Grain-boundary-rich 2D materials for attomolar-ability biochemical sensors

Next-generation high-performance biological and chemical sensors based on the emerging multitudinous two-dimensional (2D) layered materials have been attracting great attention in recent years. The performance of 2D biochemical sensors is strongly dependent on the structural defects, which provide indispensable active sites for sensitive and selective adsorption of analytes. However, achieving controllable defect engineering is still a big challenge. In the present work, we propose achieving superior biochemical sensor performance with high-surface-density grain boundaries (GBs), a kind of ubiquitous structural defects, in polycrystalline 2D thin lms, which can be controllably synthesized. As a proof-of-concept, by utilizing the high-density GBs in monolayer (1L) WS 2 lms, we fabricated a series of surface plasmon resonance (SPR) sensors for mercury ion (Hg 2+ ) detection. Our investigation has demonstrated substantial sensitivity enhancement of Hg 2+ detection down to trace attomolar-level quantication (detection limit of 1 aM), which is ascribed to the abundance of active sites on high-density GBs. This work provides a promising avenue for the design of ultra-sensitive sensors toward commercialized products based on the GB-rich 2D layered materials.

exposure to air, the GB pro les can be observable even in the OM images 60 . Figures 1a and b show the low and high magni cation AFM images of the as-grown polycrystalline 1L WS 2 lm on Si substrate after one-week exposure to air for su cient adsorption on GBs. The GB pro les can be clearly visualized between irregular-shaped WS 2 grains with the lateral dimension of less than 100 nm in average. The nano-scale grain sizes are the direct evidence of high-surface-density GBs in the CVD-grown polycrystalline 1L WS 2 lm. The atomic structure of GBs was investigated in detail by using scanning transmission electron microscopy (STEM). The low magni cation high angle annular dark eld STEM (HAADF-STEM) image of Fig. 1c clearly shows the GBs and the nano-scale grains with lateral sizes of less than 100 nm, con rming the AFM observations. In the corresponding selected area electron diffraction (SAED) of Fig. 1d, the observed diffraction rings instead of individual spots suggest that the nano-scale grains in polycrystalline 1L WS 2 lm were grown in nano-scale lateral sizes and without a preferential orientation. Statistical analysis of the grains indicates that the lateral size dominantly lies in the range from 20 to 55 nm, as shown by the inset of Fig. 1c, demonstrating the high grain density (10 10 -10 11 cm − 2 ) and high GB surface density. Atomic structure of the GBs on the as-synthesized WS 2 lm is distinguished using atomic-resolution HAADF-STEM imaging. After systematic examination of multiple boundary locations ( Supplementary Fig. S4), we con rmed that the WS 2 grains in the as-synthesized lm are indeed stitched together via GBs. It is established in previous research that the atomic make-up of GBs is diverse. In our sample, GBs are primarily consisted of 5-7 members of rings with a sulfur-rich chemical composition, as illustrated in Fig. 1e.
To demonstrate the application potential of GB-rich 2D layered materials as biochemical sensor materials, we fabricated the SPR sensors based on GB-rich 1L WS 2 lm for ultra-sensitive detection of Hg 2+ . The basic setup of our sensor device (fabrication details in Methods) and the SPR imaging system are schematically illustrated in Supplementary Fig. S3. The detection principle of SPR sensors lies in the ultra-sensitivity of SPR signals (i.e. resonance angle of θ in the present work) to minute changes in refractive index of the sensing surface. In this case, upon Hg 2+ adsorption onto GB-rich 1L WS 2 lm (sensor material), its refractive index changes, thus SPR curve shifts, providing quantitative information of Hg 2+ on the 1L WS 2 lm. By using Snell's Law and the N-layer transfer matrix method, we performed the computational simulation to pro le the vertical electric eld distribution of Au lm before and after the transferring of a 1L WS 2 lm ( Supplementary Fig. S5). Compared with bare Au, the electric eld is signi cantly enhanced upon the incorporation of GB-rich 1L WS 2 , and it is further increased to a maximum at the interface between 1L WS 2 lm and sensing medium. The 1L WS 2 integration induced electric eld enhancement between Au lm and sensing medium signi es that the proposed SPR sensor is sensitive to slight changes in sensing medium, thus strongly suggests the suitability of 1L WS 2 lm for SPR sensor. Generally, SPR signal optimization can be induced via incorporation of many other 2D layered materials onto Au lm, 65,66 which is favorable for detection sensitivity. However, this effect itself is insu cient for high sensor performance, and the fundamentally indispensable key factor is the high analyte adsorption ability of the sensor material.
To demonstrate the signi cant role of GBs in physi and/or chemi-adsorption of Hg 2+ , the CVD-grown 1L WS 2 single crystals (see Supplementary Fig. S6) based SPR sensor for Hg 2+ detection was also evaluated. Compared with the rich structural defects along GBs in polycrystalline 1L WS 2 lm, the 1L WS 2 single crystal has much fewer structural defects on the surface, but possesses edge defects. As revealed in the previous studies, 14 for the biochemical sensors based on 2D layered crystals, preferential adsorption occurs along material edges. Thus, the Hg 2+ sensing abilities of SPR sensors based on GBrich 1L WS 2 lm and 1L WS 2 single crystal were investigated in detail for comparison. Figure 2a shows the angle-resolved SPR spectra of GB-rich 1L WS 2 lm and 1L WS 2 single crystal in ultrapure water and at increasing concentrations of Hg 2+ aqueous solution (10 − 18 − 10 − 11 M). In contrast to those of the 1L WS 2 single crystal based SPR sensor, the SPR spectra based on GB-rich 1L WS 2 lm display much more prominent right-shift with increasing Hg 2+ concentration, indicating adsorption of more Hg 2+ ions. Figure 2b shows the determined resonance angle shift (∆θ) as a function of Hg 2+ concentration for the SPR sensors based on GB-rich 1L WS 2 lm, in which the extracted one of 1L WS 2 crystal is also included for comparison. Distinctly, at the same Hg 2+ concentration, the GB-rich 1L WS 2 sensor exhibits a much larger angle shift (∆θ) than that of the 1L WS 2 crystal one. The degree of angle shift is proportional to the amount of adsorbed Hg 2+ . Even at attomolar-level concentration (13 milli-degree for 10 − 18 M), the adsorption of Hg 2+ by the GB-rich 1L WS 2 lm brings about discernible change in SPR resonance angle.
Notably, the GB-rich 1L WS 2 sensor displays the wide detectable dynamic range from 10 − 11 to 10 − 18 M, covering 7 orders of magnitude. Compared with that of the 1L WS 2 single crystal sensor, the observed much larger ∆θ of GB-rich 1L WS 2 sensor indicates the substantially larger amount of Hg 2+ ions adsorbed onto GB-rich 1L WS 2 . Thus, this drastic sensor performance difference not only demonstrates the outstanding sensitivity of GB-rich WS 2 sensor device, but, more importantly, reveals the signi cant role of high-surface-density GBs in sensor performance.
Adsorption or binding sites are energetically active positions, which are widely accepted to exist in structural defects of 2D layered materials, such as vacancy, antisite, substitution, edge, and GB. 14,26,27,48-58,67−71 Structurally, although 1L WS 2 single crystals contain a certain amount of defects on surface and along edges, the surprisingly high surface-density of GBs in polycrystalline 1L WS 2 lm immensely increase the amount of defects per unit area, providing the primary analyte binding sites. Thus, for the SPR sensor based on polycrystalline 1L WS 2 lm, the observed superior sensitivity of Hg 2+ detection is considered to result from the rich GBs. Moreover, the adsorption energies of Hg around the GBs are also lower than that on pristine 1L WS 2 crystal (see Supplementary Fig. S7). This indicates the GBs behave like the Hg traps. From the charge density difference plot shown in Fig. 3b, we can nd the apparent charge transfer between Hg and GBs. There are 0.1 negative charge transfer from Hg to neighboring S atoms, indicating the formation of covalent bonds between Hg and S atoms. The formation of Hg-S bond is substantiated by X-ray photoelectron spectroscopy (XPS) measurements of the as-synthesized GB-rich 1L WS 2 lm before and after Hg 2+ detection. Figures 3c-e show the XPS spectra of W4f, S2p and Hg4f core levels before and after Hg 2+ detection. For the as-synthesized GB-rich 1L WS 2 lm, the two deconvoluted W4f peaks at 32.63 and 34.79 eV (Fig. 3c top) and the two deconvoluted S2p peaks at 162.55 and 163.71 eV (Fig. 3d top) are ascribed to WS 2 , 72 while no trace of Hg is observed (Fig. 3e top). For the GB-rich 1L WS 2 lm after Hg 2+ detection in 10 − 9 M Hg 2+ solution and then rinse with ultra-pure water for three times, obvious sign of S-Hg bonds is observable in the XPS spectrum of Hg4f core level (Fig. 3e bottom), and the main deconvoluted peaks of W4f and S2p are slightly shifted toward higher binding energies in contrast to those of the as-synthesized GB-rich 1L WS 2 lm. The two deconvoluted W 4f peaks at 33.05 and 35.22 eV can be still related to WS 2 . On the XPS spectra of S2p core level, however, in addition to the two strong deconvoluted peaks at 162.55 and 163.71 eV from WS 2 , two new weak deconvoluted peaks at 161.67 and 163.11 eV are recognized to be produced by the formation of S-Hg bonds during Hg 2+ detection. The formation of S-Hg bonds after Hg 2+ detection, which can induce changes in chemical environment for W and S atoms in the WS 2 lm, can also be the underlying origin for the observed slight shift of the XPS W4f and S2p peaks from WS 2 toward higher binding energies. The combined theoretical calculations and XPS measurements provide concrete evidence for the preferential adsorption of Hg 2+ on the GBs via the formation of S-Hg bonds. Therefore, the rich chemically active sites on GBs in polycrystalline 1L WS 2 lm serve as e cient probes for ultra-sensitive detection of Hg 2+ ions.
As one of the crucial performance criteria, the selectivity of Hg 2+ detection for the SPR sensor based on GB-rich polycrystalline 1L WS 2 lm was also assessed. The evaluation was performed by comparing the sensor responses to 10 − 12 M of Hg 2+ and common interfering ions such as Pb 2+ , Mn 2+ , Cu 2+ , Fe 3+ , Zn 2+ , Co 2+ , Cr 3+ and Mg 2+ . The determined resonance angle shifts ∆θ in Fig. 4a show that the value of ∆θ induced by Hg 2+ adsorption is at least ~ 3X higher than those produced by other ions, indicating that the GB-rich WS 2 sensor is highly selective towards Hg 2+ detection.
To further validate the superior sensing capacity of the GB-rich WS 2 lm, we compared its LOD for Hg 2+ with those of previously reported high-performance Hg 2+ sensing materials. 17,73−135 As presented in Fig. 4b. the LOD of GB-rich WS 2 lm for Hg 2+ can reach down to 1aM (~ 600 ions·cm − 3 ), according to the IUPAC guideline of 3:1 signal to noise ratio, which clearly outperforms the previous ion sensors based on conventional sensor materials.
The superior performance of GB-rich 1L WS 2 based SPR sensor in Hg 2+ detection relies on the presence of su cient active sites along the rich GBs for the preferential and e cient adsorption of Hg 2+ . Therefore, not only the polycrystalline 1L WS 2 lm, the GB-rich polycrystalline 1L and few-layer lms of any other layered metal sulphides would also be expected to serve as sensor materials for the similar ultra-sensitive detection of Hg 2+ ions. As a demonstration, the CVD-grown GB-rich polycrystalline 1L MoS 2 lm 62 was used to fabricate the SPR sensor for detection of Hg 2+ ions. As shown in the SPR spectra and the determined resonance angle shifts of Δθ ( Supplementary Fig. S8), the superior sensor sensitivity down to trace attomolar-level quanti cation (LOD of 1 aM) is also observed for the GB-rich 1L MoS 2 lm, comparable to those of the GB-rich 1L WS 2 lm, which further demonstrates that the concept of GB-rich 2D layered materials for high-performance sensors is universal and comprehensive.

Conclusion
In summary, we report CVD-grown monolayer 2D lms with high-surface-density GBs for biological and chemical sensor applications of attomolar sensing performance. Speci cally, as a proof-of-concept, SPR sensor devices based on rich GBs in polycrystalline 1L WS 2 lms present superior sensitivity of down to an ultra-low Hg 2+ LOD of 1 aM. The inherent GB-rich structure is the key factor behind the observed extraordinary sensor performances, because the abundance of active sites in GBs is essential to analyte adsorption. More importantly, DFT calculations con rm that the structural defects on GBs are primary analyte binding sites. Consequently, the discovery of GB-induced exemplary sensor performance motivates more exciting developments based on GB-rich 2D materials and constitutes an extraordinary opportunity for the advancement of ultra-sensitive sensors.

Methods
Device fabrication. SPR sensor devices are constructed of the as-synthesized TMDs lms transferred onto a 47 nm Au lm-coated cover glass substrate (MATSUNAMI GLASS, Japan; Supplementary Fig.  S1a). 2 nm Cr lm was deposited prior to Au deposition for enhanced adhesion of Au lm to cover glass substrate. The transfer of the as-grown TMDs lms were performed via a standard PMMA-based wet transfer method. PMMA was spin-coated onto the as-synthesized lm, baked at 150 °C for 180 sec, and immersed in100 °C H 2 O for 10 min to detach the PMMA-coated lm from the substrate. The Au-coated substrates were used to pick up the PMMA-coated lm. Finally, by dissolving the PMMA coating in acetone, followed by 3X ethanol rinse and 3X DI water rinse, the as-synthesized lms were successfully transferred.
Experimental Setup. The adopted SPR imaging system was based on an inverted microscope (Nikon TI2-U, Japan) and a 100× (NA 1.49) oil immersion objective. A 680 nm 10 mW He-Ne laser was used as the light source to excite surface plasmon on the gold surface.The CMOS camera (Zyla 4.2 PLUS, Andor, Belfast, UK) was used to record the SPR image at different incident angles. A stepping motor was incorporated on the optical ber to translate the incident angle of light beam. The incident angle was synchronized with the corresponding SPR image. The intensity measurement was performed using Image J software.
Sensor characterization. A silicon insert ( exiPERM®8-well reusable silicon insert, Sarstedt, Germany) is placed on the sensor device to hold analyte solutions. Angle-resolved SPR spectra was acquired in ultrapure water and then at increasing concentrations of Hg 2+ aqueous solution (10 − 18 − 10 − 11 M). At each measurement, the previous analyte solution is rst removed from chamber by pump (Kylin-Bell Lab Instruments), followed by injection of current analyte solution.
Supporting information. Detailed OM, FL, Raman, PL, AFM and TEM characterization of as-synthesized materials and detailed angle-resolved SPR spectra of measured samples are provided.

Data availability
The authors declare that the data supporting the ndings of this study are available within the paper and its Supplementary Information les.  DFT and XPS characterization. DFT calculations regarding different adsorption positions of Hg atom on grain boundary labeled out with number 1-6 (a) and charge density difference between Hg atom and WS2 with grain boundaries (b). The calculated adsorption energies of Eads at the 6 different positions are listed between (a) and (b). Yellow, blue and red balls represent S, W and Hg atoms, respectively. There is apparent charge transfer between Hg and nearby S atoms. The isosurface value is 0.0002 e/Bohr3. XPS spectra of W 4f (c), S 2p (d) and Hg 4f (e) core levels for the as-synthesized 1L WS2 lm (top) and the 1L WS2 lm after Hg2+ detection (bottom). Notably, the XPS peak at 104.54 eV (e) is ascribed to Si from wafer.