We report on the growth and characterization of Ge-doped β-Ga2O3 thin films using a solid germanium source. β-Ga2O3 thin films were grown using a low-pressure chemical vapor deposition reactor with either an oxygen or a gallium delivery tube. Films were grown on 6° offcut sapphire and (010) β-Ga2O3 substrates with growth rates between 0.5 and 22 μm/h. By controlling the germanium vapor pressure, a wide range of Hall carrier concentrations between 1017 and 1019 cm−3 were achieved. Low-temperature Hall data revealed a difference in donor incorporation depending on the reactor configuration. At low growth rates, germanium occupied a single donor energy level between 8 and 10 meV. At higher growth rates, germanium doping predominantly results in a deeper donor energy level at 85 meV. This work shows the effect of reactor design and growth regime on the kinetics of impurity incorporation. Studying donor incorporation in β-Ga2O3 is important for the design of high-power electronic devices.
I. INTRODUCTION
Power electronics based on ultrawide bandgap semiconductors (UWBGs) can lead to significant improvement in cost, performance, and energy savings.1 Recently, β-Ga2O3 has emerged as an attractive candidate for next-generation power electronics and deep-UV applications. β-Ga2O3 has a bandgap of 4.8 eV, which leads to a predicted breakdown field of 6−8 MV/cm, which would be significantly larger than other commercial power semiconductors.2 The high breakdown field results in a very large Baliga’s figure of merit (BFOM) that is ∼3500× higher than for traditional semiconductor material such as silicon. A larger value of BFOM signifies smaller on-state conduction losses, which results in improved on/off state device performance. In addition to large BFOM, β-Ga2O3 has unique features not seen in other UWBG materials, such as availability of large-area single crystal bulk substrates and orders of magnitude controllable n-type conductivity.3 Significant progress has been made since the first realization of β-Ga2O3 metal semiconductor field effect transistors in 2012.2 This includes advances in growth,4–7 fabrication,8,9 and understanding the fundamental properties of β-Ga2O3.10,11 Lateral and vertical devices with a critical breakdown field exceeding SiC and GaN have been demonstrated experimentally.12,13 Devices with power densities reaching as high as 1 GW/cm214 and breakdown voltages up to 8 kV (Ref. 8) were already realized. In addition to unipolar devices, rapid progress has been made in studying β-(AlxGa1−x)2O3/β-Ga2O3 heterostructures15–18 and integration of p-type materials with β-Ga2O3.19
Based on its intrinsic material properties, β-Ga2O3-based power devices have the potential to reach breakdown voltages as high as 100 kV.20 For reaching such high breakdown voltages, it is important to achieve high-quality, low-doped thick β-Ga2O3 drift layers. Choosing the correct epitaxial technique is critical in maximizing the performance of β-Ga2O3 power devices. The key criteria for choosing the proper thin film growth technique are growth rate, cost, material quality, and unintentional impurity incorporation. Growth of β-Ga2O3 has been realized using a variety of techniques, such as molecular beam epitaxy (MBE),21 metal-organic chemical vapor deposition (MOCVD),5–7,22,23 hydride vapor phase epitaxy (HVPE),4,24 pulsed laser deposition (PLD),25 and low-pressure chemical vapor deposition (LPCVD).26–29 Growth of β-Ga2O3 has already been studied using MBE for a variety of film orientations [(010), (−201), (001), and (110)] and dopants (Si, Sn, and Ge).30–34 A variety of lateral FETs have been fabricated using MBE-grown β-Ga2O3 films,2,18 whereas PLD is primarily utilized for studying heteroepitaxial growth of Ga2O3 polymorphs35 and achieving heavily doped n+ β-Ga2O3 layers.25 Although high-quality material can be grown using most of the techniques, MBE and PLD are not suitable for the growth of thick drift layers. Only chemical vapor-based deposition techniques can offer growth rates reaching tens of micrometers per hour. In this regard, HVPE and MOCVD are the most promising and widely used methods for studying β-Ga2O3 growth. High-quality β-Ga2O3 films with mobility values close to the theoretical maximum have already been realized using HVPE and MOCVD. Since the start, HVPE is the most studied technique for the growth of thick, low-doped drift layers. HVPE growth of β-Ga2O3 is performed using GaCl3 and O2 as precursors and SiCl4 as dopant gas.36 By growing at high temperatures and large precursor flows, growth rates up to 250 μm/h have been achieved.37 However, MOCVD is preferred for studying β-(AlxGa1−x)2O3/β-Ga2O3 heterostructures16,17 since the growth of β-(AlxGa1−x)2O3/β-Ga2O3 heterostructures is difficult using HVPE. In addition, the HVPE growth of high-quality β-Ga2O3 is largely limited to growth on (001) substrates.4 On the other hand, high-quality films with record high mobility were realized in MOCVD-grown (010) and (100) β-Ga2O3 thin films.5–7,23 Doping densities up to 1013 cm−3 are attained using HVPE by compensating the background donor impurities.
In addition to other CVD-based techniques such as MOCVD and HVPE, LPCVD has emerged as a promising growth technique for studying the growth of β-Ga2O3. LPCVD is a low-pressure growth technique using elemental gallium and oxygen as precursors. Although MOCVD and HVPE are promising for the growth of β-Ga2O3, they use an expensive process and generate harmful by-products. Unlike HVPE and MOCVD, LPCVD precursors are inert and reaction products are nontoxic. This leads to lower maintenance costs and a simpler deposition process. Since LPCVD uses elemental gallium instead of triethyl gallium, the presence of external impurities could be potentially lower in LPCVD-grown β-Ga2O3 thin films. LPCVD growth of β-Ga2O3 has been studied using (010) and (001) β-Ga2O338 and c-plane sapphire substrates.26,27,29 Depending on the configuration, growth rates up to 15 μm/h were achieved on offcut sapphire wafers.26 Typically, LPCVD growths are performed in a mass transport limited regime. In this window, the growth rate dependence on substrate orientation is generally low. High-quality β-Ga2O3 films have been realized on a variety of growth orientations such as (010), (001), and (−201).27,38 Thin films grown on c-plane sapphire contain a large amount of rotational domains, which lead to poor electrical properties. By growing on a sapphire wafer with an intentional 6° offcut, high-quality (−201) β-Ga2O3 films were also realized.27 Room temperature mobility values exceeding 100 cm2/V s were observed in both Si-doped homoepitaxial (010) and heteroepitaxial (−201) grown thin films.27,39 By growing at high temperatures (950–1050 °C), room temperature Hall mobility of 150 cm2/V s was realized for a doping density of 1.5 × 1017 cm−3.39 The measured Hall mobility values are very close to the highest reported mobility from HVPE and MOCVD thin films. Preliminary devices based on LPCVD grown β-Ga2O3 thin films show breakdown fields up to 5.4 MV/cm, which is very encouraging for the realization of high-power β-Ga2O3 devices.40 In spite of huge promise, LPCVD growth studies of β-Ga2O3 are still limited. Most studies use elemental gallium and O2 gas as growth precursors. SiCl4 has been used as the silicon dopant. Studying the effect of using different n-type dopants and studying dopant incorporation as a function of growth conditions are important to realize high-quality β-Ga2O3 thin films.
In this work, we demonstrate n-type doping of LPCVD-grown films using a solid germanium source. We developed a new technique for doping LPCVD-grown β-Ga2O3 thin films. Instead of using a gas-based dopant source, we utilized solid germanium as an n-type dopant precursor. By controlling the growth conditions, n-type doping is achieved for a range of doping concentrations. The carrier concentration, mobility, and activation energy are extracted from temperature-dependent Hall measurements. Using two different growth reactor configurations, the donor incorporation is studied as a function of growth conditions and reactor configuration. The proposed process can be extended to study different kinds of donor and acceptor impurities in β-Ga2O3.
II. EXPERIMENT
Growth of β-Ga2O3 is performed using a custom-built LPCVD reactor as shown in Fig. 1. The setup consists of single-zone furnace with a quartz tube reactor. Elemental gallium (7 N) is used as the gallium precursor. The Ga source is placed inside a quartz boat, which is positioned at the center of the reactor. Given the affinity for the oxidation of gallium, the Ga and O2 vapors need to be separated until they reach the substrate. Two different approaches have been utilized to minimize the amount of parasitic gas-phase prereactions. The first approach consists of supplying oxygen gas through a special delivery tube that physically isolates Ga and O2 molecules (setup 1) until they reach the substrate surface [Fig. 1(a)]. Argon gas is also separately injected to facilitate gallium transport. In setup 2, the gallium source is placed inside a 1-in. quartz tube that is placed inside the larger reactor [Fig. 1(b)]. The argon push gas was connected to the 1-in. tube for controlling Ga transport. The oxygen gas is supplied through a separate line in order to reduce the amount of prereaction.
Schematic of the custom-built Ga2O3 LPCVD reactor with two different configurations: (a) setup 1; (b) setup 2. (c) Cross-sectional SEM image of the sample—grown using setup 2, showing a nominal thickness of 17.5 μm.
Schematic of the custom-built Ga2O3 LPCVD reactor with two different configurations: (a) setup 1; (b) setup 2. (c) Cross-sectional SEM image of the sample—grown using setup 2, showing a nominal thickness of 17.5 μm.
In general, the unintentionally doped (UID) as-grown films were not conductive. An external dopant is required to induce n-type conductivity. Instead of using a gaseous dopant precursor, we used solid source impurity for n-type doping. Given the low vapor pressure of Ge compared to Ga at 900 °C, solid Ge pellets were used as the Ge source. The Ge source did not show noticeable signs of oxidation at the end of the growth run. Also, the Ge source did not show any memory effect between successive growth processes. The amount of doping in the film is controlled by changing the Ge source temperature and the amount of source material in the boat. Both offcut sapphire and (010) Fe-doped β-Ga2O3 substrates are used for the growth of β-Ga2O3. The samples are cleaned with acetone and iso-propyl alcohol followed by a de-ionized water rinse. The samples are placed on a quartz stage that is then placed a few centimeters from the quartz tube at the center of the furnace. The typical parameters utilized for β-Ga2O3 are growth temperature of 800–900 °C, O2 of 3–5 SCCM, argon of 70–140 SCCM, pressure of 0.5–2 Torr, and growth time of 0.5–2 h.
The cross section thickness of the as-grown film was measured using scanning electron microscopy (SEM). Room temperature and Hall measurements were used to characterize carrier concentration and mobility of Ge-doped β-Ga2O3. Ti/Au (50/50 nm) Ohmic contacts were deposited by DC sputtering on the four corners of the sample to form van der Pauw contacts. Low-temperature Hall measurements were performed for measuring low-temperature Hall mobility and donor activation energy. Ni/Au (50/50 nm) layers are deposited for obtaining Schottky contacts to the epilayer. Capacitance-voltage (CV) measurements were used to characterize the electrically active donor concentration of the doped thin films.
There are many previous reports on the presence of an unintentional second donor in β-Ga2O3.5,7,20,22 We adopted a model of two donor levels to describe our temperature-dependent Hall data. The model assumes the presence of two donor energy levels and an ionized acceptor. The charge neutrality relation relating ionized donors, electron concentration, and ionized acceptor concentration is shown in the following equation:5,20
where Nd1 and Nd2 are the densities of the two donor energy levels and Na is acceptor concentration. Ed1 and Ed2 are the donor energy levels. Ef is the Fermi level and n is the carrier concentration. kB is the Boltzmann constant and T is the measurement temperature. The Fermi level separation is calculated using the relation , where Nc is the conduction band effective density of states calculated by assuming m* = 0.28 m0. The donor energy levels Ed1 and Ed2 are obtained by fitting the temperature-dependent carrier density data from 80 to 340 K for each sample. Extending the hall measurement to lower temperatures will improve the accuracy of the fitting parameters (Nd1, Nd2, Na, Ed1, and Ed2).
III. RESULTS AND DISCUSSION
First, we focus on Ge-doped β-Ga2O3 grown using setup 1. The typical growth temperature is in the range of 800–900 °C. Typical growth rates in this configuration are in the range of 0.5–5 μm/h. All the growths were done on c-plane sapphire substrates with 6° offcut toward a-plane. Based on the literature, using a c-plane sapphire wafer with an intentional offcut suppresses the formation of rotational domains, which leads to the improved material quality of β-Ga2O3 films.27
Figure 2(a) shows temperature-dependent Hall data for films grown on a 6° offcut sapphire substrate. By adjusting the Ge source boat position, RT Hall carrier concentrations between 5 × 1017 and 1 × 1019 cm−3 were realized (samples A–D). A control sample was prepared for which no Ge dopant is used. Hall measurements revealed that the undoped sample is electrically insulating. This correlation indicates that Ge is the primary source of n-type conductivity in the LPCVD grown films. The room temperature mobility reduces from 25 to 5 cm2/V s with increasing carrier concentration. The reduction in mobility with increasing carrier concentration is attributed to higher ionized impurity scattering in heavily doped films. Similar trends have been observed in epitaxially grown β-Ga2O3. In general, the low mobility values are attributed to the high density of defects due to the heteroepitaxial nature of the growth. Further growth optimization is required to improve electron mobility. Temperature-dependent Hall measurements indicated an Arrhenius relationship between carrier concentration and temperature. By fitting the low-doped films to charge neutrality equations, activation energies between 8 and 10 meV are obtained. The Hall data could be completely explained by the presence of a single shallow energy level. Films with a lower amount of doping (samples A and B) showed Arrhenius behavior as seen in typical doped semiconductors. In samples with a high amount of doping (C and D), the carrier concentration did not freeze out at low temperatures, indicating degenerate behavior. This observation is attributed to impurity band formation usually observed in heavily doped semiconductors. Similar behavior is observed in heavily Sn-doped substrates when doping exceeded the Mott critical carrier density in gallium oxide (nmott ∼ 2 × 1018 cm−3).41 The low-temperature Hall mobility is observed to peak at 120 K, and this is due to reduced ionized impurity scattering as the measurement temperature reduces. The peak mobility of low-doped films is still comparatively lower than existing literature. This is attributed to the presence of structural defects in the heteroepitaxial β-Ga2O3 epilayer. For samples with very high doping (sample D), the Hall mobility does not show a significant variation across the measurement range.
(a) Temperature-dependent Hall carrier concentration of Ge-doped β-Ga2O3 grown using setup 1 (samples A–D). (b) Corresponding temperature-dependent Hall mobility plotted as a function of temperature.
(a) Temperature-dependent Hall carrier concentration of Ge-doped β-Ga2O3 grown using setup 1 (samples A–D). (b) Corresponding temperature-dependent Hall mobility plotted as a function of temperature.
(a) Temperature-dependent Hall carrier concentration of Ge-doped β-Ga2O3 grown using setup 2 (samples E–H). Dashed lines are fits to the charge neutrality equation. (b) Corresponding temperature Hall mobility plotted as a function of temperature.
(a) Temperature-dependent Hall carrier concentration of Ge-doped β-Ga2O3 grown using setup 2 (samples E–H). Dashed lines are fits to the charge neutrality equation. (b) Corresponding temperature Hall mobility plotted as a function of temperature.
In order to understand the effect of reactor topology on β-Ga2O3 thin film properties, β-Ga2O3 growths were also performed using setup 2 (Fig. 3). Using similar parameters, the maximum β-Ga2O3 growth rate increased sharply from <5 μm s/h to up to 22 μm s/hr. This is attributed to the increased gallium transport achieved by using the argon delivery tube (setup 2). In this study, growth of β-Ga2O3 is performed on both 6° offcut sapphire wafers (sample H) and (010) single crystal wafers (samples E, F, and G) from Kyma technologies. Room temperature Hall measurements show that the carrier concentration increased with increasing Ge source temperature, indicating that Ge is the main source of n-type conductivity. A room temperature Hall charge density of 1.5 × 1017–3 × 1018 cm−3 and Hall mobility of 20–43 cm2/V s were realized. Temperature-dependent Hall measurements were utilized to characterize the dopant density, carrier mobility, and activation energies. The low-temperature Hall data are fit to Eq. (1) from which the donor densities, acceptor density, and donor activation energies (Na, Nd1, Nd2, Ed1, and Ed2) are extracted for all the films. The details of the extracted values are listed in Table I.
Hall fit parameters of β-Ga2O3 epitaxial films grown using setup 2.
Sample name . | Sample RT Hall concentration (cm−3) . | Ed1 (meV) . | Nd1 (cm−3) . | Ed2 (meV) . | Nd2 (cm−3) . | Na (cm−3) . |
---|---|---|---|---|---|---|
E | 1.9 × 1017(010) homoepitaxy | 19.5 | 8.5 × 1016 | 85 | 4.9 × 1017 | 1.8 × 1016 |
F | 3.6 × 1017 (010) homoepitaxy | 19.5 | 1.8 × 1017 | 85 | 2 × 1018 | 8 × 1016 |
H | 1.5 × 1017 (−201) 6° offcut sapphire | 20 | 6 × 1016 | 85 | 3.2 × 1017 | 1.1 × 1016 |
Sample name . | Sample RT Hall concentration (cm−3) . | Ed1 (meV) . | Nd1 (cm−3) . | Ed2 (meV) . | Nd2 (cm−3) . | Na (cm−3) . |
---|---|---|---|---|---|---|
E | 1.9 × 1017(010) homoepitaxy | 19.5 | 8.5 × 1016 | 85 | 4.9 × 1017 | 1.8 × 1016 |
F | 3.6 × 1017 (010) homoepitaxy | 19.5 | 1.8 × 1017 | 85 | 2 × 1018 | 8 × 1016 |
H | 1.5 × 1017 (−201) 6° offcut sapphire | 20 | 6 × 1016 | 85 | 3.2 × 1017 | 1.1 × 1016 |
Unlike the films grown using setup 1, samples E, F, and G showed relatively larger values of mobility at ∼100 K. This is attributed to improved material quality due to the homoepitaxial nature of the growth. However, room temperature mobility is still relatively low compared to MBE-grown lightly Ge-doped β-Ga2O3.31 Sample G, which has a carrier density of 3 × 1018 cm−3, did not show any change with measurement temperature indicating degenerate behavior. This observation agrees well with other reports of heavily doped β-Ga2O3 with doping greater than 3 × 1018 cm−3. Low-temperature fits of the carrier concentration profile showed two different donor energy levels, a shallow Ed1 level at ∼20 meV and a deeper Ed2 level at 85 meV. Based on literature reports, the density of the shallow energy level is in general much higher compared to the deeper level (Nd1 > Nd2, Ed1 < Ed2). However, in this case, Nd2 > Nd1 for all of the three nondegenerate samples (E, F, and H). This effect is also independent of sample orientation [(010) or (−201) orientations]. However, the present deeper donor level is not seen in samples grown using setup 1. The activation energy plots show typical carrier freezeout, with no need for using a second donor in the charge neutrality fit. On the other hand, in the case of setup 2, carrier concentration fits could not be obtained without assuming a large density of Nd2. In the case of sample F, the concentration of Nd2 is 10× greater than that of Nd1. Based on the above analysis, it is clear that the density of Nd2 > Nd1. This indicates that Ge predominantly occupies Ed2 level in films grown using setup 2. Additionally, the energy levels (Ed1, Ed2) did not change significantly between various samples. However, the exact origin of the Nd1 donor levels is not clear. Based on the fitting model, the concentration of Nd1 is lower than 2 × 1017 cm−3. This could indicate the presence of an unintentional background donor unrelated to Ge. If that is the case, we would expect Nd1 to be constant on all the films. However, the concentration of Nd1 clearly increases with Ge-doping. Also, Hall measurements of UID material showed no observable n-type conductivity. This observation refutes the background donor hypothesis. We attribute Nd2 to Ge donors since such a high concentration of Nd2 can only be explained through Ge incorporation.
CV measurements were done on the as-grown films to confirm the carrier concentration of Ge-doped films. The doping depth profile of the lateral Schottky diodes is shown in Fig. 4. The carrier concentration profiles of samples E, F, and H are found close to 2 × 1017 cm−3, which is close to the measured Hall data. Given the thickness of the films and fluctuation in the Ge vapor pressure over time, there could be some doping variation with depth in the film. Nevertheless, the similarity between the CV and Hall data further confirms that Ge is the dominant source of n-type conductivity.
CV extracted apparent carrier density profiles for β-Ga2O3 films grown using setup 2 (Table I).
CV extracted apparent carrier density profiles for β-Ga2O3 films grown using setup 2 (Table I).
The above data suggest that the reactor configuration has a significant effect on donor incorporation. At low growth rates, Ge forms a shallow donor with a single energy level. When the growth rate increased through higher mass transport of Gallium, Ge prefers to occupy a deeper energy level close to 85 meV. This phenomenon is attributed to a change in dopant incorporation due to surface kinetics. A similar phenomenon has been observed in Si-doped MOCVD grown β-Ga2O3.42 According to density functional theory,43 Ge donor atom can occupy either tetrahedral or octahedral gallium sites. The calculated donor formation energy in both sites was found to be very close to each other. We hypothesize that the difference in donor energy levels between the two configurations could be explained by the difference in Ge site occupation. Depending on the growth conditions and reactor configuration, Ge seems to prefer either tetrahedral or octahedral sites, leading to difference in donor energy levels. Further quantitative measurements such as secondary-ion mass spectrometry analysis are required to directly correlate Ge donor density and carrier concentration. The Ge donor site occupation could be potentially identified using x-ray absorption spectroscopy measurements.44
IV. SUMMARY AND CONCLUSIONS
In summary, we developed a new technique for achieving n-type doping in LPCVD-grown β-Ga2O3 thin films using solid source dopants. LPCVD growth of Ge-doped β-Ga2O3 thin films is performed using solid-source germanium as a dopant source. LPCVD growth of Ge-doped β-Ga2O3 was studied using two different reactor configurations (gallium and oxygen delivery tubes). By controlling the Ge vapor pressure, carrier concentration between 1017 and –1019 cm−3 was attained for films on a 6° offcut and (010) β-Ga2O3. At low growth rates, Ge prefers to occupy a shallow donor level with Ed of 8–10 meV. When the growth rate increased beyond ∼5 μm/h, Ge prefers to occupy a shallow level Ed1 (20 meV) and a deep level Ed2 (85 meV). The density of the deeper level was found to be much larger than the shallow level Nd1. This study shows the effect of growth kinetics on Ge incorporation. Studying such phenomena is important for the design of β-Ga2O3-based high-power electronic devices.
ACKNOWLEDGMENTS
This material is based upon work supported by the Air Force Office of Scientific Research under Award No. FA9550-18-1-0507 and monitored by Ali Sayir. Any opinions, findings, conclusions, or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the United States Air Force. Praneeth Ranga acknowledges support from University of Utah Graduate Research Fellowship 2020–2021. This work was performed in part at the Utah Nanofab sponsored by the College of Engineering and the Office of the Vice President for Research. We also thank Jonathan Ogle for his support regarding Hall measurements.
DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding author upon reasonable request.