Functional polymeric passivation-led improvement of bias stress with long-term durability of edge-rich nanoporous MoS2 thin-film transistors

Nanoporous patterning of two-dimensional materials using block copolymer lithography has drawn much attention. Lateral edge exposures made by the nanoporous patterning provide electrical and optical characteristics that are different from the original materials. However, nanopatterning processes inevitably generate edge exposure and surface defects that may result in poor reliability and reproducibility of the devices. In this study, we propose a reliable method to passivate nanoporous molybdenum disulfide (MoS2) thin-film transistors (TFTs) using polymer thin films, synthesized by initiated chemical vapor deposition (iCVD) to improve the electrical stability of nanoporous MoS2 TFTs. To this end, functional polymer films of electron-donating poly(1-vinylimidzole) (pVI) and inert poly(1H,1H,2H,2H-perfluorodecyl methacrylate) (pPFDMA) were utilized as passivation layers on nanoporous MoS2 TFTs. An n-type doping effect was observed in the pVI-deposited nanoporous MoS2 film due to the electron-donating imidazole ring, whereas the inert pPFDMA efficiently stabilized the electrical characteristics of the nanoporous MoS2 TFTs. Moreover, the hydrophobic fluoroalkyl chain of the pPFDMA film efficiently prevented oxygen and moisture adsorption on the nanoporous MoS2. The superior passivation effect of the pPFDMA layer was confirmed using gate-bias stress tests and long-term storage tests under ambient conditions.


INTRODUCTION
Owing to their high carrier mobility, sharp subthreshold slope, and exceptional endurance to mechanical stress, two-dimensional (2D) materials are considered as potential candidates for applications in functional electronics, such as flexible thin-film transistors (TFTs) for active display matrices and integrated circuits [1][2][3] . Owing to their large surface-to-volume ratio, various 2D material-based sensing devices have been successfully demonstrated in the past decade, including phototransistors 4-6 , gas sensors 7,8 , and biosensors 9,10 .
The unique electrical and optical properties of 2D materials can be systematically tuned using structural engineering, making 2D materials ideal for electronic devices. Modifying the dimensions or morphology of 2D materials changes their structural properties, which imparts various exotic properties to overcome existing material limitations 11 . As a well-known example, the zero bandgap of pristine 2D graphene does not exhibit good capability for electronic applications. However, it has been modified to form nanodot 12 , nanoribbon 13 , and nanomesh 14 , achieving a finite bandgap. The nanoribbon and nanomesh graphene TFTs also exhibited superior electrical switching properties. Moreover, Kim et al. 15 demonstrated the potential to overcome the low photoluminescence of molybdenum disulfide (MoS 2 ) with an indirect bandgap by patterning the 2D materials to nanopore sheets.
Block copolymer (BCP) lithography is considered a facile but highly effective scheme for constructing nanostructured 2D materials. The self-assembled nature of BCP can provide largearea uniform patterning with regular nanosized features [16][17][18] . The simple BCP-based lithography process can obtain high-quality nanopatterns with various shapes without using complex, highcost processes, such as electron beam (e-beam) irradiation. While the pristine 2D materials are comprised of van der Waals surface without any lateral edge regions, the nanostructured 2D materials made by periodically removing the layer to the size of nanometers provide the lateral edge exposures that the pristine 2D materials fundamentally cannot have. The lateral edge exposures exhibit unique properties; depending on the edge exposure structure (i.e., armchair and zigzag edges), the interaction of electron and phonon can be controlled 19 or oxygen reduction reaction catalysts can be obtained 20 . The mentioned lateral edge exposures have been expanding the development of various applications such as optoelectronics, chemical and biomedical sensing, and catalytic activity 14,15,21,22 . Admittedly, nanopatterning can improve device performance. However, there is a severe disadvantage related to the lack of electrical stability, which results from the edge exposure and structural defects caused by the nanopatterning process. It is well known that the surface edges and defects on 2D materials are highly sensitive to oxygen and moisture in ambient environment 23 . Consequently, the adsorption of oxygen and moisture on the nanopatterned surface acts as charge trap sites, limiting electron transport, which changes the electrical characteristics of the devices 24 and causes hysteresis due to charge trapping or release 25 . Therefore, it is essential to passivate the nanopatterned surface efficiently from external contaminants using an appropriate passivation layer without damaging the 2D semiconductor.
Various inorganic and organic layers have been employed for the passivation of 2D materials using several deposition methods, including atomic layer deposition (ALD) 26,27 , pulsed laser deposition 28 , and solution-based deposition 29,30 . However, these methods can induce damage and degradation on the channel surface and affect the electrical performance of nanopatterned TFTs. The deterioration in performance is caused by the susceptibility of the unstable edges and defects of the nanopatterned surface to high temperatures and solvent environments.
In this study, we propose polymeric passivation of a nanopatterned MoS 2 TFT via initiated chemical vapor deposition (iCVD). The vapor-phase process enables solvent-free, near-roomtemperature deposition of polymer thin films without damaging or changing the electrical characteristics of the underlying 2D materials 31 . In contrast, conformal coverage of 2D material surfaces without perturbation is challenging in conventional ALD and solution-based deposition, primarily because such deposition processes are highly dependent on the surface chemical properties or prone to damage from solvent 32,33 . Moreover, ultrathin polymer films can be deposited conformally on any arbitrary 2D material surface because, during the iCVD process, the polymer film grows on the substrate surface from the input monomers adsorbed thereon 34,35 . We employed two different polymer films of poly(1-vinylimidzole) (pVI) and poly(1H,1H,2H,2H-perfluorodecyl methacrylate) (pPFDMA) to passivate nanoporous MoS 2 TFTs. We confirmed conformally deposited polymer layers on the nanoporous MoS 2 and the corresponding electrical and chemical changes in the MoS 2 using Raman spectroscopy, ultraviolet photoelectron spectroscopy (UPS), scanning transmission electron microscopy (STEM), X-ray photoelectron spectroscopy (XPS), atomic force microscopy (AFM), and current measurements. Remarkably, the deposition of the inert pPFDMA layer on the nanoporous MoS 2 exhibited exceptional passivation performance under repeated gate-bias stress and long-term air exposure, which enabled the long-term stable operation of the nanoporous MoS 2 TFT.

RESULTS AND DISCUSSION
Chemical and morphological properties of nanoporous MoS 2 Figure 1a shows a schematic illustration of a fabrication process of nanoporous MoS 2 . The multilayer MoS 2 films were transferred onto a Si/SiO 2 substrate by a mechanical exfoliation from bulk. A 10 nm-thick SiO 2 was deposited onto the MoS 2 surface to prevent surface damage of the MoS 2 during the patterning process. Random copolymer (RCP) was spin-coated onto the SiO 2 surface before the BCP coating to improve wetting interactions between the solid substrate and the BCP layer. The BCP layer was selfassembled under annealing at 230°C and a periodic nanohole pattern was obtained by selective decomposition of polymer block located in the inner holes. A hexagonally packed cylindrical BCP film (Supplementary Fig. 1) was utilized as a lithographic mask to fabricate the periodic nanoholes on the multilayer MoS 2 . The underlying RCP, SiO 2 , and MoS 2 were sequentially perforated by O 2 , SF 6 , and BCl 3 plasma reactive ion etching (RIE), respectively. Finally, remained RCP and SiO 2 layers were removed by immersing the substrate in buffered oxide etchant (BOE).
The morphological and chemical properties of nanoporous MoS 2 were explored using STEM, Raman spectroscopy, and XPS analyses. Figure 1b illustrates a STEM image of the perforated multilayer MoS 2 with periodic nanohole arrays on its surface. The average hole diameter of the MoS 2 nanoholes was 24.29 ± 1.99 nm. A cross-sectional STEM image showed vertically organized nanoholes across nine MoS 2 layers (Fig. 1c). Figure 1d illustrates a STEM image of the hexagonally packed nanoholes on the MoS 2 . The atomic configurations in the exposed edge area consist of both arm-chair and zigzag arrangements of Mo and S atoms (Fig. 1e), which may be formed by anisotropic etching under the BCl 3 plasma RIE.
Raman spectroscopy and XPS analyses were conducted to investigate the electronic and chemical structure of the nanoporous MoS 2 . In the Raman spectra (Fig. 1f), two Raman peaks of inplane (E 1 2g ) and out-of-plane (A 1g ) vibration modes were observed at 382 and 407 cm −1 , respectively. The measured Raman peak intensity of A 1g was 2.04 times higher than that of E 1 2g because the vibration of A 1g is preferentially promoted to the E 1 2g vibration for the edge-exposed MoS 2 15,21,36 . Furthermore, a small peak was also observed at 226 cm −1 , corresponding to a defect-induced longitudinal acoustic vibration at the M point of the Brillouin zone (LA(M)) 37,38 . The Raman scattering mentioned above was rarely observed in the pristine MoS 2 , whereas it was activated considerably on the perforated MoS 2 films owing to the abundant edge defects.
The XPS spectra of the nanoporous MoS 2 are shown in Fig. 1g, h with respect to the core atoms of Mo and S. The spectrum of Mo 3d was deconvoluted into three components. The peaks located at 232.4 and 235.53 eV can be assigned to Mo 6+ 3d 5/2 and Mo 6+ 3d 3/2 , respectively, indicating the existence of MoO 3 39 . The atomic fraction of Mo 6+ 3d among the total Mo ligands was approximately 10.4%. The strong doublet at 229.19 and 232.32 eV corresponding to Mo 4+ 3d 5/2 and Mo 4+ 3d 3/2 of intrinsic MoS 2 (i-MoS 2 ) represented a stoichiometric Mo-S bonding (ratio of S/Mo = 2) 40 . However, nanoporous MoS 2 involves many broken bonds at edge sites. The nonstoichiometric Mo-S bonding (ratio of S/Mo <2) was referred to the defective MoS 2 (d-MoS 2 ) peaks located at 228.59 eV (Mo 4+ 3d 5/2 ) and 231.72 eV (Mo 4+ 3d 3/2 ) 40 . The atomic fraction of d-MoS 2 (16%) was higher than that of MoO 3 . The result indicates that many edge atoms were present with dangling bonds rather than in the oxidized form. For the S 2p spectra, two peaks were observed at 162 and 163.18 eV, corresponding to S 2− 2p 3/2 and S 2− 2p 1/2 , respectively, which is attributable to Mo-S bonding 39 . No defective peak was observed in the S 2p spectra owing to the low energy resolution at low binding energy levels. Figure 2a shows a schematic of the polymer deposition process on the nanoporous MoS 2 surface. In the iCVD process, the injected monomers are adsorbed to the cooled substrate, and initiators are decomposed to form radicals by the heated filament at a temperature of less than 200°C. It triggers the growth of the polymeric film on the nanoporous surface 41 .

Polymeric passivation of nanoporous MoS 2
Since the substrate temperature was <40°C, the solvent-free vapor-phase deposition method can minimize damage to the substrate surface, which is a distinct advantage of the iCVD process, thus rendering the deposition method suitable for damage-free passivation of the nanoporous MoS 2 . Accordingly, pVI and pPFDMA were selected as candidate layers for the passivation layer of the nanoporous MoS 2 TFTs (Fig. 2b). The pVI polymer includes an electron-donating (or electropositive) imidazole ring, which can affect the charge distribution of nanoporous MoS 2 . Indeed, in our previous study, threshold voltage (V TH ) and Dirac voltage (V Dirac ) in organic and graphene field-effect transistors could be controlled systematically by exploiting the electron-donating imidazole moiety in the pVI-containing layer 42,43 . Meanwhile, pPFDMA contains an inert, long perfluoroalkyl chain. Ultrathin (~70 nm) pVI and pPFDMA layers were deposited directly on the nanoporous MoS 2 TFT device, as shown in the high-resolution transmission electron microscopy (HRTEM) images and electrical characteristics of the polymer films (Supplementary Figs. 2 and 3 and Supplementary Table 1). Owing to the mild deposition conditions, the chemical functionalities including imidazole and fluoroalkyl moieties in each monomer were fully preserved in the resulting polymer film on the MoS 2 surface, which was confirmed by Fourier transform infrared (FTIR) spectroscopy ( Supplementary Fig. 4).
Raman spectroscopy and UPS analyses were performed to investigate the variation in the electrical characteristics of the nanoporous MoS 2 films with each polymer passivation layer (Fig.  2c-e). The pristine nanoporous MoS 2 films exhibited E 1 2g and A 1g peaks at 382 and 407 cm −1 44 , respectively. The pVI-passivated nanoporous MoS 2 showed a remarkable bathochromic shift in both vibration modes (Fig. 2c), whereas no apparent peak shift was observed from the pPFDMA-deposited nanoporous MoS 2 film (Fig. 2d). We also performed the Raman analysis on the nonpatterned MoS 2 with the polymer passivation layers (Supplementary Fig. S5). The pVI layer induced a negative shift of A 1g peak of the non-patterned MoS 2 . Since the Raman peaks of nonpassivated MoS 2 have a much higher intensity than the nanoporous MoS 2 , the shift seems to be relatively small. In addition, the shift of E 1 2g was not observed because it is less sensitive to electron doping than the A 1g vibration mode 45 . On the other hand, no apparent peak shift was observed in the inert pPFDMA passivation on the MoS 2 regardless of the surface structures. The UPS spectra of pristine and polymer-passivated nanoporous MoS 2 are shown in Fig. 2e. The work function (ϕ) can be calculated using ϕ = hν -W, where hν = 21.2 eV and W denotes the secondary electron cut-off (E cut-off ) 46 . The E cut-off of the nanoporous MoS 2 was shifted toward higher binding energy with the pVI passivation layer, resulting in a lower ϕ value of the pVIpassivated nanoporous MoS 2 (3.24 eV) than that of the pristine nanoporous MoS 2 (4.03 eV) (Fig. 2f). Meanwhile, the pPFDMApassivated nanoporous MoS 2 only exhibited a marginal increase in ϕ toward 4.10 eV, resulting from electronegative fluoroalkyl side chain 47,48 . These results were derived from the distinct electrical properties of the two polymers. The electron-donating group in the pVI polymer induced an n-doping effect 42 , while the inert fluoroalkyl chain in pPFDMA induced an insignificant p-doping effect. The similar electrical effects of two polymer layers were observed in the UPS analysis on the non-patterned MoS 2 without nanoporous structure ( Supplementary Fig. 6a, b), which represents highly coherent electrical properties of the polymers on the MoS 2 films regardless of the surface structures.

Characterization of polymer-passivated nanoporous MoS 2 film
In this study, we conducted high-angle annular dark-field (HAADF)-STEM analysis on the polymer-deposited MoS 2 film to confirm the effective passivation of two different polymer layers on the nanoporous MoS 2 . Cross-sectional HAADF-STEM images with energy dispersive spectroscopy (EDS) mapping are shown in Fig. 3a-j. Both pVI and pPFDMA passivation layers were deposited conformally on the nanoporous MoS 2 surface without any notable defects (Fig. 3a, f). The EDS mapping illustrates a uniform distribution of C and N (pVI) or C and F (pPFDMA), which are clearly distinguishable from underlying Mo and S. The surface chemical composition of the pVI-and pPFDMA-passivated nanoporous MoS 2 was investigated by XPS and compared with that of pristine nanoporous MoS 2 (Fig. 3k).
Accordingly, Mo 3d and S 2p peaks were detected in the XPS survey spectrum of pristine nanoporous MoS 2 . Moreover, Si-and O-related peaks were also detected, which resulted from the Si/ SiO 2 substrate. In the polymer-passivated nanoporous MoS 2 , the intensities of the Mo 3d and S 2p peaks were smaller than those of pristine MoS 2 . Evidently, the C 1s and N 1s peak intensities increased in the pVI-passivated nanoporous MoS 2 , and the increased intensities of C 1s and F 1s peaks were obtained in the pPFDMA-passivated nanoporous MoS 2 . In fact, the XPS analysis clearly revealed that the polymer passivation layer conformally covered the nanoporous MoS 2 surface, which is completely consistent with the cross-sectional HAADF-STEM images.
To further investigate the surface properties of the polymerpassivated nanoporous MoS 2 , AFM analysis was performed (Fig. 3l). The AFM image of the pristine nanoporous MoS 2 clearly exhibited porous structure and the smooth, amorphous surface morphology was confirmed with the root-mean-square roughness (R q ) <0.5 nm, which also indicates that the pVI layer conformally covered the nanoporous MoS 2 . On the other hand, the AFM image of the pPFDMA-passivated nanoporous MoS 2 exhibited the roughened surface with the R q value >15 nm. This rough surface morphology was originated from the highly crystalline nature of pPFDMA due to the fluoroalkyl chain 49 , which can also improve the hydrophobicity of the surface 50,51 . The water contact angle (WCA) analysis also verified that the surface property of the nanoporous MoS 2 was controlled according to the deposition of the polymer films ( Supplementary Fig. 7). Therefore, it follows from the comprehensive analyses that the vapor-phase-deposited polymer passivation layers conformally covered the nanoporous MoS 2 surface without incurring defect or damage to the underlying MoS 2 , while fully retaining the chemical functionalities of each polymer film.

Electrical characteristics of nanoporous MoS 2 TFTs passivated by pVI and pPFDMA polymers
To investigate the passivation effect of the pVI and pPFDMA layers on the nanoporous MoS 2 TFTs, we measured the electrical characteristics of the devices before and after the passivation with each polymer layer. Figure 4a  (V GS ) window ranging from −40 to 40 V and drain voltage (V DS ) of 1 V. A hysteresis between forward and reverse V GS sweeps was suppressed when the MoS 2 surfaces were passivated using both polymer layers, indicating that charge trapping on the MoS 2 edge-air interface was efficiently prevented by the polymer passivation. Following the deposition of the pVI polymer on the nanoporous MoS 2 TFT, a significant increase was observed in the drain current (I DS ) in depletion regions and a slight increase in the accumulation region (Fig. 4c). In addition, a negative V TH shift of −8.4 V was detected. Meanwhile, the deposition of pPFDMA only exhibited a marginal shift in the transfer curve, except for a small decrease in the on-current (Fig. 4d). The variations in V TH and on-off ratio (I on/off ) were monitored for 12 and 15 MoS 2 TFTs passivated with each pVI and pPFDMA layer, respectively (Supplementary Fig. 9a-d and Fig. 4e). For comparison, we defined that the current level at V GS of −40 V is off-current, even if the pVI-deposited MoS 2 TFT showed metallic behavior. Generally, a strong difference was observed in V TH and I on/off before and after the deposition of pVI; in contrast, the variation was insignificant in the case of the pPFDMA-passivated device. The n-doping behavior of the pVI-deposited nanoporous MoS 2 TFT can be attributed to the electron-donating effect of the imidazole ring in pVI (Fig. 4f). On the other hand, the long fluoroalkyl chain provided a strong hydrophobicity on the polymer passivation layer 51 , which effectively prevented oxygen and water adsorption, which, in turn, minimized the perturbation of the electrical characteristics of the nanoporous MoS 2 TFTs (Fig. 4g). In other words, the pPFDMA layer can effectively passivate the nanoporous MoS 2 surface with little change in the electrical characteristics of the TFT devices. Additionally, we extracted field-effect mobilities of 12 and 15 nanoporous MoS 2 TFTs before and after pVI and pPFDMA polymer passivation ( Supplementary Fig. 9e, f), respectively. We calculated the mobility variation using an equation of (μ after − μ before )/ μ before × 100 (%), where μ before and μ after are the field-effect mobility values measured before and after the passivation, respectively. The average variations for pVI and pPFDMA are calculated as −2.1 and −1.4%, respectively. Such small variations indicate that the polymer layers rarely induce the carrier scattering at the interface between the nanoporous MoS 2 and the polymer. In the output curves measured under V GS range from −20 to 40 V with an interval of 10 V, the increase in I DS was more prominent in the lower range of V GS than the higher range for pVI-deposited nanoporous MoS 2 TFT, which conforms with the observation in the corresponding transfer curve. However, a practically identical shape of the output curve was observed for the pPFDMAdeposited device, except for a slight decrease in I DS in positive V GS ranges.

Stability tests of pPFDMA-passivated nanoporous MoS 2 TFTs
The pPFDMA layer grown by iCVD provided an outstanding passivation effect on the nanoporous MoS 2 by benefitting from its non-electrostatic properties and its ability to prevent the adsorption of water molecules from the air. We conducted gatebias stress tests and long-term storage tests under ambient air conditions to examine the passivation performance of the pPFDMA layer on the nanoporous MoS 2 TFTs. Positive bias stress (PBS), negative bias stress (NBS), positive bias illumination stress (PBIS), and negative bias illumination stress (NBIS) under a white light source of 1500 lux were conducted (Fig. 5a-d). V GS of 30 and −22 V were applied as positive and negative gate stress voltages (V Stress ), respectively. We measured the transfer curves every 10 min while giving bias stress for a total stress time of 120 min as shown in Supplementary Fig. 10. These bias stress tests were also conducted on the non-passivated nanoporous MoS 2 control TFT to facilitate comparison, as shown in Supplementary Fig. 11a-d. Compared to the positive V TH shift of the non-passivated MoS 2 TFT under the PBS test, no appreciable V TH change was observed in the passivated TFT (Fig. 5a), suggesting that the pPFDMA passivation layer can effectively prevent the charge trapping induced by the adsorption of oxygen or water in an ambient environment at the edge defect sites. In the case of NBS, a slight shift in V TH was observed in the passivated MoS 2 TFT because the electrons pre-trapped at the edge sites in the nanoporous MoS 2 were released under the negative bias stress (Fig. 5b). However, it is worth noting that the variation was considerably smaller than that of the one occurring in the non-passivated device. For the PBIS test, the V TH was positively shifted in the non-passivated nanoporous MoS 2 TFT, while the passivated TFT showed a shift of V TH toward the negative direction, albeit to a minor degree ( Supplementary Figs. 11c and 5c). The PBS test indicated that edge sites were efficiently stabilized and the adsorption of oxygen or water molecules can be prevented by pPFDMA passivation. However, the inherent trap states in the nanoporous MoS 2 induced by the edge defects captured the photoexcited hole carriers under illumination, resulting in a positive gating effect thereby resulting in a small V TH shift toward the negative direction 5,6 . The negative shift was significantly activated under the NBIS test, in which the electron release and hole trapping in the trap states of the nanoporous MoS 2 occur simultaneously. Figure 5e shows the shift in V TH of the nanoporous MoS 2 TFT passivated and non-passivated by the pPFDMA layer at each measurement, where the variation of the passivated devices is quite small compared to that for the non-passivated TFT.
Furthermore, a long-term storage test was conducted to validate the robustness of the pPFDMA-passivated nanoporous MoS 2 TFT under exposure to an ambient environment. The average V TH shift of the pPFDMA-passivated nanoporous MoS 2 TFTs was monitored over 40 days with 5-day interval, compared to the non-passivated one (Fig. 5f). The raw transfer curves of each representative device are shown in Supplementary Fig. 12. The V TH of the non-passivated devices was positively shifted with time owing to the electron trapping induced by the adsorption of oxygen and water molecules on the nanoporous MoS 2 surface. It is well known that the energetically unstable broken bonds located at the edges of the nanoporous MoS 2 channel surface prompt the adsorption of oxygen and moisture far more than normal MoS 2 , which inevitably deteriorates the device performance 52,53 . In contrast, the nanoporous MoS 2 passivated by pPFDMA does not show appreciable device degradation and has a negligible V TH shift, which verifies the excellent passivation performance of the In summary, we demonstrated an efficient passivation method for nanoporous MoS 2 using polymeric passivation layers via iCVD process. The proposed solvent-free method effectively passivated the surface edges and defects in nanoporous MoS 2 , without damaging or degrading its structure and electrical integrity. The pVI with an electron-donating imidazole ring and pPFDMA containing a fluoroalkyl chain were utilized as n-doping and inert polymer passivation layers, respectively. Both polymer passivation layers conformally covered the underlying nanostructured MoS 2 . The nanoporous MoS 2 -based TFT with the polymer passivation layers showed a significant negative V TH shift in the pVI-deposited nanoporous MoS 2 TFT, whereas the pPFDMA-deposited nanoporous MoS 2 TFT only exhibited a small reduction in the on-current. On the other hand, the long fluoroalkyl chain in the pPFDMA provided strong hydrophobicity, thus remarkably enhancing the stability of the nanoporous MoS 2 TFT in ambient air. The pPFDMAdeposited nanoporous MoS 2 TFT demonstrated outstanding stability in the PBS, NBS, PBIS, and NBIS tests compared to the non-passivated device, primarily because the pPFDMA can efficiently abate the water and oxygen adsorption on the nanoporous MoS 2 surface. In addition, only a negligible change was observed in the long-term storage test, which confirms the robustness of the pPFDMA-deposited nanoporous MoS 2 TFT. We believe that this study can provide an effective strategy for the passivation of nanostructured 2D materials. In particular, a vaporphase deposited fluoropolymer can enhance the ambient stability of nanostructured 2D materials-based TFTs, without damaging their electrical characteristics.

Fabrication of nanoporous MoS 2 TFTs
Multilayer MoS 2 nanosheets were mechanically exfoliated from bulk MoS 2 and transferred onto Si/SiO 2 substrates. The substrate was then placed in acetone overnight and washed with isopropyl alcohol for surface cleaning. A 10 nm-thick SiO 2 layer was deposited onto the surface of MoS 2 using the e-beam evaporator. Subsequently, the substrate was spin-coated with a 1 wt% toluene solution of poly (styrene-r-methyl methacrylate) (PS-r-PMMA) (M n = 8500 M w /M n = 1.45, styrene content 66%, α-hydroxyl-ω-TEMPO moiety terminated) RCP, at a speed of 3000 rpm. The substrate was annealed at 250°C for 2 hours under vacuum and washed with toluene. Next, the substrate was spin-coated with a 1 wt% toluene solution of poly (styrene-b-methyl methacrylate) (PS-b-PMMA) (PS:PMMA = 55,000:22,000, M w /M n = 1.09) BCP at a speed of 3000 r.p.m. and annealed at 230°C for 2 h under vacuum conditions. The substrate was exposed to 254 nm ultraviolet irradiation for 30 min and placed in acetic acid for 15 min to decompose the PMMA section in the BCP thin film. O 2 plasma RIE (10 sccm, 50 W, 4 s) was used to etch the RCP layer and control the nanohole size. Next, SF 6 plasma RIE (10 sccm, 200 W, 20 s) was performed to create nanoholes on the SiO 2 layer. BCl 3 plasma RIE (15 sccm, 100 W) was utilized to fabricate nanoporous MoS 2 by controlling the treatment time, depending on the thickness of MoS 2 . The final process of making nanoporous MoS 2 involved immersing the substrate in BOE for 3 s to remove the SiO 2 layer. The source and drain were prepatterned using photolithography to fabricate the nanoporous MoS 2 TFT, and Ti and Au (20 nm/100 nm) were deposited on the nanoporous MoS 2 .
Synthesis of polymer films 1-Vinylimidzole (VI) (≥99%), 1H,1H,2H,2H-perfluorodecyl methacrylate (PFDMA) (97%) monomers and tert-butyl peroxide (TBPO) (98%) initiator were purchased from Sigma-Aldrich. The polymer passivation layers were deposited using a custom-built iCVD system. For the deposition of pVI, VI monomer and TBPO initiator were injected into the chamber at flow rates of 4.4 and 1.5 sccm, respectively. The chamber pressure was maintained at 450 mTorr. For the deposition of pPFDMA, PFDMA monomer and TBPO initiator were injected into the chamber at flow rates of 0.2 sccm, and the chamber pressure was maintained at 40 mTorr. For both polymer film depositions, the substrate temperature was maintained at 40°C, and the filament was heated to 130°C to decompose the initiator into radicals.

Characterization
Raman measurements were carried out using a micro-Raman spectrometer system (ALPHA300, WITec) with a green excitation laser. Energy band structure analysis was performed using UPS (AXIS SUPRA, Kratos. Inc.) from the Korean Basic Science Institute (KBSI). UPS was performed using a He I (hv = 21.2 eV) source from a gas discharge lamp. Plan-view and crosssectional STEM images of the nanoporous MoS 2 were obtained using a STEM measurement system (HD-2300A, Hitachi) with an accelerating voltage of 200 kV. Ion milling for the cross-sectional image was conducted using a single-beam focused ion-beam (FB-2100, Hitachi). For electrical characterization of the polymer films, 50 nm-thick Al electrode was thermally deposited under high vacuum <2 × 10 −6 torr and the polymer films were deposited between top and bottom Al electrode to fabricate the metal-insulator-metal structure. The XPS spectra of the nanoporous MoS 2 were obtained using K-Alpha (Thermo Fisher Scientific) with monochromated Al Kα radiation. FTIR spectra were obtained by an ALPHA FTIR (Bruker Optics) in the absorbance mode to analyze the chemical composition. The AFM images were obtained using a scanning probe microscope (XE-100, Park Systems) to analyze the surface morphology of the nanoporous MoS 2 and polymer films. The WCA analysis was performed using a contact angle measurement system (Phoenix Series, S.E.O.). The device was sliced by a focused ion beam (Helios Nanolab 450, FEI) and the cross-sectional HRTEM images were obtained and analyzed by field emission TEM (Talos F200X, FEI) and EDS mapping to visualize the polymer passivation layers on top of the nanoporous MoS 2 . The electrical measurements of the TFTs were performed using a semiconductor characterization analyzer (4200-SCS, Keithley) equipped with a probe station for sample loading and electrode contact. We measured the same devices before and after the polymer passivation. After the measurement of the bare MoS 2 devices, we deposited a polymer layer on the devices and opened the source and drain contacts by etching the local polymer layer using the BOE solution.

DATA AVAILABILITY
The data presented in this study are available from the corresponding author upon reasonable request.