Low pressure chemical vapor deposition of-Ga 2 O 3 thin films : Dependence on growth parameters

Low pressure chemical vapor deposition (LPCVD) has been used to produce high quality β-Ga2O3 materials with controllable n-type doping. In this work, we focus on the studies of key LPCVD growth parameters for β-Ga2O3 thin films, including oxygen/carrier gas flow rates, growth temperature, pressure, and the substrate to Ga crucible distance. These growth parameters play important roles during the LPCVD β-Ga2O3 growth and determine the thin film growth rate, n-type dopant incorporation, and electron mobilities. The dependence of the growth parameters on LPCVD of β-Ga2O3 was carried out on both conventional c-plane sapphire and 6 degree off-axis (toward 〈11-20〉 direction) sapphire substrates. To better understand the precursor transport and gas phase reaction process during the LPCVD growth, a numerical model for evaluating the growth rate was developed by using a finite element method and taking into account the gas flow rate, chamber pressure, and chamber geometry. The results from this work can provide guidance for the optimization of the LPCVD growth of β-Ga2O3 with targeted growth rate, surface morphology, doping concentration, and mobility. In addition, β-Ga2O3 grown on off-axis c-sapphire substrates features with faster growth rates with higher electron mobilities within a wide growth window. © 2018 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). https://doi.org/10.1063/1.5054713 β-Ga2O3 with a bandgap of 4.6-4.9 eV represents an emerging ultrawide bandgap semiconductor, promising for radio frequency and high power device applications. The most stable β phase Ga2O3 has a complex monoclinic crystal structure with two Ga sites and three O sites. The large bandgap of β-Ga2O3 renders 2-3 times higher critical electric field (6-8 MV/cm) than those of GaN (3.3 MV/cm) and SiC (2.5 MV/cm).1–3 Consequently, β-Ga2O3 based power electronic (Baliga figure of merit, BFoM ∼ 3200) and high frequency (Johnson’s figure of merit, JFoM ∼ 2850) devices show great promises to outperform the existing technologies based on GaN (BFoM ∼ 846, JFoM ∼ 1090) or SiC (BFoM ∼ 317, JFoM ∼ 278).4 Promising progress has been made for β-Ga2O3 based devices. For example, a β-Ga2O3 enhancement mode metal-oxide-semiconductor field effect transistor (MOSFET) with breakdown voltage > 1 kV,5 a β-Ga2O3 Schottky barrier diode (SBD) with 2.3 kV breakdown voltage,6 and a low pressure chemical vapor deposition (LPCVD) β-Ga2O3 based vertical SBD with a breakdown field of 4.2 MV/cm7 have been demonstrated recently. In addition, the large bandgap corresponding to a transition wavelength at ∼250 nm enables β-Ga2O3 for optoelectronic devices operating in the deep ultraviolet (DUV) wavelength region, e.g., solar blind photodetectors.8 Another key advantage of β-Ga2O3 as compared to the existing wide bandgap semiconductors such as GaN and SiC is its availability of bulk single crystals synthesized by low cost and scalable melt-based growth methods including the floating zone method (FZ),2,9–11 Czochralski (CZ) method,12–14 and edgedefined film-fed growth (EFG) method.15,16 Currently, the assynthesized β-Ga2O3 substrates exhibit n-type conductivity APL Mater. 7, 022514 (2019); doi: 10.1063/1.5054713 7, 022514-1

β-Ga 2 O 3 with a bandgap of 4.6-4.9eV represents an emerging ultrawide bandgap semiconductor, promising for radio frequency and high power device applications.The most stable β phase Ga 2 O 3 has a complex monoclinic crystal structure with two Ga sites and three O sites.2][3] Consequently, β-Ga 2 O 3 based power electronic (Baliga figure of merit, BFoM ∼ 3200) and high frequency (Johnson's figure of merit, JFoM ∼ 2850) devices show great promises to outperform the existing technologies based on GaN (BFoM ∼ 846, JFoM ∼ 1090) or SiC (BFoM ∼ 317, JFoM ∼ 278). 4Promising progress has been made for β-Ga 2 O 3 based devices.For example, a β-Ga 2 O 3 enhancement mode metal-oxide-semiconductor field effect transistor (MOSFET) with breakdown voltage > 1 kV, 5 a β-Ga 2 O 3 Schottky barrier diode (SBD) with 2.3 kV breakdown voltage, 6 and a low pressure chemical vapor deposition (LPCVD) β-Ga 2 O 3 based vertical SBD with a breakdown field of 4.2 MV/cm 7 have been demonstrated recently.
In addition, the large bandgap corresponding to a transition wavelength at ∼250 nm enables β-Ga 2 O 3 for optoelectronic devices operating in the deep ultraviolet (DUV) wavelength region, e.g., solar blind photodetectors. 8Another key advantage of β-Ga 2 O 3 as compared to the existing wide bandgap semiconductors such as GaN and SiC is its availability of bulk single crystals synthesized by low cost and scalable melt-based growth methods including the floating zone method (FZ), 2,[9][10][11] Czochralski (CZ) method, [12][13][14] and edgedefined film-fed growth (EFG) method. 15,16Currently, the assynthesized β-Ga 2 O 3 substrates exhibit n-type conductivity

ARTICLE scitation.org/journal/apm
with a doping concentration in the order of 1-9 × 10 17 cm −3 (N d − N a ).Both intentionally doped n-type (Sn-doped) and semi-insulating (Fe-doped) β-Ga 2 O 3 substrates are commercially available with different crystal orientations.One challenge associated with β-Ga 2 O 3 is its low thermal conductivity which may require thermal management for high power device applications. 4The investigation of β-Ga 2 O 3 heteroepitaxy on foreign substrate such as sapphire can provide additional flexibility of epi-layer transferring to platforms with higher thermal conductivities.
In this work, we have performed a systematic study on the effects of various LPCVD growth parameters on the growth rate, dopant incorporation, and carrier mobility in LPCVD β-Ga 2 O 3 grown on c-plane sapphire substrates with 0 • and 6 • off-axis (towards 11-20 direction) angles.Numerical simulation based on a finite element method was used to simulate the vapor transport process and gas phase reaction in the LPCVD growth system.
Si-doped heteroepitaxial β-Ga 2 O 3 films were grown on both the conventional and off-axis (∆ a = 6 • ) sapphire substrates co-loaded in a custom-built horizontal flow LPCVD system.The system has a precise control of the temperature, gas flow rate, and pressure.High purity metallic gallium (Ga, 99.999 99%) and research grade oxygen (O 2 , 99.999%) were used as the precursors, whereas argon (Ar, 99.9999%) was used as the carrier gas.SiCl 4 (3%, balanced with Ar) was used as the n-type dopant source.The metallic Ga source was placed in a crucible inside the growth chamber, and the substrates were placed horizontally at the downstream.Prior to growth, the substrates were cleaned by an organic solvent in the sequence of acetone, IPA, then sonicated in DI water, and finally blow dried with compressed nitrogen.The room temperature doping concentration and carrier mobility of the as-grown samples were characterized by van der Pauw Hall measurement (HMS 3000 Hall measurement system).The 3D computational fluid dynamics were simulated to extract the gas flow velocity by using the COMSOL Multiphysics software.The 2D finite element method was used to numerically determine the concentration of gas species and its gradient for growth rate estimations.
In order to investigate the effects of Ar and O 2 flow rates on the growth of β-Ga 2 O 3 films, two series of growth experiments were performed with a fixed Ar flow rate of 200 SCCM and 300 SCCM, respectively.For the Ar flow rate of 200 SCCM, the O 2 flow rate increases from 5 to 20 SCCM, and for the Ar flow rate of 300 SCCM, the O 2 flow rate increases from 15 to 40 SCCM.The growth temperature was kept at 900 • C, growth time was 30 min, and SiCl 4 flow rate was set as 0.15 SCCM.Figures 1 and 2 present the effects of O 2 flow rate on the growth rate, carrier concentration, and electron Hall mobility of the as-grown β-Ga 2 O 3 films on c-plane sapphire substrates with ∆ a = 0 • and 6 • for Ar flow rates of 200 and 300 SCCM, respectively.As shown in Figs.1(a) and 2(a), with a fixed Ar flow rate, the growth rate of the films increases with an increase in O 2 flow rate, which suggests that under the investigated growth conditions the growth rate was limited by the O 2 flow rate.The films grown on off-axis substrates show faster growth rates than those on the conventional c-plane sapphire substrates with identical growth conditions.The terrace surface morphology of the off-axis substrates provides preferred incorporation sites for the adatoms which facilitate faster growth rates. 40e electron concentration in both cases shows a decrease trend as the O 2 flow rate increases, as shown in Figs.1(b) and 2(b).This can result from the faster growth rates as the O 2 flow rate increases.In LPCVD, the incorporation rate of dopant atoms per unit volume (N D ) can be predicted from the following equation: 44 The growth temperature typically plays an important role for any semiconductor material.In this study, we investigated the effects of the growth temperature on the LPCVD growth of β-Ga 2 O 3 thin films.The growth temperature is measured by the thermocouple placed at the center of the furnace.A series of samples were grown at different temperatures ranging between 820 • C and 940 • C on c-plane sapphire substrates (∆ a = 0 • and ∆ a = 6 • ).The Ar/O 2 flow rate of 300/30 SCCM was used for all the samples with a fixed dopant flow rate of 0.15 SCCM.Asshown in Fig. 3(a), the growth rate increases monotonically as the growth temperature increases.During the LPCVD growth of β-Ga 2 O 3 , the metallic Ga evaporation rate increases as the temperature increases, and therefore, the available Ga vapor transported to the substrate surface increases as the temperature increases.Overall, the growth rate of β-Ga 2 O 3 on off-axis sapphire substrate is higher than that of the conventional substrate for all the investigated temperatures.However, when the growth temperature is above 900 • C, the growth rate on off-axis sapphire substrate shows saturation.At high growth temperatures, it is reasonable to assume that the growth is mass-transport limited. 45Factors such as gas phase reaction, desorption of adatoms from the substrate surface, and decomposition of Ga

ARTICLE scitation.org/journal/apm
to play more important roles at elevated temperatures, 46 which can lead to the saturation of the growth rate.
The carrier concentration monotonically decreases as the growth temperature increases, as shown in Fig. 3(b).This is believed to be mainly due to the reduced dopant incorporation as the growth rate increases, as predicted in Eq. ( 1). 44Additionally, the growth condition such as growth temperature and chamber pressure can also affect the diffusion of precursor species in the gas phase, surface adhesion, and desorption process.The electron mobilities of β-Ga 2 O 3 films grown on off-axis sapphire substrates are higher than those grown on the conventional sapphire, mainly due to the better crystalline quality.The electron mobility of the films grown on off-axis sapphire increases as the temperature increases to 920 • C but decreases as the growth temperature increases further.Note that as the temperature increases, Ga evaporation rate increases.Meanwhile, the growth temperature also affects the gas phase reaction between Ga and O. Therefore, with the fixed Ar and O 2 flow rate, the atomic ratio of Ga and O on the substrate varies as a function of the growth temperature.As the growth temperature increases to 920 • C, the decrease of the electron mobility can be related to the increase of native defects such as vacancies.On the other hand, the mobility of the films grown on the conventional sapphire shows a weak dependence on the growth temperature.This is mainly due to the existence of high density of dislocations in the β-Ga 2 O 3 films, which limits the electron mobility.Tuning of the growth temperature does not effectively reduce the dislocation density.The electron mobility of β-Ga 2 O 3 films shows weak dependence on the carrier concentration.On the other hand, for the films grown on off-axis sapphire substrates, the electron mobilities have a stronger dependence on the carrier concentration, which indicates that the films have better crystalline quality.
From our studies, the growth pressure also plays an important role for the LPCVD growth of β-Ga 2 O 3 .In this study, we performed a controlled growth with the variation of the chamber pressure at 1.7, 3.2, 5.8, 8.8, and 11.1 Torr on c-plane sapphire substrates (∆ a = 0 • and 6 • ).The growth temperature was set at 900 • C, and the Ar/O 2 flow rate was kept at 200/15 SCCM.As shown in Fig. 4(a), the film growth rate decreases rapidly with the increase in pressure for both types of substrates, which can be due to the dominant gas phase reaction at higher growth pressure conditions. 47Note that at the same pressure, the growth rate difference between the two types of substrates are more obvious, which indicates that the growth pressure dependence on off-axis substrates is more sensitive to the growth pressure.For the films grown on the off-axis sapphire substrates, the doping concentration increases as the chamber pressure increases and reaches the peak value ].On the other hand, for the case of ∆ a = 0 • , the measured carrier concentration increases with an increase in chamber pressure up to 8.8 Torr.However, no continuous films were obtained at the pressure of 11.1 Torr and above.Therefore, carrier concentrations in these films were not included here.This can be due to the suppressed surface diffusivity of the adatoms under higher pressures.The flux of dopant species delivered to the growth surface can be determined from the basic diffusion equation, where D D is the diffusivity of the dopant species in the chamber and ∂C D /∂y is the concentration gradient of the dopant species above the growth surface (y-axis is perpendicular to the surface).The dependence of the pressure on the LPCVD growth of β-Ga 2 O 3 thin films can be resulted from the effects of both parameters: diffusivity and concentration gradient at different pressures.On the other hand, for films grown on both ∆ a = 0 • and 6 • sapphire substrates, the electron mobility reaches maximum at relatively low pressures below 3. , we observe that, with the same growth condition, the carrier concentrations on both types of substrates are similar.This is due to their similar adatom ratio of Si and Ga on both substrates at the same growth condition.
In addition to the parameters of the gas flow rate, growth temperature, and growth pressure, the distance between the substrate and the Ga crucible also plays a critical role in the LPCVD β-Ga 2 O 3 film growth.A series of growths were performed on the off-axis sapphire substrates with controlled pressures at 1.7, 5.4, 11.3, and 15.6 Torr.For each pressure, four samples were placed at horizontally different locations with respect to the Ga crucible.The growth temperature was set at 900 • C, and the gas flow rate of Ar/O 2 was set as 200/15 SCCM. Figure 5 plots the dependence of growth rate G as a function of the source to substrate distance x at different pressures.The general trend shows that the film growth rate decreases exponentially as the source to substrate distance x increases, which is mainly due to the precursor gas phase reactions.To better understand this phenomenon, we conducted a numerical simulation assuming the growth condition to be Ga rich.The growth rate is then primarily determined by the O 2 flux to the growth surface.
For the LPCVD setup, as shown in Fig. 6, the second order Fick's law was used, 48 where C, D, and v represent the mass concentration, the gas phase diffusivity of O 2 , and the average velocity of O 2 in the chamber, respectively.k is the gas phase reaction rate of oxygen.On the other hand, the diffusion flux (J) to the substrate surface can be written as Considering Eqs. ( 3) and ( 4) and assuming that all the oxygen that reaches the substrate surface, or the reactor wall is consumed or deposited, and the radial concentration gradient at the center of the chamber to be zero, the growth rate as a function of x can be written as 48 where ρ film is the volume mass density of β-Ga 2 O 3 , C 0 is the density of oxygen at the location of the Ga source, and M represents the molecular weight of the species.To fit the calculated growth rate from Eq. ( 5) over the experimental data, as shown in Fig. 5, the values of C 0 , v, D, and k are required.Among these parameters, concentration C 0 can be estimated from the ideal gas equation and diffusion coefficient D is calculated by the empirical formula expressed by Chapman-Enskog theory. 49In order to obtain v, a simulation of the gas flow in the chamber was performed using FIG.6.The schematic of horizontal LPCVD chamber illustrating the position of the metallic source, the substrate, and the gas flow direction as marked by orange arrows.The boundary conditions used in the gas transport modeling are indicated.

ARTICLE
scitation.org/journal/apmcomputational fluid dynamics (CFD).The standard gas flow rate at the inlet and the pressure at the outlet of the growth chamber were used as the boundary conditions in the CFD simulation.Figure 7(a) shows the gas velocity contour plot inside the chamber.The velocity was found to be inversely proportional to the chamber pressure, as indicated by the ideal gas law, where F mix is the mass flux of Ar/O 2 gas mixture, C mix is the concentration, and M mix is the average molar mass of Ar/O 2 mixture.
In Eq. ( 5), there are two terms that have the exponential consumption of O 2 species due to gas phase reaction.For the diffusion process between two gas species, it is known that the diffusion coefficient is inversely proportional to the gas pressure P. 50 Therefore, the first term e − π 2 Dx 4vd 2 is found to be independent of pressure P.And in the second term e − kx v , − k v is proportional to P. With higher growth pressure, the slower gas flow takes longer time to transport oxygen to the substrate, during which more oxygen is consumed via the gas phase reaction.The following equation can be obtained by including both terms: Using the four sets of experimental data, as shown in Fig. 5, the decay rate α for each pressure condition can be extracted.We find that α is proportional to the pressure, which indicates that between the two components in α, the second term e − kx v is dominant.This indicates the severe gas phase reactions during the LPCVD β-Ga 2 O 3 epitaxy.By fitting the experimental data, we extracted the first order reaction rate of O 2 k as ∼160 s −1 .With the extracted reaction rate, the partial differential equation ( 3) was solved numerically as a function of v and C 0 for each chamber pressure and O 2 flow rate.Based on the gradient of O 2 concentration, we calculated the dependence of growth rate on the O 2 flow rate and chamber pressure, as shown in Fig. 7(b).With a fixed O 2 flow rate in the growth system, the model predicts that (1) at a relatively higher pressure (>2.8 Torr), the growth rate decreases as the pressure increases, due to the strong consumption of O 2 species via the gas phase reaction; and (2) at a relatively lower pressure range (1.8-2.8Torr), the growth rate decreases as the pressure decreases.Although the gas phase reaction is suppressed at lower pressures, the O 2 concentration is lower and thus limits the growth rate.
In summary, a systematic study was performed to understand the dependence of key growth parameters on LPCVD β-Ga 2 O 3 thin films.The results reveal that the O 2 flow rate, growth temperature, growth pressure, and the distance between Ga crucible and substrate all play important roles for the LPCVD growth of β-Ga 2 O 3 , which determine the growth rate, dopant incorporation, and electron mobilities of the as-grown films.The pressure dependence studies demonstrated that gas transport and diffusion process as well as the precursor gas phase reaction are greatly influenced by the chamber pressure.The studies of the placement of the growth substrates with respect to the Ga crucible revealed an exponential decay of the growth rate along the horizontal chamber, which is mainly due to the precursor gas phase reaction.The use of off-axis sapphire substrates resulted in faster growth rates and higher electron mobilities within a wide LPCVD growth window.The studies and results from this work provide guidance for LPCVD of β-Ga 2 O 3 with targeted growth rate, doping concentration, and electron mobilities, which are indispensable for device applications.

FIG. 1 .
FIG. 1.The dependence of LPCVD n-type β-Ga 2 O 3 thin film (a) growth rate, (b) carrier concentration, and (c) electron Hall mobility on O 2 flow rate, with a constant Ar flow rate of 200 SCCM.The SiCl 4 flow rate was fixed at 0.15 SCCM.All samples were grown at 900 • C for 30 min.

FIG. 3 .
FIG. 3. The dependence of LPCVD n-type β-Ga 2 O 3 thin film (a) growth rate, (b) carrier concentration, and (c) electron Hall mobility on growth temperature.The Ar/O 2 flow rate ratio was fixed at 300/30 and SiCl 4 flow rate was fixed at 0.15 SCCM.All samples were grown for 30 min.

FIG. 4 .
FIG. 4. The dependence of LPCVD n-type β-Ga 2 O 3 thin film (a) growth rate, (b) carrier concentration, and (c) electron carrier Hall mobility in LPCVD grown β-Ga 2 O 3 thin films on chamber pressure.The samples were grown at 900 • C for 30 min with an Ar/O 2 flow rate ratio of 200/15 and a SiCl 4 flow rate of 0.15 SCCM.

FIG. 5 .
FIG. 5.The growth rate (G r ) of LPCVD β-Ga 2 O 3 thin films vs. the distance (x) between the Ga source and substrate with different chamber pressures.All samples were grown at 900 • C with 200 SCCM Ar and 15 SCCM O 2 flow rates.

− π 2 FIG. 7 .
FIG. 7. (a) The velocity contour of mixed gas flowing in the chamber during growth, obtained from 3-D CFD simulation.The Ar/O 2 flow rate ratio was set as 200/15 at the upstream of the tube.The chamber pressure at the downstream was set as 5.4 Torr.(b) The simulated β-Ga 2 O 3 thin film growth rate as a function of the oxygen flow rate and chamber pressure.The argon flow rate was fixed at 300 SCCM.The distance between the substrate and Ga precursor was kept at 5 cm.
40D represents the dopant flux to the growth surface, S D is the sticking coefficient, and G r is the growth rate.N D is inversely proportional to the growth rate G r , and our experiments show a consistent trend as predicted by Eq. (1).The Si dopant incorporation efficiency is similar for β-Ga 2 O 3 films grown on both conventional and off-axis sapphire substrates.With the Ar flow rate of 200 SCCM, as shown in Fig.1(c), the corresponding electron Hall mobility increases as O 2 flow rate increases.This can be related to the reduced carrier concentration with an increase in O 2 flow rate.The electron mobility of the film grown on off-axis sapphire substrate shows enhanced mobilities within the growth conditions investigated.The extended defects originating from the sapphire/Ga 2 O 3 interface tend to tilt and terminate within the 1-2 µm film thickness, which leads to improved crystalline quality and electron mobility.40Onthe other hand, with the Ar flow rate of 300 SCCM, the electron mobilities show a different trend.The electron mobility reaches peak at the O 2 flow rate of 30 SCCM.Note that the limiting factors for electron mobility of β-Ga 2 O 3 films grown on sapphire substrates are more complicated than those grown on native substrates due to the existence of dislocations from the lattice mismatch.The trend indicates that the electron mobilities are limited by not only impurity scattering but also other factors such as dislocations and native defects.And the β-Ga 2 O 3 films grown on the off-axis sapphire substrates still show enhanced mobilities under the investigated growth conditions.
FIG. 2. The dependence of LPCVD n-type β-Ga 2 O 3 thin film (a) growth rate, (b) carrier concentration, and (c) electron Hall mobility on O 2 flow rate, with a constant Ar flow rate of 300 SCCM.The SiCl 4 flow rate was fixed at 0.15 SCCM.All samples were grown at 900 • C for 30 min.ARTICLE scitation.org/journal/apmwhere