Influence of Oxygen Pressure on Growth of Si-Doped β(Al x Ga 1 − x ) 2 O 3 Thin Films on c-Sapphire Substrates by Pulsed Laser Deposition

Ga 2 O 3 is a deep-UV transparent semiconducting oxide being interesting for solar-blind photo detectors e

The wide bandgap semiconductor gallium oxide is more and more explored for high-power electronics, because of its promising material properties, such as beneficial breakdown voltages and large power semiconductor device figures of merit. Further applications include quantum well infrared photodetectors or deep UV-photodetectors like flame sensors for missile plume detection, biological and chemical sensors for ozone detection or gas sensors. In addition, Ga 2 O 3 finds use in touch panel displays, solar cells or optical communications, such as intra-and inter-satellite secured communication systems. [1][2][3] Ga 2 O 3 occurs in different polymorphs, denoted by α, β, γ, δ, , 4 and κ. 5 Because of it is thermodynamical stability, the monoclinic βstructure is the most studied phase, so far. The large bandgap energy of the binary oxide ranges between 4.7 and 5.0 eV. The optical anisotropy and dichroism have been investigated in detail in Ref. 6. The gap can be tuned by alloying with In 2 O 3 or Al 2 O 3 . [1][2][3] Up to now conducting (In,Ga) 2 O 3 thin films with a tailored bandgap could be produced and devices based on the ternary alloy were realized. 2,7 The most challenging part will be the fabrication of conducting (Al,Ga) 2 O 3 thin films. Since intrinsic defects such as oxygen vacancies do not contribute to the conduction of Ga 2 O 3 , 8 extrinsic doping is necessary to realize electrical devices. 9 Possible suitable cations for doping are silicon, 2,10-12 tin 2,13 or germanium 2,14 Hence these cations are part of the present study to create conducting samples. In the β-Ga 2 O 3 crystal structure, Al replaces Ga at octahedral sites, which induces changes of the crystal structure and thereby in the lattice constants as reported by Kranert et al. 15 Schmidt-Grund et al. investigated the relating bandgap dependency on the aluminum content x Al , which can be described by the formula E g = (4.811 + 2.138 x Al ) eV for 0.11 ≤ x Al ≤ 0.55. 16 In the fabrication of thin films by means of pulsed laser deposition (PLD) the oxygen pressure p(O 2 ) and growth temperature T g influence the growth process and the thin film properties independently. For binary β-Ga 2 O 3 thin films, Müller et al. observed decreasing growth rates with decreasing p(O 2 ) 12 and Zha et al. reported decreasing growth rates with increasing T g . 17 For ternary (Al,Ga) 2 O 3 thin films, Wang et al. characterized the influence of the growth temperature and showed that for increasing T g lower growth rates were observed. 18 Wakabayashi et al. showed the influence of p(O 2 ) on PLD grown (Al,Ga) 2 O 3 samples. 19 For low oxygen pressures they observed a decrease of the growth rate and a non-stoichiometric transfer of Ga z E-mail: anna.hassa@physik.uni-leipzig.de and Al atoms. They showed that this process can be suppressed by using an oxygen-radical atmosphere during growth. As a result the growth rates recovered and a stoichiometric transfer was observed. 19 Further, Feng et al. reported a non-stoichiometric cation transfer from target to sample for PLD grown (Al,Ga) 2 O 3 thin films, which was attributed to the formation of the volatile suboxide Ga 2 O. 20 The choice of oxygen pressure and/or growth temperature during PLD influence the growth rate and cation composition of (Al,Ga) 2 O 3 . Therefore, it is of great importance to understand the growth process and the formation of volatile gallium suboxides as a function of p(O 2 ) and T g . The central aspect of this work is the determination of the growth window for which stoichiometric cation incorporation occurs. Furthermore, it summarizes our attempts to fabricate electrically conducting thin films.

Experimental
The samples investigated in this study were grown by pulsed laser deposition (PLD) on (00.1) Al 2 O 3 . The ceramic targets for PLD consist of Ga 2 O 3 with 8.8 at.% Al 2 O 3 and were additionally doped with different amounts of tin, silicon or germanium to improve electrical conductivity. The growth temperature and growth pressure were changed in a wide parameter space as can be seen in Table I. For the ablation we used a KrF excimer laser (248 nm) with an energy density of 2 Jcm −2 at the target and a frequency of 15 Hz. The target to substrate distance was 10 cm. The number of laser pulses was 30300 and the resulting film thicknesses ranges between 400 and 1200 nm. Moreover, series of samples were implanted with tin or silicon using a 1 MeV Tandem accelerator from NEC. For each sample different implantation energies, for Si, 36, 160 and 600 keV and with fluencies of 1.5 × 10 13 , 6 × 10 13 , and 1.5 × 10 14 cm −2 , respectively, were used to obtain a homogeneous implantation profile. For the Sn implanted samples we applied 600 keV Table I. Range of Growth temperature T g , oxygen partial pressure p(O 2 ) and cation composition of the used PLD targets.

Target:
Growth parameters: To activate the implanted Si and Sn, the samples were annealed at approximately 840 • C for one hour in nitrogen ambient. Thin film properties as the alloy composition, structural, optical and electrical properties were investigated by means of energy-dispersive X-ray spectroscopy (EDX), X-ray diffraction (XRD), atomic force microscopy (AFM), and Hall effect measurements. The respective bandgap energies E g and layer thicknesses d were determined by transmission measurements. By extrapolation of (αhν) 2 vs. photon energy to zero the bandgap energies were estimated and d can be calculated from observable layer thickness oscillations using a refractive index of n ≈ 2. 21 There, α describes the absorption coefficient, h the Planck constant and ν the wave frequency. For the transport measurements, ohmic contacts, consisting of 30 nm thick layers of titanium, aluminum and gold (Ti/Al/Au), respectively, were thermally evaporated on the corners of the square samples through a shadow mask. Subsequently, these contacts were annealed at 500 • C for 10 minutes in a nitrogen ambient.

Results and Discussion
The influence of the oxygen pressure during growth will be discussed in the following section based on a growth series deposited at a growth temperature of 670 As the oxygen pressure in the PLD chamber increases, the kinetic energy of the atomic and molecular species arriving at the substrate decreases. The resulting smaller diffusion length leads to the formation of smaller grains, increasing the probability that some of these grains grow in an orientation different from (-201). This effect is in contrast to many other oxide film systems crystallizing in the higher symmetry cubic or hexagonal structure. 24 The correlation of oxygen pressure and crystal growth was reported by Müller et al., who observed for binary Ga 2 O 3 similar behavior. 12 Figure 2a shows the calculated growth rates (r = d/#pulses), bandgap energies and aluminum content from the (Al x Ga 1−x ) 2 O 3 :SiO 2 growth series and its dependence on the growth pressure. The bandgap  Figure 3a and show, that with decreasing growth pressure the gallium (aluminum) content x Ga (x Al ) decreases (increases) for a given growth temperature of 670 • C. The sum of all incorporated cations is always 1 and the silicon proportion ranges between 0.03 and 0.07.
For oxygen pressures between 0.04 and 0.016 mbar a nearly stochiometric transfer of the Ga, Al and Si atoms from target to layer is observed and results in similar growth rates and bandgaps (see Fig. 2). In the oxygen regime below 0.016 mbar, aluminum atoms are preferentially incorporated, because of the higher dissociation energy of the Al-O bond compared to the Ga-O bond. 25 Further, gallium forms volatile sub-oxides being desorbed. As a result, considerably lower amounts of Ga are incorporated into the layers as seen in Figure 3a. The higher aluminum content leads to the observed bandgap increase and the desorption of gallium sub-oxides leads to lower growth rates. Similar desorption processes were reported for molecular beam epi-   [26][27][28] Since the desorption is a temperature dependent process, we investigated the growth rates of samples deposited at three different oxygen pressures (0.006, 0.001 and 3 × 10 −4 mbar) and various growth temperatures (from 400 to 670 • C). For deposition of these sample series we used a target consisting of Ga 2 O 3 + 8.8 at.% Al 2 O 3 + 0.6 at.% SnO 2 . The growth rates, depicted in Figure 3b, decreases with increasing temperature showing that the formation of volatile suboxides is favorable at higher growth temperatures. For the highest investigated oxygen pressure of 0.006 mbar saturation of the growth rate with values between 24.5 to 26.2 pm/pulse is observed for growth temperatures below 550 • C.
Using AFM on areas of 3 × 5 μm 2 the surface morphology and consequently the root mean square surface roughness R q and peak-valleydistance d PV were examined. Figure 4 displays the R q and d PV values obtained from the already discussed (Al x Ga 1−x ) 2 O 3 :SiO 2 growth series. The visible surface effects can be divided into two categories for p(O 2 ) < 0.01 mbar and p(O 2 ) ≥ 0.01 mbar. For oxygen pressures below 0.01 mbar R q and d PV depend on the aluminum content.
With increasing x Al the surface becomes smoother and the roughness decreases from R q = 3.33 nm for x Al = 0.11 to R q = 0.27 nm for x Al = 0.25. For p(O 2 ) ≥ 0.01 mbar the aluminum content saturates and hence the oxygen pressure is determining the surface morphology. The surface roughness increases from R q = 1.72 nm for 0.04 mbar to R q = 4.88 nm for 0.01 mbar.
Currently, the most challenging task for the growth of ternary heteroepitaxial (Al,Ga) 2 O 3 thin films is the development of a doping strategy in order to tailor its electrical transport properties. As part of this work various tests were executed to obtain conducting thin films. All samples investigated were electrically insulating. Concerning the doping of the thin films, we doped samples during growth process in situ and additionally we used ion implantation to create conducting samples, ex situ. For the implantation process a nearly homogeneous dopant concentration of approximately (1-5) ×10 18 cm −3 was realized for both, Si and Sn, as visible in Fig. 5. Before and after annealing, however, all samples remained insulating as soon as Al was alloyed.

Conclusions
We have discussed the influence of growth conditions on the crystallinity, bandgap energy, growth rate, cation composition and surface morphology of PLD-grown (Al,Ga) 2 O 3 thin films on c-sapphire substrates. We described that the preferential incorporation of Al was in tendency observed for low oxygen pressures and/or high growth temperature, which was assigned to differences in cation oxygen bond strength and desorption of gallium sub oxide. Decreasing growth pressure lead to decreasing growth rates and low gallium content. Lower growth temperatures favor the stoichiometric cation incorporation into the thin film layer. Further, many different growth conditions were pointed out to produce insulating samples. It can be concluded that it was not possible to produce conductive samples. This issue must be solved prior to the fabrication of devices based on (Al,Ga) 2 O 3 .