Gate-Tunable In-Plane Ferroelectricity in Few-Layer SnS

Letter pubs.acs.org/NanoLett

Cite This: Nano Lett. XXXX, XXX, XXX−XXX

Gate-Tunable In-Plane Ferroelectricity in Few-Layer SnS Yang Bao,†,§ Peng Song,†,§ Yanpeng Liu,† Zhihui Chen,† Menglong Zhu,† Ibrahim Abdelwahab,† Jie Su,† Wei Fu,† Xiao Chi,† Wei Yu,† Wei Liu,† Xiaoxu Zhao,† Qing-Hua Xu,† Ming Yang,*,‡ and Kian Ping Loh*,† †

Department of Chemistry, National University of Singapore, Singapore 117543, Singapore Institute of Materials Research and Engineering, Agency for Science, Technology and Research, Singapore 117602, Singapore

‡

Downloaded via BUFFALO STATE on July 18, 2019 at 06:36:59 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Ultrathin ferroelectrics hold great promise for modern miniaturized sensors, memories, and optoelectronic devices. However, in most ferroelectric materials, polarization is destabilized in ultrathin films by the intrinsic depolarization field. Here we report robust in-plane ferroelectricity in fewlayer tin sulfide (SnS) 2D crystals that is coupled anisotropically to lattice strain. Specifically, the intrinsic polarization of SnS manifests as nanoripples along the armchair direction due to a converse piezoelectric effect. Most interestingly, such nanoripples show an odd-and-even effect in terms of its layer dependence, indicating that it is highly sensitive to changes in inversion symmetry. Ferroelectric switching is demonstrated in field-effect transistor devices fabricated on ultrathin SnS films, in which a stronger ferroelectric response is achieved at negative gate voltages. Our work shows the promise of 2D SnS in ultrathin ferroelectric field-effect transistors as well as nanoscale electromechanical systems. KEYWORDS: 2D materials, SnS, in-plane ferroelectricity, field-effect transistors, resistive switching, second-harmonic generation chieving stable ferroelectricity in ultrathin films is technologically important for the miniaturization of sensors,1 memories,2 and optoelectronic devices.3,4 For conventional ferroelectric materials such as perovskite oxides, the surface-charge-induced depolarization field becomes larger as the film gets thinner, which presents a constraint on vertical downscaling.5 The recent discovery of ferroelectricity in 2D van der Waals (vdW) materials presents a unique opportunity to investigate ferroelectricity in the ultrathin limit.6 Ferroelectricity in 2D materials can be categorized based on the direction of the spontaneous polarization as out-of-plane ferroelectricity (e.g., CuInP2S67,8 and 1T-WTe29), intercorrelated ferroelectricity (α-In2Se310−14), and in-plane ferroelectricity (e.g., SnTe15 and β′-In2Se316). For 2D ferroelectrics with either out-of-plane or intercorrelated polarization, the depolarization field still increases as the film gets thinner.8−10 In contrast, 2D materials with in-plane ferroelectricity retain the ferroelectric properties even as thickness decreases because the polarization is confined within the 2D plane and is separated by the vdW gap from adjacent layers, thus allowing it to be stable against out-of-plane perturbations. So far, the only 2D materials that have been verified experimentally to exhibit in-plane ferroelectricity are SnTe15 and β′-In2Se3,16 and their device applications remain unexplored. Here we report a systematic experimental study of the inplane ferroelectricity in 2D tin sulfide (SnS), which is predicted to exhibit a larger polarization than that of the

A

© XXXX American Chemical Society

SnTe.17−19 We have employed the molecular beam epitaxy (MBE) method to grow large-area, high-quality, few-layer SnS crystals on a range of substrates. Using optical secondharmonic generation (SHG) and piezoelectric force microscopy (PFM), we have demonstrated robust in-plane ferroelectricity in our samples under ambient conditions. Our scanning probe microscopy experiments further reveal coupling between the ferroelectric order and the lattice strain in the armchair direction of SnS via the converse piezoelectric effect. Ferroelectric field-effect transistor (FeFET) devices with SnS films as the channel materials have been fabricated, where both the polarization−voltage hysteresis curves and the SHG intensities indicate a stronger ferroelectric response at negative gate voltages. Our results provide the basis for future fieldeffect-controlled memory devices as well as electromechanical devices based on 2D ferroelectric materials. SnS adopts a low-symmetry layered orthorhombic crystal structure (space group Pnma) at room temperature, which converts to a high-symmetry cubic structure (space group Cmcm) above the Curie temperature (Tc). The Pnma phase of SnS is intrinsically in-plane polarized, whereas the Cmcm phase is nonpolar.17 The Curie temperature for monolayer SnS is ∼1200 K,17 and that for the bulk crystals is ∼800 K,20 Received: April 6, 2019 Revised: June 27, 2019 Published: June 28, 2019 A

DOI: 10.1021/acs.nanolett.9b01419 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. Growth of few-layer SnS for ferroelectric studies. (a) Schematic side view of a bilayer SnS structure. Each SnS layer is ferroelectrically polarized with a net dipole moment aligned with the x axis. Adjacent SnS layers are antiferroelectrically coupled due to the inversion symmetry in the AB-stacked structure (top panel). Schematic view of the wedding cake morphology of MBE-grown SnS crystals, which are inversion-symmetrybroken with nonzero net polarizations at the terraces (bottom panel). The dipole moment associated with each unit cell of SnS is illustrated by green arrows. (b) 3D display of the AFM image of a few-layer SnS crystal grown on mica. The inset shows the height profile along a line marked in panel b; the measured domain height of 0.6 nm matches well with the height of a SnS monolayer, suggesting a layer-by-layer growth. (c) 2D display of the AFM image of a few-layer SnS crystal grown on MgO(100). The inset shows the LEED diffraction pattern corresponding to highly oriented SnS domains. (d) Typical Raman spectrum (left panel) and the polar plot of the Ag(2) peak intensity as a function of the incident light polarization angles (right panel), collected on SnS crystals grown on MgO(100). AC and ZZ refer to the armchair and zigzag directions of SnS, respectively. (e) X-ray θ−2θ scan along the (00L)pc direction, showing only (00L) peaks for SnS, indicating that the SnS layers are perfectly parallel to the MgO(100) substrate.

SnS crystal grown on mica (film thickness determined by AFM topography at the edge; see Figure S2d), where stacked SnS domains are clearly resolved. The measured domain height of ∼0.6 nm (Figure 1b, inset) corresponds to the thickness of a single SnS layer. Highly oriented domains can be grown on epitaxial MgO(100) (AFM image shown in Figure 1c), as indicated by the sharp low-energy electron diffraction (LEED) spots (Figure 1c, inset) recorded on the films. Figure 1d shows a typical Raman spectrum of SnS films grown on MgO(100), where four sharp peaks characteristic of crystalline SnS are clearly resolved;26 the A2g Raman peak intensities have a strong correlation with the direction of the polarized light (see Figure 1d and Figure S3), showing a two-fold anisotropy with the maximum intensities along the 60 and 240° angles (armchair directions). The anisotropic Raman response of SnS on MgO(100) faithfully reproduces that of SnS single crystals,26 suggesting oriented epitaxy of the crystals on MgO. X-ray diffraction (XRD) results show that the basal plane (001̅ plane) of SnS is parallel to the MgO substrates (Figure 1e). The lack of inversion symmetry, which can be probed by the presence of optical SHG, is a prerequisite for ferroelectricity.13 Figure 2a compares the second-harmonic emission peak (λ = 610 nm) intensities generated from two samples, a 10 nm thick SnS on mica and a bulk SnS crystal, under normal incident excitation (λ = 1220 nm) and ambient conditions. The 10 nm thick film gives a prominent SHG peak (and a strong thirdharmonic generation (THG) peak; see Figure S4), whereas the bulk sample shows negligible SHG intensity. This is direct evidence that the 3D wedding cake morphology is responsible

suggesting its potential as a robust ferroelectric material. The upper panel of Figure 1a shows a schematic side view of two SnS layers. Each SnS layer contains polar Sn−S bonds aligned along the in-plane armchair direction, giving rise to the macroscopic electric polarization. SnS with even layers is coupled in an antiferroelectric manner with zero net polarization due to the inversion symmetry in the AB stacked structure; however, that with an odd number of layers lacks a perfect inversion center. Thus the inversion symmetry can be lifted in few-layer thin films with odd layers or by a perturbation such as a substrate21 or under an electric field.22−25 To study how the morphology of SnS thin films affects ferroelectric polarizations, we have grown ultrathin SnS using MBE on a range of substrates. The film thickness is controlled by varying the growth time and subsequently verified by spectroscopic ellipsometry (see Figure S1). Here we focus our attention on few-layer films because these terraced films allow us to study the odd−even effect in a convenient way. On various substrates, including MgO(100), Au(111), mica, graphene, and graphite, a Stranski−Krastanov growth mode was observed, in which stacked monolayer domains adopted a wedding cake morphology (see Figure S2 and schematic drawing in Figure 1a, lower panel). The growth morphology appears to be independent of the substrate, which is explained by the weak vdW interactions of SnS with these substrates. Because of the terraced morphology, there is an overall nonzero net polarization at terraces with odd-layer thickness. Atomic Force Microscopy (AFM) image of a seven layer thick B


Letter

Nano Letters

Figure 2. SHG and PFM studies of ferroelectricity in SnS. (a) Second-harmonic spectral intensity at 610 nm generated in reflection from a 10 nm thick SnS grown on mica (red) and a bulk SnS (blue). Samples are excited by laser pulses with a wavelength of 1220 nm. (b) Schematic view of the PFM measurement of the few-layer SnS crystals grown on corrugated graphite substrates. The SnS piezoelectric coefficient d11 is marked by yellow arrows, and the alternating current (AC) excitation field VAC is marked by the black arrow. (c) PFM topography and amplitude images of few-layer SnS grown on graphite. The PFM contrast indicates the presence of ferroelectric domains. (d) Single-point PFM amplitude (red) and phase (black) hysteresis loops taken on the sample shown in panel c. Both SHG and PFM experiments are performed under ambient conditions.

corrugation of ±10 nm is clearly observed. The corresponding PFM amplitude image (Figure 2c, lower panel) reveals that certain domains (purple color) give larger PFM signals than others due to the inhomogeneous surface morphology. We also perform PFM tip poling experiments to study the ferroelectric switching behavior by applying a direct current (DC) voltage to the sample while reading the PFM amplitudes and phases. At regions with strong piezoelectric responses, well-defined hysteresis loops can be recorded. A typical PFM amplitude response shows a butterfly loop with an opening voltage of ∼2 V, whereas the phase switches 180° at the same turning points (Figure 2d). Control experiments performed on bulk SnS crystals failed to obtain any PFM response, even after we intentionally tilted the crystal to measure the inclined surface (see Figure S5e,f). Scanning tunneling microscopy (STM) was performed on the SnS films to study the atomic structure of the polarized domains. STM imaging reveals the SnS lattice constants to be 4.31 and 4.04 Å, which correspond to the distance between the surface Sn atoms in the armchair and zigzag directions, respectively (Figure S6).28,29 For ML SnS islands grown on Au(111), domains with two different superstructures are observed (see Figure S7a). The first superstructure is made of 1D corrugations that are tilted by 43° from the armchair

for the broken inversion symmetry. The piezoelectric response of ultrathin SnS was further verified using PFM carried out in the ambient because all ferroelectric materials are also piezoelectric.12 In PFM, the sample deformation (strain) is related to the drive voltage through the converse piezoelectric effect, εj = dijEi, where εj refers to the strain in the j direction, Ei refers to the external field in the i direction, and dij is the piezoelectric coefficient. Theoretical studies show that ML SnS possesses only in-plane piezoelectric coefficients in the d11 and d12 directions (indices 1, 2, and 3 correspond to the x, y, and z directions marked in Figure 1a).13,22 This means that SnS crystals can only deform under electric fields applied exclusively along the in-plane directions.27 Therefore, layered SnS islands on flat substrates (e.g., graphene) do not respond mechanically to the vertical electric field applied by the tip, and only weak PFM signals are generated at the island edges (see Figure S5). To allow the electric field to have an in-plane component, the SnS crystals are tilted from the horizontal by using corrugated graphite as the substrate for MBE. At corrugated regions where SnS basal planes are inclined, mechanical responses in SnS are induced by the vertical electric field, leading to sizable PFM signals (Figure 2b). Figure 2c (upper panel) shows the topographic image of one SnS film (thickness ≈ 10 nm) grown on graphite, where large surface C


Letter

Nano Letters

Figure 3. Ferroelectricity in SnS studied by STM and nc-AFM. (a) (left) STM image of one ML SnS island with the nanoripple pattern. The inplane armchair and zigzag directions are determined from the lattice constants and are marked by the black arrows. The net ferroelectric polarization of this island, determined by the LCPD measurements, is marked by the white arrow. (right) Frequency shift measured as a function of the voltage, collected above the two edges marked by red and blue triangles, respectively. Measurements are performed in constant-height mode after compensating the surface tilting and thermal drifting effects. Parabolic fits and corresponding parabola peaks are indicated. (b) STM image of few-layer SnS with monolayer-thick steps, showing the nanoripple pattern in the middle terrace and the absence of nanoripple pattern in the upper and bottom terraces.(c) STM images of the nanoripple pattern observed on few-layer SnS films. With increased film thickness (from left to right), the nanoripple quasi-periodicity becomes larger. (d) DFT-calculated polarization as a function of the number of SnS layers. The STM and nc-AFM experiments are performed at 4.6 K in ultrahigh vacuum.

work functions, which is most likely due to electrical polarization. On the contrary, parabolic curves obtained at opposite edges along the zigzag direction did not show any shift (Figure S9). This is fully consistent with theoretical work17 that states that SnS is spontaneously polarized along only the in-plane armchair direction. Details of the LCPD measurements can be found in Figure S9. Figure 3b is the STM image of a few-layer SnS sample with ML steps, where the nanoripple pattern is observed only in the middle terrace. The appearance of the nanoripple pattern in alternating layers suggests an odd−even layer effect, which is a direct consequence of antiferroelectric coupling in even layers. Here we propose that the nanoripple pattern is induced by the converse piezoelectric effect. In SnS, the uniaxial strain ε1 can be related to the in-plane polarization P1 by the equation ε1 = P1 e11, where e11 is the linear piezoelectric coefficient in the armchair direction (indices 1, 2, and 3 correspond to the x, y, and z directions marked in Figure 1a).32 Our density functional theory (DFT) calculation reveals a P1 value of 2.6 × 10−10 C/m for ML SnS, and the reported e11 value is 18.1 × 10−10 C/m.32 Using these parameters, the calculated in-plane uniaxial strain is ∼14%. By applying this strain to the armchair direction of a 1 × 6 × 1 supercell of ML SnS, a rippling amplitude of ∼100 pm is reproduced by simulation (see Figure S10), which is three times larger than the observed ripple amplitude of ∼30 pm. This means the strain caused by the

direction of SnS (Figure S7b), and its origin is due to Moiré pattern induced by the lattice-misfit strain between SnS and the substrate. The observed Moiré periodicity of 1.77 nm can be reproduced by the superposition of orthorhombic SnS and hexagonal Au(111) lattices with a rotation angle of ∼7° (see Figure S8). Another superstructure, called here the nanoripple pattern (Figure S7c), contains quasi-1D corrugations with an average separation of ∼2.6 nm and is perpendicular to the armchair direction of SnS. This nanoripple pattern is not related to substrate-induced strain because once the film thickness increases above two layers, only the nanoripple pattern is observed, and Moiré pattern vanishes. On a typical ML island with the nanoripple pattern (Figure 3a, left panel), the local contact potential difference (LCPD) between the sample and tip was measured using noncontact AFM with a qPlus sensor.30,31 The vertex of the parabola obtained in the LCPD measurement reflects a minimum electrostatic force (i.e., less negative Df value) between the sample and the tip when their intrinsic work function difference is fully compensated by the bias applied across the tip−sample junction. The peak positions of two parabolic curves, obtained at the opposite edges along the armchair direction (labeled by red and blue triangles), are shown to be shifted by 16 mV (Figure 3a, right panel). Assuming that the tip work function is constant during the measurement, this shift implies that opposite edges along the armchair direction have different D


Letter

Nano Letters

Figure 4. I−V characteristics and SHG response of the ferroelectric field-effect transistor (FeFET) based on few-layer SnS. (a) I−V hysteresis curve of a SnS FET device (channel length L = 4 μm, cycled from −5 to 5 V). The voltages are swept in the order 0 → 5 → 0 → −5 → 0 V, as shown by the colored arrows with the four sweeps labeled as i, ii, iii, and iv. The coercive voltages, which refer to the minimum voltages required for ferroelectric switching, are labeled near the corresponding current peaks. Inset: Schematic illustration of the ferroelectric switching processes during the voltage sweeps i (processes 1 and 2) and ii (process 3). (b) I−V hysteresis curves of a lateral SnS memory device (channel length L = 4 μm, cycled from −5.5 to 5.5 V) measured at different gate voltages Vg. (c) Corresponding P−V hysteresis curves of the memory device at different gate voltages. (d) SHG peaks (λ = 400 nm) generated by exciting the back-gated SnS device with laser pulses (λ = 800 nm) at different gate voltages. Inset: SHG peak intensity as a function of the gate voltage.

provides the bottom gate. Two parallel Au electrodes on SnS define the channel. For nongated devices, robust I−V hysteresis curves typical of ferroelectric switching were observed from 4 up to 298 K (Figure S12). The observation of hysteresis suggests that the device resistance is changed by polarization switching of SnS. This can be attributed to the fact that the Schottky barrier width and height at the Au/SnS interface is altered by polarization reversal.11,14 The electronic band profile (and hence the conductivity) of each SnS domain is also changed by polarization flipping.35 By applying negative gate voltages, we observed a substantial increase in the magnitude of hysteresis curves. Figure 4a shows a typical I−V curve for a memory device performed in the range of +5 to −5 V when a back-gate voltage of −50 V is applied. The arrows in the figure denote the voltage sweep directions. Increasing from 0 V to positive biases (Figure 4a, sweep i), the device initially shows a low-resistance state (LRS), as the SnS channel remains a negative polarization from the previous negative bias sweep (Figure 4a, inset 1). When the positive bias (+4.3 V) is larger than the coercive field, the negatively polarized SnS domains start to reverse to positive polarizations. From +4.3 to +5 V, the device enters a high-resistance state (HRS), leading to a sudden decrease in the current in sweep i (Figure 4a, inset 2).

converse piezoelectric effect is large enough to induce spontaneous rippling in ML SnS. By forming nanoripples, the total energy of SnS is reduced as the dipole moment in the armchair direction is partially neutralized. A similar phenomenon was observed for piezoelectric ZnO nanobelts with polar surfaces,33,34 which undergo a self-coiling process to reduce the electrostatic energy. It is qualitatively observed that the ripple periodicity increases from 2.6 to 8 nm with the SnS deposition time increased from 30 s to 5 min (Figure 3c), which is consistent with the decreasing ferroelectric polarization in thicker SnS films predicted by DFT calculations (Figure 3d). Ripple formation is fully suppressed in films with a growth time of 20 min (see Figure S11), suggesting that ferroelectric polarization is fully quenched in bulk-like SnS crystals. Finally, we explore switching and gate tunability of in-plane ferroelectricity in 2D SnS by fabricating FeFET devices. Because of the special wedding cake morphology of the MBEgrown SnS, at an average thickness of 7 nm, only the bottom SnS layers merge to form a film, which significantly limits the conductivity. Therefore, 15 nm thick SnS films were chosen to fabricate devices to ensure a good channel conductivity. Devices were fabricated on 15 nm thick SnS (grown on mica) transferred on 300 nm thick SiO2 on doped Si; the latter E


Letter

Nano Letters

Figure 5. DFT calculation of the electric-field effect on bilayer SnS. (a) Atomic structure (side view) of bilayer SnS without external electric field. (b) Atomic structure (side view) of bilayer SnS with −0.4 V/Å electric field applied along the z direction, superimposed with the contour plot of the electron density redistribution induced by the electric field, projecting on the (100) plane cutting through S1−Sn2 atoms (out-of-plane charge redistribution). The blue (red) color represents electron-rich (-deficient) regions. (c,d) Contour plots (top view) of the electron density redistribution induced by the electric field, projected on the (001) plane cutting through the Sn1 atom and Sn2 atom, respectively (in-plane charge redistribution). The electron density redistribution Δρ is defined by the electron density difference between the electron density (ρE) of the SnS bilayer with electric field and the electron density (ρ0) of the SnS bilayer without the electric field (Δρ = ρE − ρ0).

curves exhibit clear current peaks corresponding to the ferroelectric switching process. Given that the source-drain voltage is swept at a constant rate (1 V s−1), the polarization− voltage (P−V) curves (Figure 4c) can be obtained by P(V) = 1 ∫ I(V ) dV , where d is the distance between the source and d drain electrodes.41,42 From the P−V curves, it is evident that the remnant polarization increases with increasing p-doping level (more negative gate voltages). Besides the enhanced polarization, both the asymmetry in I−V curves and the discontinuity in P−V curves indicate switching characteristics analogous to a diode (i.e., asymmetric conductance in forward and reverse sweeps). This effect can be ascribed to the asymmetric Schottky barriers at the Au−SnS interfaces43 or inhomogeneous doping in the channel. More data for the gated devices can be found in Figure S15. One question is whether the application of the gate voltage increases the extent of inversion symmetry breaking in the SnS sample. It has been previously reported that the inversion symmetry in bilayer MoS2 can be broken either by an external electric field23 or by the field-induced charge imbalance.44 SHG is a sensitive probe of broken inversion symmetry and an excellent tool for the investigation of ferroelectric order.13,45 We monitor the SHG intensity of the device as a function of the gate voltages (Figure 4d). Interestingly, the SHG peak (λ = 400 nm, under normal incident excitation λ = 800 nm) intensity increases with more negative gate voltages and decreases with more positive gate voltages; such a trend mirrors the trend observed in the P−V curves in Figure 4c, where a stronger ferroelectric response is observed at a more negative gate voltage. To probe the origins of the gate-dependent ferroelectricity of the SnS thin films, we performed DFT calculations on bilayer SnS, which has inversion symmetry. Figure 5a,b show

The device stays in the HRS as the voltage is decreased to zero (sweep (ii), when the net polarization remains in the same direction as the external field (Figure 4a, inset 3). Here the measured coercive voltage is ±4.3 V (voltages at the current peaks), corresponding to a coercive field of ∼10.7 kV/cm, which is comparable to that of conventional ferroelectrics such as perovskite oxides.36,37 Theoretically, polarization switching of a single SnS domain requires a high coercive field of 2.9 × 103 kV/cm.38 In our multidomain SnS film, ferroelectric switching is more likely to be governed by domain wall motion. According to a recent work, the domain wall energy of SnS is remarkably small (38 mJ/m2, much smaller than that of PbTiO3 with 169 mJ/m2), which implies that the ferroelectric domain wall is highly mobile under the influence of an external field.18 The domain wall motion can result in the growth and enlargement of SnS domains whose polarizations are aligned with the external electric field.39 The on/off ratios of the memory device, defined by the channel resistance of the HRS divided by that of the LRS, is plotted against the source-drain biases Vsd. The maximum on/off ratio of ∼100 occurs at the source-drain bias of +2.1 V and the gate voltage of −40 V, demonstrating that the device is functional as a random access memory, requiring about +2 V to read and ±4.3 V to write (see Figure S13). To investigate if the ferroelectricity in SnS can be tuned electrostatically, the I−V hysteresis curves (Figure 4b) were collected as a function of gate voltage. When a positive gate voltage is applied (Vg > 0), electrons are injected into the ptype SnS crystals,40 leading to carrier depletion. Because of the depletion of holes, the current in the channel drops significantly (see Figure S14), and the hysteresis curves become experimentally inaccessible. At negative gate voltages (−50 ≤ Vg ≤ −40 V), it is observed that the I−V hysteresis F


Letter

Nano Letters

AXS D8 Venture apparatus. Raman spectra were recorded at room temperature using the confocal WiTec alpha300 R Raman microscope. The polarized Raman measurement was achieved by placing a half-wave plate in the incident laser path. To extract the intensities of the Raman scattering, peaks were fitted by Lorentz lines after a smooth background was subtracted from the original data. Piezoelectric Force Microscopy. PFM measurements were performed using a Bruker Dimension Icon AFM in contact mode. Pt-coated silicon tips with a radius of 20 nm and a force constant of 1 N m−1 were used to measure the local switching spectrum and the PFM image. The drive frequency and drive amplitude (VAC) were ∼200 kHz and 1.5 V, respectively. To obtain the hysteresis curve, a DC voltage of ±10 V was applied to the sample, and the phase and amplitude of the PFM signal in the out-of-plane channel were recorded. Scanning Probe Microscopy. SnS films for UHV STM/ncAFM studies were grown on the Au(111) substrate. The Au(111) substrate (Mateck) was cleaned by repeated argon ion sputtering and thermal annealing. The growth of SnS films was carried out by physical vapor deposition, as detailed in the previous section. After growth, the sample was transferred into the analysis chamber, and measurements were performed at 4.5 K, using an integrated STM/nc-AFM sensor (Omicron, sensor frequency f 0 = 23 kHz, Q = 20 000) with simultaneous current/force detection capabilities. The probe tip was formed by repeated indentation into the Au surface. Device Fabrications and Measurements. SnS thin films grown on mica were first transferred onto SiO2 (300 nm)/Si substrates using the poly(methyl methacrylate) (PMMA)mediated dry transfer method. Wet transfer methods were not attempted due to the concern that etchants could damage the SnS films. Typically, a layer of PMMA was spin coated on SnS/ mica substrates and baked at 80 °C for 2 min. SnS flakes were then picked up using Scotch tape and released onto the SiO2 (300 nm)/Si substrates. SnS films grown on MgO(100) could not be picked up by the same dry transfer method, possibly due to their strong interaction with the MgO substrates. Electrodes were patterned on SnS flakes without mica beneath (confirmed by the absence of Raman signal of mica) using standard electron beam lithography and thermal evaporation of Cr/Au (2/60 nm). Electrical measurements were carried out in an Oxford Teslatron system, using a Keithley 2400 instrument to supply gate voltage and a Keithley 6430 instrument to probe the source-drain current. Second-Harmonic Generation Measurements. For nongated samples, a pulsed Yb:KGW PHAROS laser system was used as the pump of a collinear optical parametric amplifier ORPHEUS with a LYRA wavelength extension option (Light Conversion, pulse duration of 180 fs, repetition rate of 100 kHz). The excitation beam (λ = 1220 nm) was reflected by a short-pass dichroic mirror (Thorlabs DMSP1000) and focused onto the sample with a 100× (NA = 0.9) air objective from Nikon. The nonlinear emission was collected in a backscattering configuration via the same objective and detected by a spectrograph (PI Acton SP2300 by Princeton Instruments) for spectral measurements. For gate-dependent SHG measurements, we excited the flake with a mode-locked Ti:sapphire laser (Chameleon Ultra II, Coherent) working with repetition rate of 80 MHz and pulse duration of 140 fs. The laser beam was focused by a 10× objective lens (NA = 0.30) with a radius of ∼1 μm. The excitation wavelength at 800 nm output from the laser was employed to excite the sample. The SHG signals

the relaxed structure of bilayer SnS with and without an external field in the z direction. In the presence of an external field (−0.4 V/Å), the distance of the Sn atom in the top layer to its nearest S atom in the bottom layer (Sn1−S2) is elongated from 3.396 to 3.408 Å, whereas the distance of the Sn atom in the bottom layer to its nearest S atom in the top layer (Sn2−S1) is shortened from 3.396 to 3.383 Å. The asymmetry between these two interlayer Sn−S distances increases with the magnitude of the external electric field (see Figure S16), which clearly demonstrates field-induced inversion symmetry breaking in bilayer SnS. Besides the out-ofplane geometric distortions, the in-plane Sn1−S1 bond length in the top SnS layer is reduced from 2.684 to 2.681 Å, in contrast with the increased Sn2−S2 bond length (2.688 Å) in the bottom SnS layer. The different in-plane geometric distortions between the top and bottom SnS layers contribute to a net in-plane ionic polarization in the otherwise nonpolar SnS bilayer.17 Furthermore, because of the electric field, both the out-of-plane (see the contour plot in Figure 5b) and the inplane (see the contour plots in Figure 5c,d) electron density distribution in the top and bottom SnS layers become asymmetric, indicating broken wave function symmetry. The relationship between the wave function of the occupied states W n(λ)(r ) and the electronic contribution to ferroelectric e M polarization Pe(λ) is given by Pe(λ) = P( e λ) = − Ω ∑n = 0 ∫ r|Wn(λ)(r)|2dr, where e is the charge, Ω is the volume of the unit cell, r is the displacement vector, and Wn(λ)(r) is the Wannier wave function of the occupied states.46 In conclusion, we have observed robust in-plane ferroelectricity in ultrathin films of 2D SnS grown by vdW epitaxy on a range of substrates. Ferroelectric FET devices have been fabricated on 2D SnS films, where electrostatic gating was applied to modulate the ferroelectric behaviors. Compared with conventional oxide-based ferroelectric FET devices where an additional semiconducting channel layer is required in addition to the capacitor layer, the use of a ferroelectric semiconductor simplifies device integration and improves memory density. The inverse piezoelectric effect, as manifested by spontaneous rippling of the film along the direction of polarization, was observed for films with an odd number of layers. We also discovered coupling between the ferroelectric order and the anisotropic lattice (and thus the valley degree of freedom) in SnS, suggesting potential applications for polarization-dependent electromagnetic, electromechanical, and optoelectronic devices. Methods. Molecular Beam Epitaxy of SnS Films and Characterizations. SnS films were grown in an ultrahigh vacuum (UHV, base pressure of 1 × 10−10 mbar) chamber. Prior to growth, the substrate was degassed in the UHV chamber at 200 °C. While keeping the substrate at 200 °C, SnS powders were evaporated from a crucible heated to 450 °C. After growth, the UHV chamber was backfilled with Ar (1 × 10−5 mbar), and the sample was postannealed at 200 °C to obtain larger SnS domains. Grade V-1 mica substrates and grade 1 to 3 HOPG substrates were purchased from SPI supplies. Au(111) substrates were purchased from Mateck. MgO(100) substrates were purchased from MTI Corporation. Graphene substrates were prepared by transferring CVD graphene grown on Cu foils to SiO2. The surface morphology and thickness of as-grown samples were checked using a Bruker Dimension FastScan AFM in tapping mode. Single-crystal XRD was conducted in a Bruker G


Letter

Nano Letters

(8) Liu, F.; You, L.; Seyler, K. L.; Li, X.; Yu, P.; Lin, J.; Wang, X.; Zhou, J.; Wang, H.; He, H.; Pantelides, S. T.; Zhou, W.; Sharma, P.; Xu, X.; Ajayan, P. M.; Wang, J.; Liu, Z. Nat. Commun. 2016, 7, 12357. (9) Fei, Z.; Zhao, W.; Palomaki, T. A.; Sun, B.; Miller, M. K.; Zhao, Z.; Yan, J.; Xu, X.; Cobden, D. H. Nature 2018, 560 (7718), 336− 339. (10) Xue, F.; Zhang, J.; Hu, W.; Hsu, W.-T.; Han, A.; Leung, S.-F.; Huang, J.-K.; Wan, Y.; Liu, S.; Zhang, J.; He, J.-H.; Chang, W.-H.; Wang, Z. L.; Zhang, X.; Li, L.-J. ACS Nano 2018, 12 (5), 4976−4983. (11) Poh, S. M.; Tan, S. J. R.; Wang, H.; Song, P.; Abidi, I. H.; Zhao, X.; Dan, J.; Chen, J.; Luo, Z.; Pennycook, S. J.; Castro Neto, A. H.; Loh, K. P. Nano Lett. 2018, 18 (10), 6340−6346. (12) Zhou, Y.; Wu, D.; Zhu, Y.; Cho, Y.; He, Q.; Yang, X.; Herrera, K.; Chu, Z.; Han, Y.; Downer, M. C.; Peng, H.; Lai, K. Nano Lett. 2017, 17 (9), 5508−5513. (13) Xiao, J.; Zhu, H.; Wang, Y.; Feng, W.; Hu, Y.; Dasgupta, A.; Han, Y.; Wang, Y.; Muller, D. A.; Martin, L. W.; Hu, P.; Zhang, X. Phys. Rev. Lett. 2018, 120 (22), 227601. (14) Cui, C.; Hu, W.-J.; Yan, X.; Addiego, C.; Gao, W.; Wang, Y.; Wang, Z.; Li, L.; Cheng, Y.; Li, P.; Zhang, X.; Alshareef, H. N.; Wu, T.; Zhu, W.; Pan, X.; Li, L.-J. Nano Lett. 2018, 18 (2), 1253−1258. (15) Chang, K.; Liu, J.; Lin, H.; Wang, N.; Zhao, K.; Zhang, A.; Jin, F.; Zhong, Y.; Hu, X.; Duan, W.; Zhang, Q.; Fu, L.; Xue, Q.-K.; Chen, X.; Ji, S.-H. Science 2016, 353 (6296), 274. (16) Zheng, C.; Yu, L.; Zhu, L.; Collins, J. L.; Kim, D.; Lou, Y.; Xu, C.; Li, M.; Wei, Z.; Zhang, Y.; Edmonds, M. T.; Li, S.; Seidel, J.; Zhu, Y.; Liu, J. Z.; Tang, W.-X.; Fuhrer, M. S. Sci. Adv. 2018, 4 (7), No. eaar7720. (17) Fei, R.; Kang, W.; Yang, L. Phys. Rev. Lett. 2016, 117 (9), 097601. (18) Hua, W.; Xiaofeng, Q. 2D Mater. 2017, 4 (1), 015042. (19) Wu, M.; Zeng, X. C. Nano Lett. 2016, 16 (5), 3236−3241. (20) Zhao, L.-D.; Lo, S.-H.; Zhang, Y.; Sun, H.; Tan, G.; Uher, C.; Wolverton, C.; Dravid, V. P.; Kanatzidis, M. G. Nature 2014, 508, 373. (21) Wang, E.; Lu, X.; Ding, S.; Yao, W.; Yan, M.; Wan, G.; Deng, K.; Wang, S.; Chen, G.; Ma, L.; Jung, J.; Fedorov, A. V.; Zhang, Y.; Zhang, G.; Zhou, S. Nat. Phys. 2016, 12, 1111. (22) Seyler, K. L.; Schaibley, J. R.; Gong, P.; Rivera, P.; Jones, A. M.; Wu, S.; Yan, J.; Mandrus, D. G.; Yao, W.; Xu, X. Nat. Nanotechnol. 2015, 10, 407. (23) Klein, J.; Wierzbowski, J.; Steinhoff, A.; Florian, M.; Rösner, M.; Heimbach, F.; Müller, K.; Jahnke, F.; Wehling, T. O.; Finley, J. J.; Kaniber, M. Nano Lett. 2017, 17 (1), 392−398. (24) Weitz, R. T.; Allen, M. T.; Feldman, B. E.; Martin, J.; Yacoby, A. Science 2010, 330 (6005), 812. (25) Zhang, Y.; Tang, T.-T.; Girit, C.; Hao, Z.; Martin, M. C.; Zettl, A.; Crommie, M. F.; Shen, Y. R.; Wang, F. Nature 2009, 459, 820. (26) Xia, J.; Li, X.-Z.; Huang, X.; Mao, N.; Zhu, D.-D.; Wang, L.; Xu, H.; Meng, X.-M. Nanoscale 2016, 8 (4), 2063−2070. (27) Kim, S. K.; Bhatia, R.; Kim, T.-H.; Seol, D.; Kim, J. H.; Kim, H.; Seung, W.; Kim, Y.; Lee, Y. H.; Kim, S.-W. Nano Energy 2016, 22, 483−489. (28) Skelton, J. M.; Burton, L. A.; Oba, F.; Walsh, A. J. Phys. Chem. C 2017, 121 (12), 6446−6454. (29) Kim, S.-u.; Duong, A.-T.; Cho, S.; Rhim, S. H.; Kim, J. Surf. Sci. 2016, 651, 5−9. (30) Gross, L.; Mohn, F.; Liljeroth, P.; Repp, J.; Giessibl, F. J.; Meyer, G. Science 2009, 324 (5933), 1428. (31) Zhang, Q.; Li, Y. J.; Wen, H. F.; Adachi, Y.; Miyazaki, M.; Sugawara, Y.; Xu, R.; Cheng, Z. H.; Brndiar, J.; Kantorovich, L.; Š tich, I. J. Am. Chem. Soc. 2018, 140 (46), 15668−15674. (32) Fei, R.; Li, W.; Li, J.; Yang, L. Appl. Phys. Lett. 2015, 107 (17), 173104. (33) Kong, X. Y.; Ding, Y.; Yang, R.; Wang, Z. L. Science 2004, 303 (5662), 1348. (34) Kong, X. Y.; Wang, Z. L. Nano Lett. 2003, 3 (12), 1625−1631. (35) Jeong, D. S.; Thomas, R.; Katiyar, R. S.; Scott, J. F.; Kohlstedt, H.; Petraru, A.; Hwang, C. S. Rep. Prog. Phys. 2012, 75 (7), 076502.

were collected in reflection geometry via an inverted microscope (Nikon Eclipse Ti). Emission from the sample was collected with the same objective lens and routed via a bundled optical fiber to a monochromator (Acton, Spectra Pro 2300i) coupled to a charge-coupled device (Princeton Instruments, Pixis 100). There was no observed sample damage by the laser during the measurements.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.9b01419.

■

Additional experimental details and Figures S1−S16 (PDF)

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected] (M.Y.). *E-mail: [email protected] (K.P.L.). ORCID

Yang Bao: 0000-0001-9868-4946 Yanpeng Liu: 0000-0002-5265-2735 Ibrahim Abdelwahab: 0000-0002-0107-5827 Wei Liu: 0000-0002-6812-6107 Xiaoxu Zhao: 0000-0001-9746-3770 Qing-Hua Xu: 0000-0002-4153-0767 Ming Yang: 0000-0002-0876-1221 Kian Ping Loh: 0000-0002-1491-743X Author Contributions §

Y.B. and P.S. contributed equally.

Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Park Systems and Bruker for access to PFM measurements and technical support. K. P. Loh wishes to acknowledge AME-IRG grant funded by Agency for Science and Technology, Singapore, Project number: A1983c0035, Scalable Growth of Ultrathin Ferroelectric Materials for Memory Technologies. M.Y would like to acknowledge the funding support by A*STAR 2D PHROS program (SERC 1527000012).

■

REFERENCES

(1) Dagdeviren, C.; Su, Y.; Joe, P.; Yona, R.; Liu, Y.; Kim, Y.-S.; Huang, Y.; Damadoran, A. R.; Xia, J.; Martin, L. W.; Huang, Y.; Rogers, J. A. Nat. Commun. 2014, 5, 4496. (2) Garcia, V.; Fusil, S.; Bouzehouane, K.; Enouz-Vedrenne, S.; Mathur, N. D.; Barthélémy, A.; Bibes, M. Nature 2009, 460, 81. (3) Yang, S. Y.; Seidel, J.; Byrnes, S. J.; Shafer, P.; Yang, C. H.; Rossell, M. D.; Yu, P.; Chu, Y. H.; Scott, J. F.; Ager, J. W., Iii; Martin, L. W.; Ramesh, R. Nat. Nanotechnol. 2010, 5, 143. (4) Scott, J. F. Science 2007, 315 (5814), 954. (5) Junquera, J.; Ghosez, P. Nature 2003, 422, 506. (6) Cui, C.; Xue, F.; Hu, W.-J.; Li, L.-J. npj 2D Mater. Appl. 2018, 2 (1), 18. (7) Belianinov, A.; He, Q.; Dziaugys, A.; Maksymovych, P.; Eliseev, E.; Borisevich, A.; Morozovska, A.; Banys, J.; Vysochanskii, Y.; Kalinin, S. V. Nano Lett. 2015, 15 (6), 3808−3814. H


Letter

Nano Letters (36) Singh, T. B.; Marjanović, N.; Stadler, P.; Auinger, M.; Matt, G. J.; Günes, S.; Sariciftci, N. S.; Schwödiauer, R.; Bauer, S. J. J. Appl. Phys. 2005, 97 (8), 083714. (37) Kitanaka, Y.; Noguchi, Y.; Miyayama, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 81 (9), 094114. (38) Hanakata, P. Z.; Carvalho, A.; Campbell, D. K.; Park, H. S. Phys. Rev. B: Condens. Matter Mater. Phys. 2016, 94 (3), 035304. (39) Gao, P.; Britson, J.; Jokisaari, J. R.; Nelson, C. T.; Baek, S.-H.; Wang, Y.; Eom, C.-B.; Chen, L.-Q.; Pan, X. Nat. Commun. 2013, 4, 2791. (40) Tian, Z.; Guo, C.; Zhao, M.; Li, R.; Xue, J. ACS Nano 2017, 11 (2), 2219−2226. (41) Hu, L.; Dalgleish, S.; Matsushita, M. M.; Yoshikawa, H.; Awaga, K. Nat. Commun. 2014, 5, 3279. (42) Lazareva, I.; Koval, Y.; Müller, P.; Müller, K.; Henkel, K.; Schmeisser, D. J. Appl. Phys. 2009, 105 (5), 054110. (43) Yang, C. H.; Seidel, J.; Kim, S. Y.; Rossen, P. B.; Yu, P.; Gajek, M.; Chu, Y. H.; Martin, L. W.; Holcomb, M. B.; He, Q.; Maksymovych, P.; Balke, N.; Kalinin, S. V.; Baddorf, A. P.; Basu, S. R.; Scullin, M. L.; Ramesh, R. Nat. Mater. 2009, 8, 485. (44) Yu, H.; Talukdar, D.; Xu, W.; Khurgin, J. B.; Xiong, Q. Nano Lett. 2015, 15 (8), 5653−5657. (45) Denev, S. A.; Lummen, T. T. A.; Barnes, E.; Kumar, A.; Gopalan, V. J. Am. Ceram. Soc. 2011, 94 (9), 2699−2727. (46) Vanderbilt, D.; King-Smith, R. D. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 48 (7), 4442−4455.

I


Gate-Tunable In-Plane Ferroelectricity in Few-Layer SnS

Recommend Documents