Skip to content

STEM Analysis of Interfacial Defects in Cubic Boron Nitride on Diamond

FieldValue
JournalMicroscopy and Microanalysis
Published2026-07-01
AuthorsAndrew C. Lang, James L. Hart, Bethany Hudak, Matthew T. Hardy, Tyler J Erickson
DOI10.1093/mam/ozag053.570

Cyclic growth produced cubic boron nitride films with no reported 191 eV pre-peak at the B K edge in electron energy-loss spectroscopy (EELS). The authors associate the missing signal with reduced nitrogen-vacancy concentration and improved film quality. Cross-sectional electron ptychography still showed interfacial defects, nanoscale twins, and widespread nanoscale disorder at the c-BN/diamond boundary.

Each growth cycle used a boron electron-beam evaporator, an RF plasma-based atomic nitrogen source, and an RF plasma-based argon ion source. A following cycle ran with the boron shutter closed. This sequence was repeated until the desired film thickness was reached.

The paper links the altered EELS signature to the full cyclic sequence. It does not isolate the effect of boron, nitrogen, or argon during this comparison. The authors say the scheme also enables a more systematic study of ion-beam effects during growth.

Figure 1 compares EELS spectra from different c-BN growths. The cyclic samples lacked the 191 eV pre-peak associated with nitrogen vacancies. The authors describe this as elimination of the pre-peak and suppression of nitrogen vacancies. They do not report a numerical vacancy concentration. The result is therefore a spectral comparison rather than a calibrated defect density.

EELS and ptychography provide different evidence in the study. EELS addresses local chemical defect signatures. Ptychography supplies phase-sensitive structural information at the interface. Together, they show that a lower nitrogen-vacancy signal did not remove all structural disorder.

The authors place c-BN among candidate ultrawide-band-gap materials for high-power RF amplifiers and power electronics. Its band gap is about 6.4 eV. The paper also cites high thermal conductivity and high predicted saturated carrier velocity. They describe c-BN as the least mature material in that group. Synthesis remains difficult because high-quality phase-pure crystals are hard to produce and grown crystals contain many unintentional defects.

The cyclic process addresses the defect problem through a changed growth schedule. Its clearest reported effect is at the B K edge. The pre-peak serves as the paper’s marker for nitrogen vacancies. Removing that marker improves the evidence for film quality. It does not by itself establish a defect-free film.

The paper uses defect characterization to connect synthesis with proposed high-power device applications. It does not present electrical or thermal measurements for these films. The reported benefit therefore concerns film material quality rather than device performance.

Selected samples were examined in a Nion UltraSTEM 200X with a convergence angle of 27 mrad. The instrument was equipped with a MerlinEELS direct electron counting detector on a Gatan Enfinium EELS spectrometer. This setup enabled the chemical and structural analysis used to compare growths.

The workflow was post-growth and cross-sectional. It used selected specimens, so the reported observations apply to analyzed regions and datasets. The paper does not report a separate whole-film characterization.

EELS was performed first at 60 kV to preserve fine structure. That voltage was selected to retain the B K edge fine structure. Electron ptychography was performed later at 200 kV. It used a spectrometer configuration based on work by Sahib and colleagues. Multislice Electron Ptychography data were reconstructed with PtyRAD. The paper reports one reconstructed phase image from that dataset.

The reconstructed phase image directly showed the c-BN/diamond interface. The acquisition orientation was [011], and the scale bar represents 1 nm. The image contained interfacial defects, nanoscale twins, and large amounts of nanoscale disorder. These are direct observations from the reconstructed frame.

The 1 nm scale places the reported twins and defects in the nanometer regime. The paper gives no feature counts, individual defect dimensions, or depth profile. It therefore supports a qualitative assessment of local interface quality.

The ptychography result qualifies the EELS result. Cyclic growth suppressed the nitrogen-vacancy pre-peak, yet the interface still carried defects and disorder. The missing vacancy marker did not establish a structurally uniform junction.

The cited earlier study examined Mg-facilitated c-BN growth by ion-beam-assisted molecular beam epitaxy. It used post-growth cross-sectional STEM on films grown on diamond. The current paper builds on that work by introducing the cyclic growth scheme. It also combines EELS with Multislice Electron Ptychography.

The scheme reduces nitrogen vacancies and allows systematic ion-beam study during growth. The comparison in Figure 1 is spectral rather than numerical. Cyclic growths lack the 191 eV pre-peak in the reported spectra. The paper supplies no peak intensities, vacancy densities, film thicknesses, or growth yields. The improvement is established through local spectroscopy and structural imaging, not electrical performance data.

The text provides no quantitative EELS values for the earlier Mg-facilitated growth. The comparison therefore shows feature absence rather than a change in calibrated vacancy concentration.

Only selected samples received the STEM analysis. The paper does not state the number of specimens or report how many interfaces were reconstructed. The published phase image is one frame from a Multislice Electron Ptychography dataset. A single frame cannot establish the three-dimensional extent of interface features.

The authors identify three-dimensional interface structure and high through-thickness disorder as subjects for continuing work. They plan to use Multislice Electron Ptychography for both. The current text does not quantify the distribution of through-thickness disorder.

The proceedings text also omits growth temperature, pressure, quantitative film thickness, boron-to-nitrogen ratio, residual stress, and phase fraction. It gives no EELS uncertainty or sample statistics for the spectral comparison. These limits prevent a complete process-to-device comparison. The evidence supports local defect and interface conclusions.

The 191 eV EELS feature provides a possible local screening marker for nitrogen-vacancy suppression. The study shows that the marker responds to the growth schedule. A cyclic sequence that included a cycle with the boron shutter closed produced a different signature from the compared growths.

The interface image supplies a separate engineering criterion. Nitrogen-vacancy control alone did not remove twins or nanoscale disorder from the c-BN/diamond boundary. Evaluation of a c-BN device on diamond should therefore include chemical defect signals and interface structure.

The methods can examine local structural, chemical, defect, and phase/growth relationships in epitaxial c-BN. That capability is useful during process development because it can test a proposed growth change against a specific defect marker. The paper does not determine which defect controls electrical behavior or thermal transport. It also does not connect the ptychography result to device performance. These questions remain open.

The c-BN growth process requires a diamond substrate, and the microscopy workflow uses a cross-sectional specimen. The 6CCVD diamond materials catalog includes MPCVD-grown single-crystal and polycrystalline diamond plates and wafers. Engineering support can help define substrate and custom fabrication requirements for this c-BN growth and characterization application.

View Original Abstract

Ultrawide-band gap materials promise greater performance benefits over state-of-the-art wide-band gap materials in high-power RF amplifier and power electronics applications. Among the group-III nitrides, cubic boron nitride (c-BN) is one of the most promising candidate materials due to its large band gap (Eg ∼ 6.4 eV), high thermal conductivity, and high predicted saturated carrier velocity. In spite of many efforts, c-BN is still the least mature of the ultrawide-band gap materials [1]. This is primarily due to the difficulties in synthesizing high-quality phase pure c-BN crystals and the large number of unintentional defects present in grown crystals. Aberration-corrected scanning transmission electron microscopy (STEM) imaging, electron ptychography, and electron energy-loss spectroscopy (EELS) are ideally suited to examine the local structural, chemical, defect, and phase/growth relationships, in epitaxially grown c-BN. Here, we build off of our previous work [2] examining c-BN films grown on diamond with post-growth cross-sectional STEM analysis. In this work, we show the results of growth improvements obtained via a cyclic growth scheme, where c-BN growth is done using a boron e-beam evaporator, RF plasma-based atomic nitrogen source (N), and RF plasma-based argon ion source (Ar+), followed by a cycle with the boron shutter closed. This process is then repeated until the desired film thickness is achieved. The cyclic approach has reduced the concentration of N-vacancies in our films, which are identifiable in EELS via a pre-peak on the B K edge at 191eV, and enables a more systematic study of the role of the ion beams during c-BN growth. Selected samples were analyzed in a Nion UltraSTEM 200X, with a convergence angle of 27 mrad equipped with a MerlinEELS direct electron counting detector on a Gatan Enfinium EELS spectrometer. EELS analysis was initially performed at 60kV to preserve the fine-structure, and Electron Ptychography was later done at 200kV, using a unique spectrometer configuration based on the work of Sahib et al [3]. Results are presented in Figure 1, with EELS showing the elimination of the N-vacancy pre-peak on the B K edge, and a single reconstructed phase image frame from a Multislice Electron Ptychography dataset, reconstructed using PtyRAD [4], directly showing the c-BN/diamond interface, revealing interfacial defects, nanoscale twins, and large amounts of nanoscale disorder. Continuing work using Multislice Electron Ptychography will explore the 3-dimensional structure of the interface and examine the high through-thickness disorder present in the grown c-BN films. Left, EELS spectra showing the B K edge from different c-BN growths showing the lack of a pre-peak at 191eV in the cyclic c-BN growths indicating suppression of N-vacancies and a higher quality film. Right, a reconstructed phase image consisting of a single frame of a Multislice Electron Ptychography dataset of the c-BN/diamond interface. The image was acquired [011] and the scale bar in represents 1 nm.

  1. 2017 - Recent Development of Boron Nitride towards Electronic Applications [Crossref]
  2. 2022 - Mg-facilitated growth of cubic boron nitride by ion beam-assisted molecular beam epitaxy [Crossref]
  3. 2025 - Superconductivity in PrNiO2 Infinite-Layer Nickelates [Crossref]
  4. 2025 - PtyRAD: A High-Performance and Flexible Ptychographic Reconstruction Framework with Automatic Differentiation [Crossref]