Wafer-scale, epitaxial growth of single layer hexagonal boron nitride on Pt(111)

Single layer hexagonal boron nitride is produced on 2 inch Pt(111)/sapphire wafers. The growth with borazine vapour deposition at process temperatures between 1000 and 1300 K is in-situ investigated by photoelectron yield measurements. The growth kinetics is slower at higher temperatures and follows a tanh$^2$ law which better fits for higher temperatures. The crystal-quality of h-BN/Pt(111) is inferred from scanning low energy electron diffraction (x-y LEED). The data indicate a strong dependence of the epitaxy on the growth temperature. The dominant structure is an aligned coincidence lattice with 10 h-BN on 9 Pt(1$\times$1) unit cells and follows the substrate twinning at the millimeter scale.


Introduction
Two-dimensional (2D) materials are expected to become important for the application of concepts for platforms and building blocks in electronic devices beyond the silicon technology, 1,2 in nanophotonics, 3 in membrane technology, 4 for mechanical detectors 5 or as interface systems in electrochemical environments. 6 All these promises rely on the production of the materials at large scale and in sufficient quality. Quality as a degree of excellence may have several aspects, like crystallinity, homogeneity, stoichiometry, defect density etc., after growth, transfer or processing. In the present contribution to this focus on hexagonal boron nitride (h-BN) we report advances and challenges of epitaxial growth of single layer h-BN with chemical vapour deposition (CVD) in ultra-high vacuum (UHV) environment and pick platinum as a prototype growth substrate. The crystal-quality after growth is quantified from low energy electron diffraction patterns.
There are excellent reviews on CVD grown single layer h-BN, covering applications in surface science 7 and beyond. 8 First reports on the adsorption of borazine on Pt(111) and Ru(0001) surfaces date back to 1990: Paffett et al. reported ordered close-packed h-BN overlayers with saturation BN coverages corresponding to one monolayer of h-BN. 9 Work with other precursor molecules like trichloroborazine 10 revealed, compared to the 10 on 9 coincidence lattice 9 a low energy electron diffraction (LEED) pattern rotated by 30 degrees. 11 The h-BN/Pt(111) system was further investigated with low energy electron microscopy (LEEM) that gave direct insight into the growth kinetics which is based on a nucleation and growth mechanism. 12 For many applications h-BN has to be transferred from the growth substrate, where exfoliation from Pt was demonstrated in 2013. 13 The choice of the optimal substrate for growth and exfoliation depends on the bond strength of BN to the substrate, where strong bonding favours the single crystallinity of the h-BN, though strong bonding might be a disadvantage if it comes to exfoliation. The dependence of the h-BN bond strength to a given substrate 14 and the registry of the BN 15 was predicted by density functional theory (DFT). E.g. comparison between Pt and Ru indicates a weaker binding of BN to platinum, 14,16 though it appears more difficult to grow single orientation h-BN on Pt 17 as compared to Ru. 18 Upscaling is a further requirement for material production and was first addressed with the use of single crystalline films as growth substrates. 19 This opened the perspective on high quality single orientation layers of h-BN on the four inch wafer scale. 20 Here we apply Pt(111) on 2 inch wafers as substrates and optimize the growth conditions. A high-temperature growth route facilitates diffusion and stitching of coalescing islands, while using a low pressure of borazine minimizes nucleation seed densities and is necessary to promote systems with minimal residual defect numbers. 21 The h-BN quality is judged from in-situ monitoring of the growth kinetics at different temperatures with photoelectron yield measurements and the crystal-quality of LEED patterns that are recorded after growth.

Preparation of the Pt(111)/sapphire wafers
The platinum thin film preparation follows a scheme reported previously. 22 Double polished, epi-ready, c-plane, two-inch sapphire wafers were cleaned in H 3 PO 4 /H 2 SO 4 (1:3) at 200 • C for 20 min and rinsed in ultra pure water for 3 min in an overflow rinse bath. The sapphire substrates were dried using the Marangoni drying effect. Next, a 500 nm thick Pt layer was deposited on sapphire using e-beam evaporation at 550 • C with an evaporation rate of 0.02-0.03 nm/s. The base pressure of the deposition tool was about 9 × 10 −7 mbar.

Growth and characterization of h-BN/Pt(111)
The single layer h-BN is grown in an ultra-high vacuum, chemical vapour deposition (CVD) cold wall chamber for wafers up to 4-inch and was described previously. 20 Prior to all h-BN preparations the Pt/sapphire substrate was cleaned by argon sputtering, O 2 exposure and annealing cycles to 1123 K until sharp Pt(111) (1 × 1) LEED patterns were observed. Subsequently, h-BN was prepared at different substrate temperatures above 1000 K with borazine (HBNH) 3 as precursor gas at a partial pressure reading of 10 −7 mbar. At these conditions single layer growth without hydrogen incorporation is expected. X-ray photoelectron spectroscopy (XPS) on 10×10 mm 2 samples cut from the 2-inch wafers confirm this and for the

h-BN/Pt(111) growth kinetics
The CVD growth of h-BN on transition metals proceeds via nucleation and growth, as it was observed with scanning tunneling microscopy (STM) for h-BN/Ni(111) 24 or Rh(111) 25 and with LEEM for h-BN/Pt(111) 12 or Ru(0001). 26 The h-BN layer grows from homogeneously distributed nucleation centers. Later, the corresponding h-BN islands coalesce under further precursor gas exposure to a closed layer. Here we apply photoelectron yield measurements 23 for the direct recording of the growth kinetics. Compared to STM or LEEM this method is low cost and more robust in view of temperature and pressure compatibility. The insitu surface probing and an overview of the growth process is sketched in Figure 1. An ultraviolet light flash excites photoelectrons, which are collected with a biased antenna, where the detected electron yield Y is directly related to the h-BN coverage Θ.  and surface status from photoelectron yield. 23 The decrease of the yield upon heating to 700 K indicates desorption of gases from the surface. The final oxygen exposure does not further decrease the yield at 1050 K as it is observed in the first oxygen annealing cycles and marks a clean Pt sample. At this temperature and oxygen pressure the equilibrium oxygen coverage is expected to be low and the work function should not strongly increase. 27 The very small increase of the yield upon the O 2 exposure at 1050 K may be due to an increased density of single Pt atoms on the surface or due to an increase of the background signal. The significant increase of the photocurrent followed by the admittance of 10 −7 mbar (HBNH) 3 during 20 minutes at 1273 K illustrates the formation of a h-BN layer. The decrease of the yield after preparation is also due to the decrease in sample temperature that has, together with the work function an influence on the photoelectron yield. 28 In the following we will analyze the photoelectron yield curve in more detail and show on how much information on the nucleation and growth process may be retrieved.
In brackets we note that if Θ shall be determined for different temperatures Y 0 and Y 1 for the corresponding temperatures have to be known.  Substitution of p dt with dD and integration of Eq.2 yields with the initial condition Θ(0) = 0: It is seen that for small exposures the coverage Θ increases ∝ D 2 , and approaches full coverage with an exponential like in Eqn. 3, but with an offset. Figure 3 shows the almost perfect fit and the two asymptotes to one data set. The model provides a single parameter s/ √ Θ 0 =0.3 that characterizes the growth process. At this stage s and Θ 0 may not be separated and it is convenient to define the parameter s = s/ √ Θ 0 as the effective sticking coefficient. For coverages approaching one we expect all precursor material that sticks on the surface to contribute to the growth, since it reaches the rim of a h-BN patch before desorption and this would imply Θ 0 = 1. Analyzing the data sets from different growth temperatures in Figure 4 with an extended model where Θ 0 is not constant but approaches one exponentially, shows Θ 0 to be larger than one for all exposures and that the χ 2 values of the fits further decrease by about 10 %. Corresponding initial Θ 0 (D = 0) values carry additional information, but here we refrain from refining the model of Eqn. 5.  In Figure 4 four subsequent preparations, each starting from clean Pt(111) but at different growth temperatures are shown. It can be seen that s is highly non-linear with temperature, and does not decrease monotonously with preparation temperature. This indicates different growth regimes and layer qualities. We furthermore note that the goodness of fit increases with increasing temperature. As a measure for the goodness we use χ 2 n which is the χ 2 normalized with the number of data points and the standard deviation of single yield measurements before borazine admission. Large χ 2 n indicate a bad fit and a perfect fit is reached with χ 2 n =1.
Comparison of different photoemission yield measurements provides a direct measure for the reproducibility of a process and allows to forecast the expected quality of the grown material. The course of the work function that is derived from the yield is a quality measure that still may be equivocal. For validation it should be correlated with other means of quality control. In the next section we apply the inspection of the crystal-quality after growth by scanning low energy electron diffraction (x-y LEED).

x-y LEED
After each preparation where the kinetics are shown in Figure 4 the wafer surface was investigated by LEED before it was cleaned again for the next growth cycle. Figure 5 shows LEED data after the preparations of single layer h-BN on Pt(111). The dominant LEED pattern corresponds to an aligned 10 on 9 superstructure. 9 At the growth temperatures of T= 1083 K and 1203 K a minority h-BN phase that is rotated 30 degrees away from the high symmetry substrate direction is found. This reminds to the earlier study where trichloroborazine was used as a precursor. 11 Furthermore, at low temperature a "moiré-ring" with the h-BN lattice constant indicates azimuthally disordered h-BN. Such moiré-rings were also observed for h-BN/Pd(111). 29 However, with increasing preparation temperature the aligned 10 on 9 structure dominates as shown by Preobrajenski et al. 12,17 In the case of the highest growth temperature of T = 1293 K, rotational misalignment disappeared below the detection limit of our LEED instrument. This indicates that aligned h-BN may be grown on Pt(111) with a very high crystal-quality, if the growth is performed at high temperatures.
In order to quantify the crystal-quality of the h-BN on the wafer scale, the sample was scanned in front of the LEED optics on an x-y piezo motor driven stage. In Figure 6 the are checked with scanning x-y LEED that reveals the h-BN quality and twin-domains in the substrate. The best crystal-quality h-BN on Pt(111) is obtained at growth temperatures above 1000 • C.