Degradation behaviors and mechanisms of MoS2 crystals relevant to bioabsorbable electronics

Monolayer molybdenum disulfide (MoS2) exhibits unique semiconducting and bioresorption properties, giving this material enormous potential for electronic/biomedical applications, such as bioabsorbable electronics. In this regard, understanding the degradation performance of monolayer MoS2 in biofluids allows modulation of the properties and lifetime of related bioabsorbable devices and systems. Herein, the degradation behaviors and mechanisms of monolayer MoS2 crystals with different misorientation angles are explored. High-angle grain boundaries (HAGBs) biodegrade faster than low-angle grain boundaries (LAGBs), exhibiting degraded edges with wedge and zigzag shapes, respectively. Triangular pits that formed in the degraded grains have orientations opposite to those of the parent crystals, and these pits grow into larger pits laterally. These behaviors indicate that the degradation is induced and propagated based on intrinsic defects, such as grain boundaries and point defects, because of their high chemical reactivity due to lattice breakage and the formation of dangling bonds. High densities of dislocations and point defects lead to high chemical reactivity and faster degradation. The structural cause of MoS2 degradation is studied, and a feasible approach to study changes in the properties and lifetime of MoS2 by controlling the defect type and density is presented. The results can thus be used to promote the widespread use of two-dimensional materials in bioabsorption applications. The mechanism by which two-dimensional electronic materials decompose in an environment similar to that inside the human body has been identified by researchers in South Korea. Biodegradable, or transient, electronic devices disappear when no longer needed. In biomedical applications, for example, a transient sensor in the body degrades or dissolves, eliminating the need for surgery to remove it. Jong-Hyun Ahn from Yonsei University in Seoul and co-workers investigated the degradation of crystals of the two-dimensional semiconductor molybdenum disulfide (MoS2), each having the triangular shape. They showed that the rate of decomposition is dependent on the angle of misalignment between the two crystals: crystals with a larger misalignment biodegrade faster than those more closely aligned. This behavior indicates that intrinsic defects in the atomic structure of the material are the cause of the degradation. We present the degradation behaviors and mechanisms of CVD-grown monolayer MoS2 crystals relevant to bioabsorbable electronics, triggered and extended based on the intrinsic defects such as grain boundaries and point defects for their high chemical reactivity caused by broken lattice and dangling bonds. Higher misorientation angle leads to higher degradation speed. This work paves the way for lifetime modulation and bioabsorbable device application by using 2D materials.

Monolayer MoS 2 can be slowly oxidized in air or dissolved in aqueous solution after several weeks or months, resulting in its environmental degradation [12][13][14][15] . Owing to these characteristics, monolayer MoS 2 is an ideal candidate for use in bioabsorbable electronics, such as temporary biomedical sensors and therapeutic vehicles, that can be absorbed after being implanted into the human body, thereby eliminating the need for secondary surgeries to extract the implants [16][17][18][19] . The long-term cytotoxicity and immunological biocompatibility of CVD-grown monolayer MoS 2 in biofluids and tissues of live animal models have been reported 20

. Monolayer MoS 2 -based bioabsorbable and multifunctional sensors
School of Materials Science and Engineering, Ulsan National Institute of Science and Technology (UNIST), 50 UNIST-gil, Ulju-gun, Ulsan 44919, Republic of Korea Full list of author information is available at the end of the article. These author contributed equally: Xiang Chen, Sachin M. Shinde for the temporary intracranial monitoring of pressure, temperature, strain, and motion in animal models have also been demonstrated 20 . Such technologies play specific, clinically relevant roles in short-term diagnostic/therapeutic functions during recovery from traumatic brain injuries, but the details and structural cause of monolayer MoS 2 degradation are not so clear.
A fundamental understanding of the degradation behaviors and mechanisms of monolayer MoS 2 in biofluids is therefore critical to maximizing the potential benefit of such materials in a bioabsorbable electronic platform. Thermal oxidation and anisotropic etching of two-dimensional (2D) MoS 2 in air have been previously reported; however, the degradation environment differs substantially from that in biofluids not only with respect to the reaction temperature and speed but also with respect to the reactants and products [12][13][14][15][21][22][23][24][25] . Other studies have reported the dissolution of chemically/ mechanically exfoliated MoS 2 nanosheets in aqueous solution; however, nanosheets dispersed in solution have little relevance to electronics 26,27 . Fundamental studies, including the interaction between monolayer MoS 2 crystals and biofluids, the morphological evolution, and the role of grain boundaries (GBs) and point defects during degradation, have yet to be adequately conducted 6,[26][27][28][29] .
This study investigates the degradation behaviors and mechanisms of monolayer MoS 2 crystals with different crystallographic misorientation angles in phosphatebuffered saline (PBS) solution. Monolayer crystals are chosen because their basic structure enables rapid degradation and trend analysis. The results show that high-angle grain boundaries (HAGBs, >15°) preferentially degraded faster than low-angle grain boundaries (LAGBs, <15°); this phenomenon, which has not been previously reported, is referred to as "angle-dependent selective degradation." Such selective degradation results from the different dislocation types and defect densities of LAGBs and HAGBs. The degraded LAGBs and HAGBs exhibit wedge-and zigzagshaped edges, respectively. MoS 2 grains degrade and gradually disappear from the substrate. Triangular pits generated in the degraded MoS 2 grains are oriented opposite to those of parent crystals and grow into larger pits, indicating that the degradation of monolayer MoS 2 is induced by the intrinsic defects and can be extended laterally. High densities of dislocations and point defects (e.g., 5|7-fold or 4|4fold ring cores, monosulfur vacancies (V S ), or disulfur vacancies (V S2 )) lead to high chemical reactivity and fast degradation. Herein, a comprehensive study of the degradation of monolayer MoS 2 is presented; the properties and lifetime of 2D MoS 2 in biofluids are controlled by modulation of the defect type and density; and 2D TMDCs are promoted as outstanding candidates for biomedical applications, such as implantable, biocompatible, and bioabsorbable devices and systems in the body.

Synthesis of monolayer MoS 2 crystals
The atmospheric pressure CVD method was used to synthesize monolayer MoS 2 crystals. First, an SiO 2 (300 nm)/Si substrate was cleaned in acetone, isopropyl alcohol, and deionized (DI) water for 10 min each. Second, the SiO 2 /Si substrate was further treated in Piranha solution at 70°C for 2 h to hydrophilize the surface, and a 5 μM aqueous solution of perylene-3,4,9,10-tetracarboxylic acid tetrapotassium salt (PTAS) was spincoated as a seeding promoter. Third, 10 mg of MoO 3 powder (99.5%, Sigma-Aldrich) was placed in a ceramic boat and inserted into a single-zone furnace together with the upside-down SiO 2 /Si substrate on the boat. In addition, sulfur powder (99.5%, Sigma-Aldrich) was placed on another boat and loaded upstream in the furnace. The furnace was heated to 750°C within 15 min, maintained at 750°C for 30 min, and then allowed to naturally cool to room temperature. The synthesis process was protected from air and performed in an atmosphere composed of a mixture of Ar (50 sccm) and H 2 (7.5 sccm) gas.

Degradation of monolayer MoS 2 crystals
Degradation tests of monolayer MoS 2 crystals were performed in PBS solution. Notably, both Si and SiO 2 can dissolve in PBS solution after a period, resulting in MoS 2 crystals detaching from the substrate. To thoroughly study the morphological change of the MoS 2 crystals during the degradation process, the MoS 2 crystals were transferred from SiO 2 /Si to an SU-8/sapphire substrate because SU-8/sapphire is more stable in PBS solution.
First, a cleaned sapphire substrate was spin-coated with SU-8 (10% at 3500 rpm; MicroChem), followed by soft baking for 5 min at 65°C, 2 min of UV treatment, and 30 min of hard baking at 150°C to attain a thin, smooth SU-8 layer (thickness:~100 nm; root-mean-square (RMS) roughness:~0.258 nm). Second, the MoS 2 crystals were spin-coated with poly(methyl methacrylate) (PMMA) and transferred onto the SU-8/sapphire substrate using the wet transfer method. After the film had completely dried, the PMMA layer was carefully removed with hot and room temperature acetone continuously. As a result, MoS 2 crystals were transferred to the target substrate. To study the degradation process and behavior of monolayer MoS 2 crystals in PBS solution, the degradation test was performed in 60 mL of 1.0 M PBS (Sigma-Aldrich) on a hot plate, and the PBS solution was changed every 2 days to maintain its concentration. Finally, an accelerated degradation test was performed in PBS solution (pH 7.4) at 75°C for 1-7 weeks; the degraded MoS 2 samples were then removed from the PBS solution, rinsed with DI water, and gently blown dry using N 2 gas. As a reference, normal degradation was also investigated in PBS solution (pH 7.4) at 40°C over 14-28 weeks.

Characterization of monolayer MoS 2 crystals
To investigate the morphological evolution of MoS 2 crystals before and after degradation in PBS solution, optical microscopy (OM) images were captured using a Nikon Eclipse LV100ND microscope. Atomic force microscopy (AFM) images were recorded on a Park system NX10 in noncontact mode. Moreover, Raman and photoluminescence (PL) measurements were performed with a UniNanoTech Uni-RAM2 at a laser excitation wavelength of 532 nm and a laser power of 0.4 mW. To reveal the distribution of GBs between adjacent MoS 2 crystals, second-harmonic generation (SHG) nonlinear optical images were captured using a dual-mode Er-doped fiber laser (Insight Deepsee Dual; Spectra-Physics) and a confocal microscope (IX83; Olympus) after the two spatially overlapped beams were directed to a galvanometric x-y directional mirror controlled by a scanning system (Fluoview1000; Olympus). Furthermore, dark-field transmission electron microscopy (DF-TEM) images were acquired using an image aberration-corrected Titan Cube G2 operated at 80 kV. To avoid damage accumulation, the acquisition time was reduced as magnification increased. HAADF-scanning transmission electron microscopy (STEM) images were obtained with a probe aberration corrector at 80 kV. The accelerated voltage was below the knock-on threshold energy to damage MoS 2 .

Results and discussion
The degradation test in this study included four steps, as shown in Fig. 1a. First, monolayer MoS 2 crystals were synthesized on an SiO 2 /Si substrate via the CVD method. Second, the crystals were transferred to an SU-8/sapphire substrate. Third, the MoS 2 samples were placed in PBS solution for approximately 7 weeks. Finally, the MoS 2 samples were removed from the PBS solution, carefully cleaned, and characterized. Further details are provided in the Materials and methods section. Note that the SU-8 layer exhibited a smooth surface and high stability in the PBS solution, which contributes to the morphological study of MoS 2 crystals before and after degradation. In addition, the wet transfer process negligibly affected the quality of the MoS 2 crystals. Figure 1b shows the OM images of MoS 2 crystals on SiO 2 /Si and SU-8/sapphire substrates. Except for the change in contrast, the crystal surface was smooth and did not show any wrinkles or microcracks. The RMS roughness of the MoS 2 crystals increased (during the transfer process) from 0.20 to 0.62 nm because of the residual PMMA (Supplementary Figure S1). Raman and PL spectra further indicated that the monolayer nature of MoS 2 crystals was maintained after the transfer process because the difference in frequency between the E 2g and A 1g modes was maintained at 20 cm −1 (Fig. 1c). In addition, the crystals were free from doping or strain effects, as indicated by the unchanged PL peak at 1.81 eV (Fig. 1d) 8 . In other words, high-quality CVD-grown monolayer MoS 2 crystals with clean and smooth surfaces were successfully transferred to the SU-8/sapphire substrate.
For CVD-grown monolayer MoS 2 crystals, GBs are inevitably generated between two adjacent grains with crystallographic misorientation [30][31][32][33][34] . In general, LAGBs have a misorientation angle <~15°, whereas HAGBs have an angle >~15°3 5 . Because of the hexagonal atomic structure of semiconducting MoS 2 in the 2H phase, the misorientation angle ranges from 0°to 60°3 3,36 . The atomic defect structures in MoS 2 grains and GBs greatly affect the mechanical, electrical, and optoelectronic properties of MoS 2 -based devices 35,[37][38][39][40][41][42][43] . To thoroughly explore the degradation behaviors of MoS 2 grains and GBs with different inter-grain misorientation angles, we selected four samples with different angles and characterized their original morphologies and structures. Figure 1e, f show the OM and AFM images of the four samples. These crystals are well attached to the SU-8 layer with a clean surface; however, the GBs cannot be observed in these OM and AFM images due to technical limitations. To visualize the GBs, destructive methods are commonly used, including oxidation in air, ultraviolet light irradiation along with moisture exposure, spincoating of liquid crystals, and stacking of bilayers 9,36,44,45 .
Here, SHG microscopy is used as a nondestructive method to determine the crystalline orientations and the distribution of GBs 46 . Because SHG is highly sensitive to the structural symmetry, MoS 2 crystals with different orientations show different SHG responses, enabling visual distinction of GBs with two anisotropic grains ( Fig. 1g) 33 . Notably, the shape of GBs depends on the relative growth rates of the two adjacent edges, which are not easily distinguished in SHG images when the misorientation angle approaches 0°or 60°because of the 60°p eriodicity caused by the six-fold rotational symmetry of the MoS 2 lattice [33][34][35] . As demonstrated later, such a simple method of confirming the GBs enables the analysis of degradation processes.
According to previous studies, the dissolution rate of monolayer MoS 2 or Si nanomembranes increases with increasing PBS temperature 20,47 . In view of the dissolution kinetics of monolayer MoS 2 (~200 nm grain size) in PBS solution, a high concentration, pH, or temperature results in a high dissolution speed 20 . Therefore, an accelerated degradation test of CVD-grown monolayer MoS 2 crystals with different misorientation angles is performed in PBS solution (pH 7.4) at 75°C instead of 37°C, thereby eliminating the need for longer degradation time in the laboratory 20 . The corresponding OM and AFM images of MoS 2 crystals placed in the PBS solution for 1-7 weeks are shown in Fig. 2a, b. After 1 week, the MoS 2 GBs with misorientation angles of 26°, 40°, or 45°were degraded,  and the adjacent grains were separated accordingly. However, surprisingly, the MoS 2 GBs with angles of 5°or 10°were well preserved, and two contiguous grains were still connected (Fig. 2a). Such different degradation results between LAGBs and HAGBs were further confirmed in the magnified AFM images of the 5°and 45°samples, where differences were easily observed in arbitrary regions (bottom of Fig. 2a). These observations indicate that the degradation process preferentially initiates at the HAGBs rather than at the LAGBs. This unique phenomenon, which we refer to as "angle-dependent selective degradation", has never been observed or previously reported in oxidation studies of MoS 2 13,14,24 . To demonstrate this concept, a high-temperature oxidation test of monolayer MoS 2 crystals was performed on an SiO 2 substrate by heating the crystals on a hot plate at 380°C in air for various time intervals (from 90 to 210 min) (Supplementary Figure S2). After 90 min, the MoS 2 GBs with angles of 6°, 23°, or 56°were etched with similar widths, indicating that the angle-dependent selective degradation of MoS 2 GBs did not occur at an elevated temperature. By contrast, the aqueous system can provide ions with water and oxygen molecules and transport them away as products, which is impossible in the dry state. In addition, hightemperature oxidation may be limited by diffusion, rather than the elementary reactions themselves. The possible reactions that could have occurred during the degradation process are as follows: where the molybdate ion (MoO 4 2− ) is the main Mocontaining byproduct, which has been confirmed by inductively coupled plasma-mass spectrometry (ICP-MS) measurements 20,26,27 . Note that Na + and K + ions in the PBS solution may lead to lattice distortions of MoS 2 and the formation of Na 2 S (from 2H-MoS 2 to 1T-NaMoS 2 and then to soluble Na 2 S), further accelerating the degradation process 20 . According to the OM images presented in Fig. 2b, both the LAGBs and the HAGBs degraded after 3 weeks; the degradation of the whole MoS 2 grain also began, as evidenced by the numerous small pits observed inside the grains. Five weeks later, the degraded GB regions became wider, and the pits were enlarged. After 7 weeks, the MoS 2 GB regions were not easily observed. The pits merged into each other to form larger ones, and some of them even exhibited triangular-shaped pits whose orientation was opposite to that of the parent grains. These trends indicate that the MoS 2 crystals would be completely degraded in the PBS solution in a sufficient amount of time (e.g., more than 10 weeks). This process represents the entire degradation process of CVD-grown monolayer 2H-MoS 2 crystals in PBS solution.
To elucidate the degradation speed of monolayer MoS 2 crystals, the width of the degraded GB regions was measured using AFM. Figure 2c represents the height profiles across the MoS 2 GBs with angles of 2°, 16°, 48°, or 55°b efore degradation (Fig. 1f, Supplementary Figure S3). At this stage, two grains were connected without any space between them. The RMS roughness of each GB was found to be in the normal range (0.41-0.68 nm). The MoS 2 GBs with angles of 5°or 10°were still intact after 1 week; however, the MoS 2 GBs with angles of 26°, 40°, or 45°w ere selectively degraded, with gap widths of 113, 312, or 391 nm, respectively (Fig. 2d). These results indicate that the higher the HAGB angle, the higher the degradation speed. They also imply that a continuous MoS 2 monolayer with a small grain size exhibits the highest degradation rate because most of them contain HAGBs 20,44 . After 3 weeks, the width of degraded GBs increased from 527 (1°) to 996 nm (60°); 5 weeks later, they had further enlarged to between 956 (1°) and 1422 nm (60°) (Fig. 2e). The angle-dependent selective degradation occurred in the early stage (in 3 weeks or less). After both the LAGBs and HAGBs were degraded, the degradation speed of the GB regions in the lateral direction was approximately the same (~213 nm per week), as shown in Fig. 2e. Because such selective degradation started at the very beginning, HAGBs could maintain a higher speed than LAGBs, which enables control of the lifetime in biofluids via the selection of MoS 2 crystals with different misorientation angles.
Interestingly, the edges of the MoS 2 crystals were not degraded from the outside in, and the size of each crystal did not obviously decrease after 3-5 weeks (Fig. 2b). The Mo-rich synthesis condition applied in this study contributes to the formation of monolayer MoS 2 crystals with sharp, straight, and Mo zigzag edges 30 . Calculations show that the surface energy σ of a Mo zigzag edge remains smaller than that of an S zigzag edge provided that the chemical potential of sulfur (μ S ) is <−1.65 eV. The lower surface energy of a Mo zigzag edge makes it much more stable than an S edge; thus, Mo zigzag edges will be more abundant than S zigzag edges under any realistic conditions 48 . This greater abundance of Mo zigzag edges leads to their slow degradation in PBS solution compared with the degradation rates of grains or GBs. Because the center of the crystal may contain a high defect density, particularly in the form of S vacancies, degradation of the crystal occurs predominantly at these sites. This degradation process further accelerates with increasing time, and finally the crystal disappears, leaving behind some part of the crystal (Mo zigzag) edge intact at the substrate even after 5 weeks (Fig. 2b). Notably, the phenomenon of angle-dependent selective degradation is statistically significant in our research and could also be observed when the temperature of the PBS solution was reduced from 75°C to 40°C (very close to physiological conditions, 37°C). The main difference is that the degradation process becomes much slower (approximately one-tenth of that at 75°C) (Supplementary Figure S4). In addition, due to the same synthesis conditions, isolated monolayer MoS 2 crystals with a triangular shape degrade with the same tendency and speed as the polycrystalline domains (Supplementary Figure S5).
To discover the structural cause of the angle-dependent selective degradation and reveal the relation between atomic structures and degradation behaviors, TEM and STEM characterizations are performed for CVD-grown monolayer MoS 2 crystals with different misorientation angles. The high-angle annular dark-field STEM (HAADF-STEM) images of the MoS 2 crystals with angles of 5°, 16°, 27°, and 60°suspended on quantifoil TEM grids are presented in Fig. 3a. The equilateral geometry of the triangular crystals, together with sharp and straight edges, implies that the MoS 2 crystals are single crystalline (the semiconducting 2H phase) and may well have Moterminated edges 11,30 . Fig. 3b shows the dark-field TEM (DF-TEM) images and selected-area electron-diffraction (SAED) patterns taken from the regions marked with circles in Fig. 3a. GBs are formed between two adjacent grains (Grains 1 and 2), and their distributions are consistent with the SHG results in Fig. 1g. The corresponding SAED patterns inset in Fig. 3b have two sets of hexagonally arranged diffraction spots, which rotate gradually from one to the other because of the increment in the misorientation angle 31 . To examine the atomic structure along the GBs, atomically resolved HAADF-STEM was performed. Figure 3c presents the HAADF-STEM images corresponding to the areas (marked with a circle) of the four GBs in Fig. 3b.
Based on these high-resolution images, three pieces of information can be obtained. First, the MoS 2 crystals with a hexagonal lattice and no Moiré pattern have a monolayer structure. The brighter spots are the Mo atoms, and the darker spots are the two stacked S atoms because the contrast of HAADF-STEM images scales approximately as the square of the atomic number 31 . Second, the higher the misorientation angle, the greater the number of localized defects generated along the GBs, including Mo dangling bonds and tiny holes. These dangling bonds and holes are mainly caused by the incomplete growth of GBs or by slight aging because of the storage process 13,33 . Third, a similar number of point defects (V S and V S2 ) whose edges are along zigzag directions (Mo-oriented) are observed in each crystal. They are inevitable in MoS 2 crystals, as suggested by the second law of thermodynamics, and are more noticeable in CVD-grown samples because of imperfections introduced during the growth process 37 . Vacancies V S and V S2 have been reported to have the lowest formation energies (ΔE Form ) of 2.12 and 4.14 eV, respectively, among all of the defects; here, where N S_ML and N Mo_ML are the single-atom energy of Mo and S, respectively, in a perfect monolayer. The results show that S-deficient CVD-grown monolayer MoS 2 preferentially forms S vacancies, indicating that V S is the most common defect, followed by V S2 39 . The existence of such vacancies can substantially reduce the dissociation kinetic barrier by half, from~1.56 tõ 0.8 eV 14 , and make the defect regions in biofluids easier to degrade.
According to the magnified HAADF-STEM images from the continuous GB regions in Fig. 3c, more information about the evolution of edge structures and dislocations can be collected (Fig. 3d). In the LAGB case (5°), two adjacent grains with armchair edges meet and combine well with no dislocation because their complementary structures lead to the generation of new hexagonal lattices at the center and connect as one crystal in local areas. In the HAGB cases (16°and 27°), however, two grains, one an armchair edge and the other with a zigzag edge, cannot perfectly and stably combine. Thus, Mo-oriented dislocations, such as 5|7-fold ring cores, are formed to facilitate the transition from one edge to the other 30,32 . When the angle increases from 16°to 27°, the lattice mismatch becomes larger, and more 5|7-fold ring cores are generated (from 2 to 3). When the angle approaches 60°, two grains with zigzag edges come face to face, and the 5|7-fold ring cores change into 4|4-fold ring cores, although the number of cores does not change noticeably. Based on the above results, the angledependent evolvement rules about edge structures and dislocations along the MoS 2 GBs can be determined, as shown in Fig. 3e, f. In general, from LAGBs to HAGBs, such increase in dislocation number and changes in dislocation type mainly result from the various levels of lattice mismatch between two neighboring grains, which are fundamentally decided by the related edge structures (armchair or zigzag). For a perfect monolayer MoS 2 , Mo atoms are sandwiched and protected by top and bottom S atoms with no broken lattice, preventing the interaction or reaction between 2D MoS 2 and the biofluid to a certain extent. This lack of interaction causes some parts of the MoS 2 crystals to remain intact even after 7 weeks in PBS solution (Fig. 2b). Because of these GBs and point defects, CVD-grown monolayer MoS 2 crystals have different quantities of dangling bonds, which result from the growth conditions and misorientation angle. The related GB energy is proportional to the tilt angle because large tilt angles result in higher dislocation densities 32 . Such dangling bonds reduce the stability and enhance the chemical activity of these areas: more dangling bonds result in greater reactivity 38,41,42,49 . Therefore, HAGBs with more dangling bonds, holes, and ring cores exhibit faster degradation in PBS solution compared with LAGBs with fewer dangling bonds, holes, and ring cores, leading to the phenomenon of angle-dependent selective degradation. In the same manner, point defects exhibit much slower degradation than GBs because they lack sufficient dangling bonds. Above all, the degradation process of monolayer MoS 2 crystals in PBS solution is induced by intrinsic defects in the grains or along the GBs. This degradation process is suppressed in the unbroken regions, whereas it readily occurs in the regions with high defect densities.
As a starting point of the degradation process, the intrinsic defects in monolayer MoS 2 crystals play important roles in determining the time and range of degradation. However, the sequence of steps in this process needs to be discussed to determine the corresponding behaviors of these defects. Figure 4a presents the OM and AFM images of degraded MoS 2 crystals with an LAGB (1°) and a HAGB (60°) after the crystals were exposed to PBS solution for 3 weeks. Notably, the shapes of the degraded LAGB and HAGB crystals differ. The former consists of a wedge shape, whereas the latter consists of a zigzag shape. In addition, many triangular pits are generated inside the crystals; some are isolated, whereas others are connected. More interestingly, the orientation of each pit is consistent with the others but opposite to the orientation of the parent crystals. Before exploring these unusual results, we first note that LAGB and HAGB have 5|7-fold and 4|4-fold ring cores, together with their variations, Mo dangling bonds, and tiny holes. Meanwhile, V S and V S2 are the main point defects in the crystals. Based on these structures, we established four sets of models to explain the formation of the edges and pits. Figure 4b, c demonstrate the models corresponding to LAGBs, HAGBs, single-point defects, and multiple-point defects before and after degradation. The LAGB is composed of 5|7-fold ring cores and variations in a wedge shape, and the HAGB is composed of 4|4-fold ring cores and variations in a zigzag shape (Fig. 4b). After the crystals were placed in PBS solution for a certain time, these fundamental shapes extended out of the lattice perpendicular to the zigzag edges and became larger, although their general shapes remained unchanged (wedge or zigzag) (Fig. 4c). When the reaction time was sufficiently long, the GB regions with two different shapes extended from the nanoscale into the microscale and were observed with the aid of OM, which is consistent with the experimental results shown in Fig. 4a. Thus, the final shape of the degraded GBs is decided by the original shape of the dislocations along the GBs. The degradation process favors dislocations perpendicular to the zigzag Mo edge because of their higher activity compared with dislocations perpendicular to the zigzag S edge or armchair edge 24,42 . Similarly, V S2 has an initial triangular shape, opposite to that of the parent crystal, leading to the formation of a large triangular hole with the same orientation as V S2 . In the event of two adjacent V S2 , the corresponding final pits will be connected or even combined, which accelerates the degradation of MoS 2 grains. Because V S has fewer dangling bonds than V S2 , it does not degrade easily, and the process is delayed in this area, resulting in a smaller triangular hole than the one that resulted from V S2 (Fig. 4c). Due to this variation in hole size, the sizes of pits in degraded MoS 2 grains vary from small to large (Fig. 4a).
Generally, the MoS 2 surface is kinetically stable with respect to oxidation at low temperatures, and the barrier at broken edges is much lower; hence, oxidation and degradation readily occur at such sites. In addition, the formation of MoO 3 groups leads to the addition of oxygen atoms at the broken edge, continuing the oxidation and degradation process 48 . In some respects, imperfections, such as dislocations and point defects, in CVD-grown monolayer MoS 2 crystals strongly influence the degradation process, which could potentially be further controlled by defect or domain size engineering in the future 42,50 . For instance, vacancies can be created or healed to modulate the defect density and degradation speed of monolayer MoS 2 via synthetic or postprocessing technologies. The degradation of monolayer MoS 2 crystals is affected not only by external conditions, such as solution concentration, pH, and temperature, but also by internal factors, such as grain size, GB angle, and defect density, which can be modulated via the synthesis conditions.

Conclusions
To summarize, the degradation of CVD-grown monolayer MoS 2 crystals with different misorientation angles in PBS solution is elucidated, and the morphological transformation, degradation behaviors, and structural cause are systematically investigated using OM, AFM, TEM, and STEM. The degradation process of MoS 2 crystals is evidenced to be induced by intrinsic defects in the as-grown crystals, including GBs and point defects, because of the greater density of dangling bonds and higher reactivity in these regions. We also introduced the phenomenon of angle-dependent selective degradation, which explains that HAGBs are degraded preferentially at a faster rate than LAGBs because HAGBs contain more 5|7-and 4|4fold ring cores, together with more tiny holes, due to the greater lattice mismatch between two adjacent grains. Once begun, the degradation process extends perpendicular to the zigzag edge of LAGBs, HAGBs, V S , and V S2 , resulting in broken edges in a wedge or zigzag shape and triangular pits with opposite orientation to the parent crystals. These triangular pits merge into each other, resulting in the complete degradation of monolayer MoS 2 crystals. Such a defect-induced degradation process can be further controlled via defect engineering methods, such as intrinsic or artificial defect modulation. MoS 2 is only one member of a large family of 2D TMDCs; the achievements here provide a template for exploring the degradation of other 2D TMDCs in biofluids and suggest their promising applications in bioabsorbable devices and systems with controlled lifetime and performance, obviating the requirement for surgical retrieval and eliminating solid electronic waste.