The Influence of Carbon on Polytype and Growth Stability of Epitaxial Hexagonal Boron Nitride Films

Boron nitride (BN) is a promising 2D material as well as a potential wide‐bandgap semiconductor. Chemical vapor deposition (CVD) is commonly used to deposit single layers or thin films of BN, but the deposition process is insufficiently understood at an atomic scale. the CVD of BN is studied using two boron precursors, the organoboranes, triethylborane, and trimethylborane. Using high resolution (scanning) transmission electron microscopy and electron energy loss spectroscopy, this study shows that hexagonal‐BN (h‐BN) nucleates and grows epitaxially for ≈4 nm before it either polytype transforms to rhombohedral‐BN (r‐BN), turns to less ordered turbostratic‐BN or is terminated by a layer of amorphous carbon. this study proposes that the carbon in the organoboranes deposits on the epitaxially growing h‐BN surface and this either leads to the polytype transition to r‐BN, the transition to less ordered BN growth, or complete surface poisoning with carbon terminating BN growth. These results question the use of organoboranes in CVD of epitaxial BN films, and the polytype stability of h‐BN growing on graphene.


Introduction
Boron nitride (BN) has a property envelope including high thermal stability and conductivity, high resistivity, and a large bandgap.]16,17] sp 2 -BN crystalizes as differently layered crystal structures due to the weaker inter-basal plane bonds compared DOI: 10.1002/admi.202400091 to the in-plane bonds. [18]Epitaxial BN has therefore been grown in numerous reports with two different crystal structures or polytypes; hexagonal BN (h-BN) and rhombohedral BN (r-BN).h-BN has an AA'A… stacking sequence in contrast to the ABCA… of r-BN.In h-BN, the basal planes are stacked in the c-direction with a rotation of 180°in relation to each other, while they are shifted along [1 100] by 1.45 Å in r-BN. [19]The challenge in distinguishing between h-BN and r-BN in 0001-oriented films has been described by Chubarov et.al., [20] and maybe the reason for the limited attention r-BN is given compared to h-BN.
In this study, we employ transmission electron microscopy (TEM) based characterization to study CVD-grown epitaxial BN films that are grown using either TEB or TMB as the boron precursor and NH 3 as the nitrogen precursor.The detrimental effect of carbon originating from organic precursors on epitaxial BN films has been previously reported on a global scale, [34,35] we now show where it is located and its influence on epitaxial h-BN thin film growth in CVD.Film growth was conducted on both Al 2 O 3 (11 20) and Al 2 O 3 (0001) substrates, referred to as a-plane and c-plane sapphire, with the BN film nucleating on an intentionally formed AlN buffer.We study how the characteristic morphologies are correlated to the carbon content in the film.We argue that a low level of surface carbon promotes a polytype transition from h-BN to r-BN, while higher levels of surface carbon leads to less-ordered BN growth and eventually this carbon-induced surface poisoning stops the BN growth.

Results
The in-plane epitaxial relationships for BN films grown on each substrate were determined by XRD ϕ-scans in our previous study: r-BN [11 20] ∥ w-AlN [11 20] ∥ -Al 2 O 3 [0001] on the a-plane Al 2 O 3 and, r-BN [11 20] ∥ w-AlN [11 20] ∥ -Al 2 O 3 [10 10] for the c-plane Al 2 O 3 . [36]igure 1a presents a top-view SEM micrograph and Figure 1b is a cross-section TEM micrograph of BN grown on a-plane sapphire using TEB.The top-view micrograph shows a few triangular structures, henceforth referred to as Region I in Figure 1b, surrounded by less ordered growth, referred to as region II.Similar surface morphologies have been observed in several other studies on crystalline BN. [35,[37][38][39] The cross-section micrograph suggests that the triangular area is a truncated triagonal pyramid.The BN thickness in the triangular pyramidal growth region I is ≈35 nm, compared to ≈8 nm in the less ordered region II.Pyramidal BN regions were found in films grown on a-plane and c-plane sapphire grown using TMB and a-plane sapphire grown using TEB.An amorphous interlayer between the AlN buffer layer and crystalline BN is observed in all samples.This interlayer was also found in previous studies of BN thin films grown on sapphire substrates in CVD at similar growth temperatures.[37,40] The BN polytype was determined using HRTEM of region I as shown in Figure 2. The BN film initially nucleates as the h-BN polytype with the AA'…-stacking sequence.However, after ≈4 nm h-BN, the growth undergoes a polytype transition to r-BN with the ABCA…-stacking sequence, that persists throughout the epitaxial BN film growth.The BN film in Figures 1 and 2 is grown from TEB on a-plane sapphire; a similar result from a film grown using TMB is available in the supporting information (Figure S1, Supporting Information).
In region II, the film nucleates as h-BN and grows to a thickness of ≈4-10 nm, with the last few layers showing a rough less-ordered phase, after which epitaxial BN growth is terminated as observed in Figure 3.This initial h-BN nucleation and the following termination of epitaxial BN growth in the less ordered regions was observed regardless of substrate orientation and B precursor.
Figure 4 shows a film grown from TMB c-plane sapphire, containing h-BN equivalent to region II.In the samples grown with this precursor, we also observe the presence of a turbostratic BN (t-BN) phase above the terminated epitaxial h-BN, that prevails with increasing film thickness.It is also possible to identify local r-BN stacking, marked using dashed circles in the HRTEM, which failed to establish a continuous epitaxial growth.An EELS elemental depth profile of the film was acquired to explore the film chemistry given the sudden termination of the epitaxial growth.As can be observed, the film exhibits a negligible carbon content directly after the onset of h-BN growth but increases at ≈4 nm after h-BN nucleation and continues increasing with film thickness until it saturates in the t-BN.Films deposited using TMB show t-BN formation on both substrates, while there is no t-BN formation for the films grown using TEB as observed in Figures 1-3.The EELS depth profiles in the latter case show a steep increase in carbon content after the h-BN growth terminates.The EELS depth profiles for the other samples aligned with their respective HRTEM micrographs, for the less ordered  The EELS elemental depth profile for a triangular pyramidal structure in region I from BN grown from TMB on c-plane sapphire is shown in Figure 5.In contrast to region II, the pyramidal structure seen here incorporates negligible carbon.Note that this  elemental profile is acquired from the same sample as the profile shown in Figure 4.The EELS depth profiles for the pyramidal structures in the region I in the other samples, aligned with their respective HRTEM micrographs, can be found in Figures S5 and S6 (Supporting Information) in the supporting information.Figures S7 and S8 (Supporting Information) show stacked B─K, C─K, N─K, and O─K edge EELS spectra for the two regions depicted in Figures 4 and 5.
Raman measurements from the films as grown on a-plane and c-plane sapphire using both TEB and TMB are shown in Figure 6.The sp 2 -BN Raman peak at 1370 cm −1 is observed for all the grown films and appears unshifted from the corresponding peak in the reference bulk h-BN sample (top spectrum).The broad band peaking ≈1580 cm −1 observed in all samples is indicative of the G band of amorphous carbon (expected between 1500 and 1600 cm −1 ) while the D band is expected at 1350 cm −1 . [41,42]The D band from carbon is partly overlapped by the BN peak, but its contribution can still be distinguished as it extends to lower frequencies than the BN peak.

Discussion
Our results show that the carbon, originating from the employed organoboranes, influences the epitaxial growth conditions for h-BN and r-BN films.There is no literature available that compares TMB and TEB for BN epitaxy.However, a study was conducted to compare h-BN films grown from TMB and B 2 H 6 , which showed that TMB-grown BN films contained 60 times more carbon. [34]igure 6.Raman spectra taken from all BN films and a bulk BN reference sample.From top to bottom -Bulk h-BN sample, BN on c-plane sapphire grown using TMB, BN on c-plane sapphire grown using TEB, BN on a-plane sapphire grown using TMB, and BN on a-plane sapphire grown using TMB.The sp 2 -BN Raman peak at 1370 cm −1 is observed for all the grown films and appears unshifted from the corresponding peak in the reference bulk h-BN sample.The broad band peaking ≈1580 cm −1 observed in all samples is indicative of the G band of amorphous carbon.
This study supports our results that dissociating organoboranes bring carbon to the growing BN surface, which deteriorates the epitaxial growth conditions for h-BN and r-BN.Regarding the organoboranes used in our study, TMB is reported to react with H 2 to release CH 4 and BH 3 in up to three consecutive gas phase reactions, all with negative Gibbs free energy. [43]The more carbon-rich organoborane, TEB is reported to -eliminate one ethyl ligand as C 2 H 4 and approach the surface as B(C 2 H 5 ) 2 H. [44] From the results above, TEB is expected to provide more carbon to the BN surface, while TMB is expected to produce more carbon in the gas phase.A higher BN growth rate has been reported when using TMB versus TEB. [45]Additionally, we grew thicker crystalline BN films using TMB versus TEB, even after considering the difference in growth rate in the experiment.This suggests that surface poisoning from carbon accumulation occurs faster for the TEB process, compared to the TMB process.As TEB brings a more carbon-rich species to the surface, the faster accumulation would deteriorate the epitaxial growth conditions at lower BN thickness.For the epitaxial BN layers, the well-ordered h-BN, and the triangular pyramidal-shaped r-BN structures, a low carbon content is observed.Further evidenced by the thicker h-BN growth observed for region II for the films grown using TMB versus TEB as seen in S2-S4.
In a previous study on epitaxial h-BN growth on c-plane sapphire with AlN buffer, the polytype transition from h-BN to r-BN was suggested to be caused by stress relaxation from the lattice mismatch with the underlying AlN layer. [19]It has been reported that the critical thickness required for the onset of dislocations in h-BN is ≈4 nm.However, our experimental findings indicate that h-BN growth in region II can exceed 4 nm in thickness and this is the region where the h-BN to r-BN transition occurs with the presence of carbon (Figure 4), while the transition from h-BN to r-BN in region I is localized and is suggested to be driven primarily by the stress relaxation.This inconsistency between the two growth regions suggests that stress relaxation from the AlN buffer is unlikely to be the only factor contributing to the nucleation of h-BN and r-BN or the transition from h-BN to r-BN polytypes.From the literature, we find several examples of how surface carbon influences the nucleation of BN polytypes.Epitaxial growth of BN films by magnetron sputtering on graphene-covered Ru(0001) substrates resulted in r-BN growth, while growth on Ru(0001) resulted in a few-layer h-BN growth. [46]In a CVD study using ZrB 2 (0001)/4H-SiC (0001) as substrates for BN growth with TEB, r-BN nucleated directly on a carbon-containing interlayer formed on the ZrB 2 . [35]The growth turned from r-BN to t-BN as the film thickness increased.In another CVD study using TEB and 4H-SiC substrates, TEM showed that r-BN nucleates directly, without any trace of h-BN, on the Sirich (0001) plane, [19] in contrast, t-BN nucleates on the carbonrich (000 1) side of the 4H-SiC. [33]From our results and these examples from the literature, we suggest that r-BN, rather than h-BN, nucleate on surfaces that contain carbon.Furthermore, our results indicate that the co-deposition of carbon species during the growth of h-BN induces a polytype transition to r-BN (Figure 4).A study compared the thermodynamic stability of r-BN to h-BN films at CVD conditions using van der Waals corrected first principles to indicate that the h-BN lattice is less liable to expansion compared to the r-BN lattice. [47]In the presence of carbon, this suggests that r-BN has a higher tolerance to carbon impurities than h-BN and could explain why the h-BN growth transitions to r-BN growth when the carbon content increases.
Given that in our results, the epitaxial r-BN on h-BN film grew to a thickness of only 70 nm in 60 min for the TMB process, rendering a growth rate of 1.16 nm min −1 and supplemented by the study analyzing the time evolution for the surface morphology of BN films, [37] we conclude that the carbon accumulation eventually poisons the surface, terminating further BN growth.This suggests that either our BN film stopped growing well before the end of the 60 min CVD process or nucleated at a gradually decreasing rate as carbon accumulated on the surface.Support for such surface poisoning by carbon is also found in a study that analyzed the influence of growth temperature on the crystallinity of BN in CVD with TEB on c-plane sapphire with an AlN buffer.There the crystallinity then improved up to a temperature of 1500 °C but decreased significantly at 1600 °C. [28]he suggested explanation was that the crystallinity was affected by the change in the crystallinity of the AlN buffer layer.With support from carbon poisoning observed in this study, we argue that at higher temperatures, TEB should undergo a higher degree of thermal decomposition, releasing more carbon species, which are more active as CVD precursors at more elevated temperatures, increasing the potential of TEB to bring carbon species to the BN surface.
Our results have two major implications.First, we question the use of organoboranes as boron precursors in CVD of epitaxial h-BN and r-BN films.While there are numerous reports on the CVD of BN films, especially using TEB, few provide a full characterization of the grown material.[37] We suggest that a borane like B 2 H 6 might be a better, albeit possibly more problematic, boron precursor for epitaxial growth of h-BN and r-BN in CVD. [22,34]Secondly, from our results and with support from Sutter et al., [46] the growth of h-BN films exceeding a few basal planes on graphene seems unfavorable.The surface carbon of graphene, especially if the graphene is thicker than a single layer, will likely steer the growth toward r-BN instead of h-BN.Since r-BN and h-BN have somewhat different properties, e.g., different bandgaps, [18,20] and are challenging to distinguish from each other, [20] we identify the need for careful characterization for applications that require the specific BN polytype.These results highlight the relevance of epitaxial r-BN thin films for graphene technology.While identifying the CVD precursor preferences to grow polytype pure epitaxial h-BN thin films.

Conclusion
We investigated epitaxial h-BN and r-BN films grown in CVD using TMB and TEB as the boron precursors and NH 3 as the nitrogen precursor.From HRTEM and EELS profile analysis, we find that surface carbon, originating from TMB and TEB, deteriorates the epitaxial BN growth conditions.We suggest that a low level of surface carbon leads to a polytype transition from h-BN to r-BN and show that a high level of surface carbon poisons the growth surface, resulting in the growth of turbostratic BN or a layer of amorphous carbon.Our results question the use of organoboranes as a CVD precursor for epitaxial BN and the possibility of growing h-BN thicker than a few basal planes on graphene.

Experimental Section
BN films were deposited using hot wall CVD with two different boron precursors, TMB(B(CH 3 ) 3 ) (99.99% purity, Voltaix/Air Liquide Advance Materials, FL) and TEB (B(C 2 H 5 ) 3 ) (semiconductor grade quality, from SAFC Hitech).The precursor used for N was ammonia (NH 3 , 99.999%, further purified using a getter filter) with H 2 (palladium-membrane purified) as the carrier gas.The B precursors, in their respective processes, were added through a separate quartz liner along with H 2 to suppress the formation of adducts between the precursors.A pyrometer (Heitronics KT81R, calibrated using a silicon melt test) was used to monitor the growth temperature.The process pressure was maintained using a throttle valve before the process pump.
Process parameters were selected from past studies on TEB and TMB. [36,45]For both processes, the base pressure was kept below 2 × 10 −2 mbar.Carrier gas flow was maintained at 5000 sccm H 2 .The growth temperature for the process based on TMB was 1400 °C and BN films were grown for 60 min with a process pressure of 50 mbar, while the process based on TEB growth was done at 1485 °C and BN films were grown for 120 min with a process pressure of 70 mbar.The NH 3 /TMB ratio was 966, while the NH 3 /TEB ratio was 642.As observed from prior studies, the addition of silane (SiH 4 ) had proved useful to promote better crystallinity in the films grown by this process, hence it was introduced during growth using both the B precursors. [48]Silane (SiH 4 , 99.999% purity, 2000 ppm diluted in 99.9996% H 2 ) was supplied with a flow corresponding to 16.5 sccm 2 minutes before BN deposition.
The sapphire substrates were 10 × 10 mm 2 cut from 2″ 330 μm-thick on-axis wafers of either c-plane or a-plane orientation, both from The Roditi International Corp. Ltd.Before immersion into the CVD reactor, substrate cleaning was done by ultrasonication in first acetone, then ethanol at 80 °C for 3 minutes each, followed by standard clean 1 (SC1, NH 3 :H 2 O 2 :H 2 O with relative concentrations 1:1:26 at 80 °C) and standard clean 2 (SC2, NH 3 :H 2 O 2 :H 2 O with relative concentrations 1:1:22 at 80 °C). [49]The substrates were placed in the elliptically shaped susceptor coated with tantalum-carbide (TaC).The two substrates were placed next to each other in the middle of the susceptor for each CVD process.An AlN buffer has been shown to improve crystallinity in the CVD of BN films, [28,37] this AlN buffer was formed in the CVD reactor by introducing the NH 3 after temperature stabilization at 1100 °C.The temperature was then ramped up to the respective growth temperatures for the TEB and TMB processes.The NH 3 flow was maintained for 10 minutes at 1400 °C before introducing TMB and 8 minutes at 1500 °C before introducing TEB to compensate for the difference in temperature ramp-up times.
For TEM analysis, cross-sectional samples were prepared using a focus ion beam (FIB) lift-out technique, employing a CarlZeiss Cross-Beam 1540 EsB system.The BN thin film polytypes and elemental distribution were explored at the atomic scale using high resolution TEM (HRTEM) imaging and electron energy-loss spectroscopy (EELS) techniques.Characterization was performed using the Linköping double Cs corrected FEI Titan 3 60-300 operated at 300 kV.HRTEM images were recorded under negative spherical aberration imaging (NCSI) conditions with slight positive defocus, both ensuring high-resolution images with bright-atom contrast. [50]canning TEM electron energy loss spectroscopy (STEM-EELS) spectrum images of 43×96 pixels were acquired for 1 min using a 0.5 eV/channel energy dispersion, 0.2 s pixel dwell time and a collection semi-angle of 55 μrad of the employed Gatan GIF Quantum ERS post-column imaging filter.Elemental B, C, N, and O distribution maps were extracted from EELS spectrum images by background subtraction, using a power law, and choosing characteristic edges B-K (187-216 eV), C-K (285-311 eV), N-K (401-436 eV) and O-K (538-555 eV) energy loss integration windows.The EELS data presented in the figures was normalized.The figures with EELS data and the correspondingly aligned HRTEM are both taken from similar regions on the sample.
The surface morphology of the films was analyzed using Zeiss Gemini scanning electron microscopy (SEM).The images were captured using an accelerating voltage of 5 kV and an in-lens detector.
The Raman spectra were recorded using a micro-Raman setup containing a single monochromator (Jobin Yvon HR460) equipped with a CCD camera and a 600 grooves mm −1 grating resulting in a resolution of ≈5cm −1 .The 532-nm laser of ≈15 mW power is focused on the sample by an Olympus objective (100X magnification) which also collects the signal.A baseline subtraction and normalization were applied to the data in the plotted figure using Origin, the raw data was available to be viewed in the Supporting Information.

Figure 1 .
Figure 1.BN film is grown using TEB nucleating on the AlN buffer as grown on a-plane sapphire.a) The top view SEM micrograph shows scattered triangular structures in an otherwise seemingly less ordered surrounding film on the surface.b) Cross-section TEM image of such triangular pyramidal structure shows a significantly thicker BN film region for the triangular pyramidal structure compared to its less ordered surroundings.

Figure 2 .
Figure 2. Cross-section HRTEM micrograph for the triangular pyramidal BN growth observed in Figure1for the film grown on a-plane sapphire using TEB as the B precursor.An enlarged image on the right, for the selected section in this growth structure, displays the stacking sequence of the BN basal planes in detail.We observe from the AA'… stacking of the basal planes that the h-BN polytype nucleates and then undergoes a polytype transition into r-BN with the ABCA…-stacking sequence as labeled from the bottom up.The bottom right enlarged section also shows the occurrence of the first r-BN stacking basal plane on top of the h-BN growth.A horizontal white dotted line is used to indicate the thickness where the change from h-BN to r-BN is observed.

Figure 3 .
Figure3.Cross-sectional HRTEM micrograph of the BN film nucleating on AlN buffer layer, grown on a-plane sapphire using TEB as the B precursor for the less ordered film observed in Figure1.The Enlarged image to the right of the highlighted section is used to look at the BN basal planes to determine the BN polytype.We observe from the AA'… stacking of the basal planes that h-BN polytype nucleates on the interlayer above AlN and this continues without any polytype transition.Instead, a rough less ordered phase is observed close to film termination.

Figure 4 .
Figure 4. EELS depth profile aligned with HRTEM micrograph for the less ordered growth in region II shows the elemental distribution as a function of film thickness.The HRTEM micrograph is from a region with a similar film thickness as the EELS data.BN film is grown using TMB as B precursor nucleating on the interlayer above the AlN buffer layer as grown on c-plane sapphire.In the HRTEM, local r-BN stacking is identified and marked with dashed circles.

Figure 5 .
Figure 5. EELS depth profile aligned with HRTEM micrograph for the less ordered growth in the region I show the elemental distribution as a function of film thickness.The HRTEM micrograph is from a region with a similar film thickness as the EELS data.BN Film is grown using TMB as B precursor nucleating on the interlayer above the AlN buffer layer as grown on c-plane sapphire.