Synthesis of large single crystals of boron nitride for electronic and photonic applications
For more than 50 years, fine-grained polycrystalline hexagonal boron nitride (hBN) has been employed as a refractory ceramic, taking advantage of its high thermal stability, low reactivity, high thermal conductivity, high electrical resistivity, and good mechanical properties.1 Highly pure single crystals of hBN with a low defect density have the potential to make new types of electronic, optoelectronic, and nanophotonic devices feasible. Such devices require single crystals of hBN to optimize their charge transport and optical properties. Specific devices envisioned include emitters in the deep UV region—at wavelengths below 250nm—such as LEDs and diode lasers,2,3 as well as neutron detectors,4 flat optical devices, chemical sensors, and photonic interconnects.5 Single-crystal hBN is also the best substrate and gate dielectric for atomically thin 2D electronic materials such as graphene,6 phosphorene, and molybdenum disulfide.7
Much recent research has focused on the synthesis of hBN crystals. However, the crystals produced so far have been atomically thin or consisted of a few atom layers. In contrast, our research aimed to produce hBN crystals with a larger volume by growing them from solution, because this process is capable of producing crystals at near-atmospheric pressure and at temperatures of 1300–1750°C, which is well below the melting point of boron nitride.8,9 As the solvent, we chose a mixture of nickel and chromium owing to their ability to dissolve boron and nitrogen, respectively. The process was carried out under flowing nitrogen. Both hBN powder and hot pressed ceramics used as source materials, as well as elemental boron, reacted in situ with nitrogen. First, we formed a saturated solution of boron nitride in a nickel/chromium mixture by dissolving the source materials in the solvent at a high temperature. Then, the solution was cooled at 2–4°C/h, which decreased the solubility of the solute and caused hBN crystals to precipitate randomly on the surface of the solution.
Using hBN as the source material, in a typical reaction run we held the mixture at a maximum temperature of 1550°C for 24h to form a liquid solution and then cooled the solution at 4°C/h for over 50h. The crystals produced were transparent, colorless, and clear and were in the form of either flat platelets or prismatic needles. The platelets had a diameter of between several hundred microns and 1–2mm, with a layer thickness of 50–200μm: see Figure 1(a). The prismatic needles were 100–200μm long and 20–50μm wide: see Figure 1(b).
The crystals were investigated to assess their purity and structural perfection. The major impurities were carbon (1019atoms/cm3) and oxygen (1020 atoms/cm3), which originated from the source materials. These crystals produced a photoluminescence peak with a maximum energy of 5.75 eV, which only occurs for hBN crystals of high purity with low defect densities. The x-ray diffraction peaks and Raman spectroscopy peaks were narrow, which is also indicative of good structural quality. The surfaces of the crystals were smooth, with an RMS roughness as low as 0.04nm over an area of 2×2μm. We used defect-selective etching in a molten mixture of potassium hydroxide and sodium hydroxide to quantitatively estimate the densities and distributions of dislocations that were present in the hBN crystals.
We also produced hBN crystals with different boron isotope concentrations. Natural boron consists of 20% boron-10 and 80% boron-11 isotopes. Compared with hBN with the natural distribution of boron isotopes, isotopically pure hBN is predicted to have a 40% higher thermal conductivity. In addition, boron-10 has a high thermal neutron capture cross section, whereas boron-11 does not. Therefore, hBN enriched in boron-10 is better for applications involving neutron absorption. Using isotopically enriched elemental boron as the source material, we produced crystals with boron-10 concentrations ranging from 1 to 99%. A correlation was established between the concentration of boron-10 and the Raman peak position.
This work demonstrates the viability and versatility of the solution growth method for producing high-quality single crystals of hBN. In the next step of our research, we aim to more clearly establish the dependence of crystal growth on the process parameters. This should enable us to better control the crystal morphology, increase the size of individual crystals, and reduce the residual impurity concentrations. Following that, the crystals can be incorporated into devices to confirm theoretical predictions of their performance.
Kansas State University
J. H. Edgar is a university distinguished professor and head of the Department of Chemical Engineering. His research interests are in the crystal growth and epitaxy of semiconductors for electronic and optoelectronic devices.
Timothy Hoffman is a PhD candidate in chemical engineering. He received a BS in chemical engineering in 2012 from Kansas State University.