Wetting phenomena of binary liquid mixtures on chemically altered substrates.

We report measurements of the state of wetting of two liquid mixtures at coexistence near their respective critical consolute temperatures. Borosilicate glass capillary tubes were reacted with hexamethyldisilazane to produce substrates of uniform and controlled silylation. Surfaces of low coverage exhibit a series of first-order partial to complete wetting transitions and obey a short-range force scaling relation. Surfaces of high coverage yield surprising results which may be understood as a consequence of long-range forces. Disciplines Physical Sciences and Mathematics | Physics Comments At the time of publication, author Douglas J. Durian was affiliated with Cornell University. Currently, he is a faculty member at the Physics Department at the University of Pennsylvania. This journal article is available at ScholarlyCommons: https://repository.upenn.edu/physics_papers/639 VOLUME 59, NUMBER 5 PHYSICAL REVIEW LETTERS 3 AUGUST 1987 Wetting Phenomena of Binary Liquid Mixtures on Chemically Altered Substrates Douglas J. Durian and Carl Franck Laboratory of Atomic and Solid State Physics and Materials Science Center, Cornell University, Ithaca, New York I4853 (Received 17 March 1987) We report measurements of the state of wetting of two liquid mixtures at coexistence near their respective critical consolute temperatures. Borosilicate glass capillary tubes were reacted with hexamethyldisilazane to produce substrates of uniform and controlled silylation. Surfaces of low coverage exhibit a series of first-order partial to complete wetting transitions and obey a short-range force scaling relation. Surfaces of high coverage yield surprising results which may be understood as a consequence of long-range forces. PACS rlumbers: 68.10.Cr, 64.60.Fr, 64.70.Ja, 68.45.—v Close to a bulk critical point, the structure of a liquid is especially sensitive to the presence of a physical boundary through solid-liquid interactions and altered liquid-liquid coupling. Long-range, power-law forces of suSciently large exponent are irrelevant to bulk as well as certain surface critical phenomena. ' Therefore, in their investigation of wetting near a bulk critical point, Nakanishi and Fisher make the natural approximation of short-range, contact interactions only. Later workers argue that wetting is not necessarily by bulk criticality and that long-range forces cannot be neglected in real systems. Most recently, Ebner and Saam (ES) provide examples of wetting phenomena in systems near bulk criticality with both contact and power-law substrateadsorbate potentials. Their wetting phase diagrams share many features with those of Nakanishi and Fisher, but only for the case of weak long-range forces. In this Letter we present experimental results consistent with theoretical predictions for the case of shortrange plus weak long-range forces. In another liquid mixture, previously unpredicted behavior is observed which we propose may be a consequence of strong longrange forces. Our method is capillary rise in which we continuously tune the solid-liquid interaction via surface chemistry. The method and sample cells used are described by Abeysuriya, Wu, and Franck (AWF). Fluids are at liquid-vapor as well as liquid-liquid coexistence. Borosilicate capillary tubes of radii 0.134 and 0.188 mm are cleaned by sonocation with acetone, methanol, ethylenediamine tetra-acetic acid, distilled H20, nitric acid, and distilled H20 followed by vacuum baking at 120 C. The tubes are then transferred to an Ar atmosphere and subsequently handled such that they are never exposed to air. This procedure is intended to yield clean, maximally hydroxylated surfaces, free of physically adsorbed water; i.e. , with polar silanol groups, Si,OH, on the surface (subscript s). The variation of the surface field, hi, is accomplished by our partially reacting the clean hydroxylated surface with the vapor of hexamethyldisilazane (HMDS), NH[Si(CH3)3]2, according to 2[Si,OH]+ HMDS~ 2[Si,OSi(CH3) 3]+NH3. The coverage, or degree of silylation, can be controlled by the pressure of HMDS vapor or time of reaction. Presumably the heavier the silylation, the less attractive to polar fluids is the surface. A useful property of H M DS is that it scavenges physically adsorbed water. So the above reaction goes forward, albeit more slowly, in the presence of water. We find it most convenient to clean many tubes together and react them individually at 2.4 mm Hg from 1 to 500 h. Zettlemoyer and Hsing show that this is slightly below the pressure which yields monolayer coverage. Our procedure gives monotonically increasing coverage with reaction time for tubes with otherwise identical histories. Thus, our substrates are improved on those of AWF in several ways: The silylation of each capillary is uniform over its length, is well defined . chemically (no polymerization, coverage less than monolayer), and can be widely varied. However, we have not measured the actual silyl coverage resulting from our reactions. Furthermore we do not have an understanding of the solid-liquid forces necessary to predict h] as a function of this coverage. Rather, we characterize our substrates by their wetting properties. Each sample cell contains one or two hydroxylated tubes of diferent radii and up to four silylated tubes of identical radii. Capillary parameter, a =rise & radius, should be constant for identical substrates if there is no systematic error in locating the bulk meniscus. Such an error does exist on account of curvature of the bulk meniscus at the sample-cell wall and around the individual tubes. A correction requiring that hydroxylated-tube capillary parameters be identical is not used because it increases the systematic error without altering the power-law fits to be discussed. Carbon disuiftde+nitromethane (CS2 +NMe) The.— four sample cells had bulk critical temperatures in the range T, =63.052+ 0.011'C (determined visually). Receding-meniscus capillary parameters are plotted 1987 The American Physical Society 555 VOLUME 59, NUMBER 5 PHYSICAL REVIEW LETTERS 3 AUGUST 1987 versus temperature in Fig. 1 after being averaged over diferent positions of the tubes through the bulk meniscus. The hydroxylated tubes are completely wetted by the upper, nitromethane-rich (A*) phase. They exhibit capillary depression with predicted temperature dependence a = —ao I r I" ~; ao is the capillary constant and r =(T—T, )/T, is the reduced temperature with T and T, in kelvins. This is based on the well-known capillary-rise formula a ' = (2cr.p/apg)cosO, where cr, & —cro I r I" is the liquid-liquid interfacial surface tension, hp —I r I ~ is the liquid-liquid mass-density difference, g is gravitational acceleration, and 0 is the contact angle. Note the sign convention, cosO= —1 for complete wetting by N*. We find ao =48.3 ~0.5 mm and p —P=0.94+ 0.01 (cf. theoretical value of 0.935 + 0.015). This fit is shown in Fig. 1 for complete wetting by either phase; between these extremes lie states of partial wetting. Lightly silylated tubes exhibit capillary depression and a transition between partial and complete wetting by N* at a particular T (( T, ) for each capillary. Since the hydroxylated tubes are completely wetted, we use the capillary rise formula to compute cosO=a„.l„i„,q/ ahydroxylated for each silylated substrate. The predicted temperature dependence is cosO —I t I ' " far below T„. This follows from the short-range force (SRF) scaling relation" 2 —a, hi o,~cosO= cro Ir I with Pl =2 —a, —Al being the surface order-parameter exponent. We adopt a scale for hi such that the argument of F is 1 at the transition point. The small-x limit in Eq. (1) is F(x) =x because F must be both analytic around x =0 and odd by symmetry. To test this prediction, the wetting transition temperatures were estimated, then used in the making of Fig. 2. Most data fall within the plotted lines giving the result Pl —p = —0.44+ 0.06. This agrees with the recent experimental'' and theoretical ' values of —0.34+ 0.05 and —0.44 respectively. Note that tubes with a wide range of ~etting transition temperatures, I T„—T, I =0. 1 —10 K, are described by one function. In retrospect we argue that because t =hl ' and 2 —a, =p, Fig. 2 is actually a scaling plot ~/~, of cosO vs I t I /h t '. Therefore we have experimentally measured the scaling function to be F(x) =x for x ( 1 and 1 for x ~ 1. The data show no systematic deviation from this form even near the wetting transition, x=1, where the linear expansion of Eq. (1) might fail. In particular, we emphasize that the transition appears to be first order because of the discontinuity in slope' at x =1. Since our method directly compares partially wetted with completely wetted tubes, we have a much clearer view of the transition than do other capillary-rise experiments. "' Heavily silylated tubes exhibit capillary rise and no clear transition from partial to complete wetting by C* below T„ these data are not shown. AWF report this behavior and argue that short-range forces favorable to C* and long-range forces favorable to N* cause partial wetting by C* but forbid complete wetting at all T ( T, . If this is true, then SRF scaling, Eq. (1), can no longer hold. Quantitative predictions are difficult to check experimentally because mislocation of the bulk meniscus and breakdown of the capillary rise formula near T, lead to unacceptably large systematic errors in the computa-