REVIEW www.afm-journal.de Reversible Crystalline-Crystalline Transitions in Chalcogenide Phase-Change Materials Bin Liu, Kaiqi Li, Jian Zhou, and Zhimei Sun* depicted in Figure 1a,b.[11,12] Initially, crystalline PCMs are heated above their melting temperature by either a laser beam or a short-width, high-amplitude electrical pulse. Subsequently, the melted PCMs are rapidly quenched to the non-crystalline (amorphous or glassy) state. This two-step process is referred to as reset/write. During this reset/write process, a non-crystalline mark is created against the crystalline background to represent a recorded bit. The disparity in optical or electrical properties between the non-crystalline and crystalline states allows for easy retrieval of the recorded information. To revert the non-crystalline mark back to its original crystalline state, a longer, moderateamplitude laser beam or electrical pulse is employed to heat the non-crystalline PCMs just above their crystallization temperature. This heating process triggers the rearrangement of atoms, transitioning the material back to an ordered state, known as set/erase. It is necessary to highlight that both laser beams and electrical pulses operate on thermal effects to trigger the typical phase transition between non-crystalline and crystalline states of PCMs, denoted as the non-crystalline-to-crystalline phase transition (n-CCPT) in this review. After over 60 years of development, PCRAM has been considered the most technologically mature candidate for the next generation of non-volatile memory.[13–16] In 2015, Intel and Micron jointly announced the first electronic PCRAM product based on a 3D XPoint memory architecture.[17,18] Subsequently, Intel introduced Optane solid-state disk (SSD) and persistent memory,[19,20] while Micron unveiled the fastest SSD utilizing the 3D XPoint Technology.[21] Positioned between NAND flash and dynamic random access memory (DRAM) in terms of performance and cost, PCRAM is anticipated to enter the market to narrow the performance gap between storage-class memory and DRAM technologies (Figure 1c[22] ). Notably, recent advancements have demonstrated that PCMs have the potential to emulate the analog characteristics of biological synapses and neurons by leveraging continuous transitions in their electrical/optical properties.[23,24] Consequently, PCMs have emerged as a promising medium for memristive and neuro-inspired electronic/optical devices (Figure 1d).[25–27] More specifically, the partial disordering of crystalline PCMs can yield multiple reset states, meanwhile, the stepwise set process can be achieved through the Phase-change random access memory (PCRAM) is one of the most technologically mature candidates for next-generation non-volatile memory and is currently at the forefront of artificial intelligence and neuromorphic computing. Traditional PCRAM exploits the typical phase transition and electrical/optical contrast between non-crystalline and crystalline states of chalcogenide phase-change materials (PCMs). Currently, traditional PCRAM faces challenges that vastly hinder further memory optimization, for example, the high-power consumption, significant resistance drift, and the contradictory nature between crystallization speed and thermal stability, nearly all of them are related to the non-crystalline state of PCMs. In this respect, a reversible crystalline-to-crystalline phase transition can solve the above problems. This review delves into the atomic structures and switching mechanisms of the emerging atypical crystalline-to-crystalline transitions, and the understanding of the thermodynamic and kinetic features. Ultimately, an outlook is provided on the future opportunities that atypical all-crystalline phase transitions offer for the development of a novel PCRAM, along with the key challenges that remain to be addressed. 1. Introduction Since Ovshinsky filed the patent entitled Resistance Switches and the Like in 1962,[1] phase-change random access memory (PCRAM), exploiting a substantial change in electrical or optical properties of chalcogenide phase-change materials (PCMs), has become a promising technology for data storage.[2–10] The working principle of PCRAM is relatively straightforward, as B. Liu, K. Li, J. Zhou, Z. Sun School of Materials Science and Engineering Beihang University Beijing 100191, China E-mail: zmsun@buaa.edu.cn B. Liu National Key Laboratory of Spintronics Hangzhou International Innovation Institute Beihang University Hangzhou 311115, China B. Liu State Key Laboratory of Materials for Integrated Circuits Shanghai Institute of Microsystem and Information Technology Chinese Academy of Sciences 865 Changning Road, Shanghai 200050, China The ORCID identification number(s) for the author(s) of this article can be found under https://doi.org/10.1002/adfm.202407239 DOI: 10.1002/adfm.202407239 Adv. Funct. Mater. 2024, 34, 2407239 2407239 (1 of 19) © 2024 Wiley-VCH GmbH www.afm-journal.de Figure 1. PCMs and PCRAM. a,b) The working principle of PCMs for PCRAM. c) The current memory hierarchy in terms of performance and cost. PCRAM combines high speed, non-volatility, high storage density, and manufacturing cost.[50] d) Biological neuron based on PCRAM cells.[28] e) Properties comparison of PCMs with other three types of resistive switching material (RSM). Artificial neural networks and spiking neural networks are respectively abbreviated as ANN and SNN.[51] Panels (a,b) are Reproduced with permission.[12] Copyright 2007, Springer Nature. Panel (c) is Reproduced with permission.[50] Copyright 2019, Springer Nature. Panel (d) is Reproduced with permission.[28] Copyright 2016, Springer Nature. Panel (e) is Reproduced with permission.[51] Copyright 2020, Springer Nature. application of medium-amplitude multiple-short voltage pulses, reflecting a gradual crystallization of non-crystalline PCMs.[28] Furthermore, the electrical/optical properties of a given state in a single PCRAM cell depend on its excitation history, enabling the construction of intricate neural networks based on an extensive array of PCRAM cells. With the demand for increased computing power due to the rapid growth of artificial intelligence, PCRAM technology has taken a prominent role in designing and implementing specialized electronic computing systems to bolster processing in artificial neural networks,[28–33] such as applicationspecific integrated circuits,[34] neuromorphic computing,[35] and in-memory computing[36] devices. The unique storage principle gives PCRAM based on nCCPT (PCRAM-n-CCPT) a unique set of features, such as good radiation-tolerance ability,[37] high scalability,[7] and multilevel storage capability,[38] making it a technologically mature candidate for advanced non-volatile memory. Nevertheless, the n-CCPT also brings PCRAM some inherent bottlenecks (Figure 1e). First, the contradiction between thermal stability and crystallization speed.[39,40] Thermal stability refers to the ability Adv. Funct. Mater. 2024, 34, 2407239 of PCMs to resist crystallization under thermal fields. Therefore, the usual method of ensuring stability is to sacrifice speed. Another bottleneck is the high power consumption associated with the reset operation of PCMs,[41,42] wherein the melting of PCMs leads to a power-intensive process (the power consumption during the reset operation of Ge2 Sb2 Te5 -based PCRAM is ≈901.8 pJ[43] ). Additionally, PCRAM-n-CCPT faces the issue of serious resistance drift.[44] During the reset operation, a highstrain, low-order non-crystalline state was created, which undergoes spontaneous structural relaxation toward an energetically more favorable ideal glassy state (called physical aging).[45] Physical aging would lead to a resistance increase over time, viz. resistance drift, which is well-recognized and poses challenges to further optimizing PCRAM for high storage density and neuroinspired computing devices that need significant programming consistencies.[46] Despite substantial efforts have been made to understand the crystallization kinetics and structural aging of non-crystalline PCMs,[47–49] this field is challenged by the presence of varied and sometimes conflicting physical mechanisms proposed to explain these phenomena, which in turn vastly 2407239 (2 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de hinder the overall performance optimization of PCRAM-n-CCPT for universal memory and neuro-inspired computing. Recently, various atypical phase transitions such as allcrystalline transition,[52] all-noncrystalline transition,[53,54] and non-thermal amorphization[55,56] have been observed in PCMs. These atypical transitions significantly differ from the typical nCCPT in terms of microscopic mechanisms and macroscopic properties. Of particular interest is the reversible crystalline-tocrystalline phase transition (CCPT), which presents an opportunity to overcome the inherent limitations of PCRAM-n-CCPT since the poor thermal stability, slow crystallization speed, high power consumption, and severe resistance drift issues of PCMs are all related to their non-crystalline phase.[57–60] However, the CCPT is not yet a fully mature PCRAM candidate. A comprehensive understanding is crucial to facilitate its development in data memory and computing. This review delves into the key aspects of CCPT, including atomic structures, switching mechanisms, and optimization strategies. Through our discussion, we aim to enhance the community’s comprehension of PCMs exhibiting CCPT and expedite their advancement. 2. CCPT in Ternary Chalcogenide Alloys Most alloyed chalcogenide PCMs such as GeTe,[61] Sn2 Se3 ,[62] In2 Se3 ,[63] In3 SbTe2 ,[64] and Ag-In-Sb-Te,[65] are known to exhibit polymorphism, with multiple crystalline phases existing at various temperatures. However, in present PCRAM storage, only the reversible n-CCPT is utilized. For example, the canonical PCMs Ge2 Sb2 Te5 feature one non-crystalline phase and two crystalline phases (metastable cubic and stable hexagonal).[66] During thermal annealing, the non-crystalline Ge2 Sb2 Te5 was first transformed into the cubic structure at ≈150 °C, leading to a significant resistance decrease of over 4 orders of magnitude.[67,68] This large contrast in resistance serves as the basis for information storage in Ge2 Sb2 Te5 -based PCRAM-n-CCPT devices. When the cubic Ge2 Sb2 Te5 is subjected to temperatures of 320–350 °C, it transitions into the stable hexagonal structure, further reducing its resistance.[69] The phase transitions from disordered noncrystalline to cubic and then to hexagonal phases have been successfully demonstrated in Ge2 Sb2 Te5 -based PCRAM devices through current-voltage measurements, heralding an essential feature for multilevel data storage.[69] However, some studies indicated that the stable hexagonal phase may lead to compromised storage performance in terms of power consumption,[70] cyclability,[71] and operating speed.[72] Consequently, preventing the formation of the hexagonal phase has been identified as a crucial concern for PCRAM applications, leading to the exclusive utilization of the n-CCPT between non-crystalline and cubic Ge2 Sb2 Te5 for PCRAM design. Until recently, many polycrystalline phase transitions in chalcogenide PCMs were observed and investigated, foreboding their potential application in data storage devices. It has been investigated that the vacancy layers (VLs) in Ge2 Sb2 Te5 have a pronounced impact on the electrical/thermal conductivity[73–75] and the CCPT.[76–78] By aberration-corrected scanning transmission electron microscopy (STEM), Lotnyk et al.[79] observed the formation and vanish of VLs in cubic Ge2 Sb2 Te5 induced by the focused electron beam. Therefore, the work of Lotnyk et al. might be regarded as a prelude to Adv. Funct. Mater. 2024, 34, 2407239 the CCPT in chalcogenide Ge-Sb-Te alloys. However, the exact atomic arrangement in both cubic and hexagonal Ge2 Sb2 Te5 is still under debate,[80] making it difficult to accurately determine the reversible CCPT mechanism. According to Yamada and Matsunaga,[81] the 4(a) sites in cubic Ge2 Sb2 Te5 are completely occupied by Te atoms, while the 4(b) sites exhibit random occupancy by Ge, Sb, and vacancies, with probabilities determined by the actual composition of Ge2 Sb2 Te5 (Figure 2a). In contrast, utilizing high-resolution TEM, Park et al.[82] proposed that atoms at the 4(b) site are arranged in order, and Ge and Sb atoms display a tendency to align on particular planes (Figure 2b). Subsequently, Sun et al. asserted ordering within the cations and VLs of the cubic structure could result in more stable configurations compared to completely random structures.[6] Through their calculations, they demonstrated the polymorphic nature of the cubic phase, showing a spectrum ranging from structures with random vacancies to those featuring ordered VLs.[83] The hexagonal Ge2 Sb2 Te5 is a high-temperature phase with a „ space group P3m1. Its atomic structure is comprised of 9-layer stacking blocks along the [0001] axis, distinguishing it from its cubic counterpart. However, its exact atomic arrangement remains a subject of ongoing debate, with three primary proposed stacking sequences along the [0001]. I) the Petrov model,[84] -(Te-SbTe-Ge-Te-Te-Ge-Te-Sb)-, was first introduced to describe the structure of Ge1 Sb4 Te7 , with layers of GeTe situated outside the layers of Sb2 Te3 .[84] The atomic layers of Te in GeTe create vdW gaps in this model. II) the Kooi model,[85] -(Te-Ge-Te-Sb-Te-Te-Sb-TeGe)-, was proposed to describe the crystal Gex Sb2 Te3+x (x = 1, 2, 3),[85] wherein the GeTe sublayers are chemically bonded with the Sb2 Te3 sublayers and the outermost Te atomic layers form vdW gaps. The Kooi model density functional theory (DFT) calculations have indicated that the Kooi sequence is the most energyfavored configuration.[6,86–88] Nonetheless, Da Silva et al. have suggested that the stability of the configuration requires Ge and Sb atoms to reduce strain energy, leading to the atoms mixing.[89] III) the Matsunaga model:[90] -(Te-Sb/Ge-Te-Sb/Ge-Te-Te-Sb/GeTe-Sb/Ge)-. The controversy on the hexagonal Ge2 Sb2 Te5 structure extends beyond the theoretical domain to experimental observations. Studies have revealed that the electronic and atomic structures of hexagonal Ge2 Sb2 Te5 exhibit high sensitivity to the layer sequence.[91] Additionally, the experimental absorption coefficient has indicated a better fit to the Petrov sequence,[92] however, the broadening Raman peaks [93] and the non-uniform contrast of cationic layers[94,95] in HAADF-STEM images have both indicated a preference for the Matsunaga sequence.[90] Although the atomic model of Ge2 Sb2 Te5 remains under debate, the underlying mechanism of CCPT has been properly investigated based on the existing microstructure. Solid-state reactions, including CCPT, require both the correct thermodynamic conditions to provide driving forces and the right kinetic factors to determine the rate and preferred path.[96] For a reaction to occur, the initial structure should possess a higher energy to create the necessary driving force of reaction. Using Ge2 Sb2 Te5 as an example, the metastable cubic phase has a higher energy than the equilibrium hexagonal phase, thus promoting the stabilization reaction (phase transition) from cubic to hexagonal phases. It is suggested that a diffusionless controlled mechanism is likely to occur during this stabilization process due to the similar atomic sequence in the cubic and hexagonal lattice 2407239 (3 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Figure 2. The atomic model of the cubic structure of Ge2 Sb2 Te5 . a) The cubic atomic model with randomly arranged cations and vacancies. b) The cubic atomic model with alternatively arranged Ge, Sb, and vacancies. Panels are Reproduced with permission.[82] Copyright 2005, American Institute of Physics. structures.[6] This mechanism could involve various processes such as shearing martensitic movement, epitaxial growth, or vacancy ordering mechanisms.[74,97,98] By investigating the structure characteristics of the cubic Ge2 Sb2 Te5 using first-principles calculations, Kim et al.[99] proposed a transition model for the stabilization process (Figure 3a). As the temperature rises, the assembly of Sb and Ge atoms in the cubic phase on the (111) planes triggers the ordering of cations and vacancies, resulting in a contraction in lattice constants and leading to the cubic-to-hexagonal transition (Figure 3b). The stabilization process has also been scrutinized by atomicresolution energy dispersive X-ray spectroscopy and HAADFSTEM.[80,100] VLs are found frequently between the cubic and hexagonal phases and are considered precursors to vdW gaps in the hexagonal phases. EDX mapping revealed weaker Sb signals in VLs but stronger signals in neighboring Ge/Sb planes (Figure 3c). Notably, Te atoms remain nearly immobile during the transition, serving as the structural framework for both phases. These microstructure characterizations propose an energy-effective zipperlike transition mechanism (Figure 3d), Figure 3. Transition model for the stabilization process in Ge2 Sb2 Te5 . a) Atomic disorder-dependent structural stability and lattice constant, and c/a ratio-dependent total energy. b) Schematic transition model for the stabilization process in Ge2 Sb2 Te5 . c) HAADF-STEM image and EDX mapping of the transition region. d) The zipperlike transition model for the stabilization process. Panels (a,b) are Reproduced with permission.[99] Copyright 2015, American Institute of Physics. Panels (c,d) are Reproduced with permission.[100] Copyright 2019, American Physical Society. Adv. Funct. Mater. 2024, 34, 2407239 2407239 (4 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Figure 4. Phase transition from hexagonal to cubic in Ge2 Sb2 Te5 . a–f) Schematic and STEM images of an as-grown (a), a pulse-induced (b), and a repeatedly switched (c) Ge2 Sb2 Te5 NW. Metastabilization process from hexagonal g) to transitional state h) and final cubic i) phases. j) Switching performance of this NW device.[101] k) The energy diagram compares the disordered cubic and hexagonal phases of Ge1 Sb2 Te4 with varying degrees of vacancy layer formation.[103] Configurations marked by red dots are less likely to be observed in experimental settings, as they possess higher energy levels, making them less energetically favorable when compared to other configurations. Panels (a–j) are Reproduced with permission.[101] Copyright 2016, American Chemical Society. Panel (k) is Reproduced with permission.[103] Copyright 2016, American Physical Society. whereby under thermal agitation, minor Ge/Sb atoms break bonds and exchange sites with nearby vacancies in a zipperlike flipping process, leading to Sb-rich atomic arrangements in the intermediate phase. Subsequently, the relaxation process and atomic hoppings result in the creation of vdW gaps. Investigations suggest that the atomic arrangements are somewhat sequential, with Sb atoms migrating before Ge atoms. Ge and Sb atoms respectively favor the inner and the outer layers in the forming stacking blocks.[80] The atomic is arranged closer to the hexagonal structure as the degree of Ge/Sb migration in the intermediate state is greater. In the final, Sb and Ge atoms aggregate in the outer and inner layers, respectively, forming the stacking block of -Te-Sbx /Gey -Te-Gex /Sby Te-Gex /Sby -Te-Sbx /Gey -Te- (x > y), resembling the Matsunaga model.[90] The critical feature of phase transitions that are designed for random access memory is reversible, which ensures the encoded data can be repeatedly erased and written. Therefore, the primary basis for Ge2 Sb2 Te5 to integrate all-crystalline PCRAM is the reversible CCPT, viz. the stabilization reaction from cubic to hexagonal phases and the reversible reaction from hexagonal to cubic phases. Contrary to the stabilization transition, the Adv. Funct. Mater. 2024, 34, 2407239 transformation from hexagonal to cubic phases presents an apparent paradox, as the metastable cubic phase exhibits a higher Gibbs free energy (G) compared to the equilibrium hexagonal phase. The absence of a thermodynamic driving force for this transition is perplexing. Consequently, the change from hexagonal to cubic phase is typically accomplished through a process of melt-quenching amorphization followed by the crystallization of the non-crystalline phase. Remarkably, Lee et al.[101] documented the direct transition from a hexagonal to a cubic phase within a single-crystal hexagonal Ge2 Sb2 Te5 nanowire (NW) using atomicresolution STEM (Figure 4a–j). The term “metastabilization” was coined to distinguish this occurrence from the opposing stabilization process. Lee et al.[101] have delineated a two-step mechanism for metastabilization. The first step is characterized by the relocation of Ge and Sb atoms from their inherent octahedral positions within the mixed layers of Ge and Sb to the tetrahedral sites of vacancy loops (VLs), which is triggered by an applied electrical pulse. The second step involves the reconfiguration of these Ge and Sb atoms, leading to the transformation of an intermediate structure into a metastable cubic configuration. The potential CCPT process via an intermediate metastable local structure in asymmetrical Ge2 Sb2 Te5 NW was further confirmed by 2407239 (5 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de combining the nanoscale X-ray absorption fine structure (XAFS), Raman, and TEM analyses.[102] Meanwhile, a research study was carried out by Privitera et al.[103] focusing on the impact of ion irradiation on both the structural and electronic characteristics of a hexagonal Ge1 Sb2 Te4 film. They discovered that ion irradiation could also induce a metastabilization transition to a rock-salt structure, accompanied by a metal-insulator transition (Figure 4k). Through DFT simulations, Privitera et al. determined that the total energy of hexagonal Ge1 Sb2 Te4 with partial vacancy layer formation was higher compared to corresponding cubic models. These findings provided a preliminary theoretical framework, suggesting that the substantial energy cost linked to the disorder caused by vacancies in the hexagonal phase was the driving force behind the metastabilized structural transformation. Recently, Jiang et al.[78] observed vacancy disordering in hexagonal Ge1 Sb2 Te4 film as a result of electron irradiation. They attributed this disordering to the kinetic knock-on collision effects caused by high-energy electron beams, which generated displacement forces triggering vacancy diffusion. Jiang et al. demonstrated that the vacancy disordering process correlated with the metastabilization process. This transition comprised three distinct stages: I) vacancy diffusion within the trigonal phase, II) alterations in atomic stacking, and III) the elimination of planes rich in vacancy. These stages were elucidated through in situ TEM experiments and DFT nudged elastic band calculations. Furthermore, Jiang et al.[78] summarized and illustrated how vacancy ordering and disordering, induced by either heating effects or displacement forces, played a pivotal role in driving phase transitions in the Ge1 Sb2 Te4 film. 3. CCPT in Binary Chalcogenide Alloys The reversible CCPT in ternary Ge-Sb-Te alloys as discussed above, has garnered significant attention in recent years. Despite the growing interest, the practical application of CCPT in ternary Ge-Sb-Te alloys in PCRAM devices remains limited. In contrast, the reversible CCPT behavior in binary In2 Se3 has been studied extensively and is seen as more mature for potential data storage applications. In2 Se3 is characterized by a variety of crystalline phases, encompassing four prevalent ones (𝛼, 𝛽, 𝛾, and 𝛿) as well as four less frequently encountered ones (𝛼′, 𝛽′, 𝛾′, and 𝜅). Studies on the crystal structure and CCPT of In2 Se3 compounds have extensively been studied through various analytical techniques such as lab-based or synchrotron X-ray diffraction, in situ TEM, differential thermal analysis, and summaries available in works by Li et al.[63] and Han et al.[104] The phase transitions in In2 Se3 are complex, and at times, ambiguous and conflicting. The generally accepted phase transition sequence with increasing temperature is 𝛼 → 𝛽 → 𝛾 → 𝛿. The 𝛼-phase, characterized by a layered structure, and the 𝛾-phase, featuring a defective wurtzite structure, are both considered stable at room temperature.[104] In contrast, the 𝛽 and 𝛿 phases are deemed stable only within specific temperature ranges. This understanding of the phase transitions in In2 Se3 provides valuable insights into the material behavior and potential applications in PCRAM devices. In2 Se3 emerged as a material of interest for binary PCRAM design in 2005, employing the large resistance contrast between the non-crystalline and crystalline phases, in line with the typical n-CCPT framework.[105] Notably, In2 Se3 nanowires exhibit a 105 - Adv. Funct. Mater. 2024, 34, 2407239 fold variation in electric resistivity upon crystallization, offering a significantly larger resistance ratio compared to the conventional Ge2 Sb2 Te5 , which enhances data storage capabilities.[106,107] Furthermore, the higher crystalline resistivity of In2 Se3 nanowires enhances their ability to efficiently deliver switching power.[106] In the initial research on PCRAM utilizing In2 Se3 nanowires or thin films, the resistive switching characteristics were examined concerning both the duration and the magnitude of the electrical pulses applied,[105–107] the specific crystalline phase present in the set state remained elusive due to the coexistence of stable 𝛼and 𝛾-phases at room temperature. A significant breakthrough came when Huang et al. successfully fabricated an Au/In2 Se3 nanowire/Au PCRAM device and tracked the dynamic phase transition process through a reset/set operation using pulsed and DC voltage sweeps within an in situ TEM setup.[108] However, this type of PCRAM based on the typical n-CCPT within In2 Se3 falls outside the scope of this review. Upon annealing, the bulk 𝛼-In2 Se3 transforms into the 𝛽 phase at 473 K. However, when the temperature is reduced back to room temperature, the 𝛽 phase consistently returns to the 𝛼 phase.[109] Therefore, even if the bulk single-crystal In2 Se3 is used for data storage, the encoded data is volatile. In contrast to the bulk, polycrystalline In2 Se3 powders have shown the coexistence of 𝛼 and 𝛽 phases at room temperature, suggesting that the 𝛽 phase may be stabilized as the grain size is reduced.[110] Additionally, it has been noted that 𝛼-In2 Se3 can transform into the 𝛽 phase under a pressure of 0.7 GPa, and this phase change can be maintained even after the pressure is reduced back to ambient levels.[111] These two significant studies offer valuable insights into the development of a stable 𝛽 phase that can persist at room temperature. Soon, thin layers of single-crystal In2 Se3 , with thicknesses ranging from ≈87 nm down to about 4 nm, were exfoliated from crystalline powders (Figure 5[109] ). The transition from the 𝛼 to 𝛽 phase, along with the associated changes in electrical properties due to thermal heating, were detected. It has also been characterized that the structural transition from the high-resistance 𝛼 phase to the low-resistance 𝛽 phase, induced by different forms of energy input (thermal or electrical), shares a remarkable similarity.[52] Zhang et al. synthesized 2D In2 Se3 using chemical vapor deposition and discovered that In2 Se3 layers, varying from a single layer to about 20 layers, could stabilize in the 𝛽 phase with a superstructure at room temperature.[112] Furthermore, recent research has shown that preventing the oxidation of the In2 Se3 surface can lead to the persistence of the 𝛽 phase in both bulk and multi-layer flakes, even after cooling down to room temperature.[113] This finding highlights the potential use of graphene encapsulation as an innovative strategy to maintain the stability of 𝛽-In2 Se3 in open air and at higher temperatures. These findings provide the basis for designing PCRAM-CCPT devices using single-crystal In2 Se3 thin layers, which exhibit a stable 𝛽 phase even at room temperature. A test PCRAM device (Figure 6) was fabricated by layering In2 Se3 thin layers on graphene.[52] Repeatable set and reset programming was accomplished. Analysis of STEM images, selected area electron diffraction (SAED) patterns, and corresponding electron energy-loss spectroscopy (EELS) profiles revealed that the set state consists of pure 𝛽-In2 Se3 , while the reset state unexpectedly manifests as the 𝛾 phase, rather than the anticipated 𝛼-In2 Se3 . In the set state, the 𝛽 phase structure is 2407239 (6 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Figure 5. The CCPT and the associated alterations in the electrical characteristics of single-crystal In2 Se3 thin films. a) Atomic force microscopy image for the as-exfoliated In2 Se3 thin films. Electron diffraction patterns b) and Raman spectra c) are depicted for the as-exfoliated In2 Se3 thin film before and after the annealing. The micro-Raman spectra d) and the electrical resistance measurements e) are shown for an In2 Se3 thin film during both the heating and cooling cycles. f) The phase transition temperature from the 𝛼 to 𝛽 phases with a function of the layer thickness. The open circles on the plot represent the residuals from the data analysis. Panels (a–f) are Reproduced with permission.[109] Copyright 2013, American Chemical Society. Figure 6. Phase-transition characteristics of In2 Se3 -based all-crystalline PCRAM device. a) A schematic and an optical micrograph for an In2 Se3 test device. b) The current-voltage characteristics of the device for the periods before, during, and after the phase transition. c,d) STEM images along with the corresponding SAED patterns for the set and reset state of In2 Se3 devices, respectively. e) The atomic structures for the set and reset states. Panels (a–e) are Reproduced with permission.[52] Copyright 2017, Wiley. Adv. Funct. Mater. 2024, 34, 2407239 2407239 (7 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de distinguished by an octahedral bonding arrangement around each indium (In) atom. Contrastingly, the reset state is characterized by a tetrahedral arrangement, typical of the 𝛾 phase. The 𝛾 to 𝛽 phase transition involves a vacancy reconfiguration process akin to that observed during the stabilization process in Ge2 Sb2 Te5 materials.[52] The transformation from the 𝛽 to 𝛾 phase might theoretically entail the vertical migration of In atoms from their octahedral positions to the nearby van der Waals (vdW) spaces that exist across the selenium (Se) layers. Subsequently, these atoms could undergo lateral diffusion or displacive shifts toward tetrahedral sites, influenced by temperature and strain fields. Thereafter, the vacancies created by this migration would be harnessed for the ensuing structural reconfigurations that culminate in the reset state. In recent years, Mori et al. have documented a multilevel reversible CCPT phenomenon in MnTe, where the 𝛽-phase transitions to the 𝛼-phase via an intermediary phase known as 𝛽′phase. This intermediate phase is marked by minor adjustments in the positions of the tellurium (Te) atoms relative to their arrangement in the 𝛽-phase.[114–120] The transformation from 𝛽 to 𝛽′ is initiated by a puckering mechanism, causing the manganese (Mn) and Te atomic planes to move in opposite directions along the c-axis (refer to Figure 7). Subsequently, the transition from 𝛽′ to 𝛼 is initiated by a buckling mechanism, which results in pairs of Mn and Te atomic planes moving alternately in opposite directions along the [210] crystallographic direction. It is noteworthy that MnTe films not protected by a tungsten (W) capping layer failed to maintain the out-of-plane z-axis alignment during the polymorphic transformation. These observations imply that the stable two-step polymorphic transformation, which maintains the z-axis orientation, is likely affected by the confinement effect of the W capping layer on the MnTe film. Notably, the reversible displacive transformation between the 𝛽′ (wurtzite-type) and 𝛼 (nickeline-type) phases, facilitated by an atomic-plane shuffling mechanism under rapid Joule/laser heating, produces significant electrical and optical contrasts. Furthermore, this approach facilitates the creation of non-volatile memory devices that offer improved energy efficiency and quicker operational speeds when compared to conventional PCMs that experience diffusional nCCPT transitions. Given that displacive transformations are characterized by diffusionless shearing and shuffling of atoms without the need for random atomic diffusion, their rate is notably high.[114] Thus far, the transition from the 2H to the 1T′ phase in MoTe2 has been achieved through the application of heat and mechanical stress. However, the phase transitions that have been reported are volatile, which significantly restricts the potential use of MoTe2 in various applications, including memory devices, reconfigurable electronic circuits, and topological transistors. Recently, a significant breakthrough was made with the first demonstration of a reversible and non-volatile phase transition between the 2H and 1T′ phases in a Te-deficient MoTe2 .[121] The energy gap between these two phases narrows as the concentration of vacancies in Te increases, and when the concentration of Te vacancies surpasses 4%, the 1T′ phase becomes more stable than the 2H phase. In its original, unaltered state, the concentration of Te vacancies is too low for the 1T′ phase to be stable at room temperature, making the 2H phase the more stable configuration. Upon the application of an electric field, there is a localized Adv. Funct. Mater. 2024, 34, 2407239 increase in the concentration of Te vacancies as these vacancies move along the direction of the field. This increase in Te vacancy concentration further reduces the energy difference between the 2H and 1T′ phases, which can trigger the 2H-to-1T′ phase transition. When the direction of the electric field is reversed, the Te vacancies are driven back in the opposite direction. This migration leads to a decrease in the concentration of Te vacancies and an increase in the energy difference between the phases, effectively returning the material to its original 2H phase. Significantly, this phase transition process is highly controllable, exhibiting exceptional performance characteristics such as a cycling endurance surpassing 105 cycles, ultrafast switching times of ≈5 ns for set operation and 10 ns for reset operation, and remarkable retention properties exceeding 105 seconds at an elevated temperature of 85 °C.[121] Recently, Liu et al. reported the reversible cubic-tohexagonal phase transition in yttrium-doped antimony telluride (Sb2 Te3 ).[122] Their research integrated the CCPT with non-crystalline states to establish a novel multi-level storage phase-change system. This system exhibits ultralow power consumption, minimal resistance drift for the lower two states, and competitive operating speed. The authors determined that yttrium plays a crucial role in stabilizing the cubic phase and facilitating the reversible cubic-to-hexagonal transition through the sequential and directional migration of Sb atoms based on atomic-resolution structural characterization and ab initio calculations (Figure 8a–e[122] ). Researchers have provided further validation of the reversible displacive phase transformations in yttrium-doped Sb2 Te3 through the use of ab initio molecular dynamics (AIMD) simulations (Figure 8f[123] ). Specifically, the forward transformation, which is a displacive shift from the cubic to the hexagonal phase, occurs in a two-step mechanism involving shearing and contraction of the Te-Sb-Te-Sb-Te structural units. In contrast, the reverse transformation from the hexagonal back to the cubic phase requires elevated temperatures and the process of atomic diffusion. This reverse transition is set in motion by the diffusion of antimony (Sb) cations from the interlayer regions of the structural units into the interstitial spaces. The incorporation of yttrium dopants serves to enhance the thermal stability of the cubic phase, thus promoting the reverse transformation. These discoveries offer an enhanced comprehension of the displacive transformations that occur in the archetypal material Sb2 Te3 and may yield significant insights for the advancement of PRAM-CCPT technology. The controllable all-crystalline phase transitions in binary chalcogenide alloys, such as In2 Se3 , MnTe, MoTe2 , and Y-Sb2 Te3 , present promising opportunities for the development of energy-efficient and high-speed electronic/photonic phase-change storage, and neuromorphic computing devices.[122–124] 4. CCPT in Superlattice PCMs Simpson and his colleagues[15] developed a type of interfacial PCRAM, based on a superlattice consisting of GeTe and Sb2 Te3 layers with thicknesses ranging between 5 and 40 Å, separated by vdW gaps (Figure 9a,b). In the reset state, Simpson et al. observed ordered lattices and diffraction spots in the interfacial PCRAM adjacent to the electrode (Figure 9c,d). It was noted that interfacial PCRAM in the reset state exhibited a crystalline 2407239 (8 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Figure 7. MnTe-based PCRAM and its CCPT mechanism. a) Voltage-dependent resistance curve for MnTe and Ge2 Sb2 Te5 devices. Cross-sectional diagrammatic (b) and TEM c) image of the MnTe device. d) The SAED patterns are displayed for two regions: the matrix, which is outlined by the red-dotted line (top), and the active region, enclosed by the blue-dotted line (bottom). The crystal structures of MnTe are illustrated for the e) 𝛽-phase, f) 𝛽′-phase, and g) 𝛼-phase. The two-stage displacive transition mechanism is described for the phase changes from 𝛽 to 𝛽′ (h) and 𝛽′ to 𝛼 (i). Panels (a–i) are Reproduced with permission.[114] Copyright 2020, Springer Nature. structure, contrary to conventional chalcogenide alloys that are typically non-crystalline at this state. This unique observation led Simpson et al. to suggest that the transition of interfacial PCRAM was achieved through the flipping of Ge atoms between covalently and resonantly bonded sites (Figure 9e), resulting in a CCPT without melting. Compared to conventional PCRAM based on chalcogenide alloys, interfacial PCRAM offers advan- Adv. Funct. Mater. 2024, 34, 2407239 tages in power consumption, switching speed, and cycling endurance. The reduced energy cost of interfacial PCRAM is attributed to a 95% decrease in entropic losses compared to related alloys.[15] Moreover, the electro-thermal confinement aspect of interfacial PCRAM provides a significant reduction (approximately four-fold) in the effective cross-plane thermal conductivity of the superlattice when compared to polycrystalline Ge2 Sb2 Te5 .[125] 2407239 (9 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Figure 8. The process of metastabilization from the hexagonal to the cubic phase in yttrium-doped antimony telluride. a–d) HAADF-STEM image captures the intermediate state during the metastabilization process. e) The enthalpy, entropy, and energy changes of Y-Sb2 Te3 are associated with varying concentrations of cations within the vdW gap. f) The mechanism of the forward displacive transformation. Panels (a–e) are Reproduced with permission.[122] Copyright 2021, Elsevier. Panel (f) is Reproduced with permission.[123] Copyright 2023, Elsevier. Thermal properties play a more predominant role in heat confinement in mushroom cells than electrical properties.[126] Additionally, the proposed CCPT in interfacial PCRAM involves less atomic diffusion, potentially explaining its superior endurance and faster switching capabilities at the nanoscale. In conclusion, the remarkable phase-transition behavior and low power consumption of interfacial PCRAM offer a new avenue for controlling phase transitions in PCMs and enhancing the performance of PCRAM devices.[127] Chong et al.[128] were the first to introduce and implement the superlattice concept in PCRAM by utilizing alternately arranged GeTe, Sb2 Te3 superlattice layers, and vdW gaps along the z-axis, sparking a surge in research and application of superlattice PCMs following the interfacial PCRAM. The superlattice PCRAM device has recently garnered considerable attention due Adv. Funct. Mater. 2024, 34, 2407239 to its promise to reduce energy while also sustaining commendable performance characteristics, including high storage density, rapid operation speeds, extensive cyclicity, and significant electrical/optical contrast.[129–131] Nonetheless, the atomic structure and the underlying mechanisms of phase transitions within these devices remain subjects of intense discussion and debate. This ongoing discourse has been identified as a major factor that hinders advancements in performance optimization and the creation of innovative PCRAM devices that capitalize on superlattice technology. In early, Petrov[84] and Kooi[85] models that proposed to delineate the structure of Gex Sb2 Te3+x (x = 1, 2, 3) alloys were borrowed to describe the atomic configuration of the layered blocks within the GeTe/Sb2 Te3 superlattice devices (Figure 10a). Following this, Tominaga and Kolobov et al. argued that for the GeTe/ Sb2 Te3 superlattice, separating the GeTe and Sb2 Te3 sublayers to 2407239 (10 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Figure 9. The interfacial PCRAM device and operating mechanism. a) High-resolution TEM image of an as-deposited interfacial PCRAM cell. b) A meltamorphized dome forms above the TiN heater. c) The SAED pattern for the entire interfacial PCRAM structure. d) High-resolution TEM images and SAED patterns are displayed for four distinct regions. e) Two potential models are proposed to explain the 1D motion of germanium (Ge) atoms. Panels (a–e) are Reproduced with permission.[15] Copyright 2011, Springer Nature. create more vdW gaps is essential.[132] This development resulted in the formulation of the Inverted Petrov model, a modification where the order of the GeTe atomic layers is reversed in comparison to the original Petrov model. Additionally, Tominaga et al.[132] proposed the Ferro model, characterized by a bilayer of GeTe with a ferroelectric sequence. Various studies have attempted to determine the most stable phase at different temperatures. The Kooi phase was found to be the most energy-favored at lower temperatures based on AIMD simulations.[6] Conversely, at temperatures exceeding 500 K, the Inverted Petrov model and the Ferro model were identified as energetically favorable.[132] Subsequent research conducted by Yu and Robertson, which involved calculating the enthalpy variations concerning temperature using the phonon dispersion spectrum, demonstrated that the Ferro model had the greatest stability at temperatures exceeding 125 K.[133] Nonetheless, the inconsistency in results obtained from different methods has been ascribed to the inadequate consideration of Ge/Sb intermixing in simulations.[127] The compounds (GeTe)n (Sb2 Te3 )m , which exhibit a blend of Ge and Sb in the cationic layers, have been separately suggested by Karpinsky et al.[134] and Shelimova et al.[135] Subsequent research further confirmed the cationic intermixing in Ge2 Sb2 Te5 alloys.[136–139] Recent microstructure characterization studies[140–143] have demonstrated that the GeTe/Sb2 Te3 superlat- Adv. Funct. Mater. 2024, 34, 2407239 tices deposited via molecular beam epitaxy, pulsed laser deposition, or magnetic sputtering also exhibit intermixed Ge and Sb atoms. Utilizing extended X-ray absorption fine structure spectroscopy and TEM, Momand et al.[140] observed the intermixing of GeTe and Sb2 Te3 at the interfaces and formed Ge-Sb-Te alloys.[141] Notably, as revealed by X-ray diffraction and Raman spectroscopy, the GeTe blocks were found to be incorporated inside the Sb2 Te3 layers, challenging conventional models.[141] Then, a theoretical model was further put forward to explain the formation of these Ge-Sb-Te alloys when GeTe is deposited onto Sb2 Te3 in a superlattice arrangement (Figure 10b[144] ). Kolobov et al.[145] have further shown that the inversion of the Sb-Te terminating layers nearest to the vdW gaps results in a substantial alteration in the density of states. This provides a different explanation for the property differences between the set and reset states in the GeTe/Sb2 Te3 superlattices. The stability of Ge/Sb intermixing is attributed to its ability to enhance configurational entropy and reduce sublattice mismatch between GeTe and Sb2 Te3 . While traditional superlattice models are constructed based on alternating pure GeTe and Sb2 Te3 sublattices, it is suggested that simulations or calculations of the GeTe/Sb2 Te3 superlattice model should take into account the Ge/Sb intermixing.[127] Elucidating the phase transition mechanism is another critical issue for improving the performance of the interfacial PCRAM 2407239 (11 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Figure 10. Atomic structure of interfacial PCRAM and the schematic of transition mechanism. a) Simple superlattice stacking sequences in the case of (GeTe)2 (Sb2 Te3 )1 . b) A step-by-step process of Ge-Sb-Te alloy is formed during the deposition of GeTe onto Sb2 Te3 . Switching model of interfacial PCRAM based on a double c) and a single d) Ge umbrella flip. The proposed routes of the set and reset process for the switching model are proposed by Ohyanagi et al. e) and Tominaga et al. f). Panels (a,c,d) are Reproduced with permission.[140] Copyright 2015, IEEE. Panel (b) is Reproduced with permission.[144] Copyright 2016, American Chemical Society. Panel (e) is Reproduced with permission.[133] Copyright 2015, Springer Nature. and superlattice device. One of the well-known proposals for the superlattice device is the CCPT which was proposed as early as the interfacial PCRAM was invented.[15] Despite numerous hypotheses positing new switching mechanisms to account for the low switching current densities, a consensus has yet to emerge. Two primary proposals exist for the CCPT mechanism, the first involving a reversible phase transition between Petrov and Inverted Petrov models (Figure 10c), and the second proposing a transition between Ferro and Inverted Petrov models (Figure 10d). Initially, Takaura et al. introduced the reversible CCPT mechanism between Petrov and Inverted Petrov,[146,147] where the atomic layer sequence transitions from -Sb-Te-GeTe-vdW-Te-Ge-Te-Sb-Te- to -Sb-Te-vdW-Te-Ge-Ge-Te-vdW-Te/SbTe- through the flipping of Ge atoms into the vdW gaps. However, discrepancies between theoretical calculations and experimental results persist, specifically in the resistance ratio. Yu Adv. Funct. Mater. 2024, 34, 2407239 and Robertson later suggested a two-step process for Ge flipping, with calculated energy barriers indicating challenges in achieving vertical flipping (Figure 10e[133] ). The computed energy thresholds for the initial phase (vertical flipping) and the subsequent phase (lateral movement) were identified as 2.59 and 0.05 electron volts (eV), respectively. However, the substantial energy barrier associated with the first step suggests that the vertical flipping of atoms is not a likely occurrence. As a result, Song et al. proposed that the vertical flipping of Ge atoms is more probable to occur sequentially, meaning that the Ge atoms would flip individually rather than all at once in a simultaneous manner.[148] More recently, Nakamura et al. investigated the switching mechanism by constructing a GeTe/Sb2 Te3 superlattice model and proposed the metastable phase as the set state in a one-step transition process.[149,150] On the other hand, Tominaga et al.[132] put forth a reversible CCPT mechanism between 2407239 (12 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Figure 11. The evolutionary approach to designing vdW heterostructures for interfacial PCRAM. a) The algorithmic process for refining the architecture of vdW heterostructure superlattices. b) A proposed model for the diffusive atomic switching mechanism of Ge atoms. Panels (a,b) are Reproduced with permission.[151] Copyright 2016, The Royal Society of Chemistry. Ferro and Inverted Petrov models, highlighting the transitions from -Sb-Te-Ge-Te-Ge-Te-vdW-Te-Sb-Te- to -SbTe-vdW-Te-Ge-GeTe-vdW-Te-Sb-Te- through a single Ge-layer flip (Figure 10f). Significant energy barriers were identified for the vertical flipping and lateral motion steps, underscoring the challenge of achieving transitions in practice. Using a genetic algorithms (GA) method, Kalikka et al. investigated alternative structures of GeTe/Sb2 Te3 superlattice (Figure 11a[151] ), presenting novel models distinct from the established Petrov, Inverted Petrov, Ferro, and Kooi structures. Their proposed models feature stoichiometric Ge-Sb-Te layer blocks interconnected through vdW interactions, encompassing Ge3 Sb2 Te6 and Ge1 Sb2 Te4 layer blocks. They highlight the phasetransition mechanism involving Ge-atom flipping, where flipping a Ge layer in Ge3 Sb2 Te6 into the vdW interspace adjacent to Ge1 Sb2 Te4 transforms the structure into two vdW-connected Ge2 Sb2 Te5 layer blocks. The calculated energy barriers for this transformation range from 1.83–1.99 eV (Figure 11b[151] ), albeit relatively high for rapid switching. Tominaga et al. proposed that the transition in interfacial PCRAM is driven by a polarized electrical field due to its polar switching mode.[146,149] In contrast, Bolotov et al.[152] investigated the switching of electric conductance triggered by voltage pulses through the use of scanning probe techniques. They noticed a current response that was delayed, occurring between 0.05 to 10 seconds after the applica- Adv. Funct. Mater. 2024, 34, 2407239 tion of the voltage pulse, suggesting an electric field-induced switching mechanism. Conversely, Nakamura et al. contend that the driving force in nonpolar interfacial PCRAM switching is the Joule heat generated by the current rather than the electrical field effect.[149] All the above research suggested that Geatom flipping has a critical role in this CCPT of GeTe/Sb2 Te3 superlattice. Research by Hase et al. and Makino et al. utilizing coherent phonon spectroscopy investigated the transition dynamics in GeTe/Sb2 Te3 superlattice.[153–157] The changes in phonon frequency induced by femtosecond laser were detected and were attributed to the variation of Ge configuration. Furthermore, the frequency change influenced by the laser pulse polarization was further discovered, pointing to the anisotropic flipping of Ge atoms. However, the technique remains limited in capturing the atomic-level details, merely capturing transient states. In contrast, observations from HAADF-STEM characterization reveal the flipping of the Ge/Sb plane into neighboring vdW gaps within the GeTe/Sb2 Te3 superlattice induced by electron exposure.[142,158] Notably, the kinetic energy of an electron beam surpasses that of typical electrical or optical pulses used for switching. While existing models examine Ge atom flipping near vdW gaps in superlattices, they fail to account for the critical factor of Ge/Sb intermixing, significantly contributing to the behavior of the GeTe/Sb2 Te3 superlattice.[140–143] 2407239 (13 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de The proposed crystalline-crystalline transition has been discussed since the invention of the interfacial PCRAM, with some researchers considering it as a potential mechanism.[15] However, Momand et al.[140] questioned the ability of the TEM characterization provided by Simpson et al. to differentiate whether the reset state is partially or entirely crystalline. Contradictory findings to the crystalline-crystalline phase transition hypothesis, such as the commonly observed melt-quench-induced nCCPT,[159] as well as partial melting or amorphization,[160,161] strain-assisted phase transition,[162,163] and stacking fault-assisted metal-insulator transition,[164,165] have also been documented. Recent studies by Yoo et al. suggested a different perspective, proposing that the reset switching in interfacial PCRAM occurs through a transition from the GeTe/Sb2 Te3 superlattice to the cubic Ge2 Sb2 Te5 alloy followed by melt-quenching-free amorphization.[129] Interestingly, the set operation no longer reverts the material to the initial superlattice but to the FCC structure, potentially indicating that the superlattice might not be actively involved in subsequent cycling processes. This finding raises questions about the critical role of interfaces in interfacial PCRAM devices, despite the uncertainties regarding the superlattice’s exact influence on device performance. Supporting this notion, research indicates that both the switching current density and resistance drift coefficient decrease as the superlattice period thickness decreases, reflecting a higher density of interfaces.[166] In conclusion, the debate surrounding the phase change mechanism of interfacial PCRAM and superlattice phase change memory persists due to the intricate manufacturing processes and complex structural considerations. 5. Conclusion and Outlook This systematic review examines the historical development and recent advancements of PCMs featuring reversible CCPT. Analysis is conducted on the properties, structures, and potential switching mechanisms of these PCMs. One of the primary challenges in this area pertains to the ambiguity of the structure and switching mechanisms of PCRAM-CCPT. To address them, it is vital to meticulously compare the structures of the set/reset states in practical devices using advanced structural characterization techniques, such as ultrafast dark-field electron microscopy.[167] In addition to experimental approaches, AIMD has become a valuable theoretical tool for modeling the structural characteristics and dynamic behaviors of chalcogenide alloys over the past few decades.[168–173] However, the computational expense associated with AIMD hinders the transition from material-level simulations to real device-level simulations, presenting a significant obstacle that necessitates the consideration of realistic operating conditions such as active volume fluctuations and applied electric biases.[174,175] The advent of artificial intelligence has led to the development of machine learning-based interatomic potential models, derived from a range of representative configurations. These models combine the efficacy of empirical potentials with the precision of DFT, enabling a deeper understanding of the structural, chemical, and kinetic characteristics of PCMs.[176] Several machine learning potentials, including artificial neural networks and Gaussian approximation potentials (GAP), have been utilized to explore the properties and behaviors of materials such as GeTe[177–180] and Ge2 Sb2 Te5 .[181–184] Through large- Adv. Funct. Mater. 2024, 34, 2407239 scale simulations, it is now possible to elucidate the crystal structure and phase transition mechanisms of CCPT, with a particular focus on unraveling the role of interfaces within superlattice PCRAM. Recently, materials such as Ge-Sb-Te,[101] Y-Sb2 Te3 ,[122] In2 Se3 ,[185] MoTe2 ,[121] and GeTe/Sb2 Te3 superlattices[15] exhibit structural transitions between two crystalline phases potentially affording an insulator-metal transition. However, PCMs with characteristics of reversible CCPT have not received sufficient attention. Compared to typical n-CCPT, CCPT possesses many advanced phase-change natures in data storage and processing. First, CCPT does not undergo a melt-quench process during the reversible phase transition, which gives it the advantage of low power consumption and slight thermal crosstalk.[15] Second, CCPT is a diffusionless transition through shearing and shuffling of atoms and does not require random diffusion of atoms, enabling memory with lower energy, faster operation, and a more controlled phase transition process compared with the conventional diffusional n-CCPT.[114] Third, CCPT has improved cycling endurance than n-CCPT, as cycling failures, including stuck-set (caused by elemental segregation) and stuck-reset (caused by void formation), are essentially caused by atomic migration.[186] Various driving forces such as hole-wind force, electrostatic force, and crystallization-induced segregation have been identified as responsible for the atomic migration, and have a strong effect in regions where the PCMs are molten or undercooled liquid. In contrast, CCPT does not involve a molten or undercooled liquid phase, thus suppressing atomic migration and improving cycling endurance. Fourth, the optical/electrical signal of PCRAM-CCPT is more stable than that of non-crystalline chalcogenide alloys due to the resistance drift issue induced by spontaneous structural relaxation, which impedes conventional chalcogenide alloys for multilevel storage and neuro-inspired computing.[46] Due to the advanced phase-change properties, PCMs featuring reversible CCPT characteristics are poised to revolutionize the development of high-performance binary PCRAM devices. Despite the nascent stage of research on reversible CCPT and the resulting limited performance data, the potential of CCPT is already being realized in superlattice-like devices. Notably, these superlattice-based PCRAM devices demonstrate a remarkable reduction in energy consumption to just 12% of that required by comparable Ge2 Sb2 Te5 -based devices.[15] Additionally, they achieve a significant reduction in operational time by 75%. Most impressively, the cycling performance of these devices has been enhanced by an extraordinary three orders of magnitude, underscoring the promising future of CCPT in this field.[15] In addition, compared to the typical diffusional n-CCPT, the diffusionless CCPT is a more controllable phase transition process. PCMs with characteristics of CCPT thus have greater potential in achieving more quantity and precise intermediate storage states.[32] Combining the characteristics of high stability, low power consumption, and long durability, such PCMs show great application prospects in the field of integrating multilevel data storage devices and neuromorphic computing chips. Unfortunately, PCMs with reversible CCPT, especially those with large storage windows, remain very rare. More efforts should be made to explore new materials with this characteristic, to optimize synthesis strategies, to strengthen structure characterization, and to 2407239 (14 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de reveal phase transition mechanisms, and finally to achieve effective control of reversible CCPT. Material exploration aims to identify and develop new materials with inherent CCPT characteristics, that can offer improved performance metrics over existing PCMs. The direct strategy is refining the synthesis techniques to produce materials with more uniform and desirable properties, which has a great effect on improving the efficiency and endurance of CCPT-type PCRAM devices. Revealing the structural transformation mechanism of CCPT by using the most advanced structural characterization techniques and computational simulation methods is also crucial for achieving effective control over the reversible CCPT process and is essential for the practical implementation of high-performance PCRAM devices. Meanwhile, research on its application in CCPT memory devices and neuromorphic chips should also be carried out as soon as possible. Acknowledgements Financial support was provided by the National Natural Science Foundation of China (Grant Nos. 52332005 and 62304016). B.L. acknowledges support from the Open Research Fund of State Key Laboratory of Materials for Integrated Circuits (No. NKLJC-K2023-07), and Fundamental Research Funds for the Central Universities (Grant No. 62304016). Conflict of Interest The authors declare no conflict of interest. Keywords All-crystalline phase-change memory, Crystalline-crystalline, Phase transition, Phase-change materials, Phase-change memory Received: April 28, 2024 Revised: June 23, 2024 Published online: July 16, 2024 [1] S. R. Ovshinsky, (Energy Conversion Devices INC), United States Patent 3271584, 1966. [2] S. R. Ovshinsky, Phys. Rev. Lett. 1968, 21, 1450. [3] A. V. Kolobov, P. Fons, A. I. Frenkel, A. L. Ankudinov, J. Tominaga, T. Uruga, Nat. Mater. 2004, 3, 703. [4] M. Wuttig, Nat. Mater. 2005, 4, 265. [5] H. F. Hamann, M. O’Boyle, Y. C. Martin, M. Rooks, H. K. Wickramasinghe, Nat. Mater. 2006, 5, 383. [6] Z. M. Sun, J. Zhou, R. Ahuja, Phys. Rev. Lett. 2006, 96, 055507. [7] S. H. Lee, Y. Jung, R. Agarwal, Nat. Nanotechnol. 2007, 2, 626. [8] G. Atwood, Science 2008, 321, 210. [9] Z. M. Sun, J. Zhou, A. Blomqvist, B. Johansson, R. Ahuja, Phys. Rev. Lett. 2009, 102, 075504. [10] D. Loke, T. H. Lee, W. J. Wang, L. P. Shi, R. Zhao, Y. C. Yeo, T. C. Chong, S. R. Elliott, Science 2012, 336, 1566. [11] Z. M. Sun, J. Zhou, R. Ahuja, Phys. Rev. Lett. 2007, 98, 055505. [12] M. Wuttig, N. Yamada, Nat. Mater. 2007, 6, 824. [13] F. Xiong, A. D. Liao, D. Estrada, E. Pop, Science 2011, 332, 568. [14] S. W. Nam, H. S. Chung, Y. C. Lo, L. Qi, J. Li, Y. Lu, A. T. Johnson, Y. Jung, P. Nukala, R. Agarwal, Science 2012, 336, 1561. [15] R. E. Simpson, P. Fons, A. V. Kolobov, T. Fukaya, M. Krbal, T. Yagi, J. Tominaga, Nat. Nanotechnol. 2011, 6, 501. Adv. Funct. Mater. 2024, 34, 2407239 [16] H. S. Wong, S. Salahuddin, Nat. Nanotechnol. 2015, 10, 191. [17] J. Hruska, Intel, Micron reveal Xpoint, a new memory architecture that could outclass DDR4 and NAND, https://www.extremetech. com/g00/extreme/211087-intel-micron-revealxpoint-a-newmemory-architecture-that-claims-to-outclass-both-ddr4and-nand (accessed: July 2015). [18] J. Choe, Intel 3D Xpoint memory die removed from Intel optaneTM PCM (phase change memory), https://www.techinsights.com/ blog/intel-3d-xpoint-memory-die-removed-intel-optanetm-pcmphase-change-memory (accessed: May 2017). [19] Both., https://www.intel.com/content/dam/www/public/us/ en/documents/technology-briefs/what-is-optane-technologybrief.pdf (accessed: July 2024). [20] Intel Optane products, https://www.intel.com/content/www/ us/en/products/details/memory-storage/optane-memory.html, (accessed: July 2024). [21] Micron brings 3D XPoint™ technology to market with the world’s fastest SSD, https://www.nasdaq.com/press-release/micronbrings-3d-xpointtm-technology-to-market-with-the-worlds-fastestssd-2019-10-24, (accessed: October 2019). [22] S. W. Fong, C. M. Neumann, H. S. P. Wong, IEEE Trans. Electron Devices 2017, 64, 4374. [23] M. Suri, O. Bichler, D. Querlioz, B. Traoré, O. Cueto, L. Perniola, V. Sousa, D. Vuillaume, C. Gamrat, B. DeSalvo, J. Appl. Phys. 2012, 112, 054904. [24] C. Ríos, M. Stegmaier, P. Hosseini, D. Wang, T. Scherer, C. D. Wright, H. Bhaskaran, W. H. P. Pernice, Nat. Photonics 2015, 9, 725. [25] D. Kuzum, R. G. Jeyasingh, B. Lee, H. S. Wong, Nano Lett. 2012, 12, 2179. [26] C. D. Wright, P. Hosseini, J. A. V. Diosdado, Adv. Funct. Mater. 2013, 23, 2248. [27] Z. Cheng, C. Ríos, W. H. P. Pernice, C. D. Wright, H. Bhaskaran, Sci. Adv. 2017, 3, 1700160. [28] T. Tuma, A. Pantazi, M. Le Gallo, A. Sebastian, E. Eleftheriou, Nat. Nanotechnol. 2016, 11, 693. [29] Y. Wang, P. Landreman, D. Schoen, K. Okabe, A. Marshall, U. Celano, H. P. Wong, J. Park, M. L. Brongersma, Nat. Nanotechnol. 2021, 16, 667. [30] C. Varnava, Nat. Electron. 2021, 4, 454. [31] K. Roy, A. Jaiswal, P. Panda, Nature 2019, 575, 607. [32] K. Ding, J. Wang, Y. Zhou, H. Tian, L. Lu, R. Mazzarello, C. Jia, W. Zhang, F. Rao, E. Ma, Science 2019, 366, 210. [33] S. Ambrogio, P. Narayanan, H. Tsai, R. M. Shelby, I. Boybat, C. di Nolfo, S. Sidler, M. Giordano, M. Bodini, N. C. P. Farinha, B. Killeen, C. Cheng, Y. Jaoudi, G. W. Burr, Nature 2018, 558, 60. [34] J. Feldmann, N. Youngblood, M. Karpov, H. Gehring, X. Li, M. Stappers, M. Le Gallo, X. Fu, A. Lukashchuk, A. S. Raja, J. Liu, C. D. Wright, A. Sebastian, T. J. Kippenberg, W. H. P. Pernice, H. Bhaskaran, Nature 2021, 589, 52. [35] B. J. Shastri, A. N. Tait, T. Ferreira de Lima, W. H. P. Pernice, H. Bhaskaran, C. D. Wright, P. R. Prucnal, Nat. Photonics 2021, 15, 102. [36] G. Karunaratne, M. Le Gallo, G. Cherubini, L. Benini, A. Rahimi, A. Sebastian, Nat. Electron. 2020, 3, 327. [37] K. Konstantinou, T. H. Lee, F. C. Mocanu, S. R. Elliott, Proc. Natl. Acad. Sci. USA 2018, 115, 5353. [38] A. I. Khan, A. Daus, R. Islam, K. M. Neilson, H. R. Lee, H.-S. P. Wong, E. Pop, Science 2021, 373, 1243. [39] T. Wang, Y. Lu, Y. Li, Appl. Phys. Lett. 2020, 116, 141601. [40] W. Wang, D. Loke, L. Shi, R. Zhao, H. Yang, L. T. Law, L. T. Ng, K. G. Lim, Y. C. Yeo, T. C. Chong, A. L. Lacaita, Sci. Rep. 2012, 2, 360. [41] Z. Yang, B. Li, J. J. Wang, X. D. Wang, M. Xu, H. Tong, X. Cheng, L. Lu, C. Jia, M. Xu, X. Miao, W. Zhang, E. Ma, Adv. Sci. 2022, 9, 2103478. [42] K. Stern, N. Wainstein, Y. Keller, C. M. Neumann, E. Pop, S. Kvatinsky, E. Yalon, Adv. Electron. Mater. 2021, 7, 2100217. 2407239 (15 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de [43] S. Hatayama, Y. Sutou, S. Shindo, Y. Saito, Y. H. Song, D. Ando, J. Koike, ACS Appl. Mater. Interfaces 2018, 10, 2725. [44] J. Y. Raty, W. Zhang, J. Luckas, C. Chen, R. Mazzarello, C. Bichara, M. Wuttig, Nat. Commun. 2015, 6, 7467. [45] M. L. Gallo, D. Krebs, F. Zipoli, M. Salinga, A. Sebastian, Adv. Electron. Mater. 2018, 4, 1700627. [46] W. Zhang, E. Ma, Mater. Today 2020, 41, 156. [47] J. Pries, S. Wei, M. Wuttig, P. Lucas, Adv. Mater. 2019, 31, 1900784. [48] F. Rao, K. Y. Ding, Y. X. Zhou, Y. H. Zheng, M. J. Xia, S. L. Lv, Z. T. Song, S. L. Feng, I. Ronneberger, R. Mazzarello, W. Zhang, E. Ma, Science 2017, 358, 1423. [49] B. Liu, K. Li, J. Zhou, L. Wu, Z. Song, W. Zhao, S. R. Elliott, Z. Sun, J. Mater. Chem. C 2023, 11, 1360. [50] W. Zhang, R. Mazzarello, M. Wuttig, E. Ma, Nat. Rev. Mater. 2019, 4, 150. [51] Z. Wang, H. Wu, G. W. Burr, C. S. Hwang, K. L. Wang, Q. Xia, J. J. Yang, Nat. Rev. Mater. 2020, 5, 173. [52] M. S. Choi, B. K. Cheong, C. H. Ra, S. Lee, J. H. Bae, S. Lee, G. D. Lee, C. W. Yang, J. Hone, W. J. Yoo, Adv. Mater. 2017, 29, 1703568. [53] S. M. Islam, L. Peng, L. Zeng, C. D. Malliakas, D. Y. Chung, D. B. Buchholz, T. Chasapis, R. Li, K. Chrissafis, J. E. Medvedeva, G. G. Trimarchi, M. Grayson, T. J. Marks, M. J. Bedzyk, R. P. H. Chang, V. P. Dravid, M. G. Kanatzidis, J. Am. Chem. Soc. 2018, 140, 9261. [54] P. Zalden, F. Quirin, M. Schumacher, J. Siegel, S. Wei, A. Koc, M. Nicoul, M. Trigo, P. Andreasson, H. Enquist, M. J. Shu, T. Pardini, M. Chollet, D. Zhu, H. Lemke, I. Ronneberger, J. Larsson, A. M. Lindenberg, H. E. Fischer, S. Hau-Riege, D. A. Reis, R. Mazzarello, M. Wuttig, K. Sokolowski-Tinten, Science 2019, 364, 1062. [55] Z. Sun, J. Zhou, H. K. Mao, R. Ahuja, Proc. Natl. Acad. Sci. USA 2012, 109, 5948. [56] A. V. Kolobov, M. Krbal, P. Fons, J. Tominaga, T. Uruga, Nat. Chem. 2011, 3, 311. [57] C. Hu, Z. Yang, C. Bi, H. Peng, L. Ma, C. Zhang, Z. Gu, J. Zhu, Acta Mater. 2020, 188, 697. [58] R. Wang, Z. Song, W. Song, T. Xin, S. Lv, S. Song, J. Liu, InfoMat 2021, 3, 1008. [59] T. H. Lee, S. R. Elliott, Phys. Rev. Lett. 2011, 107, 145702. [60] J. Pries, C. Stenz, L. Schäfer, A. Gutsche, S. Wei, P. Lucas, M. Wuttig, Adv. Funct. Mater. 2022, 32, 2207194. [61] A. Pawbake, C. Bellin, L. Paulatto, K. Beneut, J. Biscaras, C. Narayana, D. J. Late, A. Shukla, Phys. Rev. Lett. 2019, 122, 145701. [62] G. Xing, Y. Li, X. Fan, L. Zhang, W. Zheng, D. J. Singh, J. Appl. Phys. 2017, 121, 225106. [63] J. Li, H. Li, X. Niu, Z. Wang, ACS Nano 2021, 15, 18683. [64] A. Heßler, S. Wahl, T. Leuteritz, A. Antonopoulos, C. Stergianou, C.F. Schön, L. Naumann, N. Eicker, M. Lewin, T. W. W. Maß, M. Wuttig, S. Linden, T. Taubner, Nat. Commun. 2021, 12, 924. [65] M. Zhu, W. Song, P. M. Konze, T. Li, B. Gault, X. Chen, J. Shen, S. Lv, Z. Song, M. Wuttig, R. Dronskowski, Nat. Commun. 2019, 10, 3525. [66] Q. Li, O. Levit, E. Yalon, B. Sun, J. Appl. Phys. 2023, 133, 135105. [67] L. Zheng, W. Song, S. Zhang, Z. Song, X. Zhu, S. Song, J. Alloys Compd. 2021, 882, 160695. [68] Z. He, W. Wu, X. Liu, J. Zhai, T. Lai, S. Song, Z. Song, Mater. Lett. 2016, 185, 286. [69] L. Wu, Z. Song, B. Liu, T. Zhang, F. Rao, J. Shen, F. Wang, S. Feng, J. Non-Cryst. Solids 2007, 353, 4043. [70] Y. Wang, X. Chen, Y. Cheng, X. Zhou, S. Lv, Y. Chen, Y. Wang, M. Zhou, H. Chen, Y. Zhang, Z. Song, G. Feng, IEEE Electron Device Lett. 2014, 35, 536. [71] K. Do, D. Lee, D.-H. Ko, H. Sohn, M.-H. Cho, Electrochem. SolidState Lett. 2010, 13, H284. [72] Y. H. Zheng, M. J. Xia, Y. Cheng, F. Rao, K. Y. Ding, W. L. Liu, Y. Jia, Z. T. Song, S. L. Feng, Nano Res. 2016, 9, 3453. Adv. Funct. Mater. 2024, 34, 2407239 [73] T. Siegrist, P. Jost, H. Volker, M. Woda, P. Merkelbach, C. Schlockermann, M. Wuttig, Nat. Mater. 2011, 10, 202. [74] W. Zhang, A. Thiess, P. Zalden, R. Zeller, P. H. Dederichs, J. Y. Raty, M. Wuttig, S. Blugel, R. Mazzarello, Nat. Mater. 2012, 11, 952. [75] K. S. Siegert, F. R. L. Lange, E. R. Sittner, H. Volker, C. Schlockermann, T. Siegrist, M. Wuttig, Rep. Prog. Phys. 2015, 78, 013001. [76] A. Lotnyk, T. Dankwort, I. Hilmi, L. Kienle, B. Rauschenbach, Nanoscale 2019, 11, 10838. [77] W. Peng, D. M. Smiadak, M. G. Boehlert, S. Mather, J. B. Williams, D. T. Morelli, A. Zevalkink, J. Appl. Phys. 2019, 126, 055106. [78] T.-T. Jiang, X.-D. Wang, J.-J. Wang, Y.-X. Zhou, D.-L. Zhang, L. Lu, C.L. Jia, M. Wuttig, R. Mazzarello, W. Zhang, Acta Mater. 2020, 187, 103. [79] A. Lotnyk, S. Bernütz, X. Sun, U. Ross, M. Ehrhardt, B. Rauschenbach, Acta Mater. 2016, 105, 1. [80] Y. Zheng, Y. Wang, T. Xin, Y. Cheng, R. Huang, P. Liu, M. Luo, Z. Zhang, S. Lv, Z. Song, S. Feng, Commun. Chem. 2019, 2, 13. [81] N. Yamada, T. Matsunaga, J. Appl. Phys. 2000, 88, 7020. [82] Y. J. Park, J. Y. Lee, M. S. Youm, Y. T. Kim, H. S. Lee, J. Appl. Phys. 2005, 97, 093506. [83] S. He, L. Zhu, J. Zhou, Z. Sun, Inorg. Chem. 2017, 56, 11990. [84] I. I. Petrov, R. M. Imamov, Z. G. Pinsker, Sov. Phys. Crystallogr. 1968, 13, 339. [85] B. J. Kooi, J. T. M. De Hosson, J. Appl. Phys. 2002, 92, 3584. [86] X. Yu, J. Robertson, Sci. Rep. 2016, 6, 37325. [87] W. Ibarra-Hernández, J.-Y. Raty, Phys. Rev. B 2018, 97, 245205. [88] J. Singh, G. Singh, A. Kaura, S. K. Tripathi, J. Solid State Chem. 2018, 260, 124. [89] J. L. F. Da Silva, A. Walsh, H. Lee, Phys. Rev. B 2008, 78, 224111. [90] T. Matsunaga, N. Yamada, Y. Kubota, Acta Crystallogr. B 2004, 60, 685. [91] G. Lee, S.-H. Jhi, Phys. Rev. B 2008, 77, 153201. [92] B.-S. Lee, J. R. Abelson, S. G. Bishop, D.-H. Kang, B.-k. Cheong, K.-B. Kim, J. Appl. Phys. 2005, 97, 093509. [93] G. C. Sosso, S. Caravati, C. Gatti, S. Assoni, M. Bernasconi, J. Phys.: Condens. Matter 2009, 21, 245401. [94] A. Lotnyk, U. Ross, S. Bernutz, E. Thelander, B. Rauschenbach, Sci. Rep. 2016, 6, 26724. [95] A. M. Mio, S. M. Privitera, V. Bragaglia, F. Arciprete, C. Bongiorno, R. Calarco, E. Rimini, Nanotechnology 2017, 28, 065706. [96] K. Samwer, Phys. Rep. 1988, 161, 1. [97] W. Zhang, H. S. Jeong, S. A. Song, Adv. Eng. Mater. 2008, 10, 67. [98] Y. J. Park, J. Y. Lee, Y. T. Kim, Appl. Surf. Sci. 2006, 252, 8102. [99] J. Kim, S.-H. Jhi, J. Appl. Phys. 2015, 117, 195701. [100] M. Zhu, K. Ren, L. Liu, S. Lv, X. Miao, M. Xu, Z. Song, Phys. Rev. Mater. 2019, 3, 033603. [101] J. Y. Lee, J. H. Kim, D. J. Jeon, J. Han, J. S. Yeo, Nano Lett. 2016, 16, 6078. [102] M. Ahn, J.-S. Oh, D. Park, H. Jung, S. Park, K. Jeong, D. Kim, N.E. Sung, K.-S. Lee, C.-W. Yang, M.-H. Cho, ACS Appl. Electr. Mater. 2020, 2, 2418. [103] S. M. S. Privitera, A. M. Mio, E. Smecca, A. Alberti, W. Zhang, R. Mazzarello, J. Benke, C. Persch, F. L. A. Via, E. Rimini, Phys. Rev. B 2016, 94, 094103. [104] G. Han, Z. G. Chen, J. Drennan, J. Zou, Small 2014, 10, 2747. [105] H. Lee, D.-H. Kang, L. Tran, Mater. Sci. Eng., B 2005, 119, 196. [106] B. Yu, S. Ju, X. Sun, G. Ng, T. D. Nguyen, M. Meyyappan, D. B. Janes, Appl. Phys. Lett. 2007, 91, 133119. [107] D. Kang, T. Rim, C.-K. Baek, M. Meyyappan, J.-S. Lee, Appl. Phys. Lett. 2013, 103, 233504. [108] Y.-T. Huang, C.-W. Huang, J.-Y. Chen, Y.-H. Ting, K.-C. Lu, Y.-L. Chueh, W.-W. Wu, ACS Nano 2014, 8, 9457. [109] X. Tao, Y. Gu, Nano Lett. 2013, 13, 3501. 2407239 (16 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de [110] S. Popović, A. Tonejc, B. Čelustka, B. Gržeta-Plenković, R. Trojko, J. Appl. Crystallogr. 1979, 12, 416. [111] A. M. Rasmussen, S. T. Teklemichael, E. Mafi, Y. Gu, M. D. McCluskey, Appl. Phys. Lett. 2013, 102, 062105. [112] F. Zhang, Z. Wang, J. Dong, A. Nie, J. Xiang, W. Zhu, Z. Liu, C. Tao, ACS Nano 2019, 13, 8004. [113] N. D. Ignacio, J. Fatheema, Y.-R. Jeon, D. Akinwande, Adv. Electron. Mater. 2023, 10, 2300457. [114] S. Mori, S. Hatayama, Y. Shuang, D. Ando, Y. Sutou, Nat. Commun. 2020, 11, 85. [115] S. Mori, D. Ando, Y. Sutou, Mater. Des. 2020, 196, 109141. [116] S. Mori, S. Hatayama, D. Ando, Y. Sutou, Jpn. J. Appl. Phys. 2021, 60, 045504. [117] M. Kim, S. Mori, Y. Shuang, S. Hatayama, D. Ando, Y. Sutou, Phys. Status Solidi-R 2022, 16, 2100641. [118] S. Mori, Y. Wang, D. Ando, F. Narita, Y. Sutou, Materialia 2022, 24, 101493. [119] S. Mori, H. Tanimura, T. Ichitsubo, Y. Sutou, ACS Appl. Mater. Interfaces 2023, 15, 42730. [120] Y. Yuan, Z. Xu, S. Song, Z. Song, R. Liu, J. Zhai, Appl. Surf. Sci. 2023, 640, 158362. [121] H.-K. He, Y.-B. Jiang, J. Yu, Z.-Y. Yang, C.-F. Li, T.-Z. Wang, D.-Q. Dong, F.-W. Zhuge, M. Xu, Z.-Y. Hu, R. Yang, X.-S. Miao, Mater. Horiz. 2022, 9, 1036. [122] B. Liu, K. Li, W. Liu, J. Zhou, L. Wu, Z. Song, S. R. Elliott, Z. Sun, Sci. Bull. 2021, 66, 2217. [123] K. Li, B. Liu, J. Zhou, S. R. Elliott, Z. Sun, Acta Mater. 2023, 249, 118809. [124] K. Li, L. Peng, L. Zhu, J. Zhou, Z. Sun, Mater. Sci. Semicond. Process. 2021, 135, 106052. [125] H. Kwon, A. I. Khan, C. Perez, M. Asheghi, E. Pop, K. E. Goodson, Nano Lett. 2021, 21, 5984. [126] A. I. Khan, H. Kwon, M. E. Chen, M. Asheghi, H. S. P. Wong, K. E. Goodson, E. Pop, IEEE Electron Device Lett. 2022, 43, 204. [127] X. B. Li, N. K. Chen, X. P. Wang, H. B. Sun, Adv. Funct. Mater. 2018, 28, 1803380. [128] T. C. Chong, L. P. Shi, R. Zhao, P. K. Tan, J. M. Li, H. K. Lee, X. S. Miao, A. Y. Du, C. H. Tung, Appl. Phys. Lett. 2006, 88, 122114. [129] C. Yoo, J. W. Jeon, S. Yoon, Y. Cheng, G. Han, W. Choi, B. Park, G. Jeon, S. Jeon, W. Kim, Y. Zheng, J. Lee, J. Ahn, S. Cho, S. B. Clendenning, I. V. Karpov, Y. K. Lee, J. H. Choi, C. S. Hwang, Adv. Mater. 2022, 34, 2207143. [130] J. Shen, S. Lv, X. Chen, T. Li, S. Zhang, Z. Song, M. Zhu, ACS Appl. Mater. Interfaces 2019, 11, 5336. [131] J. Tominaga, Phys. Status Solidi-R 2018, 13, 1800539. [132] J. Tominaga, A. V. Kolobov, P. Fons, T. Nakano, S. Murakami, Adv. Mater. Interfaces 2014, 1, 1300027. [133] X. Yu, J. Robertson, Sci. Rep. 2015, 5, 12612. [134] O. G. Karpinsky, L. E. Shelimova, M. A. Kretova, J. P. Fleurial, J. Alloys Compd. 1998, 268, 112. [135] L. E. Shelimova, O. G. Karpinskii, M. A. Kretova, V. I. Kosyakov, V. A. Shestakov, V. S. Zemskov, F. A. Kuznetsov, Inorg. Mater. 2000, 36, 768. [136] T. Matsunaga, R. Kojima, N. Yamada, K. Kifune, Y. Kubota, M. Takata, Appl. Phys. Lett. 2007, 90, 161919. [137] T. Matsunaga, R. Kojima, N. Yamada, K. Kifune, Y. Kubota, M. Takata, Chem. Mater. 2008, 20, 5750. [138] T. Matsunaga, N. Yamada, Phys. Rev. B 2004, 69, 104111. [139] P. Urban, M. N. Schneider, L. Erra, S. Welzmiller, F. Fahrnbauer, O. Oeckler, CrystEngComm 2013, 15, 4823. [140] J. Momand, R. Wang, J. E. Boschker, M. A. Verheijen, R. Calarco, B. J. Kooi, Nanoscale 2015, 7, 19136. [141] B. Casarin, A. Caretta, J. Momand, B. J. Kooi, M. A. Verheijen, V. Bragaglia, R. Calarco, M. Chukalina, X. Yu, J. Robertson, F. Adv. Funct. Mater. 2024, 34, 2407239 R. L. Lange, M. Wuttig, A. Redaelli, E. Varesi, F. Parmigiani, M. Malvestuto, Sci. Rep. 2016, 6, 22353. [142] A. Lotnyk, I. Hilmi, U. Ross, B. Rauschenbach, Nano Res. 2018, 11, 1676. [143] P. Kowalczyk, F. Hippert, N. Bernier, C. Mocuta, C. Sabbione, W. Batista-Pessoa, P. Noé, Small 2018, 14, 1704514. [144] R. Wang, V. Bragaglia, J. E. Boschker, R. Calarco, Cryst. Growth Des. 2016, 16, 3596. [145] A. V. Kolobov, P. Fons, Y. Saito, J. Tominaga, MRS Adv. 2018, 3, 3413. [146] T. Ohyanagi, M. Kitamura, M. Araidai, S. Kato, N. Takaura, K. Shiraishi, Appl. Phys. Lett. 2014, 104, 252106. [147] N. Takaura, T. Ohyanagi, M. Tai, M. Kinoshita, K. Akita, T. Morikawa, H. Shirakawa, M. Araidai, K. Shiraishi, Y. Saito, J. Tominaga, presented at IEEE International Electron Devices Meeting, San Francisco, CA, USA, December 2014. [148] Y.-S. Song, J. Kim, S.-H. Jhi, J. Appl. Phys. 2017, 121, 095102. [149] H. Nakamura, I. Rungger, S. Sanvito, N. Inoue, J. Tominaga, Y. Asai, Nanoscale 2017, 9, 9386. [150] N. Inoue, H. Nakamura, Faraday Discuss. 2019, 213, 303. [151] J. Kalikka, X. Zhou, J. Behera, G. Nannicini, R. E. Simpson, Nanoscale 2016, 8, 18212. [152] L. Bolotov, Y. Saito, T. Tada, J. Tominaga, Sci. Rep. 2016, 6, 33223. [153] M. Hase, P. Fons, K. Mitrofanov, A. V. Kolobov, J. Tominaga, Nat. Commun. 2015, 6, 8367. [154] M. Hase, Y. Miyamoto, J. Tominaga, Phys. Rev. B 2009, 79, 174112. [155] K. Makino, Y. Saito, P. Fons, A. V. Kolobov, T. Nakano, J. Tominaga, M. Hase, Sci. Rep. 2016, 6, 19758. [156] K. Makino, J. Tominaga, M. Hase, Opt. Express 2011, 19, 1260. [157] K. Makino, J. Tominaga, A. V. Kolobov, P. Fons, M. Hase, Appl. Phys. Lett. 2012, 101, 232101. [158] A. Lotnyk, U. Ross, T. Dankwort, I. Hilmi, L. Kienle, B. Rauschenbach, Acta Mater. 2017, 141, 92. [159] K. L. Okabe, A. Sood, E. Yalon, C. M. Neumann, M. Asheghi, E. Pop, K. E. Goodson, H. S. P. Wong, J. Appl. Phys. 2019, 125, 184501. [160] T. Ohyanagi, N. Takaura, AIP Adv. 2016, 6, 105104. [161] X.-P. Wang, X.-B. Li, N.-K. Chen, Q.-D. Chen, X.-D. Han, S. Zhang, H.-B. Sun, Acta Mater. 2017, 136, 242. [162] J. Kalikka, X. Zhou, E. Dilcher, S. Wall, J. Li, R. E. Simpson, Nat. Commun. 2016, 7, 11983. [163] X. Zhou, J. Kalikka, X. Ji, L. Wu, Z. Song, R. E. Simpson, Adv. Mater. 2016, 28, 3007. [164] N.-K. Chen, X.-B. Li, X.-P. Wang, S.-Y. Xie, W. Q. Tian, S. Zhang, H.-B. Sun, IEEE Trans. Nanotechnol. 2018, 17, 140. [165] A. V. Kolobov, P. Fons, Y. Saito, J. Tominaga, ACS Omega 2017, 2, 6223. [166] A. I. Khan, X. Wu, C. Perez, B. Won, K. Kim, P. Ramesh, H. Kwon, M. C. Tung, Z. Lee, I. K. Oh, K. Saraswat, M. Asheghi, K. E. Goodson, H. P. Wong, E. Pop, Nano Lett. 2022, 22, 6285. [167] T. Danz, T. Domröse, C. Ropers, Science 2021, 371, 371. [168] J. Akola, R. O. Jones, Phys. Rev. B 2007, 76, 235201. [169] S. Caravati, M. Bernasconi, T. D. Kühne, M. Krack, M. Parrinello, Appl. Phys. Lett. 2007, 91, 171906. [170] M. Xu, Y. Q. Cheng, H. W. Sheng, E. Ma, Phys. Rev. Lett. 2009, 103, 195502. [171] J. Kalikka, J. Akola, R. O. Jones, Phys. Rev. B 2016, 94, 134105. [172] Y. Xu, Y. Zhou, X. D. Wang, W. Zhang, E. Ma, V. L. Deringer, R. Mazzarello, Adv. Mater. 2022, 34, 2109139. [173] J. Y. Raty, M. Schumacher, P. Golub, V. L. Deringer, C. Gatti, M. Wuttig, Adv. Mater. 2018, 31, 1806280. [174] G. Wang, Z. Sun, Sci. Bull. 2023, 68, 3105. [175] Y. Zhou, W. Zhang, E. Ma, V. L. Deringer, Nat. Electron. 2023, 6, 746. [176] V. L. Deringer, M. A. Caro, G. Csányi, Adv. Mater. 2019, 31, 1902765. 2407239 (17 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de [177] G. C. Sosso, G. Miceli, S. Caravati, J. Behler, M. Bernasconi, Phys. Rev. B 2012, 85, 174103. [178] G. C. Sosso, G. Miceli, S. Caravati, F. Giberti, J. Behler, M. Bernasconi, J. Phys. Chem. Lett. 2013, 4, 4241. [179] S. Gabardi, S. Caravati, G. C. Sosso, J. Behler, M. Bernasconi, Phys. Rev. B 2015, 92, 054201. [180] G. C. Sosso, D. Donadio, S. Caravati, J. Behler, M. Bernasconi, Phys. Rev. B 2012, 86, 104301. [181] F. C. Mocanu, K. Konstantinou, T. H. Lee, N. Bernstein, V. L. Deringer, G. Csányi, S. R. Elliott, J. Phys. Chem. B 2018, 122, 8998. [182] A. P. Bartók, M. C. Payne, R. Kondor, G. Csányi, Phys. Rev. Lett. 2010, 104, 136403. [183] K. Konstantinou, F. C. Mocanu, T. H. Lee, S. R. Elliott, Nat. Commun. 2019, 10, 3065. [184] K. Konstantinou, F. C. Mocanu, J. Akola, S. R. Elliott, Acta Mater. 2022, 223, 117465. [185] T. Li, Y. Wang, W. Li, D. Mao, C. J. Benmore, I. Evangelista, H. Xing, Q. Li, F. Wang, G. Sivaraman, A. Janotti, S. Law, T. Gu, Adv. Mater. 2022, 34, 2108261. [186] D. Gao, B. Liu, Physics 2017, 47, 153. Bin Liu is an associate professor at National Key Laboratory of Spintronics, Hangzhou International Innovation Institute of Beihang University, China. He received his Ph.D. degree of Materials Science from Beihang University in 2021. His current research interests are the structure and properties of phase-change materials for random access memory. Kaiqi Li received his Ph.D. degree of Materials Science from Beihang University in 2024. His research interests are multi-scale simulations of phase-change random access memory and the transport theory of 2D devices. Jian Zhou is a professor at the School of Materials Science and Engineering, Beihang University, China. He obtained his Ph.D. from the Institute of Metal Research, Chinese Academy of Sciences in 2003. He worked at RWTH Aachen University (Germany), Royal Institute of Technology (Sweden), and Xiamen University (China), before joining Beihang University. His research interests are intermetallics, thermoelectric materials, low dimensional magnetic materials and 2D transition metal carbides and borides (MXenes and MBenes). He has published over 180 peer reviewed SCI papers. Adv. Funct. Mater. 2024, 34, 2407239 2407239 (18 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.afm-journal.de Zhimei Sun is a Cheung Kong Scholar Chair Professor at School of Materials Science and Engineering of Beihang University, China. She received her Ph.D. of Materials Science from Institute of Metal Research (CAS) in 2002, and after which she worked at RWTH Aachen University (Germany) and Uppsala University (Sweden) from 2002 to 2007, and at Xiamen University (China) from 2007 to 2013. Her research includes phase-change memory materials, high-performance structural materials and 2D transition metal carbides/borides. Adv. Funct. Mater. 2024, 34, 2407239 2407239 (19 of 19) © 2024 Wiley-VCH GmbH 16163028, 2024, 44, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adfm.202407239 by University Of Science, Wiley Online Library on [17/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com