Most writing about cubic Boron Arsenide (BAs) is about one number: how well it conducts heat. This paper, by Joon Sang Kang, Man Li, Huan Wu, Huuduy Nguyen and corresponding author Yongjie Hu of the University of California, Los Angeles, published in Applied Physics Letters 115, 122103 (2019), sets out to fill in everything else. Its premise is that a material cannot be built into real devices on the strength of its thermal conductivity alone; engineers also need to know how it absorbs and bends light, how stiff it is, and how much it swells when heated. The authors present their results as a reference table of measured properties, many of which, the authors say, had not previously been measured.
Why these properties were still unknown
The paper notes that boron arsenide was first grown in the 1950s, but that crystals clean enough to measure reliably arrived only recently. Before that, its basic characteristics rested largely on calculation, and the calculations disagreed. The band gap is the starkest example: the authors list theoretical values ranging from 0.67 to 5.5 eV. Published theoretical refractive indices also varied, which the paper attributes to differing computational assumptions.
The crystals used here were grown by chemical vapour transport, following the group's own 2018 Science paper, which had reported a thermal conductivity of 1300 W/mK. That figure appears in this paper's summary table, but it is cited from the earlier work and was not remeasured here.
What was measured, and how
Band gap. The team shone light of photon energy 1.5 to 2.2 eV through a 55 µm thick crystal and recorded how much was absorbed. Because boron arsenide is an indirect-gap semiconductor, an electron absorbing a photon also needs a phonon (a lattice vibration) to supply momentum, so the absorption edge follows a characteristic square-root relationship. Fitting a straight line to that plot gave a gap of about 1.82 eV. The authors note two complications: the phonons involved, at 0.02 to 0.04 eV, can shift the apparent gap slightly, and the absorption has a tail reaching down to about 1.5 eV, which they attribute to high-energy phonons or multi-phonon processes.
Refractive index. An 8.79 µm crystal slab was treated as a pair of parallel mirrors formed by its two faces. Light bouncing between them interferes, producing ripples in the transmission spectrum (Fabry–Perot fringes) whose spacing reveals the refractive index. Between 657 and 908 nm the index fell from 3.29 to 3.04, which the paper places close to gallium phosphide (3.2 at 900 nm) and gallium arsenide (3.7 at 876 nm).
Stiffness. A cubic crystal's elastic behaviour is fixed by three independent constants, C11, C12 and C44. The team obtained them from sound speeds. In picosecond ultrasonics, a laser pulse hitting a thin aluminium film launches a sound pulse into a 4.37 µm slab cut along the (111) direction with a focused ion beam; timing the returning echoes gave a longitudinal velocity of 8150 ± 450 m/s. A second technique patterned periodic metal lines on the surface and heated them with a pulsed laser to excite surface acoustic waves, measured at 4320 ± 120 m/s. Combining these with the known density (5.22 g/cm³) and a literature bulk modulus of 148 GPa, the paper derives C11 = 285 GPa, C12 = 79.5 GPa and C44 = 149 GPa. From these follow an averaged elastic modulus of 326 GPa, an averaged shear modulus of 128 GPa, and a Poisson's ratio of 0.22 along <100> on the (100) plane.
Thermal expansion and heat capacity. X-ray diffraction from 293 to 773 K tracked the lattice constant (4.78 Å at room temperature) as it grew. The linear expansion coefficient came out at 3.85×10⁻⁶/K at 298 K, rising to 6.06×10⁻⁶/K at 723 K; the volumetric coefficient is three times the linear one, 11.55×10⁻⁶/K. Specific heat was measured by differential scanning calorimetry, and the table gives a volumetric heat capacity of 2.09 J/cm³K. From expansion, heat capacity and bulk modulus the authors derive a room-temperature Grüneisen parameter, a measure of how strongly lattice vibrations respond to a change in volume, of 0.82.
The engineering argument
The paper's practical case rests mainly on thermal expansion. A heat spreader bonded to a chip must expand at a similar rate, or temperature cycling builds up stress, risks the layers coming apart and raises the resistance to heat crossing the interface. The authors compare boron arsenide's 3.85×10⁻⁶/K with gallium nitride's 3.94×10⁻⁶/K (a-axis, 300 K) and gallium arsenide's 5.9×10⁻⁶/K, and contrast both with diamond (1.6×10⁻⁶/K) and silicon carbide (3.2×10⁻⁶/K), which they say suffer a much larger mismatch. On stiffness they say the averaged modulus is about twice silicon's (~160 GPa) and comparable to gallium nitride's (~300 GPa), and suggest boron arsenide could suit micromachining and high-pressure uses. These are arguments from material properties; the paper builds and tests no device.
Theory alongside experiment
The group also ran first-principles (density functional theory) calculations and reports "good agreement" throughout. The numbers support that for some quantities more than others. The calculated bulk modulus (150 GPa), C11 (294 GPa), C12 (80.6 GPa) and Poisson's ratio (0.22) sit close to experiment, but the calculated C44 of 177 GPa is about 19% above the measured 149 GPa. The relaxed theoretical lattice constant, 4.7434 Å, is slightly smaller than the measured 4.78 Å. Calculated thermal expansion and heat capacity are shown as matching the measurements, partly alongside earlier heat-capacity data.
What the paper does and does not establish
The paper provides one laboratory's single, consistent set of measurements on its own crystals. Several of the derived values depend on inputs from elsewhere: the stiffness constants use a bulk modulus from earlier literature, and the thermal conductivity in the table is the group's 2018 result. Uncertainties are given for the sound velocities but not for most derived quantities. The paper does not report how sensitive these properties are to defects, although defects are known to matter greatly for thermal conductivity in this material (see Antisite Defects in Boron Arsenide).
Its band gap also differs slightly from a later figure in the dossier: the 2024 high-purity study in Strong Temperature Dependence of Thermal Conductivity in High-Purity Cubic Boron Arsenide gives 1.84 eV against 1.82 eV here. The 1.82 eV value matches the figure in the data summary reproduced on the boron arsenide entity page. For how this baseline relates to the headline conductivity claims, see Ultrahigh Thermal Conductivity in Boron Arsenide and Boron Arsenide Claims vs. Evidence: Measured Values, Patent Claims and the Theory–Experiment Gap.
Source notes & attribution
- https://arxiv.org/pdf/1911.11281
- https://doi.org/10.1063/1.5116025