Synaptic device, neuromorphic device including synaptic device, and operating methods thereof
The synapse device with a SONS structure mimics synaptic behaviors and integrates well with CMOS technology, addressing compatibility issues and enabling advanced neuromorphic computing.
Patent Information
- Application Number
- KR1020220077435
- Authority / Receiving Office
- KR · KR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-06-24
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2042-06-24
AI Technical Summary
Existing synapse devices lack sufficient mimicry of various synaptic behaviors and poor process compatibility with CMOS technology, limiting their effectiveness in neuromorphic computing systems.
A synapse device with a SONS (doped poly-silicon/blocking oxide/charge trap nitride/silicon channel) structure, where synaptic plasticity is controlled by gate voltage conditions, exhibiting characteristics like SADP, SDDP, SFDP, SNDP, and STDP, and is fully compatible with CMOS technology.
The synapse device effectively simulates biological synaptic behaviors and can be integrated seamlessly into neuromorphic devices and neural networks, enhancing neuromorphic computing systems with ultra-low power consumption.
Smart Images

Figure 112022066222613-PAT00002_ABST
Abstract
Description
Technology Field
[0001] The present invention relates to an electronic device, a device including the same, and a method of operating the same, and more specifically, to a synapse device, a neuromorphic device including the same, and a method of operating the same. Background Technology
[0002] As the scaling down of transistors reaches its limit, neuromorphic computing systems are receiving significant attention as a new concept capable of overcoming the limitations of existing von Neumann-based computer systems. Neuromorphic computing implements artificial intelligence operations by hardware-mimicking the human brain. Based on the observation that the human brain performs very complex functions while consuming only about 20W of energy, neuromorphic computing mimics the structure of the human brain itself, enabling it to perform AI operations such as association, reasoning, and recognition—which are far superior to existing von Neumann-based computing—with ultra-low power consumption.
[0003] Neuromorphic systems that enable such neuromorphic computing are composed of numerous neurons (neuronal elements) and synapses (synaptic elements), just like the human brain, and include additional circuits for signal processing and transmission. Synapses must remember the connection strength (weight) based on the correlation of spikes expressed by neurons, and, in some cases, adjust the said connection strength through processes of potentialization and depression. As synaptic elements, RRAM (resistive random access memory) and memristor-based devices have been widely studied, and recently, MOSFET (metal-oxide-semiconductor field-effect transistor)-based synaptic elements are also being researched.
[0004] However, existing synapse devices have problems and limitations, such as a lack of sufficient mimicry of the various synaptic behaviors of biological synapses and poor process compatibility with general CMOS (complementary metal oxide semiconductor) technology. The problem to be solved
[0005] The technical problem that the present invention aims to solve is to provide a synapse device that can simulate various synaptic behaviors of biological synapses and has excellent process compatibility with CMOS technology.
[0006] In addition, the technical problem that the present invention aims to solve is to provide a method of operating the synapse element.
[0007] In addition, the technical problem that the present invention aims to solve is to provide a neuromorphic device (neuromorphic system) including the synapse element.
[0008] The problems that the present invention aims to solve are not limited to those mentioned above, and other unmentioned problems will be understood by those skilled in the art from the description below. means of solving the problem
[0009] According to one embodiment of the present invention, a synapse device is provided comprising: a channel member; a charge trap layer in direct contact with the channel member; a blocking insulating layer disposed on the charge trap layer; a gate electrode disposed on the blocking insulating layer; a source connected to a first region of the channel member; and a drain connected to a second region of the channel member, wherein the post-synaptic current (PSC) of the synapse device changes and synaptic plasticity is controlled according to the application condition of a gate voltage applied to the gate electrode.
[0010] The above synapse device may have a SONS (doped poly-silicon / blocking oxide / charge trap nitride / silicon channel) structure.
[0011] The gate electrode may include silicon, the blocking insulating layer may include an oxide, the charge trap layer may include a nitride, and the channel member may include silicon.
[0012] The above synapse device may have the characteristic that charges trapped in the charge trap layer diffuse and flow out into the channel member over time due to the application of the gate voltage.
[0013] The synaptic plasticity can be controlled according to at least one of the amplitude, duration, frequency, number, and timing of the pulse of the gate voltage.
[0014] The above synapse device may have all of the following characteristics: SADP (spike amplitude dependent plasticity), SDDP (spike duration dependent plasticity), SFDP (spike frequency dependent plasticity), SNDP (spike number dependent plasticity), and STDP (spike timing dependent plasticity).
[0015] According to another embodiment of the present invention, a neuromorphic device comprising the aforementioned synapse element is provided.
[0016] The neuromorphic device may further include a CMOS peripheral circuit connected to the synapse device.
[0017] According to another embodiment of the present invention, a method for operating a synapse device is provided, comprising the steps of: providing a synapse device comprising a channel member, a charge trap layer in direct contact with the channel member, a blocking insulating layer disposed on the charge trap layer, a gate electrode disposed on the blocking insulating layer, a source connected to a first region of the channel member, and a drain connected to a second region of the channel member; and applying a gate voltage to the gate electrode of the synapse device, wherein the post-synaptic current (PSC) of the synapse device changes and synaptic plasticity is controlled according to the application condition of the gate voltage.
[0018] The above synapse device may have a SONS (doped poly-silicon / blocking oxide / charge trap nitride / silicon channel) structure.
[0019] The gate electrode may include silicon, the blocking insulating layer may include an oxide, the charge trap layer may include a nitride, and the channel member may include silicon.
[0020] The above synapse device may have the characteristic that charges trapped in the charge trap layer diffuse and flow out into the channel member over time due to the application of the gate voltage.
[0021] The synaptic plasticity can be controlled according to at least one of the amplitude, duration, frequency, number, and timing of the pulse of the gate voltage.
[0022] The above synapse device may have all of the following characteristics: SADP (spike amplitude dependent plasticity), SDDP (spike duration dependent plasticity), SFDP (spike frequency dependent plasticity), SNDP (spike number dependent plasticity), and STDP (spike timing dependent plasticity). Effects of the invention
[0023] According to embodiments of the present invention, it is possible to simulate various synaptic behaviors of biological synapses and, furthermore, to realize a synapse device and a method of operation thereof that have excellent process compatibility with CMOS technology. In particular, it is possible to realize a synapse device in which the post-synaptic current (PSC) changes depending on the application conditions of the gate voltage applied to the gate electrode and synaptic plasticity can be easily controlled.
[0024] More specifically, the synapse device according to the embodiment can control synaptic plasticity according to, for example, the amplitude, duration, frequency, number, and timing of a pulse signal applied to a gate electrode, and can exhibit all of SADP (spike amplitude dependent plasticity) characteristics, SDDP (spike duration dependent plasticity) characteristics, SFDP (spike frequency dependent plasticity) characteristics, SNDP (spike number dependent plasticity) characteristics, and STDP (spike timing dependent plasticity) characteristics. In addition, the synapse device according to the embodiment can be 100% compatible with existing CMOS technology.
[0025] Accordingly, the synapse device according to the embodiment can be used as an artificial synapse capable of replacing biological synapses and can be usefully utilized in neuromorphic devices (neuromorphic systems) and neural networks.
[0026] However, the effects of the present invention are not limited to the above effects and can be extended in various ways without departing from the technical concept and scope of the present invention. Brief explanation of the drawing
[0027] FIG. 1 is a perspective view showing a synapse device according to one embodiment of the present invention. FIG. 2 is a diagram showing the charge trap and outflow characteristics of a synapse device according to one embodiment of the present invention. Figure 3 is a diagram showing the characteristics of ion inflow and outflow in an actual biological synapse. FIG. 4 is a transmission electron microscope (TEM) image showing a stack structure that a synapse device according to one embodiment of the present invention may have. FIG. 5 is a graph showing how the post-synaptic current (PSC) of a synapse device changes when a gate voltage (pulse voltage) is applied to the gate electrode of the synapse device according to one embodiment of the present invention. Figure 6 is a characteristic evaluation graph showing the SADP (spike amplitude dependent plasticity) characteristics of a synapse device according to one embodiment of the present invention. Figure 7 is a characteristic evaluation graph showing the spike duration dependent plasticity (SDDP) characteristics of a synapse device according to one embodiment of the present invention. FIG. 8 is a characteristic evaluation graph showing the SFDP (spike frequency dependent plasticity) characteristics of a synapse device according to one embodiment of the present invention. FIG. 9 is a characteristic evaluation graph showing the SNDP (spike number dependent plasticity) characteristics of a synapse device according to one embodiment of the present invention. FIG. 10 is a characteristic evaluation graph showing the STDP (spike timing dependent plasticity) characteristics of a synapse device according to one embodiment of the present invention. FIG. 11 is a perspective view showing a circuit configuration including a synapse element according to one embodiment of the present invention. FIG. 12 is a circuit diagram showing a synaptic array device in which a synapse element according to one embodiment of the present invention is applied as a unit element. Specific details for implementing the invention
[0028] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings.
[0029] The embodiments of the present invention described below are provided to more clearly explain the present invention to those skilled in the art, and the scope of the present invention is not limited by the following embodiments, and the following embodiments may be modified in various other forms.
[0030] The terms used herein are for describing specific embodiments and are not intended to limit the invention. Terms used herein in the singular form may include plural forms unless the context clearly indicates otherwise. Additionally, the terms “comprise” and / or “comprising” used herein specify the presence of the mentioned features, steps, numbers, actions, components, elements, and / or groups thereof, and do not exclude the presence or addition of one or more other features, steps, numbers, actions, components, elements, and / or groups thereof. Furthermore, the term “connected” used herein means not only that components are directly connected, but also includes the concept of indirectly connecting components through the interposition of additional components between them.
[0031] Furthermore, when a component is described in this specification as being located "on" another component, this includes not only cases where a component is in contact with another component, but also cases where another component exists between the two components. The term "and / or" as used in this specification includes any one of the listed items and all combinations of one or more thereof. Additionally, terms of degree such as "about" and "substantially" as used in this specification are used to mean a range of numerical values or degrees or approximate values, taking into account inherent manufacturing and material tolerances, and are used to prevent an infringer from unfairly exploiting the disclosures in which precise or absolute figures provided to aid in understanding this specification are mentioned.
[0032] Embodiments of the present invention will be described in detail below with reference to the attached drawings. The sizes or thicknesses of the areas or parts depicted in the attached drawings may be slightly exaggerated for the clarity of the specification and convenience of explanation. Throughout the detailed description, the same reference numerals indicate the same components.
[0033] FIG. 1 is a perspective view showing a synapse device according to one embodiment of the present invention.
[0034] Referring to FIG. 1, a synapse element according to an embodiment of the present invention may include a channel member (10), a charge trap layer (20) in direct contact with the channel member (10), a blocking insulating layer (30) disposed on the charge trap layer (20), a gate electrode (40) disposed on the blocking insulating layer (30), a source (11) connected (contacted) to a first region (e.g., a first end) of the channel member (10), and a drain (12) connected (contacted) to a second region (e.g., a second end) of the channel member. The channel member (10) may be disposed between the source (11) and the drain (12), and the charge trap layer (20), the blocking insulating layer (30), and the gate electrode (40) may be disposed sequentially on the channel member (10). A charge trap layer (20) may be disposed between a channel member (10) and a blocking insulating layer (30), and the blocking insulating layer (30) may be disposed between a charge trap layer (20) and a gate electrode (40). Here, the charge trap layer (20) may be in direct contact with the surface of the channel member (10). When the charge trap layer (20), the blocking insulating layer (30), and the gate electrode (40) form a single stacked structure, a sidewall insulating layer (50) may be further formed on the sidewalls (both sidewalls) of the stacked structure. In FIG. 1, for convenience, parts of the charge trap layer (20), the blocking insulating layer (30), the gate electrode (40), and the sidewall insulating layer (50) are shown in a cut-off form.
[0035] The above synapse element may be formed on a predetermined substrate structure. The substrate structure may include, for example, a silicon-on-insulator (SOI) substrate. The SOI substrate may include a lower substrate layer (1), an insulating layer (2) sequentially disposed thereon, and an upper substrate layer. The upper substrate layer may include silicon, and a channel member (10), a source (11), and a drain (12) may be formed from the upper substrate layer. However, the use of the SOI substrate is exemplary, and a general semiconductor substrate other than an SOI substrate may be used. For example, the synapse element may be formed on various substrates such as a silicon substrate, a germanium substrate, a silicon-germanium substrate, or a silicon carbide substrate.
[0036] According to an embodiment of the present invention, the post-synaptic current (PSC) of the synapse device can be changed and synaptic plasticity controlled according to the application conditions of the gate voltage applied to the gate electrode (40). Depending on the application conditions of the gate voltage, the fluctuation range of the post-synaptic current (PSC) can be changed, and the synaptic plasticity can be strengthened or weakened. In this regard, the synapse device according to an embodiment of the present invention may be capable of simulating various synaptic behaviors of biological synapses. This will be explained in more detail later with reference to FIGS. 5 to 10, etc.
[0037] According to one embodiment, the gate electrode (40) may include silicon, the blocking insulating layer (30) may include oxide, the charge trap layer (20) may include nitride, and the channel member (10) may include silicon. As a more specific example, the gate electrode (40) may include polycrystalline silicon with a high concentration doped, the blocking insulating layer (30) may include silicon oxide (e.g., SiO2), the charge trap layer (20) may include silicon nitride (e.g., Si3N4), and the channel member (10) may include single-crystal silicon or polycrystalline silicon. Accordingly, the synapse device may be said to have a SONS (doped poly-silicon / blocking oxide / charge trap nitride / silicon channel) structure.
[0038] In this way, the synapse device may have a structure in which the channel member (10) and the charge trap layer (20) are in direct contact with each other without a tunnel insulation layer (e.g., tunneling oxide layer) between them. In this regard, the synapse device may exhibit characteristics similar to biological synapses. For example, the synapse device may exhibit a characteristic in which charges trapped in the charge trap layer (20) by the application of the gate voltage diffuse and flow out to the channel member (10) relatively quickly over time. This can be described as a characteristic similar to the increase in action potential due to ion inflow and the decrease in action potential due to ion outflow that occur during the neurotransmission process of actual biological synapses. This will be explained in more detail below with reference to FIGS. 2 and FIGS. 3.
[0039] Since the above synapse device has a high degree of similarity to biological synapses, based on these characteristics, just as the plasticity of actual biological synapses is controlled according to the amplitude, duration, frequency, number, and timing of the stimulus, the synaptic plasticity of the above synapse device can also be controlled in a similar manner. The synaptic plasticity of the above synapse device can be easily controlled according to at least one of the amplitude, duration, frequency, number, and timing of the pulse (pulse signal) of the gate voltage applied to the gate electrode (40). In this regard, the above synapse device may have all of the following characteristics: SADP (spike amplitude dependent plasticity) characteristics, SDDP (spike duration dependent plasticity) characteristics, SFDP (spike frequency dependent plasticity) characteristics, SNDP (spike number dependent plasticity) characteristics, and STDP (spike timing dependent plasticity) characteristics. Accordingly, the synapse device according to the embodiment can be used as an artificial synapse capable of replacing biological synapses and can be usefully utilized in neuromorphic devices (neuromorphic systems) and neural networks.
[0040] In addition, the synapse device may have excellent process compatibility with CMOS technology. Since the synapse device may have, for example, the above-described SONS structure, it may be 100% compatible with existing CMOS technology. Therefore, a CMOS-type peripheral circuit (i.e., a CMOS peripheral circuit) connected to the synapse device can be easily manufactured, and a neuromorphic device (neuromorphic system) with excellent performance can be easily manufactured.
[0041] Additionally, the synapse device may have a FinFET (fin field effect transistor) structure. A channel member (10) may be positioned to protrude upward from the upper surface of an insulating layer (2), and a charge trap layer (20), a blocking insulating layer (30), and a gate electrode (40) may be positioned to cover (encircle) the upper surface and both sides of the channel member (10). The performance of the synapse device may be further enhanced by this FinFET structure. However, the synapse device may have a planar-type transistor structure or a GAA (gate-all-around) transistor structure, etc., rather than a FinFET structure.
[0042] In addition, the thickness of the channel member (10) in the above-described synapse element may be, for example, about 1 to 50 nm, the thickness of the charge trap layer (20) may be, for example, about 1 to 10 nm, and the thickness of the blocking insulating layer (30) may be, for example, about 1 to 10 nm. However, the thickness ranges of the channel member (10), the charge trap layer (20), and the blocking insulating layer (30) are exemplary and may vary depending on the case.
[0043] The above synapse device can be described as a charge-trap device modulated by a gate, and the synaptic plasticity of the above synapse device can be controlled using the gate voltage.
[0044] FIGS. 2 and FIGS. 3 are drawings illustrating the similarity between a synapse element according to one embodiment of the present invention and an actual biological synapse.
[0045] FIG. 2 is a diagram showing the charge trap and outflow characteristics of a synapse device according to one embodiment of the present invention.
[0046] Referring to FIG. 2, a synapse device according to an embodiment of the present invention may have a structure in which the channel member (10) and the charge trap layer (20) are in direct contact with each other without a tunnel insulating layer between them. Accordingly, the synapse device may exhibit a characteristic in which charges trapped in the charge trap layer (20) diffuse and flow out to the channel member (10) within a relatively short period of time due to the application of a gate voltage (stimulation). As charges (carriers) are trapped in the charge trap layer (20), the channel current may increase rapidly, and then decrease as the charges (carriers) diffuse and flow out to the channel member (10). This may be a characteristic similar to the increase in action potential due to ion inflow and the decrease in action potential due to ion outflow that occur in the neurotransmission process of actual biological synapses.
[0047] For example, when a negative (-) voltage is applied to the gate electrode (40) as the gate voltage, a positive (+) charge, i.e., a hole, can be trapped in the channel member (10), and the trapped holes can be discharged into the channel member (10) within a relatively short time after the application of the gate voltage. In other cases, when a positive (+) voltage is applied to the gate electrode (40) as the gate voltage, a negative (-) charge, i.e., an electron, can be trapped in the channel member (10), and the trapped electrons can be discharged into the channel member (10) within a relatively short time after the application of the gate voltage.
[0048] A synapse device according to an embodiment of the present invention may have a SONS structure, and may have the aforementioned charge trapping and outflow characteristics through such a SONS structure. In the case of a transistor having a conventional SONOS (silicon / oxide / nitride / oxide / silicon) structure, since it includes a tunnel insulating layer, the characteristic of charge trapped in the trap layer flowing out by diffusion may not appear, and the ability to simulate biological synapses may be reduced. However, a synapse device according to an embodiment of the present invention may have the aforementioned charge trapping and outflow characteristics, which can be usefully utilized in simulating synapse behavior.
[0049] Figure 3 is a diagram showing the characteristics of ion inflow and outflow in an actual biological synapse.
[0050] Referring to FIG. 3, in the actual process of neurotransmission in biological synapses, characteristics of an increase in action potential due to ion inflow into the synapse (synaptic cleft) and a decrease in action potential due to ion outflow may appear. As ions flow into the synapse (synaptic cleft), the action potential may increase rapidly, and then decrease as ions outflow relatively quickly over time. This may be similar to the charge trap and outflow characteristics of the synapse device according to the embodiment described with reference to FIG. 2.
[0051] Since the above synapse device has a high degree of similarity to biological synapses, based on these characteristics, just as the plasticity of actual biological synapses is controlled according to the amplitude, duration, frequency, number, and timing of the stimulus, the synaptic plasticity of the above synapse device can also be controlled in a similar manner. The synaptic plasticity of the above synapse device can be easily controlled according to at least one of the amplitude, duration, frequency, number, and timing of the pulse (pulse signal) of the gate voltage applied to the gate electrode (40). In this regard, the above synapse device may have all of the following characteristics: SADP (spike amplitude dependent plasticity) characteristics, SDDP (spike duration dependent plasticity) characteristics, SFDP (spike frequency dependent plasticity) characteristics, SNDP (spike number dependent plasticity) characteristics, and STDP (spike timing dependent plasticity) characteristics. Accordingly, the synapse device according to the embodiment can be used as an artificial synapse capable of replacing biological synapses and can be usefully utilized in neuromorphic devices (neuromorphic systems) and neural networks.
[0052] FIG. 4 is a transmission electron microscope (TEM) image showing a stack structure that a synapse device according to one embodiment of the present invention may have.
[0053] Referring to FIG. 4, the synapse device according to the embodiment may have a SONS (doped poly-silicon / blocking oxide / charge trap nitride / silicon channel) structure. The gate electrode (40) of the synapse device may include silicon, the blocking insulating layer (30) may include oxide, the charge trap layer (20) may include nitride, and the channel member (10) may include silicon. The gate electrode (40) may include polycrystalline silicon doped to a high concentration, the blocking insulating layer (30) may include silicon oxide (e.g., SiO2), the charge trap layer (20) may include silicon nitride (e.g., Si3N4), and the channel member (10) may include single-crystal silicon or polycrystalline silicon. The above SONS structure may be, for example, an n+ poly-silicon / SiO2 / Si3N4 / p-type cystal silicon structure. However, the doping type of the gate electrode (40), the semiconductor type of the channel member (10), and the specific material composition of the charge trap layer (20) and the blocking insulating layer (30) may vary. The above synapse device may have characteristics similar to biological synapses (i.e., the potential increases and then rapidly decreases) due to the SONS structure.
[0054] FIG. 5 is a graph showing how the post-synaptic current (PSC) of a synapse device changes when a gate voltage (pulse voltage) (P1, P2) is applied to the gate electrode of the synapse device according to one embodiment of the present invention. The post-synaptic current (PSC) may be a current between a source and a drain.
[0055] Referring to FIG. 5, when a gate voltage (pulse voltage) (P1, P2) is applied to the gate electrode of the synapse device according to the embodiment, it can be seen that the post-synaptic current (PSC) exhibits behavior of increasing and decreasing similar to changes in the action potential of a biological synapse. In addition, when the gate voltages (P1, P2) (i.e., stimuli) are closely paired, it can be seen that the increase in the post-synaptic current (PSC) is prominent at the second stimulus (i.e., P2). This can be described as a paired-pulse facilitation (PPF) characteristic.
[0056] Figure 6 is a characteristic evaluation graph showing the SADP (spike amplitude dependent plasticity) characteristics of a synapse device according to one embodiment of the present invention.
[0057] Referring to FIG. 6, the gate voltage (V) applied to the gate electrode of the synapse element G It can be observed that the increase in postsynaptic current (PSC) becomes larger as the amplitude of the stimulus increases. This phenomenon, which also occurs in biological synapses, is called the SADP characteristic. In Fig. 6, V D represents the drain voltage, and t G represents the duration of the gate voltage. Gate voltage (V G ) was authorized at the 0.1 second mark.
[0058] Figure 7 is a characteristic evaluation graph showing the spike duration dependent plasticity (SDDP) characteristics of a synapse device according to one embodiment of the present invention.
[0059] Referring to FIG. 7, the gate voltage (V) applied to the gate electrode of the synapse element G Duration of )(stimulus)(t GIt can be observed that the increase in postsynaptic current (PSC) becomes larger as ) increases. This phenomenon is also seen in biological synapses and is called the SDDP characteristic. Gate voltage (V G ) was authorized at the 0.1 second mark.
[0060] FIG. 8 is a characteristic evaluation graph showing the SFDP (spike frequency dependent plasticity) characteristics of a synapse device according to one embodiment of the present invention.
[0061] Referring to FIG. 8, the gate voltage (V) applied to the gate electrode of the synapse element G It can be observed that as the frequency of stimulation increases, the amount of change in the postsynaptic current (PSC), i.e., ΔW (%), increases. This phenomenon also occurs in biological synapses and is called the SFDP characteristic.
[0062] FIG. 9 is a characteristic evaluation graph showing the SNDP (spike number dependent plasticity) characteristics of a synapse device according to one embodiment of the present invention.
[0063] Referring to Fig. 9, it can be observed that as the number of stimuli (gate voltage signals) having the same frequency increases, the increase in the post-synaptic current (PSC) of the synaptic element becomes larger. As the number of spikes of the stimulus increases from 1 to 100, the post-synaptic current (PSC) increases significantly. This phenomenon also occurs in biological synapses and is called the SNDP characteristic.
[0064] FIG. 10 is a characteristic evaluation graph showing the STDP (spike timing dependent plasticity) characteristics of a synapse device according to one embodiment of the present invention.
[0065] Referring to Fig. 10, it can be observed that the amount of change in the postsynaptic current (PSC), i.e., ΔW (%), changes depending on the timing (i.e., the interval between) when the paired stimuli (gate voltage signals) are applied. The interval (Δt) between the paired stimuli (i.e., Δt interval As ) decreases, ΔW(%) may increase. This is also observed in biological synapses and can be considered a characteristic of STDP. In particular, ΔW(%) showed a tendency to increase as Δt approached 0 in the range of approximately -15 μs to 15 μs.
[0066] As described with reference to FIGS. 6 to 10, the synapse device according to an embodiment of the present invention may have all of SADP characteristics, SDDP characteristics, SFDP characteristics, SNDP characteristics, and STDP characteristics. Accordingly, the synapse device according to the embodiment can be used as an artificial synapse capable of replacing biological synapses and can be usefully utilized in neuromorphic devices (neuromorphic systems) and neural networks.
[0067] The operation method of a synapse element according to an embodiment of the present invention is described as follows.
[0068] The method of operating the above synapse device may include the steps of providing a synapse device comprising a channel member, a charge trap layer in direct contact with the channel member, a blocking insulating layer disposed on the charge trap layer, a gate electrode disposed on the blocking insulating layer, a source connected to a first region of the channel member, and a drain connected to a second region of the channel member, and applying a gate voltage to the gate electrode of the synapse device, wherein the post-synaptic current (PSC) of the synapse device may change and synaptic plasticity may be controlled according to the application conditions of the gate voltage.
[0069] The gate electrode may comprise silicon, the blocking insulating layer may comprise an oxide, the charge trap layer may comprise a nitride, and the channel member may comprise silicon. For example, the synapse device may have a SONS (doped poly-silicon / blocking oxide / charge trap nitride / silicon channel) structure. The synapse device may have a characteristic in which charges trapped in the charge trap layer diffuse and flow out to the channel member over time due to the application of the gate voltage. Additionally, the synaptic plasticity of the synapse device may be controlled according to at least one of the amplitude, duration, frequency, number, and timing of the gate voltage pulse. The above synapse device may have all of the following characteristics: SADP (spike amplitude dependent plasticity), SDDP (spike duration dependent plasticity), SFDP (spike frequency dependent plasticity), SNDP (spike number dependent plasticity), and STDP (spike timing dependent plasticity).
[0070] It can be understood that the method of operation of the synapse device according to the above embodiment may include all structural and operational features as described with reference to FIGS. 1, 2, and FIGS. 4 to 10. Accordingly, all contents described with reference to FIGS. 1, 2, and FIGS. 4 to 10 may be applied to the method of operation of the synapse device.
[0071] When circuitously configuring a synapse, which is the connection between a pre-neuron and a post-neuron, a synapse element according to an embodiment of the present invention may be applied. A pre-neuron can input a pre-spike signal to the synapse, the synapse can transmit a predetermined synaptic signal to the post-neuron, and the post-neuron can generate a post-spike signal. Similar to how a synapse connects a pre-neuron and a post-neuron, the synapse element may serve to connect a pre-synaptic neuron circuit and a post-synaptic neuron circuit. A circuit diagram of this configuration may be as shown in FIG. 11.
[0072] FIG. 11 is a perspective view showing a circuit configuration including a synapse element according to one embodiment of the present invention.
[0073] Referring to FIG. 11, the gate electrode (40) of the synapse device may be connected to a pre-synaptic neuron circuit (N1). The source (11) of the synapse device may be connected to a post-synaptic neuron circuit (N2). A pre-spike signal from the pre-synaptic neuron circuit (N1) may be applied to the gate electrode (40), and a post-synaptic current may flow through the source (11) to the post-synaptic neuron circuit (N2). A post-spike signal may be generated from the post-synaptic neuron circuit (N2). Meanwhile, a predetermined voltage (V) is applied to the drain (12) of the above-mentioned synapse element. DS ) can be applied constantly. That is, a predetermined constant voltage (V DS ) can be applied to the drain (12).
[0074] FIG. 12 is a circuit diagram showing a synaptic array element in which a synapse element (S10) according to one embodiment of the present invention is applied as a unit element.
[0075] Referring to FIG. 12, a plurality of synapse elements (S10) may be arranged to form a plurality of columns and a plurality of rows. A plurality of first wires (W10) may be arranged, and a plurality of second wires (W20) intersecting therewith may be arranged, and a synapse element (S10) may be provided at the intersection of the first wires (W10) and the second wires (W20). A plurality of first wires (W10) may be connected to the gate electrodes of the synapse elements (S10), and a plurality of second wires (W20) may be connected to the sources of the synapse elements (S10). A first wire (W10) may be connected to a pre-synaptic neuron circuit (N10), and a second wire (W20) may be connected to a post-synaptic neuron circuit (N20). Meanwhile, a predetermined voltage (V) is applied to the drain of the synapse element (S10). DS ) can be authorized.
[0076] A pre-spike signal can be applied to the gate electrode of a synapse element (S10) through the first wiring (W10) from a pre-synaptic neuron circuit (N10). A post-synaptic current can flow to a post-synaptic neuron circuit (N20) through the source of the synapse element (S10). A post-spike signal can be generated from the post-synaptic neuron circuit (N20).
[0077] According to an embodiment of the present invention, a neuromorphic device and system can be configured by applying a synapse device according to the above-described embodiment. The neuromorphic device may include a CMOS peripheral circuit connected to the synapse device. The CMOS peripheral circuit may include a pre-synaptic neuron circuit and a post-synaptic neuron circuit, etc. The synapse device is according to an embodiment of the present invention and may have, for example, an array structure as described in FIG. 12. The configuration of a neuromorphic device and system applying a synapse device is well known, so a detailed description thereof is omitted. Since the synapse device according to an embodiment of the present invention is 100% compatible with CMOS technology, a neuromorphic device including the synapse device and the CMOS peripheral circuit can be easily implemented.
[0078] According to the embodiments of the present invention described above, it is possible to simulate various synaptic behaviors of biological synapses and, furthermore, to realize a synapse device and a method of operation thereof that have excellent process compatibility with CMOS technology. In particular, it is possible to realize a synapse device in which the post-synaptic current (PSC) changes according to the application conditions of the gate voltage applied to the gate electrode and synaptic plasticity can be easily controlled. More specifically, the synapse device according to the embodiment can control synaptic plasticity according to, for example, the amplitude, duration, frequency, number, and timing of the pulse signal applied to the gate electrode, and can exhibit all of the following characteristics: SADP (spike amplitude dependent plasticity), SDDP (spike duration dependent plasticity), SFDP (spike frequency dependent plasticity), SNDP (spike number dependent plasticity), and STDP (spike timing dependent plasticity). In addition, the synapse device according to the embodiment can be 100% compatible with existing CMOS technology. Therefore, the synapse device according to the embodiment can be used as an artificial synapse that can replace biological synapses, and can be usefully utilized in neuromorphic devices (neuromorphic systems) and neural networks.
[0079] This specification discloses preferred embodiments of the present invention. Although specific terms have been used, they are used merely in a general sense to facilitate the explanation of the technical content of the invention and to aid in understanding the invention, and are not intended to limit the scope of the invention. It is obvious to those skilled in the art that, in addition to the embodiments disclosed herein, other variations based on the technical concept of the present invention are possible. Those skilled in the art will understand that the synapse device according to the embodiment described with reference to FIGS. 1 to 12, the neuromorphic device including the same, and the method of operation thereof can be variously substituted, changed, and modified within the scope of the technical concept of the present invention. Therefore, the scope of the invention should not be determined by the described embodiments but by the technical concept described in the claims. Explanation of the symbols
[0080] * Explanation of symbols for major parts of the drawing * 1 : Lower substrate layer 2 : Insulating layer 10 : Channel missing 11 : Source 12: Drain 20: Charge trap layer 30: Blocking insulating layer 40: Gate electrode 50: Sidewall insulating layer N1: Pre-synaptic neuron circuit N2: Post-synaptic neuron circuit N10: Pre-synaptic neuron circuit N20: Post-synaptic neuron circuit S10: Synaptic element W10: 1st wiring W20: 2nd wiring
Claims
Claim 1 A synapse device comprising: a channel member; a charge trap layer in direct contact with the channel member; a blocking insulating layer disposed on the charge trap layer; a gate electrode disposed on the blocking insulating layer; a source connected to a first region of the channel member; and a drain connected to a second region of the channel member, wherein no tunnel insulating layer is formed between the channel member and the charge trap layer, and the synapse device has a SONS (doped poly-silicon / blocking oxide / charge trap nitride / silicon channel) structure, wherein the post-synaptic current (PSC) of the synapse device changes and synaptic plasticity is controlled according to the application condition of a gate voltage applied to the gate electrode. Claim 2 delete Claim 3 A synapse device according to claim 1, wherein the gate electrode comprises silicon, the blocking insulating layer comprises an oxide, the charge trap layer comprises a nitride, and the channel member comprises silicon. Claim 4 A synapse device according to claim 1, having the characteristic that charges trapped in the charge trap layer diffuse and flow out into the channel member over time by the application of the gate voltage. Claim 5 A synaptic device according to claim 1, wherein the synaptic plasticity is controlled according to at least one of the amplitude, duration, frequency, number, and timing of the pulse of the gate voltage. Claim 6 In claim 1, the synapse device is a synapse device having all of the following characteristics: SADP (spike amplitude dependent plasticity) characteristics, SDDP (spike duration dependent plasticity) characteristics, SFDP (spike frequency dependent plasticity) characteristics, SNDP (spike number dependent plasticity) characteristics, and STDP (spike timing dependent plasticity) characteristics. Claim 7 A neuromorphic device comprising a synapse element as described in any one of claims 1, 3 to 6. Claim 8 A neuromorphic device according to claim 7, further comprising a CMOS peripheral circuit connected to the synapse element. Claim 9 A method of operating a synapse device comprising: a step of providing a synapse device including a channel member, a charge trap layer in direct contact with the channel member, a blocking insulating layer disposed on the charge trap layer, a gate electrode disposed on the blocking insulating layer, a source connected to a first region of the channel member, and a drain connected to a second region of the channel member; and a step of applying a gate voltage to the gate electrode of the synapse device, wherein a tunnel insulating layer is not formed between the channel member and the charge trap layer, and the synapse device has a SONS (doped poly-silicon / blocking oxide / charge trap nitride / silicon channel) structure, and wherein the post-synaptic current (PSC) of the synapse device changes and synaptic plasticity is controlled according to the application condition of the gate voltage. Claim 10 delete Claim 11 A method of operation of a synapse device according to claim 9, wherein the gate electrode comprises silicon, the blocking insulating layer comprises an oxide, the charge trap layer comprises a nitride, and the channel member comprises silicon. Claim 12 In claim 9, the method of operation of a synapse element having the characteristic that charges trapped in the charge trap layer diffuse and flow out into the channel member over time by the application of the gate voltage. Claim 13 A method of operation of a synaptic device in which the synaptic plasticity is controlled according to at least one of the amplitude, duration, frequency, number, and timing of the pulse of the gate voltage in claim 9. Claim 14 In claim 9, the method of operation of a synapse device having all of the following characteristics: SADP (spike amplitude dependent plasticity) characteristics, SDDP (spike duration dependent plasticity) characteristics, SFDP (spike frequency dependent plasticity) characteristics, SNDP (spike number dependent plasticity) characteristics, and STDP (spike timing dependent plasticity) characteristics.
Citation Information
Patent Citations
Method of fabricating synapse memory device
KR1020170034215A
Optical synapse device and method of manufacturing thereof
KR1020220050559A
3D Semiconductor Device and Structure
US20180269229A1
Reconfigurable neuromorphic device
KR1020190135300A
A CHARGE TRAP BASED NEUROMORPHIC SYNAPTIC TRANSISTOR WITH IMPROVED LINEARITY AND SYMMETRICITY BY SCHOTTKY JUNCTIONS, AND a neuromorphic system using it
KR1020220006988A
