Atomic Layer Deposition of Ultra-Thin Crystalline Electron Channels for Heterointerface Polarization at Two-Dimensional Metal-Semiconductor Heterojunctions

: Atomic layer deposition (ALD) has emerged as a promising technology for the development of the next generation of low-power semiconductor electronics. The wafer-scaled growth of two-dimensional (2D) crystalline nanostructures is a fundamental step toward the development of advanced nanofabrication technologies. Ga 2 O 3 is an ultra-wide bandgap metal oxide semiconductor for application in electronic devices. The polymorphous Ga 2 O 3 with its unique electronic characteristics and doping capabilities is a functional option for heterointerface engineering at metal-semiconductor 2D heterojunctions for application in nanofabrication technology. Plasma-enhanced atomic layer deposition (PE-ALD) enabled the deposition of ultra-thin nanostructures at low-growth temperatures. The present study used the PE-ALD process for the deposition of atomically thin crystalline ß-Ga 2 O 3 ﬁlms for heterointerface engineering at 2D metal-semiconductor heterojunc-tions. Via the control of plasma gas composition and ALD temperature, the wafer-scaled deposition of ~5.0 nm thick crystalline ß-Ga 2 O 3 at Au/Ga 2 O 3 -TiO 2 heterointerfaces was achieved. Material characterization techniques showed the effects of plasma composition and ALD temperature on the properties and structure of Ga 2 O 3 ﬁlms. The following study on the electronic characteristics of Au/Ga 2 O 3 -TiO 2 2D heterojunctions conﬁrmed the tunability of this metal/semiconductor polarized junction, which works as functional electron channel layer developed based on tunable p-n junctions at 2D metal/semiconductor interfaces.


Introduction
Nanofabrication technologies are aimed at developing novel 2D heterointerfaces to boost the functionalities of low-power electronic devices [1]. The heterointerface polarization at metal-semiconductor junctions is one of the main approaches to the enhancement of the performance of nanoelectronics devices [2]. The high-frequency electronic devices and photonic systems particularly benefit from the polarized 2D heterointerfaces [3]. This tunable polarization plays an important role in the enhanced charge carrier separation and increased mobility at 2D electron channels in metal-semiconductor heterointerface components of high-frequency photonic devices [4]. For example, it is highly demanding in enhancing the optical-to-terahertz conversion efficiencies and, also, in elevating the performance efficiency of photonic devices via various material fabrication techniques and design technologies [5]. For example, the photoconductive electrodes are expected to mitigate the quantum efficiency of terahertz photonic systems [6]. In these devices, the generated charges are transferred through the nanoscaled heterointerfaces within a duration of picoseconds [7]. Heterointerface engineering is one of the main approaches in the fabrication rhombohedral structure and 5.3 eV bandgap has shown an improved performance in the design of Schottky barrier diodes compared with that of ß-Ga 2 O 3 [27]. Therefore, the correct design of the PE-ALD technique enables direct band engineering and the modulation of heterointerfaces for functional applications via well-selected Ga 2 O 3 structures. In the case of Ga 2 O 3 deposited by using tris (2,2,6,6-tetramethyl-3,5-heptanedionato) gallium (III) or [Ga(TMHD) 3 ] under pure O 2 plasma, the films remain fully amorphous at the ALD window of 100 • C to 400 • C [16,21]. This is a technical challenge that suppresses the application of PE-ALD for the deposition of crystalline Ga 2 O 3 films at temperatures lower than 300 • C. The present study has developed a PE-ALD recipe for the moderate-temperature deposition of a crystalline ß-Ga 2 O 3 film a few nanometers thick on wafer-scaled Au substrates for the development of photonic systems. This study shows the capability of the PE-ALD technique in the deposition of crystalline Ga 2 O 3 phases via the control of plasma gas and ALD temperature. This crystalline ß-Ga 2 O 3 film acts as 2D low-resistance photoelectron channels for facile separation and the transfer of photogenerated plasmonic charge carriers at metal-semiconductor heterojunctions. Consequently, polarized 2D structures are developed for the fabrication of low-resistance heterointerfaces for application in photoelectrodes used in high-frequency infrared photonic systems.

Materials and Methods
Four-inch-diameter Si wafers (resistivity of 1 kΩ·cm) with a 150 nm CVD grown SiO 2 layer were used as the substrate. To fabricate the plasmonic Au electrodes, electron beam (EB) evaporation technology was employed to pattern 150 nm thick crystalline Au on the SiO 2 /Si substrate. PE-ALD was employed to deposit~5.0-nm-thick gallium oxide films over Au electrodes. tris (2,2,6,6-tetramethyl-3,5-heptanedionato) gallium (III), [Ga(TMHD) 3 ] (Strem Chemicals Newburyport, MA, United States, 99%) and O 2 plasma were used in the PE-ALD process. The precursor delivery system included ALD valves and automatic pressure control valves (APC) that were kept at 150 • C. Plasma delivery was used to create a laminar flow of the precursors. The precursor pulse time was 2.0 s, followed by plasma operation for 5.0 s. Two different gas compositions of pure O 2 and a mixture of Ar and O 2 (Ar/O 2 = 4:1) were employed individually. The flow rate of the carrier gas for the precursors to the chamber was 30 sccm. The automatic pressure controller was adjusted to provide a deposition pressure of 200 mTorr. The Ga 2 O 3 films were deposited at 150 • C and 250 • C. In the subsequent PE-ALD process, the~20.0-nm-thick amorphous TiO 2 films were deposited on the Ga 2 O 3 layer by PE-ALD (tetrakisdimethylamino titanium "TDMAT" and O 2 plasma) at a temperature of 150 • C. In the PE-ALD process of TiO 2 films, the TDMAT and O 2 plasma durations were set at 0.1 s and 0.6 s, respectively. The purge time was 8.0 s. The film thickness was measured both in situ and ex situ by optical spectroscopy. The ex situ measurements were carried out on the Si witness sample using a Woollam spectroscopic ellipsometer at measurement wavelengths of λ = 200~1700 nm at incident angles of 60, 65, 70 and 75 • . The measured values were fitted with a tri-layer stack Cauchy model. A micro-Raman spectrometer (HORIBA Lab Ram ARAMIS) equipped with λ = 320 nm and 280 nm lasers was employed to extract the Raman and photoluminescence (PL) spectra of the ALD films. XPS studies were used to evaluate the chemical compositions, as well as to determine the valence band maximum (VBM) of the ALD films (XPS, Thermo Scientific theta probe, MA United States). The absorbance spectra of the samples were measured using a UV-Visible spectrometer (Shimadzu 2600, Kyoto, Japan) to evaluate the bandgap of the heterostructured films. To this aim, all films were deposited onto highly transparent glass substrates. The focused ion beam system (FIB) was employed to fabricate the lamella samples for studying the cross-sections of PE-ALD films. Field emission SEM (FESEM, JEOL 7800F) and high-resolution TEM microscopes (TEM, JEM-2100Plus) were employed to investigate the structural characteristics of the synthesized nanostructures. EDS spectrometers of the electron microscopes were employed to analyze the elemental compositions of the heterointerfaces. Hall-Effect measurements (Ecopia, Anyang, Republic of Korea) were employed to measure the conductivity of the samples.
An Autolab Metrohm (PGSTAT204) instrument was used to evaluate the performances of the devices. Tunable LED laser drivers (λ = 750 nm) in combination with the Autolab signal analyzer (PGSTAT204) were employed to measure the photoresistance of photonic devices and resistance of electron channels at metal-semiconductor heterointerfaces.

Results
3.1. PE-ALD of Ga 2 O 3 and TiO 2 Figure 1a depicts the graphical scheme of the custom-made PE-ALD machine, where the ALD cycles and recipe are employed to deposit the Ga 2 O 3 and TiO 2 films. The ALD reactor was equipped with an in situ ellipsometer, turbo pump and inductively coupled with a 300 W plasma generator fed with ultra-high purity oxygen gas (UHP, O 2 ) or Ar/O 2 compound gases. Figure 1b depicts the molecular structure of the ALD precursor of gallium oxide. The partial pressure of the precursor was 200 mTorr, and the flow rate was 30 sccm. Two different gas compositions of pure O 2 and a mixture of Ar and O 2 (Ar/O 2 = 4:1) were employed individually inside of the ALD chamber to investigate the effect the composition of gas has in the ALD process. Before each individual PE-ALD cycle, a 20 s wait time was included to reach to a stable conformal flow before using the plasma. A sequential residual flow of 20 s was employed in each individual stage. Figure 1c shows the saturation curves for the Ga 2 O 3 precursor (Ga(TMHD) 3 ) with constant O 2 plasma exposure. The results showed that a pulse time of 1 s at a pressure of 0.003 mbar was sufficient to reach saturation growth under O 2 plasma radiation ( Figure 1c). It was observed that the growth rate was independent of the composition of the gas inside of the ALD chamber. The measured growth rate was 0.1 Å/cycle (Figure 1c). The inset in Figure 1c shows the typical linear increase of the Ga 2 O 3 film thickness during 200 PE-ALD cycles. This constant growth rate was previously measured and reported for an entire temperature range between 100 • C and 400 • C, confirming a large PE-ALD window for this precursor and designed recipe [16]. The in situ ellipsometer recorded an increase in thickness of the Ga 2 O 3 film vs. the number of PE-ALD cycles. We further investigated the thickness profile of as-deposited films on the surfaces of four-inch wafers after the PE-ALD process. These ellipsometry maps were collected at the incident angle of 65 • . Figure 1d depicts the ellipsometry map of the Ga 2 O 3 films after deposition on the Au substrate. The average thickness values of the Ga 2 O 3 layer changed between 47.02 Å and 48.41 Å (Figure 1d), depicting a high level of controllability of the thickness in this ALD process. In the case of the Ga 2 O 3 film, there was only a 2.6% difference in thickness over the area of the four-inch diameter wafer ( Figure 1d). The ellipsometry map also showed the variations in thickness of the TiO 2 film on the fourinch Au wafers ( Figure 1g). In the case of TiO 2 films, the thickness variations over 70% of the ALD film were less than 1.2% (Figure 1g). With the variable angle spectroscopy technique, we used four incident angles to investigate the changes in the polarization of lights that were reflected from the surfaces of the ALD films. These values were measured and expressed via two values of Ψ (Psi) and ∆ (Delta) [28]. Figure 1e,f depict the variations of Ψ and ∆ for the Ga 2 O 3 film vs. the light wavelengths measured at four different incident angles of 60 • , 65 • , 70 • and 75 • . The same results for the TiO 2 ALD film are presented in Figure 1h,i. It was observed that the practical values of Ψ and ∆ were in high agreement with the predicted model (dashed black lines). The spectral dependencies of Ψ and ∆ were fitted in the model to extract the film thickness using the least squares regression analysis and the weighted root mean square error (MSE) function. MSE described the difference between experimental data and model predicted data. The normal fit was based on Levenberg-Marquart regression algorithm to minimize the least squares difference between the experimental and model generated data [28]. Our fitted results showed that the MSE was 4.31 arbitrary units for the Ga 2 O 3 film and 5.18 arbitrary units for the TiO 2 layer, indicating a high level of consistency between the practical results and generated model data. difference between the experimental and model generated data [28]. Our fitted results showed that the MSE was 4.31 arbitrary units for the Ga2O3 film and 5.18 arbitrary units for the TiO2 layer, indicating a high level of consistency between the practical results and generated model data.  It was observed that the PE-ALD conditions directly affected the structure and properties of deposited Ga 2 O 3 films. The ALD conditions resulted in the growth of either amorphous or crystalline structures. When the ALD process was conducted under O 2 plasma atmosphere and at a low temperature of 150 • C, the extracted film was not crystalline. The FESEM study demonstrated an ultra-smooth film, which is the characteristics of amorphous ALD films before the annealing process. The TEM instrument was employed to investigate the regions of interest in the cross-sections. The cross-sectional TEM image from the heterointerface between the Au sublayer and Ga 2 O 3 films depicted a sharp heterojunction with no clear indication of the crystalline structures ( Figure 2b). The collected SAED patterns showed halo rings (Figure 2c). The diffraction patterns from an amorphous phase will be a ring pattern, but the rings will be rather diffuse ( Figure 2c). Therefore, we estimate that this PE-ALD condition resulted in the growth of an amorphous Ga 2 O 3 thin film. The measurement of the atomic interlayer distance did not show distinctive interlayer orders and arranged structures, confirming the non-crystalline nature of this ultra-thin Ga 2 O 3 film (Figure 2c). After deposition of the TiO 2 film, we analyzed the cross-sections of the Si/SiO 2 /Au/Ga 2 O 3 -TiO 2 heterostructures with EDS equipment from a TEM machine. Figure 2d shows the cross-sectional views of the heterostructures accompanied by their corresponding EDS line elemental analysis. The detailed results of the EDS elemental line analyses showed the distribution of oxygen, titanium and gallium at the Ga 2 O 3 -TiO 2 heterointerfaces ( Figure 2e). The following EDS map analysis results in Figure 2f confirmed the growth of heterostructured Ga 2 O 3 -TiO 2 films with distinguished sharp heterointerfaces. Furthermore, it was observed that the changes in the Ar/O 2 composition at a constant ALD temperature of 150 • C did not affect the level of crystallinity of the as-deposited Ga 2 O 3 films. In the present study, it was found that the ALD parameters had major impacts on the structures of the as-deposited Ga 2 O 3 films a few nanometers thick. Here, it was observed that the increase in ALD temperature had a direct impact on the growth of the crystalline phases in the PE-ALD process.  Figure 3a shows the cross-sectional image of Au/Ga2O3-TiO2 heterostructures. A TEM study showed the distinguishable heterointerfaces between the Au substrate, Ga2O3 crystalline film and amorphous TiO2 structure. In this sample, an ALD temperature of 250 °C was employed during the deposition process of the Ga2O3 film under Ar/O2 plasma. The SEM image from the surface of the Ga2O3 film showed the growth of crystalline grains of Ga2O3 on the surface of the Au substrate. A tangible difference was observed between the surface morphology of this sample and the surface characteristics of the Ga2O3 film deposited at 150 °C and under pure O2 plasma (SEM images in Figure 3b vs. Figure 2a). The   Figure 2a). The EDS analysis of the cross-section of the Au/Ga 2 O 3 -TiO 2 heterointerfaces by TEM depicted the elemental distribution of the Ga, O , and Ti elements on the Au substrate (Figure 3c). It was observed that an increase in PE-ALD temperature did not affect the growth rate of the Ga 2 O 3 film (0.1 Å/cycle). A wide ALD window was also observed previously in the PE-ALD process of Ga 2 O 3 films by tris (2,2,6,6-tetramethyl-3,5-heptanedionato) gallium (III) or the [Ga(TMHD)3] precursor [16]. Via changing the ALD parameters, the structure of the Ga 2 O 3 films considerably changed from an amorphous to a crystalline state. At the PE-ALD temperature of 250 • C and plasma gas composition of Ar/O 2 = 4:1, fully crystalline Ga 2 O 3 films grew on the Au substrate. The ellipsometry measurements showed a refractive index of 1.78 at an incident angle of 630 nm for this film. To compare the refractive index of as-deposited crystalline Ga 2 O 3 films with the conventional annealed Ga 2 O 3 nanostructures, we further operated rapid thermal annealing (RTA) on as-deposited amorphous Ga 2 O 3 films and then measured the refractive index of these films. RTA was performed in an Ar atmosphere and at 550 • C for 5 min. A refractive index of 1.83 was measured for the annealed Ga 2 O 3 film, which was a close number to the refractive index of the as-deposited crystalline Ga 2 O 3 film of the present study. The refractive index of ultra-thin films can be affected by various factors, including stoichiometry, crystallinity and surface morphology. Since higher PE-ALD temperatures changed the level of crystallinity of the Ga 2 O 3 films, it could be concluded that the ideal refractive index of as-deposited Ga 2 O 3 film originated from a higher crystallinity level of this film (ALD temperature of 250 • C). We further measured the optical bandgap values of the as-deposited Ga 2 O 3 films via the UV-Vis technique. The bandgap value of the film deposited at 150 • C was 4.5 eV, which was similar to the bandgap of the amorphous Ga 2 O 3 film [29], while the measured bandgap of the as-deposited Ga 2 O 3 film at 250 • C was 4.71 eV, which was a close number to the reported bandgap of the ß-Ga 2 O 3 film [30]. Therefore, we used TEM equipment to generate highresolution images from the heterointerface between the Ga 2 O 3 and TiO 2 films (Figure 3d). The Ga 2 O 3 film was deposited at 250 • C, while the TiO 2 film was deposited at 150 • C to gain an amorphous morphology of this oxide of titanium. Consequently, a clear heterointerface was detected at Au/Ga 2 O 3 -TiO 2 junctions between the crystalline and amorphous structures (Figure 3e). The amorphous TiO 2 is a functional material for application in a wide range of electronic and photonic devices. Metal/semiconductor heterojunctions play a fundamental role in a wide range of functional electronic and photonic systems, including RF electronics [31], metamaterials [32], memristors and artificial synaptic systems [33][34][35]. The heterointerface between the amorphous oxide structure and metal component plays a fundamental role in the performance of nanoelectronics systems. As an example, the junction between the Au plasmonic structure and amorphous TiO 2 layer has witnessed a considerable charge trapping phenomenon due to the formation of a high Schottky barrier height at the heterointerfaces between the amorphous metal oxide semiconductor and Aubased components [36]. In this condition, the transfer of plasmonic-generated hot electrons from Au nanostructures to the TiO 2 component of a semiconductor device is hindered at low applied voltages [6,36]. Therefore, the development of low-energy electronic systems faces technical challenges. To design tunable metal-semiconductor heterointerfaces with low-resistance characteristics, we aimed to develop a crystalline two-dimensional Ga 2 O 3 electron channel layer between an amorphous TiO 2 film and an Au sublayer. Figure 3e shows the atomic arrangement at the Ga 2 O 3 -TiO 2 heterointerfaces. A clear crystalline pattern was observed in the Ga 2 O 3 film with the interlayer distance of 0.3 nm, attributed to  [37]. The interlayer distance of the crystalline plane was also measured and is shown in Figure 3f, which was close to the value of~0.3 nm. The collected SAED patterns by TEM probe showed a clear crystalline orientation, which was indexed as the (004) crystalline plane of ß-Ga 2 O 3 [37] (Figure 3f).   To investigate the characteristics of the as-deposited Ga 2 O 3 film, a Raman spectrometer was used to analyze the deposited ultra-thin films. Figure 3g depicts the Raman spectra of the Ga 2 O 3 film deposited under pure O 2 plasma (deposited at 150 • C) and the Raman spectra of the other Ga 2 O 3 sample deposited under the mixed Ar/O 2 gas with a 4:1 composition (deposited at 250 • C). The structure of the first sample was amorphous, while the second sample showed the highest level of crystallinity among all deposited samples. The Raman spectra of the as-deposited crystalline Ga 2 O 3 film showed the characteristics of ß-Ga 2 O 3 . The A 2g , A 3g , A 4g , A 5g , A 6g and C 32h characteristic Raman peaks of ß-Ga 2 O 3 are shown in Figure 3g. The A g peaks at different Raman wavenumbers were caused by the vibration of lighter oxygen atoms perpendicular to the c-axis of crystalline ß-Ga 2 O 3 [38,39]. The Raman peak at~450 cm −1 was associated with the C 32h space group of crystalline ß-Ga 2 O 3 . This peak specifically represented the bending of the GaO 4 tetrahedral structure of ß-Ga 2 O 3 [38]. Furthermore, two other peaks were distinguished at~430 cm −1 and 576 cm −1 , attributed to the vibrational mode of the Ga atoms in α-Ga 2 O 3 [40]. The obtained Raman modes from the crystalline film of Ga 2 O 3 were in close agreement with the theoretical and experimental observations of previous studies [41]. Furthermore, the observation of the characteristic Raman modes of the α-Ga 2 O 3 structure confirmed the presence of this polymorphous Ga 2 O 3 in the composition of the thin films. These peaks were not detected in the Raman spectra of the Ga 2 O 3 film deposited at 150 • C and under a pure O 2 atmosphere of the plasma. We further investigated the structure of these two films by HRXRD. The XRD spectra of the Ga 2 O 3 film deposited at 250 • C and under an Ar/O 2 plasma atmosphere (4:1) showed an individual peak related to the (400) plane of ß-Ga 2 O 3 [42]. These results were in harmony with the observed results of SAED patterns, confirming the growth of crystalline-stable ß-Ga 2 O 3 films in these specific ALD conditions. The elemental compositions of the Ga 2 O 3 films were also investigated to understand the effect of the ALD parameters on the quality of the as-deposited ALD films. Carbon was one of the main contaminants that might enter the structures of the ALD semiconductors from the organic ligands of the ALD precursors and, therefore, degrade the high quality of the ALD films. One of the main drawbacks of low-temperature ALD is the presence of residual carbon in ALD films. Therefore, high-resolution C 1s spectra were collected from the surfaces of the Ga 2 O 3 films before and after Ar sputtering in the XPS test (Figure 3i). Despite the presence of atmospheric carbon contamination on the surfaces of ALD films, a trace of residual carbon was not found in any of the deposited films at 150 • C and 250 • C. This could confirm that O 2 plasma successfully removed the metal organic ligands and their corresponding carbons during the ALD process. We also did not detect the presence of other contaminants in the XPS survey spectra of the ALD films. Figure 4a shows the high-resolution Au 4f 5/2 and Au 4f 7/2 XPS peaks at the vicinity of 87 eV and 83 eV. These peaks were collected after the deposition of~5 nm thick Ga 2 O 3 films in different ALD conditions. It was observed that the Au 4f XPS peaks shifted to higher binding energies in the samples deposited under Ar/O 2 plasma atmosphere and at the ALD temperature of 250 • C compared with those of the same samples deposited under O 2 plasma at 150 • C. A positive shift was the indication of the development of different types of heterojunctions. The morphology of the Au substrates were the same in both samples; therefore, the shift in the core level of the Au 4f peak to a higher binding energy was evidence of a strong interaction between Au and the Ga 2 O 3 thin films deposited under different ALD conditions. One can consider that the main difference between the film deposited under Ar/O 2 plasma and the other Ga 2 O 3 films (O 2 plasma) was the presence of ionic defect sites of Ga in the film deposited under Ar/O 2 [43]. Therefore, the shift of the Au 4f peak after deposition of a Ga 2 O 3 film could be due to the presence of defect sites in the PE-ALD Ga 2 O 3 film. Furthermore, it could be attributed to the rearrangement of the Au/Ga 2 O 3 heterointerfaces in the crystalline states. The Ga 2 O 3 films deposited under pure O 2 plasma at 150 • C had amorphous structures, while the film deposited under pure O 2 plasma at 250 • C had a crystalline structure. Therefore, the binding structure was totally different in these two cases. It was expected that the electron donation from the Ga 2 O 3 side to the Au substrate was responsible for this shift of the Au 4f peaks to higher binding energies. We further investigated the Ga 3d peaks in these two samples to understand the possible interactions between Au and the Ga 2 O 3 films. Figure 4b depicts the Ga 3d core-level spectra of the Ga 2 O 3 films deposited by O 2 and Ar/O 2 plasma. The Ga3d peak shifted slightly to lower binding energies for the samples deposited under Ar/O 2 plasma, which could be an indication of electron transfer between Au and the Ga 2 O 3 films (Figure 4b). This could further support the previous concept of an electron transfer from the Ga 2 O 3 films to the Au sublayer.  We further investigated the light-matter interactions in the as-deposited Ga2O3 films. Figure 4e depicts the calculated bandgap of the ß-Ga2O3-TiO2 films, which was extracted from the outcomes of the UV-Vis measurements. A bandgap of 3.35 eV was attributed to the TiO2 film, while a bandgap of 4.71 eV was characteristic of the ß-Ga2O3 film [45,46]. We further investigated the PL spectra of the ß-Ga2O3 film to better understand the photonic properties of the as-deposited crystalline ß-Ga2O3 film. The typical PL spectra of the asdeposited ß-Ga2O3 films are shown in Figure 4f. The PL spectra were characterized by several peaks from the UV to visible and near infrared regions (UV-Vis). Two sharp peaks To investigate the oxidation state of the ALD films, the O 1s spectra were collected and are shown in Figure 4c. The O 1s core level was deconvoluted into two components. The peak at 530.4 eV (upper graph) was attributed to the Ga-O bonds in the Ga 2 O 3 lattice. This peak represented the number of oxygen atoms surrounding the oxidized gallium atoms in the lattice of Ga 2 O 3 . The other peak at 531.2 eV belonged to the oxygen vacancies in the Ga 2 O 3 lattice, representing the oxygen deficiency within the Ga 2 O 3 matrix [44]. It was observed that the position of these peaks shifted to different binding energies at different ALD conditions, representing the effect of O 2 or Ar/O 2 plasma on the formation energy of Ga-O bonds in the ALD films. The shift in the positions of these peaks toward lower or higher binding energies, relative to those of other Ga 2 O 3 films, could originate from differences in stoichiometry, nature and the degree of interactions among various species of thin films. The investigation on the O 1s XPS spectra of these films showed that the film deposited under Ar/O 2 plasma had a higher level of oxygen vacancies (Figure 4c). It was possible that deposition under an Ar/O 2 atmosphere induced a higher level of Ga 3+ with an associated increase in the oxygen vacancies, especially on the surfaces of the ultra-thin Ga 2 O 3 films. We further investigated the Ti 2p peaks after the deposition of the TiO 2 film on the amorphous and crystalline Ga 2 O 3 films. Figure 4d depicts the Ti 2p peaks of the Ga 2 O 3 -TiO 2 and ß-Ga 2 O 3 -TiO 2 heterostructures fabricated at different plasma atmospheres. The study of the XPS spectra of heterostructured Ga 2 O 3 -TiO 2 films showed a gradual shift in the Ti 2p peaks to higher binding energies after deposition of the TiO 2 film on crystalline ß-Ga 2 O 3 . The crystalline ß-Ga 2 O 3 film was deposited under Ar/O 2 plasma at 250 • C. This gradual shift of the Ti 4f peaks was evidence of an electron exchange between the crystalline Ga 2 O 3 and TiO 2 films (Figure 4d).
We further investigated the light-matter interactions in the as-deposited Ga 2 O 3 films. Figure 4e depicts the calculated bandgap of the ß-Ga 2 O 3 -TiO 2 films, which was extracted from the outcomes of the UV-Vis measurements. A bandgap of 3.35 eV was attributed to the TiO 2 film, while a bandgap of 4.71 eV was characteristic of the ß-Ga 2 O 3 film [45,46]. We further investigated the PL spectra of the ß-Ga 2 O 3 film to better understand the photonic properties of the as-deposited crystalline ß-Ga 2 O 3 film. The typical PL spectra of the asdeposited ß-Ga 2 O 3 films are shown in Figure 4f. The PL spectra were characterized by several peaks from the UV to visible and near infrared regions (UV-Vis). Two sharp peaks were centered at~291 nm and~378 nm of the UV region, and two other peaks were centered at~442 nm and~474.1 nm in the blue regions (Figure 4f). A singular peak was also detected at 542.0 nm in the green region of the spectrum, and another peak was also distinguishable at 619.9 nm in the red regions. The relative intensity of the green luminescence was considerably higher than that of the peaks at the blue and red regions (Figure 4f). The PL emission could be attributed to the transition of an electron from the donor band to the acceptor and valance bands of Ga 2 O 3 [47]. In the PL spectra of Figure 4f, the major emission bands were detected at 378.0 nm (L 1 , 3.2 eV), 442.0 nm (L 2 , 2.8 eV), 474.0 nm (L 3 , 2.60 eV), 542.0 nm (L 4 , 2.20 eV) and an individual peak at~291.0 nm (L 5 , 4.26 eV). The detection of UV emission in the ß-Ga 2 O 3 film could be explained by a model that suggested the electron and holes could be de-trapped due to photoexcitation [48,49]. The migration and incidence of these electron/hole pairs formed self-trapped excitons. These excitons recombined and emitted UV photons [47,48]. A similar mechanism for UV emission was previously reported for nanostructured ß-Ga 2 O 3 [49,50]. The UV-green emission in the ß-Ga 2 O 3 structure was related to the recombination of an electron on the donor band of Ga 2 O 3 with another hole formed in the acceptor band of this material [51]. The oxygen vacancies and Ga 2+ formed a donor band, while the acceptor band was formed by a gallium vacancy and pairs of gallium-oxygen vacancy [48]. A simplified model was extracted from the PL spectra of the ß-Ga 2 O 3 film of the present study and is shown in Figure 4f. The donor band (E 1 ) was located 0.04 eV below the conduction band minimum (CBM), which was attributed to the formed oxygen vacancies [48][49][50]. The electron photoexcitation from the conduction band to the valence band was accompanied by the electron relaxation, where the electron could freely move from the conduction band to the donor band before the occurrence of the radiative recombination phenomenon. The following electron/hole recombination between the donor and acceptor bands yielded the generation of UV-green emission in the PL spectra of the ultra-thin ß-Ga 2 O 3 films (Figure 4f). The corresponding analysis of the energy levels in the bandgap of ß-Ga 2 O 3 is presented in the following lines and is depicted in Figure 4f.
The calculated value in (5) was equal to 4.26 eV, which was consistent with the energy level of the detected peak at~291 nm. This emission was related to the recombination of electrons in the donor band with the holes in the valence band edge [48]. The red emission at 619.9 eV could be attributed to the Au substrate and its coupling with the ß-Ga 2 O 3 semiconductor film. The PL spectra of the ß-Ga 2 O 3 -TiO 2 heterostructure also showed two characteristic peaks in the UV regions (Figure 4g). To investigate the type of conductivity of the as-deposited ß-Ga 2 O 3 film by the Hall-Effect measurement technique, the samples were studied at various current ranges, and the results are presented in Figure 4h. The Hall coefficient values experienced a sharp decline from positive values to negative values when the applied current increased and then remained constant for a wide range of applied currents, with slight changes. This observation was evidence of p-type conductivity, which was the characteristic property of the ß-Ga 2 O 3 film in the present study [26]. The study on the valence band maximum (VBM) of the ALD films presented valuable information on the electronic characteristics of the ALD films (Figure 4i). Our studies showed that the Ga 2 O 3 film developed at 150 • C and under pure O 2 plasma had the VBM value of 0.49 eV, while the VBM of the ß-Ga 2 O 3 film deposited at 250 • C and under Ar/O 2 plasma was −0.24 eV. This negative VBM indicated the presence of energy states extruding deeper in the band energy of the ß-Ga 2 O 3 film and was an indication of the p-type conductivity of this film [26,51]. We further studied the VBM of the heterostructured films. It was found that the remarkable difference in the VBM values of the ß-Ga 2 O 3 and TiO 2 films resulted in the shift of VBM of the Ga 2 O 3 -TiO 2 metal-oxide heterojunctions into lower values in the ß-Ga 2 O 3 -TiO 2 heterostructured films. This was more evidence of the formation of the p-n junction at 2D heterointerfaces between p-type Ga 2 O 3 and n-type TiO 2 films. Therefore, the correct materials selection and well-designed fabrication process let us develop a sharp 2D crystalline channel between Au and the amorphous TiO 2 film. In this structure, the atomicscale thin p-type interlayer ß-Ga 2 O 3 film acted as the electron polarization channel. The impediment studies resulted in insights into the heterointerface resistance at the Au/Ga 2 O 3 -TiO 2 junction. Figure 4j (Figure 4j). It is believed that the higher level of crystallinity of ß-Ga 2 O 3 was responsible for the lower resistance at the Au/ß-Ga 2 O 3 heterointerfaces. Table 1 provides detailed information on the ALD conditions and properties of the developed all-oxide heterostructured films.

Photoelectrical Characteristics
To investigate the electrical properties of the as-deposited films, a series of highfrequency photoelectrical measurements were designed and performed. Figure 5a To this aim, a λ = 750 nm near-infrared laser was used to generate light pulses with a frequency of 100 Hz. The films deposited at higher temperatures in the ALD had lower heterointerface resistance values. This observation was attributed to the higher level of crystallinity of the Ga 2 O 3 films deposited at higher ALD temperatures. The following measurements of high-frequency resistance at the Au/Ga 2 O 3 -TiO 2 heterojunction confirmed that the Au/ß-Ga 2 O 3 -TiO 2 junction showed considerably lower heterointerface resistance (Figure 5c). The typical interface resistance value of the Au/Ga 2 O 3 -TiO 2 heterojunction was almost 10 times higher than that of the Au/ß-Ga 2 O 3 -TiO 2 heterostructures at a frequency of 100 Hz (V = 0.1 V). The increased crystallinity enhanced the conductance of the ultra-thin Ga 2 O 3 films, and therefore, the ß-Ga 2 O 3 film acted as the ultra-thin gate channel at the heterointerface between the Au underlayer and amorphous TiO 2 structure. Moreover, the p-type nature of the ß-Ga 2 O 3 film was an important factor for the polarization of the electron-hole pairs at the Ga 2 O 3 -TiO 2 heterojunction. The p-type conductivity of the developed ß-Ga 2 O 3 film enabled the facile and fast separation of the charge carriers at the Au/TiO 2 heterointerfaces and, therefore, suppressed the charge trapping at the Au and amorphous TiO 2 films. When the Au/TiO 2 heterointerface was equipped with the crystalline ß-Ga 2 O 3 film, the electron-hole pairs were separated faster due to the formation of the p-n junction and moved faster to the adjacent components due to the lower resistance of the crystalline ß-Ga 2 O 3 channel. Moreover, it was observed that the type of heterointerface considerably affected the number of trapped charges at the Au/Ga 2 O 3 -TiO 2 heterointerfaces. The values of the trapped charges at a typical Au/Ga 2 O 3 -TiO 2 heterojunction were almost 100 times higher than that of the Au/ß-Ga 2 O 3 -TiO 2 heterostructure (Figure 5d,e). The lower number of trapped charges at the metal/semiconductor heterointerfaces could be attributed to the lower resistance of the ß-Ga 2 O 3 electron channels and their higher polarization capabilities due to the p-type nature of these ultra-thin crystalline layers. Generally speaking, our observations showed that the ALD-developed ultra-thin ß-Ga 2 O 3 film could sufficiently decrease the heterointerface resistance in metal-semiconductor heterointerfaces and facile the charge carrier separation at the developed p-n junction. Therefore, the developed PE-ALD technique is a capable approach for the design of tunable 2D heterointerfaces in nanofabrication technologies.
the ß-Ga2O3 electron channels and their higher polarization capabilities due to the p-type nature of these ultra-thin crystalline layers. Generally speaking, our observations showed that the ALD-developed ultra-thin ß-Ga2O3 film could sufficiently decrease the heterointerface resistance in metal-semiconductor heterointerfaces and facile the charge carrier separation at the developed p-n junction. Therefore, the developed PE-ALD technique is a capable approach for the design of tunable 2D heterointerfaces in nanofabrication technologies.

Conclusions
The PE-ALD technique was employed to develop an ultra-thin crystalline Ga2O3 film at a metal/semiconductor heterointerface to improve the high-frequency electrical performance of Au/TiO2-based devices. It was observed that the ALD parameters, including the ALD temperature and plasma gas composition, had considerable impact on the structures of the ultra-thin Ga2O3 films. It was observed that a decrease in the generated atomic oxygen in the ALD plasma had a direct effect on the formation of a thermodynamically stable ß-Ga2O3 phase. Therefore, the 5-nm-thick stable ß-Ga2O3 films were developed during the PE-ALD process under Ar/O2 plasma with a ratio of 4:1. The ALD temperature also had a direct impact on the level of crystallinity of the Ga2O3 films, where the ALD film deposited at 150 °C was totally amorphous while the Ga2O3 film developed at 250 °C showed crystalline characteristics. Consequently, the ultra-thin p-type ß-Ga2O3 film was deposited at

Conclusions
The PE-ALD technique was employed to develop an ultra-thin crystalline Ga 2 O 3 film at a metal/semiconductor heterointerface to improve the high-frequency electrical performance of Au/TiO 2 -based devices. It was observed that the ALD parameters, including the ALD temperature and plasma gas composition, had considerable impact on the structures of the ultra-thin Ga 2 O 3 films. It was observed that a decrease in the generated atomic oxygen in the ALD plasma had a direct effect on the formation of a thermodynamically stable ß-Ga 2 O 3 phase. Therefore, the 5-nm-thick stable ß-Ga 2 O 3 films were developed during the PE-ALD process under Ar/O 2 plasma with a ratio of 4:1. The ALD temperature also had a direct impact on the level of crystallinity of the Ga 2 O 3 films, where the ALD film deposited at 150 • C was totally amorphous while the Ga 2 O 3 film developed at 250 • C showed crystalline characteristics. Consequently, the ultra-thin p-type ß-Ga 2 O 3 film was deposited at 250 • C and under Ar/O 2 plasma to develop the electron channel layer at the Au/ß-Ga 2 O 3 -TiO 2 heterojunctions. This ultra-thin polarized heterointerface showed a distinguished performance as a low-energy heterointerface barrier for high-frequency electronic devices. This ALD approach for the design and fabrication of a self-polarized ultra-thin election channel layer could be expanded to the design and development of various types of heterojunctions and tunable photonic heterointerfaces for a wide range of functional applications in low-power semiconductor devices.