Dewetting Effects on Polymer-Coated Surface Acoustic Wave Vapor

Note: In lieu of an abstract, this is the article's first page. Click to increase image size Free first page. View: PDF | PDF w/ Links. Related Conten...
0 downloads 0 Views 2MB Size
Anal. Chem. 1995, 67,4015-4019

Dewetting Effects on Polymer-Coated Surface Acoustic Wave Vapor Sensors Jay W. Q r a t e * g t

Chemistry Division, Naval Research Laboratory, Washington, D.C. 20375-5000

R. Andrew McQill* Geo-Centers, Inc., 10903 Indian Head Highway, Fort Washington, Maryland 20744

Thin polymer films on surface acoustic wave device surfaces sometimes dewet the surface, leading to isolated droplets of material and a degradation in sensor performance. Dewettinghasbeen observed to lead to decreases in baseline operatiag frequencies and loss of oscillation in the worst cases. The influence of surface precleaniug methods has been examined, and, in general, plasma cleaning was found to be the method of choice for the preparation of the device surface for polymer application. Plasma cleaning results in an increase in the surface free energy and improves polymer adhesion so that dewetting is disfavored. For several years, we have been applying thin polymer films to surface acoustic wave (SAW) devices for use as vapor sensor^.^-^ (Reviews on SAW and other acoustic wave sensors can be found in refs 8-13.) Polymers are advantageous in this application because vapors are absorbed reversibly, chemical selectivity can be controlled by varying the polymer‘s chemical structure, and polymers usually form thin adherent films without dficulty. In our early empirical studies,’S6 some polymercoated sensors were ill-behaved for reasons that were not obvious at the time, and such polymers were removed from consideration without further investigation. More recently, selection strategies for sorbent polymers have been developed on the basis of solubility p r ~ p e r t i e s . ~In~ Jsome ~ cases, however, polymers with Present address: Environmental Molecular Sciences Laboratory, Pacific Northwest Laboratory, Battelle Boulevard, Richland, WA 99352. Ballantine, D. S.; Rose, S. L;Grate, J. W.; Wohltjen, H.Ana1. Chem. 1986, 58,3058-3066. Grate, J. W.; Snow, A; Ballantine, D. S.; Wohltjen, H.; Abraham, M. H.; McGill, R A; Sasson, P. Anal. Chem. 1988,60, 869-875. Grate, J. W.; Klusty, M. Anal. Chem. 1991,63, 1719-1727. Grate, J. W.; Klusty, M.; McGiII, R A; Abraham, M. H.; Whiting, G.; Andonian-Haftvan, J. Anal. Chem. 1992,64, 610-624. Grate, J. W.; Rose-Pehrsson, S. L;Venezky, D. L.; Klusty, M.; Wohltjen, H . Anal. Chem. 1993,65, 1868-1881. Rose-Pehrsson, S. L.; Grate, J. W.; Ballantine, D. S.; Jurs, P. C. Anal. Chem. 1988,60,2801-2811. Snow, A W.; Sprague, L. G.; Soulen, R L;Grate, J. W.; Wohltjen, H . J Appl. P01ym. S C ~1991, . 43, 1659-1671. D’Amico, A; Verona, E. Sens. Acfuafors1989,17, 55-66. Frye, G. C.; Martin, S. J. APpl. Specfrosc. Reo. 1991,26, 73-149. Abraham, M. H. Sens. AcfuaforsE 1991,3, 85-111. Grate, J. W.; Nieuwenhuizen, M. S.;Venema, A Sens. Mater. 1989.5,261-300. Grate, J. W.; Martin, S. J.; White, R M. Anal. Chem. 1993,65, 940A948k Grate, J. W.; Martin, S. J.; White, R M. Anal. Chem. 1993,65, 987A+

996A Grate, J. W.; McGill, R A; Abraham, M. H. PYOC. IEEE Ufruson. Symp. 1992,275-279. OW3-2700/95/0367-4015$9.00/0 Q 1995 American Chemical Society

desirable solubility properties for vapor absorption did not yield well-behaved sensors. We have now found that well-behaved sdnsors can be prepared using previously troublesome polymers, provided that sufficient attention is given to factors influencing the wetting and adhesion at the interface between the polymer and sensor surface. Problems occur when the polymer films dewet the sensor surface. Dewetting leads to a number of observable effects on thin-filmmorphology and sensor frequency signals. Wetting, spreading, and adhesion have been extensively studied and revie~ed.15-~1Three parameters that influence spreading and adhesion are the surface free energy of the solid in contact with the vapor phase, the surface free energy of the polymer or liquid in contact with the vapor phase, and the interfacial free energy between the liquid or polymer and the solid, denoted by ys, y ~ and , ~ S L respectively. , A high solid surface energy, ys, is desirable since this favors both spreading and adhesion. Clean metal and metal oxide surfaces have high surface energies. A low interfacial energy, ~ S L is, also desirable, and this is promoted by favorable interactions between the solid and the film material (e.g., van der Waals interactions and hydrogen bonding). The liquid surface tension, YL, has opposing effects on spreading and adhesion. Therefore, in practice it should not be too large or too small.18J9 Although wetting phenomena have been studied extensively for decades, the reverse process, “dewetting”,has received little attention until recently.22-26 Dewetting involves changes in the shape of a thin film that reduce the area of the film/surface interface. Reiter has shown in studies of 5-60 nm thick polystyrene films at temperatures above the static glass-to-rubber transition temperature (Th that dewetting proceeds from the (15) Zisman, W. A In Contact Angle, Wettabilifyand Adhesion; Fowkes, R M., Ed.; ACS Advances in Chemistry Series 4 3 American Chemical Society: Washington, DC. 1964; pp 1-51. (16) Zisman, W. A In Adhesion and Cohesion; Weiss, P. Ed.; Elsevier Publishing Co.: New York, 1962; pp 177-208. (17) de Gennes, P. G. Rev. Mod. Phys. 1985,57, 827-863. (18) Gray, V. R In Aspects of Adhesion; Alner, D. J., Ed.; University of London Press Ltd.: London, 1966, Vol. 2, pp 42-48. (19) Gray, V. R In Aspects of Adhesion; Alner, D. J., Ed.; University of London Press Ltd.: London, 1966;Vol. 3, pp 73-75. (20) Lee, L. H.In Adhesive Bonding; Lee, L H., Ed.; Plenum Press: New York, 1991;pp 1-30. (21) Garret, H. E. In Aspects ofAdhesion;Alner, D. J., Ed.; University of London Press Ltd.: London, 1966; Vol. 2, pp 19-41. (22) Redon. C.; Wyart, B. C.; Rondelez, F. Phys. Rev. Le#. 1991,66,715-718. (23) Reiter, G. Langmuir 1993,9, 1344-1351. (24) Shull, IC R;Karis, T.E.Langmuir 1994,10,334-339. (25) Wyart, F. B.; Daillant, J. Can. /. Pkys. 1990,68,1084-1088. (26) Wyart, R B.;Martin, P.; Redon, C. Langmuir 1993,9, 3682-3690.

Analytical Chemistty, Vol. 67, No. 21, November 1, 1995 4015

formation of pinhole defects in the thin film that increase in number and size with time.23 Eventually, the polymer breaks up into isolated droplets on the surface. We have observed that thii polymer films on sensor surfaces sometimes dewet the surface, leading to isolated droplets of material and a degradation in sensor performance. In this paper we present empirical results on polymers and sensor surfaces where dewetting has been observed and describe the effects of dewetting on sensor signals during fabrication and vapor exposures. We also describe a convenient surface cleaning method, plasma cleaning, that can alleviate these problems. EXPERIMENTAL SECTION

Materials. A solid rubbery poly(isobutylene), a solid rubbery poly(epichlorohydrin), and poly(viny1 propionate) (as a solution in toluene) were obtained from Aldrich. A slowly pourable, viscous liquid poly(epich1orohydrin) resin was obtained from Monomer/Polymer and Dajac Laboratories, Inc. Methylphenyldiphenylsiloxane copolymer (45-55%), which we shall refer to as simply a poly[ @henylmethyl)siloxanel,and poly[bis(cyanopropy1)siloxanel were obtained from Petrarch. Silar lOC, which is also a cyanopropyl-substituted polysiloxane, was obtained from Alltech. Solvents used to clean sensor devices were all HPLC grade. SAW Devices, Electronics, and Vapor Testing. The 158 MHz SAW dual delay line devices, the individual 200 MHz SAW resonators, and the oscillators used were the same as those described in previous s t u d i e ~ . ~The - ~ 158 MHz devices were used primarily for preliminary experiments on surface cleaning and characterization methods. Experiments on polymer coating and vapor testing were done using the 200 MHz SAW resonators unless otherwise specified, and only the polymercoated sampling device was exposed to vapors during testing. In this paper, frequency changes during coating or vapor exposures will always be described in terms of the absolute frequency changes occuning on the individual polymercoated sensor. Spray-coated polymer films were applied using an airbrush supplied with compressed dry nitrogen and a dilute solution of the polymer in HPLC-grade chloroform (Aldrich), exactly as in previous ~ t u d i e s . 3 Spraycoated ,~~~~ films were examined by optical microscopy with a Nikon Optiphot M microscope using reflected light Nomarski differential interference contrast. The films were examined immediately after being applied, before being exposed to vapors, and after exposure to a variety of organic vapors at -15% of saturation at room temperature. The “standard” set of vapors included 2-propanol (17 500 mg/m3), 1-butanol (3770 mg/ m3), nitromethane (16 500 mg/m3), 2-butanone (53 400 mg/m3), isooctane (45 OOO mg/m3), toluene (21 200 mg/m3), dichlorcethane (65 000 mg/m3), and water (3200 mg/m3). Vapor tests were conducted and data collected exactly as described in previous papers.3A27.28 Plasma Cleaning Method. Prior to plasma cleaning, devices were rinsed with chloroform to remove gross contamination. They then were placed in a Hanick plasma cleaner. The chamber was evacuated until the pressure was low enough to sustain a plasma (27) Grate, J. W.; Wenzel, S. W.; White, R M. Anal. Chem. 1991, 63, 15521561. (28) Grate, J. W.; Musty, M. Naval Research Laboratory Memorandum Report 6762; Naval Research Laboratory: Washington, DC, 1990.

4016 Analyfical Chemistfy, Vol. 67, No. 21, November 1, 1995

(-100 mTorr), and the rf power was turned on.B The feed gas was admitted via a needle valve, and its flow was adjusted to produce the brightest plasma. We used dry nitrogen as the feed gas, but the plasma was initially an air plasma because we turned on the power as soon as the pressure was low enough to sustain a plasma, with no effort to purge the system first. Oxygen and air plasmas are more powerful cleaners than nitrogen plasmas because of their oxidizing power. Typical cleaning time was 1520 min. It is probable that much shorter cleaning times would be sufficient, but this has not been systematically investigated. Contact Angle Measurements. Advancing contact angles were determined using triply distilled water, with the last two distillations in an all-quartz still. A platinum wire, cleaned in a flame to produce a red tip, was used to transfer a drop of water to the test surface. Several determinations were made on each surface with a RameHart contact angle goniometer, using water drops of varying sizes. RESULTS AND DISCUSSION

Surface Cleaningand Characterization. The 158 MHz dual delay line SAW devices and the 200 MHz resonator SAW devices used in this study are fabricated with Al transducers and contact pads on =‘ut quartz substrates. The entire surface, except the contact pads, is covered with a thin layer of Si02.30 It is this surface to which the polymer films must adhere. ?srpical cleaning solutions involving strong acids, bases, or oxidizers (e.g., chromic acid in H2S04, mixtures of H2S04 and HzOZ,HF solution, ammonium fluoride, or alcoholic KOH)31cannot be used on these devices because they consume the Al and/or Si02 layers. Therefore, we initially cleaned our surfaces by rinsing in organic solvents such as chloroform, acetone, and methanol.32 Many polymers can be spraycoated onto the solventcleaned devices, and the resulting vapor sensors function in a normal fashion with a stable oscillation frequency. However, problems in certain cases prompted us to find a more effective cleaning method and to examine sensor surfaces more carefully. We found that nitrogen and air plasmas were quite effective for cleaning SAW sensor surfaces without damaging the devices themselves, and we first reported this approach in ref 4. We shall refer to these surfaces as plasmacleaned to distinguish them from surfaces that are only solventcleaned. Water contact angles on our 158 MHz devices were about 60” prior to cleaning and were only slightly reduced upon rinsing in chloroform. Plasma cleaning reduced water contact angles to 0-10”; in fact, these contact angles were difficult to measure because the water spread so well. After a week of storage of the devices in the laboratory, contact angles increased to -30”, indicating that cleaned surfaces can be contaminated by adsorption of material from the air. Therefore, surfaces should preferrably (29) Our Hamck plasma cleaner was a 1974 model which we operated at power level 6. According to the manufacturer, this model corresponds to the current lower-poweredmodel PDC-3XG (25-30 W) operated at a medium rf level setting. (30) The Si02 layer is applied by a sputtering method to a thickness of -45 nm. Horine, B., Sawtek, Orlando, FL. Personal communication. (31) Caution: These aggressive cleaning solutions should be prepared and used only by knowledgeable individuals. (32) Sonication in the organic solvent is potentially more effective, but we were discouraged from using this method because it sometimes caused the epoxy that bonds the 200 MHz devices to their headers to fail, leaving the device only loosely attached to the header by the wire bonds. Similar problems occurred with the 200 MHz devices during attempts to clean them with hot vapor and solvent in a Soxhlet extractor.

Figure 1. Image of a section of a plasma-cleaned 200 MHz SAW resonator to which poly(isobulylene) has been applied by spray coating, yielding a 267 kHz frequency shift. On the section of the sensor shown, the silica overlayer rests on top of an aluminum ground plane. The pattem of the aluminum on one edge of the image gives an indication of scale: each finger is -2pm wide, as is the space in

Figure 2. Image of a section of a plasma-cleaned 200 MHz SAW resonator to which Silar 10C.a poly[bis(cyanopropyl)siloxane], has been applied by spray coating, yielding a 270 kHz frequency shift.

between fingers.

be cleaned just prior to the application of a thin film. The 200 MHz devices exhibited water contact angles of 90-95" as received. Rinsing the as received 200 MHz devices in chloroform increased their operating frequencies by 1-2 Idlz,indicating that a soluble contaminant was present Nevertheless, solvent rinsing reduced the contact angle only slightly, to -75-85O. Plasma cleaning reduced the contact angle to 0-loo. On these narrow devices, the water spread across the entire width, and contact angles could only be estimated from the water edges that advance toward the ends. Contact angles of these plasmacleaned surfaces were not significantly altered by rinsing with HPLGgrade organic solvents. The contact angle measurements clearly demonstrated that the plasmacleaned surfaces differ signiicantly from solventcleaned surfaces in their wetting properties. Dewetting Effects during Polymer F i b Applicafinn. We normally select polymers whose Tgvalues are below the sensor operating temperature in order to obtain rapid vapor diffusion and reversible responses. Polymer materials can be applied by a variety of techniques, including spin casting, Langmuir-Blodgett 0)film deposition (in selected cases), a spraycoating method using an airbrush. The latter method is particularly convenient because the acoustic device can be operated throughout the deposition process. The sensor's signal measures the amount of material applied, providing feedback for the control of the film thickness. Normally, when a SAW device is spraycoated, the oscillator frequency decreases in response to the mass of material deposited on the surface. During pauses in the coating process. the frequency drifts upward slightly as solvent evaporates. The amount of frequency shift downward during spraying is proportional to the time spent sptaying, and this proportionalityis readily discerned. We typically apply polymer until the frequency decreases by -250 Idlz, for reasons discussed in previous papers.3n This corresponds to a filmthickness of -50 nm on a 200 MHz quartz SAW device if the material is evenly distributed. Figures 1-3 provide examples of polymer morphologies on sensor surfaces. Figure 1 illustrates the morphology observed when poly0sobutylene) is applied to a plasmacleaned surface by

Figure 3. Image of a section of a solventcleaned 200 MHz SAW resonator to which Silar 1OC has been applied by spray coating. This SAW resonator was placed next to the device in Figure 2 during spray coating, and therefore has a similar amount of polymer applied.

the spraycoating method. The polymer appears in small circular domains that are numerous and overlapping. This is a typical morphology for a polymer spraycoated to a thickness of -250 kHz on clean surfaces. By contrast F i r e 3 illustrates a siloxane polymer that has a beaded appearance on a solventcleaned sensor surface. The polymer does not wet the surface. The Same polymer is shown on a plasmacleaned surface in Figure 2. In this case, the polymer is present in isolated domains but has not beaded into spheres. when the polymer does not effectively wet the surface and beads into isolated droplets, as shown in Figure 3, two anomalous effects on sensor frequency are observed. F i t the frequency drifts downunrni when spraying is interrupted, even though no more mass is being added. Normally, the frequency is shifted downward by the applied polymer and drifts slightly upward as solvent evaporates. Second, the amount of frequency shift observed for a given amount of spraying becomes less and less as the spraying process proceeds, as if the device is no longer fully sensing the material deposited. Indeed, in the worst cases, it is difficult to cause a 250 Idlz decrease in frequency before the frequency h m e s m t i c or is quenched altogether, even though Analyrical Chemistry. Vol. 67, No. 21, November 1, 1995 4017

subsequent examination under the microscope shows that a considerable amount of polymer is present. These effects were particularly severe when poly[(phenylmethyl)- and poly[bis(cyanopropyl)siloxanesl were applied onto solventcleaned sensors. These polymers are viscous liquid and stiff greasy materials, respectively. On solvent-cleaned 200 MHz devices, we could not make functional vapor sensors from these polymers since oscillation became quite erratic before the device registered a 250 kHz frequency shift. Under the optical microscope, it was clear that the siloxanes had formed isolated round droplets of material on these surfaces (e.g., see Figure 3). When these two polysiloxanes were spraycoated onto plasmacleaned devices, coating proceeded normally without the anomalous downward drift, and functional vapor sensors were obtained. Before we investigated plasma-cleaned sensor surfaces, we made some preliminary efforts to alter the surfaces with silanizing reagents. Treatment of a solventcleaned device with hexamethyldisilazane vapors at room temperature overnight signifcantly influenced the subsequent coating process. (Hexamethyldisilazane converts surface hydroxyl groups to trimethylsiloxygroups.) Using poly[(phenylmethy1)siloxanel as the test polymer, the material could be applied in the normal fashion to the usual 250 kHz thickness. The frequency did drift downward after coating, but this phenomenon was significantly reduced relative to that observed with the solvent-cleaned devices. Another solventcleaned device was treated with a freshly prepared solution made by adding dichloromethylphenylsilane to methanol. In this case, coating proceeded normally (without downward drift). Although the efficiencies of the above silanization methods on the solventcleaned surfaces are uncertain, the results demonstrate that surface modfication can influence the coating deposition process in d a c u l t cases and that the behaviors we have described are related to the surface characteristics. Since these preliminary experiments, we have developed silanization procedures for SAW devices that block surface hydroxyls and improve polymer ~etting.3~33~ A less severe example of dewetting behavior involved a poly(epichlorohydrin) resin, a viscous liquid with physical characteristics similar to those of the siloxanes. It could be easily coated onto a solvent-cleaned device to a thickness of 300 kHz. Nevertheless, slight downward frequency drift observed after pauses in the coating process suggested that a dewetting problem might exist. Results consistent with this suspicion were obtained in subsequent vapor tests (see below). The application of this same resin to a plasmacleaned device proceeded without anomalous frequency drift. Additional evidence for improved film stability on plasmacleaned surfaces can be seen by observing polymers on solventand plasma-cleaned surfaces at elevated temperatures. For example, poly(viny1 propionate) was solvent cast into continuous thin films on solventcleaned and plasmacleaned 158 MHz devices. These were heated to 85 “Cin a convection oven, and the film on the solvent-cleaned device broke up into isolated droplets within 15 min. However, the film on the plasmacleaned device remained continuous, even after 24 h at the elevated temperature. Similar effects of surface cleaning have been observed on flexural plate (33) McGill, R A,; Grate, J. W.; Anderson, M. R In Inte&ciul Design and Chemical Sensing; Mallouk, T. E., Hamson, D. J., Eds.; ACS Symposium Series 561; American Chemical Society: Washington, DC, 1994; pp 280294. (34) McGill, R. A,; Grate, J. W., manuscript in preparation.

4018 Analytical Chemistry, Vol. 67, No. 27, November 1, 1995

wave devices.35 After initial experiments where poly(viny1 propionate) dewet the FPW device surface, the device was cleaned with an HzOz/HzS04 solution.31The same polymer did not dewet this cleaned surface, even after repeated heating and cooling cycles. DeweUing Effects during Vapor Exposures. We have observed that sensors exhibiting the anomalous effects noted above during film application sometimes cease to oscillate during exposures to organic vapors. This cannot be demonstrated in the severe cases where the sensor does not function at all, but it can be observed in “borderline”cases. For example, when a dichloromethylphenylsilane-treated sensor coated with poly[ (phenylmethyl) siloxane] was exposed to vapors, it stopped oscillating. Reexposure to clean carrier gas restored oscillation, but the baseline was shifted to a lower frequency. This process was repeated many times with a variety of vapors (see Experimental Section). The shifts in baseline frequency after vapor exposures were in the same direction as the downward drift observed after the siloxanes were coated onto solvent-cleaned devices, where beading was observed. Oscillation difficulties and baseline frequency shifts were not a problem when this polymer was coated onto a plasmacleaned sensor. The sensor described above with poly(epich1orohydrin) resin deposited on a solventcleaned surface produced normal responses to most vapors at the test concentrations,but two (dichloroethane and butanone) caused it to stop oscillating. These problems did not occur when the same resin was tested on a plasmacleaned sensor. [A solid poly(epich1orohydrin) polymer we have used in past studies coats normally and responds to vapors without loss of oscillation using either solvent- or plasma-cleaned devices.] Discussion. To summarize, the following effects are associated with poor wetting and adhesion of the polymer: the material visibly beads into droplets (as observed under the optical microscope) after coating or after vapor exposures; the frequency drifts downward during pauses in spray coating and after stopping spray coating; spray-coated polymer may not produce consistent frequency decreases throughout the coating process, and in the worst cases, one may not be able to obtain a functional sensor because of loss of oscillation. Vapor exposures may quench the oscillation, and the baseline may be shifted after the vapor is removed. (The latter effect is observable in borderline cases.) Plasma cleaning sensor surfaces prior to film application can prevent these problems from arising; we have obtained wellbehaved vapor sensors from a great variety of polymers by spray coating them onto plasma-cleaned 200 MHz SAW resonators. It is presumably the high surface energy of the plasmacleaned device that favors wetting, although it is also likely that different polymer/surface interfacial free energies result when polymers are applied to plasmacleaned versus solventcleaned surfaces.36 In our experience, these effects are more likely to be observed if the polymer is a viscous liquid than if it is a solid rubber and more likely to be seen after vapor exposures than without such exposures. Presumably, lower viscosity facilitates more rapid dewetting. The physical appearance of the coated material may change with time and/or vapor exposures, depending on the viscosity and wetting properties of the polymer. Beading is observed if the polymer dewets the surface, but in other cases, (35) Grate, J. W.; Wenzel, S. W.; White, R. M. Anal. Chen. 1992,64,413-423. (36) One reviewer noted that plasma cleaning with Harrick plasma cleaners can

change surface roughness and that this might also influence wetting properties.

sharper features of the overlapping circle morphology ( T i i r e 1) soften, and in some cases polymer domains spread. The latter observation indicates that the surface forces favor wetting rather than dewetting, and a stable film and well-behaved sensor can be expected. When dewetting occurs, it indicates that the coated polymer film was not in a thermodynamically stable state. ReiteS3 investigated continuous thii polymer films where dewetting was initiated by the formation of cylindrical holes whose diameters increase with time. Our sprayed-on films are not necessarily continuous and may have holes from which dewetting may proceed. Nevertheless, Reiter's studies showed that dewetting of thii films can occur regardless of whether the film has defects to begin with or not, and our own experiments with an apparently continuous thii film of poly(viny1 propionate) confirm this. Therefore, applying a polymer film by a method that gives a continuous thin film will not necessarily prevent dewetting. In this regard, the initial film morphology is probably less important than the balance of forces that determine its equilibrium morphology. Further, we believe that it is the wetting and adhesion of the polymer on the surface, rather than simply the shape of the deposited polymer domains, that influence sensor performance. In previous studies, poly(viny1tetradecanal)coatedSAW sensors prepared with continuous thin films applied by the LB method were compared with films applied by the spraycoating method, yielding the overlapping circle morphology described above (Figure l),and these responded similarly to organic vapors3 The possibility of dewetting has a number of consequences for polymercoated SAW sensor development. The SAW devices we used are purchased and used in many other laboratories, and it is common practice for polymer films to be applied and used with no microscopic observations of film morphology or behavior. The possibility that unobserved polymer dewetting processes may

innuence sensor behavior could lead to irreproducible results or erroneous interpretations and conclusions. It has also been our observation that dewetting problems have become more common as our research has progressed to higher frequency SAW devices with thinner films? S i c e this is the general direction for SAW sensor development, attention to polymer film wetting and adhesion is likely to become increasingly important. We wish to emphasize that physically adherent films with long lifetimes can be easily made provided attention is paid to the cleaning and surface preparation of the sensor prior to coating. In previous studies, we have described polymercoated SAW vapor sensors that were repeatedly exposed to vapors over periods of months.2*5 In our experimental work, we routinely place a fluoropolyolcoated 158 MHz dual delay line SAW sensor in series aiter our test sensors whenever we test new sensors with our vapor generation sy~tem.~8~,28 The responses of this control sensor are used to confirm that the system is generating vapor properly. We have used the same sensor in this application for -5 years and countless vapor exposures with consistent performance and no problems with polymer dewetting. ACKNOWLEWMENT We acknowledge Mark Klusty for careful observations during several of the initial experiments that led to this study; William R Barger and Arthur Snow for helpful discussions on wetting, adhesion, and surface characterization; and Richard Colton for suggesting the plasmacleaning method and providing the a p paratus. This work was supported by the Office of Naval Technology/Naval Surface Warfare Center, Dahlgren, VA. Received for review March 15, 1995. Accepted August 9, 1995.m AC950262X @Abstractpublished in Advance ACS Abstracts, September 15, 1995.

Analytical Chemistry, Vol. 67, No. 21, November 1, 7995

4019