Memory and Synaptic Devices Based on Emerging 2D Ferroelectricity

Filamentary and interface switching of CMOS-compatible Ta2O5 memristor for non-volatile memory and synaptic devices

Neuro-inspired deep learning algorithms have shown promising futures in artificial intelligence. Despite the remarkable progress in software-based neural networks, the traditional von-Neumann hardware architecture has suffered from limited energy efficiency while facing unprecedented large amounts of data. To meet the performance requirements of neuro-inspired computing, large-scale vector-matrix multiplication is preferred to be performed in situ, namely compute-in-memory. Non-volatile memory devices with different materials have been proposed for weight storage as synaptic devices. Among them, HfO2-based ferroelectric devices have attracted great attention because of their low energy consumption, good complementary-metal-oxide-semiconductor (CMOS) compatibility and multi-bit per cell potential. In this review, recent trends and prospects of the ferroelectric synaptic devices are surveyed. First, we present the three-terminal synaptic devices based on the ferroelectric field effect transistor (FeFET), and discuss the switching physics of the intermediate states, the back-end-of-line integration and the 3D NAND architecture design. Then, we introduce a hybrid precision synapse concept that leverages the volatile charges on the gate capacitor of the FeFET and the non-volatile polarization on the gate dielectric of the FeFET. Lastly, we review two-terminal synaptic devices using the ferroelectric tunnel junction (FTJ) and ferroelectric capacitor (FeCAP). The design margins of the crossbar array with FTJ and FeCAP analyzed.

Two-dimensional (2D) ferroelectric materials have attracted increasing interest due to meeting the requirements of integration, miniaturization, and multifunction of devices. However, the exploration of intrinsic 2D ferroelectric materials is still in the early stage, for which the related reports are still limited, especially fewer ones prepared by chemical vapor deposition (CVD). Here, the ultrathin metal-tellurium-oxyhalide Cd7Te7Cl8O17 (CTCO) flakes as thin as 3.8 nm are realized via the selenium-induced selective-bonding CVD method. The growth mechanism has been confirmed by experiments and theoretical calculations, which can be ascribed to the induction of selective bonding of a hydrogen atom in H2O molecules by the introduction of selenium, leading to the generation of strong oxidants. Excitingly, switchable out-of-plane ferroelectric polarization was observed in CTCO flakes down to 6 nm at room temperature, which may be caused by mobile Cl vacancies. This work has implications for the synthesis and applications of 2D ferroelectric materials.

Two-dimensional (2D) ferroelectric materials have emerged as significant platforms for multi-functional three-dimensional (3D) integrated electronic devices. Among 2D ferroelectric materials, ferro-ionic CuInP2S6 has the potential to achieve the versatile advances in neuromorphic computing systems due to its phase tunability and ferro-ionic characteristics. As CuInP2S6 exhibits a ferroelectric phase with insulating properties at room temperature, the external temperature and electrical field should be required to activate the ferro-ionic conduction. Nevertheless, such external conditions inevitably facilitate stochastic ionic conduction, which completely limits the practical applications of 2D ferro-ionic materials. Herein, free-standing 2D ferroelectric heterostructure is mechanically manipulated for nano-confined conductive filaments growth in free-standing 2D ferro-ionic memristor. The ultra-high mechanical bending is selectively facilitated at the free-standing area to spatially activate the ferro-ionic conduction, which allows the deterministic local positioning of Cu+ ion transport. According to the local flexoelectric engineering, 5.76×102-fold increased maximum current is observed within vertical shear strain 720 nN, which is theoretically supported by the 3D flexoelectric simulation. In conclusion, we envision that our universal free-standing platform can provide the extendable geometric solution for ultra-efficient self-powered system and reliable neuromorphic device.

9 - Novel photoelectroactive memories and neuromorphic devices based on nanomaterials

Due to the advantages of good scalability, flexibility, low cost, ease of processing, 3D-stacking capability, and large capacity for data storage, polymer-based resistive memories have been a promising alternative or supplementary devices to conventional inorganic semiconductor-based memory technology, and attracted significant scientific interest as a new and promising research field. In this review, we first introduced the general characteristics of the device structures and fabrication, memory effects, switching mechanisms, and effects of electrodes on memory properties associated with polymer-based resistive memory devices. Subsequently, the research progress concerning the use of single polymers or polymer composites as active materials for resistive memory devices has been summarized and discussed. In particular, we consider a rational approach to their design and discuss how to realize the excellent memory devices and understand the memory mechanisms. Finally, the current challenges and several possible future research directions in this field have also been discussed.

Programmable digital polymer memories

ABSTRACTMFMPOS (Metal, Ferroelectrics, Metal, Polysilicon, Oxide, and Silicon) one-transistor (1T) ferroelectric memory devices have been fabricated. However, the yield of 1T-memory devices is lower. We find that the main problems of 1T MFMPOS memory devices are shorts, opens, no memory window, smaller memory windows and blank. In order to solve these problems, we studied the reasons resulted in the problems. Then, the integration processes for one transistor memory device were optimized. Fabrication of nMOSFET 1T memory devices starts with shallow trench isolation (STI) on p-type Si. A gate oxide is thermally grown after p-well implantation. Phosphorus ions were implanted after polysilicon gate definition for the formation of self-aligned source, drain, and n-type floating gate. A damascene process using MOCVD PGO deposition and chemical mechanical polishing (CMP) were used to avoid etching damage. Electrodes for the ferroelectric capacitor, i.e., the floating Ir bottom electrode and Pt top electrode, are deposited by E-Beam evaporation. The area ratio of the top and floating gate electrodes is 1:1. After inter-level dielectric (ILD) deposition, contact etching stops on Pt at gate and on Si at source/drain (S/D) without difficulty because of high etch rate selectively to the Pt. Finally, the high quality 1T memory devices have been made. The one-transistor memory devices showed memory windows around 2 – 3V. The memory windows are almost saturated from operation voltage of 3V. The ratios of “on” state current to the “off” state current are closed to 8 – 9 orders. The one-transistor memory devices also show a very good memory characteristics and retention properties.

Hafnium oxide (HfO2 ) is one of the mature high-k dielectrics that has been standing strong in the memory arena over the last two decades. Its dielectric properties have been researched rigorously for the development of flash memory devices. In this review, the application of HfO2 in two main emerging nonvolatile memory technologies is surveyed, namely resistive random access memory and ferroelectric memory. How the properties of HfO2 equip the former to achieve superlative performance with high-speed reliable switching, excellent endurance, and retention is discussed. The parameters to control HfO2 domains are further discussed, which can unleash the ferroelectric properties in memory applications. Finally, the prospect of HfO2 materials in emerging applications, such as high-density memory and neuromorphic devices are examined, and the various challenges of HfO2 -based resistive random access memory and ferroelectric memory devices are addressed with a future outlook.

In recent years, polymer systems have attracted significant research interest for memory device applications due to their tunable electrical properties through molecular design and good processibility. The representative classes of polymer materials used for fabrication on memories are conjugated polymers, non-conjugated polymers (functional polyimide systems or polymers with specific pendent donor or acceptor moiety) and polymer nanocomposites (metal nanoparticle or fullerene embedded). However, the effects of polymer structure on the memory characteristics have not been fully explored yet. In this thesis, the homopolymers pendant with donors or acceptors and related copolymers containing of both donor and accepter units were synthesized and evaluated on memory devices. In chapter 2, the homopolymers with pendant para-substituted dicyano- triphenylamine (D, as donor) or three different oxadiazole derivatives (2-phenyl-5-(4-vinylphenyl)-1,3,4-oxadiazole (A1), 2-(4-vinylbiphenyl)-5-(4-phenyl)- 1,3,4-oxadiazole (A2), and 2-(4-vinylbiphenyl)-5-(4-ethoxyphenyl)-1,3,4-oxadiazole (A3), as electron acceptor) and also related random copolymers consisting of pendant donor and acceptor moieties (the ratio of donor:acceptor is 8:2, 5:5 and 2:8, respectively) were synthesized by Nitroxide-Mediated Living Free Radical Polymerization (NMP) and characterized by 1H NMR spectrum and element analysis (EA). The para-substituted dicyano groups on triphenylamine improve the stability without the dimerization. The side-chain acceptor A2 shows more conjugated length relative to A1 by introducing another benzene ring, and A3 system further enhances the electronic density by adding terminal electron donating moiety (-OEt). The two thin film absorbance peaks of copolymers at 353 nm and 308 ~ 332 nm belong to the corresponding homopolymers (PD at 355 nm, PA1 at 308 nm, PA2 at 330 nm, PA3 at 332 nm, respectively) and the relative absorbance of two peaks vary with donor and acceptor ratios. The band gaps calculated from absorption edge are in the range of 3.11 ~ 3.20 eV and increase slightly with the larger acceptor content. The HOMO and LUMO energy levels of the random copolymers with different donor and acceptor ratios obtained from cyclic voltammetry (CV) are almost the same with the corresponding HOMO level of donor (-5.63 eV) and LUMO levels of acceptor (-2.63 ~ 2.81 eV), respectively, mainly due to poor hybridization of adjacent donor and acceptor which could be considered as an isolated system. The electronic properties can be well-tuned through different donor-acceptor structures and ratios of the copolymers systems. In chapter 3, the memory device of the random copolymers and homopoymers are fabricated and measured in ambient atmosphere. Different solvents (N,N-dimethyl formamide (DMF) or chlorobenzene (CB)) are chosen for polymer thin film processing. The bistable memory behavior is conducted by a simple sandwich device configuration consisted of spin-coated polymer film between indium-tin oxide (ITO) bottom electrode and aluminum (A1) top electrode. All the polymers exhibit nonvolatile flash type of switching behavior. The memory devices have low turn-on threshold voltage below 1 V and exhibits long retention time of 11000 seconds. The mechanism of the switching behavior is based on filamentary conduction with space charge limited current (SCLC) theory. As the polymer films prepare from DMF, the OFF state current during erasing scan increases from 10-5 A to 10-2 A after several WRER cycles. These phenomena are further obviously demonstrated in the case of the larger acceptor contents of the random copolymers (5:5 and 2:8 donor:acceptor ratio) and homopolymers pendant with acceptor due to the presence of the binding heteroatom (N) on acceptor with Al which contributes to filament conductions. Introducing electron donating ethoxy substituent (A3) which enhance the binding ability of N atom of oxadiazole group makes the memories difficult to turn off and also keep them on the high conductance since the stable filament channel are formed. The memory can still restore to the initial OFF state if the erasing process scan more negatively which also means the enhanced threshold voltage (from -3 V to -5 V) as the acceptor ratio increases. The polymer surface prepared from the CB solution is smooth and the current elevation of OFF state during the erasing switch could not be found. The study demonstrates that pendent donor/acceptor structures and ratio of copolymer and processing condition on the switching characteristics of memory device are fully explored.

The phase change memory (PCM) is one of the key enabling memory technologies for next-generation non-volatile memory device applications due to its high writing speed, excellent endurance, long retention time, and good scalability. However, the high power consumption of PCM devices caused by the high switching current from a high resistive state to a low resistive state is a critical obstacle to be resolved before widespread commercialization can be realized. Here, a useful approach to reduce the writing current of PCM, which depends strongly on the contact area between the heater electrode and active layer, by employing self-assembly process of Si-containing block copolymers (BCPs) is presented. Self-assembled insulative BCP pattern geometries can locally block the current path of the contact between a high resistive film (TiN) and a phase-change material (Ge2 Sb2 Te5 ), resulting in a significant reduction of the writing current. Compared to a conventional PCM cell, the BCP-modified PCM shows excellent switching power reduction up to 1/20 given its use of self-assembled hybrid SiFex Oy /SiOx dot-in-hole nanostructures. This BCP-based bottom-up process can be extended to various applications of other non-volatile memory devices, such as resistive switching memory and magnetic storage devices.

Resistive random-access memory (RRAM) devices are noticeable next generation memory devices. However, only few studies have been conducted regarding RRAM devices made of alloy. In this paper, we investigate the resistive switching behaviors of an Au/Ti/HfTiOx/p-Si memory device. The bipolar switching is characterized depending on compliance current under DC sweep mode. Good retention in the low-resistance state and high-resistance state is attained for nonvolatile memory and long-term memory in a synapse device. For practical switching operation, the pulse transient characteristics are studied for set and reset processes. Moreover, a synaptic weight change is achieved by a moderate pulse input for the potentiation and depression characteristics of the synaptic device. We reveal that the high-resistance state and low-resistance state are dominated by Schottky emissions.

We demonstrated unipolar organic bistable memory devices with 8 x 8 cross-bar array type structure. The active material for the organic non-volatile memory devices is poly(styrene-co-styrenesulfonic acid Na) (PSSANa). From the electrical measurements of the PSSANa organic memory devices, we observed rewritable unipolar switching behaviors with a stable endurance and narrow cumulative probability. Also the PSSANa memory devices exhibited a uniform cell-to-cell switching with a high ON/OFF ratio of approximately 10(5) and good retention time of approximately 10(4) seconds without significant degradation.

Redox-active ferrocene-containing iridium(III) complex for non-volatile flash memory

A continuous exponential rise has been observed in the storage and processing of the data that may not curtail in the foreseeable future. The required data processing speed and power consumption are restricted by the buses between the logic and memory devices that are characteristic of the von Neumann computing architecture. Bio‐mimicking neuromorphic computing has garnered considerable academic and industrial interest to resolve these challenges. Additionally, devices based on emerging nonvolatile memories capable of mimicking the behaviors of synapses and neurons, which are the main elements in biological computing systems (brains), are attracting significant interest from the device community. With the discovery of ferroelectricity in fluorite‐structured oxides, such as HfO2and ZrO2, which are compatible with the state‐of‐the‐art complementary‐metal‐oxide‐semiconductor processes, ferroelectric devices have rapidly evolved as the main direction of these research and development activities. Fundamental science related to fluorite‐structured ferroelectrics has been intensively studied over the last decade. At present, the focus is gradually moving to practical applications, including neuromorphic computing and advanced classical processing or memory units in the conventional von Neumann architecture. However, despite its rapid development, the wealth of recent progress in neuromorphic computing devices based on fluorite‐structured ferroelectrics has not been reviewed and systemized. This progress report comprehensively reviews and systemizes the recent progress in artificial synaptic and spiking neuron devices for neuromorphic computing based on fluorite‐structured ferroelectrics.image