Plastic deformation in nanodiamonds

Abstract

Diamond, composed of sp³ covalent carbon bonds, is renowned for its exceptional hardness, thermal conductivity, and wide bandgap, yet its intrinsic brittleness severely limits deformation and processing. Here, we report an amorphization-mediated ultralarge plasticity in nanodiamonds using a custom-designed in-situ transmission electron microscopy mechanical holder. Unlike conventional mechanisms such as dislocation motion or crystalline phase transformation, the deformation is governed by the formation of an interconnected amorphous carbon network that accommodates stress and enables cooperative grain rotation and sliding. This amorphization-mediated plasticity allows nanodiamonds to sustain compressive strains exceeding 90% without fracture. A distinct size-dependent transition is identified: ultralarge plasticity occurs only below ~13 nm, while larger diamonds deform in a brittle manner. This work provides critical insights into nanoscale mechanics and offering exciting opportunities for diamond-based nano-manufacturing, strain engineering, and advanced quantum or electronic device applications.

Nanodiamonds below ~13 nm deform plastically by forming ultrathin amorphous carbon networks, enabling over 90% compression without cracking. In situ TEM and simulations reveal this mechanism, opening routes for nanoscale diamond shaping and assembly.

Introduction

Diamond has long been renowned for its exceptional and diverse physical properties, including high hardness and stiffness, superior thermal conductivity, ultrawide bandgap, low thermal expansion coefficient, high biocompatibility, etc^1–6. Owing to these remarkable properties, diamond has found extensive applications in various fields such as mechanics, optoelectronics, biomedicine, and photonics^7–12. However, due to the strong and directional C-C sp³ covalent bonds, diamond has been traditionally characterized by high brittleness and almost non-deformability, which severely restricts its potential applications^13,14. If diamond could exhibit plasticity, it would provide an unprecedented combination of lightness and strength, surpassing many conventional alloys¹⁵. However, ceramics with ionic or covalent bonds demand significantly higher energies to break bonds and rearrange the atoms. As a result, these bulk materials are inherently brittle, except under a few special conditions¹⁶.

At bulk and sub-micrometer scales, plastic deformation in materials is typically governed by the generation and migration of dislocations, especially in metals^17,18. In a typical plastic deformation process, the density of dislocations increases due to their multiplication and accumulation¹⁹. In some ceramics, it has been reported that shear bands can facilitate plastic deformation²⁰. Recent studies have revealed that diamond can undergo both elastic and plastic deformation when its size is reduced to a few hundred nanometers^21–24. The plasticity of sub-micron scale diamond is also dominated by the generation and movement of dislocations^25–27. Theoretically, the plastic deformation behavior and mechanism in sub-10 nm scale materials deviate from those observed at bulk and sub-micrometer scales²⁷. Nanoscale materials generally exhibit low dislocation density, and in some cases, defects may even be absent due to dislocations escaping through the surface. For instance, studies have shown that plastic deformation in sub-10 nm Au and Ag nanowires is primarily governed by atomic diffusion^28,29. Considering the robust and highly directional C-C covalent bonds and reduced atomic mobility compared to metallic atoms, the plastic deformation mechanism of nanoscale diamond is expected to be significantly different from that of metal nanomaterials³⁰. Therefore, miniaturization emerges as a promising strategy to endow diamond with plasticity, providing an exciting platform for exploring novel deformation mechanisms.

In this work, we demonstrate ultralarge plastic deformation of diamond nanoparticles (NPs) using in-situ mechanical experiments conducted inside a transmission electron microscope (TEM) at room temperature. Our results show that nanoscale diamond can be compressed to a near–single-atomic-layer thickness. These findings reveal that sub-10 nm NPs exhibit distinct plastic deformation behavior, primarily driven by the formation of amorphous networks, which is different from that of typical materials. These amorphous networks act as a structural framework, binding the fractured nanograins induced by compression and facilitating their rotation and movement. This mechanism enables the ultralarge plastic deformation observed in nanoscale diamond.

Results

In-situ transmission electron microscopy mechanical measurement

The mechanical measurement and simultaneous atomic-scale observation of diamond NPs were carried out using our custom-designed in-situ TEM holder, which has been demonstrated to provide high precision and stability (Fig. 1a and Supplementary Fig. 1)^31–34. Micrometer-sized diamond indenters were affixed to the ends of two tungsten wires, with one wire mounted on the edge of the length extensional resonator (LER) prong, and the other was fixed to the holder frame. Diamond NPs, fabricated through the explosion method with a sub-10 nm diameter, were cleaned with aqua regia to ensure their purity before being dispersed onto the surface of one of the diamond indenters. The diamond NPs used in this study were primarily single-crystalline. In addition, comparative experiments were also performed on polycrystalline nanodiamonds, which exhibited similar deformation behavior.

Fig. 1. Measurement system and sample characterization.

a Schematic illustration of the experimental setup. Diamond nanoparticles (NPs) are clamped between two diamond indenters that are fixed to the ends of the tungsten (W) wires. One wire is attached to a quartz length-extension resonator (LER) and another W wire is fixed to the holder frame. The position of the LER side diamond indenter is controlled in three dimensions by a tube piezo. b Electron energy loss spectroscopy (EELS) spectra image of the diamond NP. c The distribution of EELS signal between 286–336 eV induced by transitions to the σ* orbitals due to sp³ bonding. d the distribution of EELS signal between 283–285 eV induced by transitions to the π* orbital due to the presence of sp² bonding. e 392-432 eV due to the nitrogen in the Si₃N₄ substrate. f EELS spectra of the 3 regions containing diamond NP, substrate and their overlap, showing the corresponding C K-edge and N K-edge features⁵⁹. Intensity is plotted in arbitrary units (a. u.).

During mechanical loading, the LER-side indenter was advanced toward the fixed indenter using a piezoelectric actuator, clamping a single NP for subsequent compression. The stiffness of the diamond NPs was measured using a frequency-modulation method widely used in atomic force microscopy (AFM)^35–37. The LER oscillates with an amplitude of ~20 pm, distributed along the entire manipulator and indenter assembly, resulting in an effective cyclic displacement transmitted to the NP far below the C–C bond length (~154 pm). To exclude potential contributions from high-cycle fatigue, additional compression tests were conducted under the oscillation-still condition (Supplementary Fig. 2). The applied force was calculated by integrating the stiffness over compression distance, and the associated stress was obtained by combining the measured force with the effective particle diameter using the method of Stauffer et al.³⁸.

Plastic deformation was identified by analyzing the dissipative energy of the LER, which directly correlates with the energy consumed during sample deformation, as validated in previous studies^39–41. Signal fluctuations from feedback control or noise do not affect the overall trend or conclusions. The deformation process of diamond NPs was studied using a JEM-2000VF TEM (200 kV) operating under ultrahigh vacuum conditions (10^–7 Pa). To eliminate possible beam-induced effects such as radiolysis or knock-on damage, compression tests were repeated under beam-blanking conditions at both 200 kV and 80 kV. In all cases, identical amorphization-mediated plastic deformation was observed, confirming that the deformation originates from mechanical stress rather than electron-beam irradiation. Additional analyses provided in Supplementary Section 2 (Supplementary Fig. 3) further verify that the electron beam does not influence the deformation behavior of the NPs.

Low-magnification TEM imaging (Supplementary Fig. 4) shows that the NPs have uniform size and morphology, while high magnification scanning transmission electron microscopy high-angle annular dark-field (STEM-HAADF) images (Supplementary Fig. 5) confirm their good crystallinity. Prior to compression, the electron energy loss spectroscopy (EELS) spectra of the NPs were obtained, as shown in Fig. 1b–f. The EELS spectra exhibit distinct diamond characteristics, with no detectable sp² carbon edge. The EELS mapping results further confirm the exclusive presence of sp³ carbon signals on the diamond NPs, with no evidence of sp² carbon signals in either the particles or the surrounding areas. The X-ray diffraction (XRD) and Raman spectra further confirmed the high quality of the diamond NPs (Supplementary Fig. 6).

To comprehensively understand the plastic deformation mechanism of diamond NPs, four compression cycles with varying strains were conducted (Fig. 2 and Supplementary Videos 1–4). The initially selected NP for testing had a diameter of 7.3 nm (Fig. 2a) and a clean surface free of amorphous layer. Additionally, no measurable conductivity was observed in the NPs (as shown in Supplementary Fig. 7), further confirming that the NP surface was devoid of sp² carbon^42,43. Force measurements (Fig. 2e) revealed that the initial attachment of the diamond indenter to the NPs caused a rapid increase in stiffness, followed by a gradual growth, indicating initially insufficient contact. At the onset of deformation, the dissipated energy was zero, suggesting that the NP underwent purely elastic deformation at this stage. As the NP was compressed to approximately 14% strain (~1.4 nm displacement in Fig. 2e), the dissipated energy increased to ~0.09 eV/cycle and remained constant throughout the subsequent deformation process. This continuous energy dissipation indicates ongoing plastic deformation within the NP during this period. The stress at this stage is estimated to be ~58 GPa using the method of Stauffer et al.³⁸, in agreement with previous reports^21,44. During this process, the diamond NPs showed continuous deformation with no evidence of cracks or brittle fracture, and the NP was flattened and expanded laterally with no abrupt longitudinal or lateral movements (Fig. 2 and Supplementary Fig. 8), indicating the plastic deformation occurred. Upon reaching a strain of ~25% (~2.2 nm displacement in Fig. 2e), the loading was removed, completing the initial compression process. After compression, the NP’s diameter along the compression direction was reduced to approximately 6.2 nm.

Fig. 2. Plastic deformation process of a diamond NP.

The NP is compressed for four cycles to study its deformation mechanism. a–e, f–j,k–o and p–t present the TEM images and corresponding force-displacement responses for the first, second, third and fourth compression cycles, respectively. Dashed lines in g–i highlight amorphized regions during compression. Dashed lines in l–n mark the division of the NP into small grains by amorphous carbon networks. A-D, F-I, K-M and P-S marked in e, j, o and t correspond to TEM image acquisition moment of a–e, f–j, k–o and p–t, respectively. The scale bar is 3 nm.

During the second cycle of compression, the amorphization of diamond NPs began, as indicated by the dotted line in Fig. 2g. The formation of amorphous carbon is further confirmed by nanobeam electron diffraction (NBD) and EELS spectra (Supplementary Figs. 9–11). Since no amorphous phase was observed in the diamond NP prior to mechanical testing, this amorphous region is attributed to the compression process. As compression continued, these amorphous regions expanded and multiplied (Fig. 2h, i), accompanied by lattice distortions and dislocations near the amorphous regions. In terms of force response (Fig. 2j), the initial dissipated energy was similar to that observed at the end of the first compression cycle. The dissipated energy remained constant at approximately 0.09 eV/cycle during this process, corresponding to the transformation of ordered sp³ bonds into disordered ones. The EELS spectra reveal that the amorphous regions contain both sp² and sp³ bonds, further corroborating the structural transformation (Supplementary Fig. 10). The dissipated energy decreased slightly to approximately 0.05 eV/cycle, indicating a new plastic deformation stage for the diamond NPs. When the strain reaches approximately 45% (~2.0 nm displacement in Fig. 2j), the compression in this cycle was terminated.

Further compression produced enlarged amorphous networks that segmented the NP into several smaller grains (Fig. 2k). The newly formed small grains slide along these amorphous networks as compression continues (Fig. 2l, m). FFT analysis (Supplementary Figs. 12–17) reveals reorientation of specific grains with estimated rotational angles of ~10–20°, inferred from the gradual displacement of diffraction spots. Although the nanograins are small (~2–5 nm), the sequential FFT patterns clearly demonstrate orientation evolution during compression. The absence of dislocation arrays or shear bands indicates that this reorientation arises from stress-assisted sliding and rearrangement along amorphous intergranular layers rather than from conventional dislocation motion. Eventually, compression results in the NP being composed of a mixture of small grains and amorphous carbon networks (Fig. 2n). The energy dissipated during this period is approximately 0.05 eV/cycle (Fig. 2o), significantly lower than the 0.09 eV/cycle observed during the amorphization stage. This suggests that achieving an amorphous structure (amorphization involves breaking ordered sp³ bonds) requires more energy than the movement of grains within amorphous networks. Consequently, chemical bond breaking and grain sliding result in distinct energy dissipation patterns for these two deformation processes⁴⁵. Furthermore, stiffness values at the second stage are lower than those at the first stage.

In the final compression cycle (Fig. 2p–s), grain sliding continued with dissipated energy initially at 0.05 eV/cycle (Fig. 2t). As compression continues, the dissipated energy slightly increases to around 0.06 eV/cycle, which can be attributed to the amorphization of these small grains, as shown in Fig. 2r, where the expansion of the amorphous region and disappearance of lattice structures in small grains are observed. We propose that, during this process, some small grains undergo amorphization while others continue to slide and migrate. As a result, the dissipated energy is lower than that observed during the second compression cycle when all crystals were amorphized, but slightly higher than that during the third compression cycle when all grains slipped. The ongoing amorphization of small grains is evidenced by the gradual disappearance of diamond lattice in TEM images and corresponding FFT patterns (Supplementary Figs. 12–14, 18), as well as the disappearance of diffraction spots in NBD and the appearance of diffuse rings due to the lack of long-range periodic structure (Supplementary Figs. 9 and 11). When the strain exceeds 60% (~2.5 nm displacement in Fig. 2t), the NP was almost entirely amorphous (Fig. 2s). These deformation behaviors were consistently reproduced in multiple NPs (Supplementary Figs. 2, 13–14, 16–27). Throughout the entire compression process, no nanocracks, voids, or discontinuities were observed. In-situ TEM images and videos revealed smooth and continuous structural evolution without sharp crack boundaries, and the stiffness responses lacked sudden drops, confirming deformation without brittle fracture.

To investigate the size dependence of the observed plasticity, additional in-situ compression tests were performed on nanodiamonds with diameters ranging from 13 to 100 nm (Supplementary Figs. 28–32). Nanodiamonds smaller than ~13.2 nm consistently exhibit ultralarge plastic deformation accompanied by the formation of a stable sp²/sp³ amorphous network. In contrast, larger particles (17–100 nm) deform in a brittle manner, producing sharp cracks and grain splitting without the development of an amorphous phase. These results reveal a clear size threshold: amorphization-mediated plasticity occurs only when the particle is sufficiently small for a sub-nanometer amorphous network to percolate throughout the structure. Once the diameter exceeds ~15 nm, the formation of such a continuous amorphous network becomes energetically unfavorable, and brittle fracture dominates. Similar deformation behavior was observed in both single-crystalline and polycrystalline nanodiamonds, indicating that the ultralarge deformability is independent of the initial crystallinity. Consistently, nanodiamonds compressed along different crystallographic orientations exhibit the same amorphization-mediated plastic deformation, confirming that the deformation mechanism is also independent of loading direction (Supplementary Figs. 13–14, 16, 18, 20 and 33). Furthermore, compression tests conducted with the resonant oscillation disabled yielded the same amorphization-mediated plastic deformation (Supplementary Fig. 2), confirming that the observed structural evolution arises from mechanical loading rather than cyclic fatigue.

Molecular dynamics (MD) simulations

To investigate the deformation mechanism of diamond NPs, molecular dynamics (MD) simulations were conducted to model the compression process of an 8-nm-diameter diamond NP. The simulation details are provided in the Supplementary Information. The snapshots and stress-strain curve during compression are shown in Fig. 3, where blue spheres represent diamond carbon atoms and white spheres represent amorphous carbon atoms, identified using the polyhedral template matching method. As depicted in Fig. 3a–i, the diamond NP undergoes continuous compression. Initially, small amorphous regions form within the interior or on the surface of the NP (Fig. 3b and Supplementary Fig. 34b). With increasing compressive strain, these amorphous regions multiply, coalesce, and expand to form amorphous networks that divide the diamond NPs into several smaller grains (Fig. 3c–e and Supplementary Fig. 34c–e). As compression progresses, these smaller grains exhibit relative movement (Fig. 3f–h and Supplementary Fig. 34f–h). Finally, all the diamond atoms are transformed into amorphous carbon (Fig. 3i and Supplementary Fig. 34i).

Fig. 3. Molecular dynamics simulations of diamond NP compression and the corresponding force response.

a–i The extracted structures of the NP during the compression process. Blue balls represent diamond carbon atoms and white balls represent amorphous carbon atoms. Local amorphization progresses as amorphous regions expand, encapsulating the NP into small diamond grains and ultimately resulting in complete amorphization, the scale bar is 2 nm. j Simulated stress-strain curve that considered the cross-sectional variations during the compression process. A-I correspond to the extracted structures from the MD simulations in a–i. k Radial distribution function (RDF) analysis of the diamond NP under compressive strains of 0%, 19%, 38%, 56%, 75%, and 87%. The RDF evolution indicates a gradual increase in amorphous carbon content as compression progresses, confirming strain-induced structural transformation.

The simulated stress-strain curve (Fig. 3j) reveals that the transition from elastic to plastic deformation occurs at a stress level of approximately 55 GPa, consistent with the experimental results (~58 GPa). The stress increases as amorphous regions propagate, but decreases when nanograins begin to slide. At larger strain, the stress remains nearly constant while the nanograins gradually undergo amorphization. Only a limited number of dislocations are generated during compression (Supplementary Figs. 35, 36), indicating that dislocation motion is not the dominant carrier of plasticity; instead, plastic deformation is governed by the initiation and expansion of amorphous networks. The MD results further show that amorphous carbon forms continuous ultrathin layers that partition crystalline domains into nanograins, producing a networked amorphous structure fully consistent with experimental HRTEM observations (Fig. 3 and Supplementary Fig. 34). Additional simulations performed along the [110] and [111] orientations (Supplementary Figs. 37–38) exhibit similar interconnected amorphous layers, demonstrating that this deformation mode is robust and orientation-independent.

MD simulations further show that the diamond NP can be compressed into amorphous carbon layers approaching single-atomic-layer thickness (Supplementary Fig. 39), consistent with the near–single-atomic-layer thickness observed experimentally. The radial distribution function (RDF, Fig. 3k) highlights the structural evolution from crystalline diamond to amorphous carbon. The applied cutoff radius of 5 Å is significantly larger than the lattice constant of 3.57 Å of bulk diamond. Figure 3k shows 7 peaks for uncompressed NPs corresponding to the 7^th nearest neighbors observed at interatomic distances of 1.54, 2.53, 2.97, 3.59, 3.90, 4.37 and 4.65 Å. It can be clearly seen that the RDF of the initial diamond NP significantly differs from that of the compressed one. Moreover, the RDF of the compressed NP is similar to that of amorphous carbon, confirming its transformation into the amorphous state.

During deformation, nanodiamond does not immediately transform into a fully amorphous state. Local amorphous regions coexist with nanocrystalline domains over a wide strain range, meaning that essential characteristics of diamond are retained while allowing the material to deform without fracture. The ability to accommodate amorphization-mediated plastic deformation without catastrophic failure is itself technologically valuable, as it greatly improves the processability and mechanical reliability of diamond at the nanoscale. In many material systems, amorphization can also introduce new functionalities distinct from the crystalline phase, suggesting that controlled amorphization in nanodiamond may enable additional mechanical or functional advantages.

Mechanistic Insights of the ultralarge deformation

To quantitatively resolve the evolution of bonding configurations during deformation, we performed EELS mapping under progressive compression (Fig. 4). Before compression, the C K-edge spectrum exhibits a purely sp³ fingerprint with negligible π* peak intensity. As strain increases, a distinct π* peak gradually appears, corresponding to the formation of sp² bonds. Quantitative analysis based on integration of the π* and σ* regions following Yan et al.^46,47 reveals that the sp² fraction increases from ~15% at ~16% strain to ~30% at ~45% strain, eventually saturating around 35–40% at ~72% strain. The EELS mapping further reveals that the sp² component nucleates preferentially at stress-concentrated nanoparticle ends and subsequently connects to form a continuous sub-nanometer sp²/sp³ mixed amorphous network bridging reoriented nanograins, consistent with our MD results. These results provide quantitative confirmation that the sp²/sp³ mixed amorphous network acts as the key structural carrier of ultralarge plasticity in nanodiamond.

Fig. 4. Strain-dependent EELS mapping of sp²/sp³ evolution.

a, b TEM images of the nanoparticle before and after compression (the same scale). The scale bar is 2 nm. c The C K-edge electron energy loss spectroscopy (EELS) spectra acquired under different strain levels, showing the gradual increase of the π* peak intensity indicative of progressive sp² bond formation. Intensity is plotted in arbitrary units (a. u.). d–g EELS spectrum image and π*/σ* bonding-state mapping of diamond nanoparticles during the compression process. The EELS signal between 286–336 eV corresponds to transitions to σ* orbitals arising from sp³ bonding (red), while the 283–285 eV signal corresponds to transitions to π* orbitals associated with sp² bonding (blue). The scale bar is 3 nm.

Based on these observations, we propose a deformation mechanism illustrated in Fig. 5. Initially, the diamond NP undergoes elastic deformation, reaching a maximum elastic strain of about 12% (Fig. 5a), consistent with previous reports^22,42. Subsequently, amorphization begins, requiring significant energy to transform diamond crystals into an amorphous state (Fig. 5b). The lattice structure surrounding the amorphous regions exhibits slight distortion and dislocations (Fig. 5b, c), suggesting that the amorphization may originate from the defects and lattice distortion within the diamond NPs. According to our EELS results, the amorphous region is composed of both sp² and sp³ carbon. As compression continues, the amorphization process progresses, leading to the expansion and multiplication of amorphous regions forming amorphous networks. These amorphous networks split the diamond NPs into several smaller grains (Fig. 5d). With further compression, the fine crystalline grains begin to slide, migrate, and rotate within these amorphous networks (Fig. 5e), leading to a gradual reduction in diameter along the compression direction. This stage consumes less energy than the amorphization process itself. Upon continued loading, additional amorphization of small grains occurs concurrently with grain motion, resulting in a partial recovery of dissipative energy. Finally, all nanograins become fully amorphized, and the diamond NP transforms into a nearly homogeneous amorphous state (Fig. 5f). These structural evolutions were reproducibly observed in multiple diamond NPs, as shown in Supplementary Figs. 13–14 and 16–25. In addition, we observe a clear size-dependent transition: ultralarge plasticity only occurs in nanodiamonds smaller than ~13 nm, where a continuous amorphous network can develop, whereas larger particles deform in a brittle manner without forming amorphous layers. Therefore, ultralarge plasticity is an intrinsic size-confined deformation mechanism rather than a universal property of diamond.

Fig. 5. Compressive plasticity mechanism of sub-10 nm diamond NPs and application for nanoscale assembly.

The deformation stages including: a Original NP undergoing elastic deformation. b Initiation of amorphization. c Expansion and multiplication of amorphization regions. d Formation and movement of fine grains. e Complete amorphization of the NP. f Ultralarge plastic deformation of a diamond NP compressed to atomic layer thickness, scale bar is 3 nm. g–i TEM images and corresponding illustrations reveal the plastic merging of two diamond NPs into a single NP under compression, highlighting potential applications for nanomanufacturing of nanodiamond devices. Scale bar is 3 nm. In the TEM images, red dashed outlines indicate amorphized regions, blue dashed outlines indicate lattice fringes, and “┬” marks dislocations. In the schematic, the purple and beige parts of the compression system denote the movable diamond indenter and the fixed diamond indenter, respectively.

Notably, our observations show that these amorphous networks can be very thin (less than 1 nm). In bulk diamond such ultrathin amorphous layers have negligible mechanical influence; however, at the nanoscale—where individual grains measure only a few nanometers—these networks serve as dynamic stress-accommodation pathways that bind fractured grains and facilitate cooperative grain rotation and translation. This coherent grain rearrangement mechanism is fundamentally distinct from conventional plasticity modes, such as dislocation motion in metals or shear-band formation in ceramics, as it enables stress redistribution at the atomic scale rather than merely serving as a site for fracture initiation¹³. In this study, we demonstrate that diamond NPs can be compressed to a size of about 0.4 nm with a remarkable compressive strain exceeding 90% (Fig. 5f and Supplementary Fig. 19). This result indicates that the compressed diamond NPs approach the theoretical near–single-atomic-layer limit. Although the atomic-scale structure cannot be directly resolved experimentally, high-resolution TEM images (Supplementary Fig. 40) exhibit clear near–single-atomic-layer contrast following ultralarge deformation. These contrast features most likely correspond to an ultrathin (< 1 nm) amorphous carbon layer, consistent with the monolayer thickness predicted by MD simulations.

This amorphization process fundamentally differs from conventional dislocation-mediated plasticity. The deformation in nanodiamonds proceeds via a crystalline-to-disordered transition in which the long-range diamond lattice locally collapses into a metastable amorphous phase under high compression stress. This transition produces a permanent, non-recoverable shape change upon unloading, fulfilling the mechanical definition of plastic deformation. Strain accommodation occurs through structural disordering and rearrangement within amorphous networks rather than through dislocation motion, providing an alternative pathway for stress relaxation in strongly covalent solids. Similar deformation-induced amorphization has been recognized as an effective plasticity carrier in intermetallic compounds and high-entropy alloys under extreme loading conditions^20,48,49. In contrast, plasticity previously reported in ceramics—such as submicron silicon pillars—remains dominated by defect migration, and the plastic strain is typically limited (< 10%) and primarily precedes fracture^25,26. In addition, the extent of plastic deformation found in ceramic materials is quite limited, and the mechanism of plastic deformation in nanoceramics remains unclear^25,26,50,51. Likewise, earlier studies on diamond pillars or thin films reported only limited plasticity associated with dislocation glide and twinning, without sustained plastic flow or quantitative analysis of sp²/sp³ bond evolution. Diamond, as the hardest natural material, exhibits distinct physical properties compared to other ceramics, and is considered even more resistant to deformation^21,22. However, our results reveal that when the size of the diamond nanoparticle is below ~10 nm, an entirely different deformation mechanism emerges—amorphization-mediated plasticity. The formation of a dynamically connected sp²/sp³ amorphous network transforms nanodiamonds from brittle to ductile, enabling ultralarge compressive strains without fracture. Both in-situ TEM and MD simulations confirm that this amorphous network acts as a ductile medium capable of dissipating mechanical energy, fundamentally distinct from dislocation- or crystalline–crystalline transformation–based plasticity. These results suggest that even the hardest covalent solid—diamond—can sustain extreme, ductile-like deformation through stress-induced amorphization when confined to the nanoscale, opening an unexplored regime of plastic behavior in covalent materials.

Our study not only elucidates the deformation mechanisms of diamond nanomaterials but also demonstrates a size-induced cold-welding phenomenon in ultrasmall nanodiamonds, which is fundamentally enabled by amorphization-mediated ultralarge plastic deformation. This deformation-driven processing capability is exemplified in Fig. 5g–i, where two adjacent diamond NPs were compressed and observed to gradually merge into a single larger NP, achieving cold welding and structural fusion without melting or external heating. Similarly, head-to-head compression of two diamond NPs led to direct fusion (Supplementary Fig. 41). Expanding on this, we successfully fused multiple diamond NPs into a single larger NP through controlled compression, and remarkably, this newly formed NP also exhibited ultralarge plasticity (Supplementary Figs. 42, 43). High-resolution TEM analysis (Fig. 5h, i; Supplementary Fig. 43) reveals that after fusion, the interface between two nanodiamonds remains structurally continuous, without any observable voids or separation. The merged particles share a common amorphous network that reorganizes into a nanograined structure, ensuring mechanical bonding rather than crack formation. These results establish that amorphization-enabled plasticity not only permits extreme mechanical deformation but also provides a high-precision pathway for nanoscale processing and assembly of diamond materials via cold welding⁵². Given the exceptional mechanical robustness and structural integrity of the fused NPs, this approach offers a promising route to tailor the mechanical properties and micro/nanoscale processing of carbon-based materials. This could be particularly valuable for advanced applications in nanoelectromechanical systems (NEMS), quantum information processors, and next-generation nanoelectronic devices, where precision, mechanical resilience, and material integrity are critical design factors.

Discussion

In conclusion, we have demonstrated amorphization-mediated ultralarge plasticity of sub-10 nm diamond NPs at room temperature. Using a custom-designed TEM holder, we found that diamond NPs can be compressed to several atomic layers in thickness, exhibiting an extraordinary degree of plastic deformation. By combining experimental results with MD simulations, we revealed that this plasticity is primarily governed by amorphous networks, which facilitate the sliding and cooperative rotation and movement of fine nanograins within the amorphous regions. Our findings underscore the critical role of amorphous networks as a structural binding agent that enables nanoscale diamonds to transition from brittle to ductile behavior. Moreover, we found that such amorphization induced ultralarge plasticity could be employed in the micro-nano processing of diamond, offering a novel approach for the fabrication and structural engineering of ultrahard materials. Beyond diamond, these insights provide a broader framework for enhancing plastic deformation in other brittle ceramics, paving the way for innovative nano-fabrication techniques and advanced material design.

Methods

Sample preparation and structure characterization

The experiment in this study was conducted using a home-made TEM holder (Supplementary Fig. 1a). A self-sensing AFM probe–LER was equipped in this study. The LER was excited by a periodic voltage $V^{*}=V{e}^{i\omega t}$, leading to a periodic force $F^{*}=F{e}^{i\omega t}$ ($V$ is the amplitude of the applied voltage, $F$ is the amplitude of the force, $\omega$ is frequency, $t$ is the time and $i$ is the imaginary unit). The LER was then oscillated under a picometer scale amplitude $a^{*}=a{e}^{\left(i\omega t+\phi \right)}$, where $a$ is the oscillation amplitude and $\phi$ is the phase shift between forcing and oscillations. The mechanical behavior of the target materials can be characterized by the complex impedance $k^{*}=\frac{F^{*}}{a^{*}}=k{e}^{-i\phi }$, where the real $k^{’}$ and imaginary $k^{”}$ parts (stiffness) characterize the conservative (elastic) and dissipative responses (plastic) of the materials, respectively. In fact, the energy dissipation is related to the imaginary $k^{”}$ parts induced by the plasticity^35–37,39.

Tungsten wires with diameters of 10 µm and 100 µm (99.99% purity, Nilaco) were cut into tips. The 10 µm diameter tungsten tip was glued to the edge of LER prong (3EXW-1073, STATEK), and the 100 µm diameter tungsten tip was attached to the copper plate with silver epoxy and fixed to the holder frame. A diamond particle with a diameter of ~50 µm was affixed to the end of the 100 µm diameter tungsten tip, while another ~10 µm diamond particle was attached to the end of the 10 µm diameter tungsten tip as indenters. Diamond nanoparticles (NPs), fabricated through the explosion method (Yuxing carbon material Co. Ltd) with sub-10 nm diameter, underwent thorough cleaning with aqua regia for at least 24 h to remove surface sp² carbon, and then being dispersed onto one of the diamond indenters’ surfaces, the nanodiamond particles used in this study were predominantly single-crystalline. We also performed comparative tests on polycrystalline nanodiamonds, which demonstrated comparable amorphization-mediated plastic deformation, indicating that the observed ultralarge plasticity is not affected by initial crystallinity. To estimate the mechanical response of NPs during deformation processes, the LER prong was oscillated at its resonant frequency ($f_0$, ~ 1 MHz) with an amplitude of ~20 pm by applying a sinusoidal excitation voltage via a phase-locked loop. The stiffness of the diamond NPs ($k$) was derived from the shift of the LER resonant frequency ($\Delta f$) and LER stiffness ($k_{LER}$):

$$k=2k_{LER}\frac{\Delta f}{f_0}$$ 1

The dissipative energy of LER is directly related to the energy consumed during the deformation of the sample, and is closely related to the imaginary part of the stiffness. It can be used to identify the plastic deformation in the sample, and can be calculated by the LER excitation voltage ($V\left(A\right)$):

$$E_{diss}=\frac{\pi k_{LER}{A}^2}{Q}\left(\frac{V\left(A\right)}{V_0\left(A\right)}-1\right)$$ 2

Where Q is the quality factor of the LER, $A$ is the oscillation amplitude and $V_0\left(A\right)$ is the excitation voltage of LER without sample connection.

STEM measurements were performed by a spherical-aberration-corrected scanning transmission electron microscope (JEM-ARM300F2), operating at an accelerating voltage of 300 kV. EELS mappings were acquired using a JEM-ARM300F2 microscope with a Gatan Stela detector. High-angle annular dark-field (HAADF) images of the diamond NPs are shown in Supplementary Fig. 5, which shows the excellent crystallinity of the NPs. In this study, we attempted direct tests of the diamond NP using a tungsten probe, resulting in a consistent conductance of 0 G₀ ($G_0=2e^2/h$: the quantized unit of conductance, where e is the elementary charge, and $h$ is the Planck constant). This further confirms the cleanliness of the diamond NPs, which are free from contaminations such as graphene or graphite (Supplementary Fig. 7).

In-situ TEM experiment

The measurement system is shown in Supplementary Fig. 1b. We utilized an ultra-high vacuum transmission electron microscope (UHV-TEM, JEM-2000VF, 200 kV, 10^–7 Pa) for TEM observation to ensure minimal contamination and gas adsorption on the sample. The diamond NP was clamped by adjusting the position of the indenter on the LER in contact with the indenter on the copper plate. Precise positioning of the diamond indenter’s position on the LER is achieved using a tube piezo for fine motion and an ultrasonic linear motor for coarse motion (TULA50, Technohands). Electrical excitation and monitoring of the output signal of the LER were facilitated by electrically connecting its two electrodes with coaxial cables. An oscillation controller (Nanonis OC4, SPECS) was employed to regulate the resonant frequency and amplitude of the LER at a sampling rate of 2.4 kHz. Prior to experiments, samples underwent baking at approximately 100 °C for a minimum of 24 hours to ensure cleanliness. Observations were conducted at room temperature (~300 K), with TEM images captured using a charge-coupled device (CCD) camera at intervals of 0.2 s.

MD simulation

The deformation process of spherical diamond nanoparticles was simulated using the MD techniques. The parallel code LAMMPS (64-bit 22Jul2025-MSMPI with Python)⁵³, distributed by Sandia National Laboratories, is employed for MD simulations. The nanoparticles were derived from a pristine bulk single crystal, and our focus is on particles with 8 nm diameters, as depicted in Supplementary Fig. 44. We further performed comparative MD simulations on nanodiamonds oriented along the [100], [110], and [111] zone axes, confirming that all orientations exhibit the same amorphization-mediated plastic deformation mechanism under high non-hydrostatic stress. Shrink-wrapped boundary conditions were applied in the x,y and z directions during the simulation process. The interatomic interaction force was modeled using the Tersoff force field⁵⁴, known for its ability to replicate dislocation activity observed in experiments. The temperature was maintained at 300 K using velocity rescaling, and a timestep of 1 fs was set. Classical Newton’s equation was used to describe the motion of atoms, which was solved using the velocity-Verlet algorithm. The simulation process consists of the following three steps: (1) The structure was optimized using the conjugate gradient method to ensure that each atom is in a reasonable position. (2) The nanoparticle’s structure was optimized at room temperature equilibrium (maintained at 300 K) using the NVT ensemble, with a relaxation period of 100 ps and a temperature control relaxation time set to 0.1 ps, allowing all atoms in the system to reach an appropriate structure. (3) Indentation was implemented using the NVT ensemble by placing the NP above a flat hard surface and below a flat hard indenter moving at a constant speed of 0.01 nm/ps (10 m/s), typical for indentation simulations, to calculate stress changes during compression processes and output atomic trajectory changes for observing the deformation process of diamond NP.

Visualization and analysis were conducted using the OVITO software⁵⁵. The atoms were classified utilizing the Polyhedral Template Matching (PTM) method, which enables the identification of local crystalline structures in simple condensed phases⁵⁶. PTM was applied with a rmsd value of 0.2. Dislocation analysis was performed using the Dislocation eXtraction Algorithm⁵⁷, which employs Common Neighbor Analysis to identify both the crystalline phase and dislocations within that phase.

Author contributions

S.C. and C.S. conceived the idea, C.S., Y. L. and S.C. designed the experiments, and supervised the project; J.Z. performed the in-situ TEM experiments; J.Z., C.L., Y.Z., L.X., X.L. and Y.O. performed the TEM characterization; J.Z. performed the computational studies. Y. L., C.S., S.C. and J.Z. analyzed the data and wrote the manuscript. All authors contributed to suggestions for the manuscript and reviewed the manuscript.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

All data that support the findings of this study are presented in the manuscript and Supplementary Information, or are available from the corresponding author upon request. Source data are provided with this paper⁵⁸.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70189-6.