Observation of room‐temperature out‐of‐plane switchable electric polarization in supported 3R‐MoS2 monolayers

Abstract

Two‐dimensional (2D) ferroelectrics have attracted considerable attention due to their potential in the development of devices of miniaturization and multifunction. Although several van der Waals (vdW)‐layered materials show ferroelectricity, the experimental demonstrations of ferroelectric behavior in monolayers are very limited. Here we report the observation of room‐temperature out‐of‐plane switchable electric polarization in supported MoS₂ monolayers exfoliated from 3R‐stacked bulk crystals under ambient conditions. Using in situ piezoelectric force microscopy and Kelvin probe force microscopy in a glovebox, we reveal that trapped water/ice molecules are responsible for this switchable electric polarization and this conclusion is strongly supported by theoretical simulations. It is worth noting that the water/ice trapping in the monolayers exfoliated from 2H‐stacked MoS₂ crystals is not as much as that in 3R monolayers and, consequently, the out‐of‐plane electric polarization is missing there. Our findings indicate that monolayers with a trapped single layer of polar molecules might be emerging alternatives to 2D ferroelectrics. Furthermore, the stacking sequences may bring new properties and applications to 2D vdW materials not only when we stack them up but also when we thin them down.

INTRODUCTION

Owing to their low dimensionality and large surface-to-volume ratio, two-dimensional (2D) ferroelectrics are believed to have significantly different properties compared with the traditional bulk ferroelectric counterparts.^1–4 The coexistence of switchable electric polarization and sizable electronic band gap enable applications of 2D semiconducting ferroelectrics such as field-effect transistors, sensors, photonic devices, random access memory, and solar cells.^5–10 Several layered materials have been theoretically predicted to possess ferroelectricity.^11,12 Unfortunately, very limited numbers of these have been experimentally verified. Two experimentally well-studied ferroelectric layered materials are CuInP₂S₆ and In₂Se₃.^13,14 To our knowledge, monolayer ferroelectrics are only experimentally demonstrated in 1D-MoTe (out-of-plane polarization), SnSe (in-plane polarization), and α-In₂Se₃.^15–17 For ferroelectric polarization, noncentrosymmetric structures are usually needed. Breaking centrosymmetry could naturally happen to some materials, or through the introduction of extrinsic effects, such as doping, defects, strain, functionalization, and interface.^18–21 The weak interfacial bonding, the high sensitivity to extrinsic influences, and the ease of stacking in a controlled manner facilitate the creation of ferroelectricity in van der Waals (vdW)-layered materials/heterostructures and development of predesigned multifunction.

Recently, a rational strategy for producing stable ferroelectricity in 2D systems was proposed^22,23 and then successfully demonstrated.²⁴ Breakage of the vertical centrosymmetry of the parent monolayers and, consequently, formation of out-of-plane switchable electric polarization were realized by stacking two identical monolayers of hBN in AB or BA sequence or transition-metal dichalcogenides (TMDs) such as WSe₂, MoSe₂, WS₂, and MoS₂ in a rhombohedral (R) configuration.^25–27 To visualize these local ferroelectric domains and domain walls, hBN and TMD bilayers were stacked with tiny interlayer twisting angles, which pattern the ferroelectric domains of opposite polarization into a moiré period.^28,29 The external electric field-driven flipping of polarizations and the motion of the domain walls could be facilitated by the relative sliding of the two layers. At the same time, we noticed that this sliding ferroelectricity was also observed in 2H- and 3R-like MoS₂/WS₂ hetero-bilayers.³⁰ Together with other studies on stacking sequences induced new phenomena and applications in, for example, the magic angle moiré 2D systems, it is unambiguous that stacking sequence does matter for 2D materials and devices.^31,32 Although several ferroelectrics with nanometer thicknesses have been gradually discovered, the demand for uncovering switchable ferroelectric polarization in monolayers, especially semiconducting monolayers, still remains.

In this study, we investigated switchable ferroelectric polarization in the most well-known TMD monolayers: MoS₂ monolayers. MoS₂ bulk crystals have three polymorphs, named 1T, 2H, and 3R.³³ The digits indicate the number of layers in the crystallographic unit cell and the letters T, H, and R represent the types of symmetry of tetragonal (D_3d), hexagonal (D_3h), and rhomohedral (C⁵_3v), respectively. Among them, the 1T phase is metallic and the other two are semiconducting. As schematically shown in Figure 1B, although the stacking sequences of layers are different for 2H and 3R, the individual layers are absolutely identical, meaning that the monolayer from the 2H-stacked (our notation: 2H monolayer) MoS₂ crystal is nondistinguishable from the 3R monolayer in terms of their crystal structures. This could be why researchers believe that 2H and 3R MoS₂ monolayers have the same physical properties and rarely compare them, as they are “ONE” compound. Instead of bottom-up, we focus on the stacking sequence effects on monolayers when we thin a bulk crystal down (top-down). We observed robust room-temperature out-of-plane switchable electric polarization in supported MoS₂ monolayers exfoliated from 3R-stacked bulk crystals under ambient conditions but not for the monolayers from 2H-stacked crystals. To unveil the mechanism responsible for this, we conducted systematic control experiments including sample preparation, in situ piezoelectric force microscopy (PFM), and Kelvin probe force microscopy (KPFM) inside a glovebox. Our results indicate that water/ice molecules might be trapped at the interface between the MoS₂ monolayers and substrates and this water/ice trapping is responsible for the formation of switchable ferroelectric polarization. This is also strongly evidenced by our theoretical simulation. Less amount of water/ice trapping was also found in the monolayers from 2H-stacked MoS₂ crystals and this could explain why the polarization disappears. Although the exact reasons for the differences in these “two” types of monolayers are still unknown, our findings indicate that semiconducting TMD monolayers with a trapped single layer of polar molecules might be emerging building blocks of 2D ferroelectrics. Furthermore, the stacking sequences may bring new properties and applications to 2D vdW materials not only when we stack them up but also when we thin them down.

Figure 1. Characterization of 3R- and 2H-stacked MoS2 layers by (A) optical image, scale bar, 100 μm. Inset: photoluminescence (PL) mappings, scale bar, 4 μm. (B) Schematic diagram of 2H- and 3R-stacked MoS2 lattices. Note: the monolayers of 2H- and 3R-stacked crystals are identical in theory. (C) Low-wavenumber Raman spectra of 3R-stacked MoS2 lattices. (D) Low-wavenumber Raman spectra of 2H-stacked MoS2 lattices. The filtered relay emissions are shaded in C and D.

EXPERIMENTAL SECTION Sample preparation

3R- and 2H-stacked MoS₂ bulk crystals were purchased from HQ Graphene, Inc. 3R-stacked MoS₂ crystals were also grown using a chemical vapor transport method using MoCl₅ as a transporting agent. The Mo, S, and MoCl₅ powders were mixed in a stoichiometric ratio of 9:20:1 in an argon-filled glovebox. The mixtures were sealed in an evacuated silicon tube about 20 cm in size. The sealed tubes were placed in a two-zone furnace. The reaction zone and the grown zone were heated to 850 °C and 920 °C for 30 h, respectively. Then, the reaction zone was heated to 1080 °C, whereas the grown zone was maintained at 920 °C for 150 h. Finally, the furnace was cooled to room temperature and shiny 3R MoS₂ crystals were deposited at the bottom of the tubes. Using the typical mechanical exfoliation and dry transfer processes, thin-layer samples on silicon wafers with a 300 nm thick SiO₂ cap layer and a sputtered 50 nm Au film were obtained.

Raman/photoluminescence (PL) spectroscopy and PFM/KPFM measurements

Raman and PL measurements were performed using a WITec Raman system with 532 nm laser excitation. The laser intensity was maintained below 10 μW to prevent light doping or exciton generation from high-power excitation. Atomic force microscope (AFM) experiments were carried out in a glovebox under an argon atmosphere at room temperature and ambient pressure (Icon, Bruker). A platinum/iridium (Pt/Ir)-coated Si tip (SCM-PIT-V2) was used for PFM and KPFM measurements.

Computational section

Density functional theory (DFT) calculations were performed using the Vienna Ab initio Simulation Package code and exchange–correlation functionals in the Perdew–Berke–Ernzerhof's form within generalized-gradient approximation. The projector-augmented wave potentials were used with a kinetic energy cutoff of 520 eV. The energy convergence criterion was set to 10–6 eV. The force convergence was set to 0.005 eV/Å for geometry optimization. Partial occupancies of the Kohn–Sham orbitals were allowed using the Gaussian smearing method and a width of 0.05 eV. Grimmes DFT-D3 method with the Becke–Johnson damping function was used to determine the dispersion interactions. The $$\sqrt{3}\times \sqrt{3}\times 1$$ supercell of MoS₂ and two water molecules were used to develop the MoS₂–ice model. A large vacuum region with a thickness of 25 Å was added in the z direction to reduce interaction between adjacent slabs and a 5 × 5 × 1 k-point of Monkhorst–Pack meshes at the Γ center was used. The dipole correction was applied in the z direction to compensate for the artificial field introduced by the periodic boundary conditions.

RESULTS AND DISCUSSION

To avoid the influence of measurement environments, we prepared samples containing 3R- and 2H-stacked flakes side by side. As shown in the Supporting Information: Figure S1, the thin layers of 3R-stacked MoS₂ were first exfoliated onto Au-coated SiO₂/Si. Then, the thin layers of 2H-stacked MoS₂ were exfoliated and transferred onto the same piece of the substrate right beside the 3R sample. Figure 1A shows the optical image of 3R- and 2H-stacked MoS₂ thin layers exfoliated under ambient conditions. The PL intensity mappings of the selected areas (the frames in Figure 1A) clearly show the thickness dependence. The brightest emission is from the monolayers, which have a direct band gap.^34,35 Figure 1C,D presents the low-wavenumber Raman spectra of the layers with different thicknesses (colors corresponding to the crosses in Figure 1A). The absence of interlayer shear modes (the black ones) further confirms the presence of monolayers. In contrast to the thick 2H-stacked MoS₂ layers, the thick 3R-stacked MoS₂ shows no low-wavenumber Raman modes.³⁶ This could be a unique and efficient way to distinguish 3R and 2H MoS₂ crystals. X-ray diffraction patterns also show that our crystals are pure phases of 3R and 2H MoS₂ (Supporting Information: Figure S2).³⁷ The typical PL and high-frequency Raman spectra of monolayers, few, and thick layers of MoS₂ are presented in the Supporting Information: Figures S3 and S4.³⁸

We first performed the PFM study on the thin layers exfoliated from 3R-stacked MoS₂ crystals under ambient conditions. Figure 2A shows the topographic image of the region framed in the optical picture (Figure 2D). The corresponding out-of-plane amplitude and phase images obtained through nonresonant PFM measurements are presented in Figure 2B, C. It is obvious that the electrical polarization has formed. The absence of clear domains and domain boundaries, especially in the monolayer (framed in B and C), has also been reported in other 2D ferroelectrics.^14,39 Next, we conducted single-point poling experiments on the monolayer, where a stiff cantilever with a spring constant of 3 N/m was used and the DC bias voltage was between −10 V and +10 V with an AC voltage of 3 V. The PFM hysteresis loop at individual points of the supported 3R monolayer MoS₂ shows a typical butterfly loop with clear saturated amplitude regions (Figure 2E) and an opening of ∼5 V. Figure 2F shows the corresponding phase hysteresis loop. A 180° phase difference between two out-of-plane polarization states and the corresponding phase switch tuning points can be clearly observed. These results confirm the out-of-plane switchable ferroelectric polarization in the 3R monolayer MoS₂ prepared under ambient conditions.^13,40,41 The presence of such out-of-plane switchable polarization in the supported 3R monolayer MoS₂ is remarkably robust, as revealed by our systematic PFM studies on monolayers exfoliated under ambient conditions (for more details, see the Supporting Information: Figures S3 and S6). We also investigated the monolayers exfoliated from our own synthesized 3R-stacked MoS₂ crystals (the details of method of growth are described in Section 2.1). The same results are obtained (Supporting Information: Figures S3 and S8). Moreover, in addition to the MoS₂–Au interface, the 3R monolayer MoS₂ transferred onto SiO₂/Si under ambient conditions also shows switchable polarization (Supporting Information: Figure S9). Our findings confirm that the out-of-plane switchable polarization is a general property of the supported 3R monolayer MoS₂ prepared under ambient conditions.

Figure 2. Piezoelectric force microscopy (PFM) study of 3R-stacked MoS2 on Au-coated SiO2/Si prepared under ambient conditions. (A) AFM topography. (B, C) Corresponding PFM amplitude and phase. Scale bar, 4 μm. (D) Optical image. The white box in the figure represents a single layer. Scale bar, 4 μm. (E, F) Corresponding PFM amplitude and phase hysteresis loops for monolayer 3R-stacked MoS2.

We conducted out-of-plane PFM measurements of 2H monolayer MoS₂ transferred onto Au-coated SiO₂/Si under ambient conditions as a comparison in Figure 3. In contrast to the 3R monolayers, the typical butterfly curves in the PFM amplitude plot and hysteresis loop in the phase graph disappear. Instead, a linear loop occurs, indicative of a nonferroelectric nature. Considering the fact that ideal free-standing MoS₂ 3R and 2H monolayers are identical and there should be no out-of-plane ferroelectric polarization in this single molecular layer, we believe that the observed out-of-plane switchable polarization is an extrinsic property.

Figure 3. Piezoelectric force microscopy (PFM) study of 2H-stacked MoS2 on Au-coated SiO2/Si prepared in ambient conditions. (A) AFM topography. (B, C) Corresponding PFM amplitude and phase. Scale bar, 5 μm. (D) Optical image. Scale bar, 10 μm. (E, F) Corresponding PFM amplitude and phase hysteresis loops for monolayer 2H-stacked MoS2.

All the 3R monolayers were exfoliated and transferred onto a supporting substrate under ambient conditions. Hence, we further presume that such electrical polarization may arise from the interface, specifically, the trapped water/ice molecules there. To confirm this, we exfoliated and transferred 3R-stacked MoS₂ thin layers onto Au-coated SiO₂/Si inside a glovebox and immediately performed the PFM study in the same glovebox. Figure 4 presents the PFM data. It is clear that the arid environment (humidity <0.10 ppm) suppresses the out-of-plane switchable polarization. The results prove our hypothesis that the trapped water/ice could cause the formation of out-of-plane polarization in the supported 3R monolayer MoS₂.

Figure 4. Piezoelectric force microscopy (PFM) study of 3R-stacked MoS2 ultrathin layers prepared in a glovebox. (A) AFM topography. (B, C) Corresponding PFM amplitude and phase. Scale bar, 5 μm. (D, E) Corresponding PFM amplitude and phase hysteresis loops for monolayer 3R-stacked MoS2.

With the advantages of high sensitivity and spatial resolution, KPFM has been widely used to probe the surface potential and work function of 2D materials. Therefore, KPFM is a unique and powerful tool to reveal the existence of trapped water/ice molecules, as the contact potential difference (CPD) and the work function of MoS₂ monolayers are sensitive to the molecules absorbed at the surface.^20,42 Figure 5A,C shows the topographic images of 3R-stacked MoS₂ layers prepared under dry and ambient conditions, respectively. The increase in the heights between two types of monolayers (see the insets) matches the thickness of single water/ice layer.^43,44 The corresponding CPD images are shown in Figure 5B,D. An obvious change of CPDs is clear between the monolayers with and without water/ice molecules. The CPD between the probe tip and the sample is given by CPD = $${{(\varphi }_{\text{tip}}-\varphi }_{\text{sample}}\text{)/}e$$, where e is the elementary charge, and φ_tip and φ_sample are the work functions of the KPFM tip and the sample, respectively. To ensure the consistency of our measurements, we performed KPFM on highly ordered pyrolytic graphite (HOPG) samples before each KPFM measurement. The average CPD for the HOPG reference was 300 mV (Supporting Information: Figure S10). The work function of the Pt/Ir-coated tip was first calibrated with HOPG (φ_HOPG = 4.60 eV).⁴⁵ Therefore, we determined the work function of the tip to be 4.90 eV.⁴⁶ The average CPD for the Au-coated SiO₂/Si substrate was −200 mV. The work function of the Au-coated SiO₂/Si substrate was determined to be 5.10 eV, in good agreement with the published value.⁴⁷ The surface potential difference, ΔCPD, can be calculated by $$\unicode{x02206}{\text{CPD\unicode{x000A0}=\unicode{x000A0}CPD}}_{\text{s}\text{ample}}-{\text{CPD}}_{\text{sub}\text{strate}}$$ = [($${{\varphi }_{\text{tip}}-\varphi }_{\text{sample}})-{{(\varphi }_{\text{tip}}-\varphi }_{\text{substrate}})$$]/e, where φ_substrate represents the work function of the substrate. The CPD value of 3R monolayer MoS₂ on Au-coated SiO₂/Si prepared under ambient conditions versus in a dry environment was 67 mV and the work function increased by 67 meV.

Figure 5. Investigation of the effect of the interfacial water/ice layer on the morphology and surface potential of 3R-stacked MoS2 on Au-coated SiO2/Si. (A) In situ AFM height image of 3R-stacked MoS2 on Au-coated SiO2/Si in a glovebox. (B) Corresponding spatial map of the surface potential recorded by Kelvin probe force microscopy (KPFM) height image of 3R-stacked MoS2 on Au-coated SiO2/Si outside a glovebox. (D) Corresponding spatial map of the surface potential recorded by KPFM of 3R-stacked MoS2 on Au-coated SiO2/Si outside a glovebox. Scale bar, 5 μm. (E) Model of hetero-bilayer composed of MoS2 and water monolayer. (F) Electrostatic potential energy difference ΔΦ near two opposite surfaces. Height line profile and VCPD line profile measured along the red line

To further study the trapped water/ice-induced out-of-plane electric polar behavior in the MoS₂ monolayer, we constructed a model of a hetero-bilayer composed of MoS₂ and a water monolayer, where the widely used model of a hexagon hydrogen-bonded network is adopted (Figure 5E).^48,49 Herein, the hexagonal ice monolayer is polar due to the vertical O–H bonds in half of the water molecules, giving rise to vertical polarization that can be revealed from the plane average electrostatic potential energy. The electrostatic potential energy difference ΔΦ, reaching up to ~0.72 eV near two opposite surfaces, is shown in Figure 5F, indicating the built-in electric field generated by ferroelectric polarization. The relatively smaller change in the work function in our experimental data could be due to the fact that the amount of trapped ice molecules is much lower than that in our model, where an entire continuous layer of ice molecules is anchored. The confinement effect-induced single-layer ice molecule trapping by 2D materials has been reported previously.^50,51 Our KPFM study also revealed that the changes in CPD and the work function in 2H monolayer MoS₂ prepared under ambient conditions are considerably less than those of the 3R monolayer (4.8 meV vs. 67 meV), indicating that the trapped water/ice molecules are less in the 2H monolayer, which explains the absence of obvious ferroelectric polarization in the MoS₂ 2H monolayer.

CONCLUSIONS

We observed out-of-plane switchable electric polarization in the supported MoS₂ monolayers exfoliated from 3R-stacked crystals under ambient conditions, whereas the monolayers from the 2H-stacked MoS₂ bulk crystals showed no such polar behavior. The absence of ferroelectric polarization was also found in the 3R monolayers prepared in a glovebox, indicating that the water molecules may play an important role. Our KPFM measurements and theoretical simulation indeed revealed that water/ice molecules are trapped at the interface of 3R monolayers and the substrates and further cause the out-of-plane ferroelectric polarization. The exact reasons for the top-surface layer of 3R-stacked crystals anchoring more water molecules are under investigation. The findings herein demonstrate that supported 3R monolayer MoS₂ and potentially other 3R monolayers might expand the scope of 2D ferroelectrics. This study may inspire research on emerging fundamentals and applications caused by the stacking sequences in 2D vdWs crystals. Here, the focus is not on the influence of stacking sequences on samples being stacked up (bottom-up), but those mono- or ultrathin layers being thinned down (top-down).

ACKNOWLEDGMENTS

This study was mainly supported by the National Key Research and Development Program of China (No. 2021YFA 1200800) and the Start-up Funds of Wuhan University. This study was also supported by the National Research Foundation, Singapore, under its Competitive Research Programmer (No. NRF-CRP22-2019-0007).

CONFLICTS OF INTEREST

The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT

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