The emergence of vdW synthetic ferroelectrics: fundamentals and applications

[Audio] My name is Professor Andrew Wee from the Department of Physics of the National University of Singapore. I want to firstly thank the organisers for giving me this opportunity to give this invited talk. I am going to use the text to speech function with an artificial voice, because I have a speech problem due to a medical condition. So please bear with me, especially when I try to speak in the Q and A section. My talk is entitled "The emergence of van der Waals synthetic ferroelectrics: fundamentals and applications"..

[Audio] Before I begin my talk proper, allow me to briefly introduce my NUS Surface Science Lab. We have had a series of ultrahigh (UHV) systems over the years, but currently our most used instruments are the non contact AFM, LTSTM, MBE and XPS..

[Audio] We also have access to a Class 1000 and 100 800 meter square clean room facility; which has Micro- and nanofabrication facilities for graphene and other 2D materials. This cleanroom is part of the Centre for Advanced 2D Materials (CA2DM) Additionally, we also use the soft X ray beamline at the Singapore Synchrotron Light Source (SSLS). Here, we can also do high resolution photoelectron spectroscopy (PES), X ray absorption spectroscopy, X ray magnetic circular dichroism (XMCD) and so on..

[Audio] But why are we now studying 2D materials, and how are they different from bulk materials? Removal of van der Waals interactions An increase in the ratio of surface area-to-volume ratio Confinement of electrons in a plane Band structure diagram of (left) bulk and (right) monolayer MoS2 showing the crossover from indirect to direct bandgap accompanied by a widening of the bandgap. Another effect of dimensional confinement is reduced dielectric screening between electrons and holes in semiconductors. When there is less material to screen the electric field, there will be an increase in Coulomb interaction and more strongly-bound excitons – making them more stable than excitons found in bulk materials. If the excitons are confined in a plane that is thinner than their Bohr radius (as is the case for many 2D semiconductors), quantum confinement will result in an increase in their energy compared to bulk excitons, changing the wavelength of light they absorb and emit. And Surface Science techniques are ideal for 2D materials since they are all surface and no bulk!.

[Audio] Like many other groups, we have mainly been studying transition metal di chalcogenides (T M D Cs) over the past few years. The papers above list some of our work on: 2D TMDCs & defects; Molecule-2D TMDC interfaces; 2D TMDC optical properties; and 2D TMDC magnetic properties..

[Audio] In one of our projects, we are working on P t S e2 which has potential for post silicon electronics. This is due to its high mobility, and layer dependent tunable band gap..

[Audio] This is a paper we recently published on P t S e2, by Zhang Let et al..

[Audio] I shall now move on to my talk, which is about 2D materials apart from 2D TMDCs. The scope of my talk is as follows: I will first give an Introduction to ferroelectrics I will then present Two-dimensional ferroelectricity in a single-element bismuth monolayer, done by then Research fellow Dr Gou Jian and collaborators. This work was published in the May 2023 issue of Nature. Then I will present our work on Ultrathin quantum light source with van der Waals Niobium Oxygen Dichloride crystal, done by Research Fellow Dr Guo Qiangbin and collaborators. This work was published in the Jan 2023 issue of Nature..

[Audio] The study of ferroelectrics is now more than a century old. These materials typically, upon undergoing a phase transition as the temperature decreases below a critical value, possess a permanent and hysteretic electrically switchable polarization. As a result, ferroelectrics have already had an impact on diverse applications ranging from actuators, to electronic memories The ground state of a ferroelectric is polar, implying that a prerequisite for designing new ferroelectrics should be to start from stable materials that have polar structures and so permanent polarization. However, at the nanoscale it is possible to generate ferroelectricity in meta stable materials e.g. 10 nm hafnia. As shown in the figure on the left, it's main characteristics are: It is Inversion symmetry broken It has Polarization It is Switchable or reversible.

[Audio] Previous studies have focused mostly on ferroelectric materials with 3D lattices. Recent developments in van der Waals (vdW) ferroelectrics with 2D lattices have introduced a new paradigm to the field. Their layered structure enables stable monolayer and few-layer samples to be prepared that provide an excellent platform for studying the effects of reduced lattice dimensionality on long-range ferroelectric order..

[Audio] For example: Fei et al. (PRL 2016) states that Ferroelectricity usually fades away as materials are thinned down below a critical value. But They show that the unique ionic-potential anharmonicity can induce spontaneous in-plane electrical polarization and ferroelectricity in monolayer group-IV monochalcogenides MX (M=G e, S n; X=S, S e). An effective Hamiltonian has been successfully extracted from the parametrized energy space, making it possible to study the ferroelectric phase transitions in a single-atom layer. The ferroelectricity in these materials is found to be robust and the corresponding Curie temperatures are higher than room temperature, making them promising for realizing ultrathin ferroelectric devices of broad interest. Chang et al. (Science 2016) stated that As a ferroelectric material becomes thinner, the temperature below which it develops its permanent electrical polarization usually decreases. However, they fabricated high-quality thin films of Tin Telluride that had a considerably higher transition temperature than that of the material in bulk. This finding may enable the miniaturization of ferroelectric devices..

[Audio] This Nature Materials perspective by Wang et al. "Towards two-dimensional van der Waals ferroelectrics" states that he discovery of ferroelectricity in two-dimensional (2D) van der Waals (v d W) materials has brought important functionalities to the 2D materials family, and may trigger a revolution in next-generation nanoelectronics and spintronics. In this Perspective, they briefly review recent progress in the field of 2D vdW ferroelectrics, focusing on the mechanisms that drive spontaneous polarization in 2D systems, unique properties brought about by the reduced lattice dimensionality and promising applications of 2D vdW ferroelectrics..

[Audio] The Origin of spontaneous polarization in 2D van der Waals ferroelectric systems include: Ionic-displacement-induced polarization In perovskite ferroelectric oxides such as Ba Ti O3, hybridization between the Oxygen 2p and Titanium 3d states pushes the Boron-site ions away from the geometric centres of the oxygen octahedra, giving rise to spontaneous polarization Polarization from polar molecular groups In certain molecular crystals and polymers, such as polyvinylidene fluoride (P V D F), the molecular units possess permanent dipole moments whose spontaneous ordering produces the polarization. Charge-redistribution-induced polarization In some 2D v d W materials, theory suggests that interlayer charge redistribution through hybridization between the occupied states of one layer and the unoccupied states of the neighbouring layer could induce out-of-plane electric dipoles (e.g. W T e2 bi layer h-BN, W S e2, M o S e2, W S2 and M o S2. Spin-driven polarization Ferroelectricity can also be induced by certain kinds of long-range magnetic order, as in type II multiferroics; an example is orthorhombic Tb Mn O3..

[Audio] The Applications of 2D van der Waals ferroelectric materials include: 2D vdW ferroelectric/multiferroic tunnel junctions In a ferroelectric tunnel junction (FTJ) based on traditional 3D ferroelectrics, the switchable polarization modulates the tunnelling energy barrier, resulting in tunneling electroresistance ratio (TER) values of about 10 to 10E6. 2D vdW ferroelectric transistors A ferroelectric field-effect transistor (Fe-FET) makes use of the spontaneous polarization to modulate the conductance of a semiconductive channel. 2D vdW ferroelectric-based spintronic devices Spin–orbit coupling (SOC) describes the interaction between the spin and orbit angular momentum of an electron. It has been shown that strong Rashba-type SOC exists in 2D vdW ferroelectrics such as B i T e I and Ge T e. 2D vdW ferroelectric-based topological devices The modern theory of ferroelectricity based on the Berry phase not only enables accurate calculation of the electric polarization but also bridges the gap between ferroelectricity and topology, offering a new paradigm for the electric-field control of exotic quantum geometrical phenomena..

[Audio] Some Challenges in 2D ferroelectric research include: 2D vdW multi ferroelectrics; Topological spin polar structures in 2D systems; Spin based devices in 2D systems; Polarization switching kinetics and reliabilities; Bulk photovoltaic effect in 2D vdW ferroelectrics; Large scale 2D vdW ferroelectric thim film preparation..

[Audio] I will now present our work on Two-dimensional ferroelectricity in a single-element bismuth monolayer, done mainly by then Dr Guo Jian, who is now an Assistant Professsor At Zhejiang University. This work was published in the May 2023 issue of Nature and performed in our ultrahigh vacuum (UHV) non contact AFM system..

[Audio] In this paper, we report the first observation of a single-element ferroelectric state in a black phosphorus-like bismuth layer, in which the ordered charge transfer and the regular atom distortion between sublattices happen simultaneously. Instead of a homogenous orbital configuration that ordinarily occurs in elementary substances, we found the Bismuth atoms in a black phosphorous-like Bismuth monolayer maintain a weak and anisotropic sp orbital hybridization, giving rise to the inversion-symmetry-broken buckled structure accompanied with charge redistribution in the unit cell. As a result, the in-plane electric polarization emerges in the Bi monolayer. Using the in-plane electric field produced by scanning probe microscopy, ferroelectric switching is further visualized experimentally. This emergent single-element ferroelectricity broadens the mechanism of ferroelectrics and may enrich the applications of ferroelectronics in the future. The figure below shows an artist impression of the buckled Bismuth monolayer..

[Audio] This is our model of Spontaneous symmetry breaking in single BP-Bismuth layer. The monolayer α-phase Bi has a lattice structure similar to black phosphorous. (BP-B i) thereafter. Owing to the ultra-large atomic number, Bi has a weak hybridization between the 6s and 6p orbitals so that it features partial sp2 character other than the homogenous tetrahedral sp3 configuration that exists in the black phosphorus. This brings in a small buckling (Δh) between the neighbouring sublattices with the loss of centrosymmetry. This is illustrated in Fig. 1a. The breaking of inversion symmetry allows BP-B i to adopt two domain states, either the Δh = d0 or Δh = minus d0 state. Our first-principles calculations in Fig. 1b reveal the two states can be switched to each other by crossing a small energy barrier of 43 meV per unit cell..

[Audio] In Fig. 1c, Δh adopts the right minimum of the double-well potential, and the valence band and conduction band at the Γ point are mainly contributed by the pz orbitals of A and B sublattice, respectively. When the Fermi level crosses the band gap, the valence pz orbitals at the A sublattice are fully occupied, and the pz orbitals at the B lattice are empty. In Fig. 1d, we see that in real space, this corresponds to electron transfer from sublattice B to sublattice A, leading to a spontaneously polarized character..

[Audio] Experimentally, we grew BP-B i on H O P G. Figure 1e shows a STM image of the monolayer BP-B i with some second layer nanoribbons along the [01] direction. The non contact AFM measurement indicates two different states in two neighbouring domains separated by a domain wall (Fig. 1f and g). We performed force spectroscopy measurements (the Δf(z) spectra) in Fig. 1i to find out the relative distance between the Bismuth atoms and tip apex. At the constant-height mode, the measured height difference of the turning points of the Δf(z) spectra on sublattice A and B give Δh0 = minus Δh1 = 40 pm The d I over d V spectrum at the domain wall shows the pz bands (E 1 and E ii) moving to a higher binding energy compared to the normal domain position, and a sharp peak occurs in the band gap (Fig. 1h), which will be elaborated in detail later..

[Audio] Next, we look in more detail at the In-plane polarization: Figure 2b shows the d I over d V cascade along the red dashed arrow (across an ABABA lattice) in Fig. 2a, revealing two traces of peaks. The valence band peak E 1 is clearly strong at the A sublattice (points 1, 3, 5) but weak at the B sublattice (points 2, 4), whereas the conduction band peak E ii shows the opposite behaviour. At the 2D scale, the d I over dV mapping of the valence band in Fig. 2c and conduction band in Fig. 2d show the same feature: filled pz orbitals localize at the A sublattice, while the empty pz orbitals localize at the B sublattice, confirming the predicted electron transfer from B to A..

[Audio] Typical Kelvin probe force microscopy (KPFM) measurement in Figure 2e implemented by recording the frequency shift as a function of the sample bias voltage. The electrostatic force caused by the contact potential difference (CPD) between the sample and tip changes by sweeping the bias voltage. When the CPD is totally compensated by the bias (VCPD = V*), the electrostatic force reaches the minimum, corresponding to the parabolic apex in the frequency shift curve in Fig. 2e. AFM was performed simultaneously to determine the atomic structure at the same area. See Fig. 2f. It is obvious that the two sublattices A and B have different surface potentials. The calculated electrostatic potential is shown in Fig. 2 H, which reproduces the experimental LCPD map and indicates the electron enrichment at the topmost A sublattice. By recording the bias voltage V* site by site in a lateral grid, the surface potential is mapped out and shown in Fig. 2 g. Ultimately, on the basis of the above dI over dV and LCPD measurement combined with the in-plane distorted atom structure, the in-plane polarization can be confirmed..

[Audio] We use the in-plane component of electric field from the conductive STM/AFM tip to switch the polarization of small ferroelectric domains close to the tip. See Fig 3f. During the sample bias sweeping at a specific tip height, the IV spectra are recorded as shown in Fig. 3d. While sweeping the voltage back (from positive to negative), a larger hysteretic current is maintained with a substantial conductance emerging at around 0.2 V in the gap (red series curves). According to the d I over d V curves of the domain wall (Fig. 1 H), the large in-gap current indicates the domain wall has been moved to the tip position during the application of the positive bias. When the bias voltage reaches a negative value V SW, the current jumps to the original level, suggesting the domain wall is moved back. The AFM images in Fig. 3b and c after the forwards and backwards voltage sweeping demonstrate the movement of the domain wall directly. Figure 3c shows the domain wall movement to the left side with a distance of four lattice periods after the forwards manipulation, and Fig. 3b shows the domain wall moving back to the original position after changing the bias backwards to below V SW..

[Audio] Let us look at the 180° domain walls. Besides the 180° head-to-head domain wall in the monolayer BP-B i (Figs. 1e, 3a and 4a), we also observe the conjugated 180° tail-to-tail domain wall (Fig. 4c). The AFM measurements show that the 180° head-to-head domain wall has a wall width of roughly 15 Angstroms (three unit cells, shadow area in Fig. 4e), whereas the 180° tail-to-tail domain wall has a wider width of roughly 56 Angstroms (12 unit cells, shadow area in Fig. 4g). A close determination of the band bending by extracting Ei peak and measuring the LCPD by KPFM show the same bending profile and a total band bending of about 170 meV (Fig. 4f). When turning to the tail-to-tail domain wall, the d I over dV spectra, however, show the incongruous band movement in the valence band and conduction band (Fig. 4d). The same band bending assessment methods by the Ei analysis and LCPD measurement in Fig. 4h suggest a smaller but identical bending direction to the head-to-head domain wall (Fig. 4f). Figure 4 i–l shows the calculated wall width and surface potential profile in the 180° head-to-head and tail-to-tail domain walls..

[Audio] In summary, with the combined use of STM and AFM methods, we: Confirmed the non centrosymmetric structure; Verified the in-plane polarization; and Visualized the ferroelectric switching..

[Audio] We also visualized the Ferroelectric domain walls. We studied the differences between the Head-to-head domain wall (H) and tail-to-tail domain wall (T). We found that the width of tail-to-tail domain wall is larger than the head-to-head wall.

[Audio] In the final part of this talk, we will look at Ultrathin quantum light source with another 2D van der Waals Niobium oxide dichloride (Nb O Cl2) crystal. The work was done by Research Fellow Dr Guo Qiang bin and collaborators, especially at the N U S and the C A S Key Laboratory of Quantum Information, USTC, Hefei. This work was published in the Jan 2023 issue of Nature..

[Audio] In this paper, we report a van der Waals crystal, niobium oxide dichloride (Nb O Cl2), featuring vanishing interlayer electronic coupling and monolayer-like excitonic behaviour in the bulk form, along with a scalable second-harmonic generation intensity of up to three orders higher than that in monolayer W S2. Notably, the strong second-order nonlinearity enables correlated parametric photon pair generation, through a spontaneous parametric down-conversion (SPDC) process, in flakes as thin as about 46 nm. To our knowledge, this is the first SPDC source unambiguously demonstrated in two-dimensional layered materials, and the thinnest SPDC source ever reported..

[Audio] But Why 2D T M D Cs are unsuitable? Almost all traditional entangled photon sources via SPDC process are enabled by conventional chi2 crystals like Beta barium borate and Lithium Niobate bulk crystals. 2D TMDCs have extremely high chi2 coefficients. However, the absolute chi2 efficiency of monolayer TMDs are too low to enable practical nonlinear optical device applications because the light-matter interaction length is very short due to the thinness of a monolayer. In other words, interacting substrates prevent stacking. The chi2 efficiency in such TMDC materials cannot be enhanced by increasing the thickness by stacking up many monolayers. In TMDCs the chi2 nonlinearity only exists in odd-number layers due to symmetry. TMDCs usually stack in the 2H polytype by alternate orientation of each monolayer along the c axis Interlayer electronic coupling when stacking up monolayers will greatly degrade the resulting material's chi2 nonlinearity..

[Audio] In this work, We discovered a van der Waals material niobium oxide dichloride Nb O Cl2 that can retain the monolayer properties including the chi2 nonlinearity when stacking up monolayers. In other words it has weak interlayer electronic coupling. Besides, this material always shows a non-centrosymmetric structure regardless of layer number, which is a premise for chi2 non linearity. As a result, this material shows extremely strong and scalable chi2 efficiency when stacking monolayers up. We demonstrate the first van der Waals material-based SPDC source and also the world's thinnest SPDC source (less than 50 nm) ever reported. This has the potential for constructing ultracompact on-chip quantum light sources for integrated quantum circuits..

[Audio] Nb O Cl2 crystallizes in the C2 space group with a van der Waals stacking behaviour along the a axis. See Fig. 1a and b. It has an interlayer distance of about 0.65 nm. Nb atoms exhibit a 1D Peierls distortion, resulting in a polarization along the b axis and two alternating unequal Nb–Nb distances (that is, L1 ≠ L2) along the c axis (see Fig. 1a), which can be directly observed by atomic-resolved annular dark-field scanning transmission electron microscopy (ADF-STEM) and confirmed by the corresponding simulated images. See Fig. 1c to e. Nb O Cl2 crystals can be easily exfoliated by the normal mechanical exfoliation method. Monolayer and few-layer flakes are regularly obtained as well as large rectangular thin flakes, with lateral size of up to 102 micrometers with sharp edges. This is indicative of weak interlayer interaction and strong intralayer crystallographic anisotropy..

[Audio] The layer-dependent electronic structures were characterized by STEM-based valence EELS, which is a powerful tool for studying optical excitations in nanostructures with ultrahigh spatial and energy resolutions. Figure 2a shows the normalized EELS results for different layers, in which we can see that the optical excitation starts from around1.6 eV (see enlargements in Fig. 2b) but remains at a low intensity up to 3 eV. Significant optical excitation appears only after 3 eV and comes to its first peak at around 4 eV. It is interesting to note that the optical excitation onset energy for different layers exhibits almost no energy shift (within the experimental energy resolution of 0.021 eV); that is, it is insensitive to thickness or layer number. Fig. 2c shows a comparison with other typical 2D layered materials, especially TMDCs and black phosphorus whose bandgaps evolve significantly with layer number owing to considerable interlayer coupling..

[Audio] As shown in Fig. 2d and e, the calculated quasiparticle bandgap values for the monolayer (2.27 e V) and the bulk (2.21 e V) form are also very close, which together with the Heyd–Scuseria–Ernzerhof calculations, indicates a weak interlayer electronic coupling. The calculations using GW approximation and the Bethe–Salpeter equation unveil a considerable excitonic effect in both the monolayer and bulk forms, with a high exciton binding energy of about 0.8 eV for the monolayer form and about 0.76 eV for the bulk form. The calculated optical absorption onset at about 1.5 eV for both the monolayer and bulk forms is consistent with both experimental EELS and optical absorption data. It is noteworthy that both the quasiparticle bandgap and exciton binding energy are insensitive to layer number, revealing that a monolayer-like and considerable excitonic behaviour survived in the bulk crystal. This is in sharp contrast to excitons in other 2D layered materials represented by TMDCs in which significant excitonic effects exist only in monolayers. The dispersionless valence band, resulting from the structural Peierls distortion, denotes highly localized electronic states and contributes to the large exciton binding energy. It is noted that the relatively weak optical absorption around the exciton and quasiparticle energies is attributed to the very localized Nb 4d orbitals in the valence band maximum that contribute little to optical excitation..

[Audio] To unveil a more thorough picture of the weak interlayer electronic coupling, we calculated the interlayer charge density, as shown in Fig. 2 G. The electrons are mainly localized in the intralayer (mostly on Nb and O atoms; with negligible distribution in the interlayer region, implying mainly in-plane bonding. This can be further evidenced by the interlayer differential charge density (Fig. 2 H) that is calculated by assembling the bulk system from isolated monolayers and reveals a charge redistribution process during interlayer coupling. In addition, a negligible charge redistribution can be found in the interlayer region, indicating almost no covalency in the out-of-plane direction. By contrast, significant charge distribution can be found in the interlayer region of TMDCs and black phosphorus due to stronger interlayer bonding. This rather weak interlayer coupling character in Nb O Cl2 can be understood as follows. After grabbing an electron from the Nb atom, the p shell of the Cl atom is complete and becomes inert. Consequently, the interlayer interaction (bonding) is weak as Nb and O atoms are sandwiched by Cl atoms. The ionic bond of Nb–Cl is different from the Mo–S bond in MoS2, which is more covalent. Therefore, the strong ionicity of Nb–Cl bonds together with the structural Peierls distortion led to the extremely weak interlayer electronic coupling in Nb O Cl2. Therefore, both the quasiparticle bandgap and exciton binding energy are insensitive to layer number, revealing that a monolayer-like and considerable excitonic behaviour survived in the bulk NiOCl2. This is in sharp contrast to excitons in other 2D layered in which significant excitonic effects exist only in monolayers.

[Audio] The second-order NLO response of Nb O Cl2 was investigated by SHG experiments under a back-reflection configuration, as shown in Fig. 3a. Strong emission signals at half the corresponding excitation wavelengths were observed in Fig. 3b, with a quadratic excitation power dependence, as shown in Fig. 3c..

[Audio] Fig. 3d shows the SHG response is also highly in-plane anisotropic with a maximum response along the crystal polarization direction (b axis), which can be well explained and fitted on the basis of the crystal symmetry analysis. This is unlike T M D Cs whose overall SHG response shows no polarization dependence. Fig. 3e and f show a positive scaling of SHG intensity with layer number, in a quadratic behavior, within the penetration depth and coherence length..

[Audio] The giant second-order optical nonlinearity in 2D layered Nb O Cl2 stimulates us to explore it for a quantum light source. Optical quantum information processing involves the coding, manipulation and detecting of entangled photons. Currently, the commonly used and mature method for generating entangled photons is based on a nonlinear optical process called spontaneous parametric down-conversion (S P D C) process in a nonlinear crystal (denoted chi2 crystal) in which a higher energy photon (pump photon) is fissioned into a pair of lower energy photons called signal and idler photons..

[Audio] As illustrated in Fig. 4a and b, the SPDC process was first checked on a subwavelength Nb O Cl2 flake (thickness of about 150 nm, exfoliated on a transparent sapphire substrate) with a continuous-wave laser at 404 nm. Photon pair generation was recorded by registering photon coincidences between two detectors. The normalized second-order correlation functions g (2) (τ) measured on the sample and blank substrate are presented in Fig. 4c and d, respectively. A peak at zero time delay means simultaneous arrival of one photon at each detector and thus is a signature of correlated photon pair generation. An obvious two-photon correlation peak with a peak-to-background ratio well above 2 at zero time delay (g (2) (0)) was observed in the sample but not in the substrate, unambiguously demonstrating correlated photon pair generation through the SPDC process in the sample. In addition, the g (2) (0) value under different pumping power was also measured and exhibits an inverse pump power dependence (Fig. 4e), further evidencing a photon pair generation process..

[Audio] The polarization-dependent response in Fig. 4f indicates the pump, signal and idler photons are all polarized along the crystallographic b axis. The pump-power-dependent photon pair coincidence rate was calculated from the data in Fig. 4e and follows a linear scaling relation as in Fig. 4 G, being a typical feature of the SPDC process..

[Audio] In summary, in this talk I have spoken about 2 aspects of 2D materials: Two-dimensional ferroelectricity in a single-element bismuth monolayer, and Ultrathin quantum light source with van der Waals Nb O Cl2 crystal.

[Audio] Finally, I want to acknowledge all the collaborators in these projects, especially the first authors Doctors Gou Jian and Guo Qiang bin. I want to also thank our many collaborators especially from the Institute of Physics. Chinese academy of Sciences, and Zhejiang University. I must acknowledge our funding agencies from the National Research Foundation and Ministry of Education of Singapore, and the National Natural Science Foundation of China. And finally, I want to thank you all for listening..