Skip to content

Synergistic hardening and toughening of diamond through stacking-fault engineering

FieldValue
JournalNational Science Review
Published2026-08-25
AuthorsXiaoci Ma, Qiang Tao, Min Lian, Yutong Hou, Hetian Liu
DOI10.1093/nsr/nwag546

Pre-oriented stacking faults produced a polycrystalline diamond with a Knoop hardness of 140.1 ± 3.0 GPa and an indentation fracture toughness of 16.9 ± 4.2 MPa·m^1/2. The optimized NPD8 sample had a thermogravimetric oxidation onset of about 1295 K in air. Its toughness was about 120% higher than that of conventional single-phase nanopolycrystalline diamond, while its hardness remained in the ultrahard range.

Hardness and fracture toughness were measured at a 9.8 N load. The hardness value came from the plateau of the load curve, while toughness was calculated from radial crack lengths around Vickers indents.

NPD3, synthesized at 15 GPa and 2300 K, had a Knoop hardness of 121.2 ± 2.5 GPa. NPD8’s hardness was about 16.7% higher than that of conventional nanopolycrystalline diamond at the same composition. The indenter tip remained sharp after loading, supporting the hardness measurement.

Thermogravimetry in air at 10 K/min gave single-crystal diamond an oxidation onset of 1109 K. Many reported diamond-based materials began oxidation at roughly 748 to 1121 K. Nanotwinned diamond reached an onset of about 1329 K, which NPD8 approached.

Differential scanning calorimetry detected earlier oxidation than thermogravimetry. It gave onset temperatures of 1223 K for NPD8 and 1075 K for single-crystal diamond, consistent with the greater sensitivity of calorimetry to oxidation initiation.

Graphite powder with a 20 μm particle size and 99.99% purity was pretreated at 5 GPa and 2400 K for 2 h. Hydrostatic pressure suppressed layer insertion along the c-axis and promoted lateral extension within graphite basal planes. This solid-state epitaxy transformed equiaxed particles into elongated, straight-edged flakes with reduced porosity.

The flakes formed a layered precursor assembly. Diamond nucleated epitaxially from their surfaces during conversion, inheriting their elongated morphology and in-plane orientation. A Kawai-type multi-anvil press converted the precursor at 13 to 22 GPa and 2300 to 2600 K for 30 min.

The recovered cylinders were about 1 mm in diameter and 1 mm in height. They were mechanically polished before characterization. XRD and Raman examined phase composition, while SEM examined morphology. TEM, HRTEM, and aberration-corrected HAADF-STEM examined FIB-prepared lamellae, and EELS measured bonding character. Young’s modulus was determined through Oliver-Pharr nanoindentation and pulse-echo ultrasonic measurements.

At 13 GPa and 2300 K, NPD1 remained predominantly graphite. At 13 GPa and 2600 K, NPD2 contained both graphite and diamond. At 15 GPa and 2300 K, NPD3 became phase-pure crystalline diamond with equiaxed grains and nearly no stacking faults.

At 18 GPa and 2600 K, NPD6 combined granular regions with ribbon-like domains aligned with the original graphite flakes. Its representative stacking-fault density reached about 0.24 per nm. At 22 GPa and 2600 K, NPD8 formed the ordered brick-wall architecture.

TEM measured ribbon lengths of 111 to 150 nm and widths of 9 to 12 nm, giving an aspect ratio of about 12. HAADF-STEM found fault bundles 0.4 to 0.6 nm thick and representative fault density as high as about 1.35 per nm. The authors attribute this inheritance to a shear-assisted, martensitic-like conversion under restricted diffusion.

HRTEM showed gently undulating interfaces decorated with parallel stacking faults. These interfaces interlocked neighboring ribbon domains and hindered their relative motion. NPD8 had an interface wavelength of about 1.7 nm, one order of magnitude below NPD6, with about half the amplitude.

Coherent and near-coherent boundaries displayed continuous lattice fringes across the interface. The authors linked these features to efficient stress transfer and reduced local stresses at grain junctions. Together, the interlocking domains and coherent interfaces resist elastic deformation and grain-boundary sliding.

The effective Young’s modulus was 1184 to 1194 GPa. It exceeded 1100 GPa for single-crystal diamond along (110) and 1023 GPa along (100). It approached 1195 GPa along the stiffest (111) direction. Steep nanoindentation unloading slopes and ultrasonic measurements supported the high stiffness.

Post-fracture HRTEM and HAADF-STEM found three states within about 24 nm of the crack front. These were disordered low-angle fault bundles, tensile-strained cubic diamond, and graphitic domains confined to interfacial regions.

Disordered faults had spacing of about 0.218 nm, 5.8% above the normal cubic diamond (111) spacing of 0.206 nm. Unfolded cubic diamond had a {200} spacing of 0.194 nm along [100], compared with 0.178 nm. The increase was about 9%, while the material retained cubic symmetry.

Graphitic domains had a C-C bond length of about 0.152 nm, about 7% above ideal hexagonal graphite. They extended over roughly 12 × 12 nm² and accounted for about half the bridging length. EELS showed mixed sigma*/pi* character in strained diamond and a pronounced pi* signature in graphite, confirming sp2 hybridization.

The authors interpret these structures as a three-phase crack-bridging ligament formed near the crack tip. They propose that strained diamond and disordered faults sustain bridging stresses, while sp3-to-sp2 reconstruction dissipates local energy. The post-mortem images cannot establish the transformation sequence.

At larger scales, strain gradients near fault bundles reached ±0.2. Cracks deflected, branched, and arrested while crossing the lamellae. Elongated grains sometimes bridged opposing crack faces, maintained mechanical continuity, and imposed closure stresses. These observations support a model combining local crack shielding with transformation-associated energy dissipation.

Several limits constrain how far the results can be generalized. Fracture toughness was inferred from radial cracks around indentions rather than measured during in situ crack growth. The proposed local reconstruction was reconstructed from post-mortem images, so its sequence remains unresolved.

The oxidation pathway was not verified directly. The authors linked the higher onset to oriented fault bundles and local strain, but called for in situ analysis. DFT studies on related carbon systems support the energy-dissipation concept, not the behavior of NPD8 itself.

The reported cylinders were about 1 mm across, so the paper does not establish component-scale performance. The toughness uncertainty was ±4.2 MPa·m^1/2 around a mean of 16.9 MPa·m^1/2. Twelve measurements on three separately synthesized NPD8 samples supported the reported average. The authors propose in situ nano-diffraction, 4D-STEM strain mapping, and phase-field or finite-element simulations to separate elastic shielding from transformation-assisted dissipation.

Selection implications for diamond engineering

Section titled “Selection implications for diamond engineering”

The results make defect architecture a selection criterion for damage-tolerant diamond. Hardness alone does not capture resistance to crack propagation, while the 1295 K oxidation onset provides a separate thermal limit for air exposure under the stated heating rate.

For wear and cutting concepts, the authors identify stacking-fault-engineered diamond as a candidate. NPD8 combined high hardness with a 120% toughness increase over conventional nanopolycrystalline diamond. That application case remains a proposed use because the study did not test cutting tools or wear parts.

A comparable HPHT product would need the same precursor inheritance and controlled pressure-temperature conversion. Any component claim would still require scale-up, oxidation-pathway tests, and direct fracture measurements. The practical selection criterion is therefore the controlled stacking-fault architecture, not bulk hardness alone.

View Original Abstract

Abstract Pre-oriented stacking-fault networks are programmed into polycrystalline diamond through precursor inheritance from elongated graphite flakes, producing a brick-wall architecture composed of high-aspect-ratio lamellae and dense fault bundles. This defect architecture preserves ultrahigh hardness while increasing crack resistance and oxidation stability in air. The optimized sample exhibits a Knoop hardness of 140.1 ± 3.0 GPa, an indentation fracture toughness of 16.9 ± 4.2 MPa·m1/2, and an oxidation-onset temperature of 1295 K. Correlative post-fracture high-resolution transmission electron microscopy (HRTEM), high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM), and electron energy-loss spectroscopy (EELS) analyses reveal a confined multiphase bridge near fracture origins, comprising strained cubic diamond, disordered stacking-fault-rich regions, and localized sp2-rich domains, consistent with stress-assisted local reconstruction. At larger length scales, the aligned stacking-fault architecture biases intragranular crack propagation and promotes crack deflection and crystalline bridging, thereby increasing crack-path tortuosity. These results suggest defect architecture engineering as an effective route to mitigating the hardness-toughness trade-off in diamond and point to a broader design strategy for damage-tolerant superhard covalent solids.