Langmuir 1991, 7, 2236-2242
2236
Moisture Absorption Characteristics of Organosiloxane Self-AssembledMonolayers David L. Angst* AT&T Bell Laboratories, Allentown, Pennsylvania 18103 Gary W.Simmons Zettlemoyer Center for Surface Studies and Department of Chemistry, Lehigh University, Bethlehem, Pennsylvania 18105 Received November 13,1990. In Final Form: March 11, 1991
Monolayer films derived from octadecyltrichlorosilane (OTS) and dimethyloctadecylchlorosilane (DMODCS) were prepared on silicon dioxide surfaces under molecular self-assembly conditions. The hydration state of the thermal oxide surface was found to greatly influence the degree of surface coverage and packing of the monolayers. For OTS on hydrated thermal oxide, high-quality, close-packed monolayers were formed, as evident from contact angle, ellipsometry, and ATR-IR data. For OTS on dry oxide and for DMODCS on both dry and hydrated oxide surfaces, the monolayer coverage wm significantly less. Spectroscopic adsorption isotherms for water on the OTS, DMODCS, and oxide surfaces were measured with a specially constructed attenuated total reflectance infrared (ATR-IR)vapor phase exposure test cell. This IR cell permits the detection of submonolayer amounts of water at low surface area oxide interfaces for the first time. Water adsorption on the unsilanized oxide was increased by prehydration of the surface. For the uncured silanized OTS surfaces, moisture adsorption to the surface was actually increased relative to either dry or partially hydrated unsilanized oxide surfaces, even though the outer surface was highly hydrophobic. Curing at 150 "C of the tightly packed OTS monolayer was found to significantly decrease moisture adsorption. In contrast to the behavior observed for OTS, DMODCS treatment lowered the amount of moisture adsorbed to the surface even in the uncured state. These results suggest that water absorbed by the organosiloxane monolayers is bound to hydroxyl groups in the interfacial region.
Introduction There is considerable technological interest in protecting microelectronic devices against environmentally induced failures. Plastic packaging of devices is a cost-effective option, but permeability to moisture can lead to detrimental effects on device performancelbecause of corrosion of conductors (i.e. metalization). The presence of water at the interface between the device and the organic coating can also lead to the loss of adhesion of the protective coatings2 with subsequent acceleration of the corrosion reactions leading to device failure. A specific example, reported in the literature, has shown that the corrosion of aluminum metalization in integrated circuit devices is preceded by the formation of a thin film of water at the silicon chip-epoxy interface.g-5 This thin film of water provides a conducting pathway for the movement of ionic species and corrosion products between the voltage-biased device metalization runners and leads. The existence of a thin film of water at the chip-epoxy interface in corroded integrated circuit packages has been confirmed by electrical and physical methods.'" For plastic packaging of integrated circuits to be a viable option, it is therefore necessary to use organic materials and methods of application that result in strong adhesive bonds at the interface. In addition, the bond between the coating and the substrate must be resistant to reaction with water and be effective in blocking sites on the substrate against adsorption of water. Organosilanes have been used effectively as coupling (1)Merrett, R. P.; Bryant,J. P.; Studd, R. Ret. Phys. Symp. h o c . 1983,21,73. (2)Nakagawa, 0.; Sasaki, I.; Hamamura, H.; Banjo,T. J. Electron. Mater. 1984,13 (2).231. (3)Quach, A.;'Hunter,W. L. J. Electron. Mater. 1977,6(3),319. (4) Ueell, R. J.; Smiley, S. A. Rel. Phys. Symp. h o c . 1981,19, 65. (5) Teuboeki,K.;Wakashma,Y.; Nagasima, N. Rel. Phys. Symp.hoc.
1983,21,83.
0743-7463f 91f 2407-2236$O2.50/0
agents to achieve the desirable characteristics at the substrate surface while providing a means for bonding with an organic polymer coating.6 These organosilanes have hydrolyzable functional groups on the silicon atom which permit covalent bond formation to the substrate as well as cross-linking within the organosiloxane layer. At the end of the alkyl chain, reactive functionalities are present which covalently bond to unreacted groups in the outer organic coating. Some integrated circuit packaging applications have included the use of organosilane coupling agents within the epoxy coating or at the chip-epoxy i n t e r f a ~ e .Other ~ applications of organosilane coupling agents have included glass fiber/ polymer composites' and epoxy laminate^.^ Organosilanes have also been used to modify surface hydrophobicity? to achieve biocompatibility of surfaces,10 to incorporate aromatic functional groups in monolayer films for eventual optical and electronic uses," and to improve boundary lubrication.12 A variety of tests have been used to evauate the performance of organic coatings under simulated environmental conditions; these tests include the effects of water on pull, scratch, and peel measurements of adhesion! the "pressure cooker" and other accelerated testa of the performance of integrated circuits,' and studies of the stability of chromatographic bonded phased3 during (6)Pleuddemann, E.P.Silane Coupling Agents; Plenum Press: New York, 1982. (7) Culler, S. R.; Ishida, H.;Koenig, J. L. J. ColloidlnterfoceSci. 1986, I O 9 (l),1. (8)Pleuddemann, E.P. J. Adhers. Sci. Technol. 1988,2(31,179. (9)Hertl, W.; Hair, M. L. J. Phys. Chem. 1971,75,2181. (10)Regen, S.L.;Kirszeneztejn, P.; Singh, A. Macromolecules 1983,
--.(11)Tillman, N.; Ulmnn, A.; Schildkraut, J. S.;Penner, T. L. J. Am.
16. 335.
Chem. SOC.1988,110,6136. (12)Tillman, N.; DePalma, V. Langmuir 1989,5,868. (13)Bogart, G.R.;Leyden, D. E.; Wade, T. M.; Shafer,W.; Carr, P. W. J. Chromutogr. 1989,483,209.
0 1991 American Chemical Society
Moisture Absorption of Organosiloxane Monolayers
solvent exposure. Although performance tests can assess the macroscopic behavior of a particular interface, they do not address the fundamental molecular interactions involved. Experimental procedures have recently been developed which allow the precise molecular engineering of the interfacial region.l4J6 The technique of molecular self-assembly, in which well-defined monolayer and multilayer films are spontaneously deposited on a substrate from solution, has been developed by several research gr0ups.10*11114116 In combinationwith analytical techniques such as infrared and nuclear magnetic resonance spectroscopies, ellipsometric measurements of film thickness, wettability studies, and X-ray diffraction,the self-assembly technique has been used by these groups to study the experimental variables which determine the formation and structure of these films. We have used the method of molecular self-assemblyto form monolayers and partial monolayers from two different organosilanes (octadecyltrichlorosilane, OTS, and dimethyloctadecylchlorosilane, DMODCS) on the surface of oxide films on silicon. This substrate was chosen because a passivating film of silicon oxide constitutes a significant fraction of many chip surfaces. Of particular interest was (i) the relationship between the degree of hydration of the silicon oxide surface, the structure of the organosilane molecule, and the coverage by the organosilane and (ii) the relationship between the coverage and structure of the organosilane films and the amount of water absorption. Infrared spectroscopy, contact angle measurements, and ellipsometry were used to characterize the degree of hydration of the substrate and to determine the coverage and structure of the organosilane layer. The moisture absorption characteristics of these films were then measured by using infrared spectroscopy. The surface chemistry of silica, particularly high surface areasilica, has been studied extensively and we expect the surface chemistry of the thermally grown silicon oxide films used in our studies to be similar. Since surface hydroxyl groups are most likely the reactive sites with the organosilanes (OTS and DMODCS), we present a review of the chemistry of hydrated silica surfaces. This earlier work provides a basis for the interpretation of the infrared spectra of the silicon oxide surfaces that were used as substrates for the preparation of organosilane films. We also give a brief description of the structure and chemistry of the two organosilanes used in these studies. Infrared Absorbances of Hydroxylson Silica. The infrared absorbances of SiOH groups and water on high surface area silica (powders) have been extensively studied. An excellent review of the experimental results in this area has been previously published.l6 The hydration/ dehydration chemistry of the silica surface is summarized in Figure 1, which also lists the infrared absorbance frequencies of the reactants and products. The silica surface can be dehydrated and rehydrated reversibly until a temperature of about 400 O C is reached, after which rehydration becomes extremely slow. Thus, the thermal history of the sample influences the observed infrared spectrum of silica. Organosilane Chemistry. The two organosilanes chosen for our initial studies were octadecyltrichlorosilane (OTS) and dimethyloctadecylchlorosilane(DMODCS). The structures of these two organosilanes are given in Figure 2, along with the reactions of interest. Both (14) Maoz, R.;Sagiv, J. J. Colloid Interface Sei. 1984, 100 (2), 465. (16) Allnra, D. L.;Nuzzo, R. G. Langmuir 198S, 1,52. (16) Hair, M. L.Infrared Spectroscopy in Surface Chemistry;Marcel Dekker: New York, 19137.
Langmuir, Vol. 7, No. 10, 1991 2237 H
I
0
Isolated, 3747 t 20 cm1
Hydrogen-Bonded, 3WOt90Cm-1
Physlsorbed Water, 3400 t 200 cm.1 (Water) 3520 t 200 cm' (Hydroxyl)
Figure 1. Hydration and dehydration reactions of the silica surface and the IR absorption frequenciesof the surface species. Chemisorption of water produces surface silanols, which serve as adsorption sites for water. Octadecyltrlchlorosilane
Dlmethyloctadecylchlororilane (DMODCS) cp,
cll,
(OW
(CV2 )17
(Cf2 117 CH, -Si-CHa
ci -SI -CI
dl
CI )+3H20
R
I
+1 H20
R
t3HCI
HO-Si-OH
CH, -Si-CH,
I OH
Isi02koH
Reaction with Surface
Si02
F
,
Reaction with Surface
OH
I
0 -?-R
OH
Curing
+ HCI
I OH
I
+ H20 LO-!-"
SiOz ?'?I
-R
+H,O
lSio2FoH
F
743
0 -7-R CH3
I
0 -SI-R
Figure 2. Reactions of OTS and DMODCS. Hydrolysis of the chloride group by trace amounts of water in solution to silanol is followed by condensation with surface silanols, resulting in covalent bond formation between the monolayer and the sub strate. OTS molecules can also cross-link to form polymeric species during film curing.
molecules contain at least one chlorine atom, which can be hydrolyzed by trace water in solution to form the silanol group. The primary structural differences between these two molecules is the replacement of two of the three hydrolyzable groups in OTS with methyl groups, to give the DMODCS structure. The DMODCS molecule cannot be cross-linked to form a polymeric organosiloxane, as is possible with OTS (in the presence of trace water which catalyzes the cross-linking reaction). There is also more steric hindrance at the silicon terminus in DMODCS, since the methyl groups are larger than either the chlorine atom or the hydroxyl group. The structural differences in these molecules provided for possible contrasts in the coverage, cross-linking, and structure of the resulting organosilane films. The relationship between these properties of the monolayers and the absorption of water was measured spectroscopically. Experimental Section Organosilanemonolayers were depositedonto thin Si02 layers which were thermally grown on single-crystal silicon substrates. Infrared spectroscopy, ellipsometry, and contact angle measurementa were used to characterizethe oxide surface and the surface coverage, molecular order, and surface energetics of the organosilane monolayers. The moisture absorption properties of the monolayers were determined by infrared spectroscopy as a function of in situ exposuresto water vapor. Details of the sample
Angst and Simmons
2238 Langmuir, Vol. 7, No. 10,1991 preparation and characterization techniques are presented in the following sections. Materials. Both silicon wafers and silicon ATR prisms were used as substrates. Siliconwafers obtained from Wacker Chemie were p-doped, 5-20 fl cm resistivity, highly polished integrated circuit material. The attenuated total reflectance (ATR)elements were 50 X 20 X 1mm trapezoidal prisms from Harrick Scientific. Hydrochloric acid, ammonium hydroxide, and hydrogen peroxide from Ashland Chemical (SEMI grade) were used in substrate cleaning. Deionized (DI) water of greater than 18 fl cm resistivity was obtained by passing the output of a reverse osmosispurifier through a Millipore-Qsystem of cation and anion exchange columns, followed by an organic adsorbent column. The organosilane solutions were made up in a mixture of carbon tetrachloride (CC4) and Isopar G (20% CCl, by volume). Isopar G, a branched hydrocarbon solvent, was obtained from Exxon Corp. and was passed through a column of basic alumina and filtered prior to use. Octadecyltrichlorosilane (OTS)and dimethyloctadecylchlorosilane(DMODCS)were obtained from Petrarch Systems. OTS was vacuum distilled prior to use, while DMODCS, a waxy solid, was used as received. Substrate Preparation. Wafers and ATR prisms were cleaned by using a procedure based on the "RCA clean" used in integrated circuit manufacturing." The cleaned substrates were oxidized by using a Tamarack 180-M rapid thermal annealing (RTA) unit. This oxidation step is useful in producing an oxide surface which is very clean and largely dehydroxylated. The oxidation conditions were 1150 "C for 5 min in pure oxygen. The oxides grown in this manner were typically 150-180 A thick. The oxidized substrates were then immediately analyzed as described below or were hydrated by immersion in DI water for 12 h. Self-Assembly of Monolayers. Solutions of the desired organosilane in 20 % CCl,/Isopar G were prepared 30 min prior to use. Isopar G was allowed to equilibrate overnight with a trace of water (2 drops per 100 mL) and decanted before use. The organosilane was mixed in 20% CCl,/Isopar G to give a concentration of 5 X M. The substrates were then immersed in the silanizing solution for 30 min, and rinsed 10 times each in CCl,, ethanol, and CC4again. This method of cleaningthe monolayers was found to give identical IR and ellipsometry results as did detergent scrubbing (see ref 11for a discussion of monolayer cleaning). For the OTS monolayers, no residual chlorine (from the organosilane or the CCl, rinse) was detected by X-ray photoelectron spectroscopy (PHI Model 550 ESCA and Scienta 300 ESCA). Since detergent scrubbinginvolves exposureof the monolayer to water, we avoided this treatment and relied upon repeated solvent rinsing to clean the samples. Ellipsometric Measurements. Film thickness measurements by ellipsometry provided a means of determining the amount of organosilaneadsorbed on the silica surfaces. A Gaertner L115B ellipsometer, operated at an incident angle of 70" and using a He-Ne laser, was used for thickness measurements. The thickness of the oxide was measured for each sample and then subtracted from the total thickness of the oxide plus organosilane monolayer to yield the monolayer thickness. Thirty-two points across each sample were measured and then averaged. Measurements for each sample were repeated after repositioning the sample in the ellipsometer. The average values for the oxide and monolayer thicknesses found after repositioning were always within 1 2 A of the values determined in the initial measurement. Calculations of organosilane film thicknesses were made assuming a refractive index for both the organosilane and the oxide of 1.462. This value is well established for the thermal oxide of silicon18and appears to be an appropriate value for the organosilane,based on literature values for the OTS film refractive index of between 1.45 and 1.50.11J4 Over this range of refractive indexes, the choice of the refractive index value causes the measured monolayer thickness to vary by no more than 1A.19 We estimate that the overall uncertainty in film thicknesses measured in the above manner is *3 A. (17) Kern, W. J. Electrochem. SOC.1990, 137 (6), 1887. (18) Archer, R. J. J . Opt. SOC.Am. 1962,52 (9), 970. (19) Wasserman,S. R.; Whitesides, G. M.; Tidswell,I. M.; Ocko, B. M.; Pershan, P. S.; Axe, J. D. J. Am. Chem. SOC.1989,111, 5852.
TOP VIEW
ENDiEW
Figure 3. ATR-IR vapor exposure cell. The silicon ATR substrate is clamped between the two halves of the flow cell. (1) is the IR beam entrance face into the oxidized silicon substrate (2). (3) is a Teflon gasket which seals the stainless steel body (4) to the ATR element. (5) is one of the four gas inlets/outlets, while (6) represents the organosilane-silica interface. Contact Angle Measurements. The wetting behavior of an organosilane film indicates the degree of molecular ordering present.ll Advancing contact angles were measured in this study by using a Rame-Hart NRL goniometer in the laboratory atmosphere. Between 3 and 5 drops were measured and the values averaged. Contact angle values were reproducible to 1 2 O for values less than 90° and * 3 O for larger angles. Critical surface tensions for the monolayers were determined by extrapolation of the data for the n-alkane test liquids, using a linear regression analysis, to cos 0 = 1.00, as described by Zismanm FTIR-ATR Spectroscopy. The infrared absorption peak positions, bandwidths, and dichroic ratios are useful in characterizing the molecular ordering of the monolayer films. The band intensities, in combination with the ellipsometry measurements, reveal the degree of surface coverage of the organosilane. Infrared absorption data was obtained on a Perkin-Elmer Model 1750 Fourier transform spectrometer, operated at 4-cm-' resolution. Spectra of the monolayers were obtained by subtracting the spectrum of the uncoated, oxidized ATR substrate from the spectrum of the coated substrate (1089 scans). IR dichroic ratios were determined by recording spectra with s and p polarized incident beams.l1P2l Polarization was accomplished by using a germanium polarizer (Harrick Scientific). Water Absorption Isotherms. Infrared spectroscopy was used to measure the relative amounts of water absorbed into the organosilane films a t partial pressures of water from 0 to 20 Torr (0% to 85% relative humidity a t room temperature). These measurements revealed important details regarding the effect of the organosilane structure and the oxide hydration state on the adsorption of moisture. The water vapor exposure cell, a modified version of liquid is depicted schematexposure cells described in the ically in Figure 3. Approximately half of the area of the sample was exposed to the gas stream flowing through the cell, while the remaining surface was exposed to the dry nitrogen purge of the instrument sample compartment. The gas flow apparatus for controllingthe water vapor pressure is depicted in Figure 4. The inlet side of the IR cell was connected to a general purpose humidifier (GPH) similar to the NBS Mark I1 design.23 Moisture levels between 0.4 and 20 Torr were generated via adjustment of the inlet nitrogen pressure. The moisture content of the gas stream at the outlet of the IR cell was measured with a General Eastern Model 1200APS dew point hygrometer (DPH). A Teledyne Model 310 trace oxygen analyzer (0.1 ppm sensitivity) was placed a t the outlet of the DPH to allow for continuous leak checking during operation. A five-way ball valve a t the inlet to the IR cell allowed rapid switching (20) Zisman, W. A. Adv. Chem. 1964,43,1. (21) Harrick,N. J. ATR Spectroscopy;Plenum Press: New York, 1973. (22) (a) Brauner, J. W.; Mendelsohn, R. Appl. Spectrosc. 1988,42 (l), 188. (b) Parry, D. B.; Harris, J. M. Appl. Spectrosc. 1988, 42 (6), 997. (23) Hasegawa, S.;Little, J. W. J . Res. Natl. Bur. Stand.,Sect. A 1977, 81A (l),81.
Moisture Absorption of Organosiloxane Monolayers TO VENT
IA I I
I
Langmuir, Vol. 7, No. 10, 1991 2239 H
N2
....
1st PLATEAU
,$i\O/QI\
OXIDE SURFACE
/I Figure 5. Multilayer water adsorption to the silica surface.
H,OGENERATOR
@ HIGH PRESSURE @ LOW PRESSURE TO VENT
Figure 4. Gas flow apparatus for conducting in situ water vapor exposure experiments, using ATR-IR spectroscopy.
between dry nitrogen, a HzO/N2 stream, and a DzO/N*stream . (via a second GPH filled with DzO). For the water absorptionmeasurements,a reference spectrum of the test substrate (400 scans) in dry nitrogen was subtracted from the spectrum at each moisture exposure level (100 scans). We found that water adsorptionby oxide surfacesand absorption into the monolayers was in equilibriumwithin 3 min of exposure. The area under the IR band from adsorbed water, centered at approximately 3300cm-l,was used in determiningthe adsorption isotherms. Baselines were drawn from minimum points near 3690and 2500 cm-1 for the area determinations. The areasunder the water band above 3500 cm-l and below 3000 cm-l were not used because of interferences from surface silanol and C-H absorbancesin these regions. The intensityof the 1630-cm-'bending mode for water was also used as a semiquantitativecheck of the water absorption data. This latter band, however, appears on the low-frequencytail of a large,broad absorption band centered at 1710cm-1 which is intrinsic to the ATR substratesused. During measurement of the water adsorption isotherms,slight distortions to the substrate band at 1710 cm-l occurred, which caused difficulties in drawing a reproducible baseline in this region of the IR spectrum.
Results and Discussion Infrared Spectroscopyof the Thermal Oxide. The infrared absorption characteristics of the thermal oxide as a function of hydration were found to be very similar to the results for high surface area silicas. We have been able to follow changes in the free hydroxyl, hydrogenbonded hydroxyl, and molecular water infrared absorbances as the sample was exposed to moisture. Exposure to liquid water caused a decrease in the free hydroxyl band at 3750 cm-l and increased both the molecular water bands at 3400 and 1630cm-l and the hydrogen-bonded OH band a t approximately 3690 cm-I. Our results parallel the previously reported transmission IR results for powdered silicas16 and ATR-IR studies of thin film oxide on silicon.24.25 Our interpretation of these infrared results for thin film thermal oxide is based on the earlier work of Hair'6 and later work by Scott.26 During the water soak, Si-0 siloxane bonds a t the silica surface react with water, resulting in an increase in the silanol surface concentration. The increased number of silanol groups permits increased adsorption of molecular water and causes a decrease in the number of freely vibrating SiOH groups, since some of these are now hydrogen-bonded. Observation of multilayered water adsorption on silica has been confirmed by infrared and gravimetric measurements for powdered (24) Hartatein, A.; DiMaria, D. J.; Dong, D. W.; Kucza, J. A J. Appl. Phys. 1980,51 (7), 3860. (25) Beckman, K. H.; Harrick, N. J. J.Electrochem. Soc. 1971,118 (4), 614. (26) Scott, R. P. W.; Traiman, S. J. Chromatogr. 1980, 196, 193.
silicas.27 A model for the structure of the adsorbed water multilayer is shown in Figure 5. Multilayer formation occurs through hydrogen bonding of water molecules to the water layer already adsorbed to the silica surface. We will present infrared data suggesting multilayered water adsorption to hydrated thermal oxide in the section describing the water adsorption isotherms. These infrared data suggest that the surface chemistry of the thermally grown thin film Si02 surface is very similar to that of the well-studied high surface area silicas. The initial thermal oxide surface has very few silanols present, as is also the case for powdered silicas after annealing at >a00 "C. Ellipsometric and Contact Angle Measurements. Ellipsometric measurements of film thickness and contact angle measurements of surface wettability of the OTS and DMODCS films provided a means of comparing surface coverages and molecular ordering in the organosilane m~nolayers.~~J~Jg The thickness data and selected contact angle data are summarized in Table I, along with data from the recent literature for this well-characterizedsystem (Tillman et al., refs 11 and 12). Our value for the thickness of OTS self-assembled monolayers on hydrated thermal oxide (23A) agrees quite well with Tillman's data (25A). Other investigators have measured OTS monolayer thicknesses of 23-24 A using X-ray technique^.'^^^^ For OTS on fresh (dry) thermal oxide, we have obtained a value for the film thickness of 16 A by ellipsometry. Since the thickness of an OTS monolayer with fully extended alkyl chains oriented perpendicular to the surface is 23-25 A,11J4J9J9 it is apparent that the OTS monolayer on dry oxide does not represent complete coverage. The question of whether this incomplete film represents "islands" of well-ordered molecules or uniform but incompletely packed molecules was resolved by the infrared data, as discussed later. Our values for the contact angles of several test liquids (Table I) on the OTS/hydrated oxide are also in excellent agreement with the literature values for complete OTS monolayer^.^^^^^ The data for the n-alkanes and several polar test liquids are plotted in Figure 6. For OTS on hydrated thermal oxide, a value of 20.3 dyn/cm was obtained, while for OTS on dry thermal oxide a value of 24.3 dyn/cm was found. Tillman et al." reported a best value of 20.2 dyn/cm for close-packed monolayers of OTS on silicon. Since the accuracy of these measurements is approximately fl.Odyn/cm, these data represent excellent agreement for our OTS/hydrated oxide and the Tillman" values for OTS/silicon. The higher value for OTS on dry oxide indicates that the monolayer is not tightly packed. Ellipsometric measurements of DMODCS monolayers always gave smaller thickness values than were obtained for OTS monolayers, even though both molecules have the same alkyl chain length and hence the same end-to(27) Hair, M. L.; Hertl, W. J. Phys. Chem. 1969, 73,4269. (28) Pomerantz, M.; Segmuller, A.; Netzer, L.; Sagiv, J. Thin Solid Films 1986,132, 153. (29) Tillman, N.; Ulman, A.; Penner, T. L. Langmuir 1989,5, 101.
2240 Langmuir, Vol. 7, No. 10,1991
Angst and Simmons
Table I. Ellipsometry and Contact Angle Measurements of OTS Monolayers. DMODCS OTS OTS OTS (silicon, ref 11) (hydrated oxide) (dry oxide) (hydrated oxide) 25 8 23 16 114 104 111 78 75 70 73 51 48 14 49 39 41 3 43 33 20.3 24.3 20.2 >27
thickness, A Ha0 CH~I~ bicyclohexyl hexadecane dynlcm
8, 8, 8, 8,
DMODCS (dry oxide) 5 58 42 16 3 >27
a Film thicknesses, contact angles for selected liquids (e), and critical surface tensions ycfrom Zisman plots of n-alkanes,this work and that of Tillman et al. (ref 11).
0
OTSlHYDRATED Si02 O W D R Y Si02
Table 11. C-H Infrared Absorbance Peak Positions, Bandwidths (fwhh), Band Heights, and Dichroic Ratios for the .Y (CHI) Stretch of Monolayers of OTS and DMODCS on Dry and Hydrated SO,, Solution-Phare Data, and Data from Tillman et al.11 for OTS on Silicon dichroic freq, cm-l fwhh height ratio 2919 17 0.07 1.02 OTS on hydrated Si02 OTS on dry Si02 2926 25 0.04 0.84 OTS (ref 11) 2918 16 0.03 1.04 2928 23 OTS (0.1% in CC&) DMODCS on hydrated Si02 2928 22 0.02 0.79 DMODCS on dry Si02 2928 22 0.01 0.50 2928 22 DMODCS (0.1% in CC&) ~~~
8
0.41
-.:I,
0
0.2
,
,
I
I
I
,
I
,
,
~
- 0.4
15
30
45
60
75
SURFACE TENSION (dyneslcm)
Figure 6. Zisman plots of the contact angle data for OTS monolayers. Thecosine of the advancingcontact angle is plottedversue the surface tension of the test liquid. end distance in the fully extended configuration. Thicknesses of 8 and 5 A were measured for the DMODCS monolayers formed on the hydrated and dry thermal oxide surfaces, respectively. The contact angles measured on the DMODCS monolayers were also less than the angles measured for the OTS surfaces, and approachedthe values measured for the oxide surface. The contact angle data indicate that the oxide surface is partially exposed in the DMODCS samples, since the contact angles were much lower than would be expected for a CH2-covered surface. Infrared Spectroscopy of the Silanized Thermal Oxide. For both OTS and DMODCS monolayers, the infrared data indicated that the degree of surface coverage was greatly influenced by the hydration state of the oxide. For OTS, the C-H absorbances of the monolayer formed on the fresh thermal oxide surface were only l / 2 that of monolayers formed on the oxide that had been hydrated by soaking in water for 12 h. DMODCS surface coverages were always less than for OTS. In the case of DMODCS on dry thermal oxide, the C-H absorbances were barely visible above the noise level in the spectrum. The dichroic ratio of the CH2 infrared absorbance at approximately 2920 cm-' has been used to determine the orientation of the alkyl chains of the monolayer relative to the substrate surface.11Js*30Other investigators" have reported values of 1.03-1.11 for tightly packed OTS monolayers and lower values for more disordered alkyl chains. Both the CH2 absorbance peak positions and bandwidths also reflect the degree of order in the alkyl phase of the mono1ayer.l' Tightly packed chains in a well-ordered OTS monolayer exhibit a CH2 absorbance centered at approximately 2918 cm-l, with a bandwidth (full width at halfheight, fwhh) of 16 cm-'. Shifts to higher frequency and (30)Maoz, R.; Netzer, L.; Gun, J.; Sagiv, J. New Technological Applications of Phospholipid Bilayers, Thin Films, and Vesicles;Hayward, J. A., Ed.; Plenum Press: New York, 1986.
larger bandwidths are observed as the alkyl chains become more disordered. The infrared data for OTS and DMODCS on dry and hydrated Si02, along with literature values for OTS on silicon,ll are summarized in Table 11. Our values for the OTS/hydrated Si02 film again agree quite well with literature values for tightly packed, complete monolayers. In contrast, the infrared spectra for OTS/dry Si02 and DMODCS on either surface exhibit much lower absorbances and dichroic ratios, band broadening, and shifts to higher frequencies of the C-H absorbances, relative to the tightly packed monolayer case. The infrared data indicate that the CH2 chains in the OTS/hydrated Si02 monolayer are in a more orderly arrangement than in the OTS/dry Si02 monolayer. The DMODCS monolayers exhibit even more disorder in the CH2 chains. We attribute these differences to the lower surface coverage in the more disordered monolayers. From an examination of the infrared data, it is evident that the incomplete monolayers of OTS on dry Si02 and DMODCS on both the dry and hydrated surfaces do not form islands of tightly packed molecules. This can be inferred from the C-H band broadening and shiftto higher frequencies for the incomplete monolayer^.^^*^^*^^ Thus, the incomplete OTS and DMODCS monolayers of this work can best be described as uniform films containing relatively disordered alkyl chains. This result is in agreement with results for partial monolayers examined by X-ray reflectivity m e a s ~ r e m e n t s . ~ ~ We have also examined the OH absorbance region of these samples. Upon deposition of the OTS monolayer, we observed a decrease in absorbance in the 3720-cm-' region and increases in absorbance near 3600 and 3260 cm-'. The increases in absorbance near 3600 and 3260 cm-I result from silanol groups in the organosiloxanelayer. The decrease in free hydroxyl IR absorption upon silanization may indicate interaction between the free silanols on the silica surface and the silanol groups of the monolayer. The creation of siloxane bonds between the organosilane (31)Snyder, R. G.;Straw, H. L.; Elliger, C. A. J. Phys. Chem. 1982, 86,5145. (32) Zbinden, R. Infrared Spectroscopy of High Polymers: Academic: New York, 1964.
Langmuir, Vol. 7, No. 10, 1991 2241
Moisture Absorption of Organosiloxane Monolayers
a: 3 . 0 O r
HYDRATED
3'001 0
0
4
Figure 7. Infrared spectra showing the formation of an adsorbed layer on the Si02 surface, during in situ exposure to water vapor, for water vapor exposure pressures of (a) 1.1 Torr, (b) 6.4 Torr, (c) 13.2 Torr, and (d) 19.2 Torr. and the Si02 surface could also lead to decreased free si-
lanol absorbance although the infrared data do not permit unambiguous detection of covalent bonding between the organosilane and the silica surface. Infrared Spectroscopic Water Adsorption Isotherms. The principal objective of these studies was to determine the relationship between the coverage and structure of organosilanefilms and the absorption of water. A typical set of infrared spectra (a hydrated oxide surface is used here as an example) obtained during exposure to increasing water vapor pressure is shown in Figure 7. The broad absorption band at 3000-3600 cm-l is attributed to molecular water and the intensity of this band increases with water vapor pressure. This band also shifts to higher frequency with increasing amounts of adsorbed water. We interpret this frequency shift to represent the build-up of regions of water multilayer structure^.^^ At low water vapor pressure, water is apparently present in submonolayer amounts and is hydrogen-bonded only to surface silanol groups. With increasing amounts of adsorbed water, hydrogen bonding between water molecules takes place and is accompanied by a shift to higher frequency of the water infrared absorbance. These results are similar to those obtained previously for water absorption by powdered silicas in the near-IR frequency range,% For the silanized surface, no water-induced changes in the CH infrared absorption bands were observed, indicating that water adsorption did not significantly change the packing of the monolayer alkyl chains. This result suggests that water adsorption on silanized surfaces takes place at the organosilane-oxide interface. In Figure 8 we have plotted the area of the infrared band of the OH stretch of molecular water, versus water vapor pressure, for dry and hydrated Si02 surfaces, and for the OTS-and DMODCS-treated surfaces. The Si02 samples were hydrated by soaking in DI water at the indicated temperatures and times. The substrates used for the monolayers were hydrated in DI water at room temperature for 12 h before film deposition. For a convenient comparison of the relative amounts of adsorbed water for each surface,Table I11lists the data corresponding to 6 Torr water vapor exposure (multiple entries represent data from separate samples). Significantly more water adsorption was found for the hydrated thermal oxides than on the freshly oxidized surface. This result is consistent with studies for high surface area silicasn in which it was reported that the amount of water adsorbed on silica which had many hv(33) Franks, Felii Water: A Comprehemive Treathe;Plenum Press: New York, 1982. (34) Klier, K.; Zettlemoyer, A. C. J. Colloid Interface Sci. 1977,58, 216.
HYDRATED DRY
A
CURED. HYDRATED
CURED. HYDRATED 0 CURED: DRY
2.00
PRESSURE (Torr)
Figure 8. Adsorption isotherms for water on SO2 and organosilane monolayer surfaces. The area of the infrared absorbance from adsorbed water is plotted versus the water vapor pressure. Table 111. Infrared Absorbance Peak Areas from Surface-Adsorbed Water, Adsorbed on the Test Surfaces after Equilibration for 15 min at 6 Torr Water Vapor Pressure dry Si02 cured DMODCS/drv Si09 DMODCS/dry SiO; dry SiO2, after 2 h in air at 60% re1 hum cured OTS/dry Si02 OTS/dry Si02 DMODCS/hydrated Si02 cured DMODCS/hydrated Si02 cured/hydrated Si02 cured OTS/hydrated Si02 hydrated Si02 OTS/hydrated Si02
HzO area (IR) 0.21,0.36 0.45 0.49 0.62 0.66.0.68 0.95 1.01 1.02 1.10 1.24,1.15 1.26,1.23,1.38 1.81,1.61,1.75
droxyl groups and a preadsorbed layer of surface water was greater than the amount adsorbed on dried silica. If the freshly oxidized silicon surface was exposed to humid room air for 2 h before measurement of the adsorption isotherm, the isotherm indicated higher amounts of adsorbed water. We attribute this behavior to preadsorption of atmospheric water prior to the IR measurements and to the possible hydrolysis of some surface siloxane bonds, although our measurement could not unambiguously detect the formation of hydrogen-bonded silanols after air exposure. The kinetics of the reaction of water with the oxide surface are evidently slow at room temperature, sincewe observed relatively low water adsorption on the freshly oxidized silicon surface even after exposure of the oxide to water vapor. The dry silicon oxide surfaces silanized with DMODCS essentially maintained the low water adsorption characteristics of the bare oxide surface (see Figure 8 and Table 111). The amount of water adsorbed on a hydrated oxide silanized with DMODCS was less than the bare hydrated oxide surface but greater than the bare dry oxide surface. DMODCS apparently is effective in blocking some sites (i.e. surface silanols) for water adsorption. In contrast with DMODCS, when either the hydrated or dry oxide surface was silanized with OTS, the amount of water adsorption was higher than the bare surfaces. This result was unexpected since the OTS monolayers exhibited the highest coverage and highest degree of order of any of the surfaces. Furthermore, the OTS silanized hydrated oxide surface was the most hydrophobic (at least macroscopi-
2242 Langmuir, Vol. 7, No. 10, 1991
Table IV. Effect of OTS Polymerization (Solution) and Curing (Film) on Water Absorption polymerization time 0 min 4h 240 h 0 min/cured 4 h/cured 240 h/ cured
Ra
8b
0.95 1.10 1.11 0.94 1.13 1.09
98 104 114 92 93 98
Hz0 (6)' 1.02 0.88 0.86 0.58 0.50
0.29
0 R,dichroic ratio (s/p) of the CH2 absorbance. b 8, water contact angle. e H20 (6), amount of water absorbed at 6 Torr (IR area). I
cally). The increased water adsorption capacity of the OTS/hydrated oxide surface compared to the bare surface may be attributed to the fact that although some surface silanols are blocked by the covalent bonding of the OTS molecule to the oxide surface, the OTS layer itself has silanol groups present. There is consequently a net gain in sites (silanols) for water adsorption. Curing the OTS surface overnight at 150 "C reduced the number of silanols through cross-linking between OTS molecules, which resulted in a significant decrease in the water adsorption capacity of the surface. Curing the DMODCS-treated surface had no effect on water adsorption characteristics since cross-linking and reduction in the number of silanols does not occur in this case. These results suggested that a reduction in water absorption by OTS monolayers could be realized by reducing the number of silanols. To test this possibility, monolayers were deposited from an OTS solution in which the organosilane was first allowed to polymerize for various times before deposition. In this experiment, a comparison was made between films deposited immediately after preparation of the solution and films deposited from the same solution after 4 h and after 10 days had elapsed. Although no fogging of the solution from coagulation or precipitation was observed, the solution was sonicated for 10 min before each use to ensure homogeneity. Data obtained from these samples are presented in Table IV. Water absorption by OTS monolayers was reduced both by polymerization of the organosilane prior to film deposition and by curing of the deposited film. In support of this finding, it has also been reported recently that the performance of chromatographic columns was enhanced by using a polymerized stationary phase on silica.35 The chromatographic data indicated a reduction in polar adsorption sites for the polymerized phase relative to monomer phases. Conclusions Monolayers of octadecyltrichlorosilane (OTS) and dimethyloctadecylchlorosilane (DMODCS)were deposited onto the thermal oxide of silicon under molecular selfassembly conditions, using both freshly oxidized and hydrated silicon oxide surfaces. We found that closepacked, high-quality monolayerswere only formed for the case of OTS on the hydrated oxide surface. Ellipsometry, contact angle, and IR-ATR data for OTS/hydrated oxide prepared under our experimental conditions agreed with previously published data and confirmed that close-packed monolayers were formed for this system. Trifunctionality at the silicon terminus of the alkylorganosilane was found to be necessary to form tightlypacked monolayers. The ability to form cross-links at the oxide surface, between OTS molecules as well as to the oxide surface, apparently facilitated the formation of ordered
I
I
I
Figure 9. Schematic representation of the uncured OTS interfacial region, showing several possible sites for water adsorption.
monolayers. In contrast, steric hindrance of the methyl groups at the interface and absence of cross-linking in the case of DMODCS presumably interfered with ordered monolayer formation. The amount of water adsorbed onto the oxide and at the monolayer-oxide interface was measured by using infrared spectroscopy. We believe the water absorption behavior of the organosilane-silica interface can be interpreted in terms of the diagram shown in Figure 9., Vapor-phase water can penetrate even a tightly packed, fully covered OTS surface, despite the hydrophobicity observed macroscopically. Water which penetrates the outer alkyl surface binds at the interfacial region to silanol groups attached to both the silica surface and the OTS molecules. Curing at 150 "C, which has been shown via NMR data to result in cross-linking of organosilane molecules and covalent bond formation to the silica surface,36decreases the number of silanol groups available for water adsorption. Thus, uncured OTS monolayers actually increase interfacial moisture adsorption relative to the unsilanized surface (data of Table 111).After curing, water absorption at the OTS/silica interface is greatly decreased, since the number of silanols present at the interface is reduced. In contrast to silanization with OTS, DMODCS treatment does not result in an increase in the number of silanol groups at the interface. On the hydrated surface, DMODCS treatment removes some of the oxide silanol groups, resulting in decreased moisture adsorption. On the dry oxide surface, DMODCS treatment does not significantly affect moisture adsorption. Thus, although the outer surfaces of the uncured OTS and DMODCS monolayers are hydrophobic, the amount of moisture adsorption at the interface depends on the number of unreacted silanols available to serve as adsorption sites. The effect of curing at 150 "C was minimal for DMODCS, because curing does not change the number of silanol groups present at the DMODCS/oxide interface. Polymerization of the OTS molecules in solution also was found to decrease water absorption. Dry Si02, and surfaces prepared from OTS (cured) or DMODCS deposited on dry Si02, exhibited the least moisture adsorption of the test surfaces investigated. Within 2 h of moist air exposure, however, the dry unsilanized Si02 surface had chemisorbed enough water to significantly increase the amount of water adsorbed during measurement of the spectroscopic adsorption isotherm. The organosilane-coatedsurfaces provide better protection of the Si02 surface from atmospheric water, provided that the coating does not increase the number of silanols present in the interfacial region. Registry No. Si02, 7631-86-9;HzO, 7732-18-5;CH& 7511-6;bicyclohexyl, 92-51-3;hexadecane, 544-76-3.
(35) Payne,K. M.; Tarbet, B. J.;Bradshaw, J. S.;Markides, K. E.; Lee, M. L. Anal. Chem. 1990,62, 1379.
(36)Caravajal, G. S. Dissertation, Colorado State,1986.