Manuscript Received 1-18-89 Stress Reduction of Ion-Beam-Sputtered Mixed-Oxide Coatings by Baking B. J. Pond, J. I. DeBar, C.K. Carniglia and Tilak Raj Martin Marietta Astronautics Group, Laser Systems Technology Thin films deposited by ion-beam sputter deposition (IBSD) typically have a high compressive stress. This stress can be reduced for certain materials by cosputtering with another material [1]. The stress can be further reduced by baking the films in air after coating. Films of zirconia (ZrO2) and silica (SiO2) were prepared by IBSD from hot-pressed oxide targets using argon as the sputter gas. Films consisting of a mixture of silica and zirconia were prepared by sputtering from both targets simultaneously. Calorimetry measurements at 351 nm showed that the absorption in the mixed-oxide films was lower than the absorption in the zirconia film. A compressive stress of 219 kpsi was observed for the zirconia film and of 112 kpsi for the silica film. All of the mixed-oxide films had lower stress. Those films with silica fractions between 10% and 50% had stresses in the range of 40-50 kpsi. This stress could be reduced even further by baking the coated parts for several hours at 300°C. For mixed-oxide films with a silica fraction less than 50%, the stress of the films after baking was tensile. In particular, the film with 10% silica was changed from a compressive stress of 46 kpsi to a tensile stress of 23 kpsi by the baking process. Similar results were observed for a mixed-oxide film consisting of zirconia and alumina (A1203). These results indicate that a stress-free multilayer coating may be achievable by IBSD. cosputtering; ion beam; optical coatings; refractive index; silica; Key words: stress; zirconia Ion-beam sputter deposition (IBSD) has been shown to be a viable process for producing high-energy laser coatings [2-4]. Thin-film optical coatings made by IBSD have a higher density and lower impurity levels than conventionally evaporated films. This is due to the higher energy of the sputtered particles condensing on the substrates. The average kinetic energy of sputtered particles is approximately 5-10 electron volts, whereas the kinetic energy of conventionally evaporated species is approximately one-tenth of an electron volt [5]. These energetic particles dislodge adsorbed impurities from the substrate surface and from the coating as it is being deposited. One limitation of coatings produced by IBSD is the high compressive stress of the films [3,6-9]. Stress can be a significant problem with thick coatings, since the mechanical forces scale with thickness. These forces can distort the substrate and cause the film to delaminate. In mirrors for optical systems, the deformed surfaces can result in considerable beam distortion. Numerous studies have been conducted to investigate the cause of stress and to determine methods of modifying the stress in thin films. However, most of these studies have been conducted with sputtered metal films [7-16]. Recently, several studies have shown that the stress in dielectric thin films can be altered by coevaporating two materials [17-18]. One of these studies also showed a change in the grain structure of the coatings which correlated with the changes observed in the stress [17]. In the case of electron-beam (E-beam) evaporated zirconia, a change was observed in the crystalline phase, the microstructure, the grain size and the stress of the films when a glass former such as silica was added using coevaporation [18]. This paper reports on the cosputtering of zirconia (ZrO2) with either silica (SiO2) or alumina (A1203) using IBSD. The effects of cosputtering on the stress and the optical properties of the films are presented. A total of 12 different compositions were investigated. Eleven of the films were zirconia/silica with the silica fraction ranging from 0% to 100% in increments of approximately 10%, and one of the films was zirconia/ alumina with an alumina fraction of approximately 10%. The addition of either alumina or silica to the zirconia was found to reduce the stress observed in the zirconia films. The effect on the stress of post-deposition baking was investigated to determine whether the stress in the films could be reduced even further. The films were baked at 300°C for three hours, and a noticeable change was observed in the stress of the films. A multilayer coating was fabricated using zirconia/silica mixed-oxide material as the high-index material and silica as the low-index material. After the coating was completed, it was baked at 300°C for several hours, and a change was observed in the stress which corresponded to the change observed in the stress of the single-layer films. The following section discusses the experimental procedures used in depositing, analyzing, and baking the films. This is followed by a section discussing the results. 2. Experimental The films were fabricated by IBSD in a cryopumped 20-inch bell jar using two Kaufmantype ion sources to sputter material simultaneously from two separate targets. The zirconia and alumina targets were in the form of hot-pressed oxide material, and the silica target was in the form of fused silica. All of the targets were 17.5-cm in diameter. The base pressure of the vacuum system was 5 x 10-7 torr or lower. The ion energy used with both sources was 1000 eV. The sputter gas used was argon, and a partial pressure of oxygen of 3.0 x 10-5 torr was supplied directly into the chamber to achieve stoichiometric films [19]. An equal pressure of argon was used in each ion source and the total chamber pressure was 3.1 x 10-4 torr. The substrate temperature rose above ambient during deposition due to radiant heating from the ion sources. The highest temperatures were in the range 40°-60°C. The films were deposited on substrates held in a circular rack which was rotated about its axis. The film thicknesses were monitored by an optical monitoring system using front surface reflection monitoring. The various compositions of the single-layer films were obtained by adjusting the ion beam current incident on each target. The system was calibrated by depositing separate single-layer films of alumina, silica and zirconia using a fixed ion beam current for a fixed time. The thicknesses of these layers (denoted dao, dso and do for the alumina, silica and zirconia films, respectively) were determined using the optical methods described below. To obtain a film with a given volume fraction f of silica, the currents Is and Iz for the silica and zirconia ion beams are related by Is = Iz (dzo/dso) f/(1 - f). (1) For compositions with silica fractions less than 0.5, Iz was set to its maximum value, and eq. (1) was used to determine Is. Iz was determined using eq. (1). given volume fraction of alumina. For f > 0.5, Is was set to its maximum value, and A similar relation was used to obtain a film with a This deposition technique was used to give a series of single-layer coatings with approximate silica volume fractions equally spaced from 0% to 100% in 10% increments. The actual composition of each film was determined separately using analytical techniques. Single-layer films of two different thicknesses were fabricated sequentially under identical conditions for each of the zirconia/silica compositions. The thinner films were approximately 100 nm in physical thickness, and these films were used for analysis by Rutherford back scattering (RBS) [20] to determine the actual composition of the films. thicker films had an optical thickness of approximately 5 quarter waves at 550 nm. The thicker films were used for stress analysis, measurements of optical properties, and absorption measurements by laser calorimetry. The deposition technique described above was also used to make a zirconia/alumina mixed-oxide coating with an approximate alumina volume fraction of 10%. Only a thick film of this composition was made, and the actual composition of this film was not determined by RBS. The multilayer coating was a 25-layer quarter-wave-stack high reflector centered at 1.3 μm. The high-index material was a zirconia/silica mixed-oxide film, with an approximate silica volume fraction of 10%, and the low-index material was silica. Alternate layers of the two materials were deposited in the same manner as the single-layer coatings. The stress of the films was determined by measuring the curvature of a thin fused-silica disc coated with a single layer of the given film material [21]. The stress discs for the single-layer coatings were 2.54 cm in diameter and 0.38-mm thick, and for the multilayer film, 2.54 cm in diameter and 0.50-mm thick. The curvature was measured with a Fizeau interferometer using the 508-nm green light from a cadmium lamp. The sensitivity of the method was + 4 kpsi. In order to reduce the data, the physical thickness of the films was needed. For the single-layer films, this thickness was calculated as part of the spectral determination of the optical constants of the films. For the multilayer coating the thickness was determined using the optical constants for each material and the theoretical optical thickness of each layer in the coating design. Thus, the determination of the stress in the single layer films is susceptible to errors in the measured refractive index of the films. In the multilayer films, it is also susceptible to errors in the optical thickness of each layer. The envelope method [22,23] was used for determining the optical constants and the physical thicknesses of the films. This method is based on the analysis of spectral scans of transmittance and reflectance of the single-layer films. These scans were made on a Cary-2300 dual-beam spectrophotometer covering the visible, near-UV, and near-IR wavelength ranges. The envelope method of analysis allows the determination of the refractive index n, the extinction coefficient k, and the physical thickness d of the films. The method has been modified to allow for the determination of the degree of inhomogeneity An/n of slightly absorbing films [24]. This additional information provides useful insight into the structure of the films [25,26]. The precision in the determination of n was estimated to be 0.01, while the uncertainty in k varied from about 0.0015 at 1000 nm to about 0.0003 at 400 nm. Established procedures [22] were used for making the transmittance and reflectance measurements necessary for the determination of n, k and An/n. For these measurements, the films were deposited on UV-grade fused silica substrates with a 2.54-cm diameter and a 1-mm thickness. The transmittance measurements were made by first scanning an uncoated fused silica substrate and then scanning the coated part. The transmittance was determined by taking the ratio of the two scans. The reflectance measurements were made from the single coated surface at near-normal incidence. (The deviation from normal incidence was ignored.) The reflectance from the second surface was eliminated by using an index-matching fluid to attach a piece of fused silica to the back side of the sample being measured. The second surface of this piece of fused silica was frosted to diffuse the reflected light. The necessary data were taken manually from the spectral scans and used to determine the optical constants and physical thicknesses of the films. Absorption of the single-layer films at 351 nm was measured using a laser calorimeter [27,28]. The substrates used were fused-silica stress discs, 2.54 cm in diameter and 0.38-mm thick. The uncoated substrates had a absorption of 0.0150%-0.0200%. The values for the coated samples include the absorption of the substrate. The chemical compositions of the zirconia/silica single-layer films were determined from RBS measurements conducted at Charles Evans and Associates*. The RBS method provides a quantitative chemical analysis of most elements without any external standards. Detection sensitivity is not high for light elements like carbon and oxygen. A 2.2-MeV He++ ion beam was directed at the sample, and the backscattered ions were analyzed for energy and flux. The RBS results provided measured values of the relative fraction of silicon and zirconium in the films, thus allowing the atomic percentages of silica and zirconia to be determined to an accuracy of about ± 1%. Charles Evans and Associates, 301 Chesapeake Drive, Redwood City, California, 99063. The post-deposition baking was conducted in an air furnace. The parts were placed in the furnace, and slowly heated to 300°C. The single-layer films were baked at 300°C for three hours, and the multilayer film was baked at 300°C for eight hours. The parts were then cooled slowly to room temperature. 3. Results A summary of the measured compositions and optical properties of the films is presented in table 1. The first column lists the estimated volume fraction of zirconia based on the as-deposited ion-beam currents. The second column gives the atomic percent of zirconia as measured by RBS for the zirconia/silica single layers. The following graphical presentations of the zirconia/silica results are in terms of atomic percent of zirconia as measured by RBS. The zirconia/alumina data points are included in these graphs, but this composition was only determined by the deposition parameters. The remainder of table 1 presents the optical data [n, k, (An/n)ay and layer thickness]. These data are discussed below. Table 1. Summary of film properties for cosputtered films. The measured stress values are presented as a function of measured zirconia fraction in figure 1. The stress of the zirconia/alumina film is indicated by the filled-in data point at a zirconia fraction of 90%. It can be seen that there are three distinct regions in the graph. The first region is for a zirconia film, which has a high compressive stress. The second region is for compositions with a zirconia fraction between approximately 90% and 35%. In this region, the stress values are low and relatively constant over a wide range of compositions. The third region is for a zirconia fraction less than 35%. In this region, the stress gradually increases towards the stress of the silica films. The major observation that can be made from these results is that the high stress observed in zirconia films is reduced significantly by the addition of silica. Similar results were observed when alumina was added to the zirconia. The stress in the zirconia/ alumina film was not as low as in the corresponding zirconia/silica film, although the difference was not significant. It can be seen that the addition of zirconia to silica has a different effect than the addition of silica to zirconia. The reduction in the stress of the films is believed to be caused by a modification of the structure and reduction in the grain size. However, this does not rule out other possible mechanisms of reduction in stress. In previous work it was shown that the zirconia film has a polycrystalline structure whereas the zirconia/silica mixed-oxide films have an amorphous structure [1]. The zirconia/alumina film also has an amorphous structure. The addition of either alumina or silica to the zirconia disrupts the crystalline structure and results in an amorphous film. The measured stress values of the baked films are presented as a function of composition in figure 2. The stress of the zirconia/alumina film is indicated by the solid data point. A positive value indicates compressive stress and a negative value indicates tensile stress. The baking process modified the stress of all the films including the zirconia film and the silica film. The stress was less compressive and, in some compositions, the stress actually changed from compressive to tensile when the films were baked. The films with a zirconia fraction between 90% and 50% were tensile after baking, while films with a zirconia fraction less than 50% were compressive. It is interesting to note that the composition with a zirconia fraction of approximately 50% had the lowest stress after it was baked. The change in stress due to baking is believed to be caused by an internal rearrangement including diffusion of vacancies. The baking temperature of 300°C is lower than the annealing temperatures reported for IBSD zirconia films [29]. The total change in stress between the unbaked films and the baked films is presented in figure 3 as a function of composition. The change in stress of the zirconia/alumina film is indicated by the solid data point. This result shows that the change in stress varies almost linearly with composition. Thus, the change in stress is proportional to the concen tration of zirconia in the film. A theoretical method was used to estimate the stress of the multilayer coating both before and after it was baked [30]. This calculation was based on the stress values determined from the single-layer coatings and the estimated thickness of the layers in the coating. The stress calculated for the multilayer coating before it was baked was 84 kpsi and the measured stress for the actual film was 98 kpsi. The calculated stress of the baked film was 39 kpsi and the measured stress for the actual film was 55 kpsi. The actual values |