Introduction

Diamond, as the hardest natural material, is widely used in the field of superhard processing, in the form of polycrystalline diamond (PCD, composite of binder and diamond particles prepared by high temperature and pressure, HPHT), coating, and single crystal diamond (SCD) tools1,2,3,4,5,6,7. In the face of rapid advancements across modern industries such as semiconductor8, aerospace9, and mechanical processing10, there is a significantly growing demand for new engineering materials that combine hardness, wear-resistance, and large-size for high-precision machining under extreme conditions. Despite diamond’s remarkable properties, current techniques cannot simultaneously achieve inch-scale, binder-free, ultrahard diamond with hardness surpassing that of natural diamond, particularly in special or complex shapes. This limitation underscores the need for new pathways to develop advanced diamond materials that address the growing demands of modern industries.

The hardness values of typical natural diamond range from 60 to 120 GPa11. In recent years, it possible to produce diamonds that can surpass the hardness of natural ones through nano-structuring strategies. For instance, Irifune et al.12 successfully synthesized ultrahard sintered nano-polycrystalline diamond from graphite under static HPHT with a hardness of 110-140 GPa. Nanocrystalline13,14 and ultrananocrystalline diamond (UNCD)15,16 films, characterized by nanoscale grain sizes, typically achieve enhanced hardness through grain boundary strengthening mechanisms. Owing to these mechanical advantages, both NCD and UNCD coatings have been widely employed in cutting and machining tools for their high hardness, excellent wear resistance, and low friction performance17,18,19,20. However, the mechanical performance of these films is often constrained by factors such as limited thickness and grain-boundary-related degradation. To overcome these limitations, recent studies have shifted toward atomic-level defect engineering, with nanotwinning emerging as an especially effective strategy. Huang et al.1 successfully synthesized nano-twinned diamonds with an average twin thickness of 5 nm and a hardness of ~ 200 GPa, which is twice that of natural single crystal diamonds. In the latest study, by controlling the size of onion-like carbon precursors and inducing high-pressure phase transformation, nano-twinned diamonds with an average twin thickness of 2.3 nm were synthesized, achieving a record-breaking hardness of 276 GPa21. However, as a thermodynamic equilibrium method, it demands exceptionally high temperature and pressure, making the synthesis of inch-grade ultrahard diamonds highly challenging. Chemical vapor deposition (CVD), on the other hand, is another widely used method for preparing diamonds by dissociating the gas source to produce plasma, which is usually a localized equilibrium environment during the stable growth of the diamond. In principle, the effective growth space of CVD diamond is much larger, and it offers great flexibility to grow inch-scale diamonds with various features. For instance, via microwave plasma chemical vapor deposition (MP-CVD) of diamond films on silicon wafers, Jing et al.22 recently obtained free-standing, ultrathin flexible diamond membranes with large-area (up to 2-inch) by exfoliation using sticky tape. However, attempts to produce inch-scale diamond wafers with hardness above 200 GPa by the CVD method have so far been unsuccessful.

In this work, we design and develop a customized MP-CVD diamond growth system, as illustrated in Fig. 1. By integrating a fast-response gas-switching control system and a customized-designed chamber into the CVD equipment23, this setup enables precise modulation of the plasma environment during growth, allowing controlled introduction of defects within the diamond lattice. Utilizing this approach, we synthesize large-area diamond wafers on various substrates and investigate their structural and mechanical characteristics. Transmission electron microscopy (TEM) and related techniques reveal extremely high-density stacking fault (SF) structures (defect density of 4.3×10¹² cm⁻²), indicating that high-density stacking-fault diamonds (HSD) are produced by this method. We further explore how the modulation of growth technics governs defect formation and contributes to the mechanical performance of the resulting films. These insights lay the groundwork for scalable fabrication of defect-engineered diamond materials with tailored properties for advanced applications.

Fig. 1: Synthesis and characterization of the inch-scale ultrahard diamond wafers using customized microwave plasma chemical vapor deposition (MP-CVD) method.
Fig. 1: Synthesis and characterization of the inch-scale ultrahard diamond wafers using customized microwave plasma chemical vapor deposition (MP-CVD) method.
Full size image

a Schematic of the polycrystalline diamond growth process with cyclic nitrogen doping, illustrating changes of gas compositions and the corresponding plasma groups with the rapid flow switching. MFC: Mass Flow Controller. b Comparation of CN group signal (inset showing the magnified spectrum in the blue-shaded region) and C₂ group signal during the nitrogen valve open and close periods within a single cycle detected by optical emission spectrometer. c Intensity variation of optical emission spectrometer characteristic peaks for C₂ and CN groups over 3000 growth cycles. d Temperature monitoring of the MP-CVD chamber over 3000 growth cycles, with temperature fluctuations corresponding to the cyclic doping process. e Photograph of a typical 3-mm-thick, 5-inch free-standing polycrystalline diamond wafer grown by the cyclic nitrogen doping process. f XRD pattern of the growth wafer, with characteristic crystal plane peaks consistent with the standard diamond structure (red bars), with the (111) plane showing the strongest reflection. g Raman spectrum of the diamond wafer, indicating the coexistence of sp³, sp², and Trans-polyacetylene (TPA, refer to ultrananocrystalline diamond grain67) signals. h SEM image of the diamond wafer, showing a pyramidal surface morphology (indicated by orange dashed lines); representative SEM image were obtained from five independent experiments showing similar results. Source data are provided as a Source Data file.

Results

Inch-scale diamond wafer with pulsed local non-equilibrium growth

Introducing high-density defect structures has been proven to be an effective way for enhancing the hardness of various materials, including diamond1,24,25,26. In the context of polycrystalline diamond film fabrication, previous studies have shown that incorporating argon or employing methane-rich gas mixtures can significantly influence diamond growth kinetics and defect formation27,28,29,30. Among various dopants, nitrogen impurity could be used to reduce grain size31,32 and introduce defects33. Building on these insights, we believe that the key to introducing high density nanostructured defects into CVD-diamond is lies in effectively controlling the nitrogen doping process. To accomplish this, we developed a customized technique for MP-CVD diamond growth: achieving local non-equilibrium growth of diamond through high-frequency cyclic pulsed nitrogen doping, as shown in Fig. 1a. Although cyclic gas-phase doping has been previously applied to modulate electrical properties of diamond, such as in phosphorus doping strategies34,35,36, our work uniquely applies high-frequency, short-duration nitrogen pulses to induce defects during diamond growth, introducing a new pathway for defect-driven mechanical enhancement. Specifically, by utilizing a high-speed gas valve control system, the nitrogen gas flow switching cycles (with a pulse width of 6 s, see Methods for details) were achieved. Each cycle consists of two phases: the presence and absence of nitrogen gas. These rapid, cyclic fluctuations in gas composition continuously alter the plasma group, thereby creating a local non-thermodynamic equilibrium growth environment (shifting the ratio of CN and C2 radicals periodically, as indicated by the optical emission spectrometer results in Fig. 1b and c), and the periodic variations of growth temperature in Fig. 1d). Using this unique process, we achieved ultra-stable diamond growth on diamond substrates for over 104 cycles. This process ultimately produced a ~ 3-mm-thick, 5-inch free-standing polycrystalline diamond wafer (Fig. 1e) with controllable thicknesses (see Supplementary Fig. 1, where thickness increased linearly with the number of cycles. The growth rate is 22.1 nm/cycle.). The XRD pattern of the PCD film (Fig.1f) exhibits a strongest (111) diffraction peak at 44° (predominantly exposed crystal plane), along with weaker (220) and (311) peaks and a very faint (400) peak. The narrow full width at half maximum of the (111) peak suggests high crystallinity. The Raman spectrum (Fig. 1g) shows three characteristic peaks at 1195 cm-¹, 1332.5 cm-¹, and 1520 cm-¹. The sharp peak at 1332.5 cm-¹ corresponds to the first-order Raman mode of sp³-bonded diamond. The 1195 cm-¹ peak is associated with defect-related or ultrananocrystalline diamond features. The 1520 cm-¹ peak may be attributed to nitrogen-induced local lattice distortions, distinct from the typical G band of sp² carbon32,37,38. Surface morphology characterization by Scanning Electron Microscopy (SEM) revealed that most polycrystalline grains exhibit a pyramid-like surface morphology (Fig. 1h), indicating that each grain predominantly exposed four equivalent {111} planes (consistent with the XRD strongest peak intensity (Fig. 1f) and electron backscatter diffraction (EBSD) result and related discussions (Supplementary Fig. 2). We can precisely adjust the periodic alternating nitrogen doping duration per cycle, and a series of polycrystalline diamond films have been successfully obtained (see Method for specific growth parameters; for ease of discussion, these samples are labeled as HSD-1 to HSD-4).

Achieving hardness over 200 GPa in inch-scale diamond wafers

We then conducted multiple indentation tests (seven times for each sample) on HSD wafers, as well as on conventionally synthesized single-crystalline diamond (SCD) and polycrystalline diamond (PCD). As a case example, Fig. 2a shows the load-hardness curve for the HSD-1 sample under various loads. Vickers hardness tests were conducted under 1 kgf (9.8 N) load to obtain asymptotic hardness39,40, producing indentations with diagonal lengths of approximately 9-10 μm, corresponding to the mesoscopic scale. Multiple indentations were randomly distributed across the sample surface to ensure measurement accuracy and uniformity. Based on this load value, we calculated the Vickers hardness and fracture toughness using standard formulas1,41,42 (see Methods) for all samples, with comparisons shown in Figs. 2b and 2c, respectively. Surprisingly, all HSD wafers exhibited a significant increase in hardness compared to SCD with (100) plane and PCD (Fig. 2b). Furthermore, as the nitrogen-doped diamond growth duration per cycle in step 2 (as described in the Methods section) decreased, the hardness of the HSD wafers increased significantly. Notably, the HSD-1 wafer, which synthesized with the shortest nitrogen doping time per cycle—just 6 seconds—achieved an average Vickers hardness value of 208.3 GPa. The 6-second nitrogen-doped layer duration was optimized through systematic experiments, which showed that shortening the doping time increases hardness without compromising toughness, achieving an optimal balance of mechanical properties. This impressive value is comparable to the reported 5 nm-thick nano-twinned diamond1 and is twice as hard as typical single-crystalline diamond. This meso-scale hardness measurement reliably reflects the macroscopic mechanical performance of the material, especially given the relatively uniform microstructure (Supplementary Fig. 3) and little variation in hardness (180 GPa-210 GPa) observed across the wafer. We also compared the hardness of HSD-1 grown on different substrates (PCD, silicon and SCD) (see Supplementary Fig. 4a) and observed no significant difference in hardness, indicating that this property is intrinsic (not influenced by the substrate). In addition, the comparison of fracture toughness in Fig. 2c and Supplementary Fig. 4b indicates that the increase in hardness does not significantly compromise the toughness of the HSD wafers.

Fig. 2: Mechanical properties of the high-density stacking-fault diamond (HSD) and its comparison with other diamonds.
Fig. 2: Mechanical properties of the high-density stacking-fault diamond (HSD) and its comparison with other diamonds.
Full size image

a Hardness of HSD as a function of applied load. Beyond 9.8 N, the hardness HV decreases to the asymptotic values of ~208.3 GPa. The inset shows the typical indentation morphology of HSD-1 and the diagonal lengths of the indentation and crack. b, c, Vickers hardness (HV) (b) and fracture toughness (KIc) (c) for HSD-1 to HSD-4, (100) single crystalline diamond (SCD) and polycrystalline diamond (PCD). d Statistical plot of diamond hardness as a function of sample diameter (The square shape of the samples according to its longest side as the diameter). e Demonstration of HSD growth on the surface of a tungsten carbide end mill. A very uniform superhard diamond coating (bottom left) was grown using this method on a standard tungsten steel mill (top left), with enlarged surface images of tungsten steel and diamond coatings in red boxes on the right; representative result from three independent experiments showing similar morphology. f Chemical vapor deposition (CVD) SCD without scratches (top left). The diagram of using HSD-1 to scratch a single crystal diamond (bottom left). The clear scratch morphology of the surface of the single crystal diamond is scratched by a cutting tool made of HSD-1 (right). g The scratch depth morphology (left) was obtained by laser confocal scanning microscope of the red box in figure f. Depth profile (right) of the transverse surface of the scratch of the red line in the left figure. h Comparison of abrasive ratio across polycrystalline cubic boron nitride (c-BN), sintering-PCD, PCD film, SCD and HSD samples. The error bars in (a–c) represent the standard deviations from seven parallel measurements. Source data are provided as a Source Data file.

Over the years, the experience of synthesizing ultrahard diamonds has shown us a clear trade-off: the harder the diamond, the more demanding the synthesis conditions, resulting in a more limited (and often smaller) sample size. Here, our analysis of the correlation between hardness values and sample diameters for previously reported ultrahard diamonds (see Fig. 2d) indicates that the size of these diamonds has not exceeded 11 mm, regardless of whether the diamond type is single-crystalline (the hardness range of 70-160 GPa)43,44,45,46, polycrystalline (110-140 GPa)12,47,48,49,50, nanocrystalline (60-145 GPa)51,52,53,54, or nano-twinned with a twin thickness 5 nm (~200 GPa)1. While diamonds with an average twin thickness of 2.3 nm can achieve a hardness of up to 276 GPa21, the sample diameter was only ~1 mm. Encouragingly, the HSD samples in this study overcome previous size limitations, achieving a hardness of 208.3 GPa and a diameter of up to 5 inches.

This periodic nitrogen-doping process can not only be applied to large-scale planar substrates, but can also be extended to commonly used tools, where uniform diamond coatings can be achieved by optimizing the holder design and plasma exposure conditions. Figure 2e provides a typical example of HSD coating on common engineering tools. After cobalt removal and deposition of a nanocrystalline diamond transition layer, the HSD coating was successfully grown on the surface of the tungsten carbide end mill. SEM images confirm the formation of a uniform and continuous HSD layer across the tool surface. Figure 2f, g exhibit that the HSD sample, when used directly as a cutting tool, can easily cut high-quality single-crystal diamond. Additionally, we compared the wear resistance of various common ultrahard materials used in fabricating cutting tools (see Fig. 2h). The results indicate that the HSD-1 sample, which exhibited the highest hardness, had ~250 times greater wear resistance than polycrystalline c-BN (the most commonly used processing material) and was ~7 times more resistant to wear than polycrystalline diamond (the surface morphology of the samples after wear tests is provided in Supplementary Fig. 5).

The thermal stability of the HSD-1 sample was characterized by thermogravimetry curves measured in air. At a heating rate of 10 °C/min, the onset oxidation temperature was ~760 °C from the thermogravimetry (TG) curve and the differential scanning calorimetry (DSC) curve (Supplementary Fig. 6). The thermal oxidation temperature is comparable to that of natural diamond (770 °C from TG or 720 °C from DSC)1 and greater than that of nanocrystalline diamond (~680 °C)55. This result indicates that the stacking fault structure has excellent thermal stability.

Ultra-dense, three-dimensional interlocked stacking fault network

The successful growth of the HSD-1, which exhibits a hardness exceeding 200 GPa, indicates that our unique nano-structuring strategy for producing larger and harder diamonds is effective. To further investigate the origin of the hardness enhancement in the diamond wafer, we conducted an in-depth structural characterization of HSD-1 and other samples using TEM, ranging from nanoscale to atomic scale. The bright-field TEM image in Fig. 3a highlights a key feature within a grain of HSD-1: the presence of extremely high-density, intersecting planar defect structures along two equivalent {111} planes (see the selected area electron diffraction (SAED) inset in the bottom right), which are distinct from previously reported nano-twinned structures1, as they exhibit significantly shorter lengths (average length = 24.7 nm, see Supplementary Fig. 7) and narrower widths. High-resolution TEM (HRTEM) imaging (Fig. 3b) further confirmed that these crystal defects are primarily stacking faults. Specifically, the stacking faults appear in nanometer-sized, bundle-like structures, with each bundle containing numerous individual faults (as indicated by the red arrows in Fig. 3b). These bundles are evenly distributed along two equivalent {111} planes with a high degree of intersection (marked by yellow dashed circles in Fig. 3b), some of which form the well-known Lomer-Cottrell Locks56 (see corresponding atomic-resolution Scanning Transmission Electron Microscopy (STEM) images in Supplementary Fig. 8). We calculated the average stacking fault interval spacing (corresponding to the thickness of nanotwins) along one direction in HSD-1 to be 6.3 nm across multiple areas, as shown in Fig. 3c. And the calculated stacking fault density is 4.3×10¹² cm⁻² (for detailed calculation procedure see Supplementary Information). The shorter nitrogen doping duration per cycle, the higher stacking fault density within the diamond film, as evidenced by the TEM images and corresponding calculated stacking fault densities from HSD-1 to HSD-4 (Supplementary Fig. 9). However, a high-density stacking fault structure can only be achieved when the frequency of periodic nitrogen doping exceeds a certain threshold. This increased stacking fault density directly contributes to higher hardness values, as demonstrated in Fig. 2a and Supplementary Fig. 10. The exceptionally high density of defect structures (corresponding to small stacking fault spacing) and the associated high hardness are consistent with the previously proposed hardening mechanism, which involves the combined contributions of the Hall-Petch and quantum confinement effects57. In addition to typical stacking fault configurations (Supplementary Fig. 11), we also observed several overlap structures (see Supplementary Fig. 12) within the bundles, suggesting that the planar defects are not only short in length and narrow in width but also have very small planar defect areas (averaging 7.3 nm2, see Supplementary Fig. 13 for HRTEM views and analysis along the [111] zone axis), resulting in the formation of an ultra-dense, three-dimensional interlocked stacking fault network (Fig. 3d). As shown in Fig. 1h and Supplementary Fig. 2, the majority of the grains in the HSD sample are oriented along the <001> direction. This means that during indentation, the indenter interacts simultaneously with four equivalent {111} planes. And this interlocked network, composed of ultra-dense, intersecting stacking faults uniformly distributed across these four {111} planes (see Fig. 3d), causes dislocations to accumulate at these positions, hindering dislocation movement and effectively preventing deformation, thereby increasing hardness.

Fig. 3: Microstructural analysis of stacking faults in the high-density stacking-fault diamond (HSD).
Fig. 3: Microstructural analysis of stacking faults in the high-density stacking-fault diamond (HSD).
Full size image

a TEM bright-field image of a single grain of HSD-1, with the corresponding SAED pattern shown in the bottom right. b HRTEM image of HSD-1 with the zone axis along [110]. The stacking fault bundle structures are marked by red arrows, and the stacking faults intersections are highlighted with yellow dashed circles. c Statistical distribution of stacking fault spacing in HSD-1. d Three-dimensional schematics illustrating the ultra-dense, three-dimensional interlocked stacking fault network within the sample, along with a stacking fault pyramid structure formed by four equivalent {111} planes, which is oriented perpendicular to the growth direction. e EELS comparison between Area 1 (stacking fault bundle region, red box) and Area 2 (perfect-crystalline region, blue box) in the corresponding ADF image; representative spectra were obtained from three independent experiments showing similar results. f ABF-STEM images comparing the lattice structures of undoped and nitrogen-doped stacking fault, along with corresponding lattice models. g DFT calculated GSFE curves on glide-set plane of undoped diamond and N-doped diamond. Representative images in (a) and (b) are taken from ten independent measurements, showing consistent structural characteristics. Source data are provided as a Source Data file.

This ultra-dense, three-dimensional interlocked stacking fault network has never been reported in diamond materials before. This unique structure is a direct result of the local non-equilibrium growth caused by the disturbance of high-frequency heterogeneous atoms in the plasma, facilitated by our specialized rapidly periodic alternating nitrogen doping cycling technique. The periodic introduction of nitrogen perturbs the plasma and temperature environment, inducing localized non-equilibrium conditions that promote stacking fault formation22. We then further conducted elemental analysis to investigate the distribution of nitrogen in the sample. As shown in the electron energy loss spectroscopy (EELS) results in Fig. 3e, in addition to the diamond C-K edge detected in the perfect-crystalline region (Area 2 in the inset annular dark field (ADF)-STEM image), significant π* and N-K edge signals are also observed in the stacking fault bundle region (Area 1), indicating that nitrogen atoms preferentially accumulate at the stacking faults (also confirmed by energy dispersive x-ray spectroscopy (EDS) results, see Supplementary Fig. 14). Moreover, using atomic-resolution annular bright field (ABF)-STEM (Fig. 3f), we found that the atomic stacking fault layers of HSD-1 exhibit noticeable local tilting, in contrast to the undoped polycrystalline diamond. Therefore, we performed density functional theory (DFT) calculations to determine the formation energies of undoped stacking faults and nitrogen-doped stacking faults in diamond (Fig. 3g). The DFT results reveal that the generalized stacking fault energy (GSFE) of undoped diamond is 5.68 J/m2, consistent with literature reports58, whereas nitrogen doping leads to a pronounced reduction to 2.23 J/m2. These findings indicate that the local non-equilibrium dynamic growth process, induced by periodically alternating nitrogen doping, continuously perturbs the CVD growth environment of diamond, thereby facilitating the formation of nitrogen-doped stacking faults. The stacking fault formation process can be speculated base on the atom stacking growth mode. In the high-frequency cyclic pulsed local non-equilibrium plasma environment, the periodically doped nitrogen will react with atomic and molecular hydrogen and carbon atoms to form very stable molecules HCN and free radicals CN38,59. The CN radical, acting as a diatomic adsorbent, rapidly forms a stable structure of four atomic islands, which form the nucleus of the subsequent layer of growth on the {111} surface55. As nitrogen accumulates, the lattice becomes distorted. When interstitial nitrogen atoms aggregate into lamellae within the diamond lattice, extrinsic stacking faults are formed through relaxation. Simultaneously, during the preparation process, the dynamic plasma caused by high-frequency periodic nitrogen doping generates energy fluctuations, causing some carbon atoms to overcome the constraints of surrounding atoms and migrate elsewhere. As a result, vacancies form at the original positions. When these vacancies aggregate into plates/disks, intrinsic stacking faults may be formed through collapse relaxation. To sum up, the interstitial nitrogen atoms and local dynamic plasma disturbance in the rapidly periodic nitrogen doping result in a large number of stacking faults.

The nitrogen-doped stacking faults also exhibit relative thermal stability, as evidenced by in situ heating TEM observations (Supplementary Fig. 15), where the thermal degradation of the HSD sample initiates from the perfect-crystalline region rather than the faulted area.

In summary, we have proposed an innovative MP-CVD technique that enables the synthesis of polycrystalline diamond wafers of various thicknesses and sizes (up to 5 inches), while achieving a high hardness exceeding 208.3 GPa. This remarkable enhancement in hardness is primarily attributed to the controllable formation of an extremely high density of stacking faults caused by high-frequency pulsed local non-equilibrium growth, as the rapidly periodic alternating nitrogen doping strategy significantly reduces the formation energy of stacking faults. Constitutionally, the interstitial nitrogen atoms and local dynamic plasma disturbance cause a large number of stacking faults. This method has led to an unprecedented nanostructured diamond architecture—a highly dense, three-dimensional interlocked stacking fault network—that provides extraordinary resistance to extreme mechanical deformation. This ultrahard CVD diamond technology holds immense potential for a wide range of industrial applications, particularly as cutting tools and coatings for machining the hardest known materials, including diamond, c-BN, and composites containing ultrahard phases. This work provides a new paradigm for enhancing the mechanical performance of diamond, enabling its broader application in aerospace, electronics, and high-precision manufacturing through a defect-engineered strengthening mechanism involving dynamic plasma regulation and stacking fault control.

Methods

Sample preparation

A two-step growth strategy was employed to fabricate large-size, high-thickness free-standing diamond wafers. In the first step, a 5-inch-diameter, 3.5 mm-thick free-standing polycrystalline diamond wafer was deposited on a graphite substrate using a direct current arc plasma jet chemical vapor deposition technique, which offers a high growth rate suitable for thick diamond synthesis. The diamond growth rate ranged from 10 to 15 μm/h23 from the center to the edge of the wafer. The wafer was then mechanically grinded and polished to a final thickness of ~2.8 mm with a smooth surface. In the second step, an ultrahard diamond layer (~200 μm thick) was deposited on the polished wafer using a 915 MHz MP-CVD system with periodically modulated nitrogen doping, under the following growth conditions: microwave power of 11-12 kW, chamber pressure of 11-11.5 kPa, and CH₄/H₂/N₂/O₂ flow rates of approximately 45/900/15/1.5 sccm. The temperature change caused by the gas cycle change is shown in Fig.1d. During diamond deposition, H₂ maintains a stable plasma and etches the non-diamond phase, while CH₄ provides a carbon source. N₂ doping disrupts the atomic arrangement, while a small amount of O₂ improves the crystal quality by oxidizing the non-diamond carbon. The MP-CVD equipment integrates a fast-response gas flow control system, enabling automatic, periodic adjustments of growth parameters during the diamond deposition process. The programmed gas flow sequence for periodic nitrogen doping is as follows:

(1) Turn on the nitrogen gas flow for 48 seconds.

(2) Turn on the methane gas flow for “x” seconds

(3) Turn off the nitrogen and methane gas flow for 48 seconds.

(4) Turn on the methane gas flow for 48 seconds.

(5) Turn off the methane gas flow for 42 seconds.

These five steps represent one complete cycle for the deposition of HSD films. The durations of step (1) and (3)-(5) were experimentally optimized. The gas flow rates for H₂, CH₄, O₂, and N₂ were set at 300, 15, 0.5, and 5 sccm, respectively. In this work, the mechanical properties of the diamond films were analyzed by varying the nitrogen doping time (“x” in step 2), which was set to 6, 48, 96, and 144 seconds, with the corresponding diamond samples labeled HSD-1 to HSD-4, respectively. Throughout the growth process, optical emission spectroscopy was used to monitor plasma variations, while an infrared thermometer was employed to track temperature changes of the sample.

Characterization of mechanical properties

The hardness and fracture toughness of the prepared diamond films was measured using a Vickers hardness tester (Wilson-WolpertTM Microindentation Tester 401MVD).

The hardness, Hv, was calculated using the following formula1,41,42:

$${H}_{v}=1854.4\times \frac{P}{{a}^{2}}$$
(1)

where a is the average length of the two diagonals of the indentation, a1 and a2, measured in micrometers (μm), and P is the applied load in newtons (N). The load used in this work was 9.8 N. The unit of the hardness Hv obtained by substituting the values of P and a is GPa.

Fracture toughness was calculated using the following formula1,41,42:

$${K}_{{{{\rm{Ic}}}}}=\left(0.016\pm 0.004\right){\left(\frac{E}{{Hv}}\right)}^{\frac{1}{2}}\left(\frac{P}{{b}^{\frac{3}{2}}}\right)$$
(2)

where b = \(\frac{b1+b2}{4}\), with b1 and b2 representing the lengths of the two crack diagonals in the indentation, and E is the elastic modulus of the material, with the elastic modulus for diamond samples taken as 1000 GPa1,43. The indentation and crack diagonal lengths were observed and measured using a laser scanning confocal microscopy (OLS4000, Olympus, Japan).

The wear resistance of the samples was evaluated using a fully automated diamond abrasion ratio tester (Changchun Hengyue Electronics Co., Ltd., model HYMH-251). The diamond samples (HSD-1 to HSD-4 grown on PCD substrate) were tilted at 45 degrees and subjected to sliding friction against a diamond composite grinding wheel (13 m/s). The feed rate was set to 0.005 mm every 40 seconds, with the friction load controlled at 300 g. After the test (2.5 hours), the abrasion ratio was determined by the ratio of the abrasion amount of the grinding wheel to the abrasion amount of the samples. A higher abrasion ratio value indicates stronger wear resistance.

DFT calculation

The spin-polarized DFT calculations were performed using the Vienna ab initio simulation package (VASP)60,61 to elucidate the impact of nitrogen doping on the GSFE of the diamond layer. All bulk structures were sourced from the Materials Project database62. The projector augmented wave (PAW) method61, with a cut-off energy of 500 eV, in conjunction with the Perdew-Burke-Ernzerhof (PBE)63 functional, was employed in the DFT calculations. The GSFE is calculated as follows64,65: \({\gamma }_{d}=\frac{{E}_{d}-{E}_{0}}{{A}_{0}}\), where, E0 and Ed denote the total energy of the system before and after the slip relative displacement vector d, respectively, and A0 is the area of the stacking fault. A 112-atom slab model containing an intrinsic stacking fault along the energetically favorable (111) plane was constructed. The model measures approximately 5 × 10 × 13 Å3, with a 15 Å vacuum layer along the z-direction to avoid interactions between periodic images. The topmost and bottom layers were fixed, while the atoms in the other layers were allowed to relax only perpendicular to the slip plane. All models were fully relaxed with an energy convergence criterion of 10−4 eV and a force convergence criterion of 0.02 eV/Å. The models were constructed using Monkhorst-Pack k-point meshes of (9×6×1)66.

Microstructural characterization

X-ray diffraction measurements were performed using a Philips X’pert MRD system with a monochromated Cu-Kα1 X-ray source and a Ge (220) four-crystal monochromator. Raman spectrum was collected using a Renishaw inVia-Qontor micro-Raman imaging system. SEM images were obtained using a field emission scanning electron microscope (Zeiss Gemini 300) equipped with an EBSD detector (Oxford Instruments). TEM samples were prepared using a Thermo Fisher Scientific Helios 5 CX dual-beam system. TEM and STEM images, EDS mappings, and EELS results were acquired using a double Cs-corrected scanning transmission electron microscope (FEI Titan G2 60-300) equipped with a Gatan EELS system and a SuperX EDS system, operating at 300 kV. The in situ heating TEM experiments were carried out using a DENSsolutions Wildfire holder with a MEMS-based heating chip.