We report on the properties of gallium oxide (Ga2O3) thin films deposited on c-plane sapphire substrates using radio frequency magnetron sputtering under various conditions. The parameters varied included the composition of the deposition gas, the substrate temperature, and postdeposition annealing temperature. The optical characteristics obtained by UV-VIS spectroscopy showed excellent transparency of 90%–95% for all the films obtained. The structural and compositional properties of the films were determined using x-ray diffraction and energy dispersive spectrometry measurements. The films deposited in Ar at 400 °C showed diffraction peaks at 18.6°, 37.2°, and 58.2°, which are attributed to diffraction peaks from , , and planes of β-Ga2O3. Postdeposition annealing in N2 at 400–900 °C did not make any improvement in the crystalline quality of the films. The addition of tin in the films produced transparent films whose optical bandgaps decreased with increasing tin concentration in the films.
In recent years, gallium oxide (Ga2O3) has emerged as a more attractive semiconductor material for fabrication of electronic and optoelectronic devices. It is now known to exist in five polymorphs, namely, corundum (α), monoclinic (β), defective spinel (γ), and orthorhombic (ε and δ), with the β form being more stable over a wide temperature range.1,2 The larger bandgap energy of about 4.5–5.3 (depending on the polytype) for Ga2O3 compared to that of 4H-SiC (3.26 eV) and GaN (3.44 eV) translates into exponential improvements of many critical device parameters for high power electronic device applications.3 For instance, the critical electric field breakdown in V/cm is found to vary with bandgap according to the relation
where a is a constant whose value is ∼105 and the exponent n can vary from 2.0 to 2.5.4 This gives an estimated value of 8 MV/cm for β-Ga2O3, much higher than that for 4H-SiC (2.5 MV/cm) or GaN (3.3 MV/cm).5,6 For this reason, Ga2O3 devices would withstand higher electric fields with reduced energy loss, would require less material, and would accommodate higher temperature operations leading to improved system efficiency. Ga2O3 would usher the next generation of electronic and optical devices with significantly improved performances. Interest in Ga2O3 as a material for power devices has further been accelerated by the successful growth of large single crystals, which would potentially translate to easier wide scale adaptation and cost-effective devices.7 Furthermore, the excellent transparency of Ga2O3 over a wavelength range covering visible down to 250 nm makes it suitable for a host of photonic device applications including transparent conductors and deep ultraviolet (UV) sensors.8 Development of SiC and GaN materials for device applications have made significant progress in part due to achievement of their n- and p-type doping. However, similar to other wide bandgap transparent oxide semiconductors such as ZnO with doping asymmetry, it is relatively easy to obtain n-type conductivity in β-Ga2O3 by doping with group IV elements such as Si, Sn, or Ge. On the other hand, realization of effective p-type conductivity in β-Ga2O3 has presented a significant challenge, due to limitations arising from the high formation energy of the native acceptors as well as the low formation energy of compensating native donors and self-trapped holes.1,9 Furthermore, as it is a newer material in development, only limited studies have been carried out on the long term stability of β-Ga2O3 devices to determine appropriate isolation and surface passivation schemes for improved stability.10 Other oxide semiconductors such as ZnO have shown surface sensitivity to oxygen and water vapor requiring appropriate passivation to ensure device stability.
Epitaxial growth of high quality Ga2O3 thin films has been demonstrated by various methods including chemical vapor deposition, pulsed lased deposition, molecular beam epitaxy, metalorganic chemical vapor deposition, and magnetron sputtering.1,11–17 Among the different fabrication methods for depositing oxide thin films, magnetron sputtering is the preferred method in many industrial processes because it is an established technique, suitable for large area fabrication and roll-to-roll processing.18 Additionally, magnetron sputter deposition is a versatile technique that accurately preserves homogeneous composition and stoichiometry reproducibly, achieving better step coverage and thickness uniformity.19 Sputter deposition has been a key enabler for technology advancement of several material structures including superlattices and complex oxides. This technique has been successfully employed in realizing high quality epitaxial films of complex oxides and other materials at a more cost-effective manner.20 Moreover, sputtering technique does not rely on the reactions of complex precursor molecules but on purely physical momentum transfer process. The sputtering technique allows for a more purely thermodynamic study of the deposition of the undoped/doped β-Ga2O3 material. While other physical vapor processes such as pulsed laser deposition and molecular beam epitaxy also do not involve many complex precursors, sputtering is a much faster and cheaper method that would greatly speed up the discovery process.
In general, epitaxial thin films of Ga2O3 is obtained when deposition is carried on c-plane sapphire substrates using ultrahigh purity (UHP) Ar at substrate temperatures of 300–600 °C by magnetron sputtering or pulsed laser deposition.17,21,22 Using magnetron sputtering, Sun et al. obtained β-Ga2O3 films with the best crystalline quality when the films were deposited on c-plane sapphire held at 500 °C with average transmittance greater than 90%, and the optical bandgaps that varied from 4.68 to 4.94 eV.23 Similar results were obtained by others.24,25 These reports mainly focused on the films obtained by optimizing the substrate temperatures for films grown in UHP Ar. In other investigations, β-Ga2O3 films sputter-deposited at room temperature were found to be amorphous, but transformed to crystalline form after annealing at 500–900 °C.16,26,27 The properties of β-Ga2O3 thin films have been found to vary with the growth methods and parameters used, perhaps due to the nature of accompanying defects varying with the growth methods used.28 Earlier work on Ga2O3 films indicate that growth of the film in reducing conditions form semiconductors while growth in oxidizing conditions form insulators.29,30 It is therefore important to study the properties of films obtained under various conditions. In this report, we present results obtained on Ga2O3 films deposited on c-plane sapphire substrate by magnetron sputtering under various conditions. The objective of the investigation was to determine optimized deposition conditions for Ga2O3 thin films using magnetron sputtering, a relatively inexpensive technique commonly used in many industrial processes. The films were produced by changing the substrate temperature from 20 to 900 °C, the deposition gas composition, and the postdeposition annealing treatment. Films deposited in UHP Ar on substrates held at 400 °C showed superior quality exhibited by x-ray diffraction (XRD) peaks belonging to , , and planes of β-Ga2O3. This optimum condition for substrate temperature shifted to a higher value when different mixtures of UHP Ar and O2 were used for the deposition. Postdeposition annealing in UHP N2 at temperatures up to 900 °C did not improve but instead worsened the quality of the films formed. The addition of tin produced films whose optical bandgaps decreased with increasing amount of tin in the films.
The surfaces of the c-plane double-side polished sapphire substrates were first prepared by degrease-cleaning in boiling acetone, alcohol followed by rinsing in de-ionized (DI) water. The samples were then dipped in buffered HF acid for 10 min and rinsed again in DI water, dried and mounted on a substrate heater, loaded in a vacuum chamber, and evacuated to 2 × 10−7 Torr by a turbo molecular pump. Right before deposition, the sapphire substrates were subjected to thermal desorption by heating at 500 °C for 30 min in UHP O2, a process previously found essential for sapphire substrate preparation.31,32 All the gases mentioned in this report for deposition or for annealing were of UHP grade (>99.99% pure). The films in this study were deposited from a 2-in.-diameter high purity (99.95%) Ga2O3 ceramic target, and a distance of about 8 cm was maintained between the target and the substrate. A 15-min presputtering was performed to remove any impurities from the target, followed by a deposition of about 120 min for each sample. The gas pressure was set at 10 mTorr, a flow rate of 5.0 standard cubic centimeters per minute at standard temperature and pressure (STP), and a radio frequency (RF) power of 100 W, and the thicknesses of the resulting films were 145–200 nm. The first set of samples prepared were of films deposited at different substrate temperatures from 20 to 900 °C in (i) Ar and (ii) a mixture consisting of 80% Ar and 20% O2. A 2-in.-diameter substrate heater (Cat # 10 1446 MOD from Heatwave lab, Inc., Watsonville, CA) was used in which the temperature is determined using a thermocouple placed in direct contact with the substrate. Further investigation was carried on the film deposited in Ar at 400 °C. This sample was diced into several pieces and annealed in N2 at 400–900 °C for 15 min. The second set of samples was of films deposited at 500 °C, but using different percent mixtures of Ar/O2 comprising 100/0, 80/20, 50/50, 20/80, and 0/100. The third set were samples formed by co-deposition of Ga2O3 and Sn on a substrate held at 500 °C with Ar used as the process gas. The Sn atoms, sputtered from a pure Sn target at different currents (8.0, 20, and 25 mA) of the direct current power supply, were introduced to serve as n-type dopants to the Ga2O3 films.
The microstructures of the resulting films were characterized using XRD measurements with a Bruker-Prospector diffractometer (Bruker AXS, Inc., Madison, Wisconsin, USA) fitted with a high brightness Cu Incoatec microsource. The optical property of the films was investigated using the Ocean Optics QE65000 spectrometer with a deuterium halogen UV-VIS-NIR light source (DT-MINI-2-GS). Out of these, we obtained the spectral transmission characteristics and the optical bandgap of Ga2O3 films, as further explained later in Sec. III of this report. The compositional property of the films was determined from energy dispersive spectrometry (EDS) measurements using a JEOL JIB-4500 scanning electron microscope fitted with Apollo XV detector (EDAX Inc.). To determine the thickness of the films, several images of the cross section of each film were obtained using a backscattering detector in the JEOL JSM 7600F field emission scanning electron microscope.
III. RESULTS AND DISCUSSION
Figure 1 shows the XRD scans of the β-Ga2O3 films deposited on c-plane sapphire substrates at different temperatures from 20 to 900 °C using (a) Ar and (b) a mixture of 80% Ar and 20% O2. The thickness of each film is indicated in the corresponding legend. For the films deposited in Ar, the substrate temperature of 400 °C gives the best crystalline quality with appearance of higher intensity diffraction peaks at 18.3°, 37.1°, and 57.8° corresponding to , , and planes of β-Ga2O3, respectively. This film has a thickness of 134 nm, less than that of the rest of the films in this group. Films deposited at temperatures higher than 400 °C were polycrystalline and showed additional peaks at 30.0° and 63.7°, which belong to the (4 0 0) and (8 0 0) planes of β-Ga2O3, respectively.23,25,33,34 Additional peak appearing at 43.8° could be associated with a reflection from the (3 1 1) plane. The film deposited at room temperature (20 °C) showed no diffraction peak, indicating amorphous quality, while the appearance of a tiny peak is observed on the film deposited at 200 °C. Meanwhile, the films deposited using a mixture of 80% Ar and 20% O2 has better crystalline formation at a substrate temperature of 500 and 550 °C. Polycrystalline formation indicated by appearance of additional peaks at 30.0°, which belong to the (4 0 0) plane of β-Ga2O3, is observed for films deposited at temperatures higher than 400 °C for Fig. 1(a) and higher than 550 °C for Fig. 1(b). The scans shown in Fig. 1(c) are for films deposited at 500 °C using different mixtures of Ar and O2. These results indicate that the optimum substrate temperatures for formation of high quality crystalline β-Ga2O3 films is 400 °C when using Ar, and 500–550 °C when using 80% Ar mixed with 20% O2. It would appear that high quality crystalline β-Ga2O3 films can be realized on c-plane sapphire substrates using different mixtures of Ar and O2, but at different substrate temperatures. The variation in the thickness of the films in the above data could be a result of nonuniformity in the film thickness across each sapphire substrate (¼ of a 2-in. wafer) during deposition. Typically, this thickness nonuniformity is reduced by rotating the substrate during deposition, and in our case, this capability was not available. Nevertheless, Zhang et al and Sun et al showed that the increase in the thickness of Ga2O3 films only increases the intensity of XRD peaks.35,36 Therefore, due to the variation in the thickness of our film, we make no comparison of the XRD peak intensities. Several researchers have reported optimum temperatures of 300–600 °C for sputter-deposited β-Ga2O3 films on c-plane sapphire substrates. Akazawa found 300 °C to be the optimum temperature when β-Ga2O3 was deposited under O2 gas flow on sapphire c-plane.25 Other researchers found 500 °C to be the optimum substrate temperature for crystallization with a clear out-of-plane orientation of β-Ga2O3 with the sapphire substrate using Ar.23,37,38 Achievement of high quality crystalline β-Ga2O3 films deposited on c-plane Al2O3 is made possible by their relatively small lattice mismatch of 6.6% and nearly equal thermal expansion coefficients (αc) of 3.15 × 10−6 K−1 (for Ga2O3) and 4.3 × 10−6 K−1 (for Al2O3).39,40
The film microstructure is established by the mobility of the sputtered atomic species on the surface of the substrate, which in turn is determined by several factors in the sputtering process. Proper optimization of these parameters can, therefore, be used to tailor the film quality. Explanation for preferential growth directions of sputtered films has been proposed comprising the thermodynamic, kinetic, and atomistic models.41 In the first model, the minimum energy of the substrate-film system (thermodynamic equilibrium) determines the growth direction of the film.42 In the kinetic model, the eventual orientation of the film is that in which the species have the least mobility, caused by the substrate temperature.43 In the atomistic model, orientation of the films deposited at high temperatures is determined by the thermodynamics of the growth planes, in which the one with the lower energy is preferred.44 For these reasons, the change in the deposition gas composition will in turn produce sputtered species with varied kinetic energies, and the change in substrate temperature will contribute to the kinetics of the species as they condense on the surface of the substrate. In either case, optimum film microcrystalline structures are obtained at different sputtering conditions.
Table I shows the elemental composition obtained by EDS analysis of Ga2O3 films deposited at 500 °C using different Ar/O2 mixtures. The results show that using higher Ar content in the deposition gas produces correspondingly higher Ga content in the films. This could be due to increased preferential sputtering of Ga arising from increased kinetic energy of the sputtering ionic species when more Ar atoms with larger mass compared to O2 is present in the gas mixture. Meanwhile, Table II shows the EDS elemental composition of the films deposited at different temperatures using an Ar/O2 mixture of 80/20%.
% ratio .
|Atomic % .||Ga/O .|
|Ga(L) .||O(K) .||Al(K) .|
% ratio .
|Atomic % .||Ga/O .|
|Ga(L) .||O(K) .||Al(K) .|
|Atomic % .||.|
|Ga(L) .||Ga/O .||Al(K) .||Ga/O .|
|Atomic % .||.|
|Ga(L) .||Ga/O .||Al(K) .||Ga/O .|
These data indicate that as the substrate temperature is increased, the Ga content in the film reaches a peak value at Tsub = 500 °C and then decreases. As explained above, the change in substrate temperature contributes to the kinetics of the sputtered species as they condense on the surface of the substrate. Higher substrate temperatures above 500 °C seems to be unfavorable for condensation of the Ga species onto the substrate, hence the decrease in its content in the film. The Al in the EDS results is due to its presence in sapphire substrate. The contribution of oxygen from both the film and the sapphire substrate as well as lack of a calibration standard limits inference of quantitative compositions of the films using the EDS data.
The plots shown in Figs. 2(a)–2(c) are of data from Ga2O3 films deposited in Ar at different temperatures. Figure 2(a) shows the optical transmittance of these films, which fall at 90–95%. The transmittance of 100% in this plot is from a sapphire substrate without any film. Similar range of transmittance data is obtained for films deposited with the various mixtures of Ar and O2 in this investigation. In order to determine the optical bandgap of the films, a plot of (αhν)2 vs photon energy (hν)—called the Tauc plot—is constructed, as shown in Fig. 2(b). The underlying assumption is that β-Ga2O3 has a direct energy bandgap, in which case, the absorption changes with the bandgap energy according to the expression
where α is the absorption coefficient, hν is the energy of the incident photon, B the absorption edge width parameter, and Eg is the optical bandgap.37,45 The absorption coefficient of the film is determined from the expression
where t is the thickness and A is the absorbance of the film. The bandgap is obtained by extrapolating the linear portion of the plot of (αhν)2 vs hν to the hν axis.Figure 2(c) shows the optical bandgap as a function of the substrate temperature for the films deposited in Ar. These data show that the bandgap decreases from a value of 5.03 eV for the film deposited at 20 °C to 4.96 eV for the film deposited at 600 °C after which the bandgap starts to increase. Each data point represents the average from three runs and the error bars came from the largest deviations from the average values. We observed a similar trend of decrease in bandgap with increase in substrate temperature in films deposited using a mixture of 80% Ar and 20% O2 (data are not shown). Sampath Kumar et al. observed similar trend in which the optical bandgap of β-Ga2O3 deposited using Ar was found to decrease with increase in the deposition temperature.37 The higher values of the bandgap at lower substrate temperatures were attributed to the combined effect of the presence of excess O2 or amorphous nature of the film, which could explain the case for the films we have obtained. In the reported study, films deposited above 300 °C were found to be stoichiometric. Figure 2(d) shows the variation of bandgap with the composition of Ar in the Ar/O2 gas used in the deposition. These films were deposited at 500 °C. The figure shows that the bandgap reaches a maximum value of 5.06 eV at a percent Ar/O2 gas composition of 50/50. Takakura et al. reported a similar trend in β-Ga2O3 films deposited on silicon substrates at room temperature using different compositions of Ar/O2.46 The values they obtained were 5.04, 5.08, and 5.05 eV for compositions of Ar/O2 = 0, 0.4, and 0.6, respectively, which peaks at about Ar/O2 = 0.5 as confirmed by our data shown in Fig. 2(d).
Figure 3(a) shows the XRD scans of Ga2O3 films doped with different concentrations of Sn. These films were deposited at 500 °C in Ar and then annealed in O2 using rapid thermal processor for 3 min at 500 °C following the deposition. The indicated atomic(weight) % of 2.4(6.4), 3.1(9.2), and 3.7(10.2) of the films were determined from measurements using energy dispersive spectroscopy. Also indicated are the thicknesses of the films. During the simultaneous deposition of Ga2O3 and Sn, the power used in depositing Ga2O3 was kept constant for the three samples prepared, while the power used for depositing the Sn film was varied. We note here that the Sn atomic % depends only on the flux of the sputtered Sn and Ga2O3, while the film thickness depends on the duration of the sputtering as well as on the location of the piece used due to nonuniformity across the wafer. The peak with a stronger intensity at 37.7° and the broader one with less intensity at 29.8° correspond to the and (4 0 0) planes of the β-Ga2O3, respectively, as previously discussed. The appearance of the (4 0 0) peak, indicating polycrystalline microstructure of the film, is likely a result of introduction of a significant amount of Sn in the film and the increase in the thickness of the film. The optical transmittance of the films is shown in Fig. 3(b) where it can be seen that the films maintain high transparency. The Tauc plots of these films (not shown) were constructed, and the optical bandgaps obtained is shown in the inset of Fig. 3(b). The bandgap decreases with increase in the amount of Sn in the film. The decrease in bandgap as the concentration of the Sn is increased could be related to the bandgap renormalization effect.47,48 The introduced Sn donor atoms in the films could create energy levels slightly below the conduction band edge leading to changes in the band structure of Ga2O3 and narrowing its bandgap. We performed four-point probe measurements to investigate the electrical properties of these films. All the films, including the ones we attempted to dope with Sn, were found to be highly resistive, with sheet resistance in the order of 109 Ω. This implies that the doping attempt was not successful, perhaps due to unoptimized growth conditions leading to the formation of various defects, SnOx inclusions, inhomogeneous tin incorporation, etc. Additional work is underway to optimize this growth process to improve the electrical conductivity of the film.
The effect of postdeposition annealing was investigated. Films obtained by depositing in Ar at 400 °C were annealed in N2 at different temperatures from 400 to 900 °C for 15 min. The XRD scans of the annealed films are shown in Fig. 4(a) and indicate that the films maintain the β-Ga2O3 orientation. The film annealed at 900 °C, however, exhibited additional XRD peaks at 30.0° and 35.3° belonging to (4 0 0) and (1 1 1) planes of β-Ga2O3, respectively.49 Figure 4(b) shows the variation of the integrated intensity of the peak at 37.2° with the annealing temperature. The data show a gentle decrease in the intensity up to annealing temperature of 500 °C, after which, there is a rapid decrease in intensity, indicating a worsening of the crystalline quality of the film. This suggests that improvement of the crystalline quality of the films is achieved with the deposition at 400 °C and that postdeposition annealing only worsens the crystal quality of the films.
IV. SUMMARY AND CONCLUSIONS
Ga2O3 thin films were deposited on c-plane sapphire substrates by RF magnetron sputtering under different substrate temperatures and different Ar/O2 mixtures. The optimum substrate temperature for the films deposited in Ar was found to be 400 °C, while that for the films deposited in 80% Ar + 20% O2 was 500 °C. The XRD scans of these films showed higher intensity diffraction peaks at 18.3°, 37.1°, and 57.8° belonging to , , and planes of β-Ga2O3, respectively. Films deposited at other temperatures showed either amorphous or polycrystalline properties with additional XRD peaks. Postdeposition annealing of these films at temperatures up to 900 °C in N2 did not produce any further improvement in the crystalline quality of the films. The optical bandgaps of these films were found to decrease with increasing substrate temperature. For the films deposited in different Ar/O2 mixtures, the bandgap reached a maximum value of 5.06 eV at a gas composition of 50% Ar + 50% O2. The addition of increasing amounts of Sn dopants resulted in films with decreasing bandgaps, a consequence of the renormalization effect.
This research was supported by funds from the University Research Council, the College of STEM, and the College of Graduate Studies at Youngstown State University. One author (E.V.) wishes to acknowledge funding support from the National Science Foundation (Grant No. HRD-1432950). Various instrumentations used in this work were acquired with funds from the National Science Foundation (Grant Nos. DMR-1229129, DMR- 1337296, DMR-1006083, and ECCS-0622086).
The data that support the findings of this study are available from the corresponding author upon reasonable request.