In the ever-changing field of memory technologies, ferroelectric memory transistors are innovative devices with features that can potentially improve data processing and storage hardware systems. Such features include power efficiency, non-volatility, high endurance, non-destructive readout, and faster read/write speed, which are made possible through the special properties of ferroelectric materials. Ferroelectric memory technology can be quickly adopted across different industries by being easily integrated into current semiconductor manufacturing processes, such as CMOS technology, creating opportunities for future scaled devices.

An arguable limitation of traditional ferroelectric materials, however, is that their polarization tends to vanish at the nanoscale regime. Excitingly, a recently discovered capability of two-dimensional ferroelectric materials has been revealed such that ferroelectric domains can move or slide along the van der Waals interface. This phenomenon is called sliding ferroelectricity, which introduces the possibility of dynamically controlling and manipulating the distribution of ferroelectric domains, leading to a novel prototype for memory design.

Such material with interesting characteristics deserves further exploration. Hence, in our featured study, we have demonstrated a polarity-switchable epitaxial rhombohedral (3R)-stacked MoS₂ as a sliding ferroelectric channel on ferroelectric semiconductor field-effect transistors (FeS-FETs). The developed transistor device is portrayed in Figure 1a. Here, we used a back-gated structure consisting of a 30-nm thick SiO₂ and a 10-nm thick hBN layers, forming a double-layer dielectric, and an ST-3R MoS₂ ferroelectric channel.

Sliding ferroelectricity on the FeS-FET device

An important stage in fabricating the FeS-FET device is setting up the 3R-MoS₂ film channel into a switchable ferroelectric material during the chemical vapor deposition (CVD) growth process. Formation of domain boundaries in 3R-MoS₂ films is necessary to possess the ability to switch polarized domains; however, this is rare in most epitaxial 3R MoS₂ films. In the paper, we featured a strategy to increase the chance of domain boundaries appearing in the material, that is, through forming regions of unstable AA phase by performing a growth process with relative low-temperature tempering under molybdenum-deficient conditions. By doing so, an eventual decomposition of some portions of the AA phase region into phases with lower stacking energies, namely AB (polarization up) and BA (polarization down) domains, occur, consequently forming the desired AB/BA domains embedded in the AA (unpolarized) matrix (Figure 1b). Such structure exhibits shear transformation (ST)-induced dislocations near domain boundaries due to the contrast between the misalignment-free AB/BA stacked domains and the highly dislocated AA phase (Figure 1c). The presence of these ST-induced dislocations in the 3R-MoS₂ material allows it to possess sliding ferroelectric properties, which can be initiated by an external electric field. Through piezoelectric force microscopy (PFM) and polarization-electric field (P-E) measurements, the switchable electric polarization of the device was verified at both localized and device levels, respectively.

The low coercive field of the ferroelectric device

Coercive field is the minimum field across the ferroelectric material, in our case, the 3R MoS₂, that can trigger the movement of domain walls, i.e., trigger polarization switching. By employing the ferroelectric sample in a metal-insulator-semiconductor-metal (MISM) architecture, since the semiconducting nature of the ferroelectric material needs to be insulated to effectively facilitate measurement, a polarization switching (coercive) field of 0.036 V nm^-1 was measured. We note that this recorded coercive field is lower than typical thin-film ferroelectrics and twisted 3R WSe₂ bilayers, which contain pinning points that constrain the sliding of domain boundaries. We, therefore, view this as a result of the lack of structural pinning points in the ST-3R MoS₂ material due to the parallel-patterned ST-induced dislocations, eliminating a limitation in the domain walls’ movements.

Challenges when addressing sliding ferroelectricity

We note that the polarization strength of 0.56 μC cm⁻² in bilayer 3R MoS₂ is relatively weak, 1~2 orders of magnitudes lower than that of conventional ferroelectric materials. It is also noteworthy that the polarization of sliding ferroelectricity is at a similar level to intrinsic dielectric polarization. To exclusively study the behavior of sliding ferroelectricity switching, we adopt techniques that have been viewed as the most reliable ways to investigate ferroelectric materials, including P-E loops, PFM, and positive-up negative-down (PUND) methods. Based on the combined results, the ferroelectric switching behavior of our ST-3R MoS₂ is confirmed, alongside precise estimation of coercive fields and polarization strength.

The FeS-FET device performance

Subsequent current-voltage measurements reveal the device’s switchability, reliability, and repeatability. In particular, gate voltage levels of at least ~ 2 to 3 V are found to be enough to flip the polarization of the ferroelectric channel from one state to the other. Its repeatable memory window persists even after 50 cycles of hysteresis measurements, and its capability to retain memory states or polarization is expected to last in the order of years even after removing the electric field at the gate. Such long retention we regard to have been caused by the straight arrangement of the dislocations, enabling the domain boundaries to stay fixed for an extended period.

Moving Forward

The continuous demand for enhanced technology pushes researchers to design and develop new devices that can meet the requirements of modern-day applications. Desirable devices possess smaller sizes with faster response, power efficiency, and reliability. That’s when sliding ferroelectric materials can come into play, particularly in memory device implementation. This work started with a dedicatedly designed synthesis process, verification of the physical origin behind ferroelectricity switching, and eventually, demonstration of the prototype sliding ferroelectric memory. Our ST-3R MoS₂ demonstrates sliding ferroelectricity, banking on shear transformation-induced dislocations. With a thickness of about two atomic layers, the device is a promising component that can fit into the requirements of state-of-the-art CMOS technology, e.g., sub-3 nm nodes, reducing short-channel effects and cutting down leakage current in the OFF state. Given our findings and related advancements in two-dimensional, TMD-based devices, we envision a wider application of 2D materials in electronic systems, e.g., data processing and storage systems, in the future, encompassing memory units and logic circuits and paving the way for all-2D-material-based nanoelectronics.

For more details regarding our research, we encourage everyone to refer to our article.