Material challenges for nonvolatile memory Cite as: APL Mater. 10, 090401 (2022); https://doi.org/10.1063/5.0111671 Submitted: 19 July 2022 • Accepted: 05 August 2022 • Published Online: 08 September 2022 Bhagwati Prasad, Stuart Parkin, Themis Prodromakis, et al. COLLECTIONS Paper published as part of the special topic on Materials Challenges for Nonvolatile Memory ARTICLES YOU MAY BE INTERESTED IN A comprehensive review on emerging artificial neuromorphic devices Applied Physics Reviews 7, 011312 (2020); https://doi.org/10.1063/1.5118217 Recent developments and the future perspectives in magnetoelectric nanocomposites for memory applications APL Materials 10, 010901 (2022); https://doi.org/10.1063/5.0076106 Next generation ferroelectric materials for semiconductor process integration and their applications Journal of Applied Physics 129, 100901 (2021); https://doi.org/10.1063/5.0037617 APL Mater. 10, 090401 (2022); https://doi.org/10.1063/5.0111671 10, 090401 © 2022 Author(s). APL Materials EDITORIAL scitation.org/journal/apm Material challenges for nonvolatile memory Cite as: APL Mater. 10, 090401 (2022); doi: 10.1063/5.0111671 Submitted: 19 July 2022 • Accepted: 5 August 2022 • Published Online: 8 September 2022 Bhagwati Prasad,1,a) Stuart Parkin,2,b) Themis Prodromakis,3,c) Chang-Beom Eom,4,d) Jordi Sort,5 , 6,e) and J. L. MacManus-Driscoll7,f) AFFILIATIONS 1 Materials Engineering Department, Indian Institute of Science, Bengaluru, KA, 560012, India 2 Max Planck Institute of Microstructure Physics, 06120 Halle, Germany 3 Centre for Electronics Frontiers, School of Engineering, University of Edinburgh, Edinburgh, EH9 3BF, United Kingdom 4 Department of Materials Science and Engineering, University of Wisconsin-Madison, Madison, Wisconsin 53706, USA 5 Departament de Física, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain 6 Institució Catalana de Recerca i Estudis Avançats (ICREA), Passeig Lluís Companys 23, 08010 Barcelona, Spain 7 Department of Materials Science and Metallurgy, University of Cambridge, 27 Charles Babbage Road, Cambridge CB3 0FS, United Kingdom Note: This paper is part of the Special Topic on Material Challenges for Nonvolatile Memory. a) Author to whom correspondence should be addressed: bpjoshi@iisc.ac.in b) Electronic mail: parkin@mpi-halle.mpg.de c) Electronic mail: t.prodromakis@ed.ac.uk d) Electronic mail: ceom@wisc.edu e) Electronic mail: jordi.sort@uab.cat f) Electronic mail: jld35@cam.ac.uk Published under an exclusive license by AIP Publishing. https://doi.org/10.1063/5.0111671 This Special Issue is focused on the material challenges for The starting point for the development of non-volatile memory emerging nonvolatile memory technologies. Over the last few (NVM) technology is to find a suitable material that shows two or decades, several materials have been explored for memory applica- more stable states of physical properties that can be switchable by tions. There is an urgent need for material innovations for devel- external electrical, mechanical, thermal, magnetic, or other stimuli. oping emerging data storage technologies, driven by the inherent Several switching mechanisms have been investigated significantly limitations of current computational systems, e.g., the “memory over the last two decades in various two-terminal devices, such wall,” and high power consumption. The “memory wall” problem as resistive random-access memory (RRAM), ferroelectric tunnel of the current von-Neumann architecture is that the time to move junction (FTJ), phase-change RAM (PCRAM), and spin-transfer and store data is orders of magnitude slower than the computa- torque-based magnetic random-access memory (STT-MRAM), with tion speed. The computational architecture includes several types the promise of scalable, low-power operating, high-density non- of memories, such as the static random-access memory (SRAM), volatile memory (NVM) solutions. Various material systems, rang- with a response time of ∼1 ns and virtually unlimited endurance but ing from metallic binary oxides, complex oxides, chalcogenides, with low areal density due to the six-transistor configuration, or the nitrides, etc., mostly in thin-film form, have been explored for these dynamic random-access memory (DRAM), with a response time of devices. Initially, crossbar arrays have been proposed and experi- ∼10 ns. The areal density of DRAM is higher than SRAM because of mentally tried for the fabrication of the high-density memory archi- its one-transistor one-capacitor configuration. Storage class memory tecture. However, the crosstalk between the neighboring memory (SCM) is a new level in the hierarchy, being between main memory units during the write/read operation results in unwanted switching (DRAM) and storage memory and it is non-volatile. Although the in these devices. To overcome this issue, one transistor–one resistor response times of DRAM and SRAM are much higher than SCM, (1T–1R) and one selector–one resistor (1S–1R) structures have been they are volatile and so do not retain information when powered investigated. For selector purposes, the ovonic threshold switching down. The higher density, lower cost, and higher-speed SCM can (OTS) or metal–insulator transition (MIT) characteristics of several function as a non-volatile DRAM.1 chalcogenides and oxide materials have been explored.2 APL Mater. 10, 090401 (2022); doi: 10.1063/5.0111671 10, 090401-1 Published under an exclusive license by AIP Publishing APL Materials EDITORIAL scitation.org/journal/apm The growth of multilayer complex heterostructures is typically a highly explored material for phase-change memory as it offers required for these memory devices, which is the foundation for revolutionary properties that enable non-volatile storage. Electri- the fabrication process. Mostly sputtering, chemical vapor depo- cal heating can cause this material to change “phases” between sition and atomic layer deposition (ALD) techniques have been crystalline (c) and amorphous (a) states, which is employed for effi- used to grow these materials. Among these processes, ALD enables cient data storage. Here, the crystalline phase represents the low low temperature, uniform, large-area deposition of various mate- resistance state (LRS) or ON state, and the amorphous phase rep- rials. Controlling the thickness of the films down to a monolayer resents the high resistance state (HRS) or OFF state. Besides the over the large-scale sample is possible with ALD. Integration of data storage application, phase change materials are also used as a these devices with the complementary metal-oxide-semiconductor selector in MRAM and PCRAM applications. For selector applica- (CMOS) process is another key step. tions, the volatile ovonic threshold switching (OTS) characteristics A further critical consideration wherein the development of of these materials are being used. Normally, under the electric field, non-volatile memory can play a key role relates to the emergence the amorphous phase of these materials turns on/off the electric of the Artificial Intelligence (AI) era, where a large amount of data current, without notable phase transition. Although the PCRAM needs to be processed to bring automation in smart systems in and RRAM have been well-explored, the device-to-device variabil- various fields, such as healthcare, automobile, manufacturing indus- ity in these technologies is unfavorable for reliable data storage try, advertising industry, and security applications. To keep pace applications. with the ever-growing need for AI systems, fast access to vast Among the new memory technologies, the spin-transfer, amounts of data with low operation energy is required. However, torque-based MRAM technology has emerged as a promising alter- the current computation architecture is not capable to cope with native non-volatile memory solution due to its low-error rate. How- these needs due to the “memory wall” problem. Introducing high- ever, the low endurance, difficulty in switching well below 10 ns, speed, low-power, non-volatile memory closer to the process part and high switching current issues are challenging aspects to be would be a potential solution to tackle this issue. A new computing addressed before making STT-MRAM a standalone memory tech- paradigm—beyond CMOS, such as in-memory and neuromorphic nology. Changing the device structure from a two-terminal STT computing—would be a better approach to tackle this issue. The to a three-terminal Spin–Orbit Torque (SOT) device resolves these key advantage of using neuromorphic computing is its parallel data issues, as unlike in STT-MRAM, read and write current paths are processing capability that enables adaptive learning, similar to the different in the SOT-MRAM device, and the current density to human brain, and hence can perform complex tasks, such as pattern accomplish the SOT effect could be significantly lower than that of recognition and real-time data analysis. the STT effect. Additionally, the ultrafast switching of perpendic- Among all new memory technologies, RRAM devices with mul- ular magnetization is possible with the in-plane spin polarization tilevel states have been investigated for memory, neuromorphic carrying spin-currents in the SOT layer. Typically, the heavy met- computing, and synaptic device applications. In a typical RRAM als (HMs) that show a spin Hall effect (SHE), such as Pt, W, Ta, device or memristor, resistive switching is accomplished by the and Hf, are being used as a SOT layer in SOT-MRAM devices. electromigration of ions and defects, such as oxygen vacancies. The charge-to-spin conversion efficiency, known as the spin-Hall The resistance levels of these devices depend upon the past and angle (SHA), of the SOT material, is the key to realizing the present applied external stimuli. It has been demonstrated that energy-efficient SOT-MRAM devices. In addition to heavy metals, one can mimic the functionality of the biological synapses with several other material systems, such as topological insulators, com- such multilevel resistive states with spike-time-dependent plasticity plex oxides, etc. have been explored for higher SHA. Beside the and long-term synaptic potentiation/depression. Formulation and bulk SHE in the heavy metal and the interfacial spin–orbit cou- manipulation of defects in the material enables resistive switch- pling effect play a role in the generation of spin current in FM/HM ing behavior. In complex oxide systems, the formation of oxygen bilayer structures. On the one hand, the magnitude of SHE decreases vacancies can disrupt the charge, spin, and orbital ordering of the with decreasing the thickness of the HM below a certain thick- materials and hence can modulate the physical properties, includ- ness determined by the spin diffusion length; on the other hand, it ing resistive switching. Recently, the role of oxygen vacancies in increases due to the dominant interfacial spin–orbit coupling with the stabilization of the ferroelectric phase in ultrathin films of decreasing HM thickness. Therefore, an optimal thickness of HM Zr-doped HfO2 systems has also been demonstrated. In fact, one is required to achieve a higher SHE at the FM/HM interface in can achieve binary or even multiple resistive states by sandwich- SOT-MRAM devices. ing such ultrathin films as a tunnel barrier between two metallic Magnetic tunnel junctions with perpendicular magnetic electrodes. anisotropy (p-MTJ) are now primarily being used in MRAM tech- Changing the phase of the material from crystalline to amor- nology for high density, low-power consuming non-volatile memory phous or vice versa via electric current is another approach to applications. Fabrication of the commercial memory products needs achieve the nonvolatility in the resistive states of the devices. Such CMOS compatibility of the MRAM devices, especially in terms of phase change materials (PCM) have also been well explored for the thermal budget for the back-end-of-line integration. Typically, non-volatile memory as well as neuromorphic computing appli- HMs are being used as a seed or SOT layer in the electric current- cations. The fundamental idea underlying data storage in phase- driven STT and SOT MRAM devices. Among HMs, only Mo and change memory is the principle of change in resistance rather than beta-W can maintain the PMA in HM/CoFeB/MgO-based p-MTJ charge. The typical phase change materials are chalcogenides, such devices at 400 ○ C or higher temperature. Although STT and SOT as GeTe, GeS, GeSe, SiTe, Sb2 Te3 , and Ge2 Sb2 Te5 . Germanium have proven to be elegant ways of switching the MRAM devices, the Antimony Tellurium (GST) (usually Ge2 Sb2 Te5 and GeSb2 Te4 ) is higher power consumption due to the Joule heating effect caused APL Mater. 10, 090401 (2022); doi: 10.1063/5.0111671 10, 090401-2 Published under an exclusive license by AIP Publishing APL Materials EDITORIAL scitation.org/journal/apm by the passage of electric current is still a major concern. To over- demonstrated formingless bipolar RS and self-rectifying behavior come this problem, electric field control of magnetism is found via fluorine anion migration, dramatically lowering sneak current in to be an alternative route to manipulate the resistance state of crossbar array RRAM. They verified this experimentally with XPS, MRAM devices. FT-IR, and Hall measurements in the presence of F anion passivated In this Special Issue, we compile a series of articles on emerging oxygen vacancies in the film. memory technologies, particularly on RRAM, PCRAM, and MRAM. Cox et al.7 also took a step forward toward better designing These articles provide an overview of the ongoing research in these of ReRAM devices by providing an elusive key to understand- fields and constitute the guidelines for the development of novel ing and addressing the problem of ReRAM device reliability. They materials and devices. revealed that the device interacts with the ambient environmental RRAM is categorized into two types based on the nature of conditions while in operation. Their findings demonstrated not just the ions involved in the switching mechanism: oxygen vacancy- factual proof of moisture injection into devices, but also a direct RAM (OxRAM) and conductive bridge RAM (CBRAM). To create a link between the amount of oxygen injected from ambient moisture conductive bridge, CBRAM uses ionization of metal electrodes (for and the microstructure of the oxide layer. The authors developed an example, Ni, Ag, and Cu) unlike OxRAM where ionization of oxy- efficient secondary ion mass spectrometry (SIMS) in situ normaliza- gen is taking place. This way, CBRAM has faster switching speed. tion approach for monitoring chemical changes between electrodes However, CBRAM often suffers from switching instability. Thus, for and switching layers. They demonstrated it for SiOx devices, but the improving switching stability, the insertion of a metal layer between method is widely relevant to any device with a rough surface that can the electrode and switching layer was discovered as the most feasible host oxygen exchange. approach. Simanjuntak et al.3 found that inserting Cr and Ti films Furthermore, resistive switching (RS) emerged as an attrac- below the Ag electrode creates ternary oxides, CrZnOx and TiZnOy , tive alternative also for neuromorphic computing. Several materials, that act as a barrier layer to restrict Ag species drift during switching including titanium oxide, hafnium dioxide, tantalum oxide, and and facilitate stable switching. aluminum oxide, have already been proposed for the RS layer. In addition to standard metal oxides (TiO2 , HfO2 , and Most of the devices are conductive filament (CF) memristors. How- Al2 O3 ) with MIM structures, for RRAM applications, Lei ever, CF-type memristors have certain drawbacks in neuromorphic et al.4 created a unique stable hybrid Hf-based hydroquinone computing due to their natural abrupt switching nature. Thus, (Hf-HQ) film deposited using molecular layer deposition (MLD) for compatibility with a thin-film transistor (TFT), Pereira et al.8 for flexible hybrid RRAM devices. The resistive switching (RS) fabricated a a-Indium-gallium-zinc-oxide (a-IGZO) based memris- behavior of the hybrid device (Pt/Hf-HQ/Al2 O3 /TiN) was designed, tor, such as Mo/a- IGZO/Ti/Mo, for a reduced oxygen concentra- and properties such as retention and durability have been thor- tion. These findings reveal the impact of RS device compatibility oughly investigated. Unlike halide perovskite RRAM devices, with thin-film transistors (TFT). which have a high ON/OFF ratio, these hybrid memory units Proceeding further, in synaptic characteristics of memristor offer electroforming-free RS behavior, with improved stability and devices, analog and digital memristors have various circuit func- repeatability. tionalities (operational speed, current level, voltage range, etc.). In a similar manner, aside from standard oxide research, The research done by Simanjuntak et al.9 gave not only an insight attention has shifted to SiO2 and its associated suboxide (SiOx ), into establishing an equivalent synaptic response for dictating an which provides several advantages in terms of processing, cost, analog and digital response of memristor devices but also pro- manufacture, and compatibility with CMOS. It was proposed that vided a way for realizing convertible digital-analog memristors, the formation of voids, low-density areas, and/or redistribution which might open new possibilities in electronic circuits and sys- of O create a columnar structure, which helps in the forma- tems. Their experimental data revealed that the maximum work- tion of conductive filaments (CFs). To realize the full potential ing current level of memristor devices controls their synaptic of these devices, it is important to exploit the formation of col- response. umn boundaries (CB). These boundaries are crucial for achiev- De Araujo et al.10 proposed a novel resistive switching device, ing good resistance switching characteristics in SiOx (ReRAM) i.e., the correlated electron-based RAM (CeRAM), to prevent the devices. Patel et al.5 studied the formation and structure of colum- unpredictability of resistive memory produced by the stochastic nar boundaries that arise due to SiOx layers being templated filamentary activity. CeRAM is a unique resistive switching tech- by metal surface roughness. Here metal/SiOx /metal stacks are nology that may be utilized to allow analog neuromorphic comput- created by sputter deposition, and to understand the nature of ing and increased memory density, or as an extra CMOS process columnar structures in SiOx and the interaction between metal modification. electrodes and the SiOx layer, high-resolution TEM (HRTEM), Furthermore, the electroforming process is followed by a electron energy-loss spectroscopy (EELS), energy-dispersive x-ray significant oxygen release and temperature effects that damage the spectroscopy (EDX), and density functional theory (DFT) modeling device’s nanostructure and decouple the active zone from the rest of are used. the device, making it difficult to integrate these systems in numer- Another research upgrade for RRAM was to considerably ous device topologies. Thus, for better development of non-volatile increase its reliability and to fabricate sneak current-free cross- memory and neuromorphic computing, Acevedo et al.11 provide bar array architecture (CAA) memory devices via controlling many methods for reducing damage while preserving the device’s oxygen vacancies. Kim et al.6 presented a simple Atomic Layer structural integrity, chemistry, and multi-mem behavior, which are Deposition (ALD) technique for a novel RS material for RRAM required for memory or neuromorphic computer hardware. They devices: fluorine-doped titania (F: TiO2 ). The Pt/F: TiO2 /Pt device produced oxidized La1/2 Sr1/2 Mn1/2 Co1/2 O3−x (LSCMO) thin films on APL Mater. 10, 090401 (2022); doi: 10.1063/5.0111671 10, 090401-3 Published under an exclusive license by AIP Publishing APL Materials EDITORIAL scitation.org/journal/apm conducting Nb:SrTiO3 (NSTO) substrates with out-of-plane (001) Furthermore, a two-terminal GST (Ge2 Sb2 Te5 ) based memris- and (110) orientations using pulsed laser deposition (PLD), and tive device was built to implement synaptic functions by Go et al.16 conducted a thorough investigation of the electrical behavior of In their work, when using a memristive device with just nanopil- multi-mem NSTO/LSMCO/Pt devices, with varying crystallinities, lars deposited, this newly constructed artificial synapse exhibits out-of-plane orientation, and stimulation, which display a robust paired-pulse facilitation/depression (PPF/PPD) behavior as well as memristive behavior. multistate and rapid long-term potentiation/depression LTP/LTD Besides RRAM, introducing an ultrathin ferroelectric film as in a GST-based device, with more conductance states than pre- a tunnel barrier between two metallic electrode layers is another viously described GST-based synapses. They demonstrated cell way to achieve a large resistive switching in the devices. Although characteristics and electro-thermal simulations for RIMS (Rapid- the tunnel electroresistance effect has been demonstrated with ultra- operating-time, Intermediate bias-range, Multiple-states, Several- thin perovskite-based ferroelectric tunnel junctions for more than synaptic-functions) synapse utilizing this memristive device. As a 15 years, technological interest in this field has increased recently result of this work, it will be possible to analyze both temporal and because of the discovery of ferroelectricity in CMOS process- geographical data more effectively. friendly material, doped HfO2 films. The tunnel electroresistance Although phase change materials have been studied signifi- effect has also been demonstrated in ultrathin doped HfO2 ferro- cantly, the ovonic threshold switching (OTS) mechanism in these electric tunnel barriers. However, the electroresistance varies from materials is not yet fully understood. The OTS characteristics of device to device due to the presence of the grain boundaries in these materials are being used as a selector in several memory tech- these ferroelectric films that provide conducting paths for the charge nologies, including MRAM and PCRAM. In view of this, Gu et al.17 motions. To overcome this problem, Long et al.12 used an ultrathin used first-principle calculations to provide an in-depth understand- (∼1 nm) SrTiO3 film as a capping layer on a Hf0.5 Zr0.5 O2 (HZO) ing of the structural and electronic property of a simple OTS film epitaxially grown on SrTiO3 and GdScO3 substrates, that led to material, the amorphous (a-) SiTe. Their study revealed how two more robust ferroelectric tunnel junctions with large endurance and identical materials, GeTe and GeSe, behave differently as GeTe is an higher yield. ovonic memory switching (OMS) material rather than OTS material, Zhou et al.13 used the Z-scan technique to explore the phase unlike GeSe. shift of GST material under a terahertz quantum cascade laser (QCL) In STT-MRAM, the reduction of the barrier height (com- emitting at around 2.5 THz. To evaluate the phase transition of GST, monly known as delta) between two stable magnetization states TEM, and X-ray diffraction (XRD) characterization were performed. leads to a lowering of the switching current but at the cost of poor Numerical simulations based on 3D finite elements and electro- retention. Therefore, finding the right balance is always challeng- magnetic heating were performed to compute the temperature and ing. To address this issue, recently the concept of double magnetic transmission changes, in which both simulation and experiment tunnel junctions (DMTJ) has been proposed as an alternative for verified that the a-GST film can be phase changed to c-GST under energy-efficient STT switching, where the storage layer is sand- 1.5 mW terahertz power. wiched between two tunnel barriers with a hard bottom reference To extend the capacity of PCRAM, Sun et al.14 focused on layer and a hard top polarizer layer. Interestingly, this device struc- doping Sb2 Te3 with Sc forming Sc0.2 Sb1.8 Te3 (SST) that increases the ture not only decreases the switching current but also enhances the SET speed up to ∼0.7 ns. However, for embedded memory, ther- data retention by increasing the PMA of the storage layer. How- mal stability is not suitable for SST. Further indium-doped GST ever, getting sufficient PMA on the top polarizer, dealing with alloys were shown to have high thermal stability. They presented a stray magnetic fields with two high PMA layers (top polarizer thorough theoretical work on InTe based on ab initio simulations, and bottom reference), and the difficulty to fabricate such DMTJ with calculations performed using density functional theory (DFT) devices remain challenging. Hazen et al.18 have solved these issues based on molecular dynamics. The time step was set to 2 fs. The elec- by introducing a simpler and thinner assisting layer, instead of tronic structure computations and chemical bonding investigations the thick top polarizer layer in DMTJ, with the MgO capping were performed utilizing the Vienna Ab initio Simulation Package layer. They have found the figure of merit defined by the ratio of (VASP) code and the Local Orbital Basis Suite. This study will serve thermal stability (delta) vs critical current, in their DMTJ devices as a springboard for further research on phase-change devices that to be nearly double that of single MTJ devices with a diameter use both heat and mechanical-induced transitions. 80–100 nm. In addition to this, the use of GST in nanophotonic elements CMOS integration is another challenge in MRAM technology was experimentally demonstrated by Lazarenko et al.15 Their work development. For STT devices and even for SOT devices, Ta has focuses on a numerical and experimental investigation of the atten- been widely used as a seed layer for PMA as well as a spin current uation coefficient dependency for the a-GST and c-GST lengths source layer for SOT-MRAM. However, Ta has poor thermal sta- (100 nm–20 μm), on the silicon nitride rib waveguide at the telecom bility at a temperature above 350 ○ C, and therefore its integration wavelength of 1550 nm, as well as the experimental identification of into the CMOS process is difficult. Although beta W is a superior the GST cell range with dominant light scattering. The transmission material for SOT application in terms of efficiency and thermal sta- spectra for the two constructed O-ring resonators (ORRs) before and bility, its growth is a challenging task. Therefore, a novel material after a-GST/SiO2 and c-GST/SiO2 synthesis were measured. They that can replace beta W is needed for the CMOS integration of found that the most energy-efficient devices in this situation should SOT-based MRAM devices. In view of this, Ranjan et al.19 demon- have GST cell lengths of more than 2 μm. As a result, compromises strated a thermally stable PMA up to 425 ○ C in rhenium (Re) with an must be made throughout the creation of nanophotonic devices, and SHA of 0.065 ± 0.003 and SOT switching current density of around each task must be approached differently. 1.36 × 1011 A/m2 . Besides the larger SHA, lower resistivity is sought APL Mater. 10, 090401 (2022); doi: 10.1063/5.0111671 10, 090401-4 Published under an exclusive license by AIP Publishing APL Materials EDITORIAL scitation.org/journal/apm to reduce the Joule heating effect during the writing. Interestingly, computing systems. We hope the contents of this issue will pro- the resistivity of the Re is nearly half compared to Ta and beta-W. vide several new directions for materials research on non-volatile Further exploration of the new materials may provide even better memory. SOT material. The interfacial spin–orbit coupling at the FM/HM interface We would like to thank the editorial staff at APL Materials has been explored significantly for achieving the large SHE. How- who supported us all behind the scenes, especially Jessica Trudeau. ever, the role of the interface between the HM and substrate was This collection of papers could not have been put together without overlooked. Some theoretical works have recently suggested a sig- their hard work and dedication. We also thank Astha Khandelwal nificant contribution of the insulator/HM interface to the SOT (Indian Institute of Science, Bengaluru) and Markus Hellenbrand effect, particularly when the thickness of the HM is less than (University of Cambridge) for their help in editing and proofreading the spin diffusion length. The common substrates that are being this article. used for SOT-MRAM are insulating materials such as amorphous SiOx , single-crystalline SrTiO3 , MgO, etc., where the HM is sit- ting on the substrate. Therefore, besides the FM/HM interface, the interfacial spin–orbit coupling at HM/insulating substrate interface AUTHOR DECLARATIONS would be another approach to tuning the SOT effect. Recently, Conflict of Interest Choi et al.20 have done such experiments with the Pt/substrate interface by investigating the dependence of Pt thickness on the The authors have no conflicts to disclose. SOT effect, where Co/Pt bilayers were grown on amorphous SiOx and single-crystalline SrTiO3 and LaAlO3 substrates. Unlike the conventional bulk SHE, weaker thickness dependence of the SHE REFERENCES was found in the samples grown on single crystalline substrates, 1 M. Si, H. Y. Cheng, T. Ando, G. Hu, and P. D. Ye, “Overview and outlook of which suggests that the interfacial spin–orbit coupling at the emerging non-volatile memories,” MRS Bull. 46, 946 (2021). HM/substrate interface can be exploited to develop efficient SOT 2 T. Schenk, M. Pešić, S. Slesazeck, U. Schroeder, and T. Mikolajick, “Memory based devices. technology—A primer for material,” Rep. Prog. Phys. 83, 086501 (2020). Fabricating a three-terminal SOT-MRAM device is challeng- 3 F. M. Simanjuntak et al., “Formation of a ternary oxide barrier layer and its role in ing, particularly avoiding the side redeposition during the etch- switching characteristic of ZnO-based conductive bridge random access memory ing process that leads to the shutting effect in MTJ. Ren et al.21 devices,” APL Mater. 10, 031103 (2022). 4 have addressed this issue by carefully optimizing the ion beam J. Lei, W.-J. Ding, C. Liu, D. Wu, W.-M. Li, and A.-D. Li, “Polymerized hybrid etching angle and time along with other device fabrication pro- Hf-based hydroquinone/Al2 O3 bilayer structure by molecular/atomic layer depo- cesses in three-terminal in-plane SOT devices. Besides this, with sition for non-volatile resistive random access memory,” APL Mater. 9, 121110 (2021). the insertion of thin Hf interlayers into the heterostructure, they 5 K. Patel, J. Cottom, A. Mehonic, W. H. Ng, A. J. Kenyon, M. Bosman, and have demonstrated the enhancement in thermal stability and reduc- A. L. Shluger, “The nature of column boundaries in micro-structured silicon oxide tion in the switching current density of the fabricated devices. nanolayers,” APL Mater. 9, 121107 (2021). As per the micromagnetic simulation, the insertion of thin Hf 6 M. Kim et al., “Resistive switching properties for fluorine doped titania fabricated increases the magnetic anisotropy at one interface, while decreasing using atomic layer deposition,” APL Mater. 10, 031105 (2022). the strength of the Dzyaloshinskii–Moriya interaction on another 7 H. R. J. Cox et al., “A nanoscale analysis method to reveal oxygen exchange interface, thereby decreasing the switching current density while between environment, oxide, and electrodes in ReRAM devices,” APL Mater. 9, enhancing the thermal stability. 111109 (2021). In order to make energy-efficient spintronic devices by elimi- 8 M. E. Pereira et al., “Tailoring the synaptic properties of a-IGZO memristors for nating the Joule heating effect, the magnetoelectric coupling route artificial deep neural networks,” APL Mater. 10, 011113 (2022). 9 has been explored with several systems for the last two decades. F. M. Simanjuntak et al., “Conduction channel configuration controlled digital Wu and MacManus-Driscoll22 have provided an overview of the and analog response in TiO2 -based inorganic memristive artificial synapses,” APL recent developments in several magnetoelectric nanocomposite sys- Mater. 9, 121103 (2021). 10 tems with future perspectives for magnetoelectric RAM (MERAM). C. A. P. De Araujo et al., “Universal non-polar switching in carbon-doped They have covered the topic from basic mechanisms to mate- transition metal oxides (TMOs) and post TMOs,” APL Mater. 10, 040904 (2022). 11 rials and applications. Among several nanocomposite systems, W. R. Acevedo et al., “Optimization of the multi-mem response of topotactic redox La1/2 Sr1/2 Mn1/2 Co1/2 O3−x ,” APL Mater. 10, 011111 (2022). the 3-1 type oxide composite films are promising candidates for 12 X. Long et al., “Enhanced electroresistance endurance of capped Hf0.5 Zr0.5 O2 low-power spintronics. They set out the challenges and direc- ultrathin epitaxial tunnel barriers,” APL Mater. 10, 031114 (2022). tions for achieving high-performance MTJ-based MERAM from 13 K. Zhou et al., “Phase change of Ge2 Sb2 Te5 under terahertz laser illumination,” 3-1 composite systems and explained how and why the use of self- APL Mater. 9, 101113 (2021). assembled triple 3-1 composite systems can enable a large converse 14 S. Sun, B. Zhang, X. Wang, and W. Zhang, “Density dependent local structures magnetoelectric effect at room temperature without the application in InTe phase-change materials,” APL Mater. 9, 121105 (2021). of magnetic field biasing. 15 P. Lazarenko et al., “Size effect of the Ge2 Sb2 Te5 cell atop the silicon nitride Overall, this Special Issue covers the recent advances in RRAM, O-ring resonator on the attenuation coefficient,” APL Mater. 9, 121104 (2021). PCRAM, MRAM, and MERAM. The authors have demonstrated 16 S. X. Go, T. H. Lee, S. R. Elliott, N. Bajalovic, and D. K. Loke, “A fast, low- several novel ways of tackling the materials challenges for the energy multi-state phase-change artificial synapse based on uniform partial-state development of emerging memory technologies and neuromorphic transitions,” APL Mater. 9, 091103 (2021). APL Mater. 10, 090401 (2022); doi: 10.1063/5.0111671 10, 090401-5 Published under an exclusive license by AIP Publishing APL Materials EDITORIAL scitation.org/journal/apm 17 20 R. Gu et al., “Structural features of chalcogenide glass SiTe: An ovonic threshold G. Choi et al., “Thickness dependence of spin–orbit torques in Pt/Co structures switching material,” APL Mater. 9, 081101 (2021). on epitaxial substrates,” APL Mater. 10, 011105 (2022). 18 21 D. S. Hazen et al., “Real time investigation of double magnetic tunnel junction H. Ren, S.-Y. Wu, J. Z. Sun, and E. E. Fullerton, “Ion beam etching dependence with a switchable assistance layer for high efficiency STT-MRAM,” APL Mater. of spin–orbit torque memory devices with switching current densities reduced by 10, 031104 (2022). Hf interlayers,” APL Mater. 9, 091101 (2021). 19 22 A. Ranjan et al., “Spin–orbit torque driven perpendicular magnetization switch- R. Wu and J. L. MacManus-Driscoll, “Recent developments and the future ing in Re/CoFeB/MgO with high thermal stability,” APL Mater. 10, 011104 perspectives in magnetoelectric nanocomposites for memory applications,” APL (2022). Mater. 10, 010901 (2022). APL Mater. 10, 090401 (2022); doi: 10.1063/5.0111671 10, 090401-6 Published under an exclusive license by AIP Publishing