method and employing the corrections given by Harkins.' It waa found for many of these liquids that only after repeated percolation through Florid adsorbent did the ring method give the maximum and reproducible surface tensions listed in Table I. The lower, less precise, valuea were undoubtedly due to the adsorption of polar contaminants on the platinum ring. Molecular distillation of a few of the liquids waa necessary to remove minor amounts of isomers and homologs. This waa never fully achieved for nonyl- and dinonylnaphthalene. although molecular weight determinations showed that they (1) S. Ka-ifman and C. R. Sindeterry, J . ColIoid Sci.. 12. 465 (1957). (2) W. C. Bigelow, 3. L. Pickett and W. A. Zisman. ibid.. 1, 513 (1946). (3) J. E. I'ourlg, Avitrdian J . Ctum., 8, 173 (1955). :4) V. Hang. Y. L. W e ? . I). Bootcin and H. Harrison, C o n o a o n . 10, 343 (1954). :5) E. G . Sbafrin irnd W. A . Zisman, 3. Phya. Chem.. 64, 519 (1960). \ti) S. fia~~ftiisr. and C. R. Siudeterry. J . Colloid Sci.. 10. 138 (less). (7)W. D. f i a r k i n i and W. F. Jordan, J . Am. Ctum. Soc.. 6 1 , I751 i1930).
Alkylbiphenyls 1,l-diphenylethane 2-methylbibenz yl isopropylbiphenyl ( m h , meta) amylbiphenyl diphenyldodecane Miscellaneous Liquida water methyl iodide 1-bromonaphthalene bpropylbiphenyl (Oruro, me&) bie(2-ethylhexyl) sebacate octane
37.7 36.4 34.8 34.9 33 5 72.8 50.8 44.6 34 s 31.1 "1
s
gested the presence of adsorbed organic contamination and required the alumina polishing procedure to be repeated. The specimens were dried in an evacuated rhamber at 7% 80". The platinum surface waa prepared by polishing one face of a dust one inch in diameter and one-axteenth inch thick. Thia surface waa cleaned of organic contamination before each experiment by heating the disc to redness in a gae flame. A specularly reflective polytetrafluom-ethylene surface waa prepared by cutting one face of a small block with a microtome and then preasing the fresh surface against a amooth gless p h t e at 180' and IO00 p.s.i.6 The polyethylene was obtained as a powder and molded under h a t and m u r e to obtain a specularly reflective surface. T f e adsorbed s+onate films were obtained by expoaing stainlepasteel specunens for 16 to 20 hours at 20" to solutions containing 0.5% of adsorbate in bpropylbiphenyl. The specimens then were withdrawn and the solution allowed to retract from their surfaces. Octadecylamine monolayers were deposited on the platinum suatacca by ademption from a 0.1% solution in hexadecane and isolated by retraction. Contact Angle Measurement.-Using a telescope-goniometer. the sessile drop method waa employed to determine advancing contact anglea when they were above 10". A drop I mm. in diameter waa plsoed on a surface and then advanced by two aucce88ive additions of liquid. Three such drop were observed on both aides for each liquid, giving a total of twelvr measurements. The average deviation waa lese than two degrees. It was necesssry that contact angles considerably : e ~ s than 10" be measured with some precision. It is difficult to make messurementa with a teleacape-goniometer rn this low (8) R. C. Bowera. W. C. Clinton and W. A . Zisrnsn. 3. Appl. Phys 34, 1088 (1953;.
.
Oct., 1961
ADSORPTION OF OIL-SOLUBLE SULFONATESAT METAL-OILINTERFACE
range but a micro-interference technique proved to be useful. B metallographic microscope fitted with s 540 mp light aource for vertical incident light waa focused at the drop edge. When the contact angle waa lesa than ten degees it wm possible to resolve the interference bands resulting from normal reflection from the upper and lower s u r f m s of the thin liquid wedge. The interference atterns were photographed and the band spacing meaeurefto obtain the contact angle. It was found that contact anglea measured in this way on two separate drops on the same surface rarely differed by more than one-half degree. All contact angle mwurements and surface tension mHssurements were carried out at 20'. Robbins and LaMer also have reported an interference method for contact angle messurement.g Numerous attempts were made to detect any tendency for the adsorbed sulfonste molecules to dissolve into the wetting liquids. These efforts were unsuccessful. The deliberate addition of sulfonate to the wetting liquids did not result in contact angles different from those obtained using the pure liquids. There was no observable change with tune in the contact-angles for drops allowed to remain on the retracted films for aa long as 20 hours. The resistance of the adsorbed films to solvent dewr tion was further tested by placing a drop of wetting liquicfon a surface and noting the contact angle. This initial drop then waa retracted after a few moments and the contact angle determined for a aecond drop on the same area. No significant difference waa apparent in these instances.
Results The purpose of this work was to study the properties of the films adsorbed on stainless steel from solutions containing dinonylnaphthalenesulfonic acid or its metal salts. The most important property of such films is the surface energy, but this cannot be measured directly; inferences concerning the surface energy of solids are most conveniently derived from contact angle or wetting phenomena. Although some information can be gained by comparing the contact angles of a single liquid, such as methylene iodide, on various surfaces, a more reliable and convenient index of surface energy is furnished by the determination of the critical surface t,ension of wetting, yo, the surface tension of a liquid which wil:l just spread on the solid surface. Zisman and co-workerss have determined yo by measuring the contact angles, 6, of a suitably chosen series of liquids on the surface and plotting the cosine of the contact angle as a function of the surface tension, y ~ v of, the liquids. They found cos 6 to be a linear function of y ~ and v extrapolated the data to cos 0 = 1 to obtain ye. For the study of polytetrafluoroethylene surfaces and adsorbed monolayers of straight chain polar-non-polar compounds, they measured the contact angles of a series of 72slkanes from CISto Ce.*o*llThese liquids, hawever, all spread on the films formed by the dinonylnaphthalene sulfonates. Zisman, et al., when studying surfaces of higher energy than the paraffin chain monolayers, used a series of miscellaneous liquids ranging in surface tension from 30 to 70 dynes/ ~ . ! m . : ~ 2but ~ 1 9in the present work this series was fc,:md to give poorly reproducible results on several oi the fibs. However, two series of alkyl-subutit u t d aromatic compounds were found which had s:ifficiectly high surface tensions, and which gave 1ine:x plots of COS d TS. y ~ for v the organic surfaces ,'J) M . T I . Robbins and 1 '. K. LaMrr, . I Colloid . Sei.. lS, 151 (1960). (10) ii. W. F.?x and W. A. Zieman, ibid.. S, 614 (1950). ; 1. i K. G Shafrix and W. A. Zisrnan, d i d . , 7, 16fi (1952). 21 IT. W. F o x and W. 4. Ziaman, ibid., 7 , 438 (1952). 31 A. R. %Ninon a n a W. A. Zimmn. J . Phua. Chem.. 68. 503 (1954).
tiMA-
ETHYLENE
Goo0 GHXlr_J___I_ ! 250 300 350 400 SURFACE TENSION, q v , l d y W d .
1685
-Iso ---lgoI 410
Fig. l.-"he spreading behavior of a .aeries of alkylnaphthalenes, 0,and a series of alkylbiphenyls, A, on various oleophobic surfaces.
previously investigated by Zisman and co-workers and also for the soap films of this study (Fig. 1). One series consisted of alkylnaphthalenes and the other of hydrocarbons containing two non-fused benzene rings. The alkylnaphthalenes were selected for most of the work because they have structures closer to that of the dinonylnaphthalene sulfonate radical of which the outer surface of the monolayer was presumed to be composed. The surfaces studied and the conditions of film formation were selected to provide information on (1) the variation'of yofor a given surface with the choice of liquid series, (2) the effect of water present during adsorption on the properties of the iilm formed by a given sulfonate and (3) the effect of different cations on the nature of the films. Figure 1 summarizes data for the two series of alkylaromatic hydrocarbons on polytetrafluoroethylene, octadecylamine monolayer, sodium dinonylnaphthalene sulfonate monolayer and polyethylene. These data show that the liquid series rank the critical surface tensions of solid polytetrafluoroethylenc, octadecylamine monolayer, and solid polyethylene surfaces in the order previously assigned by measurements with miscellaneous liquids (Table 11) and by friction studies.14 The alkylbiphenyl series indicates a lower value of yo than does the alkylnaphthalene series, the dlfference being greater for surfaces of lower critical surface tension. For the siirface coated by sodium dinonylnaphthalene sulfonate, the difference is comparable with the uncertainty of the extrapolation. The value of yc for the sodium sulfonate monolayer indicated by the two alkyl aromatic series is approximately 29 dynes/cm. , which is distinctly lower than yo for polyethylene but higher than for octadecylnmine monolayer. Figure 2 compares the data for the alkylnaphthalenes and for the miscellaneous series of liquids on copper dinonylnaphthalene sulfonate monolayer. The alkylnaphthalene data have a steeper slope and extrapolate to a yo higher by three dynes than that indicated by the miscellaneous series. Data (14) W. A. Zmman in "Friction and Wear." edited bv R. Daviea. Elsedar Publishing Co., Amsterdam, 1959. p. 110.
WILLARDD. BASCOM AND C. R. SINGLETERRY
1686
VOl. 65
TABLEI1 THECRITICALSURFACE TENSIONS, yc, OF VARIOUSOLEOPHOBIC SURFACES Surface
Surface constitution
n.4lkanes
71, dynes/cm.
Ref.
Polytetrafluoroethylene -CFz18 10 Octadecylamine monolayer -CH322 11 Dinonylnaphthalene sulfonate -CH3--, -CH, monolayer" Polyethylene -CHZPolystyrene -CH*-, CeHs Isolated by retraction from water-saturated isopropylbiphenyl solution.
for the miscellaneous liquids on the other soap films were not reproducible and are not reported. In Fig. 3 the wetting data for the alkylnaphthalenes on surfaces obtained from a solution of dinonylnaphthalenesulfonic acid and from a solution of the copper soap of this acid indicate a slightly smaller value of yo for the acid film. The water content of the solutions from which the sulfonate films were deposited had a distinct effect on the properties of the monolayer, the magnitude of the effect being specific to the cation present. The pertinent data are shown in Fig. 4, which has been divided into two sections to avoid confusion in the plots. The solid line in the upper section refers to the composite data for all the sulfonate films retracted from water-saturated oil solution. For such films the wetting data, within experimental uncertainty, gave the same linear relationship for all of the metal sulfonates. Adsorption from anhydrous solutions, on the other hand, gave surfaces for which the data fell on distinctly different lines which are presented as dotted lines in Fig. 4a. The difference from the corresponding water-saturated solution was greatest for the barium compound and, as indicated in Fig. 4b, was negligible for copper dinonylnaphthalene sulfonate. Separate experiments demonstrated that the results were not influenced by variations of the relative humidity with which the metal surface was equilibrated before deposition of the film. The time required for film adsorption to be completed for any given soap solution was found to be greater than five hours for the sodium soap in water-saturated solution but less than ten minutes form the dry sodium soap solution and also less than ten minutes for the copper soap in either watersaturated or anhydrous oil solution. This was determined by observing when a-methylnaphthalene gave a maximum contact angle against surfaces retracted from the solution a t various time intervals. It also was found that reduction of the sodium soap concentration from 0.5 to 0.0005 weight 70 in water-saturated oil solution did not diminish the maximuni contact angle obtainab;e with a-methylnaphthahme.
Mkc. liquide
Ref.
18-20 18-24 26 (Cu)
10 11
31
12
Alkyl.41kylnaphthaleneab biphenyls h
0 12 29 (Na)
0 10 29 (Na) :i 2
35 13 This investigation.
not previously employed for this purpose. As Zisman and Fox pointed out1*in their first paper on the subject, yc is not the surface tension, ysv, of the solid, but is smaller by the magnitude of the interfacial tension, y s ~between , the solid and that liquid which just spreads on the solid surface. This can be seen from the Young-Dupre equation which relates the interfacial tensions and cos 8 for a liquid on a solid surface in equilibrium with the saturated vapor of the wetting liquid
Extrapolation of the wetting data for a series of liquids on a given solid to cos 8 = 1 gives yo which may be thought of as the surface tension of a hypothetical member of the series that just spreads. Equation 1 then becomes Y~ = ySv
- ySL, a t cos e = 1
( 2)
In this instance, y s ~is the interfacial tension against the solid surface for the liquid that just spreads and will be small provided y s for ~ the actual members of the series of wetting liquids is small or becomes small as the contact angle decreases. This interfacial tension is believed to be small when the molecular structure of the wetting liquids approaches that of the outer layer of the solid surface as the contact angle approaches zero. This belief is supported by several lines of evidence. Two liquids having similar molecular structures are usually miscible, Le. , the interfacial tension between them is zero. The critical surface tensions of a number of organic solids are not far removed from the surface tension of liquids that closely resemble them in molecular structure. As an example, the wetting data for the alkylbiphenyls on polyethylene (Fig. 1) extrapolate to a yc value slightly higher than 31 dynes/cm. Fox, et uZ.,'~ determined the surface tension of a high molecular weight liquid polyethylene to be 30.7. This close agreement between y ~ vand yo for polyethylene implies that for the alkylbiphenyl liquids ~ S becomes very small in the extrapolation to cos 0 = 1. It is possible to arrive a t the same conclusions if one interprets the interfacial tensions between Discussion non-polar phases as resulting from the difference The Significance of yc Values.-The use of the between the magnitudes of unsatisfied molecular experimental quantity yc as an indication of interaction or dispersion forces a t the surfaces of surface properties or surface configuration of ad- two phases. London16 has related these forces of sorbed molecules needs to be examined critically H. W. Fox, E. F. Hare and W. A. Zisman, J. P h y s . Chem.. when yc must be obtained by the extrapolation of 69,(15) 1097 (1955). contact angle data from a series of hydrocarbons (16) F. London, Trans. Faraday Soc.. 33, 8 (1937).
L
Oct., 1961
168i
\DiiORPTIOiV OF OIL-SOLUBLE: S U L F O N A T E S A T AIE CAL-OIL I N T E R F A C E
interaction for non-polar liquids with their molecular polarizability, so the difference in polarizability of the atomic groups a t an interface may be taken to reflect, the imbalance of dispersion forces. This difference will be least when the atomic groups in the two substances forming the interface are most nearly the same. There may, of course, be residual differences because of different packing densities in the two phases or because of the entropy effect associated with confining the molecules in the solid surface to fixed positions. When the solid surface is made up of oriented polar-non-polar molecules, as it is in this study of sulfonate adsorption, comparisons of molecular structure must be made between the average surface of the liquid and the atomic groups composing the outermost part of the monolayer. Zisman has demonstrated that the surface properties of close-packed films are determined by the nature of the outermost atomic groups and not by the dipole forces associated with the polar heads held on the substrate surface.6 The criteria that must be met for a series of liquids to be useful in the determination of the yc for non-polar solids are (1) that their wetting behavior on the solid surface give a linear relation between cos e and YLV and (2) that the molecular structures of the liquids approach that of the atomic groups prevailing a t the solid surface with decreasing surface tension of the liquids. There is one further limitation. It was demonstrated by Zisman and co-workers'? and also in the present investigation that it was not possible to obtain an imambiguous extrapolation to cos e = 1 of the wetting data for a group of alkylbenzenes on octadecylamine monolayer. This difficuity may well be a result of the very narrow range of surface tensions of the alkylbenzenes. Even though the solid-liquid interfacial tensions for a group of liquids against a non-polar solid may be small, this quantity does not necessarily change in the same way as the surface tension changes with decreasing contact angle. This irregularity may be neglected if the surface tensions of the liquids cover a relatively wide range, i e . , if the increments in surface tensior, arc large compared with the probable change in ysL between the same members. However, the alkylbenzenes 'haw surface tensions between 28.5 and 30.5 dynes/cm., a much smaller range than that covered by the alkynaphthalenes, alkylbiphenyls or the n-alkanes 'Therefore, it is cwncluded that a third condition for a Peries of liquids to be useful for the deterniinalion of y,. is that their surface tensions vary systematically c)ver a substantial range. The three conditions nweswry for a series of wetting liquids are qeen t t i he s:iti>,fiedwhen the n-alkanes dre nsed as hitiids on n paraffin surface or on :I 1 rr,nn,Jl:iyer. They ais0 are catisylnapt7;hairne Iiqnids are observed on monolayers tf tht. ~li?tir?yln:tphth:~lenr sulfonites, ;;\it (I,v ~ 4 t h t j m e :iikvi-arom:L!it*s :ire tipplied to :I I ~ t h r y l surinc.e In thc 1
I 250
3 0
,
-
~
40 0
500 60G SURFACE TENSION, YLy,(dyre/cmi.
d
I
w
'0 0
Fig. 2.---The spreading behavior of a series of alkylnaphthalenes and a series of miscellaneous liquids on copper dinonylnaphthalene sulfonate monolayer retracted from a water-saturated solution of the sulfonate in isopropylbiphenyl.
30 0
35 0
400
SURFACE TENSION ,rLV, (dymkml.
Fig. 3.-The spreading behavior of a series of alkyinaphthalenes on dinonylnaphthalenesulfonic acid monolayers, 0,and on copper dinonylnaphthalene sulfonate monolayer, 0, retracted from water-saturated isopropylbiphenyl solution.
E0
30C
15 c
~qr c
SURFACE TEVSION t,,idyne/m'
Fig. 4.-The effect of wnter in the adsorbate solution and the choice of soap cation on the spreading behavior of alkylnaphthalenes on dinonylnaphthalene sulfonate monolayers.
functions and liquids which contain the more polarizable methylene groups and aromatic rings. I: is, therefore, not surprising that extrapoiation of the contact angle data for the alkylnaphthalenes or the dlkylbiphenyls on an octadecylaminc monolayer -a wwli ni an m- gives mnch smaller numbers for the critical ,.nrface ; I r t w e r i d i d Ytirfacw trnsion than were obtained with the +alk:mes. p d h h l e methyl or perfluoro 7'1-11sconfirms the generalization made by Zimmi 81-
1688
WILLARD D. BABCOM AND C. R. SINGLETERRY
that from among the various series of wetting liquids that can be employed on a given non-polar surface, the surface tension of the solid will be best approximated by the series giving the highest value of Yc. I n spite of the limitation imposed by the uncertain magnitude of YSL, the critical surface tensions obtained on the non-polar surfaces listed in Table I1 using the n-alkanes, alkylnaphthalenes, alkylbiphenyls and the miscellaneous liquids all agree in ranking the surfaces in the same order of increasing surface tension. Therefore, the differences between various monolayers prepared from the dinonylnaphthalene sulfonates which are revealed by rC values estimated with the alkylnaphthalenes may be accepted with considerable confidence to represent real differences in the surface energies of these monolayers resulting from differences in the closeness of packing of the adsorbed molecules. Soap Monolayers.-There is little doubt that the sulfonate films isolated and studied in this investigation are monolayers. This belief is based on the fact that the T~ values for those adsorbed films are all lem than that for the surface of polyethylene which is comprised primarily of methylene groups. Also, the ‘yo values for the soap films are all greater than the value obtained on the methylrich surface of a close-packed octadecylamine monolayer. Thus, the surface properties of the sulfonate monolayers correspond to those of an admixture of methyl and methylene groups. This requires that the molecules be adsorbed on the stainless steel surfaces by attachment of their polar sulfonate heads with their hydrocarbon tails oriented more-or-less normal to the surface. The molecules must be sufficiently close-packed so that the nonyl groups shield the naphthalene rings and t,he polar sulfonate groups from interaction with the spreading liquids. This is similar to the orientation and packing that has been observed for the monolayers adsorbed from oil solutions of many other polar-non-polar materials. Multilayer adsorption of the soap would involve only van der Waals or dispersion force interaction between the soap molecules in solution and the hydrocarbon end-groups of the initially adsorbed soap. The energy available from such an attachment would be considerably less than the energy necessary to release the ionic interaction that holds a soap molecule in the core of a soap micellel’ so that such a second layer could not be in equilibrium with a solution containing micelles. It has been experimentally determined here that, for the soap monolayers isolated from anhydrous oil solution, the critical surface tensions determined using the alkylnaphthalenes are specific to the soap cation. On the other hand, surfaces isolated from water-saturated oil solution have the same critical surface tension regardless of the soap cation, and this common value of ye is lower than the ycvalues obtained for the various monolayers from anhydrous systems. If we interpret differences in yo as indicating differences in packing of the molecules in the adsorbed films we must conclude that the cation
Vol. 65
determines the closeness of packing in monolayers from anhydrous sulfonate solution and that the monolayers obtained from water-saturated oil solutions of all the sulfonates have essentially the same packing and are generally more closely packed than the monolayers obtained from anhydrous oil solution. The effect of water in these experiments is not believed to result from hydrolysis of the sulfonate to give the free acid. Baker and Singleterry’* have shown that the salts of dinonylnaphthalenesulfinic acid are not significantly hvdrolyzed in oil solution in the presence of excess water and carbon dioxide. The effect of water on sulfonate adsorption is controlled by the water content of the solutions and not by the previous history of the metal surface with respect to adsorbed water. This is evidenced by the fact that the experimental result could not be altered by equilibrating the metal specimens with various relative humidities prior to their contact with either water-saturated or anhydrous sulfonate solution. That is, a surface carrying adsorbed water, upon exposure to anhydrous sulfonate solution, develops a monolayer which is indistinguishable from that deposited on an intensively dried surface. It must be concluded either that adsorbed water does not alter the binding of the monolayer or that the anhydrous sulfonate system removes the water from the surface during the establishment of adsorption equilibrium. The latter alternative is more probable. The first water solubilized by sulfonates1ghas a high apparent density and low vapor pressure which suggests it is firmly held in the soap micelles. The amount of water required for a monolayer on all surfaces of the metal button is only 0.01% of that corresponding to one molecule per mole of sulfonate present in the adsorbate solution. In the face of competition with sulfonate molecules for adsorption sites on the metal it may be expected that the partition of water between the surface and the micelles will leave much less than a monolayer of water molecules directly adsorbed on the metal. Thus in the presence of anhydrous solutions, trace amounts.of water introduced during experimental manipulations will have negligible effects on the adsorbed monolayer. Sulfonate soap adsorption from anhydrous systems then must involve adsorbate molecules in which the cation is not hydrated but is more-or-less strongly associated with the anion. Attachment of the adsorbate on the metal-oxide surface could involve specific coordination of the cation with oxygen of the oxide surface or a general dipole interaction between the polar sulfonate group of the soap moIecuIe and the polar metal surface. Consequently, it is not surprising that the closeness of packing of the various sulfonate monolayers from anhydrous oil solution is determined by the soap cation. The tendency to coordinate with surface oxygen would vary with the cation, and the cation size would determine the charge separation and (IS) H. R. Baker and C. R. Singleterry, lnd. En#. Chcm., 48, 1190 (1956).
(17) W. D. Bancom. 8. Kaufman and C. R. Sindeterry, Filth World Petroleum Congrea, Section VI, Paper 18. 195%
(19) J. B. Mathew and E. Hirshhorn, J . Colloid Sei.. 13, 4F5 (1957).
THERMAL FORCES ON AEROSOL PARTICLES
Oct., 1961
thus the dipole strength of the various sulfonate groups. There is no simple correlation between the -yo values for the various monolayers and the probable coiirdinating tendency of the cations or the probable dipole strengths of the various sulfonates. It is possible that the two mechanisms are both operative but to different relative degrees for different soaps. In. water-saturated solution the cation will be hydrated; this will increase the charge separation in the sulfonate ion pair as well aa saturate the cotirdination tendency of the cation. The ion-pair may dissociate sufEciently to permit the sulfonate anion to be adsorbed independently. The data are not adequate, however, to distinguish this possibility from a simple dipole interaction between the hydrated sulfonate ion pair and the oxide surface. It is also possible that water adsorbed on the metal oxide is involved in linking the sulfonate molecules to the surface. The water present on the metal oxide will be in equilibrium with water in the micelles and the exact amount held on the surface will depend upon the ability of water to compete with sulfonate for adsorption sites and the ability of the surface to compete with the micelles for water. Acknowledgments.-Grateful acknowledgment is inade of the contiiderable assistance and encouragement given us b y Dr. W. A. Zisman and the members of his research group, Mrs. E. G. Shafrin, M n . M. K. Bemett and Dr. N. L. Jarvis.
DISCUSSION E. D. GODDARD (Lever Brothers Company) .-The authors express the view that the adsorbed surfactant is in monolayer form and adv,ance an explanation of this view which is baaed on energetic considerations involving the presence of micelles in the solut,ion. Haa adsorption been carried out from solutions where there are no micelles present,, if need be from water-free solutions or from very &lute surfactant solutions? Although it might be that a monolayer is the equilibrium adsorbed species when the solution is in contact with metal, this might well change when the liquid retracts and a situation obtains which is somewhat similar to that in multilayer deposition by the Langmuir-Blodgett technique. W. D. BascoM.---The suggestion is, of course, that above the critical micelle concentration (CMC) the sulfonate soaps are adsorbed a t the air/organic liquid interface and that aa
1689
the adsorbate Solution retracta from the metal oxide Burface
a second monolayer ia deposited. We have not studied the adsorption of theae sulfonates at concentrations below their CMC since this would involve the use of inordinately dilute adsorbate solutions (1 X lo-’ m./l.). However, the presence of any appreciable surface exceaa of solute at the air/
organic liquid interface should be detectable from the difference in surface tension between the solutions and the pure solventa. Measurement of the surface tension of a 1% cop er dinonylnaphthalenesulfonate solution in isopropylbipgenyl using the pendent drop method did not differ from the value obtained for the solvent. A. S. MICHAELS(Massachusetta Institute of Technology). -Is the difference between the “critical surface tension” of an aromatic surface sufliciently M e r e n t from that of an aliphatic surface to justify your conclusions about molecular orientations in these monolayers? W. D. Bascox-The critical surface tension of polyethylene is 4 dynes/cm. less than the value obtained using comparable liquids on polystyrene (Table In. This difference is many times greater than the limiting difference in ye that could be called experimentally significant. A. S. MICRAELS.-HOW do you reconcile your conclusions about molecular packing in these monolayers with L. S. Bartell’s observations on the wettability of incomplete monoIayersi.e., small ditTerences in contact angle often were observed over rather large changes in fractional surface coverage. W. A. ZISMAN(U. S. Naval Resesrch Laboratory).-In L. S. Bartell’s work no change in contact angle waa observed over an ap arent large change in fractional coverage because he used a {quid which was able to fill the rea in the adsorbed film with adlineated hydrocarbon mof%es. Hence, he formed a mixed film and didn’t know that because both solute and solvent molecules in the monolayer were exosing -CHI groups to the contacting liquid sessile drop. is work does not contradict that reported by me or in the present paper. W. D. BASCOM.--L:S. Bartell has reported a more or less proportional and consiatent decrease in eontact angle with monolayer depletion for wetting liquids that would not be expected to “fit” into the spa= left by the removed adsorbate molecules. The wetting liquids employed in our study of the sulfonate soaps could not participate in such quasi-crystalline penetration into the monolayer. HOWARDSHEFFEB(Union College).-Which isomeric dinonylnaphthalenesulfonic acids were used in the studies? Do ou plan to do further work varying the studies of the diazy lnaphthaleneaulfonic acids? W. D. BascoM.--The alkyl groups were on the same ring and the sulfonic acid groups were on the other ring. Thee? producta are difficult to pre are and are a mixture of isomers, so no further work along t i w e linea is planned.
€i
THERMAL FORCES ON AEROSOL PARTICLES’ BY C. F. SCHADTAND R. D. CADLE Stanford Research Institute, Menlo Park, C a l i f m i u Rccuvcd March 7,lWf
A modified Millikan oil drop apparatus has been used to investigate thermal forces on individual aerosol particles of varying &e and widely varying thermal conductivity. The resulta confirm previous indications that thermal forces on particlea of high thermal conductivity are very much larger than predicted by Epetein’s equation. Thermal forcea on particlea whose size ia about that of the mean free path are in fair agreement with predictions from Cawood’s equation, regardleas of thermal conductivity.
Introduction One of the more useful tools for collecting aerosol particles in the 0.1 to 1.0 p size region is the thermal precipitator. Ac’cording to the generally accepted (1) This work wan supported by Grant E 3 0 of the U. 9. Publio Health Service.
radiometer theory of Epstein,f particles having high thermal COnductivities should be Very difficult to collect by this method. However, no such difEculties have been reported in the literature- In a previous study of the efficiency of a thermal pre(2)
P. 9. Epstein, 2.Phyaik, 64,537
(1929).
