Direct measurement of tensile mechanical properties of few-layer hexagonal boron nitride (h-BN)

Hexagonal boron nitride (h-BN), known as “white graphene,” contains alternating boron and nitrogen atoms in a honeycomb lattice structure, which possess high thermal conductivity, ultra-wide bandgap, and low dielectric coefficient.^1–6 h-BN has been widely employed as the encapsulation, dielectric layer, or substrate for two-dimensional (2D) material-based field-effect transistors (FETs), tunneling devices, and nanocomposites.^7–10 Recent studies have revealed that h-BN possesses a distinctive combination of optical properties including defects-related single-photon emission, natural hyperbolic behavior, and bright deep ultraviolet (DUV) emission, which hold potential for applications from infrared nanophotonic devices to DUV emitters.^11–15 On the other hand, h-BN demonstrates excellent mechanical properties such as high Young's modulus and fracture strength.^{16, 17} It has been reported that monolayer h-BN possesses inherent fracture toughness and stable crack propagation capabilities, which are related to the asymmetric edge elastic characteristics, making its fracture behavior distinctly different from that of graphene.¹⁸ Furthermore, monolayer h-BN has been proven to have a great capacity for elastic deformation, which exhibits an elastic strain as high as 6.2%.¹⁹ These exceptional mechanical properties not only contribute to enhancing the service life of the h-BN-based devices but also make elastic strain engineering possible to tune the optoelectronic properties of h-BN.²⁰ For example, elastic tensile straining was applied for reversible tuning of the quantum emission characteristics and related photonic properties of few-layer h-BN.²¹ It is noteworthy that due to their superior mechanical stiffness and dielectric properties compared to monolayers, few/multi-layer h-BN are preferred over monolayers in practical applications.^22,23 However, current experimental measurement of the tensile mechanical properties of h-BN focuses on monolayer h-BN, and the experimental assessments of the tensile straining limits of h-BN with varying layer numbers are still notably absent.

To date, the prevailing method for assessing the mechanical properties of 2D materials has been the atomic force microscope (AFM)-based nanoindentation, allowing for the measurement of Young's modulus and strength.^24,25 However, it remains unreliable for determining the tensile straining limit and fracture strength of these materials.²⁶ In recent years, the micro-electromechanical system (MEMS)-based uniaxial tensile technology has emerged as a promising approach for accurately measuring the tensile mechanical properties of 2D materials.^27,28 Nevertheless, conducting uniaxial tensile experiments on mechanically exfoliated 2D nanosheets remains highly challenging, especially in terms of nondestructive and precise sample transfer. To address this challenge, here, we developed a sophisticated experimental protocol to achieve the quantitative measurement of mechanical properties of the suspended few-layer h-BN with different layer numbers, highlighted by a deterministic sample transfer method. Young's modulus and fracture strain of single-crystalline few-layer h-BN were obtained from the in situ tensile testing. Post-mortem analysis of the crack pattern revealed an abnormal layer number-dependent fracture behavior, indicating a reduction in toughness with the increase of layers, which could be interpreted by the interaction between the strong interlayer coupling and intralayer covalent bonding. This study not only attests the intrinsic mechanical robustness of the h-BN nanosheets but, more importantly, suggests a previously undiscovered tough-to-brittle transition mechanism in multi-layer h-BN.

Few-layer h-BN flakes were exfoliated onto a piece of SiO_x/Si wafer (pre-treated by oxygen plasma to remove contamination and increase surface adhesion) from bulk h-BN flakes synthesized by a high-pressure/high-temperature (HPHT) method,²⁹ after which the few-layer-thick sheets were selected based on optical contrast examination and inspected by Raman spectroscopy (WITec alpha300R) [Fig. 1(a)]. The sharp Raman peak centered at 1366 cm⁻¹ corresponds to the G band frequency of h-BN, indicating the high quality of the exfoliated thin h-BN sheets.^16,30 The single crystalline nature of the h-BN flakes was also confirmed by high-resolution transmission electron microscopy (HRTEM) characterization (JEOL JEM-2100F, 200 kV), with the selected area electron diffraction (SAED) pattern showing one set of hexagonal spots and sample-wide uniformity [Fig. 1(b)].

To perform the in situ tensile test, we need to transfer the chosen h-BN flake onto a push-to-pull (PTP) nanomechanical device [Fig. 1(d)],³¹ which raised a major technical challenge for a damage-free transfer method, considering the mechanical vulnerability of nanometer-thin h-BN, the micrometer-sized sample and target region, and the suspended target device setup. Here, we adopted a modified micromanipulator-assisted wedging transfer method with high efficiency and accuracy [Fig. 1(c)].³² To be specific, a thin polymethyl-methacrylate (PMMA) stamp (prepared by spin coating) was first attached to the h-BN flake through a hot press process (120 °C) as the intermediate carrier substrate, after which the Si wafer was slowly wedged into DI water with an inclination of about 30°, allowing water to penetrate the interface between the hydrophobic PMMA/h-BN stack and the hydrophilic SiO_x surface. During the wedging cleavage, the PMMA/h-BN stack will gradually peel off from the Si substrate and finally floating on the water surface. Next, the stack was transferred onto the PTP device and precisely positioned under the optical microscope by a tungsten needle loaded on a micromanipulator. A thin layer of water between the stack and the PTP device was preserved during the alignment to ensure smooth movement and avoid material damage due to direct contact. The h-BN flake was deposited onto the tensile region of the PTP device once the water layer evaporated, and the capillary effect provides robust attachment between the material and substrate, followed by a 100 °C annealing that further enhanced the adhesion. The final suspended h-BN device was obtained after the removal of PMMA using acetone and a critical point drying process (Leica EM CPD300), which could avoid surface tension damage to the nanometer thick material. To conduct quantitative mechanical measurement to the few-layer h-BN, the samples suspended on the PTP device were further trimmed by focused ion beam (FIB) (FEI Scios DualBeam, Ga⁺), to cut the to-be-tested flake into desired size and geometry and remove unwanted flakes covering gaps besides the tensile region [Figs. 1(d)–1(e)]. To minimize the sample damage from ion beam irradiation, moderate voltage and current (8 kV, 2 pA) were used.

A scanning electron microscope (SEM)-based in situ tensile test platform (Hysitron PI 85 PicoIndenter inside an FEI Quanta 450 Field Emission SEM) was used to study the elastic properties of the h-BN samples, enabling simultaneous acquisition of high-resolution, load–displacement data and high-magnification visualization of the deformation process. The tensile strain of the sample was calculated by a digital image correlation method based on the captured video frames. The gauge section was selected within the freestanding region of the sample between the clamps. Normally, the van der Waals interaction between the PTP substrate and the h-BN flakes is strong enough to prevent any slippage (as confirmed by experimental observation), meaning that the edges of the tensile gap of the PTP device coincide with the clamping edges, thus, define the gauge length (∼2.5 μm), as indicated by the yellow dashed lines in Figs. 2(a)–2(e). The tensile tests were performed under displacement control mode with a loading rate of 2 nm/s (corresponding to a sample strain rate of ∼8 × 10^–4 s⁻¹).

A series of loading–unloading cyclic tensile straining were applied to the free-standing h-BN nanosheet with increasing displacement, of which the results for one sample are shown in Fig. 2. The pristine sample was pendulous over the device gap and, thus, in pre-compression before fully stretched, as indicated by the wrinkles formed parallel to the clamping edges due to instability of the thin film [Fig. 2(a)]. As the tensile gap widened under indentation, the suspended h-BN sample was tightened to a flat state and then underwent tensile straining. The sample demonstrated fully elastic recovery with a maximum engineering strain of 3.5% before fracture [Figs. 2(b)–2(e)]. The fracture strain is lower than that of monolayer h-BN and graphene (∼6%),^19,27 possibly due to a higher defect density presented in the tested sample (the monolayers are more sensitive to defects that may lead to premature failure before test). A sudden failure with typical brittle fracture morphology was observed, and the post-mortem SAED analysis reveals a crack path along the armchair direction of the h-BN [Fig. 2(f); Fig. S1 in the supplementary material].

The corresponding load–displacement curves from the indenter are shown in Fig. 2(g), which agrees well with the observe deformation patterns. The initial linear region with smaller slope explains the intrinsic stiffness of the PTP device since the pre-compressed sample was not bearing any loading. The following rapid increase in the curve slope indicates the stretching and tightening of the h-BN sheet, and the inflection point marked with an orange arrow defines the beginning of sample straining (zero strain point). Once the sample was fully stretched, a second linear region with significantly larger slope was disclosed, which corresponds to the total stiffness of the PTP device and the h-BN. The sudden drop in the final loading cycle marks the brittle failure of the sample, and the remaining curve aligns with the initial linear stage where solely the PTP device was driven.

By subtracting the intrinsic stiffness (linear force-displacement response) of the PTP device measured from the load-–displacement data, the tensile force F on the sample can be isolated and converted to engineering stress by $σ= F W t$⁠, with sample width W measured from the SEM image and sample thickness t measured by AFM (Bruker dimension icon) (Fig. S2 in the supplementary material). Since it is reasonable to assume a sample-wide uniform strain distribution,^19,27 the stress–strain curve can be obtained [Fig. 2(h)], yielding a 3D Young's modulus of E_3D = ∼720 GPa and a fracture strength of ∼20 GPa. Based on the as-developed protocol, we performed tensile tests on a series of h-BN flakes with various thicknesses (Fig. 3). We noted that for nanometer-thin h-BN nanosheets, a more accurate approach to determine their thickness may be through HRTEM observation of the rolling edges due to their extreme flexibility (Fig. S3 in the supplementary material), which directly gives the layer number that can be converted to thickness by multiplying the interlayer spacing of 0.333 nm.³³ From the stress–strain curves acquired from samples without major defects, their Young's moduli, tensile strengths, and break elongations (fracture strains) were extracted and are summarized in Figs. 4(a)–4(c).

The calculated results from 3- to 8-layer h-BN show that Young's moduli are 573.8 ± 101.4 GPa and tensile strengths are 11.7 ± 5.9 GPa, which are lower than the previous experimental values of 865 ± 73 and 70.5 ± 5.5 GPa from AFM-based nanoindentation of few-layer single-crystalline h-BN and theoretical values from DFT calculation.^16,34 On the other hand, the fracture strength of CVD-grown single-crystalline monolayer h-BN has been measured as 7.9 ± 2.5 GPa,¹⁸ which is much lower due to its intrinsically high defect level commonly induced by CVD growth. A comparison of Young's modulus of h-BN measured by different methods is shown in Table S1 in the supplementary material. As for our mechanically exfoliated h-BN samples, the diminished mechanical performance compared with that from AFM nanoindentation can be attributed to the inevitable sample damage during the transfer process and the extensive edge defects induced by FIB irradiation. An amorphous region with a width of ∼25 nm was found along the Ga⁺ cutting edge (Fig. S4 in the supplementary material), with considerably lower mechanical robustness comparing with its crystalline form,^35,36 thus causing premature sample failure during the tensile test.

From the recorded videos, the crack propagation paths and fracture morphologies of the tested samples were also analyzed and compared. Although the calculated strength values show no obvious thickness dependence, we observed a difference in their crack patterns. Crack deflection and branching with segmented fracture edges were found in h-BN sheets with fewer layers (less than eight), clearly different from the eight-layer h-BN flakes presented above demonstrating a straight and sharp fracture path with no branches [Figs. 4(d)–4(g)]. The schematics in Fig. 4(h) illustrate the crack propagation paths of hBN with different number of layers. For few-layer hBN, the crack propagation is dominated by the intrinsic toughness of hBN, so the cracks are obviously deflected, and the cracks in different layers are probably staggered, which corresponds to higher fracture toughness. Comparing with the few-layer hBN, the crack propagation of multilayer hBN does not have obvious deflection, and the crack paths of different layers are close to straight lines. The results show that h-BN with different thicknesses will exhibit different fracture behaviors, and we speculate that there could be a transition from brittleness to toughness as the thickness decreasing for h-BN, especially when considering the recently reported high fracture toughness in monolayer h-BN.¹⁸ This proposed toughen-to-brittle transition in multilayer h-BN contrasts with the toughening discovered in multilayer graphene, which is attributed to asynchronous cracking due to weak interlayer coupling.^37–39 Here, we believe this fresh-new phenomenon can be explained by the interaction between the van der Waals interface and in-plane polar covalent bonds.^40,41 More specifically, the strong interlayer interaction as well as the resultant suppressed interlayer slippage in multilayer h-BN hinder the interplanar crack propagations, thus limiting elastic strain energy dissipation. This surmise is also evidenced by the absence of h-BN layer delamination along the crack edge. In addition, it is also possible that as the layer number increases, the out-of-plane deformation of the crack tip may be restrained by the strong interlayer bindings and further reduces the toughness.

In summary, an experimental protocol was developed for quantitative in situ tensile test of few-layer single-crystalline h-BN, including sample preparation, highly accurate transfer, and uniform straining. The h-BN nanosheets with thickness from three to eight layers had Young's moduli of 573.8 ± 101.4 GPa and tensile strengths of 11.7 ± 5.9 GPa, with elastic strains up to 3.5%. In addition, a layer number-dependent fracture behavior was observed, with the arising of toughening feature as the h-BN becoming thinner. These findings advance the understanding on tensile properties of few-layer h-BN and shed light on the design of h-BN-based electronics and structural materials.

See the supplementary material for additional figures (Figs. S1–S4; Table S1) and description for the supplementary video.

This work was supported by NSFC/RGC Joint Research Scheme (No. N_HKU159/22), Shenzhen-Hong Kong-Macau Technology Research Program (Type C, No. SGDX2020110309300301), and Research Grants Council of the Hong Kong Special Administrative Region, China under Grant No. RFS2021-1S05.

The authors have no conflicts to disclose.

Jingzhuo Zhou: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Writing – original draft (equal); Writing – review & editing (equal). Mengya Zhu: Investigation (equal); Methodology (equal); Writing – original draft (equal). Ying Han: Investigation (equal); Methodology (equal). Xuefeng Zhou: Investigation (equal). Shanmin Wang: Investigation (equal). Juzheng Chen: Writing – review & editing (equal). Hao Wu: Writing – review & editing (equal). Yuan Hou: Conceptualization (equal); Investigation (equal); Writing – original draft (equal); Writing – review & editing (equal). Yang Lu: Conceptualization (equal); Funding acquisition (equal); Supervision (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding authors upon reasonable request.