# Fileset

[s41467-025-63358-6.pdf](https://mdr.nims.go.jp/filesets/f50464f8-7016-4a3c-a8ed-9de3b181f634/download)

## Creator

Bingjie Wang, Chuanli Yu, Yifan Jiang, Chong Tian, Jiamin Tian, Shuo Li, Zheng Fang, Menglan Li, Weilong Wu, Zhaohe Dai, [Takashi Taniguchi](https://orcid.org/0000-0002-1467-3105), [Kenji Watanabe](https://orcid.org/0000-0003-3701-8119), Qing Chen, Xianlong Wei

## Rights

[Creative Commons BY-NC-ND Attribution-NonCommercial-NoDerivs 4.0 International](https://creativecommons.org/licenses/by-nc-nd/4.0/)

## Other metadata

[Dielectric strength weakening of hexagonal boron nitride nanosheets under mechanical stress](https://mdr.nims.go.jp/datasets/e3e3ce78-34be-4771-9edd-aaada199bc31)

## Fulltext

Dielectric strength weakening of hexagonal boron nitride nanosheets under mechanical stressArticle https://doi.org/10.1038/s41467-025-63358-6Dielectric strength weakening of hexagonalboron nitride nanosheets undermechanical stressBingjie Wang 1, Chuanli Yu2, Yifan Jiang1, Chong Tian3, Jiamin Tian1, Shuo Li1,Zheng Fang1, Menglan Li1,4, Weilong Wu1, Zhaohe Dai 2, Takashi Taniguchi 5,Kenji Watanabe 6, Qing Chen 1 & Xianlong Wei 1Hexagonal boron nitride (hBN) nanosheets have become the most promisingcandidates as gate dielectric and insulating substrates for two-dimensional(2D) material-based electronic and optoelectronic devices. While mechanicalstress in hBN nanosheets is often either intrinsically or intentionally intro-duced for 2D material-based devices during device fabrication and operation,the dielectric strength of hBN nanosheets under mechanical stress is stillelusive. In this work, the dielectric strength of hBN nanosheets in ametal/hBN/metal structure is systematically studied when mechanical stress normal tonanosheets is applied. Thedielectric strengthof hBNnanosheets is found tobeweakenedwith lower breakdown strength, shorter breakdown time, and largerleakage current under the mechanical stress with the order of 100MPa, andthe weakening is more remarkable for thinner nanosheets. The thickness-dependent weakening of dielectric strength is attributed to the thickness-dependent stress gradient in hBN nanosheets. Furthermore, the ability of hBNnanosheets to block leakage current can be significantly degraded bymechanical stress even for thick nanosheets up to 41.3 nm. The results indicatethat it is highly important to eliminate mechanical stress in high-performance2D material-based devices employing hBN nanosheets as 2D insulators.Hexagonal boron nitride (hBN) nanosheets are considered the mostpromising two-dimensional (2D) insulator and are used as gatedielectric1,2, insulating substrates3,4, insulating barrier5,6 and encap-sulation layers7–10 for 2D material-based electronic and optoelec-tronic devices. Due to the atomically smooth interface that isrelatively free of dangling bonds and charge traps in hBN/2Dmaterialheterostructures3, field-effect transistors with graphene and 2Dsemiconductor channels show boosted performances, includingultrahigh carrier mobility, ultralow charge inhomogeneity, andultralong mean free paths if the channel is supported or encapsu-lated by high-quality hBN nanosheets2,6,8,11. Furthermore, hBNencapsulation can be used to preserve the intrinsic physical prop-erties of 2Dmaterials. For example, superconductivity inmagic-anglegraphene andMott insulator in trilayer graphene were observed withhBN encapsulation9,10. As a 2D insulator, hBN nanosheets thereforeplay important roles in achieving high-performance 2D material-based devices and revealing unique physical phenomena related to2D materials.Received: 2 November 2024Accepted: 14 August 2025Check for updates1Key Laboratory for the Physics and Chemistry of Nanodevices, School of Electronics, Peking University, Beijing, China. 2Department of Mechanics andEngineering Science, College of Engineering, Peking University, Beijing, China. 3State Key Laboratory for Artificial Microstructures and Mesoscopic Physics,School of Physics, Peking University Yangtze Delta Institute of Optoelectronics, Peking University, Beijing, China. 4Academy for Advanced InterdisciplinaryStudies, Peking University, Beijing, China. 5Research Center for Materials Nanoarchitectonics, National Institute for Materials Science, Tsukuba, Japan.6Research Center for Electronic and Optical Materials, National Institute for Materials Science, Tsukuba, Japan. e-mail: weixl@pku.edu.cnNature Communications |         (2025) 16:8078 11234567890():,;1234567890():,;http://orcid.org/0000-0001-8397-3761http://orcid.org/0000-0001-8397-3761http://orcid.org/0000-0001-8397-3761http://orcid.org/0000-0001-8397-3761http://orcid.org/0000-0001-8397-3761http://orcid.org/0000-0002-5205-089Xhttp://orcid.org/0000-0002-5205-089Xhttp://orcid.org/0000-0002-5205-089Xhttp://orcid.org/0000-0002-5205-089Xhttp://orcid.org/0000-0002-5205-089Xhttp://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0002-1467-3105http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0003-3701-8119http://orcid.org/0000-0002-7919-5159http://orcid.org/0000-0002-7919-5159http://orcid.org/0000-0002-7919-5159http://orcid.org/0000-0002-7919-5159http://orcid.org/0000-0002-7919-5159http://orcid.org/0000-0002-1181-9500http://orcid.org/0000-0002-1181-9500http://orcid.org/0000-0002-1181-9500http://orcid.org/0000-0002-1181-9500http://orcid.org/0000-0002-1181-9500http://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-63358-6&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-63358-6&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-63358-6&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1038/s41467-025-63358-6&domain=pdfmailto:weixl@pku.edu.cnwww.nature.com/naturecommunicationsMechanical deformation with stress generally in the range of0–1000MPa usually accompanies the fabrication and operation of thedevices reliant onhBN/2Dmaterial heterostructures, particularly state-of-the-art electronic devices like stacked nanosheet gate-all-around(GAA) field effect transistors (FETs)12, vertical-transport FETs(VTFETs)13, 2D flexible/strain FETs14 and flexible neuromorphicdevices15. Todate,most 2DFETs basedonhBNdielectric layers still relyon mechanical transfer and stacking procedure16–18. Deformations inthe forms of wrinkles, bubbles, and buckling are commonly observedin transferred 2D materials19, so it is difficult to construct hBN/2Dmaterial heterostructures free of mechanical stress, especially forlarge-area construction. Furthermore, the difference in thermalexpansion coefficient between hBN and its surrounding materials wasfound to induce irreversible straining in hBN nanosheets whenheated20,21. Mechanical stress in hBN is therefore introduced if it istreated by thermal annealing, which is a general procedure in fabri-catinghBN/2Dmaterial heterostructures and their devices11.Moreover,in most of the studies in advanced electronic devices involving 3Dstacked architectures like GAA, structure-induced strain is widelypresent22, and the variationof the intrinsic dielectric strength of hBNasa stacked part is usually ignored. Finally, due to the atomic thicknesswith high flexibility, hBN nanosheets have already been explored foruse as insulating substrates and encapsulation layers for 2D material-based flexible devices, where mechanical stress in hBN is inevitableduring device preparation and operation23,24. Dielectric strength ofhBN under mechanical stress is therefore highly important for devel-oping 2D material-based devices.Mechanical stress-induced dielectric strength weakening of con-ventional silicon dioxide (SiO2) or other dielectric films is wellstudied25,26, whereas that of 2D hBN nanosheets has not been studied.The elusive dielectric strength of hBN nanosheets under mechanicalstress is mainly attributed to the challenges of the simultaneous andreliable determination of its dielectric strength and mechanical stressin experiments. A few experimental methods, including electricalmeasurements on stretchable substrates27 and indentation tests withconductive atomic force microscopy (C-AFM) probes28, were devel-oped to perform simultaneous electrical and mechanical measure-ments on ultrathin 2D materials. While the dielectric strength ofpolymers (i.e. polyimide, polyethylene terephthalate, etc.) used forstretchable substrates is generally smaller than that of hBN29,30, thedielectric strength of hBN cannot be tested on these stretchable sub-strates. As for indentation with AFM probes, it is hard to accuratelydetermine the indented area and mechanical stress because of thecomplex contact mechanics of sharp AFM probes31,32. More impor-tantly, AFM tests are generally performed at atmospheric pressure,which makes the electrical contact between the AFM probe and 2Dmaterials ambiguous andunstable becauseof reactive interfacial waterand oxygenmolecules, especially when dense electrical current passesthrough the contact33,34. Despite the challenge of AFM indentationtests, C-AFM has been applied to study dielectric breakdown of hBNwithout intended mechanical stress. Several important phenomenahave been observed in dielectric breakdown of hBN, including layer-by-layer breakdown35,36, non-volatile resistive switch37, B-vacancy-dominated defect mechanisms38, interlayer molecular bridges39,electrode-affected leakage current40, and so on.In this work, the dielectric strength of hBN nanosheets undermechanical stress is systematically studied by an in situ scanningelectron microscopy (SEM) method. With the method, the dielectricstrength of hBN nanosheets in a metal/hBN/metal structure is sys-tematicallymeasuredwhenwell-determinedmechanical stress normalto nanosheets is applied. Dielectric strength weakening of hBNnanosheets is observed to show lower breakdown strength, shorterbreakdown time, and larger leakage current prior to breakdown whenthey are subjected to normal mechanical stress with the order of100MPa. The results indicate that it is highly important to eliminatemechanical stress in high-performance 2D material-based devicesemploying hBN nanosheets as 2D insulators.Results and discussionDielectric strength weakening in ramped voltage stress testsHigh-quality hBN nanosheets mechanically exfoliated from bulk hBNcrystals are used in this work (see Fig. S1 for structural characterizationof hBN nanosheets). To facilitate multiple dielectric strength mea-surements on hBN nanosheets with the same thickness, large hBNnanosheets with uniform thickness are first transferred onto a SiO2/Sisubstrate with an Au/Ti coating film, and an array of 1μmwide circularislands in the structure of metal/hBN/metal are then fabricated fromthe nanosheets (see Figs. 1a, and S2 for device structure and fabrica-tion). Dielectric strength measurements are performed at room tem-perature and inside the vacuum chamber of an SEM (see Fig. S3 formore details of the experimental setup). The substrate with metal/hBN/metal islands is fixed to a spring stage that is used tomeasure theforceapplied to the islands (Fig. 1b). Tomeasure the dielectric strengthof hBN nanosheets under normal mechanical stress, a tungsten (W)electrical probe with a radius of ~200 nm is manipulated to indent thetop metal layer of an island and measure the electrical current acrossthe metal/hBN/metal structure under an applied voltage (see Fig. 1cand Supplementary Movie 1). The compression force applied to hBNnanosheets is obtained through the formula F = kd, where k is theelastic coefficient of the spring stage (see Fig. S4 in SupplementaryInformation), and d is the retraction distance of the substrate that canbe directly measured by SEM imaging. Due to the bending stiffness ofthe topmetal layer, the indented area S of the hBNnanosheets is largerthan that of the indenter with a nonuniform stress distribution.According to our recent work41,42, a well-determined apparent normalstress (σ) can be obtained by finite element simulation (FEM) (seeFig. S5 in Supplementary Information). Dielectric strength and normalmechanical stress of hBN nanosheets can therefore be well andsimultaneously determined in our experiments.The dielectric strength of hBN nanosheets is firstly studied byramped voltage stress (RVS) tests, where the electrical current acrossthe metal/hBN/metal structure is recorded while increasing theapplied voltage linearly with time until dielectric breakdown. Eachcurrent (I)–voltage (V) curve is measured from a different islandsample without any previous measurements (referred to as multi-sample RVS tests), which enables us to exclude the possible memoryeffects of previous measurements. To investigate the effect of thick-ness on the dielectric strength of hBN nanosheets, the dielectricstrength of hBN nanosheets with thicknesses in the range of about5–40 nm is studied (see Fig. S6 for thickness measurements by AFMimaging). Figure 1d shows the typical I–V curves of hBN nanosheetswith three different thicknesses (4.6, 11.5 and 41.3 nm) under zeronormal mechanical stress. The curves show a common feature thatobvious leakage current is observed at a voltage (VLC) before thedielectric breakdown of hBN nanosheets at a larger voltage (VBD),where leakage current experiences a sudden jump. Breakdownstrength (EBD) is therefore obtained by EBD = VBD/ tBN, where tBN is thethickness of the hBN nanosheets. The strength of the electric fieldcorresponding to VLC is defined as ELC =VLC/tBN. The voltage intervalbetween VLC and VBD is usually regarded as the stage of progressivebreakdown of hBN nanosheets, in which defects are randomly gener-ated and the leakage current increases quickly. For thick nanosheetswith 41.3 nm thickness, the leakage current shows a progressiveincrease with the voltage until breakdown. In contrast, the leakagecurrent of thin nanosheets with 4.6 and 11.5 nm thicknesses is highlyfluctuating with less dependence on voltage before breakdown. Theabove breakdown behaviors observed at zero stress agree well withthose observed in previous C-AFM experiments28,43.What will happen if hBN is subjected to mechanical stress whileresisting the external electric field? To investigate the effects ofArticle https://doi.org/10.1038/s41467-025-63358-6Nature Communications |         (2025) 16:8078 2www.nature.com/naturecommunicationsmechanical stress on the dielectric strength of hBN nanosheets, weperformed RVS tests on the nanosheets of the three thicknesses underdifferent normal stresses. Comparative I–V curves of RVS tests underzero and ~400MPa normal stress are shown in Fig. 1e. Twophenomenaare observed. First, I–V curves of different samples for each thicknessshowquite differentVLC and VBD under zero stress, especially for thoseof thin nanosheets of 4.6 nm. Second, despite the scattering in VLC andVBD among different samples, I–V curves for thin nanosheets of 4.6 nmare obviously left shifted with smaller VLC and VBD when ~400MPanormal stress is applied. In contrast, the normal stress causes a lessobvious decrease of VLC and VBD for the nanosheets with 11.5 and41.3 nm thickness.To get insights into the effects of normal mechanical stress ondielectric strength of hBNnanosheets, EBD and ELC of hBNnanosheetsare extracted from I–V curves of multi-sample RVS tests under dif-ferent stress, as shown in Figure 1f and g. For those samples with thesame thicknesses, EBD and ELC exhibit a similar dependence on nor-mal mechanical stress, consistent with their roles as sequentialindicators of dielectric breakdown. A small ELC is generally associatedwith a small EBD. For thin nanosheets with 4.6 nm thickness, EBD andELC exhibit average values of 15.7 and 12.6MV/cm, respectively, atzero normal stress. An EBD as high as 15.7MV/cm at zero stressindicates the high quality of our hBN nanosheets. The average valuesof 8.6 and 8.0MV/cm around 400MPa are obtained for EBD and ELC,Fig. 1 | Experimental setup andmulti-sample ramped voltage stress (RVS) testson the dielectric strength of hBN nanosheets under mechanical stress. a SEMimage showing the array of metal/hBN/metal (M/hBN/M) islands for dielectricstrength tests. b Schematic drawing showing the measurement setup, where ametal/hBN/metal island is indented by a tungsten (W) probe, and the compressionforce is measured by the spring stage. c SEM image showing breakdown test on ametal/hBN/metal island indented by a W probe. Pseudo-colors were used to dis-tinguish themetal film and hBNnanosheet.d Typical current (I)–voltage (V) curvesof RVS tests for 4.6, 11.5, and 41.3 nm hBN nanosheets without mechanical stress.The voltage for the onset of leakage current is defined as VLC, and the voltage forthe onset of dielectric breakdown is defined as VBD. e Repeatedly tested I–V curvesof RVS tests for 4.6, 11.5, and 41.3 nm hBN nanosheets with 0 and ~400MPa stress,respectively. Each I–V curve is measured from a different island sample (multi-sample RVS test). fTheplots of breakdown strength (EBD) as a function of stress fornanosheets of 4.6 nm (i), 11.5 nm (ii), and 41.3 nm (iii). g The plots of electric fieldstrength (ELC) corresponding to VLC as a function of stress for nanosheets of 4.6,11.5, and 41.3 nm. The number of island samples corresponding to the 4.6, 11.5, and41.3 nm nanosheets is 37, 42, and 52, respectively. Each I–V curve or data point infigures e–g comes from a different island sample. The black squares represent theaverage EBD or ELC at zero stress, with the error bars represented by standarddeviations (SD). The round dots and circles represent EBD and ELC under differentstress. In figures for 11.5 and 41.3 nm, the dashed lines are linearfits of experimentaldata. Source data are provided as a Source Data file.Article https://doi.org/10.1038/s41467-025-63358-6Nature Communications |         (2025) 16:8078 3www.nature.com/naturecommunicationsrespectively. Both EBD and ELC show an overall decrease with theincrease of normal stress, especially in the range of 0–200MPa. Forthe nanosheets with 11.5 nm thickness, EBD and ELC exhibit averagevalues of 11.4 and 10.6MV/cm, respectively, at zero normal stress.Both EBD and ELC decrease slowly with the increase of normal stresswith an overall rate of 0.38MV/cm per 100MPa in the range of0–500MPa. For thick hBN nanosheets with 41.3 nm thickness, EBDand ELC exhibit the average values of 10.2 and 8.9MV/cm, respec-tively, at zero normal stress, and show a negligible decrease with theincrease of normal stress.To display the thickness-dependent dielectric strengthweakeningof 2D hBN, the presently measured EBD of hBN nanosheets evolvingwith different thicknesses are shown in Fig. 2 togetherwith those in theprevious reports28,35,36. As shown in Fig. 2a, thicker nanosheets arefound tohave smaller EBDat zero normal stress. Both the valueof EBDatzero normal stress measured in our work and its dependence onthickness agree well with those in the previous reports. It can be seenfromFig. 2b, c that thebreakdown strengthof hBNnanosheetswith 4.6and 11.5 nm thickness is obviouslyweakenedby the normalmechanicalstress of about 400MPa, with more significant weakening for thinnernanosheets, while thick hBN nanosheets with 41.3 nm thickness canretain their breakdown strength under the normal stress up to about400MPa in multi-sample RVS tests.The dielectric strength of hBN nanosheets shows obvious ran-domness among different samples, as shown in Fig. 1. To obtain thedielectric strength of hBN nanosheets statistically, the breakdownstrength of hBN nanosheets is analyzed by cluster fitting (see Fig. S7 inSupplementary Information)44,45, which can give the characteristicstrength of electric field (E63.2%) corresponding to cumulative break-down probability of 63.2%. At zero mechanical stress, the nanosheetswith 4.6, 11.5, and 41.3 nm thickness show E63.2% of 15.1, 11.0, and10.4MV/cm, respectively. This is in good agreement with the varyingtrend of average EBD with thickness as shown in Fig. 2a. In contrast,whenmechanical stress of ~400MPa is applied, E63.2% decreases to 8.9,10.1, and 9.5MV/cm, respectively. In agreement with the averagebreakdown strength, E63.2% is weakened by mechanical stress, withthinner nanosheets showing more remarkable weakening. While thenanosheets with 4.6 nm thickness show larger E63.2% than that of 11.5and 41.3 nm nanosheets at zero normal stress, they show smaller E63.2%than that of the latter at ~400MPa normal stress.Despite negligible effects on the EBD and ELC of 41.3 nm-thickhBN nanosheets in multi-sample RVS tests, normal mechanicalstress is found to have a critical effect on leakage current of thethick nanosheets in single-sample RVS tests. A single-sample RVStest is to perform RVS measurements repeatedly on the same islandsample to assess its dielectric strength over multiple cycles. Asshown in Fig. S8, single-sample RVS tests on thicker hBN samplesgenerally show two different phenomena: hard breakdown withirreversible dielectric performance, and soft breakdown withreversible dielectric performance. Since a hard breakdown causesall subsequent RVS cycles to follow the same short circuit in I–Vcurves, single-sample RVS tests with soft breakdown are employedto observe the continuous evolution of leakage current anddielectric strength of hBN nanosheets.Figure 3a and b show the sequential I–V curves of single-sampleRVS tests on 41.3 nm nanosheets with the normal stress of 0 and490MPa, respectively. Repeated RVS tests were performed for 14 and13 cycles, respectively. EBD and ELC of the I–V curves and their differ-ence ΔE = EBD−ELC are shown in Fig. 3c and d. Two phenomena areobserved. First, repeated RVS tests at the same point of a sample resultin smaller and smaller ELC and no obvious decrease in EBD for both 0and 490MPa normal stress. Leakage current, therefore, appears atsmaller and smaller voltages for repeated RVS tests. Second, the nor-mal stress can speed up the reduction of ELC and significantly increasethe leakage current. While ΔE increases to 4.26MV/cm after 14 repe-atedRVS testswith zeronormal stress,ΔE increases to 10.72MV/cmfor13 repeated tests with 490MPa normal stress. Importantly, ELCdecreases bymore than 10 times to <1MV/cm, and leakage current canprogressively increase to 10μA before the breakdown, which is oneorder of magnitude larger than that (<1μA) with zero stress, afterrepeatedRVS tests formore than 10 times in the latter case. The resultsindicate that normal mechanical stress can significantly degrade theability of hBN nanosheets to block leakage current in cyclic voltageapplication, even for thick nanosheets up to 41.3 nm.Dielectric strength weakening in constant voltage stress testsThe dielectric strength of hBN nanosheets is then studied by constantvoltage stress (CVS) tests, where electrical current across the metal/hBN/metal structure under a fixed voltage is recorded with time.Notably, since EBD varies with the thickness of samples as shown inFig. 2, applying the same strength of the electric field to the sampleswith different thicknesses will result in quite different behaviors ofleakage current: while thick samples with lower EBD exhibit instanta-neous dielectric breakdown, thin samples with higher EBD exhibitnegligible leakage current for a long time (>2000 s). In order to focuson the effect of mechanical stress on leakage current, we chose aspecific voltage slightly less than the VBD to each sample to obtaincurrent (I)–time (T) curveswithin a reasonable time (typically <1000 s),Ref [36]Ref [28]Ref [35]EBD(MV/cm)ELC(MV/cm)Thickness (nm)EBD(MV/cm)This work (0 MPa)This work (400 100 MPa)This work (0 MPa)This work (400 100 MPa)EBDELCEBDabcThis work (0 MPa)1015202510152 10 801015Fig. 2 | Summary of the breakdown strength EBD and ELC as a function of thethickness of hBN. a Comparison of EBD at zero stress measured in this work withthe data from previous reports28,35,36. The horizontal axis is the thickness of hBN.b Our measured EBD in this work without stress (red square) and with themechanical stress of 400 ± 100MPa (blue square) at different thicknesses. c Ourmeasured ELC from 102 different samples in this work, without stress (yellowsquare) andwith themechanical stress of 400 ± 100MPa (green square) at differentthicknesses. The number of 4.6, 11.5, and 41.3 nm samples we tested without stressis 24, 14, and 41, respectively. The number of 4.6, 11.5, and 41.3 nm samples with thestress of 400 ± 100MPa are 5, 12, and 6, respectively. The square points representthe average values of ourmeasured EBD or ELC fromdifferent samples, and the errorbars are represented with standard deviations. The dashed lines in (b) and (c) areguides for the eyes. Source data are provided as a Source Data file.Article https://doi.org/10.1038/s41467-025-63358-6Nature Communications |         (2025) 16:8078 4www.nature.com/naturecommunicationswhich not only ensures us to observe obvious leakage current but alsodoes not induce immediate breakdown.Typical I–T curves of CVS tests for hBN nanosheets with differentthicknesses andunder different normal stresses are shown in Fig. 4a–c.Each I–T curve is measured from a different island sample (multi-sample CVS tests). Four phenomena are observed in Fig. 4a–d. First,remarkable leakage current can be observed after a period of time(defined as TLC) even though the applied voltage is smaller thanVBD. Inaddition to the strength of the electricfield, the applied duration of theelectric field therefore contributes to the dielectric strength degrada-tion of hBN nanosheets, which agrees well with previous reports39,46.Therefore, an electric voltage smaller than VBDmay induce remarkableleakage current after a long duration of voltage application. Thischaracter will cause a serious problem to the long-term reliability ofhBN nanosheet-based devices. Second, the onset of leakage current ofthin nanosheets with thicknesses of 4–6 and 10–13 nm under fixedvoltage can be significantly accelerated by normal stress with smallerTLC (Fig. 4a, b). For the thin sampleswith a thickness of 4–6 nm,TLC canbe shortened from around 500 s to <20 s when a stress of about600MPa is applied. Third, the stress-induced acceleration of the onsetof leakage current is more significant for thinner samples. Figure 4dshows the TLC extracted from more than 20 CVS tests under differentstress conditions in Fig. 4a, b. While 4–6 nm-thick samples exhibit adecreasing rate of TLC being about 0.9 s/MPa, 10–13 nm-thick samplesexhibit a smaller decreasing rate of about 0.6 s/MPa. Fourth,mechanical stress shows no regular effect on the onset of leakagecurrent of thick samples with a thickness of 37–42 nm (Fig. 4c), indi-cating that the effect of mechanical stress on the onset of leakagecurrent is negligible for such thick nanosheets.In addition to multi-sample CVS tests, single-sample CVS tests arealsoperformed to investigate the effect of normal stress on the leakagecurrent of hBN nanosheets. In single-sample CVS tests, I–T curvesunder different normal stresses are recorded from the same samples.Figure 4e, f show the I–T curves of two different samples in 8.9 nmthickness when they were stressed and unstressed. It can be seen thatboth the magnitude of the leakage current and its increasing rate withtime are significantly increased when subjected to normal stress. Asindicated by the red dashed lines in Fig. 4f, the I–T curves near theonset of obvious leakage currents can be described by the power lawI / ðT � TLCÞC , whereC is the exponent used tomeasure the increasedrate of leakage current, similar to that of voltage stress-induced leak-age current (SILC) in the previous reports47. In addition to smaller TLC,it can be seen that both samples exhibit larger Cwhen they are loaded,indicating a larger increasing rate of leakage current under normalstress. Therefore, both multi-sample and single-sample CVS testsindicate that normal mechanical stress can weaken the dielectricstrength of thin hBN nanosheets by inducing larger leakage currentand a faster increase of leakage current.Mechanism for thickness-dependent dielectric strengthweakeningBoth RVS and CVS tests demonstrate the mechanical stress-induceddielectric strength weakening of hBN nanosheets, with thinnernanosheets showing more obvious weakening. Mechanical stress-induced dielectric strength weakening has been observed in tradi-tional dielectrics. Zeller and Steve et al. have found that mechanicalstress can effectively reduce the breakdown strength of SiO2 films25,48.They have speculated that the effects of electric field and mechanicalstress on breakdown failure in dielectrics are basically the same, andfailure occurs when the local electric field and mechanical stressaccumulate to a limit. Recently, Mario et al. reported that the pressureapplied by the probe on the top of the metal/insulator/metal (MIM)memristor can reduce the breakdown voltage49. In the study ofdielectric ceramics, Li et al. calculated the vonMises stress distributionand mentioned that strain reduces the breakdown electric field bypromoting microcrack initiation and accelerating local discharge50.0 10 20 30 40 50 601E−71E−61E−51E−4)A(tnerruCVoltage (V)0 10 20 30 40 50 601E−71E−61E−51E−4Voltage (V)Current (A)0 2 4 6 8 10 12 14036912E)mc/VM(Cycles (#)036912150 2 4 6 8 10 12 14036912E (MV/cm)Cycles (#)03691215EBDELCEBDELCTest cycles1 3141 1dca baPM 094 ssertS   aPM 0 ssertSΔE = EBD-ELC ΔE = EBD-ELCthBN=41.3 nm thBN=41.3 nm∆E(MV/cm)∆E(MV/cm)Test cyclesaPM 094 ssertS   aPM 0 ssertSFig. 3 | Single-sample RVS tests on the dielectric strength of hBN nanosheetsundermechanical stress. a, b Sequential I–V curves of single-sample RVS tests onthe dielectric strength of 41.3 nm-thick nanosheets with 0MPa (a) and 490MPa (b)stress. In single-sample RVS tests, repeated RVS tests are performed on the sameisland sample. c, d EBD (purple horizontal lines), ELC (green horizontal lines), andtheir difference (ΔE = EBD−ELC, solid points) of the sequential I–V curves of single-sample RVS tests in a (c) and b (d). The solid lines in c and d are the exponentialfitting of ΔE. Source data are provided as a Source Data file.Article https://doi.org/10.1038/s41467-025-63358-6Nature Communications |         (2025) 16:8078 5www.nature.com/naturecommunicationsAlong with traditional dielectric research, the electro-fracturemechanics model48 and the filamentary electromechanical break-down model51 based on fracture mechanics have been proposed andcontinuously improved. Regarding stress-induced overall compres-sion in thickness and subsequent strength enhancement of the electricfield, the thickness compression of hBN nanosheets is calculated to beless than about 2% (see Section 9 of Supplementary Information). Sucha small thickness compression cannot make hBN show remarkabledielectric strength weakening by the enhancing strength of theelectric field.So, what is the mechanism responsible for the mechanical stress-induced dielectric strength weakening in the 2D hBN? Progressivebreakdown of hBN under an electric field can be attributed to variousmechanisms, including anode hole injection52, anode hydrogenrelease53, atomic migration54, collision ionization55, and so on. Thesedefect-based factors are directly related to the application of theelectric field and the subsequent flow of electrical current, so wetherefore summarize them as the electric field-induced defects.Dielectric breakdown of hBN without mechanical stress is directlyrelated to the formation and accumulationof the electric field-induceddefects. Similar to an electric field, mechanical stress can inducedefects in 2D materials from the atomic level to the crystallographiclevel, including lattice deformation56, dislocations57, and cracks58. Inaddition to the above defects induced by mechanical stress, mechan-ical stress may accelerate the aggregation and expansion of pre-existing electric field-induced defects59,60. We here refer to all thesemechanical factors that are likely to cause dielectric breakdown asmechanical stress-induced defects. The mechanical stress-induceddefects are expected to result in dielectric breakdown in a similar wayto that of electric field-induced defects. When an hBN nanosheet issubjected to both electric field and mechanical stress, the effects ofelectric field and mechanical stress on dielectric breakdown will becoupled together and superimposed, showing accelerated breakdowncompared with the case without mechanical stress. Therefore,mechanical stress-induced dielectric strength weakening in 2D hBN isattributed to accelerated formation and accumulation of defects bymechanical stress.To further illustrate themechanism responsible for the thickness-dependent dielectric strength weakening in 2D hBN, we conductedFEM simulations to study the internal stress distribution in hBNnanosheets. Figure 5a shows the distribution of the von Mises stress,which is closely related to the defect accumulation and failure ofmaterials61, in 40 nm-thick hBN nanosheets. It can be seen that thestressed area in the hBN nanosheet is larger than that where uniaxialstress is applied because of the bending stiffness of the topmetal layer.Importantly, the stress at the bottom surface of the nanosheets issmaller than that at the upper surface, indicating a stress gradient inthe thickness direction. The gradual attenuation of stress is caused bythe gradual expansion of the stressed area from the top to the bottomand is found to be strongly dependent on the thickness. The ratio ofthe stress at the bottom surface (Pbottom) to that at the upper surface(Ptop) is found to decreasewith the thickness increase (Fig. 5b). Pbottom/Ptop of the von Mises stress is 73.3% for a 4 nm-thick nanosheet anddecreases to 35.0% for a 40 nm-thick nanosheet. The ratio of normalstress (S33) shows a similar decrease with the increase in thickness. Thethickness-dependent stress gradient is thought to be responsible forthe thickness-dependent dielectric strength weakening.Dielectric breakdown of hBN nanosheets under mechanical stressis schematically shown in Fig. 5c. For thin nanosheets, internal stressaccumulation and mechanical stress-induced defects (see green dots)show similar density across the thickness because of a small mechan-ical stress gradient, so the progressive breakdown across the thickness0 200 400 60002004006002e-71e-610 1001E−81E−70 50 100 1501E−71E−61E−51E−410 100 10001E−71E−61E−51E−410 100 10001E−71E−61E−51E−410 100 10001E−71E−61E−51E−4Time (s)Current (A)8.9 nm12 VTime (s))A( tnerruC4~6 nmTime (s)Current (A)10~13 nmTime (s)Current (A)4~6 nm10~13 nmT LC(s)Stress (MPa)a b cd e fTime (s)Current (A)400~600 MPa0 MPa100~300 MPaMulti-sample CVS test    Bias: 5 V37~42 nm400~600 MPa0 MPa100~300 MPaMulti-sample CVS test    Bias: 13 V Multi-sample CVS test    Bias: 30 V0 MPa0 MPa8.9 nm13 VC=0.6C=1.5C=1.4C=2.3Single-sample CVS testSingle-sample CVS test539 MPa 188 MPaFig. 4 | Constant voltage stress (CVS) tests on the dielectric strength of hBNnanosheets under mechanical stress. a–c Current (I)–time (T) curves of multi-sample CVS tests on hBN nanosheets with thicknesses of (a) 4–6 nm, (b) 10–13 nm,and (c) 37–42 nm when different stress is applied. The green, yellow, and bluecurves correspond to the stress range of 0, 100–300, and 400–600MPa, respec-tively. The time for the onset of the obvious leakage current is defined as TLC. d Theplots of TLC in (a) and b versus stress for hBN nanosheets with 4–6 and 10–13 nm.The dashed lines are a linearfit of the data points. e I–T curves of single-sample CVStests on 8.9 nm-thick hBN nanosheets when they are stressed (dark blue/yellowlines for 13/12 V) and unstressed (light blue/yellow lines for 13/12 V). f I–T curveszoomed in near the onset of obvious leakage current in (e). The reddashed lines arepower function fitting of I–T curves with the formula I =A(T−TLC)C, where A and Care constants. The value ofC is shown for each curve. Source data are provided as aSource Data file.Article https://doi.org/10.1038/s41467-025-63358-6Nature Communications |         (2025) 16:8078 6www.nature.com/naturecommunicationscan be easily sped up by mechanical stress. In contrast, for thicknanosheets, the internal stress and mechanical stress-induced defectsnear the bottom surface are much smaller than those near the uppersurface because of the large internal stress gradient. So, comparedwith thin nanosheets, it is more difficult to achieve dielectric break-down across the whole thickness under mechanical stress in thicknanosheets. Thicker nanosheets are less susceptible to themechanicalstress-induced dielectric strength weakening. Thickness-dependentdielectric strength weakening observed in our experiments cantherefore be well explained by the mechanism that internal stressaccelerates the formation and accumulation of defects in 2D hBN andthus accelerates the breakdown process. This further supports theproposed mechanism for mechanical stress-induced dielectricstrength weakening in 2D hBN.Importantly, the attenuation of internal stress along the thicknessdirection is closely related to the layered structure of hBN nanosheets.For comparison, we calculate the stress distribution in isotropic SiO2thin films with different thicknesses (Figs. 5b and S9). Even thoughstress attenuation is also observed in SiO2 films, hBN exhibits muchmore obvious attenuation of mechanical stress than that of SiO2,especially in von Mises stress. When the thickness increases from 4 to40nm, similar to that in our experiments, Pbottom/Ptop of the vonMisesstress in SiO2 decreases from 96.7% to 82.3%, while it decreases from73.3% to 35.0% in hBN. In addition to the more obvious attenuation ofvon Mises stress in hBN for a specific thickness, stress attenuation inhBN also exhibits a stronger thickness dependence than that in SiO2.The phenomena are attributed to the larger lateral expansion ofstressed areas in anisotropic layered structures than that in isotropicSiO2. Our observed thickness-dependent dielectric strength weaken-ing of hBN nanosheets under mechanical stress is therefore closelyrelated to their layered structures.In summary, the dielectric strength of hBN nanosheets undermechanical stress is systematically studied by multi-sample/single-sample RVS/CVS tests. The dielectric strength of hBN nanosheets isfound to be weakened with lower breakdown strength, shorterbreakdown time and larger leakage current under normal mechanicalstress. The dielectric strength weakening is found to show thicknessdependence,withmore remarkableweakening for thinner nanosheets,which is closely related to the layered structures of hBN nanosheets.The ability of hBNnanosheets to block leakage current in cyclic voltageapplication can be significantly degraded by mechanical stress, evenfor thick nanosheets up to 41.3 nm. The results indicate that hBNnanosheets are not ideal insulators for 2D material-based devicesexposed tomechanical stress, and special cautions are needed in caseswith possible mechanical stress, especially for those on flexiblesubstrates.MethodsSample preparationAu/Ti film (thickness 200/20 nm) is firstly deposited on SiO2/Si sub-strate (~285 nm/500μm in thickness) using electron beamevaporation(Texas Instruments DE400) method. High-quality hBN nanosheetswith different thicknesses (<50nm) are obtained by the mechanicalexfoliation method from hBN bulk crystals synthesized by the high-pressure high-temperature method62 and transferred to the substratewith an Au film using thermal-release tape. The thickness of hBNnanosheets is measured by atomic force microscopy (AFM). hBNnanosheetswith uniform thickness in a large area are selectedunder anoptical microscope. An array of Au/Ti metal disks with a thickness of160/20 nm and a diameter of 1μm is prepared on the surface of thehBNnanosheets usingmicrofabrication techniques, including electronbeam lithography (Raith150 Two), electron beam evaporation, and lift-off process. The array of metal/hBN/metal islands is obtained byetching away hBN surrounding the metal disks through reactiveion etching (ETCHLAB 200) (O2/CHF3, 4/40 sccm, 60W, 5min) usingthe metal disks as the marks (see Fig. S2 in SupplementaryInformation).Dielectric strength tests under mechanical stressDielectric strength tests are performed at roomtemperature inside thevacuum chamber (~10−3 Pa) of an SEM (FEI Quanta 600F) equippedwith two nanomanipulators (Kleindiek MM3A). The substrate with anarray of metal/hBN/metal islands is fixed to a spring stage that is heldby a nanomanipulator. A tungsten (W) probe held by another nano-manipulator is manipulated to indent a metal/hBN/metal island toapply normal mechanical pressure to the hBN nanosheets. Electricalvoltage and current of metal/hBN/metal islands are sourced andmeasured by using a semiconductor parameter analyzer (Keithley4200-SCS) through the two nanomanipulators.Reporting summaryFurther information on research design is available in the NaturePortfolio Reporting Summary linked to this article.Data availabilityAll raw data generated during the study are available from the corre-sponding author upon request. Source data are provided withthis paper.VThin hBNThick hBNNormal stressNormal stressMechanical stress-induced defectsElectric field-induced defectsVcAuNormal stress 400 MPavon Mises stressTihBNa200 nmtBN335 MPa0 MPatBN (nm)hBN-S33 stresshBN-von Mises stressPbottom / PtopbSiO2-S33 stressSiO2-von Mises stress4 10 40 1000%20%40%60%80%100%Fig. 5 | Simulation of mechanical stress in hBN nanosheets. a Cloud diagram ofvon Mises stress distribution in a hBN nanosheet with 40nm thickness based onfinite element method (FEM) simulation. In the simulation, an island sample withthe same metal/hBN/metal structure as that in our experiments is indented with400MPa pressure in the center area with 200nm in diameter, similar to that in ourexperiments. b The ratio of the simulated stress at the bottom surface (Pbottom) tothat at the upper surface (Ptop) of hBN and SiO2 nanosheets with differentthicknesses. Both von Mises stress (solid dots) and the c-axis S33 stress (hollowdots) are displayed. Source data are provided as a Source Data file. c Schematicdrawing showing dielectric breakdown (meandering lines) in thin (upper figure)and thick (bottom figure) hBN nanosheets under electric field and mechanicalstress. The defects induced by the electric field and mechanical stress are repre-sented by purple and green dots, respectively.Article https://doi.org/10.1038/s41467-025-63358-6Nature Communications |         (2025) 16:8078 7www.nature.com/naturecommunicationsReferences1. Knobloch, T. et al. The performance limits of hexagonal boronnitride as an insulator for scaled CMOS devices based on two-dimensional materials. Nat. Electron. 4, 98–108 (2021).2. Sasama, Y. et al. High-mobility p-channel wide-bandgap transistorsbased on hydrogen-terminated diamond/hexagonal boron nitrideheterostructures. Nat. Electron. 5, 37–44 (2022).3. Dean, C. R. et al. Boron nitride substrates for high-quality grapheneelectronics. Nat. Nanotechnol. 5, 722–726 (2010).4. Dean, C. R. et al. Multicomponent fractional quantum Hall effect ingraphene. Nat. Phys. 7, 693–696 (2011).5. Wang, J. et al. High mobility MoS2 transistor with low Schottkybarrier contact by using atomic thick h-BN as a tunneling layer.Adv.Mater. 28, 8302–8308 (2016).6. Britnell, L. et al. Field-effect tunneling transistor based on verticalgraphene heterostructures. Science 335, 947–950 (2012).7. Purdie, D. G. et al. Cleaning interfaces in layered materials hetero-structures. Nat. Commun. 9, 5387 (2018).8. Wang, L. et al. One-dimensional electrical contact to a two-dimensional material. Science 342, 614–617 (2013).9. Cao, Y. et al. Correlated insulator behaviour at half-filling in magic-angle graphene superlattices. Nature 556, 80–84 (2018).10. Chen, G. et al. Signatures of tunable superconductivity in a trilayergraphene moiré superlattice. Nature 572, 215–219 (2019).11. Yankowitz,M.,Ma,Q., Jarillo-Herrero, P. & LeRoy, B. J. van derWaalsheterostructures combining graphene and hexagonal boronnitride. Nat. Rev. Phys. 1, 112–125 (2019).12. Zeng, S., Liu, C. & Zhou, P. Transistor engineering based on 2Dmaterials in the post-silicon era. Nat. Rev. Electr. Eng. 1,335–348 (2024).13. Yang, Q. et al. Steep-slope vertical-transport transistors built fromsub-5 nm Thin van der Waals heterostructures. Nat. Commun. 15,1138 (2024).14. Akinwande, D., Petrone, N. & Hone, J. Two-dimensional flexiblenanoelectronics. Nat. Commun. 5, 5678 (2014).15. Sangwan, V. K. & Hersam, M. C. Neuromorphic nanoelectronicmaterials. Nat. Nanotechnol. 15, 517–528 (2020).16. Jin, C. et al. Interlayer electron–phonon coupling in WSe2/hBNheterostructures. Nat. Phys. 13, 127–131 (2017).17. Geim, A. K. & Grigorieva, I. V. Van der Waals heterostructures.Nature 499, 419–425 (2013).18. Shimazaki, Y. et al. Strongly correlated electrons and hybrid exci-tons in a moiré heterostructure. Nature 580, 472–477 (2020).19. Khestanova, E., Guinea, F., Fumagalli, L., Geim, A. K. & Grigorieva, I.V. Universal shape and pressure inside bubbles appearing in vander Waals heterostructures. Nat. Commun. 7, 12587 (2016).20. Bera, K., Chugh, D., Tan, H. H., Roy, A. & Jagadish, C. Non-thermaland thermal effects on mechanical strain in substrate-transferredwafer-scale hBN films. J. Appl. Phys. 132, 104303 (2022).21. Kriegel,M. A.,Omambac, K.M., Franzka, S.,Meyer zuHeringdorf, F.-J. &Horn-vonHoegen,M. Incommensurability andnegative thermalexpansion of single layer hexagonal boron nitride. Appl. Surf. Sci.624, 157156 (2023).22. Tang, Z. et al. A steep-slope MoS2/graphene Dirac-source field-effect transistor with a large drive current. Nano Lett. 21,1758–1764 (2021).23. Lee, G.-H. et al. Flexible and transparent MoS2 field-effect transis-tors on hexagonal boron nitride–graphene heterostructures. ACSNano 7, 7931–7936 (2013).24. Georgiou, T. et al. Vertical field-effect transistor based ongraphene–WS2 heterostructures for flexible and transparent elec-tronics. Nat. Nanotechnol. 8, 100–103 (2013).25. Jeffery, S., Sofield, C. J. & Pethica, J. B. The influence of mechanicalstress on the dielectric breakdown field strength of thin SiO2 films.Appl. Phys. Lett. 73, 172–174 (1998).26. Choi, Y. S., Park, H., Nishida, T. & Thompson, S. E. Reliability ofHfSiON gate dielectric silicon MOS devices under [110] mechanicalstress: time dependent dielectric breakdown. J. Appl. Phys. 105,044503 (2009).27. Cho, C. et al. Strain-resilient electrical functionality in thin-filmmetal electrodes using two-dimensional interlayers. Nat. Electron.4, 126–133 (2021).28. Hattori, Y., Taniguchi, T., Watanabe, K. & Nagashio, K. Comparisonof device structures for the dielectric breakdown measurement ofhexagonal boron nitride. Appl. Phys. Lett. 109, 253111 (2016).29. Cook, J. T. et al. Temperature-dependent dielectric properties ofpolyimide (PI) and polyamide (PA) nanocomposites. IEEE Trans.Nanotechnol. 20, 584–591 (2021).30. Zhou, J. et al. Temperature-dependent breakdown and pre-breakdown conduction of polyethylene terephthalate. J. Phys. D:Appl. Phys. 55, 365302 (2022).31. Luan, B. & Robbins, M. O. The breakdown of continuummodels formechanical contacts. Nature 435, 929–932 (2005).32. Stifter, T., Marti, O. & Bhushan, B. Theoretical investigation of thedistance dependence of capillary and van der Waals forces inscanning force microscopy. Phys. Rev. B 62, 13667–13673 (2000).33. Day, H. C. & Allee, D. R. Selective area oxidation of silicon with ascanning force microscope. Appl. Phys. Lett. 62, 2691–2693 (1993).34. Avouris, P., Martel, R., Hertel, T. & Sandstrom, R. AFM-tip-inducedand current-induced local oxidation of silicon and metals. Appl.Phys. A 66, S659–S667 (1998).35. Hattori, Y., Taniguchi, T.,Watanabe, K. &Nagashio, K. Layer-by-layerdielectric breakdown of hexagonal boron nitride. ACS Nano 9,916–921 (2015).36. Ranjan, A. et al. Dielectric breakdown in single-crystal hexagonalboron nitride. ACS Appl. Electron. Mater. 3, 3547–3554 (2021).37. Shi, Y. et al. Electronic synapses made of layered two-dimensionalmaterials. Nat. Electron. 1, 458–465 (2018).38. Pan, C. et al. Coexistence of grain-boundaries-assisted bipolar andthreshold resistive switching in multilayer hexagonal boron nitride.Adv. Funct. Mater. 27, 1604811 (2017).39. Ranjan, A. et al. Molecular bridges link monolayers of hexagonalboron nitride during dielectric breakdown. ACS Appl. Electron.Mater. 5, 1262–1276 (2023).40. Shen, Y. et al. Two-dimensional-materials-based transistors usinghexagonal boron nitride dielectrics and metal gate electrodes withhigh cohesive energy. Nat. Electron. 7, 856–867 (2024).41. Wang, B. et al. Large and pressure-dependent c-axis piezoresistivityof highly oriented pyrolytic graphite near zero pressure. Nano Lett.24, 4965–4971 (2024).42. Li, J., Zhang, G., Wang, L. & Dai, Z. Indentation of a plate on a thintransversely isotropic elastic layer. Acta Mech. Solida Sin. 38,331–340 (2024).43. Palumbo, F. et al. A review on dielectric breakdown in thin dielec-trics: silicon dioxide, high-K, and layered dielectrics. Adv. Funct.Mater. 30, 1900657 (2020).44. Shimizu, T. et al, A new aspect of time-dependent clusteringmodelfor non-uniform dielectric TDDB. 2016 IEEE International ReliabilityPhysics Symposium (IRPS), pp. 3A-4-1-3A-4-10 (IEEE, Pasadena, CA,USA, 2016)45. Beek, S. V. et al. Four point probe ramped voltage stress as anefficient method to understand breakdown of STT-MRAM MgOtunnel junctions. 2015 IEEE International Reliability Physics Sympo-sium (IRPS), pp. MY.4.1-MY.4.6 (IEEE, Monterey, CA, USA, 2015).46. Pazos, S. et al. High-temporal-resolution characterization revealsoutstanding random telegraph noise and the origin of dielectricbreakdown in h-BN memristors. Adv. Funct. Mater. 34, 2213816(2024).47. Mannequin, C. et al. Stress-induced leakage current and trap gen-eration in HfO2 thin films. J. Appl. Phys. 112, 074103 (2012).Article https://doi.org/10.1038/s41467-025-63358-6Nature Communications |         (2025) 16:8078 8www.nature.com/naturecommunications48. Zeller, H. R. & Schneider, W. R. Electrofracture mechanics ofdielectric aging. J. Appl. Phys. 56, 455–459 (1984).49. Zuo, Y. et al. Effect of the pressure exerted by probe station tips inthe electrical characteristics of memristors. Adv. Electron. Mater. 6,1901226 (2020).50. Li, J. et al.Grain-orientation-engineeredmultilayer ceramiccapacitorsfor energy storage applications. Nat. Mater. 19, 999–1005 (2020).51. Fothergill, J. C. Filamentary electromechanical breakdown. IEEETrans. Electr. Insul. 26, 1124–1129 (1991).52. Hattori, Y., Taniguchi, T., Watanabe, K. & Nagashio, K. Impact ioni-zation and transport properties of hexagonal boron nitride in aconstant-voltage measurement. Phys. Rev. B 97, 045425 (2018).53. Hattori, Y., Taniguchi, T., Watanabe, K. & Nagashio, K. Anisotropicdielectric breakdown strength of single crystal hexagonal boronnitride. ACS Appl. Mater. Interfaces 8, 27877–27884 (2016).54. Jiang, L. et al. Dielectric breakdown in chemical vapor depositedhexagonal boron nitride. ACS Appl. Mater. Interfaces 9,39758–39770 (2017).55. Guiot, V. et al. Avalanche breakdown in GaTa4Se8−xTex narrow-gapMott insulators. Nat. Commun. 4, 1722 (2013).56. Rooney, A. P. et al. Anomalous twin boundaries in two dimensionalmaterials. Nat. Commun. 9, 3597 (2018).57. Ly, T. H., Zhao, J., Cichocka, M. O., Li, L.-J. & Lee, Y. H. Dynamicalobservations on the crack tip zone and stress corrosion of two-dimensional MoS2. Nat. Commun. 8, 14116 (2017).58. Yang, Y. et al. Intrinsic toughening and stable crack propagation inhexagonal boron nitride. Nature 594, 57–61 (2021).59. Kruv, A. et al. On the impact of mechanical stress on gate oxidetrapping. 2020 IEEE International Reliability Physics Symposium(IRPS), pp. 1-5 (IEEE, Dallas, TX, USA, 2020).60. Schneider, G. A. Influence of electric field andmechanical stresseson the fracture of ferroelectrics. Annu. Rev. 37, 491–538 (2007).61. Vinod, S. et al. Low-density three-dimensional foam using self-reinforced hybrid two-dimensional atomic layers. Nat. Commun. 5,4541 (2014).62. Taniguchi, T. & Watanabe, K. Synthesis of high-purity boron nitridesingle crystals under high pressure by using Ba–BN solvent. J.Crystallogr. Growth 303, 525–529 (2007).AcknowledgementsThis work was supported by the National Natural Science Foundation ofChina (Grant Nos. 62350040 and 11890671). The authors acknowledgethe support of the Nanofabrication Laboratory of Peking University. K.W.and T.T. acknowledge support from the JSPS KAKENHI (Grant Nos.21H05233 and 23H02052) and World Premier International ResearchCenter Initiative (WPI), MEXT, Japan.Author contributionsX.W. conceived the idea and supervised the investigation. B.W., X.W.,and Q.C. designed the experiments. B.W. fabricated the devices andperformed dielectric strength measurements. B.W., C.Y. and Z.D. per-formed FEM simulations and analysis. Y.J., M.L. andW.W. contributed todevice fabrication. C.T. contributed useful discussions. J.T., C.Y. andS.L.took TEM and AFM images. Z.F. facilitated the force measurement. T.T.and K.W. provided bulk hBN crystals. X.W., B.W. and Q.C. wrote themanuscript. All the authors read the manuscript and providedcomments.Competing interestsThe authors declare no competing interests.Additional informationSupplementary information The online version containssupplementary material available athttps://doi.org/10.1038/s41467-025-63358-6.Correspondence and requests for materials should be addressed toXianlong Wei.Peer review information Nature Communications thanks Mark Bissett,and the other, anonymous, reviewer(s) for their contribution to the peerreview of this work. A peer review file is available.Reprints and permissions information is available athttp://www.nature.com/reprintsPublisher’s note Springer Nature remains neutral with regard to jur-isdictional claims in published maps and institutional affiliations.Open Access This article is licensed under a Creative CommonsAttribution-NonCommercial-NoDerivatives 4.0 International License,which permits any non-commercial use, sharing, distribution andreproduction in any medium or format, as long as you give appropriatecredit to the original author(s) and the source, provide a link to theCreative Commons licence, and indicate if you modified the licensedmaterial. Youdonot havepermissionunder this licence toshare adaptedmaterial derived from this article or parts of it. The images or other thirdparty material in this article are included in the article’s CreativeCommons licence, unless indicated otherwise in a credit line to thematerial. If material is not included in the article’s Creative Commonslicence and your intended use is not permitted by statutory regulation orexceeds the permitted use, you will need to obtain permission directlyfrom the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.© The Author(s) 2025Article https://doi.org/10.1038/s41467-025-63358-6Nature Communications |         (2025) 16:8078 9https://doi.org/10.1038/s41467-025-63358-6http://www.nature.com/reprintshttp://creativecommons.org/licenses/by-nc-nd/4.0/http://creativecommons.org/licenses/by-nc-nd/4.0/www.nature.com/naturecommunications Dielectric strength weakening of hexagonal boron nitride nanosheets under mechanical stress Results and discussion Dielectric strength weakening in ramped voltage stress tests Dielectric strength weakening in constant voltage stress tests Mechanism for thickness-dependent dielectric strength weakening Methods Sample preparation Dielectric strength tests under mechanical stress Reporting summary Data availability References Acknowledgements Author contributions Competing interests Additional information