Amorphous, fully-stoichiometric molybdenum oxide for high performance nonvolatile resistive switching memory: The role of stoichiometry on synaptic plasticity

Amorphous, fully-stoichiometric molybdenum oxide for high performance nonvolatile resistive switching memory: The role of stoichiometry on synaptic plasticity | 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 Research Article Amorphous, fully-stoichiometric molybdenum oxide for high performance nonvolatile resistive switching memory: The role of stoichiometry on synaptic plasticity Gion Kalemai, Konstantinos Aidinis, Michael-Alexandros Kourtis, and 2 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6829733/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 04 Dec, 2025 Read the published version in Discover Materials → Version 1 posted 11 You are reading this latest preprint version Abstract Transition metal oxides (TMOs) are a promising class of materials for neuromorphic computing and processing systems demonstrating a variety of resistive switching (RS) mechanisms. However, little is known about the correlation between its stoichiometry and RS. This study is focused on the development and characterization of amorphous molybdenum oxide memristors with different stoichiometry. Fully-stoichiometric (MoO 3 ) and hydrogenated sub-stoichiometric (H-MoO 3 − x ) amorphous molybdenum oxide thin films were developed via a hot-wire chemical vapor deposition system. Both, stoichiometric and hydrogenated sub-stoichiometric molybdenum oxide devices showed good resistive switching behavior. However, the fully-stoichiometric memristor exhibited better RS properties with endurance of 250 cycles, ON/OFF ratio ~ 10 3 and high retention of almost 3·10 4 s, compared with the poor RS behavior of the device based on the H-MoO 3 − x film. This impressive memristive behavior could be attributed to the excess of oxygen atoms in the case of fully-stoichiometric memristor in respect to the sub-stoichiometric H-MoO 3 − x which play crucial role in the conductive behavior of the device. The high reproducibility observed in MoO 3 -based memristor highlights their potential for practical applications and scalability. Additionally, the outstanding features of the MoO 3 memristor demonstrated through its long-term potentiation (LTP), long-term depression (LTD), and spike-timing dependent plasticity (STDP) indicate that the fully stoichiometric molybdenum oxide memristor has significant potential for simulating biological synapses, opening doors to a new era in neuromorphic computing applications. Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Introduction Non-volatile memristive (NVMs) devices have attracted much attention due to the high-speed and low-energy processing and storage required by the huge amount of data that annually produced 1 . The existing silicon-based memory technologies are encountering major obstacles including slow operational speeds, limited storage capacity, and high energy consumption 2 – 4 . Therefore, there is a strong demand for groundbreaking memory technologies that offer extremely high speed, ultra-long retention, high capacity, and minimal energy use, utilizing innovative principles, materials, and structures 5 – 7 . Memristors are a category of NVMs that are recognized for their low energy consumption, high storage capacity and capability to mimic the synaptic plasticity of the brain making them promising candidates for neuromorphic computing 7 – 11 . A memristor is a two-terminal electronic device based on the resistive switching (RS) effect from the low resistance state (LRS) to the high resistance state (HRS). In a representative structure, the RS layer is sandwiched between two metals, serving as the top and bottom electrodes 12 . When a voltage is applied to the device, a conductive filament (CF) is formed (disrupted) within the RS layer leading the HRS-LRS (LRS-HRS) transition. The CF can be vacancy- or metallic-mediated assigned to the migration of oxygen anion species 13 or the electrochemical metallization 14 , respectively. A variety of materials has been investigated as resistive switching layers in memristors including transition metal oxides (TMOs) 15 , 16 , 2D materials 17 , 18 , organic semiconductors 19 , 20 and inorganic-organic halide perovskites 21 , 22 . TMOs are a class of materials widely used in memristors, whose RS behavior is determined by the oxygen vacancy distribution within the oxide layer. Among them molybdenum oxide (MoO 3 ) is a wide bandgap TMO that has been applied not only in memristors 23 but also in a variety of electronic applications including sensors, catalysis, solar cells, and light emitting diodes 24 – 29 . MoO 3 is an exceptional functional layer for memristor fabrication showing also artificial synaptic properties 30 due to its abundant, low-cost, and environmentally friendly nature 31 along with its outstanding electrical and thermal stability, advantageous for memristive devices 32 . Another advantage of MoO 3 is the ease of tuning its electrical properties by controlling the stoichiometry depending on the development procedure and method 33 , 34 . The presence of oxygen vacancies significantly influences the conductive mechanism of the memristor. Therefore, the proper amount of oxygen in the resistive switching layer is essential in order to achieve the desired memristor performance. There are a few reports over the last ten years regarding the development of amorphous molybdenum oxide functional layer in memristors 35 , 36 . Xue et al demonstrated the RS behavior and synaptic plasticity of a memristor using amorphous MoO x prepared at room temperature by magnetron sputtering 35 . Recently, solution-processed MoO 3 -based memristors reported excellent nonvolatile properties upon annealing at 250 o C, which didn’t affect the amorphous nature of the metal oxide 36 . On the other hand, most of the reports on molybdenum oxide-based memristors are referred to crystalline MoO 3 nanobelts 37 , 38 , 39 , microrods 40 , heterostructures with other metal oxides 41 or transition metal dichalcogenodes 42 , 43 , and preparation methods using high temperatures during deposition 44 . More recently, Rajesh et al. 45 investigated the polycrystalline Mo x O y structure deposited at different oxygen/argon (O/Ar) concentration using magnetron sputtering altering the amount of oxygen vacancies in the Mo x O y films. The impact of the oxygen vacancies on the artificial synapses was also demonstrated reporting that the Au/Mo x O y /FTO artificial synaptic device exhibited excellent neuromorphic characteristics. However, crystalline MoO 3 are less uniform layers compared to amorphous films due to grain boundaries limiting high-density integration 46 . In this work, the influence of stoichiometry of amorphous molybdenum oxide on the resistive switching behavior of memristors is investigated. Therefore, vertical two-terminal memristors using amorphous fully-stoichiometric (MoO 3 ) and hydrogenated sub-stoichiometric (H-MoO 3 − x ) as functional layers were fabricated. Both, fully-stoichiometric and hydrogenated sub-stoichiometric molybdenum oxide devices demonstrated favorable resistive switching characteristics. Nevertheless, the fully-stoichiometric memristor showcased superior RS performance, with an endurance of 250 cycles, an impressive retention time of nearly 3·10 4 seconds and an ON/OFF ratio of approximately 10 3 , in respect with the poor RS behavior exhibited by the device employing the H-MoO 3 − x film. This remarkable memristive performance can be attributed to the abundance of oxygen vacancies present in the fully-stoichiometric memristor compared to the sub-stoichiometric H-MoO 3 − x , which are essential for the formation of the conductive filament in the RS layer of the device. The high level of reproducibility found in MoO 3 -based memristors underscores their potential for real-world applications and scalability. Furthermore, the remarkable capabilities of the MoO 3 memristor showcased through its long-term potentiation (LTP), long-term depression (LTD), and spike-timing dependent plasticity (STDP). This reveals that the fully stoichiometric molybdenum oxide memristor holds immense promise for emulating biological synapses, paving the way for a new era in neuromorphic computing applications. 2. Experimental Details 2.1 Thin Film Deposition and Device Fabrication Memristors were fabricated following the metal-insulator-metal (MIM) structure on fluorine-doped tin oxide (FTO) used as the bottom electrode of the devices. Prior to molybdenum oxide deposition, glass/FTO substrates were cleaned using ultrasonication bath of deionized water, acetone, and isopropyl alcohol for 10 minutes each. Fully-stoichiometric (MoO 3 ) and hydrogenated sub-stoichiometric molybdenum oxide (H-MoO 3 − x ) thin films were then deposited on FTO substrates using a homemade hot-wire chemical vapor deposition system (HW-CVD), as already described elsewhere 47 . In brief, FTO substrates were placed on an aluminum susceptor in a stainless-steel reactor. 2.5 cm upwards of the FTO substrates a molybdenum (Mo) wire with diameter of 0.5 mm was placed between two copper leads. The deposition began with the reactor’s evacuation down to 10 mTorr using a mechanical pump. For the development of fully-stoichiometric molybdenum oxide high purity oxygen (O 2 ) gas 99.999% was used, while the deposition of hydrogenated sub-stoichiometric molybdenum oxide was taken place using forming gas; a gas mixture of 90% nitrogen and 10% hydrogen. MoO 3 and H-MoO 3 − x films were deposited by setting the base pressure at 80 mTorr and simultaneously heating the Mo wire at 560 o C. The thickness of the prepared films was 100 nm. Finally, the memristors were completed with the deposition through thermal evaporation of a 100 nm-thick aluminum (Al) layer serving as the top electrode. 2.2 Thin Film and Device Characterization The molybdenum oxide films optical properties were obtained using a Perkin Elmer Lambda 40 UV/Vis spectrometer recording the transmittance and absorbance spectra in the wavelength range of 200–900 nm. A Bruker Tensor 27 Fourier transform infrared (FTIR) spectrometer equipped with a DTGS detector was used to record FTIR spectra of the deposited films using transmittance mode. The X-ray diffractograms (XRD) of the MoO 3 and H-MoO 3 − x films was taken using a Smart Lab Rigaku diffractometer (θ/θ scan) with a CuKA radiation (3 kW). The surface morphology of the samples was investigated by scanning electron microscopy (SEM) using a JEOL 7401f FESEM microscope. The electrical characterization of the fabricated memristors was performed using a VersaSTAT4 potentiostat. The bias was applied to the top electrode (Al) maintaining the FTO-bottom electrode connected to the ground. 3. Results and Discussion 3.1 Thin Film Characterization 100 nm-thick fully-stoichiometric (MoO 3 ) and hydrogenated sub-stoichiometric (H-MoO 3 − x ) films were prepared using a hot-wire chemical vapor deposition system, as described in details in the Experimental section. It is demonstrated that both films are amorphous, as revealed from the XRD measurements shown in Figure S1 a (Supporting Information), where no XRD peaks appeared in the diffractograms of MoO 3 and H-MoO 3 − x . Furthermore, the surface nanomorphology of the deposited molybdenum oxide films is composed of grain-like structure, as it can be seen in SEM images, shown in Figure S1 b and c for the MoO 3 and H-MoO 3 − x film, respectively. The surface of the fully-stoichiometric molybdenum oxide consists of large grains with dimensions of ~ 40 nm (Figure S1 b). On the other hand, the H-MoO 3 − x film is composed by smaller grains compared to the MoO 3 sample with dimensions of ~ 25 nm (Figure S1 c). Figure 1 d shows the photoluminescence (PL) emission spectra of MoO 3 and H-MoO 3 − x films at an excitation wavelength of 325 nm. In the case of H-MoO 3 − x , the first weak peak is found at 335 nm, which is located close to the near band edge (NBE) emission and can be attributed to the free exciton recombination. In addition, the broad peak at 363 nm in the PL spectrum of H-MoO 3 − x can be assigned to the Mo 5+ d xy – d yz band transition, while the weak peak at 469 nm is mainly due to the deep-level transitions caused by the existence of defects and oxygen vacancies. In the case of MoO 3 sample, the peak at 469 nm is more pronounced indicating the excess of oxygen vacancies compared with the H-MoO 3 − x . These oxygen vacancies may be beneficial on the resistive switching effect. The stoichiometry and electronic properties were estimated using X-ray (XPS) and ultraviolet (UPS) photoelectron spectroscopy measurements presented in previous work 47 demonstrating that the molybdenum oxide film deposited in O 2 environment was fully-stoichiometric, while forming gas resulted in the hydrogenation of the sub-stoichiometric molybdenum oxide. Moreover, changes in the UPS spectra of the MoO 3 and H-MoO 3 − x were observed, where the work function (W F ) is significantly reduced upon metal oxide hydrogenation. Occupied states inside the band gap of the H-MoO 3 − x were also appeared. These gap states resulted in reduced energy gap of around 2.4 eV compared with that of MoO 3 (3 eV), as evidenced from the Tauc plot presented in Figure S2a in the Supporting Information. Furthermore, the hydrogenated sub-stoichiometric molybdenum oxide exhibited reduced transmittance in the visible region, as shown in Figure S2b (increased absorbance, Figure S2c) in respect to the MoO 3 film. In order to further investigate the structure of the prepared molybdenum oxide samples, FTIR transmittance measurements were performed on MoO 3 and H-MoO 3 − x films. The FTIR transmittance spectra of the MoO 3 and H-MoO 3 − x are shown in Figure S1 e and S1f, respectively. For both samples, the fingerprint region located in the wavenumber range of 400–1100 cm − 1 contains three transmittance bands indicating different types of Mo – O bonding in MoO 3 lattice structure. Particularly, the bands at 635 cm − 1 and 561 cm − 1 are assigned to Mo – O bending vibration mode of the symmetric bridging oxygen bonds to three Mo atoms (O – Mo3). The bands observed in the region 880–650 cm − 1 are attributed to the stretching vibration of the asymmetric bridging oxygen connected to two neighboring Mo atoms (O – Mo2), while the peaks at 951 cm − 1 and 905 cm − 1 are assigned to the stretching mode of the terminal oxygen (Mo = O bond) 48 – 51 . Upon hydrogenation, a shift of Mo = O bond to higher frequency values was observed. Moreover, the O – Mo2 band was shifted to lower frequency values suggesting the incorporation of hydrogen atoms in the molybdenum oxide structure. 3.2 MoO 3 - and H-MoO 3 − x -based memristor’s characterization In order to investigate the influence of the stoichiometry of the amorphous molybdenum oxide films on the memristive behavior, memristors based on the fully-stoichiometric and hydrogenated sub-stoichiometric films were fabricated with the device structure FTO/MoO 3 or H-MoO 3 − x /Al, shown as inset in Fig. 1 a and 1 b, respectively. The FTO (fluorine-doped tin oxide) serving as the bottom electrode was grounded, while the top aluminum (Al) electrode was biased. Current – voltage (I – V) characteristic curves were recorded by performing cycle sweeps of voltage (– 6V → 0V → + 3V → 0V → – 6V). A typical bipolar resistive switching (RS) memory behavior is observed for both memristors as revealed from recorded I – V curves presented in Fig. 1 a and 1 b for the devices based on amorphous MoO 3 and H-MoO 3 − x films. However, in the case of the fully-stoichiometric-based device, the memristive behavior is more pronounced. Initially the devices are in the high resistance state (HRS). When a forward voltage was applied between the bottom and the top electrode, the transition from the HRS to the lower resistance state (LRS) occurred resulting in the ON switching, also referred as SET. More importantly, the transition from HRS to LRS in MoO 3 -based device is more abrupt (Fig. 1 a) than in the case of the memristor with H-MoO 3 − x (Fig. 1 b), where the current gradually increases (decreases) in the SET (RESET) operation. Next, the devices returned to the HRS (RESET) by applying a bias of opposite polarity. To examine the repeatability and non-volatility of the devices, several repeated cycles of sweep voltage from − 6 V to + 3 V were employed to the devices. Figure 1 a and 1 b presents the semi-logarithmic I – V characteristic curves for consecutive 250 cycles. It is observed that both memristors exhibited good repeatability maintaining stable resistive switching behavior even after 250 cycles. Figure S3a and S3b represents the distribution of RESET and SET voltage, respectively, of 250 cycles for the MoO 3 -based device exhibiting the best memristive behavior, where the estimated standard deviation values (σ) are 0.40 V and 0.33 V, respectively. The MoO 3 -based memristor demonstrates also low threshold voltages with mean value of 0.96 V and − 2.01 V. Furthermore, both memristors maintain their resistance states steady as revealed from the endurance plots, shown in Fig. 1 c and 1 d for the devices with the MoO 3 and H-MoO 3 − x , respectively, indicating the reproducibility of the devices upon cycle repeat. From the estimated cumulative probability shown in Fig. 1 e and 1 f, the MoO 3 device exhibits high HRS/LRS ratio of 10 3 , while memory window of the H-MoO 3 − x memristor is very small (HRS/LRS = 2.5). Figure 1 g and 1 h represents the retention performance of LRS and HRS of the fully-stoichiometric and hydrogenated sub-stoichiometric molybdenum oxide based memristive devices, respectively, after applying a pulse with amplitude of + 1.5 V for SET, -4 V for RESET, and setting the reading voltage V read at 0.1 V. Reliable non-volatility is clearly seen in both cases, where no significant change observed after 2.8·10 4 sec. To set more light into the underlying conduction mechanism of our best performed MoO 3 memristor, the log(I) – log(V) plots of the LRS and HRS was fitted and presented in Fig. 2 . It is observed that I-V characteristic curve of the HRS shown in Fig. 2 a and 2 b consists of three distinguished regions; i) the ohmic region, where the current is linearly related to the applied voltage ( \(I \propto V\) ) originated by the thermally generated free carriers, ii) Schottky emission (SE), assigned to the oxygen vacancies at the MoO 3 /Al interface, and iii) the space charge limited current (SCLC), where the current is followed by a quadratic term ( \(I \propto {V^2}\) ) attributed to the traps that are filled by charges. Specifically, at low voltages of HRS, the slope estimated by fitting the experimental data is ~ 1 suggesting ohmic conduction behavior (Fig. 2 a). When the applied voltage increases, the Schottky model well fits the I-V characteristic curve (Fig. 2 b). The charges localized at the interface between the MoO 3 and the bottom electrode (FTO), are trapped by the HRS layer’s empty trap sites of MoO 3 . As the applied voltage further increases, the empty trap sites are gradually occupied until fully completed by electrons. The slope of fitting model significantly increases suggesting the dominance of the SCLC conduction mechanism. On the other hand, the LRS region is well fitted by ohmic and SE models, as presented in Fig. 2 c and 2 d. Ohmic conduction mechanism (slope equal to 1) is dominant at low voltages, while at higher voltages the slope of fitting curve is 3.6 indicating that the conduction mechanism obeys the SE model, which is attributed to the barrier at the FTO/MoO 3 interface. Same results also obtained in the case of H-MoO 3 − x , as shown in log(I) – log(V) and Ln(I) – V 1/2 plots of the LRS and HRS presented in Figure S4a-d. However, as the W F of the fully-stoichiometric molybdenum oxide is higher than that of hydrogenated sub-stoichiometric the Schottky barrier height increases resulting in an increase in the HRS, which therefore leads to much larger ON/OFF ratio. Figure 3 illustrates the switching mechanism in the MoO 3 -based memristor. In the pristine state (Fig. 3 a) without the application of electrical bias, the oxygen vacancies are randomly distributed in the MoO 3 layer, and thus the device is on HRS. When a positive bias is applied to the FTO electrode, partial reduction of the molybdenum oxide occurs and oxygen vacancies accumulate at the bottom electrode. The migration of the oxygen vacancies forms a conductive filament (Fig. 3 b), which facilitates the electron transport between the two electrodes through the MoO 3 film. The current abruptly jumps and then gradually increases resulting in the HRS to LRS transition. The sudden change in resistive switching is more noticeable in the MoO 3 functional layer than in H-MoO 3 − x , which is attributed to the higher number of oxygen vacancies as indicated by the PL measurements. In the case of H-MoO 3 − x the lesser oxygen vacancies result in a gradually HRS to LRS transition. On the other hand, when the polarity of the applied voltage is changed, the oxygen vacancies are suppressed and the conductive filament is disrupted (Fig. 3 c). In this case, a Schottky barrier gap is formed between the tip of the conductive filament and the top electrode. The MoO 3 -based memristor showed also a multilevel resistive switching due to the oxygen-rich conductive filament formed during the LRS and Schottky gaps during the disruption of the conductive filament. By applying different RESET voltages of -5V and − 2V, we can modulate the Schottky barrier (Fig. 3 d) getting different HRS, as presented in Fig. 3 e. It is observed that increasing the RESET voltage, the Schottky gap also increases. 3.3 MoO 3 memristor’s synaptic behavior In order to investigate the long-term stability of the MoO 3 -based memristor, two sets of scanning voltages, one positive (+ 1V) and one negative (-1V), were applied to the device, and the output current was recorded. Figure 4 a shows the evolution of the MoO 3 memristor as a function of the alternative positive and negative voltage sweeps applied to the device. The positive and negative voltage sweeps were applied on the device 10 times, sequentially, totally 100 times. In particular, the first positive voltage sweep was conducted applying the positive voltage for 10 times, which results in an increase of the current, and thus the conductance. Next, the negative voltage sweep was followed for 10 times. It is observed that the absolute value of current gradually decreases in the negative direction, leading to the increase of the conductance. Therefore, the synaptic weight could be enhanced (suppressed) upon the application of serial excitatory (inhibitory) spikes, which is analogous to the biological synapses. The synaptic plasticity, named as synaptic potentiation, which is the ability to modulate and retain the synaptic weight representing the level of learning and memory divided into short-term (STP) and long-term (LTP) potentiation 52 is also studied. It is noted that the short-term (STD) and long-term (LTD) depression represent the depression of synaptic weight 53 . To simulate the LTP and LTD of the MoO 3 -based device, 80 positive and 80 negative pulses were sequentially applied. When 80 continuous positive programming pulses are applied the current, and thus the conductance, of the device increases, as shown in Fig. 4 b. After the application of 80 continuous negative programming pulses, the current of the MoO 3 memristor decreases. It is suggested that the post-synaptic current could be bidirectional controlled by the sequentially application of positive and negative pulses, which is essential for application in neuromorphic computing, neural networks, and biophysics 54 . By adjusting the input voltage, the conductance of the device can be also regulated. Figure 54c shows the MoO 3 memristor response to mixed potentiating and depressing pulse voltage with different amplitude and same pulse width (500 µs). The conductance change, ΔG, was estimated using the Eq. ( 1 ): $$\Delta G=\frac{{{G_1} - {G_2}}}{{{G_1}}}100\%$$ 1 Where G 1 is the conductance recorded before each pulse voltage, and G 2 is the conductance after the pulse. It is observed that, the conductance changes of the device are grater (smaller) upon the application of larger (smaller) positive pulse. The same phenomenon is also observed in the application of negative pulse, suggesting that the device can imitate the biological synapses 55 . Figure 4 d shows the spike timing dependent plasticity (STDP) characteristics of the FTO/MoO 3 /Al synaptic device. STDP can regulate the synaptic weight through the timing between the pre-synaptic and post-synaptic pulse 56 . Δt (Δt = t post – t pre ) represents the time between the application of pre-synaptic (with amplitude of -0.4 V and width of 500 µs) and post-synaptic (with amplitude of + 0.4 V and width of 500 µs) neuron pulse. t post and t pre are the time for applying post-synaptic and pre-synaptic pulse, respectively. In the case of Δt < 0, meaning that the pre-synaptic pulse comes after the post-synaptic pulse, the synaptic weight gradually decreases. When the post-synaptic pulse comes after the application of pre-synaptic pulse (Δt > 0), the synaptic weight gradually increases. A strong functional relationship between the Δt and the change of the synaptic weight is also observed, where the change of the synaptic weight versus the Δt can be fitted with an exponential decay function. The LTP (for Δt > 0) and LTD (for Δt < 0) are well emulated, demonstrating that the FTO/MoO 3 /Al memristor is analogous to biological systems. 4. Conclusions In summary, the influence of the stoichiometry of molybdenum oxide films on the memristive behavior was investigated. Therefore, memristors based on the fully-stoichiometric (MoO 3 ) and hydrogenated sub-stoichiometric (H-MoO 3 − x ) were fabricated. The devices exhibited excellent memristive behavior including good endurance and retention. In addition, the MoO 3 device exhibited higher ON/OFF ratio of ~ 10 3 compared with that of the H-MoO 3 − x based device (~ 2.5) attributed to the formation and destruction of conductive filament due to the excess of oxygen vacancies in the MoO 3 film. Furthermore, the excellent long-term potentiation (LTP), long-term depression (LTD), and spike timing dependent plasticity (STDP) of the MoO 3 memristor was demonstrated. Therefore, it can be concluded that the fully-stoichiometric molybdenum oxide memristor shows a great potential on mimicking biological synapses for neuromorphic computing applications. Declarations Data availability The authors confirm that the data supporting the findings of this study are available within the article. If someone wants to request the data from this study, the corresponding author (Anastasia Soultati) should be contacted. Funding No funding. Corresponding author Correspondence to Anastasia Soultati, [email protected] Ethics declarations Not applicable. Consent to participate Not applicable. Consent for publication Not applicable. Competing interests The authors declare no competing interests. References Hwang CS. Prospective of semiconductor memory devices: from memory system to materials. Adv Electron Mater. 2015;1:1400056. Bertolazzi S, Bondavalli P, Roche S, et al. Nonvolatile memories based on graphene and related 2D materials. Adv Mater. 2019;31:1806663. Li J, Zhang H, Ding Y, Li J, Wang S, Zhang DW, Zhou P. A non-volatile and gate based on Al 2 O 3 /HfO 2 /Al 2 O 3 charge-trap stack for in-situ storage applications. Sci Bull. 2019;64:1518–24. Ding Y, Liu L, Li J, Cao R, Jiang Y-G, Liu C, Liu Q, Zhou P. A semi-floating memory with 535% enhancement of refresh time by local field modulation. Adv Funct Mater. 2020;30:1908089. Vu Q, Shin Y, Kim Y, et al. Two-terminal floating-gate memory with van der Waals heterostructures for ultrahigh on/off ratio. Nat Commun. 2016;7:12725. Li Y, Zhang ZC, Li J, et al. Low-voltage ultrafast nonvolatile memory via direct charge injection through a threshold resistive-switching layer. Nat Commun. 2022;13:4591. Wang R, Song Z, Song W, Xin T, Lv S, Song S, Liu J. Phase-change memory based on matched Ge-Te, Sb-Te, and In-Te octahedrons: improved electrical performances and robust thermal stability. InfoMat. 2021;3:1008–15. Sun K, Chen J, Yan X. The future of Memristors: Materials Engineering and Neural Networks. Adv Funct Mater. 2021;31:2006773. Rajasekaran S, Simanjuntak FM, Chandrasekaran S, Panda D, Saleem A, Tseng TY. Flexible Ta 2 O 5 /WO 3 -Based Memristor Synapse for Wearable and Neuromorphic Applications. IEEE Electron Device Lett. 2022;43(1):9–12. Yan X, Zhao Q, Chen AP, et al. Vacancy-Induced Synaptic Behavior in 2D WS 2 Nanosheet-based Memristor for Low-Power Neuromorphic Computing. Small. 2019;15:1901423. Ju JH, Seo S, Baek S, Lee D, et al. Two-Dimensional MXene Synapse for Brain-Inspired Neuromorphic Computing. Small. 2021;17:2102595. Khan R, Ilyas N, Shamim MZM, et al. Oxide-based resistive switching-based devices: fabrication, influence parameters and applications. J Mater Chem C. 2021;9:15755–88. Yildirim H, Pachter R. Mechanistic Analysis of Oxygen Vacancy-Driven Conductive Filament Formation in Resistive Random Access Memory Metal/NiO/Metal Structures. ACS Appl Mater Interfaces. 2018;10(11):9802–16. Kim M-S, Park E, Kim S-G, et al. Highly Reliable Electrochemical Metallization Threshold Switch Through Conductive Filament Engineering Using Two-Dimensional PtSe 2 Insertion Layer. Adv Mater Interfaces. 2023;10(7):2202296. Carlos E, Branquinho R, Martins R, Kiazadeh A, Fortunato E. Recent Progress in Solution-Based Metal Oxide Resistive Switching Devices. Adv Mater. 2021;33:2004328. Cho D-Y, Luebben M, Wiefels S, Lee K-S, Valov I. Interfacial Metal-Oxide Interactions in Resistive Switching Memories. ACS Appl Mater Interfaces. 2017;9(22):19287–95. Lee D, Kim S-M, Park J-C, Jung Y, Lee S, Lee BH, Lee S. Enhancing reliability in oxide-based memristors using two-dimensional transition metal dichalcogenides. Appl Surf Sci. 2025;679:161216. Zhou Z, Yang F, Wang S, et al. Emerging of two-dimensional materials in novel memristor. Front Phys. 2022;17:23204. Casula G, Busby Y, Franquet A, Spampinato V, Houssiau L, Bonfiglio A, Cosseddu P. A flexible organic memory device with clearly disclosed resistive switching mechanism. Org Electron. 2019;64:209–15. Zhou Z, Mao H, Wang X, et al. Transient and Flexible Polymer Memristors Utilizing Full-Solution Processed Polymer Nanocomposites. Nanoscale. 2018;10(31):14824–29. John RA, Demirağ Y, Shynkarenko Y, et al. Reconfigurable halide perovskite nanocrystal memristors for neuromorphic computing. Nat Commun. 2022;13:2074. Li B, Xia F, Du B, Zhang S, Xu L, Su Q, Zhang D, Yang J. 2D Halide Perovskite for High-Performance Resistive Switching Memory and Artificial Synapse Applications. Adv Sci. 2024;11:2310263. Arita M, Kaji H, Fujii T, Takahashi Y. Resistance Switching Properties of Molybdenum Oxide Films. Thin Solid Films. 2012;520(14):4762–67. Medeiros SASL, Oliviera ALM, Duarte TM. a-MoO 3 Micro- and Nanoparticles as Catalysts for Biofuel Production. ACS Appl Nano Mater. 2025;8(9):4339–53. Sanchez-Fernandez LS, Guzmán Vargas A, de La Amador LO. Manufacture and study of a chemiresistive gas sensor using MoO 3 as a sensitive film. MRS Adv. 2025;10:592–97. Gong Y, Dong Y, Zhao B, Yu R, Hu S, Tan Z. Diverse applications of MoO 3 for high performance organic photovoltaics: fundamental, processes and optimization strategies. J Mater Chem A. 2020;8:978–09. Hong J, Xu Z, Hu T, Lee S, Gui M, Kahn A, Rand BP. Evaporated organic-MoO 3 composite hole transport layers towards stable perovskite solar cells. EES Sol. 2025;1:107–14. Choi DK, Kim DH, Lee CM, et al. Highly efficient, heat dissipating, stretchable organic light-emitting diodes based on a MoO 3 /Au/MoO 3 electrode with encapsulation. Nat Commun. 2021;12:2864. Nie L, Fan J, Li Y, Xiang C, Zhang T. Direct Optical Patterning of MoO 3 Nanoparticles and Their Application as a Hole Injection Layer for Solution-Processed Quantum Dot Light-Emitting Diodes. ACS Appl Nano Matter. 2024;7(8):9499–06. Sunny MM, Thamankar R. Spike rate dependent synaptic characteristics in lamellar multilayered alpha-MoO 3 based two-terminal devices – efficient way to control the synaptic amplification. RSC Adv. 2024;14:2518–28. Alizadeh S, Hassanzadeh-Tabrizi SA. MoO 3 fibers and belts: Molten salt synthesis, characterization and optical properties. Ceram Int. 2015;41:10839–43. Balendhran S, Walia S, Nili H, Ou J, Zhuiykov S, Kaner R, Sriram S, Bhaskaran M. Kalantar-zadeh K. Two-dimensional molybdenum trioxide and dichalcogenides. Adv Funct Mater. 2013;23:3952–70. Reddy K, Mhamane N, Ghosalya M, Gopinath C. Mapping Valence Band and Interface Electronic Structure Changes during the Oxidation of Mo to MoO 3 via MoO 2 and MoO 3 Reduction to MoO 2 : A NAPPES Study. J Phys Chem C. 2018;122:23034–44. Cheng C, Wang A, Humayun M, Wang C. Recent advances of oxygen vacancies in MoO 3 : preparation and roles. Chem Eng J. 2024;498:155246. Xue Q, Wang Y-C, Wei X-H. Synaptic plasticity of room-temperature fabricated amorphous MoO x film based memristor. Appl Surf Sci. 2019;479:469–74. Martins RA, Carlos E, Kiazadeh A, Martins R, Deuermeier J. Low-Temperature Solution-Based Molybdenum Oxide Memristors. ACS Appl Eng Mater. 2024;2:298–04. Du H, Chen J, Tu M, Luo S, Li S, Yuan S, Gong T, Huang W, Jie W, Hao J. Transition from nonvolatile bipolar memory switching to bidirectional threshold switching in layered MoO 3 nanobelts. J Mater Chem C. 2019;7:12160. Shan X, Wu Z, Xie Y, Lin X, Zhou B, Zhang Y, Yan X, Ren T, Wang F, Zhang K. Centimetre-scale single crystal α-MoO 3 : oxygen assisted self-standing growth and low-energy-consumption synaptic devices. Nanoscale. 2023;13:1200. Tan Z, Yin X, Guo X. One-dimensional memristive device based on MoO 3 nanobelt. Appl Phys Lett. 2015;2:023503. Patil SR, Mullani NB, Kamble BB, Tayade SN, Kamat RK, Park TJ, Kim D-K, Dongale TD. Forming-free and multilevel resistive switching properties of hydrothermally synthesized hexagonal molybdenum oxide microrods. J Mater Sci : Mater Electron. 2021;32:12490–502. Guo F, Liu Y, Zhang M, Yu W, Li S, Zhang B, Hu B, Li S, Sun A, Jiang J, Hao L. VO 2 /MoO 3 Heterojunctions Artificial Optoelectronic Synapse Devices for Near-Infrared Optical Communication. Small. 2024;20:2310767. Yadav R, Poudyal S, Rajarapu R, Biswal B, Barman PK, Kasiviswanathan S, Novoselov KS, Misra A. Low Power Volatile and Nonvolatile Memristive Devices from 1D MoO 2 -MoS 2 Core-Shell Heterostructures for Future Bio-Inspired Computing. Small. 2024;20:2309163. Kalemai G, Verykios A, Chatzigiannakis G, et al. Highly Robust Double Memristive Device Based on Perovskite/Molybdenum Oxide-Sulfide Compound Heterojunction System. Adv Electron Mater. 2025;11:2400433. Rasool A, Amiruddin R, Mohamed IR, Kumar MCS. Fabrication and Characterization of Resistive Random Access Memory (ReRAM) Devices Using Molybdenum Trioxide (MoO 3 ) as Switching Layer. Superlattices Microstruct. 2020;147:106682. Rajesh VM, Dayal G, Gondhalekar J, Jinesh KB. From Hebbian learning to pattern recognition: The role of oxygen vacancies in the synaptic responses of magnetron sputtered Mo x O y devices. Mate. Sci. Sem. Proc. 2025;188:109194. Wang ZQ, Xu HY, Li XH, Yu H, Liu YC, Zhu XJ. Synaptic Learning and Memory Functions Achieved Using Oxygen Ion Migration/Diffusion in an Amorphous InGaZnO Memristor. Adv Funct Mater. 2012;22:2759. Vasilopoulou Μ, Douvas ΑΜ, Georgiadou DG, Et. The Influence of Hydrogenation and Oxygen Vacancies on Molybdenum Oxides Work Function and Gap States for Application in Organic Optoelectronics. J Am Chem Soc. 2012;134:16178–87. Layegh M, Ghodsi FE, Hadipour H. Experimental and theoretical study of Fe doping as a modifying factor in electrochemical behavior of mixed-phase molybdenum oxide thin films. Appl Phys A. 2020;126:14. Qi Y, Chen W, Mai L, Zhu Q, Jin A. Synthesis and Electrochemical Performance of PEO Doped Molybdenum Trioxide Nanobelts. Int J Electrochem Sci. 2006;1:317–23. Eda KJ. Raman spectra of hydrogen molybdenum bronze, H 0.30 MoO 3 Solid. State Chem. 1992;98:350. Ajito K, Nagahara LA, Tryk DA, Hashimoto K, Fujishima A. J Phys Chem. 1995;99:16383–88. Ohno T, Hasegawa T, Tsuruoka T, Terabe K, Gimzewski JK, Aono M. Short-term plasticity and long-term potentiation mimicked in single inorganic synapses. Nat Mater. 2011;10:591–95. Voglis G, Tavernarakis N. The role of synaptic ion channels in synaptic plasticity. EMBO Rep. 2006;7:1104–10. Yu S. Neuro-Inspired Computing with Emerging Nonvolatile Memorys. Proc. IEEE Inst. Electr. Electron. Eng. 2018;106:260–85. Huo Q, Yang Y, Wang Y, Lei D, et al. A computing-in-memory macro based on three-dimensional resistive random-access memory. Nat Electron. 2022;5(7):469–77. Bi GQ, Poo MMJ. Synaptic Modifications in Cultured Hippocampal Neurons: Dependence on Spike Timing, Synaptic Strength, and Postsynaptic Cell Type. J Neurosci Off J Soc Neurosci. 1998;18(24):10464–72. Additional Declarations No competing interests reported. Supplementary Files SupportingInformation.docx Cite Share Download PDF Status: Published Journal Publication published 04 Dec, 2025 Read the published version in Discover Materials → Version 1 posted Editorial decision: Revision requested 15 Jul, 2025 Reviews received at journal 04 Jul, 2025 Reviewers agreed at journal 03 Jul, 2025 Reviewers agreed at journal 23 Jun, 2025 Reviews received at journal 19 Jun, 2025 Reviewers agreed at journal 12 Jun, 2025 Reviewers invited by journal 11 Jun, 2025 Editor invited by journal 11 Jun, 2025 Editor assigned by journal 09 Jun, 2025 Submission checks completed at journal 09 Jun, 2025 First submitted to journal 05 Jun, 2025 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-6829733","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":470189821,"identity":"f16e7ba0-5cd1-456f-941a-92321cd5b923","order_by":0,"name":"Gion Kalemai","email":"","orcid":"","institution":"National Center for Scientific Research Demokritos","correspondingAuthor":false,"prefix":"","firstName":"Gion","middleName":"","lastName":"Kalemai","suffix":""},{"id":470189822,"identity":"1ab3924f-3fe8-44f0-b0c5-38a9ee220657","order_by":1,"name":"Konstantinos Aidinis","email":"","orcid":"","institution":"Ajman University","correspondingAuthor":false,"prefix":"","firstName":"Konstantinos","middleName":"","lastName":"Aidinis","suffix":""},{"id":470189823,"identity":"b89de4b1-524c-480b-91ec-f1d22cc84647","order_by":2,"name":"Michael-Alexandros Kourtis","email":"","orcid":"","institution":"National Center for Scientific Research Demokritos","correspondingAuthor":false,"prefix":"","firstName":"Michael-Alexandros","middleName":"","lastName":"Kourtis","suffix":""},{"id":470189824,"identity":"e78ec3be-74b9-445e-b8b8-32941dd77647","order_by":3,"name":"Dimitris Davazoglou","email":"","orcid":"","institution":"National Center for Scientific Research Demokritos","correspondingAuthor":false,"prefix":"","firstName":"Dimitris","middleName":"","lastName":"Davazoglou","suffix":""},{"id":470189825,"identity":"7bbb426a-ca77-47d7-890b-eb7272dea993","order_by":4,"name":"Anastasia Soultati","email":"data:image/png;base64,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","orcid":"","institution":"National Center for Scientific Research Demokritos","correspondingAuthor":true,"prefix":"","firstName":"Anastasia","middleName":"","lastName":"Soultati","suffix":""}],"badges":[],"createdAt":"2025-06-05 13:38:36","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6829733/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6829733/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s43939-025-00452-y","type":"published","date":"2025-12-04T15:57:34+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":84817040,"identity":"23cdce26-e20e-4f62-8c0f-bc38f792135e","added_by":"auto","created_at":"2025-06-17 15:47:20","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":407264,"visible":true,"origin":"","legend":"\u003cp\u003eI-V switching curves of the memristor based on a) MoO\u003csub\u003e3\u003c/sub\u003e and b) H-MoO\u003csub\u003e3-x\u003c/sub\u003e for 250 consecutive cycles of sweep voltages. The device structures are presented as insets. Endurance characteristics of the c) MoO\u003csub\u003e3\u003c/sub\u003e and d) H-MoO\u003csub\u003e3-x\u003c/sub\u003e memristors under 250 SET/RESET switching cycles. Cumulative probability plots of HRS and LRS of the e) MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3-x\u003c/sub\u003e memristor. Retention performance of the LRS and HRS of the memristors based on g) MoO\u003csub\u003e3\u003c/sub\u003e and h) H-MoO\u003csub\u003e3-x\u003c/sub\u003e after SET and RESET operation. The pulse SET and RESET voltages was 1.3 V and -3 V, respectively, while the V\u003csub\u003eread\u003c/sub\u003e was 0.1 V.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6829733/v1/714cd78d8ed96e0e412b8f99.png"},{"id":84819881,"identity":"1a32aed7-2396-4daf-99b9-409b1b86f243","added_by":"auto","created_at":"2025-06-17 16:03:21","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":151899,"visible":true,"origin":"","legend":"\u003cp\u003eI-V curve analysis for conductive mechanism of the MoO\u003csub\u003e3\u003c/sub\u003e-based device during a) and b) SET and c) and d) RESET operation.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6829733/v1/18a3e58284b41f4c8b2301f5.png"},{"id":84818821,"identity":"5a8c1f3c-f5d0-473f-a4d9-955d66bd3d86","added_by":"auto","created_at":"2025-06-17 15:55:20","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":109744,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic representation of the RS mechanism of FTO/MoO\u003csub\u003e3\u003c/sub\u003e/Al memristor a) at the initial state, and during b) LRS, c) HRS with V\u003csub\u003eRESET\u003c/sub\u003e of -5 V and d) HRS with V\u003csub\u003eRESET\u003c/sub\u003e of -2 V. e) Multilevel endurance characteristics of the MoO\u003csub\u003e3\u003c/sub\u003e-based memristor under different RESET voltage values.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6829733/v1/400961322c2605ec053d16c6.png"},{"id":84817045,"identity":"f0bdf8a1-9b4c-4bb5-baeb-0a443170a7b0","added_by":"auto","created_at":"2025-06-17 15:47:21","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":128986,"visible":true,"origin":"","legend":"\u003cp\u003ea) Evolution of device current as a function of alternative positive and negative voltage sweeps applied on the FTO/MoO\u003csub\u003e3\u003c/sub\u003e/Al memristor. b) Potentiation and depression in the synaptic memristor when 80 identical positive pulse voltages and 80 negative pulse voltages are applied on the FTO/MoO\u003csub\u003e3\u003c/sub\u003e/Al memristor. c) Response of the MoO\u003csub\u003e3\u003c/sub\u003e-based memristor to mixed potentiating and depressing pulse voltage with various amplitude values. d) STDP characteristics of the FTO/MoO\u003csub\u003e3\u003c/sub\u003e/Al synaptic device.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6829733/v1/62d73ce5d340a38955a76255.png"},{"id":97723839,"identity":"ea96fc43-74f7-415b-90cb-0d6ffe84a137","added_by":"auto","created_at":"2025-12-08 16:08:33","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1347732,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6829733/v1/0b3ec56c-5888-47e6-801e-1880c4951e59.pdf"},{"id":84817046,"identity":"2472d932-7e64-4d92-9b57-2c37116f434f","added_by":"auto","created_at":"2025-06-17 15:47:21","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":630408,"visible":true,"origin":"","legend":"","description":"","filename":"SupportingInformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-6829733/v1/fb273236dd63fd1684bd6026.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Amorphous, fully-stoichiometric molybdenum oxide for high performance nonvolatile resistive switching memory: The role of stoichiometry on synaptic plasticity","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eNon-volatile memristive (NVMs) devices have attracted much attention due to the high-speed and low-energy processing and storage required by the huge amount of data that annually produced\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. The existing silicon-based memory technologies are encountering major obstacles including slow operational speeds, limited storage capacity, and high energy consumption\u003csup\u003e\u003cspan additionalcitationids=\"CR3\" citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. Therefore, there is a strong demand for groundbreaking memory technologies that offer extremely high speed, ultra-long retention, high capacity, and minimal energy use, utilizing innovative principles, materials, and structures\u003csup\u003e\u003cspan additionalcitationids=\"CR6\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. Memristors are a category of NVMs that are recognized for their low energy consumption, high storage capacity and capability to mimic the synaptic plasticity of the brain making them promising candidates for neuromorphic computing\u003csup\u003e\u003cspan additionalcitationids=\"CR8 CR9 CR10\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eA memristor is a two-terminal electronic device based on the resistive switching (RS) effect from the low resistance state (LRS) to the high resistance state (HRS). In a representative structure, the RS layer is sandwiched between two metals, serving as the top and bottom electrodes\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. When a voltage is applied to the device, a conductive filament (CF) is formed (disrupted) within the RS layer leading the HRS-LRS (LRS-HRS) transition. The CF can be vacancy- or metallic-mediated assigned to the migration of oxygen anion species\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e or the electrochemical metallization\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e, respectively. A variety of materials has been investigated as resistive switching layers in memristors including transition metal oxides (TMOs)\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e, 2D materials\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e,\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e, organic semiconductors\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e and inorganic-organic halide perovskites\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eTMOs are a class of materials widely used in memristors, whose RS behavior is determined by the oxygen vacancy distribution within the oxide layer. Among them molybdenum oxide (MoO\u003csub\u003e3\u003c/sub\u003e) is a wide bandgap TMO that has been applied not only in memristors\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e but also in a variety of electronic applications including sensors, catalysis, solar cells, and light emitting diodes\u003csup\u003e\u003cspan additionalcitationids=\"CR25 CR26 CR27 CR28\" citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e. MoO\u003csub\u003e3\u003c/sub\u003e is an exceptional functional layer for memristor fabrication showing also artificial synaptic properties\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e due to its abundant, low-cost, and environmentally friendly nature\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e along with its outstanding electrical and thermal stability, advantageous for memristive devices\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. Another advantage of MoO\u003csub\u003e3\u003c/sub\u003e is the ease of tuning its electrical properties by controlling the stoichiometry depending on the development procedure and method\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e,\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e. The presence of oxygen vacancies significantly influences the conductive mechanism of the memristor. Therefore, the proper amount of oxygen in the resistive switching layer is essential in order to achieve the desired memristor performance.\u003c/p\u003e \u003cp\u003eThere are a few reports over the last ten years regarding the development of amorphous molybdenum oxide functional layer in memristors\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Xue et al demonstrated the RS behavior and synaptic plasticity of a memristor using amorphous MoO\u003csub\u003ex\u003c/sub\u003e prepared at room temperature by magnetron sputtering\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Recently, solution-processed MoO\u003csub\u003e3\u003c/sub\u003e-based memristors reported excellent nonvolatile properties upon annealing at 250 \u003csup\u003eo\u003c/sup\u003eC, which didn\u0026rsquo;t affect the amorphous nature of the metal oxide\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eOn the other hand, most of the reports on molybdenum oxide-based memristors are referred to crystalline MoO\u003csub\u003e3\u003c/sub\u003e nanobelts\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e,\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e, microrods\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e, heterostructures with other metal oxides\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e or transition metal dichalcogenodes\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e,\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e, and preparation methods using high temperatures during deposition\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e. More recently, Rajesh et al.\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e investigated the polycrystalline Mo\u003csub\u003ex\u003c/sub\u003eO\u003csub\u003ey\u003c/sub\u003e structure deposited at different oxygen/argon (O/Ar) concentration using magnetron sputtering altering the amount of oxygen vacancies in the Mo\u003csub\u003ex\u003c/sub\u003eO\u003csub\u003ey\u003c/sub\u003e films. The impact of the oxygen vacancies on the artificial synapses was also demonstrated reporting that the Au/Mo\u003csub\u003ex\u003c/sub\u003eO\u003csub\u003ey\u003c/sub\u003e/FTO artificial synaptic device exhibited excellent neuromorphic characteristics. However, crystalline MoO\u003csub\u003e3\u003c/sub\u003e are less uniform layers compared to amorphous films due to grain boundaries limiting high-density integration\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIn this work, the influence of stoichiometry of amorphous molybdenum oxide on the resistive switching behavior of memristors is investigated. Therefore, vertical two-terminal memristors using amorphous fully-stoichiometric (MoO\u003csub\u003e3\u003c/sub\u003e) and hydrogenated sub-stoichiometric (H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e) as functional layers were fabricated. Both, fully-stoichiometric and hydrogenated sub-stoichiometric molybdenum oxide devices demonstrated favorable resistive switching characteristics. Nevertheless, the fully-stoichiometric memristor showcased superior RS performance, with an endurance of 250 cycles, an impressive retention time of nearly 3\u0026middot;10\u003csup\u003e4\u003c/sup\u003e seconds and an ON/OFF ratio of approximately 10\u003csup\u003e3\u003c/sup\u003e, in respect with the poor RS behavior exhibited by the device employing the H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e film. This remarkable memristive performance can be attributed to the abundance of oxygen vacancies present in the fully-stoichiometric memristor compared to the sub-stoichiometric H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e, which are essential for the formation of the conductive filament in the RS layer of the device. The high level of reproducibility found in MoO\u003csub\u003e3\u003c/sub\u003e-based memristors underscores their potential for real-world applications and scalability. Furthermore, the remarkable capabilities of the MoO\u003csub\u003e3\u003c/sub\u003e memristor showcased through its long-term potentiation (LTP), long-term depression (LTD), and spike-timing dependent plasticity (STDP). This reveals that the fully stoichiometric molybdenum oxide memristor holds immense promise for emulating biological synapses, paving the way for a new era in neuromorphic computing applications.\u003c/p\u003e"},{"header":"2. Experimental Details","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Thin Film Deposition and Device Fabrication\u003c/h2\u003e \u003cp\u003eMemristors were fabricated following the metal-insulator-metal (MIM) structure on fluorine-doped tin oxide (FTO) used as the bottom electrode of the devices. Prior to molybdenum oxide deposition, glass/FTO substrates were cleaned using ultrasonication bath of deionized water, acetone, and isopropyl alcohol for 10 minutes each. Fully-stoichiometric (MoO\u003csub\u003e3\u003c/sub\u003e) and hydrogenated sub-stoichiometric molybdenum oxide (H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e) thin films were then deposited on FTO substrates using a homemade hot-wire chemical vapor deposition system (HW-CVD), as already described elsewhere\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e. In brief, FTO substrates were placed on an aluminum susceptor in a stainless-steel reactor. 2.5 cm upwards of the FTO substrates a molybdenum (Mo) wire with diameter of 0.5 mm was placed between two copper leads. The deposition began with the reactor\u0026rsquo;s evacuation down to 10 mTorr using a mechanical pump. For the development of fully-stoichiometric molybdenum oxide high purity oxygen (O\u003csub\u003e2\u003c/sub\u003e) gas 99.999% was used, while the deposition of hydrogenated sub-stoichiometric molybdenum oxide was taken place using forming gas; a gas mixture of 90% nitrogen and 10% hydrogen. MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e films were deposited by setting the base pressure at 80 mTorr and simultaneously heating the Mo wire at 560\u003csup\u003eo\u003c/sup\u003eC. The thickness of the prepared films was 100 nm. Finally, the memristors were completed with the deposition through thermal evaporation of a 100 nm-thick aluminum (Al) layer serving as the top electrode.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Thin Film and Device Characterization\u003c/h2\u003e \u003cp\u003eThe molybdenum oxide films optical properties were obtained using a Perkin Elmer Lambda 40 UV/Vis spectrometer recording the transmittance and absorbance spectra in the wavelength range of 200\u0026ndash;900 nm. A Bruker Tensor 27 Fourier transform infrared (FTIR) spectrometer equipped with a DTGS detector was used to record FTIR spectra of the deposited films using transmittance mode. The X-ray diffractograms (XRD) of the MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e films was taken using a Smart Lab Rigaku diffractometer (θ/θ scan) with a CuKA radiation (3 kW). The surface morphology of the samples was investigated by scanning electron microscopy (SEM) using a JEOL 7401f FESEM microscope. The electrical characterization of the fabricated memristors was performed using a VersaSTAT4 potentiostat. The bias was applied to the top electrode (Al) maintaining the FTO-bottom electrode connected to the ground.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results and Discussion","content":"\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Thin Film Characterization\u003c/h2\u003e \u003cp\u003e100 nm-thick fully-stoichiometric (MoO\u003csub\u003e3\u003c/sub\u003e) and hydrogenated sub-stoichiometric (H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e) films were prepared using a hot-wire chemical vapor deposition system, as described in details in the Experimental section. It is demonstrated that both films are amorphous, as revealed from the XRD measurements shown in Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003ea (Supporting Information), where no XRD peaks appeared in the diffractograms of MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e. Furthermore, the surface nanomorphology of the deposited molybdenum oxide films is composed of grain-like structure, as it can be seen in SEM images, shown in Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eb and c for the MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e film, respectively. The surface of the fully-stoichiometric molybdenum oxide consists of large grains with dimensions of ~\u0026thinsp;40 nm (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eb). On the other hand, the H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e film is composed by smaller grains compared to the MoO\u003csub\u003e3\u003c/sub\u003e sample with dimensions of ~\u0026thinsp;25 nm (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003ec). Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed shows the photoluminescence (PL) emission spectra of MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e films at an excitation wavelength of 325 nm. In the case of H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e, the first weak peak is found at 335 nm, which is located close to the near band edge (NBE) emission and can be attributed to the free exciton recombination. In addition, the broad peak at 363 nm in the PL spectrum of H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e can be assigned to the Mo\u003csup\u003e5+\u003c/sup\u003e d\u003csub\u003exy\u003c/sub\u003e \u0026ndash; d\u003csub\u003eyz\u003c/sub\u003e band transition, while the weak peak at 469 nm is mainly due to the deep-level transitions caused by the existence of defects and oxygen vacancies. In the case of MoO\u003csub\u003e3\u003c/sub\u003e sample, the peak at 469 nm is more pronounced indicating the excess of oxygen vacancies compared with the H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e. These oxygen vacancies may be beneficial on the resistive switching effect.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe stoichiometry and electronic properties were estimated using X-ray (XPS) and ultraviolet (UPS) photoelectron spectroscopy measurements presented in previous work\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e demonstrating that the molybdenum oxide film deposited in O\u003csub\u003e2\u003c/sub\u003e environment was fully-stoichiometric, while forming gas resulted in the hydrogenation of the sub-stoichiometric molybdenum oxide. Moreover, changes in the UPS spectra of the MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e were observed, where the work function (W\u003csub\u003eF\u003c/sub\u003e) is significantly reduced upon metal oxide hydrogenation. Occupied states inside the band gap of the H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e were also appeared. These gap states resulted in reduced energy gap of around 2.4 eV compared with that of MoO\u003csub\u003e3\u003c/sub\u003e (3 eV), as evidenced from the Tauc plot presented in Figure S2a in the Supporting Information. Furthermore, the hydrogenated sub-stoichiometric molybdenum oxide exhibited reduced transmittance in the visible region, as shown in Figure S2b (increased absorbance, Figure S2c) in respect to the MoO\u003csub\u003e3\u003c/sub\u003e film.\u003c/p\u003e \u003cp\u003eIn order to further investigate the structure of the prepared molybdenum oxide samples, FTIR transmittance measurements were performed on MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e films. The FTIR transmittance spectra of the MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e are shown in Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003ee and S1f, respectively. For both samples, the fingerprint region located in the wavenumber range of 400\u0026ndash;1100 cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e contains three transmittance bands indicating different types of Mo \u0026ndash; O bonding in MoO\u003csub\u003e3\u003c/sub\u003e lattice structure. Particularly, the bands at 635 cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e and 561 cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e are assigned to Mo \u0026ndash; O bending vibration mode of the symmetric bridging oxygen bonds to three Mo atoms (O \u0026ndash; Mo3). The bands observed in the region 880\u0026ndash;650 cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e are attributed to the stretching vibration of the asymmetric bridging oxygen connected to two neighboring Mo atoms (O \u0026ndash; Mo2), while the peaks at 951 cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e and 905 cm\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e are assigned to the stretching mode of the terminal oxygen (Mo\u0026thinsp;=\u0026thinsp;O bond)\u003csup\u003e\u003cspan additionalcitationids=\"CR49 CR50\" citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u003c/sup\u003e. Upon hydrogenation, a shift of Mo\u0026thinsp;=\u0026thinsp;O bond to higher frequency values was observed. Moreover, the O \u0026ndash; Mo2 band was shifted to lower frequency values suggesting the incorporation of hydrogen atoms in the molybdenum oxide structure.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e3.2 MoO\u003csub\u003e3\u003c/sub\u003e- and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e-based memristor\u0026rsquo;s characterization\u003c/h2\u003e \u003cp\u003eIn order to investigate the influence of the stoichiometry of the amorphous molybdenum oxide films on the memristive behavior, memristors based on the fully-stoichiometric and hydrogenated sub-stoichiometric films were fabricated with the device structure FTO/MoO\u003csub\u003e3\u003c/sub\u003e or H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e/Al, shown as inset in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb, respectively. The FTO (fluorine-doped tin oxide) serving as the bottom electrode was grounded, while the top aluminum (Al) electrode was biased. Current \u0026ndash; voltage (I \u0026ndash; V) characteristic curves were recorded by performing cycle sweeps of voltage (\u0026ndash; 6V \u0026rarr; 0V \u0026rarr; + 3V \u0026rarr; 0V \u0026rarr; \u0026ndash; 6V). A typical bipolar resistive switching (RS) memory behavior is observed for both memristors as revealed from recorded I \u0026ndash; V curves presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb for the devices based on amorphous MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e films. However, in the case of the fully-stoichiometric-based device, the memristive behavior is more pronounced. Initially the devices are in the high resistance state (HRS). When a forward voltage was applied between the bottom and the top electrode, the transition from the HRS to the lower resistance state (LRS) occurred resulting in the ON switching, also referred as SET. More importantly, the transition from HRS to LRS in MoO\u003csub\u003e3\u003c/sub\u003e-based device is more abrupt (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea) than in the case of the memristor with H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb), where the current gradually increases (decreases) in the SET (RESET) operation. Next, the devices returned to the HRS (RESET) by applying a bias of opposite polarity.\u003c/p\u003e \u003cp\u003eTo examine the repeatability and non-volatility of the devices, several repeated cycles of sweep voltage from \u0026minus;\u0026thinsp;6 V to +\u0026thinsp;3 V were employed to the devices. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb presents the semi-logarithmic I \u0026ndash; V characteristic curves for consecutive 250 cycles. It is observed that both memristors exhibited good repeatability maintaining stable resistive switching behavior even after 250 cycles. Figure S3a and S3b represents the distribution of RESET and SET voltage, respectively, of 250 cycles for the MoO\u003csub\u003e3\u003c/sub\u003e-based device exhibiting the best memristive behavior, where the estimated standard deviation values (σ) are 0.40 V and 0.33 V, respectively. The MoO\u003csub\u003e3\u003c/sub\u003e-based memristor demonstrates also low threshold voltages with mean value of 0.96 V and \u0026minus;\u0026thinsp;2.01 V. Furthermore, both memristors maintain their resistance states steady as revealed from the endurance plots, shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed for the devices with the MoO\u003csub\u003e3\u003c/sub\u003e and H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e, respectively, indicating the reproducibility of the devices upon cycle repeat. From the estimated cumulative probability shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ef, the MoO\u003csub\u003e3\u003c/sub\u003e device exhibits high HRS/LRS ratio of 10\u003csup\u003e3\u003c/sup\u003e, while memory window of the H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e memristor is very small (HRS/LRS\u0026thinsp;=\u0026thinsp;2.5). Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eg and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eh represents the retention performance of LRS and HRS of the fully-stoichiometric and hydrogenated sub-stoichiometric molybdenum oxide based memristive devices, respectively, after applying a pulse with amplitude of +\u0026thinsp;1.5 V for SET, -4 V for RESET, and setting the reading voltage V\u003csub\u003eread\u003c/sub\u003e at 0.1 V. Reliable non-volatility is clearly seen in both cases, where no significant change observed after 2.8\u0026middot;10\u003csup\u003e4\u003c/sup\u003e sec.\u003c/p\u003e \u003cp\u003eTo set more light into the underlying conduction mechanism of our best performed MoO\u003csub\u003e3\u003c/sub\u003e memristor, the log(I) \u0026ndash; log(V) plots of the LRS and HRS was fitted and presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. It is observed that I-V characteristic curve of the HRS shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb consists of three distinguished regions; i) the ohmic region, where the current is linearly related to the applied voltage (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(I \\propto V\\)\u003c/span\u003e\u003c/span\u003e) originated by the thermally generated free carriers, ii) Schottky emission (SE), assigned to the oxygen vacancies at the MoO\u003csub\u003e3\u003c/sub\u003e/Al interface, and iii) the space charge limited current (SCLC), where the current is followed by a quadratic term (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(I \\propto {V^2}\\)\u003c/span\u003e\u003c/span\u003e) attributed to the traps that are filled by charges. Specifically, at low voltages of HRS, the slope estimated by fitting the experimental data is ~\u0026thinsp;1 suggesting ohmic conduction behavior (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea). When the applied voltage increases, the Schottky model well fits the I-V characteristic curve (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). The charges localized at the interface between the MoO\u003csub\u003e3\u003c/sub\u003e and the bottom electrode (FTO), are trapped by the HRS layer\u0026rsquo;s empty trap sites of MoO\u003csub\u003e3\u003c/sub\u003e. As the applied voltage further increases, the empty trap sites are gradually occupied until fully completed by electrons. The slope of fitting model significantly increases suggesting the dominance of the SCLC conduction mechanism. On the other hand, the LRS region is well fitted by ohmic and SE models, as presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed. Ohmic conduction mechanism (slope equal to 1) is dominant at low voltages, while at higher voltages the slope of fitting curve is 3.6 indicating that the conduction mechanism obeys the SE model, which is attributed to the barrier at the FTO/MoO\u003csub\u003e3\u003c/sub\u003e interface. Same results also obtained in the case of H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e, as shown in log(I) \u0026ndash; log(V) and Ln(I) \u0026ndash; V\u003csup\u003e1/2\u003c/sup\u003e plots of the LRS and HRS presented in Figure S4a-d. However, as the W\u003csub\u003eF\u003c/sub\u003e of the fully-stoichiometric molybdenum oxide is higher than that of hydrogenated sub-stoichiometric the Schottky barrier height increases resulting in an increase in the HRS, which therefore leads to much larger ON/OFF ratio.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e illustrates the switching mechanism in the MoO\u003csub\u003e3\u003c/sub\u003e-based memristor. In the pristine state (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea) without the application of electrical bias, the oxygen vacancies are randomly distributed in the MoO\u003csub\u003e3\u003c/sub\u003e layer, and thus the device is on HRS. When a positive bias is applied to the FTO electrode, partial reduction of the molybdenum oxide occurs and oxygen vacancies accumulate at the bottom electrode. The migration of the oxygen vacancies forms a conductive filament (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb), which facilitates the electron transport between the two electrodes through the MoO\u003csub\u003e3\u003c/sub\u003e film. The current abruptly jumps and then gradually increases resulting in the HRS to LRS transition. The sudden change in resistive switching is more noticeable in the MoO\u003csub\u003e3\u003c/sub\u003e functional layer than in H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e, which is attributed to the higher number of oxygen vacancies as indicated by the PL measurements. In the case of H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e the lesser oxygen vacancies result in a gradually HRS to LRS transition. On the other hand, when the polarity of the applied voltage is changed, the oxygen vacancies are suppressed and the conductive filament is disrupted (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). In this case, a Schottky barrier gap is formed between the tip of the conductive filament and the top electrode. The MoO\u003csub\u003e3\u003c/sub\u003e-based memristor showed also a multilevel resistive switching due to the oxygen-rich conductive filament formed during the LRS and Schottky gaps during the disruption of the conductive filament. By applying different RESET voltages of -5V and \u0026minus;\u0026thinsp;2V, we can modulate the Schottky barrier (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed) getting different HRS, as presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee. It is observed that increasing the RESET voltage, the Schottky gap also increases.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e3.3 MoO\u003csub\u003e3\u003c/sub\u003e memristor\u0026rsquo;s synaptic behavior\u003c/h2\u003e \u003cp\u003eIn order to investigate the long-term stability of the MoO\u003csub\u003e3\u003c/sub\u003e-based memristor, two sets of scanning voltages, one positive (+\u0026thinsp;1V) and one negative (-1V), were applied to the device, and the output current was recorded. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea shows the evolution of the MoO\u003csub\u003e3\u003c/sub\u003e memristor as a function of the alternative positive and negative voltage sweeps applied to the device. The positive and negative voltage sweeps were applied on the device 10 times, sequentially, totally 100 times. In particular, the first positive voltage sweep was conducted applying the positive voltage for 10 times, which results in an increase of the current, and thus the conductance. Next, the negative voltage sweep was followed for 10 times. It is observed that the absolute value of current gradually decreases in the negative direction, leading to the increase of the conductance. Therefore, the synaptic weight could be enhanced (suppressed) upon the application of serial excitatory (inhibitory) spikes, which is analogous to the biological synapses.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe synaptic plasticity, named as synaptic potentiation, which is the ability to modulate and retain the synaptic weight representing the level of learning and memory divided into short-term (STP) and long-term (LTP) potentiation\u003csup\u003e\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e is also studied. It is noted that the short-term (STD) and long-term (LTD) depression represent the depression of synaptic weight\u003csup\u003e\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e. To simulate the LTP and LTD of the MoO\u003csub\u003e3\u003c/sub\u003e-based device, 80 positive and 80 negative pulses were sequentially applied. When 80 continuous positive programming pulses are applied the current, and thus the conductance, of the device increases, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb. After the application of 80 continuous negative programming pulses, the current of the MoO\u003csub\u003e3\u003c/sub\u003e memristor decreases. It is suggested that the post-synaptic current could be bidirectional controlled by the sequentially application of positive and negative pulses, which is essential for application in neuromorphic computing, neural networks, and biophysics\u003csup\u003e\u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eBy adjusting the input voltage, the conductance of the device can be also regulated. Figure\u0026nbsp;54c shows the MoO\u003csub\u003e3\u003c/sub\u003e memristor response to mixed potentiating and depressing pulse voltage with different amplitude and same pulse width (500 \u0026micro;s). The conductance change, ΔG, was estimated using the Eq.\u0026nbsp;(\u003cspan refid=\"Equ1\" class=\"InternalRef\"\u003e1\u003c/span\u003e):\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e\n$$\\Delta G=\\frac{{{G_1} - {G_2}}}{{{G_1}}}100\\%$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWhere G\u003csub\u003e1\u003c/sub\u003e is the conductance recorded before each pulse voltage, and G\u003csub\u003e2\u003c/sub\u003e is the conductance after the pulse. It is observed that, the conductance changes of the device are grater (smaller) upon the application of larger (smaller) positive pulse. The same phenomenon is also observed in the application of negative pulse, suggesting that the device can imitate the biological synapses\u003csup\u003e\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed shows the spike timing dependent plasticity (STDP) characteristics of the FTO/MoO\u003csub\u003e3\u003c/sub\u003e/Al synaptic device. STDP can regulate the synaptic weight through the timing between the pre-synaptic and post-synaptic pulse\u003csup\u003e\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e\u003c/sup\u003e. Δt (Δt\u0026thinsp;=\u0026thinsp;t\u003csub\u003epost\u003c/sub\u003e \u0026ndash; t\u003csub\u003epre\u003c/sub\u003e) represents the time between the application of pre-synaptic (with amplitude of -0.4 V and width of 500 \u0026micro;s) and post-synaptic (with amplitude of +\u0026thinsp;0.4 V and width of 500 \u0026micro;s) neuron pulse. t\u003csub\u003epost\u003c/sub\u003e and t\u003csub\u003epre\u003c/sub\u003e are the time for applying post-synaptic and pre-synaptic pulse, respectively. In the case of Δt\u0026thinsp;\u0026lt;\u0026thinsp;0, meaning that the pre-synaptic pulse comes after the post-synaptic pulse, the synaptic weight gradually decreases. When the post-synaptic pulse comes after the application of pre-synaptic pulse (Δt\u0026thinsp;\u0026gt;\u0026thinsp;0), the synaptic weight gradually increases. A strong functional relationship between the Δt and the change of the synaptic weight is also observed, where the change of the synaptic weight versus the Δt can be fitted with an exponential decay function. The LTP (for Δt\u0026thinsp;\u0026gt;\u0026thinsp;0) and LTD (for Δt\u0026thinsp;\u0026lt;\u0026thinsp;0) are well emulated, demonstrating that the FTO/MoO\u003csub\u003e3\u003c/sub\u003e/Al memristor is analogous to biological systems.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. Conclusions","content":"\u003cp\u003eIn summary, the influence of the stoichiometry of molybdenum oxide films on the memristive behavior was investigated. Therefore, memristors based on the fully-stoichiometric (MoO\u003csub\u003e3\u003c/sub\u003e) and hydrogenated sub-stoichiometric (H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e) were fabricated. The devices exhibited excellent memristive behavior including good endurance and retention. In addition, the MoO\u003csub\u003e3\u003c/sub\u003e device exhibited higher ON/OFF ratio of ~\u0026thinsp;10\u003csup\u003e3\u003c/sup\u003e compared with that of the H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e based device (~\u0026thinsp;2.5) attributed to the formation and destruction of conductive filament due to the excess of oxygen vacancies in the MoO\u003csub\u003e3\u003c/sub\u003e film. Furthermore, the excellent long-term potentiation (LTP), long-term depression (LTD), and spike timing dependent plasticity (STDP) of the MoO\u003csub\u003e3\u003c/sub\u003e memristor was demonstrated. Therefore, it can be concluded that the fully-stoichiometric molybdenum oxide memristor shows a great potential on mimicking biological synapses for neuromorphic computing applications.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors confirm that the data supporting the findings of this study are available within the article. If someone wants to request the data from this study, the corresponding author (Anastasia Soultati) should be contacted.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNo funding.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorresponding author\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCorrespondence to Anastasia Soultati, [email protected]\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics declarations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\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"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eHwang CS. Prospective of semiconductor memory devices: from memory system to materials. Adv Electron Mater. 2015;1:1400056.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBertolazzi S, Bondavalli P, Roche S, et al. Nonvolatile memories based on graphene and related 2D materials. Adv Mater. 2019;31:1806663.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi J, Zhang H, Ding Y, Li J, Wang S, Zhang DW, Zhou P. A non-volatile and gate based on Al\u003csub\u003e2\u003c/sub\u003eO\u003csub\u003e3\u003c/sub\u003e/HfO\u003csub\u003e2\u003c/sub\u003e/Al\u003csub\u003e2\u003c/sub\u003eO\u003csub\u003e3\u003c/sub\u003e charge-trap stack for in-situ storage applications. Sci Bull. 2019;64:1518\u0026ndash;24.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDing Y, Liu L, Li J, Cao R, Jiang Y-G, Liu C, Liu Q, Zhou P. A semi-floating memory with 535% enhancement of refresh time by local field modulation. Adv Funct Mater. 2020;30:1908089.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVu Q, Shin Y, Kim Y, et al. Two-terminal floating-gate memory with van der Waals heterostructures for ultrahigh on/off ratio. Nat Commun. 2016;7:12725.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi Y, Zhang ZC, Li J, et al. Low-voltage ultrafast nonvolatile memory via direct charge injection through a threshold resistive-switching layer. Nat Commun. 2022;13:4591.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang R, Song Z, Song W, Xin T, Lv S, Song S, Liu J. Phase-change memory based on matched Ge-Te, Sb-Te, and In-Te octahedrons: improved electrical performances and robust thermal stability. InfoMat. 2021;3:1008\u0026ndash;15.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun K, Chen J, Yan X. The future of Memristors: Materials Engineering and Neural Networks. Adv Funct Mater. 2021;31:2006773.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRajasekaran S, Simanjuntak FM, Chandrasekaran S, Panda D, Saleem A, Tseng TY. Flexible Ta\u003csub\u003e2\u003c/sub\u003eO\u003csub\u003e5\u003c/sub\u003e/WO\u003csub\u003e3\u003c/sub\u003e-Based Memristor Synapse for Wearable and Neuromorphic Applications. IEEE Electron Device Lett. 2022;43(1):9\u0026ndash;12.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYan X, Zhao Q, Chen AP, et al. Vacancy-Induced Synaptic Behavior in 2D WS\u003csub\u003e2\u003c/sub\u003e Nanosheet-based Memristor for Low-Power Neuromorphic Computing. Small. 2019;15:1901423.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJu JH, Seo S, Baek S, Lee D, et al. Two-Dimensional MXene Synapse for Brain-Inspired Neuromorphic Computing. Small. 2021;17:2102595.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhan R, Ilyas N, Shamim MZM, et al. Oxide-based resistive switching-based devices: fabrication, influence parameters and applications. J Mater Chem C. 2021;9:15755\u0026ndash;88.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYildirim H, Pachter R. Mechanistic Analysis of Oxygen Vacancy-Driven Conductive Filament Formation in Resistive Random Access Memory Metal/NiO/Metal Structures. ACS Appl Mater Interfaces. 2018;10(11):9802\u0026ndash;16.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim M-S, Park E, Kim S-G, et al. Highly Reliable Electrochemical Metallization Threshold Switch Through Conductive Filament Engineering Using Two-Dimensional PtSe\u003csub\u003e2\u003c/sub\u003e Insertion Layer. Adv Mater Interfaces. 2023;10(7):2202296.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCarlos E, Branquinho R, Martins R, Kiazadeh A, Fortunato E. Recent Progress in Solution-Based Metal Oxide Resistive Switching Devices. Adv Mater. 2021;33:2004328.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCho D-Y, Luebben M, Wiefels S, Lee K-S, Valov I. Interfacial Metal-Oxide Interactions in Resistive Switching Memories. ACS Appl Mater Interfaces. 2017;9(22):19287\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLee D, Kim S-M, Park J-C, Jung Y, Lee S, Lee BH, Lee S. Enhancing reliability in oxide-based memristors using two-dimensional transition metal dichalcogenides. Appl Surf Sci. 2025;679:161216.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou Z, Yang F, Wang S, et al. Emerging of two-dimensional materials in novel memristor. Front Phys. 2022;17:23204.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCasula G, Busby Y, Franquet A, Spampinato V, Houssiau L, Bonfiglio A, Cosseddu P. A flexible organic memory device with clearly disclosed resistive switching mechanism. Org Electron. 2019;64:209\u0026ndash;15.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhou Z, Mao H, Wang X, et al. Transient and Flexible Polymer Memristors Utilizing Full-Solution Processed Polymer Nanocomposites. Nanoscale. 2018;10(31):14824\u0026ndash;29.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJohn RA, Demirağ Y, Shynkarenko Y, et al. Reconfigurable halide perovskite nanocrystal memristors for neuromorphic computing. Nat Commun. 2022;13:2074.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi B, Xia F, Du B, Zhang S, Xu L, Su Q, Zhang D, Yang J. 2D Halide Perovskite for High-Performance Resistive Switching Memory and Artificial Synapse Applications. Adv Sci. 2024;11:2310263.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eArita M, Kaji H, Fujii T, Takahashi Y. Resistance Switching Properties of Molybdenum Oxide Films. Thin Solid Films. 2012;520(14):4762\u0026ndash;67.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMedeiros SASL, Oliviera ALM, Duarte TM. a-MoO\u003csub\u003e3\u003c/sub\u003e Micro- and Nanoparticles as Catalysts for Biofuel Production. ACS Appl Nano Mater. 2025;8(9):4339\u0026ndash;53.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSanchez-Fernandez LS, Guzm\u0026aacute;n Vargas A, de La Amador LO. Manufacture and study of a chemiresistive gas sensor using MoO\u003csub\u003e3\u003c/sub\u003e as a sensitive film. MRS Adv. 2025;10:592\u0026ndash;97.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGong Y, Dong Y, Zhao B, Yu R, Hu S, Tan Z. Diverse applications of MoO\u003csub\u003e3\u003c/sub\u003e for high performance organic photovoltaics: fundamental, processes and optimization strategies. J Mater Chem A. 2020;8:978\u0026ndash;09.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHong J, Xu Z, Hu T, Lee S, Gui M, Kahn A, Rand BP. Evaporated organic-MoO\u003csub\u003e3\u003c/sub\u003e composite hole transport layers towards stable perovskite solar cells. EES Sol. 2025;1:107\u0026ndash;14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChoi DK, Kim DH, Lee CM, et al. Highly efficient, heat dissipating, stretchable organic light-emitting diodes based on a MoO\u003csub\u003e3\u003c/sub\u003e/Au/MoO\u003csub\u003e3\u003c/sub\u003e electrode with encapsulation. Nat Commun. 2021;12:2864.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNie L, Fan J, Li Y, Xiang C, Zhang T. Direct Optical Patterning of MoO\u003csub\u003e3\u003c/sub\u003e Nanoparticles and Their Application as a Hole Injection Layer for Solution-Processed Quantum Dot Light-Emitting Diodes. ACS Appl Nano Matter. 2024;7(8):9499\u0026ndash;06.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSunny MM, Thamankar R. Spike rate dependent synaptic characteristics in lamellar multilayered alpha-MoO\u003csub\u003e3\u003c/sub\u003e based two-terminal devices \u0026ndash; efficient way to control the synaptic amplification. RSC Adv. 2024;14:2518\u0026ndash;28.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAlizadeh S, Hassanzadeh-Tabrizi SA. MoO\u003csub\u003e3\u003c/sub\u003e fibers and belts: Molten salt synthesis, characterization and optical properties. Ceram Int. 2015;41:10839\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBalendhran S, Walia S, Nili H, Ou J, Zhuiykov S, Kaner R, Sriram S, Bhaskaran M. Kalantar-zadeh K. Two-dimensional molybdenum trioxide and dichalcogenides. Adv Funct Mater. 2013;23:3952\u0026ndash;70.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eReddy K, Mhamane N, Ghosalya M, Gopinath C. Mapping Valence Band and Interface Electronic Structure Changes during the Oxidation of Mo to MoO\u003csub\u003e3\u003c/sub\u003e via MoO\u003csub\u003e2\u003c/sub\u003e and MoO\u003csub\u003e3\u003c/sub\u003e Reduction to MoO\u003csub\u003e2\u003c/sub\u003e: A NAPPES Study. J Phys Chem C. 2018;122:23034\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheng C, Wang A, Humayun M, Wang C. Recent advances of oxygen vacancies in MoO\u003csub\u003e3\u003c/sub\u003e: preparation and roles. Chem Eng J. 2024;498:155246.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eXue Q, Wang Y-C, Wei X-H. Synaptic plasticity of room-temperature fabricated amorphous MoO\u003csub\u003ex\u003c/sub\u003e film based memristor. Appl Surf Sci. 2019;479:469\u0026ndash;74.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMartins RA, Carlos E, Kiazadeh A, Martins R, Deuermeier J. Low-Temperature Solution-Based Molybdenum Oxide Memristors. ACS Appl Eng Mater. 2024;2:298\u0026ndash;04.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDu H, Chen J, Tu M, Luo S, Li S, Yuan S, Gong T, Huang W, Jie W, Hao J. Transition from nonvolatile bipolar memory switching to bidirectional threshold switching in layered MoO\u003csub\u003e3\u003c/sub\u003e nanobelts. J Mater Chem C. 2019;7:12160.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShan X, Wu Z, Xie Y, Lin X, Zhou B, Zhang Y, Yan X, Ren T, Wang F, Zhang K. Centimetre-scale single crystal α-MoO\u003csub\u003e3\u003c/sub\u003e: oxygen assisted self-standing growth and low-energy-consumption synaptic devices. Nanoscale. 2023;13:1200.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTan Z, Yin X, Guo X. One-dimensional memristive device based on MoO\u003csub\u003e3\u003c/sub\u003e nanobelt. Appl Phys Lett. 2015;2:023503.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePatil SR, Mullani NB, Kamble BB, Tayade SN, Kamat RK, Park TJ, Kim D-K, Dongale TD. Forming-free and multilevel resistive switching properties of hydrothermally synthesized hexagonal molybdenum oxide microrods. J Mater Sci : Mater Electron. 2021;32:12490\u0026ndash;502.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo F, Liu Y, Zhang M, Yu W, Li S, Zhang B, Hu B, Li S, Sun A, Jiang J, Hao L. VO\u003csub\u003e2\u003c/sub\u003e/MoO\u003csub\u003e3\u003c/sub\u003e Heterojunctions Artificial Optoelectronic Synapse Devices for Near-Infrared Optical Communication. Small. 2024;20:2310767.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYadav R, Poudyal S, Rajarapu R, Biswal B, Barman PK, Kasiviswanathan S, Novoselov KS, Misra A. Low Power Volatile and Nonvolatile Memristive Devices from 1D MoO\u003csub\u003e2\u003c/sub\u003e-MoS\u003csub\u003e2\u003c/sub\u003e Core-Shell Heterostructures for Future Bio-Inspired Computing. Small. 2024;20:2309163.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKalemai G, Verykios A, Chatzigiannakis G, et al. Highly Robust Double Memristive Device Based on Perovskite/Molybdenum Oxide-Sulfide Compound Heterojunction System. Adv Electron Mater. 2025;11:2400433.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRasool A, Amiruddin R, Mohamed IR, Kumar MCS. Fabrication and Characterization of Resistive Random Access Memory (ReRAM) Devices Using Molybdenum Trioxide (MoO\u003csub\u003e3\u003c/sub\u003e) as Switching Layer. Superlattices Microstruct. 2020;147:106682.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRajesh VM, Dayal G, Gondhalekar J, Jinesh KB. From Hebbian learning to pattern recognition: The role of oxygen vacancies in the synaptic responses of magnetron sputtered Mo\u003csub\u003ex\u003c/sub\u003eO\u003csub\u003ey\u003c/sub\u003e devices. Mate. Sci. Sem. Proc. 2025;188:109194.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang ZQ, Xu HY, Li XH, Yu H, Liu YC, Zhu XJ. Synaptic Learning and Memory Functions Achieved Using Oxygen Ion Migration/Diffusion in an Amorphous InGaZnO Memristor. Adv Funct Mater. 2012;22:2759.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVasilopoulou Μ, Douvas ΑΜ, Georgiadou DG, Et. The Influence of Hydrogenation and Oxygen Vacancies on Molybdenum Oxides Work Function and Gap States for Application in Organic Optoelectronics. J Am Chem Soc. 2012;134:16178\u0026ndash;87.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLayegh M, Ghodsi FE, Hadipour H. Experimental and theoretical study of Fe doping as a modifying factor in electrochemical behavior of mixed-phase molybdenum oxide thin films. Appl Phys A. 2020;126:14.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQi Y, Chen W, Mai L, Zhu Q, Jin A. Synthesis and Electrochemical Performance of PEO Doped Molybdenum Trioxide Nanobelts. Int J Electrochem Sci. 2006;1:317\u0026ndash;23.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEda KJ. Raman spectra of hydrogen molybdenum bronze, H\u003csub\u003e0.30\u003c/sub\u003eMoO\u003csub\u003e3\u003c/sub\u003e Solid. State Chem. 1992;98:350.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAjito K, Nagahara LA, Tryk DA, Hashimoto K, Fujishima A. J Phys Chem. 1995;99:16383\u0026ndash;88.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOhno T, Hasegawa T, Tsuruoka T, Terabe K, Gimzewski JK, Aono M. Short-term plasticity and long-term potentiation mimicked in single inorganic synapses. Nat Mater. 2011;10:591\u0026ndash;95.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVoglis G, Tavernarakis N. The role of synaptic ion channels in synaptic plasticity. EMBO Rep. 2006;7:1104\u0026ndash;10.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYu S. Neuro-Inspired Computing with Emerging Nonvolatile Memorys. Proc. IEEE Inst. Electr. Electron. Eng. 2018;106:260\u0026ndash;85.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuo Q, Yang Y, Wang Y, Lei D, et al. A computing-in-memory macro based on three-dimensional resistive random-access memory. Nat Electron. 2022;5(7):469\u0026ndash;77.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBi GQ, Poo MMJ. Synaptic Modifications in Cultured Hippocampal Neurons: Dependence on Spike Timing, Synaptic Strength, and Postsynaptic Cell Type. J Neurosci Off J Soc Neurosci. 1998;18(24):10464\u0026ndash;72.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"discover-materials","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"dime","sideBox":"Learn more about [Discover Materials](https://www.springer.com/journal/43939)","snPcode":"","submissionUrl":"","title":"Discover Materials","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6829733/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6829733/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTransition metal oxides (TMOs) are a promising class of materials for neuromorphic computing and processing systems demonstrating a variety of resistive switching (RS) mechanisms. However, little is known about the correlation between its stoichiometry and RS. This study is focused on the development and characterization of amorphous molybdenum oxide memristors with different stoichiometry. Fully-stoichiometric (MoO\u003csub\u003e3\u003c/sub\u003e) and hydrogenated sub-stoichiometric (H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e) amorphous molybdenum oxide thin films were developed via a hot-wire chemical vapor deposition system. Both, stoichiometric and hydrogenated sub-stoichiometric molybdenum oxide devices showed good resistive switching behavior. However, the fully-stoichiometric memristor exhibited better RS properties with endurance of 250 cycles, ON/OFF ratio\u0026thinsp;~\u0026thinsp;10\u003csup\u003e3\u003c/sup\u003e and high retention of almost 3\u0026middot;10\u003csup\u003e4\u003c/sup\u003e s, compared with the poor RS behavior of the device based on the H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e film. This impressive memristive behavior could be attributed to the excess of oxygen atoms in the case of fully-stoichiometric memristor in respect to the sub-stoichiometric H-MoO\u003csub\u003e3\u0026thinsp;\u0026minus;\u0026thinsp;x\u003c/sub\u003e which play crucial role in the conductive behavior of the device. The high reproducibility observed in MoO\u003csub\u003e3\u003c/sub\u003e-based memristor highlights their potential for practical applications and scalability. Additionally, the outstanding features of the MoO\u003csub\u003e3\u003c/sub\u003e memristor demonstrated through its long-term potentiation (LTP), long-term depression (LTD), and spike-timing dependent plasticity (STDP) indicate that the fully stoichiometric molybdenum oxide memristor has significant potential for simulating biological synapses, opening doors to a new era in neuromorphic computing applications.\u003c/p\u003e","manuscriptTitle":"Amorphous, fully-stoichiometric molybdenum oxide for high performance nonvolatile resistive switching memory: The role of stoichiometry on synaptic plasticity","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-17 15:47:16","doi":"10.21203/rs.3.rs-6829733/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-07-15T09:22:21+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-07-04T04:16:51+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"166594366172510234383480297593650254361","date":"2025-07-03T08:35:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"245840560454764620561847183223122334036","date":"2025-06-23T05:43:57+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-06-19T13:26:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"176117346339775503350774204579772669108","date":"2025-06-12T06:54:06+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-06-11T12:10:37+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-06-11T11:43:22+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-06-09T04:47:29+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-06-09T04:45:41+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Materials","date":"2025-06-05T13:34:45+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-materials","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"dime","sideBox":"Learn more about [Discover Materials](https://www.springer.com/journal/43939)","snPcode":"","submissionUrl":"","title":"Discover Materials","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"640999f7-9c9d-41db-a113-1d3943461728","owner":[],"postedDate":"June 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-08T16:01:11+00:00","versionOfRecord":{"articleIdentity":"rs-6829733","link":"https://doi.org/10.1007/s43939-025-00452-y","journal":{"identity":"discover-materials","isVorOnly":false,"title":"Discover Materials"},"publishedOn":"2025-12-04 15:57:34","publishedOnDateReadable":"December 4th, 2025"},"versionCreatedAt":"2025-06-17 15:47:16","video":"","vorDoi":"10.1007/s43939-025-00452-y","vorDoiUrl":"https://doi.org/10.1007/s43939-025-00452-y","workflowStages":[]},"version":"v1","identity":"rs-6829733","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6829733","identity":"rs-6829733","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}