Scalable and sustainable N-Si-Ge-Te Ovonic threshold switching devices for energy-efficient artificial neuron applications

Scalable and sustainable N-Si-Ge-Te Ovonic threshold switching devices for energy-efficient artificial neuron applications | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Scalable and sustainable N-Si-Ge-Te Ovonic threshold switching devices for energy-efficient artificial neuron applications Jong-Souk Yeo, Siwon Park, Chaebin Park, Su-Bong Lee, Sangyeop Kim This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7764312/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Neuromorphic computing, inspired by biological nervous systems, yields high energy efficiency and data throughput by integrating computation and storage within memory crossbar arrays. A key requirement for neuromorphic hardware is an artificial neuron capable of low-power, high-frequency operation. Ovonic threshold switch (OTS) devices have attracted attention due to their scalability and intrinsic capacitance, enabling simple circuitry to demonstrate leaky integrate-and-fire (LIF) behavior. This study proposes a sustainable and scalable OTS device fabricated with non-toxic, industry-friendly materials and provides insights into the roles of individual elements in bond formation, correlating with enhanced electrical performance (J off = 2.3 ∙ 10 − 8 MA/cm²). Finally, our optimized NSGT OTS device demonstrates low-power spiking operation, achieving 0.56 pJ/µm 2 per spike. These findings establish stoichiometric guidelines for designing high-performance Te-based OTS devices for energy-efficient neuromorphic computing. Physical sciences/Materials science/Materials for devices/Electronic devices Physical sciences/Materials science/Nanoscale materials/Electronic properties and materials Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction The rapid growth of artificial intelligence (AI) applications has intensified the demand for energy-efficient, high-throughput computing architectures that overcome the technical limitations of conventional von-Neumann architecture 1 . Neuromorphic or brain-inspired computing, modelled on the principles of biological neurons and synapses in human thinking, has emerged as a promising approach, offering notable performance gains in energy efficiency and data throughput 2 – 5 . In-memory crossbar architectures, analogous to biological synapses, enable simultaneous storage and computation of data, significantly reducing memory bandwidth requirements and enhancing throughput and energy efficiency in data-intensive AI workloads 6 – 8 . As the core element of neuromorphic computing hardware, an artificial neuron processes incoming electrical spike signals from pre-synaptic neurons and determines their transmission to interconnected post-synaptic neurons 9 . A crucial aspect of such artificial neurons is their ability to operate within low-power, high-frequency domains, ensuring applicability for neuromorphic computing. Extensive research has explored artificial neurons using scalable electronics, ranging from simple transistor latches 10 and complex CMOS circuits 11 to resistive-switching devices, to achieve neuro-morphic behavior with low energy overheads 12 , 13 . Among various approaches, artificial neurons based on Ovonic threshold-switch (OTS) 14 devices have recently attracted attention due to scalability, fast switching, and low energy consumption 15 , 16 . Their intrinsic capacitance, derived from the metal-insulator-metal (MIM) structure, further enables a simple one selector-one resistor (1S-1R) scheme to emulate the leaky integrate-and-fire (LIF) dynamics of ideal neurons. A key challenge in advancing OTS-based artificial neurons lies in material engineering to optimize device properties for low-energy neuromorphic operations 17 . Since the origin of OTS behavior remains unresolved, material optimization of OTS devices requires precise control of material stoichiometry and comprehensive simulations to understand the correlation between material properties and device characteristics within the amorphous chalcogenide structure. Numerous efforts have been made to enhance OTS device performance through various chalcogenide systems. Binary chalcogenide OTS devices have demonstrated outstanding switching performances and stabilities 15 , 18 – 20 , and revealed composition-dependent behaviors associated with each element’s distinct role 21 – 23 . These insights have further progressed the design of multi-element chalcogenide systems, from ternary to quaternary OTS devices, enabling more comprehensive optimization of OTS performances 24 – 27 . However, reliance on toxic dopants such as arsenic, in dramatically improving OTS performances, particularly regarding the balance between selectivity and thermal stability, raises sustainability concerns 28 , 29 . This study examines the elemental contributions in quaternary N-Si-Ge-Te (NSGT) OTS devices through material analysis and ab initio simulations, thus optimizing to demonstrate devices with outstanding performance, low off-state current density (2.3 ∙ 10 − 8 MA/cm²) and high thermal stability (> 400°C). Comparative characterizations from binary Si-Te (ST), ternary Si-Ge-Te (SGT) to quaternary NSGT stoichiometric systems reveal how each element uniquely shapes bond configurations and molecular orbitals, enhancing electrical performance and thermal stability. The effects of each bond were analyzed via density of states (DoS) and charge density of each bond based on Density Functional Theory (DFT), with main bond models informed by spectral analysis. Furthermore, OTS pulse measurements further indicate potential for neuromorphic applications. This study establishes stoichiometric guidelines for tailoring selector characteristics. Results and Discussions Te-based OTS devices for low-power artificial neuron Spiking neural networks (SNNs) replicate the operating principle of biological neurons: 1) dendrites (memristors) receive pre-synaptic spike signals from the previous counterparts, 2) soma/OTS neurons integrate and fires these signals above a threshold, and 3) axons (OTS devices with high selectivity) acts as artificial neurons when an appropriate circuit is designed (Fig. 1a). Neuromorphic hardware for SNN operation primarily comprises synaptic devices (memristive devices), which physically store conductance g W for weight multiplication, and neuronal devices (threshold switches), which control input signal propagation to subsequent layers. The development of artificial neural networks (ANNs) is driven by the need for energy-efficient, low-power computing. Achieving this objective requires artificial neurons that operate at low voltage and current, necessitating the precise selection of OTS device elements. Among OTS-applicable chalcogen elements (S, Te, Se), tellurium is notable for low-V th , low-I off operation 43,44 . Furthermore, compound elements N, Si, and Ge ensure environmental sustainability and CMOS-compatibility, avoiding toxic dopants such as arsenic (Fig. 1b). A key performance limitation in amorphous chalcogenide-based OTS devices is the trade-off between thermal stability (T x , crystallization temperature) and off-current density (J off , off-current at half-V th divided by device area), as shown in Fig. 1c, Table T1 18,20,23,29-33 . Thermal stability is defined as the temperature at which the device loses threshold-switching property due to irreversible crystallization. Off-current is defined as the sub-threshold current flow, governed by Poole-Frenkel conduction, which is mainly influenced by critical device dimensions such as area and thickness, as well as material parameters as trap distribution and electrical bandgap 45 . To focus on the materialistic view of engineering OTS devices, off current density can also be considered as the criteria for examining OTS performance. Analysis of previously studied chalcogenides highlights the trade-off as the major bottleneck to achieving high stability and energy efficiency in OTS devices. In this work, material characterization and ab-initio simulation are conducted to address this limitation, guiding the material design of optimized NSGT devices. The proposed NSGT OTS device demonstrates ultra-low off-current density (J off = 2.3 ∙ 10 -8 MA/cm²) and high thermal stability (T x > 400°C) compatible with back-end fabrication. It also exhibits low-power leaky LIF neuron behavior, with normalized power consumption per spike reaching 0.56 pJ/μm 2 (Fig. 1d) 34-42 . The device yields tunable spiking frequency depending on the load resistor, up to maximum frequency of 8.6 MHz, outperforming many previously reported artificial neuron devices. (detailed in Table T2, Supplementary Information) Device structure and electrical characteristics The OTS device is fabricated in a via-hole structure with a global 100 nm thick tungsten (W) bottom and top electrodes with the switching medium as an amorphous chalcogenide layer (ST, SGT, and NSGT), forming a vertical MIM structure. The effective device dimension is formed by selectively etching the dielectric SiO 2 layer with a feature area of 25 μm 2 (Fig. 2a; fabrication details in Fig. S1, Supplementary Information). Cross-sectional analysis via scanning electron microscopy (SEM; Fig. 2b) and high-resolution transmission electron microscopy (HR-TEM; Fig. 2c) verifies uniform deposition of the chalcogenide layer (~20 nm thickness). Fast-Fourier transform (FFT) analysis of the amorphous chalcogenide layer reveals the absence of crystalline sites with no observable diffraction peaks. (Fig. 2c, inset). OTS devices exhibit a sharp decrease in resistance at the threshold voltage (V th ) to switch to the ON state, and a subsequent increase in resistance at the hold voltage (V h ) to return to the OFF state. Representative DC I-V measurement of ST, SGT, and NSGT devices (feature area of 25 μm 2 and 20 nm thickness), as illustrated in Fig. 2d, reveals a discrete trend in sub-threshold I off depending on the constituent elements, with ST having the highest I off (26 μA), followed by SGT (0.74 μA), and NSGT with the lowest I off . The NSGT device exhibits outstanding I-V characteristics, having a high-resistance state (HRS) with I off = 47 nA (@ 1/2V th ) and a rapid transition to a low-resistance state (LRS) at V th = 0.75 V. Cumulative device measurements confirm low cell-to-cell variation across ST, SGT, and NSGT devices, with average I off values of 60 μA (ST), 5.29 μA (SGT), and 39 nA (NSGT), respectively (Fig. 2e; cumulative I-V curves in Fig. S2, Supplementary Information). The device demonstrates a 5 ns switching delay time from LRS to HRS under a triangular pulse. (Fig. S3, Supplementary Information). Thermal stability in SGT (Fig. 2d) decreased to < 300 °C, as shown by XRD analysis of equivalent chalcogenide films annealed to high temperatures. Crystallization peak at 28° was observed in SGT film of X-ray Diffraction (XRD) pattern, losing the film’s as-deposited amorphous state. In contrast, no crystallization peak appeared in ST and NSGT films, even at temperatures above 400°C. This behavior is attributed to low bond enthalpies 46 of Ge (H Ge-Ge = 186 kJ/mol, H Ge-Te = 192 kJ/mol) calculated using Pauling’s relation. On the other hand, ST film possesses high bond enthalpies of Si (H Si - Te = 215 kJ/mol), which is believed to result in a higher thermal stability compared to SGT. Finally, N-doping is performed to optimize device parameters 47,48 , forming Si-N (H Si - N = 395 kJ/mol) and Ge-N (H Ge-Ge = 354 kJ/mol) bonds that stabilize the chalcogenide’s microstructure. This modification enhances the device performance, yielding a selectivity of 3.9 ∙ 10 5 , a delay time of 5 ns after arrival of V th , and thermal stability > 400°C, satisfying back-end-of-line (BEOL) requirements. Optical bandgaps and local structures of amorphous ST, SGT, and NSGT This section investigates why NSGT OTS selectors exhibit superior selectivity and thermal stability compared to ST and SGT OTS. Structural differences among the three OTS materials, which may influence the properties of OTS selectors, are analyzed using X-ray photoelectron spectroscopy (XPS) and Raman scattering spectroscopy. In the Te 3d region, Te 2- peaks (Te-Te bonds, located at 583.6 eV and 573.2 eV) 48-50 and Te 4+ bonds (located at 586.8 eV and 576.5 eV) 48,51-53 are consistently observed across ST, SGT, and NSGT films (Fig. 3a). Te-Te bonds dominate in ST (57%), compared to SGT (39%) and NSGT (7%) (Fig. 3c). Raman analysis between 50 cm – 1 and 200 cm – 1 regions shows a similar trend: peaks A and B correspond to the E g vibrational mode of Te-Te bonds 54,55 , peak D to the A 1 expansion mode of the Te-Te bond 54-56 , and peak E to the E TO mode of Te helical chain 48,54,56,57 . Detailed wavenumbers and area fractions for each deconvoluted Raman peak are provided in Table T3, Supplementary Information. The fraction of Raman peaks related to Te-Te bonds (peaks A, B, D, and E) decreases from ST (93%) to SGT (51%) and NSGT (43%), reflecting Ge-Te bond formation through Ge-doping. Peak C at 123 cm – 1 corresponds to Ge-Te units 53,57-59 in SGT and NSGT by Ge-doping. XPS analysis indicates Te-rich compositions in SGT and NSGT, consistent with tetrahedral Ge-Te bond configurations, such as GeTe 4 and GeTe 3 Ge 59 . In particular, GeTe 4 forms a stable structural motif stabilized by sp 3 hybridization, with Ge’s four valence electrons fully shared 60,61 . Unlike Ge-doping, N-doping does not induce direct bonding with Te; instead, N incorporates as Ge-N and Si-N, confirmed in Ge 3d, N 1s, and Si 2p spectra of NSGT (Fig. S4, Supplementary Information). Although not directly bonded to Te, N substitution at Te sites in the Ge-Te configuration can modify the structure and reduce Te-Te bond fractions. These structural variations significantly alter the electronic structure, producing distinct optical bandgaps, as determined by the linear fitting of the near absorption edge in the Tauc plots of the UV-vis spectroscopy (Fig. 3g) for each material. The optical bandgap increases from 0.725 eV (ST) to 0.856 eV (SGT) and 1.583 eV (NSGT). In OTS materials, subthreshold conduction 62 is governed by the Poole-Frenkel mechanism 45 , which is bandgap-dependent. The observed bandgap trend in this study aligns well with the I off tendency in Fig. 2d–e and is attributed to the combined effects of bond configuration variations and quantitative changes in bond fractions. Further insights are provided through ab initio calculations. As confirmed in existing studies, Te is essential for OTS operation. However, off-current behavior varies with the compound material and composition 19,20,23,44,63 . In this study, a consistent trend is observed, where reducing Te–Te bonds increase the optical bandgap and decreases I off . This is attributed to Te–Te bonds containing lone pairs (LPs) with weakly bound electrons that may have influenced conduction in the subthreshold region 14,23,64 . Furthermore, the thermal stability of OTS material based on amorphous chalcogenides is generally determined by its crystallization temperature (T x ), since nonlinear conduction typically occurs in the amorphous state 23 . T x of OTS material can be predicted using the ratio of bond enthalpies of constituent bond states, based on theoretical models 46 (Method C1, Supplementary Information). Although the model does not precisely predict absolute values, it reliably indicates trends. SGT, containing Ge-related bonds with relatively lower bond enthalpy, exhibits a lower T x than ST. In contrast, NSGT, containing only N-related bonds with relatively higher enthalpy, shows a higher T x , indicating that its high thermal stability originates from the presence of bonds with high bond enthalpy, consistent with previous research 47 . This trend aligns with XRD results (Fig. 2f) after annealing up to 400°C, the reference point for BEOL processes. Structural models and f irst-principles simulations of amorphous ST, SGT, and NSGT Density Functional Theory (DFT) calculations are performed to examine bond configuration changes and the effects of Ge/N doping on the mobility gap using SIESTA (Spanish Initiative for Electronic Simulations with Thousands of Atoms). The Generalized-Gradient Approximation (GGA-PBE) functional is employed to describe the exchange and correlation of the systems. Amorphous material structures (Fig. 4a–c) are generated using the melt-quench-relaxation method (see Methods section for details). The light blue curve in structural models represents the charge density distribution with an isovalue of 0.5 e/ų. The calculated radial distribution functions (RDFs) of all three structures confirm the absence of long-range order, a typical feature of amorphous materials (Fig. S5 in Supplementary Information). Structural models are verified by their agreement with bond distributions reported in previous spectroscopic studies. The amorphous ST, SGT, and NSGT networks exhibit Te–Te bonds (Fig. 4g), corresponding to the dominant peak in all RDFs. Moreover, Ge–Te tetrahedral units (Fig. 4e) and N-related bonds (Fig. 4f), highlighted by red squares in Fig. 4a–c, are also reflected in the RDFs. The calculated electronic density of states (DOS) of ST, SGT, and NSGT are presented in Fig. 4g–i. Consistent with previous studies, amorphous chalcogenide materials exhibit a mobility gap with an exponential distribution between the conduction and valence states 43,49,65,66 . Importantly, the mobility gap increases progressively from ST to SGT and NSGT, and this trend agrees well with the optical bandgap extracted from the Tauc plot (Fig. 3e). The absolute values of the mobility gap are lower than experimentally measured optical bandgaps, due to the bandgap underestimation of GGA functional. This progressive widening of the mobility gap indicates that further structural/electronic interpretation is necessary to examine local bonding arrangements and their correlation with electronic states governing electronic properties in OTS devices. In Fig. 4g, conduction and valence band edges of ST are dominantly formed by Te p-orbital as in projected DoS (PDoS, dashed line). This matches well with the low charge density distribution (0.58 e/ų) of the Te chain, denoted by the double blue curve near the Te atoms, which corresponds to lone-pair from Te p-orbital (p-LP). The PDoS of the s-orbital is observed far below the valence band maximum (Fig S6, Supplementary Information). The trap state observed just under the conduction band edge in ST is mainly attributed to the lone pairs of Te atoms in the Te–Te bond (Te-1 – Te-2), and the high contribution of these non-bonding lone pairs in the PDoS is clearly visible 49 . The inset of Fig. 4g further supports this interpretation by showing spatially localized charge density concentrated around Te-1 – Te-2. Such shallow trap states are known to facilitate carrier leakage in the subthreshold region of OTS devices, linking the electronic structure of ST directly to its relatively high off-current. With Ge-doping, combined orbital states between Ge and Te appear in the PDoS of SGT (Fig. 4h), reducing the contribution of the Te p-LP. Consistent with Fig. 3a–b, Ge incorporation decreases the density of Te–Te bonds and promotes the formation of Ge–Te bonds. The major structural motifs in SGT include a Ge-centered Ge₂Te₃ tetrahedral configuration, in which Te-3 and Te-4 participate in corner-sharing connections with neighboring Ge atoms. The PDoS confirms that these Te orbitals participate in Ge–Te bonding and contribute to the occupied states above the valence band edge, rather than forming trap states under the conduction band edge as in ST. Even the terminal Te atom (Te-5), which is not part of a Te–Te chain but instead terminates through a Ge connection, exhibits a valence-band-tail state rather than the deeply localized non-bonding LP state observed in ST. This redistribution of Te-LP-driven states leads to a wider mobility gap in SGT and explains the observed reduction of off-current. NSGT also shows a reduction in Te p-LP and a corresponding widening of the mobility gap. However, unlike ST and SGT, which exhibit trap states predominantly under the conduction band edge, NSGT exhibits an additional trap state between the Fermi level and the valence band edge. N-doping induces strong electron localization around N atoms 48 , as confirmed by the charge density distribution. It appears to substitute at the corners of the Ge–Te structural units, where its high electronegativity induces structural distortions. Consequently, the Ge–Te unit with N atoms slightly changes in the Ge-centered bond angle (Fig. S7, Supplementary Information). The trap states are therefore attributed to localized states arising from distortions in the Ge–Te hybrid orbital and rearrangements of Te p-LP states. Nevertheless, because these N-driven trap states lie closer to the valence band rather than under the conduction band edge, they tend to capture hole carriers rather than facilitate leakage, providing a mechanism for reducing the off-current rather than increasing it. Across the full ST → SGT → NSGT series, the evolution of Te lone-pair states shows a strong correlation with the formation of valence-band tails. Direct visualization of representative local structures reveals that the charge density associated with Te-LP decreases nearly linearly from ST to NSGT. In parallel, the slope of the valence-band tail in the DOS becomes steeper, resulting in an expanded mobility gap. Initially, in ST, Te-LP generates trap states under the conduction band, but as Te atoms become increasingly involved in Ge–Te or N–Ge–Te bonding environments, these Te-LP-driven states migrate toward the valence-band-tail region, reducing their contribution to shallow trap states and thus broadening the mobility gap. This establishes a direct structural–electronic–device link: Te-LP-driven shallow trap states produce low valence-band-tail slopes and high off-current, whereas suppression and relocation of LP states steepen the tail, widen the mobility gap, and improve off-current performance. It is well known that OTS properties are strongly affected by trap states located within the mobility gap 43,49,65,66 . In our selector device as well, Te-LP-induced traps govern the off-current behavior. In ST, the shallow Te-LP trap with a small slope in the valence-band tail increases the off-current, whereas Ge and N incorporation progressively suppresses Te-LP-driven states, which in turn widens the mobility gap and suppresses the off-current. Therefore, control over Te-LP—not only the composition of constituent elements but the local bonding topology that governs the spatial distribution of LP states—serves as the electronic-structure basis for improving the on/off ratio of Te-based OTS materials. Energy- e fficient n euronal a pplication of the NSGT OTS d evice Spiking neural networks (SNNs) are energy-efficient, low-power networks that propagate information via spike signals across neuron layers. A critical requirement for realizing SNN into a hardware system is an artificial neuron capable of operating at low voltage and current. Resembling biological neurons, such an artificial neuron can be fabricated with a pre-neuron synapse resistor (R s ), an OTS device representing the soma’s threshold determining the spike propagation, and a load resistor (R L ) (Fig. 5a). Our optimized NSGT OTS device, featuring low subthreshold current and operating voltage, exhibits typical LIF neuronal characteristics with high energy efficiency and controlled spiking (Fig. 5b). LIF behavior involves a leaky charging of the OTS’s intrinsic capacitance, in which leaky firing is mostly suppressed by the optimization of N-Si-Ge-Te OTS device having low subthreshold current, followed by an abrupt firing and a slow discharging to load resistor. The overall circuit can thus be paralleled to RC circuit model with series resistance R s , membrane capacitance C OTS , and load resistance R L , with reset switch as the OTS device. The fabricated OTS-based artificial neuron exhibits tunable spike characteristics depending on load resistor (R L ), synaptic resistor (R s ), and voltage amplitude (V in ) (Fig. 5c–e). In Fig. 5c, spike frequency increases with decreasing load resistor (R L ), consistent with the time constant relation ( ) governing the discharging profile of individual spikes. FFT analysis (Fig. 5d) reveals first harmonic peaks for each R L condition, indicating an inverse relation between R L and spike frequency, due to higher R L inducing slower current discharge as proportional to R L C OTS . Likewise, the effect of synapse resistor (R S ) on spike frequency, governing the charging profile of an individual spike, reveals that higher R S induces slower charging to C OTS , similarly leading to lower spike frequency (detailed in Fig. S8, Supplementary Information). Furthermore, increasing the applied voltage amplitude raises the spike probability and frequency of the NSGT artificial neuron (Fig. 5e). Such a trend is due to the increased current inflow to the parasitic capacitor as input voltage is increased, which stimulates the charge-discharge behavior to higher frequencies with more frequent spikes. Throughout the observations, the probabilistic nature of the OTS artificial neuron is consistently observed, which is presumably caused by the amorphous atomic structure of the chalcogenide film changing the switching path at every cycle, resulting in a subtle fluctuation of the threshold voltage, and leading to varying LIF behavior. Conclusion This study investigates the impact of stoichiometric variations on the electrical and thermal properties of Te-based OTS materials composed of environmentally friendly elements (Ge, Si, N; excluding As and Sb). In all three materials, Te p-lone pairs dominate the band edges, while Ge doping forms Ge–Te tetrahedral structures that reduce their influence and widen the mobility gap. Although N does not directly bond with Te, it lowers the overall Te p-lone pair content and distorts the GeTe structure, creating mid-gap states that trap carriers in this p-type material and effectively suppress the off-current density down to 2.3 ∙ 10 − 8 MA/cm 2 . Compared with previous reports, the proposed artificial neuron with optimized leakage properties highlights its potential to operate at ultralow energy of 0.56 pJ/µm 2 and ultrahigh spike frequency of 8.6 MHz, despite large device area. Overall, the results imply strong potentials for the optimized NSGT materials to be applicable for scalable, sustainable and energy-efficient computing systems. Methods Device fabrication and electrical characterization The fabricated device structure consists of a dry-oxidized Si wafer substrate with a universal bottom W electrode (100 nm) deposited by DC magnetron sputtering. SiO 2 dielectric barrier is deposited by plasma-enhanced chemical vapor deposition (PE-CVD), selectively etched by ICP-RIE for feature area formation, followed by deposition of chalcogenide layer (20 nm) via reactive RF magnetron co-sputtering of Si, Ge, and Te targets in Ar + N 2 reactive atmosphere. Finally, the top W electrode (100 nm) is deposited by DC magnetron sputtering (Fig. S1 , Supplementary Information). Fabricated OTS device’s I-V characteristic is analyzed using Keithley 4200–SCS parametric analyzer under I c.c . = 1 mA. Delay time is measured in AC pulse sweep mode via pulse measurement unit (PMU) for pulse generation and measurement. The time resolution of the measurement unit of 5 ns limits the minimum detectable delay to 5 ns. X-ray diffraction for thermal stability evaluation Thermal stability is assessed by depositing chalcogenide films on SiO 2 wafer substrates and annealing at T max ranging from R.T. to 400°C in increments of 100°C inside a vacuum chamber with a rapid thermal processing unit. Annealing conditions are as follows: ramp rate = 1°C/s, T max hold time = 300 s. After annealing treatment, its crystallinity is measured using Rigaku XRD instrument. X-ray photoelectron spectroscopy (XPS) and Raman spectroscopy measurements For microstructural analysis, sample films are deposited onto W-deposited SiO 2 wafer substrates using reactive RF co-sputtering. XPS measurements are conducted using the Thermo Scientific K-Alpha X-ray photoelectron spectrometer system with a monochromatized Al Kα source at an operating voltage of 12 kV, operating current of 3 mA, pass energy of 200 eV, and sampling area of 400 µm. Peak binding energies are referenced to NIST Standard Reference Database 20 v4.1. Raman spectroscopy measurements are conducted using Nanobase XPER-RF Raman spectrometer, with laser wavelength of 532 nm. Cs-corrected transmission electron microscope measurement OTS device is prepared as a lamella sample via FIB (focused ion-beam) milling. JEM-ARM 200F transmission electron microscope with Cs-correction is used under an operating voltage of 200 kV and emission current of 117.7 mA. Atomic model simulation First-principles calculations are performed using the SIESTA package within the Materials Square 67 platform by Virtual Lab Inc. Amorphous OTS material models are generated using molecular dynamics simulations in LAMMPS 68 with a melt-quench-relaxation method. The structure is initially heated from 298 K to 1200 K over 50 ps and then quenched from 1200 K to 300 K at a cooling rate of 10 K/ps. The model is further relaxed with gamma-point sampling of the Brillouin zone at 200 K to reach an optimized atomic configuration (atomic force < 0.02 eV/Ang). Each cell contained 64 atoms, reflecting the composition as characterized by XPS. The electron-electron exchange-correlation energy is treated using generalized gradient approximation (GGA) functional in Perdew–Burke–Ernzerhof (PBE) form. Cutoff energy is set to 300 eV, and SCF tolerance value to 10 − 5 . Declarations Acknowledgements The authors acknowledge financial supports from the Ministry of Trade, Industry, and Energy (MOTIE) and Korea Evaluation Institute Industrial Technology (KEIT) (Project No. 10080625), the Korea Semiconductor Research Consortium (KSRC) program, Samsung Electronics Co., Ltd. (Project No. IO2102021-08356-01), the BK21 FOUR (Fostering Outstanding Universities for Research) funded by the Ministry of Education (MOE) and National Research Foundation (NRF) of Korea, and the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT). (No.2023R1A2C2006811). Funding Declaration Open Access funding is enabled by the BK21 FOUR (Fostering Outstanding Universities for Research), funded by the Ministry of Education (MOE) and National Research Foundation (NRF) of Korea. Author Contributions S.P. and C.P. initially conceived the concept and conceptualized the specific experiment. S.P. performed the DFT calculations and characterized the OTS materials using XRD, XPS, Raman, and UV-vis measurements, with support from S.-B.L. and S.K. The manuscript was prepared by S.P., C.P., S.-B.L., and J.-S.Y. S.P. and C.P. designed the figures. J.-S.Y. guided the research. All authors have discussed the results and given comments on the manuscript. Competing Interests The authors declare no competing interests. Data Availability All data supporting the findings of this study are present in the main article or in supplementary information files. Original data or any additional data related to the study are available from the corresponding author upon request. References Ivanov, D., Chezhegov, A., Kiselev, M., Grunin, A. & Larionov, D. Neuromorphic artificial intelligence systems. Frontiers in Neuroscience 16 (2022). https://doi.org/10.3389/fnins.2022.959626 Upadhyay, N. K. et al. Emerging Memory Devices for Neuromorphic Computing. Advanced Materials Technologies 4 , 1800589 (2019). https://doi.org/10.1002/admt.201800589 Mehonic, A. et al. Memristors—From In‐Memory Computing, Deep Learning Acceleration, and Spiking Neural Networks to the Future of Neuromorphic and Bio‐Inspired Computing. Advanced Intelligent Systems 2 , 2000085 (2020). https://doi.org/10.1002/aisy.202000085 Aguirre, F. et al. Hardware implementation of memristor-based artificial neural networks. Nat Commun 15 , 1974 (2024). https://doi.org/10.1038/s41467-024-45670-9 Roy, K., Jaiswal, A. & Panda, P. Towards spike-based machine intelligence with neuromorphic computing. Nature 575 , 607-617 (2019). Yang, J. J., Strukov, D. B. & Stewart, D. R. Memristive devices for computing. Nat Nanotechnol 8 , 13-24 (2013). https://doi.org/10.1038/nnano.2012.240 Xia, Q. & Yang, J. J. Memristive crossbar arrays for brain-inspired computing. Nature Materials 18 , 309-323 (2019). https://doi.org/10.1038/s41563-019-0291-x Wang, Z. et al. Resistive switching materials for information processing. Nature Reviews Materials 5 , 173-195 (2020). https://doi.org/10.1038/s41578-019-0159-3 Han, J. K., Yun, S. Y., Lee, S. W., Yu, J. M. & Choi, Y. K. A Review of Artificial Spiking Neuron Devices for Neural Processing and Sensing. Advanced Functional Materials 32 , 2204102 (2022). https://doi.org/10.1002/adfm.202204102 Pazos, S. et al. Synaptic and neural behaviours in a standard silicon transistor. Nature 640 , 69-76 (2025). https://doi.org/10.1038/s41586-025-08742-4 Indiveri, G. et al. Neuromorphic silicon neuron circuits. Front Neurosci 5 , 73 (2011). https://doi.org/10.3389/fnins.2011.00073 Zhang, X. et al. An artificial spiking afferent nerve based on Mott memristors for neurorobotics. Nature Communications 11 (2020). https://doi.org/10.1038/s41467-019-13827-6 Park, S. O., Jeong, H., Park, J., Bae, J. & Choi, S. Experimental demonstration of highly reliable dynamic memristor for artificial neuron and neuromorphic computing. Nat Commun 13 , 2888 (2022). https://doi.org/10.1038/s41467-022-30539-6 Ovshinsky, S. An introduction to ovonic research. Journal of Non-Crystalline Solids 2 , 99-106 (1970). Jia, S. et al. Ultrahigh drive current and large selectivity in GeS selector. Nat Commun 11 , 4636 (2020). https://doi.org/10.1038/s41467-020-18382-z Lee, J. et al. An Artificial Tactile Neuron Enabling Spiking Representation of Stiffness and Disease Diagnosis. Advanced Materials 34 , 2201608 (2022). https://doi.org/10.1002/adma.202201608 Ding, Y. et al. Engineering Spiking Neurons Using Threshold Switching Devices for High-Efficient Neuromorphic Computing. Front Neurosci 15 , 786694 (2021). https://doi.org/10.3389/fnins.2021.786694 Chekol, S. A., Yoo, J. & Hwang, H. Thermally Stable Te-based Binary OTS Device for Selector Application. 2018 Non-Volatile Memory Technology Symposium (NVMTS) (2018). https://doi.org/10.1109/NVMTS.2018.8603103 Yoo, J. et al. Te-based binary OTS selectors with excellent selectivity (105 ) endurance (108 ) and thermal stability (450C). 2018 IEEE Symposium on VLSI Technology (2018). https://doi.org/10.1109/VLSIT.2018.8510681 Koo, Y., Lee, S., Park, S., Yang, M. & Hwang, H. Simple binary ovonic threshold switching material SiTe and its excellent selector performance for high-density memory array application. IEEE Electron Device Letters 38 , 568-571 (2017). https://doi.org/10.1109/LED.2017.2685435 Kim, S. D. et al. Effect of Ge Concentration in GexSe1-x Chalcogenide Glass on the Electronic Structures and the Characteristics of Ovonic Threshold Switching (OTS) Devices. ECS Solid State Letters 2 , Q75-Q77 (2013). https://doi.org/10.1149/2.001310ssl Velea, A. et al. Te-based chalcogenide materials for selector applications. Sci Rep 7 , 8103 (2017). https://doi.org/10.1038/s41598-017-08251-z Koo, Y. & Hwang, H. Zn1− xTex ovonic threshold switching device performance and its correlation to material parameters. Scientific Reports 8 , 11822 (2018). Cheng, H. Y. et al. An Ultra High Endurance and Thermally Stable Selector based on TeAsGeSiSe Chalcogenides Compatible with BEOL IC Integration for Cross-Point PCM. 2017 IEEE International Electron Devices Meeting (IEDM) (2017). https://doi.org/10.1109/IEDM.2017.8268310. Verdy, A. et al. High Temperature Stability and Performance Analysis of N-doped Ge-Se-Sb based OTS Selector Devices (supplementary). 2018 IEEE International Memory Workshop (IMW) (2019). https://doi.org/10.1109/IMW.2018.8388834 Laguna, C. et al. Ge Content Optimization in Ge(SbSe)N OTS Materials for Selector Applications. IEEE Transactions on Electron Devices 69 , 6277-6283 (2022). https://doi.org/10.1109/ted.2022.3203368 Verdy, A. et al. Effect of Nitrogen on the Amorphous Structure and Subthreshold Electrical Conduction of GeSeSb‐Based Ovonic Threshold Switching Thin Films. physica status solidi (RRL) – Rapid Research Letters 14 (2019). https://doi.org/10.1002/pssr.201900548 Gu, R. et al. How arsenic makes amorphous GeSe a robust chalcogenide glass for advanced memory integration. Scripta Materialia 218 , 114834 (2022). https://doi.org/10.1016/j.scriptamat.2022.114834 Lee, M.-J. et al. A plasma-treated chalcogenide switch device for stackable scalable 3D nanoscale memory. Nature communications 4 , 2629 (2013). https://doi.org/10.1038/ncomms3629 Avasarala, N. S. et al. in 2018 IEEE Symposium on VLSI Technology. 209-210 (IEEE). Li, X., Yuan, Z., Lv, S., Song, S. & Song, Z. Extended endurance performance and reduced threshold voltage by doping Si in GeSe-based ovonic threshold switching selectors. Thin Solid Films 734 , 138837 (2021). https://doi.org/10.1016/j.tsf.2021.138837 Verdy, A. et al. in 2018 IEEE International Memory Workshop (IMW). 1-4 (IEEE). Yoo, J. et al. 3D stackable and scalable binary ovonic threshold switch devices with excellent thermal stability and low leakage current for high‐density cross‐point memory applications. Advanced Electronic Materials 5 , 1900196 (2019). https://doi.org/10.1002/aelm.201900196 Lee, M. et al. Simple artificial neuron using an ovonic threshold switch featuring spike-frequency adaptation and chaotic activity. Physical Review Applied 13 , 064056 (2020). https://doi.org/10.1103/PhysRevApplied.13.064056 Lee, D. et al. Various threshold switching devices for integrate and fire neuron applications. Advanced Electronic Materials 5 , 1800866 (2019). Ding, Y. et al. Engineering spiking neurons using threshold switching devices for high-efficient neuromorphic computing. Frontiers in neuroscience 15 , 786694 (2022). https://doi.org/10.3389/fnins.2021.786694 Wang, L., Zhang, L., Hua, S., Fu, Q. & Guo, X. Leaky integrate-and-fire and oscillation neurons based on ZnO diffusive memristors for spiking neural networks. Science China Materials 68 , 1212-1219 (2025). https://doi.org/10.1007/s40843-024-3236-6 Choi, H. et al. Spiking neural network integrated with impact ionization field‐effect transistor neuron and a ferroelectric field‐effect transistor synapse. Advanced Materials 37 , 2406970 (2025). https://doi.org/10.1002/adma.202406970 Han, J.-K. et al. Mimicry of excitatory and inhibitory artificial neuron with leaky integrate-and-fire function by a single MOSFET. IEEE Electron Device Letters 41 , 208-211 (2019). https://doi.org/10.1109/LED.2019.2958623 Han, J.-K. et al. Cointegration of single-transistor neurons and synapses by nanoscale CMOS fabrication for highly scalable neuromorphic hardware. Science Advances 7 , eabg8836 (2021). https://doi.org/10.1126/sciadv.abg8836 Dutta, S., Kumar, V., Shukla, A., Mohapatra, N. R. & Ganguly, U. Leaky integrate and fire neuron by charge-discharge dynamics in floating-body MOSFET. Scientific reports 7 , 8257 (2017). https://doi.org/10.1038/s41598-017-07418-y Tuma, T., Pantazi, A., Le Gallo, M., Sebastian, A. & Eleftheriou, E. Stochastic phase-change neurons. Nature nanotechnology 11 , 693-699 (2016). https://doi.org/10.1038/NNANO.2016.70 Jia, S. et al. Ultrahigh drive current and large selectivity in GeS selector. Nature Communications 11 , 4636 (2020). https://doi.org/10.1038/s41467-020-18382-z Zhao, Z. et al. Chalcogenide ovonic threshold switching selector. Nano-Micro Letters 16 , 81 (2024). https://doi.org/10.1007/s40820-023-01289-x Ielmini, D. & Zhang, Y. Analytical model for subthreshold conduction and threshold switching in chalcogenide-based memory devices. Journal of Applied Physics 102 (2007). https://doi.org/10.1063/1.2773688 Lankhorst, M. Modelling glass transition temperatures of chalcogenide glasses. Applied to phase-change optical recording materials. Journal of non-crystalline solids 297 , 210-219 (2002). https://doi.org/10.1016/S0022-3093(01)01034-1 Gu, R. et al. Nitrogen: A promising doping strategy for high-performance ovonic threshold switching selectors. Journal of Alloys and Compounds 1005 , 176011 (2024). https://doi.org/10.1016/j.jallcom.2024.176011 Lee, S.-B., Park, C., Park, S.-H., Lee, Y. & Yeo, J.-S. Effect of nitrogen doping on characteristics of SiTe Ovonic threshold switch for selectors. Materials Science in Semiconductor Processing 178 (2024). https://doi.org/10.1016/j.mssp.2024.108470 Hatayama, S. et al. Origins of midgap states in Te-based Ovonic threshold switch materials. Acta Materialia 258 , 119209 (2023). https://doi.org/10.1016/j.actamat.2023.119209 Robert, B., Flaud, V., Escalier, R., Mehdi, A. & Vigreux, C. XPS study of Ge-Se-Te surfaces functionalized with organosilanes. Applied Surface Science 607 , 154921 (2023). https://doi.org/10.1016/j.apsusc.2022.154921 Reese, M. O. et al. Intrinsic surface passivation of CdTe. Journal of Applied Physics 118 (2015). https://doi.org/10.1063/1.4933186 Li, H.-Y. & Lii, K.-H. Hydrothermal synthesis, crystal structures, and X-ray photoelectron spectroscopy of lead tellurium (iv) and tellurium (vi) oxycompounds: Ba 3 PbTe 6 O 16 and Na 2 Pb 9 (μ 6-O) 2 (Te 2 O 10) 2. Dalton Transactions 50 , 8189-8195 (2021). https://doi.org/10.1039/D1DT00566A Lan, W., Cao, L., Fu, Y., Fang, J. & Wang, J. Structure and optical properties of GeTe film controlled by amorphous to crystalline phase transition. Vacuum 197 , 110847 (2022). https://doi.org/10.1016/j.vacuum.2021.110847 Jagannatha, K. & Das, C. Switching, optical, Raman, and Morphological investigations on Sn doped SiTe glasses and thin films. Thin Solid Films 759 , 139472 (2022). https://doi.org/10.1016/j.tsf.2022.139472 Rao Gunti, S. & Asokan, S. Thermodynamic, Raman and electrical switching studies on Si15Te85-xAgx (4≤ x≤ 20) glasses. Journal of Applied Physics 111 (2012). https://doi.org/10.1063/1.3682759 Zhao, Z. et al. Volatile to Non‐Volatile Switching Transition in Chalcogenides. Advanced Functional Materials , 2423940 (2025). https://doi.org/10.1002/adfm.202423940 Kumar, R. A., Mousa, H. A., Lekha, P. C., Mahmoud, S. T. & Qamhieh, N. in Journal of Physics: Conference Series. 012018 (IOP Publishing). Ji, X. et al. Titanium-induced structure modification for thermal stability enhancement of a GeTeTi phase change material. RSC Advances 5 , 24966-24974 (2015). https://doi.org/10.1039/C4RA11504J Andrikopoulos, K. et al. Raman scattering study of the a-GeTe structure and possible mechanism for theamorphous to crystal transition. Journal of physics: condensed matter 18 , 965 (2006). https://doi.org/10.1088/0953-8984/18/3/014 Kolobov, A. V., Fons, P. & Tominaga, J. Local instability of p-type bonding makes amorphous GeTe a lone-pair semiconductor. Physical Review B—Condensed Matter and Materials Physics 87 , 155204 (2013). https://doi.org/10.1103/PhysRevB.87.155204 Kolobov, A., Fons, P. & Tominaga, J. Understanding phase-change memory alloys from a chemical perspective. Scientific reports 5 , 13698 (2015). https://doi.org/10.1038/srep13698 Velea, A. et al. Te-based chalcogenide materials for selector applications. Scientific reports 7 , 8103 (2017). https://doi.org/10.1038/s41598-017-08251-z Koo, Y., Baek, K. & Hwang, H. in 2016 IEEE symposium on VLSI technology. 1-2 (IEEE). Kastner, M., Adler, D. & Fritzsche, H. Valence-alternation model for localized gap states in lone-pair semiconductors. Physical Review Letters 37 , 1504 (1976). https://doi.org/10.1103/PhysRevLett.37.1504 Wu, R. et al. The role of arsenic in the operation of sulfur-based electrical threshold switches. Nature Communications 14 , 6095 (2023). https://doi.org/10.1038/s41467-023-41643-6 Noé, P. et al. Toward ultimate nonvolatile resistive memories: The mechanism behind ovonic threshold switching revealed. Science Advances 6 (2020). https://doi.org/10.1126/sciadv.aay2830 Inc., V. L. Materials Square , 2017.01.01). Plimpton, S. Fast parallel algorithms for short-range molecular dynamics. Journal of computational physics 117 , 1-19 (1995). https://doi.org/10.1006/jcph.1995.1039 Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryCommNat.docx Supplementary Information Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7764312","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":564561684,"identity":"6a3e0c34-ff00-4bcd-9c7c-70e7dfcea85f","order_by":0,"name":"Jong-Souk Yeo","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEUlEQVRIie2RMUsDMRiGvxioS6Srh4P+hByBQKlwf6WlEJcDB5eAy0HhukRdK/gj/AnKB+0SnA8q2EkQOpxbxKW5OwWXYEfBPEPyEt6HL/ABRCJ/EFJ0N2/yuo20PUc7KZTvpHzTtHuHPx7CCp1d4ZvTKPhyurhk+jnr7wOpHajz4MfMkxoyi5LbhVox+zq+nQJNDOSDIqTMcylIiae8yuXqoMQRR4AjAM2DU7ySfjTKy0ZeeCXLEOjnL4pY+6bkFZPUB3JPoeen5GHFWEmZPROJVZPkzuJ4jqQcGK6CSjoz4t3pYXq9xMd6ozHr3yBWTk/CSvG1jpMHgD3WTS66PQU49qup2+B7xIWLkUgk8o/ZAtbSVtD4ikfYAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0001-6452-2404","institution":"Yonsei University","correspondingAuthor":true,"prefix":"","firstName":"Jong-Souk","middleName":"","lastName":"Yeo","suffix":""},{"id":564561685,"identity":"55f56bdf-6ed2-4195-b089-c65d186a0888","order_by":1,"name":"Siwon Park","email":"","orcid":"https://orcid.org/0009-0002-2695-9829","institution":"Yonsei University","correspondingAuthor":false,"prefix":"","firstName":"Siwon","middleName":"","lastName":"Park","suffix":""},{"id":564561686,"identity":"4273b42b-3d70-4348-8028-a6d309d13735","order_by":2,"name":"Chaebin Park","email":"","orcid":"","institution":"Yonsei University","correspondingAuthor":false,"prefix":"","firstName":"Chaebin","middleName":"","lastName":"Park","suffix":""},{"id":564561687,"identity":"3a9e8607-6d36-4168-8253-703f2059448d","order_by":3,"name":"Su-Bong Lee","email":"","orcid":"","institution":"Yonsei University","correspondingAuthor":false,"prefix":"","firstName":"Su-Bong","middleName":"","lastName":"Lee","suffix":""},{"id":564561688,"identity":"7ef7201e-19b2-45c3-88eb-bbe3312a5ff2","order_by":4,"name":"Sangyeop Kim","email":"","orcid":"","institution":"Yonsei University","correspondingAuthor":false,"prefix":"","firstName":"Sangyeop","middleName":"","lastName":"Kim","suffix":""}],"badges":[],"createdAt":"2025-10-02 07:00:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7764312/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7764312/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99312086,"identity":"e183b2b0-c26e-490c-814c-71ca31eedbdb","added_by":"auto","created_at":"2025-12-31 16:17:53","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3486567,"visible":true,"origin":"","legend":"","description":"","filename":"ManuscriptCommNat.docx","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/36fdb363acd08d57bc600180.docx"},{"id":98961036,"identity":"e24b8a37-715b-4f81-9b83-94d20626d928","added_by":"auto","created_at":"2025-12-24 17:27:39","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6382,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSMAT251155T.json","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/027df56a0441229851167974.json"},{"id":99311206,"identity":"151109ba-cce1-44f9-afee-f772feadfd5d","added_by":"auto","created_at":"2025-12-31 16:14:06","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2365329,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryCommNat.docx","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/13c3be335c632d9b5c880aa3.docx"},{"id":99311501,"identity":"44d6475f-57d9-4309-b870-79f3ebce3619","added_by":"auto","created_at":"2025-12-31 16:15:55","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":143121,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSMAT251155T0enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/048f4d49c7d97244c652a61c.xml"},{"id":98961037,"identity":"0006117a-8f08-4f96-b88a-edf6d3d739ff","added_by":"auto","created_at":"2025-12-24 17:27:39","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":530928,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/c172490ac1b1661c6732af77.png"},{"id":99311479,"identity":"4ef3cb22-a075-430c-b5cd-dcb805e16e43","added_by":"auto","created_at":"2025-12-31 16:15:45","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":645208,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/360972b042022841100601a7.png"},{"id":98961050,"identity":"1c44a95f-9f18-4e4f-bb09-16aaef09d0b2","added_by":"auto","created_at":"2025-12-24 17:27:39","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":663726,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/a7aff50484f1b368f5a4e477.png"},{"id":99311247,"identity":"5cf6846a-33f0-4d25-93f5-9155ebae984b","added_by":"auto","created_at":"2025-12-31 16:14:11","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1105251,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/e62fc4d27289633f3fd15326.png"},{"id":99311203,"identity":"05f25553-df14-4504-b8d6-7585ea4fc87a","added_by":"auto","created_at":"2025-12-31 16:14:06","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":394968,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/2589389249e9396b2a55dcd0.png"},{"id":99311119,"identity":"830bdaa1-56a7-4bb3-8c4f-bf2668774b8b","added_by":"auto","created_at":"2025-12-31 16:13:54","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":113732,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/969114b90121427f6ae06f78.png"},{"id":98961042,"identity":"44e86093-7e7a-4073-8df0-ae550943b9bb","added_by":"auto","created_at":"2025-12-24 17:27:39","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":97853,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/687c8f70cf3747baa6f2ff44.png"},{"id":98961052,"identity":"4664b24f-8795-4667-bdb2-3d98aff15ca2","added_by":"auto","created_at":"2025-12-24 17:27:39","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":142940,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/ab4d15f12a4217f3a82d75e6.png"},{"id":99311151,"identity":"0135f271-b96b-416b-a69a-ee2153958a22","added_by":"auto","created_at":"2025-12-31 16:13:59","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":200731,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/4c0fd84c1400376225a13d8e.png"},{"id":99311543,"identity":"5a3b7fd5-6087-4304-9456-6a05944f70f0","added_by":"auto","created_at":"2025-12-31 16:16:02","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":92350,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/fd7cb27fc9e0f95e351542c7.png"},{"id":98961053,"identity":"f082f8ee-0660-4762-8785-b8f8719f792b","added_by":"auto","created_at":"2025-12-24 17:27:39","extension":"xml","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":139475,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSMAT251155T0structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/c004bb8d2babfa91bbca408f.xml"},{"id":99311280,"identity":"a79a6d62-5261-4e21-bed6-59357c451acf","added_by":"auto","created_at":"2025-12-31 16:14:13","extension":"html","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":156311,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/c6f53461dc0925930342ec3c.html"},{"id":98961033,"identity":"6ec58a18-550a-4a2a-ad1e-05e0bf3bcdd6","added_by":"auto","created_at":"2025-12-24 17:27:38","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":233763,"visible":true,"origin":"","legend":"\u003cp\u003eOverview of neuromorphic hardware and performances of the proposed OTS devices.\u003cstrong\u003e a \u003c/strong\u003eStructure of a biological neuron comprising dendrite (synapses), soma, and axon, and a schematic of a spiking neural network (SNN) consisting of memristor synapses and OTS neurons. \u003cstrong\u003eb\u003c/strong\u003e Periodic table of elements used as OTS materials in previous studies and in this work. \u003cstrong\u003ec\u003c/strong\u003e Comparison map for glass transition temperature (T\u003csub\u003ex\u003c/sub\u003e) and off-current (J\u003csub\u003eoff\u003c/sub\u003e) density of this work and the previous reports\u003csup\u003e18,20,23,29-33\u003c/sup\u003e.\u003cstrong\u003e d\u003c/strong\u003e Spiking frequency (MHz) versus power consumption per unit area per spike (pJ/μm\u003csup\u003e2\u003c/sup\u003e) of the artificial neuron device based on NSGT in this work and the previous reports\u003csup\u003e34-42\u003c/sup\u003e.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/664ea47ba92d2d45c1e5bebd.png"},{"id":99311228,"identity":"115bb732-d926-40c3-bf78-975476c89854","added_by":"auto","created_at":"2025-12-31 16:14:10","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":356618,"visible":true,"origin":"","legend":"\u003cp\u003eDevice structure and electrical characteristics.\u003cstrong\u003e a \u003c/strong\u003eSchematic of the fabricated OTS device with a chalcogenide layer thickness of 20 nm. \u003cstrong\u003eb\u003c/strong\u003e Cross-sectional SEM image of a single device. \u003cstrong\u003ec \u003c/strong\u003eCross-sectional\u003cstrong\u003e \u003c/strong\u003eobservation of amorphous NSGT chalcogenide layer by high-resolution TEM, and FFT of the selective TEM region corresponding to the polycrystalline W electrode and amorphous NSGT layer.\u003cstrong\u003e d\u003c/strong\u003e Representative DC I-V profiles of the three chalcogenide systems with identical feature areas. \u003cstrong\u003ee\u003c/strong\u003e The statistical distribution of I\u003csub\u003eoff\u003c/sub\u003e from 20 OTS devices of each material. \u003cstrong\u003ef\u003c/strong\u003e XRD patterns of annealed ST, SGT, and NSGT films.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/7f79a3dfdc8e9f422fa463c5.png"},{"id":99311235,"identity":"5adfa48d-9cc5-4be2-8e48-4ce2e2826344","added_by":"auto","created_at":"2025-12-31 16:14:10","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":174673,"visible":true,"origin":"","legend":"\u003cp\u003eOptical bandgap and bond configuration distributions of amorphous ST, SGT, and NSGT.\u003cstrong\u003e a \u003c/strong\u003eXPS spectra of Te 3d. \u003cstrong\u003eb\u003c/strong\u003e Raman spectra between 50 cm\u003csup\u003e–1\u003c/sup\u003e~150 cm\u003csup\u003e–1\u003c/sup\u003e. \u003cstrong\u003ec\u003c/strong\u003e Normalized peak area ratio of XPS Te 3d orbital. \u003cstrong\u003ed\u003c/strong\u003e Normalized peak area ratio from Raman spectra. \u003cstrong\u003ee\u003c/strong\u003e The Tauc plot and optical bandgap from UV-vis transmittance spectra of ST, SGT, and NSGT.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/47de2607288feee206974923.png"},{"id":99311435,"identity":"bb79cfcc-9a29-4bea-be3e-d0eb8bdf0b6c","added_by":"auto","created_at":"2025-12-31 16:15:06","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":314840,"visible":true,"origin":"","legend":"\u003cp\u003eStructural models and first-principles simulations of amorphous ST, SGT, and NSGT. Amorphous structural models of a ST, bSGT, and c NSGT. DoS around the mobility gap, with solid/dashed lines indicating the PDoS of Si, Te, Ge, and N sites in d ST, e SGT, and f NSGT. PDoS of major bond configurations showing significant contributions to trap states and band edges within g ST, hSGT, and i NSGT.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/25d600aa9d33b879a7446ff4.png"},{"id":98961043,"identity":"4420521c-064b-498d-85d3-c420c7f68c16","added_by":"auto","created_at":"2025-12-24 17:27:39","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":279203,"visible":true,"origin":"","legend":"\u003cp\u003eLow-power artificial neuron characteristics of NSGT OTS device.\u003cstrong\u003e a \u003c/strong\u003eStructure of the artificial neuron based on the fabricated OTS device consisting of a pre-neural synapse resistor, an OTS device, and a load resistor. \u003cstrong\u003eb\u003c/strong\u003e LIF behavior of the fabricated artificial neuron. \u003cstrong\u003ec \u003c/strong\u003eSpiking frequency of artificial neuron depending on load resistor (R\u003csub\u003eL\u003c/sub\u003e = 2.4 kΩ, 5.1 kΩ, 10.0 kΩ) \u003cstrong\u003ed\u003c/strong\u003e FFT of the measured output spikes. \u003cstrong\u003ee \u003c/strong\u003eSpiking probability of artificial neuron depending on driving voltage (V\u003csub\u003ein\u003c/sub\u003e of 0.93~1.20 V).\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/a6e7c838503702328968ec43.png"},{"id":102298644,"identity":"54f2c559-5cb5-43ed-a5e3-8bdc5e75638c","added_by":"auto","created_at":"2026-02-10 10:55:30","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2091943,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/d82ffe73-f55a-4da7-8b67-ded9e8d4f9c2.pdf"},{"id":98961039,"identity":"c82df27f-3303-4f07-9a9c-380601daef59","added_by":"auto","created_at":"2025-12-24 17:27:39","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":2365329,"visible":true,"origin":"","legend":"Supplementary Information","description":"","filename":"SupplementaryCommNat.docx","url":"https://assets-eu.researchsquare.com/files/rs-7764312/v1/a903f2e176c4ed22d0c5d8d6.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Scalable and sustainable N-Si-Ge-Te Ovonic threshold switching devices for energy-efficient artificial neuron applications","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe rapid growth of artificial intelligence (AI) applications has intensified the demand for energy-efficient, high-throughput computing architectures that overcome the technical limitations of conventional von-Neumann architecture\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Neuromorphic or brain-inspired computing, modelled on the principles of biological neurons and synapses in human thinking, has emerged as a promising approach, offering notable performance gains in energy efficiency and data throughput\u003csup\u003e\u003cspan additionalcitationids=\"CR3 CR4\" citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. In-memory crossbar architectures, analogous to biological synapses, enable simultaneous storage and computation of data, significantly reducing memory bandwidth requirements and enhancing throughput and energy efficiency in data-intensive AI workloads\u003csup\u003e\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. As the core element of neuromorphic computing hardware, an artificial neuron processes incoming electrical spike signals from pre-synaptic neurons and determines their transmission to interconnected post-synaptic neurons\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. A crucial aspect of such artificial neurons is their ability to operate within low-power, high-frequency domains, ensuring applicability for neuromorphic computing.\u003c/p\u003e \u003cp\u003eExtensive research has explored artificial neurons using scalable electronics, ranging from simple transistor latches\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e and complex CMOS circuits\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e to resistive-switching devices, to achieve neuro-morphic behavior with low energy overheads\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. Among various approaches, artificial neurons based on Ovonic threshold-switch (OTS)\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e devices have recently attracted attention due to scalability, fast switching, and low energy consumption\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Their intrinsic capacitance, derived from the metal-insulator-metal (MIM) structure, further enables a simple one selector-one resistor (1S-1R) scheme to emulate the leaky integrate-and-fire (LIF) dynamics of ideal neurons. A key challenge in advancing OTS-based artificial neurons lies in material engineering to optimize device properties for low-energy neuromorphic operations\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. Since the origin of OTS behavior remains unresolved, material optimization of OTS devices requires precise control of material stoichiometry and comprehensive simulations to understand the correlation between material properties and device characteristics within the amorphous chalcogenide structure.\u003c/p\u003e \u003cp\u003eNumerous efforts have been made to enhance OTS device performance through various chalcogenide systems. Binary chalcogenide OTS devices have demonstrated outstanding switching performances and stabilities\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan additionalcitationids=\"CR19\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e, and revealed composition-dependent behaviors associated with each element\u0026rsquo;s distinct role\u003csup\u003e\u003cspan additionalcitationids=\"CR22\" citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. These insights have further progressed the design of multi-element chalcogenide systems, from ternary to quaternary OTS devices, enabling more comprehensive optimization of OTS performances\u003csup\u003e\u003cspan additionalcitationids=\"CR25 CR26\" citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. However, reliance on toxic dopants such as arsenic, in dramatically improving OTS performances, particularly regarding the balance between selectivity and thermal stability, raises sustainability concerns\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThis study examines the elemental contributions in quaternary N-Si-Ge-Te (NSGT) OTS devices through material analysis and ab initio simulations, thus optimizing to demonstrate devices with outstanding performance, low off-state current density (2.3 ∙ 10\u003csup\u003e\u0026minus;\u0026thinsp;8\u003c/sup\u003e MA/cm\u0026sup2;) and high thermal stability (\u0026gt;\u0026thinsp;400\u0026deg;C). Comparative characterizations from binary Si-Te (ST), ternary Si-Ge-Te (SGT) to quaternary NSGT stoichiometric systems reveal how each element uniquely shapes bond configurations and molecular orbitals, enhancing electrical performance and thermal stability. The effects of each bond were analyzed via density of states (DoS) and charge density of each bond based on Density Functional Theory (DFT), with main bond models informed by spectral analysis. Furthermore, OTS pulse measurements further indicate potential for neuromorphic applications. This study establishes stoichiometric guidelines for tailoring selector characteristics.\u003c/p\u003e"},{"header":"Results and Discussions","content":"\u003cp\u003e\u003cstrong\u003eTe-based OTS devices for low-power artificial neuron\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSpiking neural networks (SNNs) replicate the operating principle of biological neurons: 1) dendrites (memristors) receive pre-synaptic spike signals from the previous counterparts, 2) soma/OTS neurons integrate and fires these signals above a threshold, and 3) axons (OTS devices with high selectivity) acts as artificial neurons when an appropriate circuit is designed (Fig. 1a). Neuromorphic hardware for SNN operation primarily comprises synaptic devices (memristive devices), which physically store conductance \u003cem\u003eg\u003csub\u003eW\u003c/sub\u003e\u003c/em\u003e for weight multiplication, and neuronal devices (threshold switches), which control input signal propagation to subsequent layers.\u003c/p\u003e\n\u003cp\u003eThe development of artificial neural networks (ANNs) is driven by the need for energy-efficient, low-power computing. Achieving this objective requires artificial neurons that operate at low voltage and current, necessitating the precise selection of OTS device elements. Among OTS-applicable chalcogen elements (S, Te, Se), tellurium is notable for low-V\u003csub\u003eth\u003c/sub\u003e, low-I\u003csub\u003eoff\u003c/sub\u003e operation\u003csup\u003e43,44\u003c/sup\u003e. Furthermore, compound elements N, Si, and Ge ensure environmental sustainability and CMOS-compatibility, avoiding toxic dopants such as arsenic (Fig. 1b).\u003c/p\u003e\n\u003cp\u003eA key performance limitation in amorphous chalcogenide-based OTS devices is the trade-off between thermal stability (T\u003csub\u003ex\u003c/sub\u003e, crystallization temperature) and off-current density (J\u003csub\u003eoff\u003c/sub\u003e, off-current at half-V\u003csub\u003eth\u003c/sub\u003e divided by device area), as shown in Fig. 1c, Table T1\u003csup\u003e18,20,23,29-33\u003c/sup\u003e. Thermal stability is defined as the temperature at which the device loses threshold-switching property due to irreversible crystallization. Off-current is defined as the sub-threshold current flow, governed by Poole-Frenkel conduction, which is mainly influenced by critical device dimensions such as area and thickness, as well as material parameters as trap distribution and electrical bandgap\u003csup\u003e45\u003c/sup\u003e. To focus on the materialistic view of engineering OTS devices, off current density can also be considered as the criteria for examining OTS performance. Analysis of previously studied chalcogenides highlights the trade-off as the major bottleneck to achieving high stability and energy efficiency in OTS devices.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn this work, material characterization and ab-initio simulation are conducted to address this limitation, guiding the material design of optimized NSGT devices. The proposed NSGT OTS device demonstrates ultra-low off-current density (J\u003csub\u003eoff\u0026nbsp;\u003c/sub\u003e= 2.3 ∙ 10\u003csup\u003e-8\u003c/sup\u003e MA/cm²) and high thermal stability (T\u003csub\u003ex\u003c/sub\u003e \u0026gt; 400°C) compatible with back-end fabrication. It also exhibits low-power leaky LIF neuron behavior, with normalized power consumption per spike reaching 0.56 pJ/μm\u003csup\u003e2\u003c/sup\u003e (Fig. 1d)\u003csup\u003e34-42\u003c/sup\u003e. The device yields tunable spiking frequency depending on the load resistor, up to maximum frequency of 8.6 MHz, outperforming many previously reported artificial neuron devices. (detailed in Table T2, Supplementary Information)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDevice structure and electrical characteristics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe OTS device is fabricated in a via-hole structure with a global 100 nm thick tungsten (W) bottom and top electrodes with the switching medium as an amorphous chalcogenide layer (ST, SGT, and NSGT), forming a vertical MIM structure. The effective device dimension is formed by selectively etching the dielectric SiO\u003csub\u003e2\u003c/sub\u003e layer with a feature area of 25 μm\u003csup\u003e2\u003c/sup\u003e (Fig. 2a; fabrication details in Fig. S1, Supplementary Information). Cross-sectional analysis via scanning electron microscopy (SEM; Fig. 2b) and high-resolution transmission electron microscopy (HR-TEM; Fig. 2c) verifies uniform deposition of the chalcogenide layer (~20 nm thickness). Fast-Fourier transform (FFT) analysis of the amorphous chalcogenide layer reveals the absence of crystalline sites with no observable diffraction peaks. (Fig. 2c, inset).\u003c/p\u003e\n\u003cp\u003eOTS devices exhibit a sharp decrease in resistance at the threshold voltage (V\u003csub\u003eth\u003c/sub\u003e) to switch to the ON state, and a subsequent increase in resistance at the hold voltage (V\u003csub\u003eh\u003c/sub\u003e) to return to the OFF state. Representative DC I-V measurement of ST, SGT, and NSGT devices (feature area of 25 μm\u003csup\u003e2\u003c/sup\u003e and 20 nm thickness), as illustrated in Fig. 2d, reveals a discrete trend in sub-threshold I\u003csub\u003eoff\u003c/sub\u003e depending on the constituent elements, with ST having the highest I\u003csub\u003eoff\u003c/sub\u003e (26 μA), followed by SGT (0.74 μA), and NSGT with the lowest I\u003csub\u003eoff\u003c/sub\u003e.\u0026nbsp;The NSGT device exhibits outstanding I-V characteristics, having\u0026nbsp;a high-resistance state (HRS)\u0026nbsp;with I\u003csub\u003eoff\u003c/sub\u003e = 47 nA (@ 1/2V\u003csub\u003eth\u003c/sub\u003e)\u0026nbsp;and a rapid transition\u0026nbsp;to\u0026nbsp;a low-resistance state (LRS)\u0026nbsp;at V\u003csub\u003eth\u003c/sub\u003e = 0.75 V. Cumulative device measurements confirm low cell-to-cell variation across ST, SGT, and NSGT devices, with average I\u003csub\u003eoff\u003c/sub\u003e values of 60 μA (ST), 5.29 μA (SGT), and 39 nA (NSGT), respectively (Fig. 2e; cumulative I-V curves in Fig. S2, Supplementary Information). The device demonstrates a 5 ns switching delay time from LRS to HRS under a triangular pulse. (Fig. S3, Supplementary Information).\u003c/p\u003e\n\u003cp\u003eThermal stability in SGT (Fig. 2d) decreased to \u0026lt; 300 °C, as shown by XRD analysis of equivalent chalcogenide films annealed to high temperatures. Crystallization peak at 28° was observed in SGT film of X-ray Diffraction (XRD) pattern, losing the film’s as-deposited amorphous state. In contrast, no crystallization peak appeared in ST and NSGT films, even at temperatures above 400°C. \u0026nbsp;This behavior is attributed to low bond enthalpies\u003csup\u003e46\u003c/sup\u003e of Ge (H\u003csub\u003eGe-Ge\u003c/sub\u003e = 186 kJ/mol, H\u003csub\u003eGe-Te\u003c/sub\u003e = 192 kJ/mol) calculated using Pauling’s relation. On the other hand, ST film possesses high bond enthalpies of Si (H\u003csub\u003eSi\u003c/sub\u003e\u003csub\u003e-\u003c/sub\u003e\u003csub\u003eTe\u003c/sub\u003e = 215 kJ/mol), which is believed to result in a higher thermal stability compared to SGT. Finally, N-doping is performed to optimize device parameters\u003csup\u003e47,48\u003c/sup\u003e, forming\u0026nbsp;Si-N (H\u003csub\u003eSi\u003c/sub\u003e\u003csub\u003e-\u003c/sub\u003e\u003csub\u003eN\u003c/sub\u003e = 395 kJ/mol) and Ge-N (H\u003csub\u003eGe-Ge\u003c/sub\u003e = 354 kJ/mol) bonds that stabilize the chalcogenide’s microstructure. This modification enhances the device performance, yielding a selectivity of 3.9 ∙ 10\u003csup\u003e5\u003c/sup\u003e,\u0026nbsp;a\u0026nbsp;delay time of 5 ns after arrival of V\u003csub\u003eth\u003c/sub\u003e, and thermal stability \u0026gt; 400°C, satisfying back-end-of-line (BEOL) requirements.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOptical bandgaps and local structures of amorphous ST, SGT, and NSGT\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis section investigates why NSGT OTS selectors exhibit superior selectivity and thermal stability compared to ST and SGT OTS. Structural differences among the three OTS materials, which may influence the properties of OTS selectors, are analyzed using X-ray photoelectron spectroscopy (XPS) and Raman scattering spectroscopy. In the Te 3d region, Te\u003csup\u003e2-\u0026nbsp;\u003c/sup\u003epeaks (Te-Te bonds, located at 583.6 eV and 573.2 eV)\u003csup\u003e48-50\u003c/sup\u003e and Te\u003csup\u003e4+\u003c/sup\u003e bonds (located at 586.8 eV and 576.5 eV)\u003csup\u003e48,51-53\u003c/sup\u003e are consistently observed across ST, SGT, and NSGT films (Fig. 3a). Te-Te bonds dominate in ST (57%), compared to SGT (39%) and NSGT (7%) (Fig. 3c). Raman analysis between 50 cm\u003csup\u003e–\u003c/sup\u003e\u003csup\u003e1\u003c/sup\u003e and 200 cm\u003csup\u003e–\u003c/sup\u003e\u003csup\u003e1\u003c/sup\u003e regions shows a similar trend: peaks A and B correspond to the E\u003csub\u003eg\u003c/sub\u003e vibrational mode of Te-Te bonds\u003csup\u003e54,55\u003c/sup\u003e, peak D to the A\u003csub\u003e1\u003c/sub\u003e expansion mode of the Te-Te bond\u003csup\u003e54-56\u003c/sup\u003e,\u0026nbsp;and peak\u0026nbsp;E to\u0026nbsp;the\u0026nbsp;E\u003csub\u003eTO\u003c/sub\u003e mode of Te helical chain\u003csup\u003e48,54,56,57\u003c/sup\u003e.\u0026nbsp;Detailed\u0026nbsp;wavenumbers\u0026nbsp;and area fractions for\u0026nbsp;each deconvoluted Raman peak\u0026nbsp;are\u0026nbsp;provided in Table T3, Supplementary Information.\u0026nbsp;The fraction of Raman peaks related to Te-Te bonds\u0026nbsp;(peaks\u0026nbsp;A, B, D, and E)\u0026nbsp;decreases\u0026nbsp;from ST (93%)\u0026nbsp;to\u0026nbsp;SGT (51%)\u0026nbsp;and\u0026nbsp;NSGT (43%), reflecting\u0026nbsp;Ge-Te\u0026nbsp;bond formation through\u0026nbsp;Ge-doping. Peak C at 123 cm\u003csup\u003e–\u003c/sup\u003e\u003csup\u003e1\u003c/sup\u003e corresponds to Ge-Te units\u003csup\u003e53,57-59\u003c/sup\u003e in SGT and NSGT by Ge-doping. XPS analysis indicates Te-rich compositions in SGT and NSGT, consistent with tetrahedral Ge-Te bond configurations, such as GeTe\u003csub\u003e4\u003c/sub\u003e and GeTe\u003csub\u003e3\u003c/sub\u003eGe\u003csup\u003e59\u003c/sup\u003e. In particular, GeTe\u003csub\u003e4\u003c/sub\u003e forms a stable structural motif stabilized by sp\u003csup\u003e3\u003c/sup\u003e hybridization, with Ge’s four valence electrons fully shared\u003csup\u003e60,61\u003c/sup\u003e. Unlike Ge-doping,\u0026nbsp;N-doping does not induce direct\u0026nbsp;bonding with Te; instead, N\u0026nbsp;incorporates\u0026nbsp;as Ge-N and Si-N,\u0026nbsp;confirmed in Ge 3d,\u0026nbsp;N 1s, and Si 2p\u0026nbsp;spectra\u0026nbsp;of NSGT\u0026nbsp;(Fig. S4,\u0026nbsp;Supplementary Information).\u0026nbsp;Although not directly bonded to Te,\u0026nbsp;N\u0026nbsp;substitution\u0026nbsp;at\u0026nbsp;Te sites in the Ge-Te configuration\u0026nbsp;can modify\u0026nbsp;the structure\u0026nbsp;and\u0026nbsp;reduce\u0026nbsp;Te-Te\u0026nbsp;bond fractions.\u0026nbsp;These\u0026nbsp;structural variations significantly alter the electronic structure,\u0026nbsp;producing distinct\u0026nbsp;optical bandgaps, as determined by the\u0026nbsp;linear fitting of the near absorption edge in the Tauc plots of the UV-vis spectroscopy (Fig.\u0026nbsp;3g) for each material.\u0026nbsp;The optical bandgap increases\u0026nbsp;from 0.725 eV (ST)\u0026nbsp;to\u0026nbsp;0.856 eV (SGT)\u0026nbsp;and\u0026nbsp;1.583 eV (NSGT).\u0026nbsp;In OTS materials,\u0026nbsp;subthreshold\u0026nbsp;conduction\u003csup\u003e62\u003c/sup\u003e is governed by the Poole-Frenkel mechanism\u003csup\u003e45\u003c/sup\u003e, which\u0026nbsp;is\u0026nbsp;bandgap-dependent.\u0026nbsp;The\u0026nbsp;observed\u0026nbsp;bandgap\u0026nbsp;trend\u0026nbsp;in this\u0026nbsp;study aligns\u0026nbsp;well with the I\u003csub\u003eoff\u003c/sub\u003e tendency in Fig. 2d–e and is attributed to the combined effects of bond configuration variations and quantitative changes in bond fractions. Further insights are provided through ab initio calculations.\u003c/p\u003e\n\u003cp\u003eAs confirmed in existing studies, Te is essential for OTS operation. However, off-current behavior varies with the compound material and composition\u003csup\u003e19,20,23,44,63\u003c/sup\u003e. In this study, a consistent trend is observed, where reducing Te–Te bonds increase the optical bandgap and decreases I\u003csub\u003eoff\u003c/sub\u003e. This is attributed to Te–Te bonds containing lone pairs (LPs) with weakly bound electrons that may have influenced conduction in the subthreshold region\u003csup\u003e14,23,64\u003c/sup\u003e. Furthermore, the thermal stability of OTS material based on amorphous chalcogenides is generally determined by its crystallization temperature (T\u003csub\u003ex\u003c/sub\u003e), since nonlinear conduction typically occurs in the amorphous state\u003csup\u003e23\u003c/sup\u003e. T\u003csub\u003ex\u003c/sub\u003e of OTS material can be predicted using the ratio of bond enthalpies of constituent bond states, based on theoretical models\u003csup\u003e46\u003c/sup\u003e (Method C1, Supplementary Information). Although the model does not precisely predict absolute values, it reliably indicates trends. SGT, containing Ge-related bonds with relatively lower bond enthalpy, exhibits a lower T\u003csub\u003ex\u003c/sub\u003e than ST. In contrast, NSGT, containing only N-related bonds with relatively higher enthalpy, shows a higher T\u003csub\u003ex\u003c/sub\u003e, indicating that its high thermal stability originates from the presence of bonds with high bond enthalpy, consistent with previous research\u003csup\u003e47\u003c/sup\u003e. This trend aligns with XRD results (Fig. 2f) after annealing up to 400°C, the reference point for BEOL processes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cstrong\u003eStructural models and\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ef\u003c/strong\u003e\u003cstrong\u003eirst-principles simulations of amorphous ST, SGT, and NSGT\u003c/strong\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDensity Functional Theory (DFT) calculations are performed to examine bond configuration changes and the effects of Ge/N doping on the mobility gap using SIESTA (Spanish Initiative for Electronic Simulations with Thousands of Atoms). The Generalized-Gradient Approximation (GGA-PBE) functional is employed to describe the exchange and correlation of the systems. Amorphous material structures (Fig. 4a–c) are generated using the melt-quench-relaxation method (see Methods section for details). The light blue curve in structural models represents the charge density distribution with an isovalue of 0.5 e/ų. The calculated radial distribution functions (RDFs) of all three structures confirm the absence of long-range order, a typical feature of amorphous materials (Fig. S5 in Supplementary Information). Structural models are verified by their agreement with bond distributions reported in previous spectroscopic studies. The amorphous ST, SGT, and NSGT networks exhibit Te–Te bonds (Fig. 4g), corresponding to the dominant peak in all RDFs. Moreover, Ge–Te tetrahedral units (Fig. 4e) and N-related bonds (Fig. 4f), highlighted by red squares in Fig. 4a–c, are also reflected in the RDFs.\u003c/p\u003e\n\u003cp\u003eThe calculated electronic density of states (DOS) of ST, SGT, and NSGT are presented in Fig. 4g–i. Consistent with previous studies, amorphous chalcogenide materials exhibit a mobility gap with an exponential distribution between the conduction and valence states\u003csup\u003e43,49,65,66\u003c/sup\u003e. Importantly, the mobility gap increases progressively from ST to SGT and NSGT, and this trend agrees well with the optical bandgap extracted from the Tauc plot (Fig. 3e). The absolute values of the mobility gap are lower than experimentally measured optical bandgaps, due to the bandgap underestimation of GGA functional. This progressive widening of the mobility gap indicates that further structural/electronic interpretation is necessary to examine local bonding arrangements and their correlation with electronic states governing electronic properties in OTS devices.\u003c/p\u003e\n\u003cp\u003eIn Fig. 4g, conduction and valence band edges of ST are dominantly formed by Te p-orbital as in projected DoS (PDoS, dashed line). This matches well with the low charge density distribution (0.58 e/ų) of the Te chain, denoted by the double blue curve near the Te atoms, which corresponds to lone-pair from Te p-orbital (p-LP). The PDoS of the s-orbital is observed far below the valence band maximum (Fig S6, Supplementary Information). The trap state observed just under the conduction band edge in ST is mainly attributed to the lone pairs of Te atoms in the Te–Te bond (Te-1 – Te-2), and the high contribution of these non-bonding lone pairs in the PDoS is clearly visible\u003csup\u003e49\u003c/sup\u003e. The inset of Fig. 4g further supports this interpretation by showing spatially localized charge density concentrated around Te-1 – Te-2. Such shallow trap states are known to facilitate carrier leakage in the subthreshold region of OTS devices, linking the electronic structure of ST directly to its relatively high off-current.\u003c/p\u003e\n\u003cp\u003eWith Ge-doping, combined orbital states between Ge and Te appear in the PDoS of SGT (Fig. 4h), reducing the contribution of the Te p-LP. Consistent with Fig. 3a–b, Ge incorporation decreases the density of Te–Te bonds and promotes the formation of Ge–Te bonds. The major structural motifs in SGT include a Ge-centered Ge₂Te₃ tetrahedral configuration, in which Te-3 and Te-4 participate in corner-sharing connections with neighboring Ge atoms. The PDoS confirms that these Te orbitals participate in Ge–Te bonding and contribute to the occupied states above the valence band edge, rather than forming trap states under the conduction band edge as in ST. Even the terminal Te atom (Te-5), which is not part of a Te–Te chain but instead terminates through a Ge connection, exhibits a valence-band-tail state rather than the deeply localized non-bonding LP state observed in ST. This redistribution of Te-LP-driven states leads to a wider mobility gap in SGT and explains the observed reduction of off-current.\u003c/p\u003e\n\u003cp\u003eNSGT also shows a reduction in Te p-LP and a corresponding widening of the mobility gap. However, unlike ST and SGT, which exhibit trap states predominantly under the conduction band edge, NSGT exhibits an additional trap state between the Fermi level and the valence band edge. N-doping induces strong electron localization around N atoms\u003csup\u003e48\u003c/sup\u003e, as confirmed by the charge density distribution. It appears to substitute at the corners of the Ge–Te structural units, where its high electronegativity induces structural distortions. Consequently, the Ge–Te unit with N atoms slightly changes in the Ge-centered bond angle (Fig. S7, Supplementary Information). The trap states are therefore attributed to localized states arising from distortions in the Ge–Te hybrid orbital and rearrangements of Te p-LP states. Nevertheless, because these N-driven trap states lie closer to the valence band rather than under the conduction band edge, they tend to capture hole carriers rather than facilitate leakage, providing a mechanism for reducing the off-current rather than increasing it.\u003c/p\u003e\n\u003cp\u003eAcross the full ST → SGT → NSGT series, the evolution of Te lone-pair states shows a strong correlation with the formation of valence-band tails. Direct visualization of representative local structures reveals that the charge density associated with Te-LP decreases nearly linearly from ST to NSGT. In parallel, the slope of the valence-band tail in the DOS becomes steeper, resulting in an expanded mobility gap. Initially, in ST, Te-LP generates trap states under the conduction band, but as Te atoms become increasingly involved in Ge–Te or N–Ge–Te bonding environments, these Te-LP-driven states migrate toward the valence-band-tail region, reducing their contribution to shallow trap states and thus broadening the mobility gap. This establishes a direct structural–electronic–device link: Te-LP-driven shallow trap states produce low valence-band-tail slopes and high off-current, whereas suppression and relocation of LP states steepen the tail, widen the mobility gap, and improve off-current performance.\u003c/p\u003e\n\u003cp\u003eIt is well known that OTS properties are strongly affected by trap states located within the mobility gap\u003csup\u003e43,49,65,66\u003c/sup\u003e. In our selector device as well, Te-LP-induced traps govern the off-current behavior. In ST, the shallow Te-LP trap with a small slope in the valence-band tail increases the off-current, whereas Ge and N incorporation progressively suppresses Te-LP-driven states, which in turn widens the mobility gap and suppresses the off-current. Therefore, control over Te-LP—not only the composition of constituent elements but the local bonding topology that governs the spatial distribution of LP states—serves as the electronic-structure basis for improving the on/off ratio of Te-based OTS materials.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEnergy-\u003c/strong\u003e\u003cstrong\u003ee\u003c/strong\u003e\u003cstrong\u003efficient\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003en\u003c/strong\u003e\u003cstrong\u003eeuronal\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ea\u003c/strong\u003e\u003cstrong\u003epplication of\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;the\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;NSGT OTS\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ed\u003c/strong\u003e\u003cstrong\u003eevice\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSpiking neural networks (SNNs) are energy-efficient, low-power networks that propagate information via spike signals across neuron layers. A critical requirement for realizing SNN into a hardware system is an artificial neuron capable of operating at low voltage and current. Resembling biological neurons, such an artificial neuron can be fabricated with a pre-neuron synapse resistor (R\u003csub\u003es\u003c/sub\u003e), an OTS device representing the soma’s threshold determining the spike propagation, and a load resistor (R\u003csub\u003eL\u003c/sub\u003e) (Fig. 5a). Our optimized NSGT OTS device, featuring low subthreshold current and operating voltage, exhibits typical LIF neuronal characteristics with high energy efficiency and controlled spiking (Fig. 5b). LIF behavior involves a leaky charging of the OTS’s intrinsic capacitance, in which leaky firing is mostly suppressed by the optimization of N-Si-Ge-Te OTS device having low subthreshold current, followed by an abrupt firing and a slow discharging to load resistor. The overall circuit can thus be paralleled to RC circuit model with series resistance R\u003csub\u003es\u003c/sub\u003e, membrane capacitance C\u003csub\u003eOTS\u003c/sub\u003e, and load resistance R\u003csub\u003eL\u003c/sub\u003e, with reset switch as the OTS device.\u003c/p\u003e\n\u003cp\u003eThe fabricated OTS-based artificial neuron exhibits tunable spike characteristics depending on load resistor (R\u003csub\u003eL\u003c/sub\u003e), synaptic resistor (R\u003csub\u003es\u003c/sub\u003e), and voltage amplitude (V\u003csub\u003ein\u003c/sub\u003e) (Fig. 5c–e). In Fig. 5c,\u0026nbsp;spike frequency\u0026nbsp;increases with decreasing\u0026nbsp;load resistor (R\u003csub\u003eL\u003c/sub\u003e), consistent with the\u0026nbsp;time constant relation (\u003cimg width=\"76\" height=\"20\" src=\"data:image/png;base64,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\" alt=\"image\"\u003e) governing the discharging profile of individual spikes.\u0026nbsp;FFT\u0026nbsp;analysis\u0026nbsp;(Fig.\u0026nbsp;5d) reveals first\u0026nbsp;harmonic peaks\u0026nbsp;for\u0026nbsp;each R\u003csub\u003eL\u003c/sub\u003e condition, indicating an inverse relation between R\u003csub\u003eL\u003c/sub\u003e and spike frequency, due to higher R\u003csub\u003eL\u003c/sub\u003e inducing slower current discharge as proportional to R\u003csub\u003eL\u003c/sub\u003eC\u003csub\u003eOTS\u003c/sub\u003e. Likewise, the effect of synapse resistor (R\u003csub\u003eS\u003c/sub\u003e) on spike frequency, governing the charging profile of an individual spike, reveals that higher R\u003csub\u003eS\u003c/sub\u003e induces slower charging to C\u003csub\u003eOTS\u003c/sub\u003e, similarly leading to lower spike frequency (detailed in Fig. S8, Supplementary Information). Furthermore, increasing the applied voltage amplitude raises the spike probability and frequency of the NSGT artificial neuron (Fig. 5e). Such a trend is due to the increased current inflow to the parasitic capacitor as input voltage is increased, which stimulates the charge-discharge behavior to higher frequencies with more frequent spikes. Throughout the observations, the probabilistic nature of the OTS artificial neuron is consistently observed, which is presumably caused by the amorphous atomic structure of the chalcogenide film changing the switching path at every cycle, resulting in a subtle fluctuation of the threshold voltage, and leading to varying LIF behavior.\u0026nbsp;\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study investigates the impact of stoichiometric variations on the electrical and thermal properties of Te-based OTS materials composed of environmentally friendly elements (Ge, Si, N; excluding As and Sb). In all three materials, Te p-lone pairs dominate the band edges, while Ge doping forms Ge\u0026ndash;Te tetrahedral structures that reduce their influence and widen the mobility gap. Although N does not directly bond with Te, it lowers the overall Te p-lone pair content and distorts the GeTe structure, creating mid-gap states that trap carriers in this p-type material and effectively suppress the off-current density down to 2.3 ∙ 10\u003csup\u003e\u0026minus;\u0026thinsp;8\u003c/sup\u003e MA/cm\u003csup\u003e2\u003c/sup\u003e. Compared with previous reports, the proposed artificial neuron with optimized leakage properties highlights its potential to operate at ultralow energy of 0.56 pJ/\u0026micro;m\u003csup\u003e2\u003c/sup\u003e and ultrahigh spike frequency of 8.6 MHz, despite large device area. Overall, the results imply strong potentials for the optimized NSGT materials to be applicable for scalable, sustainable and energy-efficient computing systems.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eDevice fabrication and electrical characterization\u003c/h2\u003e \u003cp\u003eThe fabricated device structure consists of a dry-oxidized Si wafer substrate with a universal bottom W electrode (100 nm) deposited by DC magnetron sputtering. SiO\u003csub\u003e2\u003c/sub\u003e dielectric barrier is deposited by plasma-enhanced chemical vapor deposition (PE-CVD), selectively etched by ICP-RIE for feature area formation, followed by deposition of chalcogenide layer (20 nm) via reactive RF magnetron co-sputtering of Si, Ge, and Te targets in Ar\u0026thinsp;+\u0026thinsp;N\u003csub\u003e2\u003c/sub\u003e reactive atmosphere. Finally, the top W electrode (100 nm) is deposited by DC magnetron sputtering (Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Supplementary Information). Fabricated OTS device\u0026rsquo;s I-V characteristic is analyzed using Keithley 4200\u0026ndash;SCS parametric analyzer under I\u003csub\u003ec.c\u003c/sub\u003e. = 1 mA. Delay time is measured in AC pulse sweep mode via pulse measurement unit (PMU) for pulse generation and measurement. The time resolution of the measurement unit of 5 ns limits the minimum detectable delay to 5 ns.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eX-ray diffraction for thermal stability evaluation\u003c/h2\u003e \u003cp\u003eThermal stability is assessed by depositing chalcogenide films on SiO\u003csub\u003e2\u003c/sub\u003e wafer substrates and annealing at T\u003csub\u003emax\u003c/sub\u003e ranging from R.T. to 400\u0026deg;C in increments of 100\u0026deg;C inside a vacuum chamber with a rapid thermal processing unit. Annealing conditions are as follows: ramp rate\u0026thinsp;=\u0026thinsp;1\u0026deg;C/s, T\u003csub\u003emax\u003c/sub\u003e hold time\u0026thinsp;=\u0026thinsp;300 s. After annealing treatment, its crystallinity is measured using Rigaku XRD instrument.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eX-ray photoelectron spectroscopy (XPS) and Raman spectroscopy measurements\u003c/h2\u003e \u003cp\u003eFor microstructural analysis, sample films are deposited onto W-deposited SiO\u003csub\u003e2\u003c/sub\u003e wafer substrates using reactive RF co-sputtering. XPS measurements are conducted using the Thermo Scientific K-Alpha X-ray photoelectron spectrometer system with a monochromatized Al Kα source at an operating voltage of 12 kV, operating current of 3 mA, pass energy of 200 eV, and sampling area of 400 \u0026micro;m. Peak binding energies are referenced to NIST Standard Reference Database 20 v4.1. Raman spectroscopy measurements are conducted using Nanobase XPER-RF Raman spectrometer, with laser wavelength of 532 nm.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eCs-corrected transmission electron microscope measurement\u003c/h2\u003e \u003cp\u003eOTS device is prepared as a lamella sample via FIB (focused ion-beam) milling. JEM-ARM 200F transmission electron microscope with Cs-correction is used under an operating voltage of 200 kV and emission current of 117.7 mA.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eAtomic model simulation\u003c/h2\u003e \u003cp\u003eFirst-principles calculations are performed using the SIESTA package within the Materials Square\u003csup\u003e\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e\u003c/sup\u003e platform by Virtual Lab Inc. Amorphous OTS material models are generated using molecular dynamics simulations in LAMMPS\u003csup\u003e\u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e\u003c/sup\u003e with a melt-quench-relaxation method. The structure is initially heated from 298 K to 1200 K over 50 ps and then quenched from 1200 K to 300 K at a cooling rate of 10 K/ps. The model is further relaxed with gamma-point sampling of the Brillouin zone at 200 K to reach an optimized atomic configuration (atomic force\u0026thinsp;\u0026lt;\u0026thinsp;0.02 eV/Ang). Each cell contained 64 atoms, reflecting the composition as characterized by XPS. The electron-electron exchange-correlation energy is treated using generalized gradient approximation (GGA) functional in Perdew\u0026ndash;Burke\u0026ndash;Ernzerhof (PBE) form. Cutoff energy is set to 300 eV, and SCF tolerance value to 10\u003csup\u003e\u0026minus;\u0026thinsp;5\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors acknowledge financial supports from the Ministry of Trade, Industry, and Energy (MOTIE) and Korea Evaluation Institute Industrial Technology (KEIT) (Project No. 10080625),\u0026nbsp;the\u0026nbsp;Korea Semiconductor Research Consortium (KSRC) program, Samsung Electronics Co., Ltd. (Project No. IO2102021-08356-01),\u0026nbsp;the BK21 FOUR (Fostering Outstanding Universities for Research) funded by the Ministry of Education (MOE) and National Research Foundation (NRF) of Korea, and\u0026nbsp;the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT). (No.2023R1A2C2006811).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding Declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eOpen Access funding is enabled by\u0026nbsp;the BK21 FOUR (Fostering Outstanding Universities for Research), funded by the Ministry of Education (MOE) and National Research Foundation (NRF) of Korea.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eS.P. and C.P. initially conceived the concept and conceptualized the specific experiment. S.P. performed the DFT calculations and characterized the OTS materials using XRD, XPS, Raman, and UV-vis measurements, with support from S.-B.L. and S.K. The manuscript was prepared by S.P., C.P., S.-B.L., and J.-S.Y. S.P. and C.P. designed the figures. J.-S.Y. guided the research. All authors have discussed the results and given comments on the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data supporting the findings of this study are present in the main article or in supplementary information files. Original data or any additional data related to the study are available from the corresponding author upon request.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eIvanov, D., Chezhegov, A., Kiselev, M., Grunin, A. \u0026amp; Larionov, D. Neuromorphic artificial intelligence systems. \u003cem\u003eFrontiers in Neuroscience\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e (2022). https://doi.org/10.3389/fnins.2022.959626\u003c/li\u003e\n\u003cli\u003eUpadhyay, N. K.\u003cem\u003e et al.\u003c/em\u003e Emerging Memory Devices for Neuromorphic Computing. \u003cem\u003eAdvanced Materials Technologies\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 1800589 (2019). https://doi.org/10.1002/admt.201800589\u003c/li\u003e\n\u003cli\u003eMehonic, A.\u003cem\u003e et al.\u003c/em\u003e Memristors\u0026mdash;From In‐Memory Computing, Deep Learning Acceleration, and Spiking Neural Networks to the Future of Neuromorphic and Bio‐Inspired Computing. \u003cem\u003eAdvanced Intelligent Systems\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 2000085 (2020). https://doi.org/10.1002/aisy.202000085\u003c/li\u003e\n\u003cli\u003eAguirre, F.\u003cem\u003e et al.\u003c/em\u003e Hardware implementation of memristor-based artificial neural networks. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 1974 (2024). https://doi.org/10.1038/s41467-024-45670-9\u003c/li\u003e\n\u003cli\u003eRoy, K., Jaiswal, A. \u0026amp; Panda, P. Towards spike-based machine intelligence with neuromorphic computing. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e575\u003c/strong\u003e, 607-617 (2019). \u003c/li\u003e\n\u003cli\u003eYang, J. J., Strukov, D. B. \u0026amp; Stewart, D. R. Memristive devices for computing. \u003cem\u003eNat Nanotechnol\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 13-24 (2013). https://doi.org/10.1038/nnano.2012.240\u003c/li\u003e\n\u003cli\u003eXia, Q. \u0026amp; Yang, J. J. Memristive crossbar arrays for brain-inspired computing. \u003cem\u003eNature Materials\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 309-323 (2019). https://doi.org/10.1038/s41563-019-0291-x\u003c/li\u003e\n\u003cli\u003eWang, Z.\u003cem\u003e et al.\u003c/em\u003e Resistive switching materials for information processing. \u003cem\u003eNature Reviews Materials\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 173-195 (2020). https://doi.org/10.1038/s41578-019-0159-3\u003c/li\u003e\n\u003cli\u003eHan, J. K., Yun, S. Y., Lee, S. W., Yu, J. M. \u0026amp; Choi, Y. K. A Review of Artificial Spiking Neuron Devices for Neural Processing and Sensing. \u003cem\u003eAdvanced Functional Materials\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 2204102 (2022). https://doi.org/10.1002/adfm.202204102\u003c/li\u003e\n\u003cli\u003ePazos, S.\u003cem\u003e et al.\u003c/em\u003e Synaptic and neural behaviours in a standard silicon transistor. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e640\u003c/strong\u003e, 69-76 (2025). https://doi.org/10.1038/s41586-025-08742-4\u003c/li\u003e\n\u003cli\u003eIndiveri, G.\u003cem\u003e et al.\u003c/em\u003e Neuromorphic silicon neuron circuits. \u003cem\u003eFront Neurosci\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 73 (2011). https://doi.org/10.3389/fnins.2011.00073\u003c/li\u003e\n\u003cli\u003eZhang, X.\u003cem\u003e et al.\u003c/em\u003e An artificial spiking afferent nerve based on Mott memristors for neurorobotics. \u003cem\u003eNature Communications\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e (2020). https://doi.org/10.1038/s41467-019-13827-6\u003c/li\u003e\n\u003cli\u003ePark, S. O., Jeong, H., Park, J., Bae, J. \u0026amp; Choi, S. Experimental demonstration of highly reliable dynamic memristor for artificial neuron and neuromorphic computing. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 2888 (2022). https://doi.org/10.1038/s41467-022-30539-6\u003c/li\u003e\n\u003cli\u003eOvshinsky, S. An introduction to ovonic research. \u003cem\u003eJournal of Non-Crystalline Solids\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, 99-106 (1970). \u003c/li\u003e\n\u003cli\u003eJia, S.\u003cem\u003e et al.\u003c/em\u003e Ultrahigh drive current and large selectivity in GeS selector. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 4636 (2020). https://doi.org/10.1038/s41467-020-18382-z\u003c/li\u003e\n\u003cli\u003eLee, J.\u003cem\u003e et al.\u003c/em\u003e An Artificial Tactile Neuron Enabling Spiking Representation of Stiffness and Disease Diagnosis. \u003cem\u003eAdvanced Materials\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, 2201608 (2022). https://doi.org/10.1002/adma.202201608\u003c/li\u003e\n\u003cli\u003eDing, Y.\u003cem\u003e et al.\u003c/em\u003e Engineering Spiking Neurons Using Threshold Switching Devices for High-Efficient Neuromorphic Computing. \u003cem\u003eFront Neurosci\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 786694 (2021). https://doi.org/10.3389/fnins.2021.786694\u003c/li\u003e\n\u003cli\u003eChekol, S. A., Yoo, J. \u0026amp; Hwang, H. Thermally Stable Te-based Binary OTS Device for Selector Application. \u003cem\u003e2018 Non-Volatile Memory Technology Symposium (NVMTS)\u003c/em\u003e (2018). https://doi.org/10.1109/NVMTS.2018.8603103\u003c/li\u003e\n\u003cli\u003eYoo, J.\u003cem\u003e et al.\u003c/em\u003e Te-based binary OTS selectors with excellent selectivity (105 ) endurance (108 ) and thermal stability (450C). \u003cem\u003e2018 IEEE Symposium on VLSI Technology\u003c/em\u003e (2018). https://doi.org/10.1109/VLSIT.2018.8510681\u003c/li\u003e\n\u003cli\u003eKoo, Y., Lee, S., Park, S., Yang, M. \u0026amp; Hwang, H. Simple binary ovonic threshold switching material SiTe and its excellent selector performance for high-density memory array application. \u003cem\u003eIEEE Electron Device Letters\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 568-571 (2017). https://doi.org/10.1109/LED.2017.2685435\u003c/li\u003e\n\u003cli\u003eKim, S. D.\u003cem\u003e et al.\u003c/em\u003e Effect of Ge Concentration in GexSe1-x Chalcogenide Glass on the Electronic Structures and the Characteristics of Ovonic Threshold Switching (OTS) Devices. \u003cem\u003eECS Solid State Letters\u003c/em\u003e \u003cstrong\u003e2\u003c/strong\u003e, Q75-Q77 (2013). https://doi.org/10.1149/2.001310ssl\u003c/li\u003e\n\u003cli\u003eVelea, A.\u003cem\u003e et al.\u003c/em\u003e Te-based chalcogenide materials for selector applications. \u003cem\u003eSci Rep\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 8103 (2017). https://doi.org/10.1038/s41598-017-08251-z\u003c/li\u003e\n\u003cli\u003eKoo, Y. \u0026amp; Hwang, H. Zn1\u0026minus; xTex ovonic threshold switching device performance and its correlation to material parameters. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 11822 (2018). \u003c/li\u003e\n\u003cli\u003eCheng, H. Y.\u003cem\u003e et al.\u003c/em\u003e An Ultra High Endurance and Thermally Stable Selector based on TeAsGeSiSe Chalcogenides Compatible with BEOL IC Integration for Cross-Point PCM. \u003cem\u003e2017 IEEE International Electron Devices Meeting (IEDM)\u003c/em\u003e (2017). https://doi.org/10.1109/IEDM.2017.8268310.\u003c/li\u003e\n\u003cli\u003eVerdy, A.\u003cem\u003e et al.\u003c/em\u003e High Temperature Stability and Performance Analysis of N-doped Ge-Se-Sb based OTS Selector Devices (supplementary). \u003cem\u003e2018 IEEE International Memory Workshop (IMW)\u003c/em\u003e (2019). https://doi.org/10.1109/IMW.2018.8388834\u003c/li\u003e\n\u003cli\u003eLaguna, C.\u003cem\u003e et al.\u003c/em\u003e Ge Content Optimization in Ge(SbSe)N OTS Materials for Selector Applications. \u003cem\u003eIEEE Transactions on Electron Devices\u003c/em\u003e \u003cstrong\u003e69\u003c/strong\u003e, 6277-6283 (2022). https://doi.org/10.1109/ted.2022.3203368\u003c/li\u003e\n\u003cli\u003eVerdy, A.\u003cem\u003e et al.\u003c/em\u003e Effect of Nitrogen on the Amorphous Structure and Subthreshold Electrical Conduction of GeSeSb‐Based Ovonic Threshold Switching Thin Films. \u003cem\u003ephysica status solidi (RRL) \u0026ndash; Rapid Research Letters\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e (2019). https://doi.org/10.1002/pssr.201900548\u003c/li\u003e\n\u003cli\u003eGu, R.\u003cem\u003e et al.\u003c/em\u003e How arsenic makes amorphous GeSe a robust chalcogenide glass for advanced memory integration. \u003cem\u003eScripta Materialia\u003c/em\u003e \u003cstrong\u003e218\u003c/strong\u003e, 114834 (2022). https://doi.org/10.1016/j.scriptamat.2022.114834\u003c/li\u003e\n\u003cli\u003eLee, M.-J.\u003cem\u003e et al.\u003c/em\u003e A plasma-treated chalcogenide switch device for stackable scalable 3D nanoscale memory. \u003cem\u003eNature communications\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 2629 (2013). https://doi.org/10.1038/ncomms3629\u003c/li\u003e\n\u003cli\u003eAvasarala, N. S.\u003cem\u003e et al.\u003c/em\u003e in \u003cem\u003e2018 IEEE Symposium on VLSI Technology.\u003c/em\u003e 209-210 (IEEE).\u003c/li\u003e\n\u003cli\u003eLi, X., Yuan, Z., Lv, S., Song, S. \u0026amp; Song, Z. Extended endurance performance and reduced threshold voltage by doping Si in GeSe-based ovonic threshold switching selectors. \u003cem\u003eThin Solid Films\u003c/em\u003e \u003cstrong\u003e734\u003c/strong\u003e, 138837 (2021). https://doi.org/10.1016/j.tsf.2021.138837\u003c/li\u003e\n\u003cli\u003eVerdy, A.\u003cem\u003e et al.\u003c/em\u003e in \u003cem\u003e2018 IEEE International Memory Workshop (IMW).\u003c/em\u003e 1-4 (IEEE).\u003c/li\u003e\n\u003cli\u003eYoo, J.\u003cem\u003e et al.\u003c/em\u003e 3D stackable and scalable binary ovonic threshold switch devices with excellent thermal stability and low leakage current for high‐density cross‐point memory applications. \u003cem\u003eAdvanced Electronic Materials\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 1900196 (2019). https://doi.org/10.1002/aelm.201900196\u003c/li\u003e\n\u003cli\u003eLee, M.\u003cem\u003e et al.\u003c/em\u003e Simple artificial neuron using an ovonic threshold switch featuring spike-frequency adaptation and chaotic activity. \u003cem\u003ePhysical Review Applied\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 064056 (2020). https://doi.org/10.1103/PhysRevApplied.13.064056\u003c/li\u003e\n\u003cli\u003eLee, D.\u003cem\u003e et al.\u003c/em\u003e Various threshold switching devices for integrate and fire neuron applications. \u003cem\u003eAdvanced Electronic Materials\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 1800866 (2019). \u003c/li\u003e\n\u003cli\u003eDing, Y.\u003cem\u003e et al.\u003c/em\u003e Engineering spiking neurons using threshold switching devices for high-efficient neuromorphic computing. \u003cem\u003eFrontiers in neuroscience\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 786694 (2022). https://doi.org/10.3389/fnins.2021.786694\u003c/li\u003e\n\u003cli\u003eWang, L., Zhang, L., Hua, S., Fu, Q. \u0026amp; Guo, X. Leaky integrate-and-fire and oscillation neurons based on ZnO diffusive memristors for spiking neural networks. \u003cem\u003eScience China Materials\u003c/em\u003e \u003cstrong\u003e68\u003c/strong\u003e, 1212-1219 (2025). https://doi.org/10.1007/s40843-024-3236-6\u003c/li\u003e\n\u003cli\u003eChoi, H.\u003cem\u003e et al.\u003c/em\u003e Spiking neural network integrated with impact ionization field‐effect transistor neuron and a ferroelectric field‐effect transistor synapse. \u003cem\u003eAdvanced Materials\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 2406970 (2025). https://doi.org/10.1002/adma.202406970\u003c/li\u003e\n\u003cli\u003eHan, J.-K.\u003cem\u003e et al.\u003c/em\u003e Mimicry of excitatory and inhibitory artificial neuron with leaky integrate-and-fire function by a single MOSFET. \u003cem\u003eIEEE Electron Device Letters\u003c/em\u003e \u003cstrong\u003e41\u003c/strong\u003e, 208-211 (2019). https://doi.org/10.1109/LED.2019.2958623\u003c/li\u003e\n\u003cli\u003eHan, J.-K.\u003cem\u003e et al.\u003c/em\u003e Cointegration of single-transistor neurons and synapses by nanoscale CMOS fabrication for highly scalable neuromorphic hardware. \u003cem\u003eScience Advances\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, eabg8836 (2021). https://doi.org/10.1126/sciadv.abg8836\u003c/li\u003e\n\u003cli\u003eDutta, S., Kumar, V., Shukla, A., Mohapatra, N. R. \u0026amp; Ganguly, U. Leaky integrate and fire neuron by charge-discharge dynamics in floating-body MOSFET. \u003cem\u003eScientific reports\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 8257 (2017). https://doi.org/10.1038/s41598-017-07418-y\u003c/li\u003e\n\u003cli\u003eTuma, T., Pantazi, A., Le Gallo, M., Sebastian, A. \u0026amp; Eleftheriou, E. Stochastic phase-change neurons. \u003cem\u003eNature nanotechnology\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 693-699 (2016). https://doi.org/10.1038/NNANO.2016.70\u003c/li\u003e\n\u003cli\u003eJia, S.\u003cem\u003e et al.\u003c/em\u003e Ultrahigh drive current and large selectivity in GeS selector. \u003cem\u003eNature Communications\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 4636 (2020). https://doi.org/10.1038/s41467-020-18382-z\u003c/li\u003e\n\u003cli\u003eZhao, Z.\u003cem\u003e et al.\u003c/em\u003e Chalcogenide ovonic threshold switching selector. \u003cem\u003eNano-Micro Letters\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 81 (2024). https://doi.org/10.1007/s40820-023-01289-x\u003c/li\u003e\n\u003cli\u003eIelmini, D. \u0026amp; Zhang, Y. Analytical model for subthreshold conduction and threshold switching in chalcogenide-based memory devices. \u003cem\u003eJournal of Applied Physics\u003c/em\u003e \u003cstrong\u003e102\u003c/strong\u003e (2007). https://doi.org/10.1063/1.2773688\u003c/li\u003e\n\u003cli\u003eLankhorst, M. Modelling glass transition temperatures of chalcogenide glasses. Applied to phase-change optical recording materials. \u003cem\u003eJournal of non-crystalline solids\u003c/em\u003e \u003cstrong\u003e297\u003c/strong\u003e, 210-219 (2002). https://doi.org/10.1016/S0022-3093(01)01034-1\u003c/li\u003e\n\u003cli\u003eGu, R.\u003cem\u003e et al.\u003c/em\u003e Nitrogen: A promising doping strategy for high-performance ovonic threshold switching selectors. \u003cem\u003eJournal of Alloys and Compounds\u003c/em\u003e \u003cstrong\u003e1005\u003c/strong\u003e, 176011 (2024). https://doi.org/10.1016/j.jallcom.2024.176011\u003c/li\u003e\n\u003cli\u003eLee, S.-B., Park, C., Park, S.-H., Lee, Y. \u0026amp; Yeo, J.-S. Effect of nitrogen doping on characteristics of SiTe Ovonic threshold switch for selectors. \u003cem\u003eMaterials Science in Semiconductor Processing\u003c/em\u003e \u003cstrong\u003e178\u003c/strong\u003e (2024). https://doi.org/10.1016/j.mssp.2024.108470\u003c/li\u003e\n\u003cli\u003eHatayama, S.\u003cem\u003e et al.\u003c/em\u003e Origins of midgap states in Te-based Ovonic threshold switch materials. \u003cem\u003eActa Materialia\u003c/em\u003e \u003cstrong\u003e258\u003c/strong\u003e, 119209 (2023). https://doi.org/10.1016/j.actamat.2023.119209\u003c/li\u003e\n\u003cli\u003eRobert, B., Flaud, V., Escalier, R., Mehdi, A. \u0026amp; Vigreux, C. XPS study of Ge-Se-Te surfaces functionalized with organosilanes. \u003cem\u003eApplied Surface Science\u003c/em\u003e \u003cstrong\u003e607\u003c/strong\u003e, 154921 (2023). https://doi.org/10.1016/j.apsusc.2022.154921\u003c/li\u003e\n\u003cli\u003eReese, M. O.\u003cem\u003e et al.\u003c/em\u003e Intrinsic surface passivation of CdTe. \u003cem\u003eJournal of Applied Physics\u003c/em\u003e \u003cstrong\u003e118\u003c/strong\u003e (2015). https://doi.org/10.1063/1.4933186\u003c/li\u003e\n\u003cli\u003eLi, H.-Y. \u0026amp; Lii, K.-H. Hydrothermal synthesis, crystal structures, and X-ray photoelectron spectroscopy of lead tellurium (iv) and tellurium (vi) oxycompounds: Ba 3 PbTe 6 O 16 and Na 2 Pb 9 (\u0026mu; 6-O) 2 (Te 2 O 10) 2. \u003cem\u003eDalton Transactions\u003c/em\u003e \u003cstrong\u003e50\u003c/strong\u003e, 8189-8195 (2021). https://doi.org/10.1039/D1DT00566A\u003c/li\u003e\n\u003cli\u003eLan, W., Cao, L., Fu, Y., Fang, J. \u0026amp; Wang, J. Structure and optical properties of GeTe film controlled by amorphous to crystalline phase transition. \u003cem\u003eVacuum\u003c/em\u003e \u003cstrong\u003e197\u003c/strong\u003e, 110847 (2022). https://doi.org/10.1016/j.vacuum.2021.110847\u003c/li\u003e\n\u003cli\u003eJagannatha, K. \u0026amp; Das, C. Switching, optical, Raman, and Morphological investigations on Sn doped SiTe glasses and thin films. \u003cem\u003eThin Solid Films\u003c/em\u003e \u003cstrong\u003e759\u003c/strong\u003e, 139472 (2022). https://doi.org/10.1016/j.tsf.2022.139472\u003c/li\u003e\n\u003cli\u003eRao Gunti, S. \u0026amp; Asokan, S. Thermodynamic, Raman and electrical switching studies on Si15Te85-xAgx (4\u0026le; x\u0026le; 20) glasses. \u003cem\u003eJournal of Applied Physics\u003c/em\u003e \u003cstrong\u003e111\u003c/strong\u003e (2012). https://doi.org/10.1063/1.3682759\u003c/li\u003e\n\u003cli\u003eZhao, Z.\u003cem\u003e et al.\u003c/em\u003e Volatile to Non‐Volatile Switching Transition in Chalcogenides. \u003cem\u003eAdvanced Functional Materials\u003c/em\u003e, 2423940 (2025). https://doi.org/10.1002/adfm.202423940\u003c/li\u003e\n\u003cli\u003eKumar, R. A., Mousa, H. A., Lekha, P. C., Mahmoud, S. T. \u0026amp; Qamhieh, N. in \u003cem\u003eJournal of Physics: Conference Series.\u003c/em\u003e 012018 (IOP Publishing).\u003c/li\u003e\n\u003cli\u003eJi, X.\u003cem\u003e et al.\u003c/em\u003e Titanium-induced structure modification for thermal stability enhancement of a GeTeTi phase change material. \u003cem\u003eRSC Advances\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 24966-24974 (2015). https://doi.org/10.1039/C4RA11504J\u003c/li\u003e\n\u003cli\u003eAndrikopoulos, K.\u003cem\u003e et al.\u003c/em\u003e Raman scattering study of the a-GeTe structure and possible mechanism for theamorphous to crystal transition. \u003cem\u003eJournal of physics: condensed matter\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 965 (2006). https://doi.org/10.1088/0953-8984/18/3/014\u003c/li\u003e\n\u003cli\u003eKolobov, A. V., Fons, P. \u0026amp; Tominaga, J. Local instability of p-type bonding makes amorphous GeTe a lone-pair semiconductor. \u003cem\u003ePhysical Review B\u0026mdash;Condensed Matter and Materials Physics\u003c/em\u003e \u003cstrong\u003e87\u003c/strong\u003e, 155204 (2013). https://doi.org/10.1103/PhysRevB.87.155204\u003c/li\u003e\n\u003cli\u003eKolobov, A., Fons, P. \u0026amp; Tominaga, J. Understanding phase-change memory alloys from a chemical perspective. \u003cem\u003eScientific reports\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 13698 (2015). https://doi.org/10.1038/srep13698\u003c/li\u003e\n\u003cli\u003eVelea, A.\u003cem\u003e et al.\u003c/em\u003e Te-based chalcogenide materials for selector applications. \u003cem\u003eScientific reports\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 8103 (2017). https://doi.org/10.1038/s41598-017-08251-z\u003c/li\u003e\n\u003cli\u003eKoo, Y., Baek, K. \u0026amp; Hwang, H. in \u003cem\u003e2016 IEEE symposium on VLSI technology.\u003c/em\u003e 1-2 (IEEE).\u003c/li\u003e\n\u003cli\u003eKastner, M., Adler, D. \u0026amp; Fritzsche, H. Valence-alternation model for localized gap states in lone-pair semiconductors. \u003cem\u003ePhysical Review Letters\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 1504 (1976). https://doi.org/10.1103/PhysRevLett.37.1504\u003c/li\u003e\n\u003cli\u003eWu, R.\u003cem\u003e et al.\u003c/em\u003e The role of arsenic in the operation of sulfur-based electrical threshold switches. \u003cem\u003eNature Communications\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 6095 (2023). https://doi.org/10.1038/s41467-023-41643-6\u003c/li\u003e\n\u003cli\u003eNo\u0026eacute;, P.\u003cem\u003e et al.\u003c/em\u003e Toward ultimate nonvolatile resistive memories: The mechanism behind ovonic threshold switching revealed. \u003cem\u003eScience Advances\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e (2020). https://doi.org/10.1126/sciadv.aay2830\u003c/li\u003e\n\u003cli\u003eInc., V. L. \u003cem\u003eMaterials Square\u003c/em\u003e, 2017.01.01).\u003c/li\u003e\n\u003cli\u003ePlimpton, S. Fast parallel algorithms for short-range molecular dynamics. \u003cem\u003eJournal of computational physics\u003c/em\u003e \u003cstrong\u003e117\u003c/strong\u003e, 1-19 (1995). https://doi.org/10.1006/jcph.1995.1039\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7764312/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7764312/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eNeuromorphic computing, inspired by biological nervous systems, yields high energy efficiency and data throughput by integrating computation and storage within memory crossbar arrays. A key requirement for neuromorphic hardware is an artificial neuron capable of low-power, high-frequency operation. Ovonic threshold switch (OTS) devices have attracted attention due to their scalability and intrinsic capacitance, enabling simple circuitry to demonstrate leaky integrate-and-fire (LIF) behavior. This study proposes a sustainable and scalable OTS device fabricated with non-toxic, industry-friendly materials and provides insights into the roles of individual elements in bond formation, correlating with enhanced electrical performance (J\u003csub\u003eoff\u003c/sub\u003e = 2.3 ∙ 10\u003csup\u003e\u0026minus;\u0026thinsp;8\u003c/sup\u003e MA/cm\u0026sup2;). Finally, our optimized NSGT OTS device demonstrates low-power spiking operation, achieving 0.56 pJ/\u0026micro;m\u003csup\u003e2\u003c/sup\u003e per spike. These findings establish stoichiometric guidelines for designing high-performance Te-based OTS devices for energy-efficient neuromorphic computing.\u003c/p\u003e","manuscriptTitle":"Scalable and sustainable N-Si-Ge-Te Ovonic threshold switching devices for energy-efficient artificial neuron applications","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-24 17:27:34","doi":"10.21203/rs.3.rs-7764312/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"45f35f8f-d22c-4f9b-8518-2208d4d5f4cc","owner":[],"postedDate":"December 24th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":60111149,"name":"Physical sciences/Materials science/Materials for devices/Electronic devices"},{"id":60111150,"name":"Physical sciences/Materials science/Nanoscale materials/Electronic properties and materials"}],"tags":[],"updatedAt":"2026-02-05T18:56:57+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-24 17:27:34","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7764312","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7764312","identity":"rs-7764312","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}