Synthesis of high-quality monolayer tungsten disulfide with chlorophylls and its application for enhancing bone regeneration

Due to the population explosion of the 21st century, nearly one billion people are over 64 years of age and bone fracture is one of the most frequent problems facing both sexes because of osteoporosis. However, difficulty in enhancing bone regeneration to repair bone fracture poses challenges and thus, a two-dimensional monolayer material (i.e. tungsten disulfide (WS2)) could be one of the candidates offering a possible solution to the problem. Here, we prepare high-quality monolayer WS2 thin sheets in a large quantity with the assistance of extracted chlorophyll molecules, the natural pigment used in photosynthesis, via a liquid-phase exfoliation method. Then, the exfoliated WS2 sheets were mixed with polycaprolactone (PCL)/calcium silicate (CS) to form a biocompatible WS2-based composite. The in vivo experiments show that the bone regeneration of the WS2-based composite was 120% superior to commercially available mineral trioxide aggregate (MTA) bone cement. Moreover, the mechanical properties of the WS2-based composite exhibited ~300% enhancement over PCL/CS, which is one of the most commonly used bone regeneration materials. Our findings highlight the prospects for the composite of WS2 towards the improvement of bone regeneration applications.


INTRODUCTION
In 2020, the world's population is close to eight billion 1 . Humankind is facing two critical challenges: population ageing and global climate change 2 . Regarding the first, nearly one billion people are over 64 years of age, and bone fracture is one of the most frequent problems for all people in this demographic because of osteoporosis 3 . In response, many synthetic biocompatible materials have been developed and used in bone tissue engineering. However, most biocompatible materials are insufficient in their mechanical properties 4 , lack bioactivity, have uncontrolled degradation rates 5 , and thus, cannot help with native bone regeneration for certain physiological conditions 6 . Therefore, developing biocompatible materials with superior mechanical properties and osteogenic bioactivity has attracted significant attention in the field of bone tissue regeneration 7,8 . To do so, discovering functional materials is fundamental to resolving the issues mentioned above.
Transition metal dichalcogenides (TMDs) have attracted much attention over the past ten years due to their intrinsic properties [9][10][11] . As the most representative TMD, molybdenum disulfide (MoS 2 ) has been intensively studied. Although tungsten disulfide (WS 2 ) has a crystal structure analogous to that of MoS 2 , it exhibits superior biocompatibility and physical and photoelectrochemical properties [12][13][14] . Lalwani et al. 15 demonstrated that WS 2 nanotubes can uniformly disperse in biodegradable polymers and are excellent reinforcing agents compared with carbon-based materials (e.g., carbon nanotubes). However, in vitro experimental results for WS 2 nanotubes are lacking. Raichman et al. used a highly electrophilic acidic Vilsmeier-Haack reagent to produce polycarboxylated functional WS 2 nanotubes that can be dispersed in polar liquids 16 . In 2018, they also found that the functionalised WS 2 is biocompatible and may come into contact with blood, indicating a high degree of biocompatibility 17 . However, there have only been a limited number of studies focussing on the preparation of WS 2 thin sheets via liquidphase exfoliation and the application of these sheets 18-21 . There has yet to be found a simple, eco-friendly, facile method for producing WS 2 thin sheets in large quantities owing to its insolubility in common solvents, and a low degree of compatibility with common polymers 16 . In this study, the authors report on an environmentally friendly and effective strategy for scalable production of high-quality WS 2 thin sheet suspensions with the assistance of a natural pigment that exists in abundance-chlorophyll molecules. With the strategy used in this study, exfoliated WS 2 thin sheets can be produced in high concentrations. They can be thoroughly mixed with polycaprolactone (PCL) and calcium silicate (CS), which are Food and Drug Administration (FDA) approved materials, to form a WS 2 /PCL/ CS biocompatible composite possessing an extraordinary bone regeneration efficiency of 120%, which is superior to commercially available mineral trioxide aggregate (MTA) bone cement, and it is non-toxic for stem cells. To the best of the authors' knowledge, the current study is the first on a WS 2 /PCL/CS composite to demonstrate an excellent degree of bone regeneration efficiency.

RESULTS AND DISCUSSION
Stability of WS 2 thin sheet suspension Images of the chlorophyll-assisted exfoliated WS 2 thin sheet suspensions are shown in Fig. 1a. Figure 1b shows that no decay in the absorbance of the suspension was observed, indicating the superior stability of this study's chlorophyll-assisted liquid-phase exfoliation method in the scalable production of WS 2 thin sheets. Figure 1c shows the optical absorbance spectra of the exfoliated WS 2 thin sheet suspensions and extracted chlorophyll solution, which exhibited characteristic excitonic transitions. According to the related empirical equation 22 , the layer number (N) of the sheets in the dispersions was close to unity. The low-resolution TEM ( Fig. 2a and Supplementary Fig. 1) images show that the bulk WS 2 was exfoliated into thin sheets with the help of extracted chlorophyll molecules. Supplementary Fig. 2 shows a high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image, and the elemental mapping results, indicating that the W and S atoms were homogeneously distributed. The atomic percentages of W and S were 28.9% and 54.7%, respectively. In the HRTEM image with atomic resolution (Fig. 2b), the sharp edge corresponds to a monolayer single-crystal WS 2 structure. The fast Fourier transformation (FFT) in Fig. 2c and Supplementary Fig. 3 exhibit reflection spots, while the pattern of the thin sheets indicates that the crystallinity and hexagonal structure were not significantly affected by probe sonication. Figure 2d shows the unpolarised Raman spectrum measured at 532 nm of excitation. The dominant peaks at~350 and~420 cm −1 were attributed to the E 1 2g and A 1g modes, respectively. Most importantly, the B 2g 1 phonon (~310 cm −1 ) peak has been reported to correspond to layer-layer interaction, which is active for multilayers but inactive for monolayers 23 . Figure 2d shows that the B 2g 1 phonon peak did not appear in the exfoliated highquality WS 2 thin sheets, indicating that they are monolayer in structure [23][24][25][26][27] . These results validate the successful large-scale production of monolayer WS 2 thin sheets having a relatively high crystallinity.
To further examine the chemical composition of the sheets, X-ray photoelectron spectroscopy (XPS) was utilised to measure the binding energy of W and S. Figure 3a shows the core-level spectrum of W 4 f and W 5p on the sheets. Three prominent peaks appear at 32.9, 35.1, and 37.8 eV, which correspond to the W 4f 7/2 , W 4f 5/2, and W 5p 3/2 components, respectively. The peak positions correspond to a trigonal prismatic configuration of W atoms and are identical to those of the semiconducting-phase WS 2 grown Fig. 1 Synthesis of high-quality WS2 thin sheet suspension via chlorophyll-assisted liquid-phase exfoliation. a Images of 1000 mL suspensions of bulk WS 2 in acetone with extracted chlorophyll molecules (sample A) and without extracted chlorophyll molecules (sample B); b Comparative stability of sample A (□) and sample B (Δ). 'A 0 ' represents the initial absorbance of the suspension, and 'A' depicts the measured absorbance after standing for certain periods of time; and c Characteristic absorbance of exfoliated WS 2 thin sheets and extracted chlorophyll molecules.   through other methods [28][29][30] . Meanwhile, the peak at 35.8 eV is ascribed to the characteristics of W 6+ . The appearance of W 6+ peaks might be on account of limited surface oxidisation of the sheets during exfoliation [30][31][32] . Similarly, in the core-level spectrum of S 2p, Figure 3b shows two prominent peaks at 162.2 and 163.2 eV, corresponding to the S 2p 3/2 and S 2p 1/2 components of WS 2 , respectively. These results are superior to that of electrochemically exfoliated WS 2 sheets that oxidise the pristine charge of W 4+ to W 5+ and W 6+ 33 . To gain further insight into the exfoliation mechanism, FL, FL lifetime, XPS and zeta potential measurements were conducted. The FL spectra of the extracted chlorophyll solution were measured on excitation at 610 nm, as shown by the black line in Supplementary Fig. 4a. Then, the WS 2 powder was added into the extracted chlorophyll solution without using sonication. The FL intensity of the extracted chlorophyll molecules were fully quenched as shown by the blue line in Supplementary Fig. 4a (Supplementary Video). Supplementary Fig.  4b shows the FL lifetime result of the extracted chlorophyll solution and exfoliated WS 2 thin sheets. The sheets exhibited no FL intensity, indicating that the FL property of the extracted chlorophyll molecules was quenched by the adsorbed WS 2 thin sheets. Supplementary Fig. 4c depicts the core-level spectrum of N 1s of the sheets. The binding energy locations of 399.7 and 401.6 eV represent the functional groups of the extracted chlorophyll [34][35][36] . The lack of a peak being located at 397.2 eV is ascribed to the absence of W-N bond formation between the sheets and the extracted chlorophyll molecules. According to the FL, FL lifetime, and N1s XPS data, the extracted chlorophyll molecules were physically adsorbed on the surface of the sheets. Moreover, the zeta potential of the sheets was −10.8 mV, as shown in Fig. 3c. The negative surface potential value is evidence of electrostatic repulsive forces between each of the sheets, resulting in a stable suspension. Xu et al. 19 utilised substoichiometric amounts of lithium to exfoliate bulk WS 2 to 1-2layer WS 2 nanosheets. However, this exfoliation process requires harsh conditions. Adilbekova et al. 37 used a less environmentally friendly alkaline solvent to exfoliate bulk WS 2 into thin sheets. However, the size of the sheets was only several tenths of a nanometre in length, which does not provide sufficient mechanical strength for bone regeneration. Coleman et al. 20 reported that the surface tension of an N-methyl-2-pyrrolido (NMP) solvent was in the range of 30-40 mJ m −2 , which is close to that of transition metals dichalcogenides. However, most of the WS 2 thin sheets prepared through this method were too thick. Furthermore, NMP is a highly toxic and high-temperature solvent which is not suitable for scalable production of WS 2 thin sheets. To the best of our knowledge, the current study is the first to demonstrate the scalable production of sub-micrometre sized monolayer WS 2 thin sheets with the assistance of extracted chlorophyll molecules under mild conditions. According to density functional theory (DFT) calculations, WS 2 has the potential to be amongst the most efficient electrocatalysts because its hydrogen adsorption free energy (ΔG H* ) on the WS 2 edge site is close to thermoneutral, which is similar to Pt-group metals. Liu et al. 38 demonstrated that synthesised WS 2 exhibited an onset potential of −130 mV and a Tafel slope of 80 mV dec −1 . Voiry et al. 18 found that Li-assisted exfoliated WS 2 thin sheets can be synthesised yielding a Tafel slope of 60 mV dec −1 . It is expected that the preparation of monolayer WS 2 sheets will significantly affect the catalytic performance. The electrochemical performance of the exfoliated monolayer WS 2 sheets, as the electrocatalyst on an Au electrode, were conducted using a three-electrode configuration in a 0.5 M H 2 SO 4 electrolyte. Supplementary Fig.  5a shows the polarisation curves of the assembled WS 2 films with different assembly times. The onset potential and the Tafel slope of the sample labelled 'WS 2 assembly 24 h' were −118 mV (at the current density of 1 mA cm −2 ) and 60 mV dec −1 ( Supplementary  Fig. 5b), respectively. Interestingly, unlike MoS 2 , bulk WS 2 in its semiconducting form is not perceived as a good hydrogen evolution reaction (HER) electrocatalyst 18,30,[38][39][40] , but here a superior electrocatalytic performance was achieved with the chlorophyll-assisted exfoliated WS 2 thin sheets. Either a too short or a too long an assembly time led to a decrease in HER activity on the WS 2 film. A possible reason for this is that <24 h was too short a time to form a compact self-assembled film, however, increasing the incubation to 36 h resulted in the stacking of the exfoliated sheets 30 . The reduced onset potential and lower Tafel slope of the WS 2 assembled 24 h sample indicate that the free energy of the hydrogen binding to the chlorophyll-assisted exfoliated WS 2 thin sheets is closer to the thermodynamic equilibrium 30 . The Tafel slope of the 36 h-assembled films was around 95 mV dec −1 , which is most likely related to the higher internal resistance of the film 41,42 . In order to further understand the reason for the enhanced electrocatalytic performance, XPS analysis was carried out. Supplementary Fig. 5c shows that the WS 2 assembly 24 h sample had two sets of binding energies. The S 2p 3/2 binding energy at around 162.7 eV represented the original S in WS 2 , and an S 2p 3/2 peak at 161.8 eV could be attributed to the S of the WS 2 assembled on the Au surface. It has been demonstrated that the electron transfer from electron-rich Au to electron-deficient S causes a decrease in binding energy for S, attesting to the presence of the S-Au bond 32,43 . Taking this into account, since the binding energy of 161.8 eV closely matches the binding energy of the S-Au bond 44 , the authors believe that the peak of 161.8 eV should be attributed to the formation of the S-Au bond during the assembly process. Supplementary Fig. 5d shows a good degree of stability, where the electrocatalytic activity did not greatly change after 20 h. Therefore, the authors believe that the chlorophyllassisted exfoliated WS 2 thin sheets have promise as a watersplitting electrocatalyst.
To enhance the HER performance of the exfoliated WS 2 thin sheets, their photoelectrochemical behaviour was further examined. Figure 4a shows the polarisation curves of the samples labelled 'WS 2 assembly 24-h-light off' and 'WS 2 assembly 24-hlight on'. Figure 4b shows that the latter sample exhibited a 277% current density increase at the potential of −50 mV. This demonstrates that the sample exhibited a superb degree of photoelectrocatalytic activity for water-splitting because the exfoliated WS 2 thin sheets can absorb visible light between 600 and 700 nm, as mentioned above. After loading limited PtNPs onto WS 2 , the corresponding photocurrent showed a significant increase to 4.524 mA cm −2 at −50 mV under visible-light illumination (the blue dash-dot line in Fig. 4c). Surprisingly, the onset potential of the sample labelled 'WS 2 assembly 24-h-light on-with PtNPs' was further improved to −14 mV (at the current density of 1 mA cm −2 ), as shown in Fig. 4d, which is far superior to values reported in other studies for WS 2 -based materials ( Supplementary  Fig. 6). Here, PtNPs play two roles in the photoelectrocatalytic reaction: as an electron sink to trap photo-excited electrons from the WS 2 sheets for water reduction 45,46 , and as a provider of efficient channels for electron transportation for direct water reduction 47,48 .
Physical properties of WS 2 /PCL/CS scaffold The potential use of chlorophyll-assisted exfoliated WS 2 thin sheets in a wide range of applications requires the predetermination of risks they pose in living organisms 49 . Therefore, the sheets were subjected to a series of experiments to evaluate their biocompatibility and bone regeneration. To mimic bone composition, bio-ceramics of CS and PCL polymers have been widely used [50][51][52][53][54] . These materials are commercially available, biodegradable, biocompatible, and approved by the FDA 55 . After fully mixing these materials with exfoliated WS 2 thin sheets to form a WS 2 /PCL/CS composite, a three-dimensional scaffold was successfully created by using three-dimensional printing. The printed scaffolds had a well-designed morphology (Fig. 5a). The interconnected macro-pore structure promotes cell in-growth and nutrient transportation, and their bioactive components can lead to osteogenesis and angiogenesis 56 . To understand the homogeneity of the chlorophyll-assisted exfoliated WS 2 thin sheets in the WS 2 /PCL/CS composite, non-destructive Raman measurements were used. Figure 5b shows that the mapping spectrum of the scaffold had a uniform WS 2 signal intensity, indicating that the exfoliated WS 2 thin sheets were homogeneously distributed in the WS 2 /PCL/CS composite. To further investigate the reinforcement effect of the sheets in the scaffold, compression testing was carried out. Figure 5c shows that the compression strength was enhanced from 2.0 MPa for a PCL/CS scaffold to 6.0 MPa for the WS 2 /PCL/CS scaffold, a 300% improvement. The Young's modulus of the PCL/CS scaffold was 32 MPa. The Young's modulus of the composite was significantly improved to 145 MPa, which is nearly a 450% improvement. The results of the compressive modulus and strength of the scaffold reached the range of human cancellous bone tissue 57 . Such reinforcement-enhancement is related to the original properties of the exfoliated sheets 58 , indicating the superior degree of dispersion of the sheets in the biocompatible matrix. In addition to understanding the biocompatibility and bioactivity of WS 2 /PCL/CS scaffold, precipitated spherical mineral aggregates were observed on the surfaces of the PCL/CS and WS 2 /PCL/CS scaffolds using FE-SEM. Full mineral aggregates coating both types of scaffold showed that apatite formation was not inhibited by the added WS 2 (Fig. 5d). Apatite layers are responsible for the biocompatibility and bioactivity of CS-based materials that are part of the interface of the scaffold and natural bone tissue 59 .
In vitro cell behaviour effects of WS 2 /PCL/CS scaffold To evaluate the effect of cells on WS 2 /PCL/CS scaffold, we used the human mesenchymal stem cells (hMSCs) to analyse Col I secretion, cell viability, alkaline phosphatase (ALP), and cell morphology. Figure 6a shows the amount of Col I adsorbed onto the scaffolds of a cell culture well (the control, abbreviated as Ctl), the PCL/CS, and the WS 2 /PCL/CS. The concentration of Col I on the WS 2 /PCL/ CS scaffold was over two times higher than the result of Ctl, indicating the effective binding of Col I on the scaffold with the added WS 2 . Col I contains several cell-binding sub-units, which in turn bind to components of cell membranes to positively influence the cell behaviour, such as adhesion and proliferation 60 . An understanding of the biocompatibility of the WS 2 /PCL/CS scaffold is required, as the efficiency of cell viability is one of the critical issues affecting the results of tissue regeneration. Figure 6b shows significant enhancement of the viability of the cultured hMSCs on the WS 2 /PCL/CS scaffolds, as compared to the Ctl at days 1, 3, and 7 (p < 0.05). This was especially true for the WS 2 / PCL/CS scaffolds as a nearly 30% enhancement of the cell viability, as compared with the PCL/CS scaffolds, was observed at Day 7. This is attributed to the WS 2 /PCL/CS scaffold offering better microenvironmental conditions for hMSCs viability 56 . To evaluate the osteoblastic differentiation efficiency in detail, the useful marker of ALP activity was performed. Figure 6c shows that the ALP activity for all scaffolds increased with time, and the ALP activity from hMSCs in WS 2 /PCL/CS scaffolds was significantly higher than that of the Ctl and the PCL/CS scaffold at days 3, 7, and 14 (p < 0.05), indicating that the WS 2 /PCL/CS scaffold ably supports the early steps of osteoblastic differentiation in hMSCs. Figure 6d shows that hMSCs had a better adhesion to the WS 2 /PCL/CS scaffold than to the PCL/CS scaffold after 12 h. The authors believe that the WS 2 /PCL/CS scaffold can adsorb more Col I on the surface so enhancing the initial cell attachment ability. Various studies have demonstrated that Col I can regulate the attachment, proliferation, and osteogenic differentiation of stem cells 56 . Overall, the cell adhesion and viability data demonstrate that the chlorophyllassisted exfoliated WS 2 thin sheets did not exhibit any cytotoxic responses to the hMSCs. Thus, this study opens possibilities in the development of chlorophyll-assisted exfoliated WS 2 thin sheets for tissue regeneration applications.
In vivo bone regeneration efficacy of WS 2 /PCL/CS scaffold In vivo experiments were carried out ( Supplementary Fig. 8) by evaluating bone regeneration in femoral defects in rabbits after the implantation of PCL/CS, WS 2 /PCL/CS, and MTA scaffolds. Figure 7a shows micro-computed tomography (µCT) images of the three types of scaffolds after implantation at four and eight weeks. The implanted WS 2 /PCL/CS scaffold specimen generated much more bone around the scaffold compared to the PCL/CS and MTA scaffolds. The BV/TV (Fig. 7b) and Tb.Th (Fig. 7c) values for the WS 2 /PCL/CS scaffold were much higher than those of the PCL/CS and MTA scaffolds (p < 0.05). In comparison to MTA, the best therapeutic efficiency (scored as the extent of in vivo bone regeneration) was exhibited by the WS 2 /PCL/CS scaffold. The results of the undecalcified specimens of PCL/CS, WS 2 /PCL/CS, and MTA scaffolds stained by the haematoxylin and eosin (HE) stain, Von Kossa (VK) stain, and Masson's trichrome (MT) stain are shown in Fig. 7d. The HE staining showed the tissue regeneration was well formed within the defect area in the WS 2 /PCL/CS scaffold, whereas tissue was minimal in the PCL/CS scaffold. The MT staining in the WS 2 /PCL/CS group at eight weeks shows that the collagen area and the lumen of the blood vessels supported active osteogenesis around the scaffolds. VK staining is known to be highly visible in formed bone tissue 61 . In particular, the images of the WS 2 /PCL/CS scaffold in Fig. 7d exhibit the most intense VK staining, which indicates much more extensive bone regeneration in the WS 2 /PCL/CS sample, demonstrating that the formed bone was generated around the bone defect.
In summary, this study demonstrates an eco-friendly and easy route to the scalable production of high-quality WS 2 thin sheets for bone regeneration and photoelectrochemical hydrogen generation applications. The extracted chlorophyll is a useful pigment from nature for preparing a high-concentration WS 2 thin sheet dispersion with an excellent degree of stability in a volatile solvent. The self-assembled exfoliated WS 2 thin sheets on an Au surface exhibited unusual HER photoelectrocatalytic activity with an onset potential of −14 mV and a Tafel slope of 49 mV dec −1 , the lowest onset potential achieved with such a low Tafel slope value that has been reported to date. According to the in vivo experiments, the bone regeneration of the WS 2 /PCL/CS scaffold was 120% superior to that of commercially available MTA material. The mechanical properties of the WS 2 /PCL/CS scaffold exhibited 300% enhancement over PCL/CS, which is one of the most commonly used materials. The WS 2 added to the PCL/CS scaffold enabled the scaffold to interact with cells, resulting in excellent osteoblast adhesion and apatite deposition. The authors thus believe that these versatile WS 2 thin sheets may provide an opportunity to help resolve issues related to population ageing and global climate change.

Preparation of chlorophyll extracts
Green leaves (20 g) were ground in a mortar and pestle and 500 mL acetone was poured into the mortar. After standing for 8 h, the chlorophyll extracts were filtered through a 0.22 μm pore size polyvinylidene fluoride membrane to remove impurities. The filtered solution was centrifuged at 3000 rpm for 30 min and the supernatants of the chlorophyll extract solution were collected. The concentration of the chlorophyll extracts was~5 mg L −1 .
Chlorophyll-assisted liquid-phase exfoliation of WS 2 WS 2 powder (5 g), 10 mL of extracted chlorophyll solution, and 240 mL acetone were added to a 500 mL glass serum bottle at a controlled temperature, and sonicated with 150 W horn sonication (Q700, QSONICA) in pulse mode for 5 h. To measure the concentration of the WS 2 suspension, 100 mL of solution was filtered through a 0.2 μm polyvinylidene fluoride membrane, and the weight of the WS 2 measured. The suspension concentration of the exfoliated WS 2 thin sheets was~1 mg mL −1 .

Synthesis of PtNPs
The platinum nanoparticles (PtNPs) were synthesised in glassware with a magnetic stir bar. 1 mL of 16 mM hexachloroplatinate (H 2 PtCl 6 ), 1 mL of 40 mM tri-sodium citrate, and 38 mL of deionised water were separately added to a 200 mL two-necked bottle and stirred for a half-hour to obtain a colourless mixture. Then, 200 μL of 50 mM sodium borohydride (NaBH 4 ) was added slowly into the two-necked bottle. The colourless mixture instantaneously turned to brownish-yellow. Finally, the brownish-yellow solution was continuously stirred at room temperature for 1 h. The solution was then stored in the dark until use. All glassware used in the PtNPs synthesis experiments was thoroughly cleaned with aqua regia, rinsed with deionised water (Elga Ltd., High Wycombe, UK), and then dried in an oven before use.

Electrochemical measurements
HER experiments were measured using a CHI7279E electrochemistry workstation. Electrochemical measurements were carried out with a threeelectrode configuration using 0.5 M H 2 SO 4 as an electrolyte, and an Ag/ AgCl electrode and a graphite electrode as the reference and counter electrodes, respectively. Linear sweep voltammetry (LSV) was performed at a rate of 1 mV s −1 for the polarisation curves. The I-t curve test was carried out with a controlled bias voltage.

Preparation of the calcium silicate (CS) powder
The method used for the preparation of CS powder has been described elsewhere 62 . In brief, CaO, SiO 2 , and Al 2 O 3 powders were used as matrix materials (composition: 70% CaO, 25% SiO 2 , and 5% Al 2 O 3 ). The oxide mixtures were sintered at 1400°C for 2 h using a high-temperature furnace, and then the CS powder was ball-milled in ethyl alcohol using a centrifugal ball mill (S100, Retsch, Hann, Germany) for 6 h.

Fabrication of the WS 2 /PCL/CS composite scaffolds
The CS matrices were produced using the thermal pressing method 63 . In a previous study 64 , we demonstrated that the ratio of PCL/CS = 50/50 had better mechanical properties and cell behaviours. First, 47.5 mg PCL was placed in a 150°C oven for 2 h. Then, 47.5 mg CS powder and 5 mg of the exfoliated WS 2 thin sheets were suspended in 5 mL alcohol and dropped into the PCL. The sample was labelled is 5% WS 2 /PCL/CS. Supplementary  Fig. 7 shows the cell viability of different ratios of the WS 2 scaffold. The 5% WS 2 /PCL/CS composite showed a suitable condition for cell viability and hence, it was used in the study. The WS 2 /PCL/CS composites were loaded into a steel syringe and heated to 100°C in a three-dimensional printer (3DP) system (BioScaffolder 3.1, GeSiM, Grosserkmannsdorf, Germany) to produce the printable WS 2 /PCL/CS composites, which were injected by applying a pressure of 500 kPa to produce a 400 µm diameter straight line. Seven lines were printed in parallel with a gap of 400 µm between the lines, forming one layer. The 3DP scaffold was plotted layer-by-layer, up to 16 layers, for bone engineering applications. The weight of the 16 layers scaffold was 0.4 g. Mechanical testing was done in accordance with ASTM D695-02a. The EZ Test machine (Shimadzu, Kyoto, Japan) was used to determine the mechanical properties of each scaffold at a loading rate of 1 mm min −1 . The scaffolds were printed into a 6 mm×6 mm × 10 mm rectangle. The average values and standard deviations (s.d.) of Young's modulus, and the maximum compressive strength were evaluated from the recorded stress-strain curves. Data are presented as mean ± s.d., n = 6 for each group.

Cell adhesion and viability
All 3DP scaffolds were soaked in 75% ethanol followed by irradiation for 1 h under UV light before the cell culture experiment. The human mesenchymal stem cells (hMSCs) were purchased from Sciencell Research Laboratories (Sciencell, Carlsbad, CA) and developed in medium (Sciencell). The cultured medium consisted of 500 mL of basal medium, 25 mL of foetal bovine serum, 5 mL of mesenchymal stem cell growth supplement and 5 mL of penicillin/streptomycin solution. The cells were seeded onto the specimens at a concentration of 5 × 10 4 cells per scaffold. After culturing for 1 and 3 h, the cultured medium was collected, and Col I secreted from hMSCs was analysed by an enzyme-linked immunosorbent assay kit (ELISA, Invitrogen), following the instructions in the manufacturer's manual. In addition, cell viability was determined using PrestoBlue Fig. 6 In vitro cell behaviour effects of WS 2 /PCL/CS scaffold. a Col I secretion; b Viability; and c ALP activity of hMSCs cultured on PCL/CS and WS 2 /PCL/CS scaffolds. '*' indicates a significant difference (p < 0.05) compared to Ctl. ' # ' indicates a significant difference (p < 0.05) compared to the PCL/CS scaffold; d F-actin filaments (green) and nuclei (blue) staining of hMSCs cultured on scaffold at different time-points. Scale bar is 200 µm. assay (Invitrogen) after being directly cultured for 1, 3, and 7 days. The PrestoBlue solution (30 μL) and 300 μL DMEM were added separately to each of the 96-well plates and incubated for 30 min, after which, 100 μL was extracted from each well and placed into a fresh 96-well plate and the absorbance was measured at 570 nm with a reference wavelength of 600 nm. Cells cultured without scaffold were used as the control.

Osteogenic differentiation
After cell seeding, the osteogenic differentiation was induced using a specific osteogenic medium from StemPro™ (osteogenesis differentiation kit, Invitrogen) and ALP secretion was measured on days 3 and 7. The ALP activity of each sample was characterised using p-nitrophenyl phosphate as the substrate with 1 mol L −1 diethanolamine buffer. The reaction was terminated by adding 5 N NaOH and the absorbance recorded at 405 nm. Analysis of blank disks was used as the control.

Fluorescent staining
The cultured cells were washed with PBS, fixed in 4% paraformaldehyde for 15 min, and then permeabilised with a 0.1% Triton X-100 PBS solution at 25°C. The F-actin filaments were labelled with phalloidin conjugated to Alexa Fluor 488 dye (Invitrogen) for 1 h. The cell nuclei were stained with 300 nM 4′,6-diamidino-2-phenylindole (DAPI) for 30 min. After washing thoroughly, the morphology was imaged using a white-light laser confocal microscope (Leica TCS SP8 X, Leica Microsystems, Heidelberg, Germany).

In vivo bone regeneration
All in vivo studies and protocols were done in accordance with relevant laws and regulations. The animal use protocol listed below has been reviewed and approved by the Institutional Animal Care and Use Committee of China Medical University, Taichung, Taiwan. The white New Zealand rabbits (2.0-2.5 kg) underwent surgery to create a standard critical-sized femoral defect. The PCL/CS, WS 2 /PCL/CS, and commercially available CS-based product (MTA) scaffolds (d = 7 mm and h = 8 mm) were implanted into the defects to assess the bone regeneration capacity. Afterwards, the periosteum and related tissues were sutured, and the rabbits were kept under close examination until they recovered from the effects of the anaesthesia. They initially limped but were able to walk with a proper gait after four weeks. They had free access to food and water during the period of observation. After the four weeks, the rabbits were euthanised through CO 2 asphyxiation. The scaffolds and bone specimens of the rabbits were then harvested and subsequently fixed using 10% formalin. Images of all in vivo specimens were taken with a high 360°s patial resolution micro-CT camera (µ-CT, SkyScan 1076, Skyscan Inc., Kontich, Belgium). The SkyScan software was used to recognise bone tissue and analyse the bone volume per tissue volume (BV/TV) and trabecular thickness (Tb.Th) in the 3D model. The specimens were then embedded in OCT (KMA-0100-00A, CellPath Ltd., Newtown, Wales, UK), without being decalcified, and sectioned (6 µm). The sections were stained with a haematoxylin and eosin stain kit, a modified Masson's Trichrome stain kit (ScyTek Lab., West Logan, UT, USA), and a Von Kossa kit (ScyTek) in accordance with the manufacturers' instructions.

Characterisation
The exfoliated 2D material thin sheets were deposited onto a silicon substrate for Raman and attenuated total reflectance Fourier-transform infrared spectroscopy (ATR-FTIR) studies. Raman spectra were recorded using an iHR550 spectrometer (Horiba Jobin Yvon) with a 532 nm laser source in ambient conditions, a laser power of <100 mW, a laser spot-size of 5 μm, and the signal of silicon at 520.7 cm −1 as a reference. ATR-FTIR data were recorded with a Perkin-Elmer Frontier infrared spectrometer equipped with a deuterated-triglycine sulfate detector. The UV-vis spectra were recorded on a Unicam UV-300 UV-vis spectrophotometer. The fluorescence (FL) spectra were measured using an FL spectrophotometer (Hitachi F-7000, Japan) at room temperature. A multi-well plate reader (TECAN Infinite Pro M200; Tecan, Männedorf, Switzerland) was used for molecular biology analysis. The compressive strength of the 3DP scaffolds was tested on an EZ-Test machine (Shimadzu, Kyoto, Japan) under a loading rate of 1 mm min −1 . The images of the exfoliated WS 2 thin sheets were obtained using a low-resolution transmission electron microscope (LRTEM; JEOL JEM-2100) and a high-resolution transmission electron microscope (HRTEM; Hitachi H-7100, JEM-ARM200F and JEM-ARM200FTH). The morphology of the composite paper was imaged by a field emission scanning electron microscope (FE-SEM; JEOL JSM-7600F).

Statistical analysis
A one-way analysis of the variant statistical data was used to evaluate the significance of the differences between the means in the measured data. A Scheffe's multiple comparison test was used to determine the significance of the deviations in the data for each specimen. In all cases, the results were considered to be statistically significant with a p-value < 0.05.

Reporting summary
Further information on experimental design is available in the Nature Research Reporting Summary linked to this paper.