A Photoelectric-Stimulated MoS2 Transistor for Neuromorphic Engineering.

The von Neumann bottleneck has spawned the rapid expansion of neuromorphic engineering and brain-like networks. Synapses serve as bridges for information transmission and connection in the biological nervous system. The direct implementation of neural networks may depend on novel materials and devices that mimic natural neuronal and synaptic behavior. By exploiting the interfacial effects between MoS2 and AlOx, we demonstrate that an h-BN-encapsulated MoS2 artificial synapse transistor can mimic the basic synaptic behaviors, including EPSC, PPF, LTP, and LTD. Efficient optoelectronic spikes enable simulation of synaptic gain, frequency, and weight plasticity. The Pavlov classical conditioning experiment was successfully simulated by electrical tuning, showing associated learning behavior. In addition, h-BN encapsulation effectively improves the environmental time stability of our devices. Our h-BN-encapsulated MoS2 artificial synapse provides a new paradigm for hardware implementation of neuromorphic engineering.

1. Introduction

The challenges of traditional computing architectures stem from storage capacity limitations and the high cost of specific data transfer speeds between memory and processors, the so-called von Neumann bottleneck [1-5]. With the advent of the artificial intelligence and big data era, this dilemma is becoming more profound. Brain-inspired neuromorphic engineering is different to the von Neumann architecture, combining memory and computation, with efficient energy utilization, and flexible adaptive and massively parallel computing capabilities [6]. It may achieve unprecedented technological breakthroughs, fundamentally overcoming the von Neumann bottleneck [7, 8]. Artificial synapses, just as those in the biological nervous system [9], play an important role in connecting various neuron blocks as the basic units of neuromorphic engineering [10]. Constructing new, stable, reliable, and efficient artificial high-performance synaptic devices is essential for neuromorphic engineering and neural network computing [11]. Many artificial synaptic devices have been reported, including oxide electric double layer [12-14], ionic liquid/gel transistors [15-20], memristors [21-29], phase-changed memory [30-34], and ferroelectric transistors [35-37]. Also, the unique internal and interfacial structures of two-dimensional (2D) materials, as well as their electrical and optical properties [38-40], make them promising candidates for complex neuromorphic engineering [41-45]. In addition, optical modulation can establish a connection between the external environment and the brain through the visual system [46-48], and combining effective optoelectronic modulation is critical for neuromorphic engineering applications, such as artificial eyes and super vision [49-51].

Here, we demonstrate an efficient photoelectrical tunable h-BN-encapsulated MoS2 synaptic transistor with basic synaptic functions. Furthermore, under electrical modulation, we successfully simulate the impressive Pavlov classical conditioning experiment through Vbg tuning, which realizes the acquisition, extinction, and recovery function of associated learning. Due to the h-BN encapsulation, our devices exhibit superior environmental time stability. Our h-BN-encapsulated MoS2 artificial synaptic transistor provides a novel paradigm for neuromorphic engineering based on 2D materials.

2. Results

First, we fabricated an h-BN-encapsulated MoS2 synaptic transistor on an AlOx/Si substrate, which simulates synaptic behavior by photoelectric stimulation, as shown in Figures 1(a) and 1(b). 2D layered h-BN and MoS2 were prepared by mechanical exfoliating. The surface morphology of our device was characterized by scanning electron microscopy (SEM) and atomic force microscopy (AFM), as shown in Figure 1(c) and , respectively, showing a typical channel width of 10 μm, a length of 15 μm, and the thickness of the MoS2; h-BN was approximately 1.7 and 7 nm. The Raman spectrum shows the characteristic peaks of both materials: Raman shift of the MoS2 characteristic peak is 385,405 cm−1 (Figure 1(e)) and the h-BN is 1366 cm−1 (), which is consistent with previous reports. Figure 1(d) shows the Raman mapping of h-BN-encapsulated MoS2 synaptic transistor at 405 cm−1, the channel MoS2 exhibits intense intensity, and the h-BN/MoS2 overlap region is more strongly correlated with h-BN encapsulation, where the black and gray dashed areas represent the h-BN/MoS2 overlap region and channel MoS2, respectively. A significant peak was observed in the PL spectrum of MoS2 at 1.88 eV photon energy (Figure 1(f)), which is consistent with the band gap of multilayer MoS2. Then, we studied the behavioral characteristics of our h-BN-encapsulated MoS2 synaptic transistor under electrical modulation. Figure 2(a) shows the Ids-Vbg curves of the h-BN-encapsulated MoS2 synaptic transistor with Vds of 0.1, 0.5, and 1 V. The back gate voltage was swept from -6 to 8 V, then swept back, and a noticeable clockwise hysteresis loop was observed, which may be due to charge trapping between the MoS2 and AlOx interfaces. The statistical distribution of the maximum value of the memory window indicates that the memory window of most devices is 2~3 V (see the statistics of 80 devices in in Supplementary Materials). The transfer curves of the h-BN/MoS2/h-BN control devices show no hysteresis window, since the bottom h-BN isolates the interface effect of MoS2 and AlOx (see Figures , the schematic diagram of the control devices, micrograph of the control device, and transfer curves of the control devices in Supplementary Materials). Owing to top encapsulated h-BN, the stability of our devices has been significantly improved (see , output curves and stability of h-BN-encapsulated MoS2 synaptic transistor in Supplementary Materials) [52-55]. We explored the optimal base and pulse voltages for device operation in electrical mode for excitatory and inhibitory synapses, with reference to gain (A5/A1, the amplitude of the postsynaptic current caused by spike is denoted by A) of five consecutive pulses and long-term synaptic weight changes (ΔW/W, calculated by (I − I0)/I0∗100%, where I0 and I represent the current states before and after the application of the pulse signal, respectively. Before applying the pulse signal, we select the average value at the 5th second as I0. After the pulse signal is applied, the average value of the 40th second is selected as I, and the pulse signals are applied at the same time. For excitatory synapses, the gain was maximized when Vbg base was -3 V and pulse was -4 V (pulse duration of 10 ms, interval of 200 ms), as shown in Figure 2(b) (no synaptic excitability of the h-BN/MoS2/h-BN control devices under the same Vbg base and pulse conditions, see in Supplementary Materials). For inhibitory synapses, excitatory spike stimulation was first performed, and then fixed base, incremental Vbg pulse was applied, and gain and weight changes were reduced, that is, the depression effect gradually strengthened, and 8 V was selected as the inhibitory spike (duration of 10 ms, interval of 200 ms), as shown in Figure 2(c) (no synaptic inhibition of the h-BN/MoS2/h-BN control devices under the same Vbg base and pulse conditions, see in Supplementary Materials). Figure 2(d) shows the frequency plasticity of inhibitory synapses with fixed duration of 10 ms and number of 10, and the gain gradually decreases as the frequency increases. Figure 2(e) depicts postsynaptic current characteristics under 30 cumulative excitatory and inhibitory spiking stimulations (duration of 10 ms, interval of 200 ms), which exhibits long-term potentiation and inhibition under electrical mode (the h-BN/MoS2/h-BN control devices have no LTP and LTD characteristics under the same Vbg base and pulse, see in Supplementary Materials). Furthermore, Figure 2(f) shows extracted PSC from excitatory and inhibitory spikes, where electrical potentiation and inhibition are clearly observed. The number-dependent facilitation and depression under electrical stimulation are shown in in Supplementary Materials. The electrical potentiation and inhibition effects under electrical stimulation are attributed to the charges trapping and detrapping at the MoS2-AlOx interfaces. The statistical distribution of the maximum value of the excitatory index indicates that the excitatory index of most devices can reach 500-700% (see the statistics of 80 devices in in Supplementary Materials). Under forward bias (Vbg pulse of 8 V), the oxygen vacancy trapping states in AlOx move toward the channel, trapping the electrons in MoS2, causing channel current to decrease, corresponding to synaptic inhibition. While under reverse bias (Vbg pulse of -4 V), oxygen ions in AlOx move toward MoS2, and the oxygen vacancy trapping states release trapped electrons, resulting in increased channel current, which corresponds to synaptic potentiation (see , physical mechanism under electrical stimulation in Supplementary Materials).

Figure 1

The h-BN-encapsulated MoS2 synaptic transistor for neuromorphic engineering. (a) Schematic of h-BN-encapsulated MoS2 synaptic transistor. (b) Schematic diagram of biological neurons and synapses as a bridge of neuronal connections. (c) False-color SEM image of h-BN-encapsulated MoS2 synaptic transistor. (d) The Raman mapping of h-BN-encapsulated MoS2 synaptic transistor at 405 cm−1, where the black and gray dashed areas represent the h-BN/MoS2 overlap region and channel MoS2, respectively. (e) Raman shift of the MoS2 characteristic peak is 385,405 cm−1. (f) A significant peak was observed in the PL spectrum of MoS2 at 1.88 eV photon energy, which is consistent with the band gap of multilayer MoS2.

Figure 2

Characteristics of h-BN-encapsulated MoS2 synaptic transistor under electrical stimulation. (a) Transfer curve under different Vds. (b) Selecting the optimal base and pulse for the excitatory synapse by gain, when base of -3 V and pulse of -4 V, the maximum gain is obtained. (c) Selecting the optimal pulse of inhibitory synapse by gain and long-term synaptic weight change, the maximum inhibition effect and weight change are obtained when pulse is 8 V. (d) Frequency plasticity of inhibitory synapses, and the gain gradually decreases as the frequency increases. (e) Accumulation of postsynaptic current characteristics under 30 excitatory and inhibitory pulse stimulations. (f) Postsynaptic current characteristics as a function of progressive excitatory and inhibitory pulse stimulation numbers, showing long-term potentiation and inhibition effects.

The realization of the association learning is of great significance for neuromorphic engineering. Pavlov's dog classical conditioning experiment is a typical associative learning experiment in physiology [56-58]. In Pavlov's dog experiment, food is called unconditional stimulation (US), while the bell and salivation are called neutral stimulation (NS) and unconditional response (UR), respectively. Food can cause salivation, while bell ringing alone does not cause salivation. Combining the bell with food, that is, after the bell rings, the dog is fed with food, also causes salivation [57, 59]. Pavlov's dog classical conditioning experiment can be simulated on the proposed h-BN-encapsulated MoS2 synaptic transistor by efficient electrical modulation, as shown in Figure 3. Vbg (base, pulse) of (-5, -4 V) applied to the presynaptic gate is considered to be “bell” (NS), and Vbg (base, pulse) of (-3, -4 V) is considered “food” (US). The postsynaptic source drain channel current acts as synaptic weight, and the synaptic weight of 20 nA is defined as the threshold for the “salivation” response (UR). After a single training, only the “bell” ringing does not cause salivation, but after repeated training, the “bell” ringing can also cause “salivation,” which shows the same effect as feeding “food.” At this point, an association is established between “bell” and “food,” and the corresponding NS “bell” is converted to conditional stimulation (CS), causing a conditional response (CR) that triggers “salivation” similar to US, which is called acquisition. After a long time or reset operation, “salivation” no longer occurs when there is only CS, which means that the association between CS and US is extinct/forgotten. However, after training again, “salivation” occurs again when the “bell” rings only, that is, the association is recovered. In addition, we found that due to the existence of acquisition, the current of single training after recovery is significantly higher than the previous single training, which has exceeded the threshold and “salivation” occurs.

Figure 3

Pavlov's dog classical conditioning experiment implemented by h-BN-encapsulated MoS2 synaptic transistor. Pavlov's dog classical conditioning experiments can be simulated on the proposed h-BN-encapsulated MoS2 synaptic transistor by efficient electrical modulation. Vbg (base, pulse) of (-5, -4 V) applied to the presynaptic gate is considered to be “bell” (NS), and Vbg (base, pulse) of (-3, -4 V) is considered “food” (US). The postsynaptic source drain channel current acts as synaptic weight, and the synaptic weight of 20 nA is defined as the threshold for the “salivation” response.

In addition to electrical modulation, optical spikes also enable efficient regulation of our h-BN-encapsulated MoS2 synaptic transistor, which uses laser pulses as the photogate to adjust the channel conductance (synaptic weight), as shown in Figure 4(a). Figure 4(b) shows the single-laser pulse characteristics (532 nm, duration of 100 ms, power of 50 mW/cm2) of the synaptic transistor at Vbg of 0, -5, and -10 V, which significantly affect the reference current (the single-laser pulse characteristics of our synaptic transistor at 473,655 nm and the single-laser pulse current versus time at three wavelengths with Vds of 1 V are detailed in in Supplementary Materials). Besides, variation of postsynaptic current amplitude under different Vbg(0, -5, -10 V) and single-laser pulses with different wavelengths (473, 532, 655 nm) is shown in Figure 4(c). We found that the PSC amplitude increases significantly with Vbg, where different wavelengths have little effect on the PSC amplitude (both of μA), which may be due to the excitation of the h-BN-encapsulated MoS2 synaptic transistor at each wavelength, resulting in photocarrier accumulation in the channel. Specifically, photogenerated carriers (electron-hole pairs) are generated and separated in the top h-BN under laser duration, in which photogenerated electrons are transferred to MoS2, resulting in an increase in channel current. With the cumulative number of laser pulses, the electrons in MoS2 increase continuously, and the channel current appears to be nonvolatile, corresponding to the LTP behavior of neural synapses (see the physical mechanism under optical stimulation in in Supplementary Materials). Moreover, paired pulse facilitation (PPF) is a dynamic increase in neurotransmitter release that is thought to be critical in biosynaptic function simulations [60], where presynaptic-induced EPSC amplitude decreases with increasing two consecutive pulse intervals (Δt). Figure 4(d) describes the postsynaptic current response when a pair of consecutive laser pulses (532 nm, duration of 100 ms, 50 mW/cm2 of power, Vbg and Vds are 0 and 1 V) is applied. For small Δt, the postsynaptic current is further enhanced, resulting in A2 > A1, corresponding to typical synaptic PPF characteristics. Figure 4(e) exhibits the PPF index A2/A1 as a function of the interval time (Δt), where the red dashed line represents the fitting curve of the double exponential decay function (Equation (1)) [61]. C1 and C2 are the initial magnitudes of the fast and slow phases, and t1 and t2 are the characteristic relaxation times of the phases. For our h-BN-encapsulated MoS2 synaptic transistor, t1 and t2 are about 3.0 and 247.4 ms, respectively, which is faster than in most previous work and is consistent with the relaxation time in biological synapses [16, 20, 62–65]. Moreover, we demonstrate long-term synaptic potentiation and inhibition effects under photoelectric modulation with Vbgand Vds of 0 and 1 V, i.e., implementing optical potentiation (532 nm, laser duration 100 ms, power 50 mW/cm2) and electrical inhibition (Vbg pulse 3 V, duration 50 ms) behaviors in sequence, as shown in Figure 4(f). The optimal Vbg pulses for inhibition under optical stimulation are explored in in Supplementary Materials, and 50 laser-stimulated LTP, followed by 50 electrical stimulation LTD characteristics, are shown in of Supplementary Materials, respectively.

Figure 4

Basic synaptic characteristics of h-BN-encapsulated MoS2 transistor under optical stimulation. (a) Schematic diagram of h-BN-encapsulated MoS2 synaptic transistor under optical stimulus. (b) Single-laser pulse characteristics under different Vbg (0, -5, -10 V). (c) Variation of postsynaptic current amplitude under different Vbg (0, -5, -10 V) and single-laser pulses of different wavelengths (473, 532, 655 nm). (d) Typical paired laser pulse facilitation characteristics. (e) PPF characteristics as a function of paired laser pulse intervals. (f) Postsynaptic current characteristics are a function of excitatory laser pulses and inhibitory electrical pulse stimulation, which also shows long-term potentiation and depression.

Subsequently, we demonstrate the optical neural plasticity of the h-BN-encapsulated MoS2 synaptic transistor. Figure 5(a) exhibits a typical synaptic LTP of our h-BN-encapsulated MoS2 synaptic transistor under optical stimulation (532 nm, laser duration and intervals are 100 ms and 1 s, power of 50 mW/cm2, laser number of 50, Vbg and Vds are 0 and 1 V), and Figure 5(b) is an enlargement of the dotted circle region in Figure 5(a). Gain (A/A1) variation of different wavelengths (473, 532, 655 nm) and pulse numbers under laser stimulation (532 nm, laser duration and interval are 100 and 400 ms, power of 50 mW/cm2, Vbg and Vds are 0 and 1 V) are demonstrated in Figure 5(c), which accumulates the laser pulse numbers, and the gains under three wavelength stimuli gradually increase and tend to saturate. Besides, we demonstrate that with the increase of the laser spiking number, the long-term synaptic weight changes at different wavelengths also gradually increase, indicating the synaptic connections are strengthened, as shown in Figure 5(d). And we found that the ΔW/W induced by the 532 nm laser spike is the largest, that is, the strongest synaptic connection strength, and the weakest was at 655 nm, which may be attributed to the larger the wavelength, the smaller the energy under the same conditions, and the fewer photogenerated carriers are generated, resulting in the weakest connection strength. However, 532 nm may more easily excite our h-BN-encapsulated MoS2 synaptic transistor than 473 nm, by concentrating more photogenerated carriers. Figure 5(e) shows the gain (A50/A1) variation of different wavelengths and laser powers under optical modulation (laser duration and interval are 100 and 400 ms, Vbg and Vds are 0 and 1 V). Abnormally, as the laser power increases, the synaptic gain decreases, which may be attributed to the incremental power intensity causing slight damage to the channel material and degradation of performance. Finally, we also demonstrate the synaptic gain as a function of wavelength and laser frequency (1-50 Hz, duration of 100 ms, power of 50 mW/cm2, Vbg and Vds are 0 and 1 V), which increases with frequency and has a maximum at 50 Hz, as shown in Figure 5(f). PSC, PPF, gain, and ΔW/W tuning all demonstrate the flexibility and diversity of synaptic plasticity in our h-BN-encapsulated MoS2 synaptic transistor. Besides, comparing the performance of h-BN-encapsulated MoS2 artificial synapse with other 2D-based artificial synaptic devices, including organic and inorganic materials (such as PEDOT:PSS, CsPbBr3, Pentacene, EMIM-TFSI, PVA, MoS2, WSe2, graphene, h-BN, and BP) demonstrates the superiority of our devices (see in Supplementary Materials). The acceptable switching power consumption is estimated to be 80 pJ per spike, which is two orders of magnitude lower than the traditional CMOS [66] and even down to femtojoule when the Vds is 0.1 V, close to the human brain [16]. The h-BN-encapsulated MoS2 artificial synaptic transistor provides a novel paradigm for neuromorphic engineering based on 2D materials.

Figure 5

Optical neural plasticity of h-BN-encapsulated MoS2 synaptic transistor. (a) Typical long-term potentiation of h-BN-encapsulated MoS2 synaptic transistor under optical stimulation. (b) Magnification of the dotted circle area in (a). (c) Gain variation of different wavelengths (473, 532, 655 nm) and pulse numbers under laser stimulation. (d) Long-term synaptic weight changes at different wavelengths (473, 532, 655 nm) and pulse numbers under laser stimulation. (e) Gain variation of different wavelengths and laser powers under optical modulation. (f) Gain as a function of laser wavelength and frequency in optical modulation mode.

3. Discussion

In conclusion, our breakthrough, efficient, photoelectrical tunable, diverse h-BN-encapsulated MoS2 synaptic transistor demonstrates basic synaptic functions including EPSC, PPF, LTP, LTD, synaptic gain, frequency, and weight plasticity. In addition, under electrical modulation, we successfully simulated the Pavlov classical conditioning experiment and realized the associated learning function. It is worth mentioning that due to the h-BN encapsulation, our devices have superior environmental stability. Our synaptic transistor provides an unparalleled perspective on novel 2D material-based neuromorphic engineering and brain-like computing.

4. Materials and Methods

4.1. Preparation of the h-BN-Encapsulated MoS2 Synaptic Transistor

We fabricated the h-BN-encapsulated MoS2 synaptic transistor on an AlOx/Si substrate, which simulates synaptic behavior by photoelectric stimulation, as shown in Figures 1(a) and 1(b). Firstly, two-dimensional layered h-BN and MoS2 were prepared by mechanical exfoliation, and their thicknesses were determined by an atomic force microscope (AFM) to be about 7 and 1.7 nm, as shown in . The h-BN was placed on top of the MoS2 by wet transfer using polyvinyl alcohol (PVA) as a sacrificial layer to construct an h-BN/MoS2 heterojunction and depositing the 30 nm source-drain electrodes by electron beam evaporation (EBE) to form the synaptic transistor. The detailed manufacturing process of the h-BN-encapsulated MoS2 synaptic transistor is shown in .

4.2. Device Characterization and Measurement

The surface morphology of our device was characterized by scanning electron microscopy (SEM) and atomic force microscopy (AFM), as shown in Figure 1(c) and , respectively, showing a typical channel width of 10 μm, a length of 15 μm, and the thickness of the MoS2; h-BN was approximately 1.7 and 7 nm. Besides, the 2D layered material was characterized by Raman and PL spectroscopy, as shown in Figures 1(e) and 1(f) and . MoS2 shows two strong peaks near 385 and 405 cm−1, corresponding to the in-plane (E2g) and vertical (A1g) vibration models. The Raman spectrum of h-BN reveals a peak near 1366 cm−1 and also corresponds to the in-plane (E2g) vibration model. The Raman mapping of h-BN-encapsulated MoS2 synaptic transistor at 405 cm−1 is shown in Figure 1(d), the channel MoS2 exhibits intense intensity, and the h-BN/MoS2 overlap region is more strongly correlated with h-BN encapsulation, where the black and gray dashed areas represent the h-BN/MoS2 overlap region and channel MoS2, respectively. A distinct peak was observed in the PL spectrum of MoS2 at 1.88 eV photon energy, which is consistent with the band gap of multilayer MoS2. All electrical measurements of our device were performed on the cascade probe station and Keithley 4200A semiconductor analyzer, and optical measurements were performed on the TTL/analog-modulated multiwavelength (655, 532, 473 nm) laser system.