Chemical vapor deposition of sp 2-boron nitride films on Al 2 O 3 ( 0001 ) , ( 11 2 0 ) , ( 1 1 02 ) and ( 10 1 0 ) substrates

Thin films of boron nitride in its sp2-hybridized form (sp2-BN) have potential use in UV-devices and dielectrics. Here, we explore chemical vapor deposition (CVD) of sp2-BN on various cuts of sapphire; Al2O3(112̅0), Al2O3(11̅02), Al2O3(11̅00) and Al2O3 (0001) using two CVD processes with different boron precursors; triethylborane (TEB) and trimethylborane (TMB). Fourier transform infrared spectroscopy (FTIR) showed that sp2-BN grows on all the sapphire substrates, using X-ray diffraction (XRD), 2θ/ω diffractograms showed that only Al2O3(112̅0) and Al2O3(0001) renders crystalline films and using phi(ɸ)-scans the growth of rhombohedral polytype (r-BN) films on these substrates is confirmed. These films are found to be epitaxially grown on an AlN interlayer with a higher crystalline quality for the films grown on the Al2O3(112̅0) substrate which is determined using omega(ω)-scans. Our study suggests that Al2O3(112̅0) is the most favorable sapphire substrate to realize the envisioned applications of r-BN films.

We have previously employed an AlN buffer layer to facilitate epitaxial growth of sp 2 -BN on c-cut sapphire. 13,21 AlN growth on r-cut Al2O3 also showed a-plane (1120) oriented growth suggesting growth orientations can be altered based on the substrate used. 22 Other studies of AlN growth on c-cut and a-cut Al2O3 showed that the growth direction was identical in both cuts of Al2O3. 23 This is relevant since the surface nitridation of Al2O3 is observed and expected for thermal CVD at such high temperatures.
Notably, a recent study on the surface properties of the substrate comparing h-BN growth using diborane (B₂H₆) on a-cut and c-cut Al2O3 revealed the former to be less likely to modify under growth conditions in the absence of an AlN buffer layer. 19 Its noteworthy that when B₂H₆, a carbon free boron precursor, was used, as opposed to trimethylboron (TMB) and triethylboron (TEB), to deposit BN without nitridizing the sapphire, h-BN films are observed. This study aims to investigate possible growth of sp 2 -BN using thermal CVD, comparing the difference of growth on a-cut, r-cut, m-cut and c-cut substrates. This growth is done using two different boron precursors TMB and TEB which have been investigated in previous studies. 17,18 Crystalline growth in the polytype r-BN was observed in these studies. The most suitable substrates for specific growth applications will be realized as we identify the major characteristic difference between each substrate.
The boron precursors, in their respective processes, are added through a separate quartz liner along with H2 to avoid the formation of intermediate adducts prematurely. A pyrometer (Heitronics KT81R, calibrated by a silicon melt test) is used to monitor the growth temperature. The process pressure is maintained using a throttle valve. As observed from prior experience, the addition of silane (SiH4) has proved useful to promote better crystallinity in the films grown by this process hence added while growing using both precursors. 24 Optimized processes parameters are selected from previous studies on TEB and TMB. 17,18 For all processes, the base pressure is kept below 2 x 10 -2 mbar. Carrier gas flow is maintained at 5000 sccm H2. The growth temperature for the process based on TMB was 1400°C and BN is grown of sp 2 -BN 60 min with a process pressure of 50 mbar, while the process based on TEB growth was done at 1500°C and BN is grown for  25 The substrates are then placed in the heating part of the reaction cell(susceptor), this is a elliptically shaped and this susceptor is coated with tantalum-carbide (TaC). The depositions on the different substrates are done separately as they are always placed in the same position in the susceptor for each process. Prior to BN deposition, the α-Al2O3 substrates were heated to 1100 °C for 5 min in H2 gas, after which NH3 was introduced and the temperature is ramped to the selected growth temperatures and maintained for 10 min to form an in situ aluminum nitride buffer layer as previously reported. 9,16

B. Film characterization
Fourier transform infrared spectroscopy (FTIR) reflectance spectra were measured using a Bruker Vertex70 FTIR spectrometer, with a globar MIR light source and a DLaTGS detector, the software used here is Bruker OPUS 7.5. The incident spolarized light at an angle of 60° with respect to the sample surface normal. The spectra were acquired at room temperature, after a 30 min N2 purge, with 2 cm -1 resolution and averaged over 50 scans. A thin film of gold was used as reference. The FTIR peaks were fitted using a Lorentzian profile and linear base line using SciDAVis software (version 1.22). X-ray diffraction (XRD) was used to investigate the structural properties and the crystallographic relationship to the buffer layer and the film orientation. All diffractograms were recorded using Cu Kα radiation (Cu Kβ removed by a nickel filter).
The 2θ/ω diffractograms were recorded in a PANAlytical X'Pert Pro diffractometer with a Bragg-Brentano HD and 1/2° slit as primary optics and X'celerator detector with a 5 mm anti scatter slit on the secondary side. The azimuthal scans (ɸ-scans) were recorded in a Phillips X'Pert MPD diffractometer with crossed slits (2 × 2 mm 2 ) and 1/2° slit as primary optics and proportional detector (PW1711/96) equipped with parallel plate collimator on the secondary side. Same instrumentation is used to obtain XRD ω-scan ToF-ERDA spectra data is then converted into relative atomic concentration profiles using the Potku code. 26 The surface morphology of the sp 2 -BN films was analyzed using scanning electron microscopy (SEM)1550 Gemini. The microscope was operated with conventional and immersion lens (in-lens) secondary electron detectors and an accelerating voltage of 3 kV.  suggested to indicate the crystallinity of these samples. 28 Since, for amorphous material there is a variation in the frequency of the vibrational modes which gives broader peaks compared to crystalline films, suggesting that the sp 2 -BN films grown on r-cut and m-cut sapphire are amorphous. precursor have comparatively lower intensities. From previous studies we know that samples grown using the TMB precursor have higher growth rate of sp 2 -BN and hence more material is deposited during growth which is reflected in the peak intensity. In figures 2(c) and 2(d) there are no BN peaks, confirming the FTIR results that the BN films present are amorphous on r-cut and m-cut sapphire. In figure 2(d), i.e. the r-cut sample shows AlN (112 ̅ 0) growth, this growth orientation on this substrate has been observed seen in past studies 22,31 and for the m-cut samples in fig. 2(d) AlN(1000) and (101 ̅ 3) growth is observed, the latter of which has been reported previously. 32 observed in previous studies, hence 6 peaks are expected. 13,17 Since sp 2 -BN diffracts poorly, longer scan times were used for these ɸ-scans which is why a background subtraction and signal smoothening is applied. The low intensity for a few peaks observed is due to the high tilt sensitivity of the sample at higher ɸ angles. Twinning of AlN grown on c-cut Al2O3 is observed. 33 XRD ω-scans on the films grown using TMB provide crystal properties and shows the difference in crystalline quality of the films grown. The a-cut sample shows lower FWHM of 0.985° compared to 1.155° of the c-cut sample which is a difference of about 15%. Further confirming suggestions of a more suitable substrate in a-cut Al2O3 for the growth of sp 2 -BN. 19 Small deviation from the expected ω peak angle suggests very slight buckling in the film surface with respect to the substrate. Broadness of the peaks suggest a less ordered film with respect to the substrate. Samples grown using TMB are compared in the ɸ-scans and ω-scans due to the difference in amount of sp 2 -BN present as described above. Figure. 4. XRD ω-scans of the crystalline films grown on c-cut and a-cut Al2O3 using the TMB precursor. Dotted line corresponds to the ω-angle which is half of the 2θ-angle for the sp 2 -BN (000ℓ) peak at 26.5 Figure 5. ToF-ERDA analysis is used to calculate the elemental compositions of the samples, (a) Samples grown using TMB precursor (b) Samples grown using TEB precursor. Here the substrates used, Al2O3 (0001), Al2O3 (112 ̅ 0), Al2O3 (11 ̅ 02), Al2O3 (101 ̅ 0) are called c-cut, a-cut, r-cut and m-cut respectively As observed from the ToF-ERDA results in figures 5(a) and 5(b), the samples grown using TEB as the boron precursor show more consistent stoichiometry and a better nitrogen to boron (N/B) ratio compared to the TMB grown samples. For the TEB samples, growth on the c-cut substrates display the least amount of oxygen contamination. Samples grown using TMB as the boron precursor display varying quantities of oxygen in the films, at this stage it is assumed to be a function of the difference in the amount grain boundaries on the films, which leads to oxidation post deposition. The m-cut sample grown using the TMB precursor displays film growth with least amounts of oxygen which could potentially be due to the denser amorphous microstructure on this cut. Looking at the film surface structure would provide further insight (Fig. 6). Hydrogen content is very low (<0.2%) and thus omitted. Overall, lower quantities of C and higher quantities of B is seen on the film grown using TEB compared to TMB which agrees with results from past studies. 34 Figure 6. Plan-view SEM micrographs of film growth on all cuts of Al2O3 (top to bottom) grown using TMB (left column) and TEB (right column) as the boron precursor.

D. Surface Morphology
Plan-view SEM micrographs reveal that for the substrates that showed r-BN growth, being the c-cut and a-cut Al2O3 grown using TEB precursor, forms films with previously observed triangular shaped grains of comparable sizes surrounded by a less ordered material with substantial whisker growth. 21 For the samples grown using TMB, the c-cut sample show similar crystal islands but not shaped as consistently as when deposited from TEB. On the a-cut sample, there are no ordered crystals observed on the plan-view, which could be attributed to deposition of disordered material on the topmost layers. The plan-view SEM of sp 2 -BN grown films on the r-cut and m-cut Al2O3 is consistent with the expected amorphous film structure suggested in the XRD (Fig.2) and FTIR ( Fig.1) results.

E. Crystallinity
From XRD results on specifically the c-cut and a-cut Al2O3, we can conclude that the out of plane crystalline growth direction for r-BN is still identical, being sp 2 -BN (000ℓ) and sp 2 -BN (0002ℓ), between the two substrates. These substrates also show identical AlN 0002 and AlN 0004 peaks, this result was also observed in a past study on AlN growth on these Al2O3 substrates. 23 The potential for a new growth direction was not realized, but a-cut Al2O3 showed comparable crystal properties to the c-cut with better crystalline quality as supported by the ω-scan (Fig. 4) for the growth of r-BN films.
These results agree with suggestions from a recent surface study of BN growth on a-cut Al2O3, which described it to be more suitable for the growth of h-BN. 19 The presence of AlN 0002 and AlN 0004 peaks on both c-cut and a-cut Al2O3 evidently dictates the epitaxial r-BN film growth on these substrates. Supported by previous studies that have shown that h-BN initially nucleates epitaxially followed by r-BN epitaxial film growth on AlN (0001) grown on c-cut sapphire. 35 In contrast, for r-cut and m-cut Al2O3 we don't detect AlN 0002 and AlN 0004 peaks, instead we observe AlN (112 ̅ 0) and mostly AlN (101 ̅ 3) growth on these cuts respectively. Since the crystal direction of the AlN film dictates the growth direction of the r-BN film, we would expect growth of crystalline BN films on these substrates. Instead, a-BN growth is observed on r-cut and m-cut substrates.
We speculate that this could be explained due to the higher surface energies of the offaxis BN planes preventing crystalline BN film growth for these substrates.

IV. CONCLUSIONS
We report the differences in CVD growth of sp 2 -BN films with TMB and TEB on different Al2O3 substrates (Al2O3(112 ̅ 0), Al2O3(11 ̅ 02), Al2O3(101 ̅ 0) and Al2O3 (0001) called a-cut, r-cut, m-cut and c-cut respectively). This result is promising towards improving crystal quality of r-BN for envisioned applications and studies in the future.