Flexible organic synaptic device based on poly (methyl methacrylate):CdSe/CdZnS quantum-dot nanocomposites

A synaptic device that functionally mimics a biological synapse is a promising candidate for use as an electronic element in a neuromorphic system. In this study, flexible electronic synaptic devices based on poly (methyl methacrylate) (PMMA):CdSe/CdZnS core-shell quantum-dot (QD) nanocomposites are demonstrated. The current-voltage characteristics for the synaptic devices under consecutive voltage sweeps show clockwise hysteresis, which is a critical feature of an artificial synaptic device. The effect of the CdSe/CdZnS QD concentration on the device performance is studied. The flexible electronic synaptic devices under bending show the similar and stable electrical performances. The memory retention measurements show that the e-synapse exhibits long-term potentiation and depression. The carrier transport mechanisms are analyzed, and thermionic emission and space-charge-limited-current conduction are found to be dominant.


Results and Discussion
show schematics of the structure of a synapse and the structure of the synaptic device used in this work, respectively, and Fig. 1(c) shows a cross-sectional SEM image of the fabricated synaptic device. The SEM image shows that the PMMA:CdSe/CdZnS QD layer, which has a thickness of 35 nm, is uniformly deposited on the ITO bottom electrode. The PEDOT:PSS layer with a thickness of about 80 nm was used to enhance the adhesion between the active layer and the ITO electrode. Figure 1(d) presents the PL spectrum and the absorption spectrum for the CdSe/CdZnS QDs, which clearly indicate the high quality of the CdSe/CdZnS QDs used in this work.
Note that synthesizing QDs with a highly uniform size-distribution is difficult. Thus, we only focus on the effects of the number of CdSe/CdZnS QDs in the nanocomposite on the device performance. The effects of the amount of CdSe/CdZnS QDs on the device performance were studied by measuring the I-V characteristics of the synaptic devices for various amounts of PMMA:CdSe/CdZnS QDs, and the results are shown in Fig. 2(a-e) for devices 1, 2, 3, 4, and 5 with 20, 33.3, 42.9, 50, and 55.6 wt% of CdSe/CdZnS QDs in the nanocomposite layers, respectively. For the current measurements, five consecutive positive voltage sweeps were applied to the synaptic devices. As shown in Fig. 2, the amount of CdSe/CdZnS QDs in the PMMA matrix clearly has an effect on the electrical properties of the synaptic devices. Device 1 under dual voltage (0-3-0 V) sweeps showed a relatively large current in comparison with other devices, and the shape of its I-V appears to be linear, as shown in Fig. 2(a). However, device 1 cannot be used as synaptic device because the difference in current between five consecutive applied voltage sweeps is inadequate. For devices 2 and 3 shown in Fig. 2(b,c), respectively, the current consistently decreases with increasing number of voltage sweeps. However, the current differences for devices 2 and 3 are smaller than that of device 4 shown in Fig. 2(d). For device 4, the maximum current clearly decreases with increasing number of dual voltage sweeps, with the maximum currents at 3 V for five consecutive voltage sweeps being 1.56 × 10 −7 , 8.95 × 10 −8 , 7.25 × 10 −8 , 6.52 × 10 −8 , and 6.10 × 10 −8 A, respectively. The dual voltage sweeps in Fig. 2(d) show clockwise hysteresis, which is a critical feature of an artificial synaptic device. In Fig. 2(e), the dual voltage sweeps for device 5 show that the maximum current at 3 V is decreased and the pinched hysteresis is unstable. Especially, as shown in Fig. 2(a-e), the highest currents in the synaptic devices with different amounts of QDs steadily decrease with increasing amount of CdSe/CdZnS QDs in the active hybrid layer. Furthermore, the I-V data for the dual voltage sweeps with synaptic device 4 show excellent pinched hysteresis behavior similar to that of a biological synapse. Therefore, in the following discussion, we use the data for device 4 for a detailed analysis of the device characteristics.
A more specific operation of the ITO/PEDOT:PSS/PMMA:CdSe/CdZnS QDs/Al synaptic device is shown in Fig. 3. Figure 3 Fig. 3(b). The device's conductance under the positive voltage sweep ( Fig. 3(a)) can be seen to be decreased; however, after the ensuing negative voltage stimulation ( Fig. 3(b)), the original high conductivity can be seen to have been recovered. Similarly, in Fig. 3(c), the current under the consecutive positive voltage sweeps at 3 V decreases from 1.95 × 10 −7 to 7.12 × 10 −9 A. The inset presents the I-V curves under consecutive negative voltages sweeps, and the current at −3 V decreases from 1.44 × 10 −7 to 1.10 × 10 −8 A. When the I-V curves for the first positive-negative voltage stimulation are compared, the maximum currents are found to be similar, indicating that the synaptic device is rewritable. Figure 3(d-f) present data showing the stability of the synaptic device under bending. This flexible synaptic device is clearly stable under bending because the pinched hysteresis behaviors and the electrical characteristics of the bent device are similar to those for the flat device. Figure 4(a) shows a series of voltage stimulations with five consecutive positive voltage sweeps followed by five consecutive negative voltage sweeps being applied to the synaptic device. Set and Reset conditions can be seen in V-t and the I-t data. Thus, one can conclude that the synaptic device is rewritable with repeated switches between low-resistance and high-resistance states. Figure 4(b,c) present the I-t curves when a 100 cycles of pulse voltage (3 V, 100 ms) are applied to the flat and the bent device, respectively. The inset of Fig. 4(b) presents the pulse voltage applied to the electronic synaptic device. As shown in Fig. 4(b,c), the current in the flat device under pulse voltage decreases from 9.03 × 10 −8 to 5.11 × 10 −8 A and that in the bent device decreases from 2.04 × 10 −7 to 9.98 × 10 −8 A. Note that the final current in the bent state is a little higher than that in the flat state. A possible reason is that under bending, the PMMA:CdSe/CdZnS QD layer is stretched. As a result, the thickness of the active layer is reduced, and the distance between neighboring QDs in the vertical direction is reduced, leading to an increase in the current. Figure 5 shows the retention characteristics of the ITO/PEDOT:PSS/PMMA:CdSe/CdZnS QDs/Al synaptic device measured at room temperature and a reading voltage of 0.3 V after one, four, and five consecutive positive voltage sweeps have been applied to the device. As shown in the graphs, the currents at different conductivity states are maintained for up to 10 4 s, indicating that the ITO/PEDOT:PSS/PMMA:CdSe/CdZnS QDs/Al synaptic device has a high stability. These results indicate that the depression of our device can be attributed to a long-term depression behavior. Note that the current drops significantly in 1000 s for the first sweep. A possible reason is   www.nature.com/scientificreports www.nature.com/scientificreports/ that the electrons are more easily captured by the QDs because the QDs have previously only captured a small number of electrons. As a result, the current drops significantly in the beginning. The result is similar to the I-V cure shown in Fig. 2(d), in which the current decreases significantly in the beginning. However, for other cases after the fourth and the fifth positive voltage sweeps have been applied to the device, the capture of electrons by the QDs is more difficult because the QDs have already captured a significant number of electrons.
The I-V characteristics of the synaptic device were studied further to understand the carrier transport mechanisms at work in the device, and the results are presented in Fig. 6. The results obtained by fitting the first sweep of the I-V curves are presented in Fig. 3(a,b). We used the field-assisted thermionic emission model (Eq. 1) and the space-charged-limited-current (SCLC, Eq. 2) model to fit the I-V curves 30,31 : where I, V, A, T, ε, ϕ, k, and q represent the current, applied voltage, Richardson's constant, absolute temperature, dielectric permittivity, barrier height, Boltzmann's constant, and electronic charge, respectively. Figure 6(a) shows the I-V curve when a positive voltage (0-3 V) is applied to the electronic synaptic device. The inset shows the fitted I-V graph under a positive consecutive voltage sweep, and a linear relationship between ln (I) and V 1/2 is seen at voltages from 0 to about 1.6, indicating that the carrier transport in the synaptic device is dominated by thermionic emission in region I of the graph in Fig. 6(a). On the other hand, for region II of the graph in Fig. 6(a), as shown in Fig. 6(b), the slope of ln (I) versus ln (V) positive voltage curve for voltages from 1.7 to 3 V is 9.88, indicating that the carrier transport in the synaptic device is dominated by SCLC conduction,  www.nature.com/scientificreports www.nature.com/scientificreports/ which results from the electron capture by QDs. Figure 6(c,d) show the fitted graphs under negative consecutive voltage (0-3 V) for the first sweeps. The linear relationship between ln (I) and V 1/2 at voltages from 0 to −1.6 V (region I), as seen in the inset of Fig. 6(c), and the linear relationship between ln (I) and ln (V) at voltages from −1.7 to −3.0 V (region II), as seen in Fig. 6(d), indicate that the mechanisms are the same as those in Fig. 6(a,b).

Conclusion
Flexible electronic synaptic devices based on PMMA:CdSe/CdZnS QD nanocomposites were fabricated. We observed that the current differences in the devices depended significantly on the CdSe/CdZnS QDs concentration in the nanocomposites. The I-V characteristics of the synaptic devices showed clockwise pinched hysteresis behavior similar to a biological synapse. Furthermore, the synaptic devices exhibited excellent bending stability because the electrical characteristics after bending were similar to those for the devices in the flat state. The carrier transport mechanisms in the synaptic devices were dominated by thermionic emission and SCLC conduction.