Thermodynamic Analysis of Hexagonal Boron Nitride Crystal Growth from Metal Solutions

Hexagonal boron nitride (hBN) has many unique properties such as a strong interaction with thermal neutrons, a wide energy bandgap (>6.0 eV), and a graphite-like structure. These properties make it a good candidate for compact, efficient, and low-cost neutron detectors, deep UV light emitting diodes, and in graphene-based electronics as a substrate and dielectric. Thus, producing high quality hBN single crystals would have major implications for national defense (detection of nuclear weapons by neutron detectors), health (sterilization of water by radiation from deep UV LEDs), and computing (high-speed graphene electronics).

Currently, we are producing hBN single crystals by precipitation from a molten nickel-chromium solvent at temperatures between 1450 °C and 1550 °C under 1 bar of flowing nitrogen. To date, we have grown single crystals up to 3mm in diameter and hundreds of microns thick. However, the residual impurity concentrations in these crystals is still too high for many electronic and optoelectronic device applications. Specifically, the carbon and oxygen concentrations are typically on the order of 10¹⁹ cm^-3 and 10²⁰ cm^-3 respectively. To solve this, the chemical reactions involving carbon and oxygen (the main impurities) and simultaneous phase equilibrium were analyzed using thermodynamics to determine how process conditions may be adjusted to reduce the impurity concentrations. Aside from all of the elements required for hBN crystal growth (B, N, Ni and Cr), hydrogen was also included in the study to test its potential as a reducing agent. Ideally, hydrogen will react with metal oxides and carbides to form volatile oxygen-containing or carbon-containing species, which would be removed in the system exhaust. For this analysis, chemical equilibrium constants were calculated from Gibb’s energy of formation data available from the JANAF Thermochemical Tables. All relevant reactions were coupled, and the equilibrium condition was solved for through an iterative method based on extents of reactions. The vapor phase was treated as an ideal gas and activity coefficients for components in the liquid phase were calculated from excess Gibb’s Energy expressions which were determined via the Calphad method.

We predict that, if the temperature is kept above 1400°C and the ambient partial pressure of water vapor in the system is kept below ~10^-6 bar, all of the oxides may be reduced by hydrogen. The maximum partial pressure of water vapor increases with temperature to a value of ~10^-5 bar at 1700°C. This range of conditions is practically feasible, thus adding hydrogen is a viable option for the removal of oxygen. Boron oxide is the most stable oxide (as compared to the oxides of nickel and chromium), so will be the most difficult to reduce with hydrogen. However, under normal process conditions, B₂O₃ will readily evaporate, providing an alternative mechanism for the removal of oxygen. The reaction of hydrogen with carbon to form volatile methane would provide a mechanism for the removal of carbon, but is thermodynamically unfavorable, and is thus unlikely to succeed. Other potential reactions for the removal of carbon are being considered, but have not been thoroughly studied.