Taking into consideration the sequence of thermal treatment process and the excellent performance the memory exhibits, the configuration of device in this work is designed into bottom-gate, top-contact p-channel ONFGM as shown in Fig. 1. In this cross-sectional structure model, pentacene with high hole-mobility^[19,20] is covered on the tunneling insulator as an active layer and polysilicon is inserted into the oxide layer as a charge storage medium. Gold with a high workfunction is used as the source/drain contact which is propitious to injection and extraction of holes between the metal electrode and the organic semiconductor channel. The control gate directly contacts the heavily doped n-type silicon. The detailed physical and structural parameters are listed in Table 1.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. Schematic cross section of ONFGM based on pentacene with a bottom-gate, top-contact structure.

Table 1.

Table 1.

Table 1. Physical and structural parameters used for device simulation. . Parameter Symbol Value Channel length L/μm 10, 20, 30 Channel width W/mm 2 Thickness of source and drain electrode (Au) ST, DT/nm 40 Thickness of organic semiconductor TS/nm 60 Thickness of tunneling insulator TT/nm 10 Thickness of floating gate TF/nm 20 Thickness of gate insulator TOX/nm 300 Dielectric constant of gate insulator (SiO2) εs 3.9 Dielectric constant of organic semiconductor 3 Schottky barrier height φB/eV 0.1 ∼ 0.3 Band gap of organic semiconductor Eg/eV 1.8 ∼ 2.2 Work function (Au) WF/eV 5.10 ∼ 5.47 Impurity concentration of gate region n+/cm−3 1 × 1019 Lowest unoccupied molecular orbital level of pentacene LUMO/eV –3.2 Highest occupied molecular orbital level of pentacene HOMO/eV –5.0

Table 1. Physical and structural parameters used for device simulation. .

The geometry of the ONFGM device resembles that of OTFT except the tunneling insulator and floating gate that are sandwiched between the organic semiconductor and the gate insulator. Based on the investigation of p-channel OTFTs,^[21,22] the charge carriers transporting mechanism is strongly dependent on the temperature, electric field and holes traps between the organic semiconductor channel, electrode and gate dielectric layer.^[23,24] Therefore, the simulations for the charge carrier transport of the ONFGM device are carried out by solving Poisson’s equation, continuity equation and holes drift-diffusion equation, which are described as follows:^[25–28]

In this model, μ_p0 is the low field mobility and E_h0 represent the activation energy for hole (E = 0) which is equal to 0.018 eV, β is the hole Poole–Frenkel factor, and γ is the fitting parameter. For obtaining prominent simulation results, the contact between electrode and organic semiconductor are regarded as a Schottky barrier (0.1 eV ∼ 0.3 eV). Moreover, in an organic semiconductor material, the effective density of states at the lowest unoccupied molecular orbital (LUMO) level is considered to be the same as that at the highest occupied molecular orbital (HOMO) level.^[30,31] In addition, the hot carrier injection and Fowler–Nordheim (FN) tunneling model are utilized for simulating holes injected into or extracted from the floating gate during P/E operation. The thermal emission current J and tunneling current J_FN are respectively given as follows:^[32,33]

Based on the models above, the electrical characteristics of the ONFGM device are simulated and calculated reasonably through Newton’s iteration method. Figure 2 shows the flow chart of numerical simulation.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. Flow chart of numerical simulation.

The simulations of the output and transfer characteristics of our present ONFGM device are performed, and the simulated results indicate that the device features typical p-channel field-effect behavior in hole-accumulation mode during V_DS ranging from 0 V to −20 V and V_GS from 0 V to −20 V in steps of −5 V as shown in Fig. 3(a). In comparison with the experimental data,^[34] the simulated results of output curves are in the appropriate order of magnitude and the current values are superior to that of Au-nanocrystal memory within a certain voltage range. Figure 3(b) shows the simulated transfer curves for the channel length (L) ranging from 10 μm to 30 μm. The regulation effect of gate on channel current increases with channel length decreasing from 30 μm to 10 μm. The inset shows the behavior of a small fraction of holes accumulating at the channel in the case of low-ranging V_GS which is possibly attributed to the induced weak electric field of trapped charge carriers generated from the floating gate. This behavior is in good agreement with that in the literature.^[35]

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. (a) Output characteristic curves of ONFGM with a channel length of 10 μm. (b) Comparison of the transfer characteristic curves with different channel lengths. The inset shows the transfer curves at low gate voltages.

According to the above analysis, by employing the hot hole injection model and the FN tunneling model, the (P/E) characteristics of the present ONFGM are simulated and the effects associated with shifts in transfer curves are examined through a series of P/E operations in a sequence of programing voltages (V_P) and erasing voltages (V_E), respectively. For all of the P/E operations, the drain-source voltage (V_DS) is maintained at 0 V and the P/E time is set to be 5 s, holes are majority charge carriers in the present ONFGM which is operated in the p-channel as mentioned in Subsection 3.1.

Figure 4(a) shows the energy level of the present ONFGM device, where E₀ is the vacuum energy level, LUMO and HOMO are the lowest unoccupied molecular orbital level and the highest occupied molecular orbital level of pentacene, which are −3.2 eV and −5.0 eV, respectively.^[36] During programing operations, a high negative voltage is applied to the gate, holes accumulate at the interface between channel and SiO₂, and the energy band of the ultrathin SiO₂ layer is extremely bended so that holes are injected from the channel into the poly-Si floating gate (Fig. 4(b)). Once holes are injected in the floating gate, a shift towards the threshold voltage (ΔV_T) will be induced as depicted in Fig. 5(a).

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (a) Energy level diagram of ONFGM without applying voltage, (b) energy band diagrams of ONFGM during programming, and (c) erasing operations.

Fig. 5.

Figure Option ViewDownloadNew Window

Fig. 5. (a) Transfer curves of the ONFGM device in the initial and a series of post-P/E operations at various of VP/VE voltages in a time of 5 s. The inset shows the shifts in transfer curves during P operations at VGS ranging from −5 V to −30 V. (b) Corresponding variations of charge stored in the poly-Si floating gate in logarithmic coordinates and the inset in linear coordinates during P operation and (c) E operation. (d) Memory windows for various VP and VE voltages.

It is clearly revealed that the transfer curves shift rapidly in the negative direction with the increase of V_P from −45 V to −60 V, and then more slowly as V_P increases from −60 V to −70 V. To further confirm that the shift effect of the transfer curve is attributed to the trapped holes in the floating gate, the variations of poly-Si floating gate charge are simulated as shown in Fig. 5(b) (logarithmic coordinates). The variations of poly-Si floating gate charge show that the holes injected from the channel to the poly-Si floating gate first increase rapidly with the increase of V_P, and then gradually increase slowly as V_P increases from −60 V to −70 V which is consistent with the shifted effect of the transfer curve. This also demonstrates that the charge stored in the floating gate will not increase even though the programing time is very long. Based on the equation Δn = ΔQ/e (where Δn is the number of transferred charge carriers, ΔQ is the total charge quantity of transferred charge carriers and e is the elementary charge), the number of transferred holes is ≈ 10¹¹ which is slightly inferior to that of CoFe₂O₄ NPs-structure memory in Ref. [17].

After a series of programing operations, a series of erasing operations are performed. As depicted in Fig. 5(a), it is obviously observed that the transfer curves shift towards the positive direction rather than return to the initial state with the V_E increasing from 45 V to 70 V. Moreover, after erasing operations, the values of the shifts in transfer curves towards the positive direction are larger than towards the negative direction in the case of post-programing operations. These phenomena also result from the variations of charge carriers stored in the poly-Si floating gate. As depicted in Fig. 5(c), holes are gradually extracted from the poly-Si floating gate to the channel, which induces a shift of transfer curves towards the positive values. It is notable that the value of charge stored in the poly-Si floating gate decreases to zero with the erasing time increasing from ×10⁻⁵ s to 1×10⁻⁴ s, which suggests that the transfer curves should return to the initial state. However, it is interesting that the value of charges changes into a negative value with the increase of erasing time, by which the transfer curves continually shift towards the positive direction. Owing to a high barrier height of 1.9 (eV) between the LUMO of pentacene (3.2 eV) and the work function of Au (5.1 eV), electrons are limited to be injected into the channel.^[37] In this case, the holes stored in the floating-gate are first ejected from the floating-gate to the channel at large positive gate voltage, and then electrons at the interface between organic semiconductor and SiO₂ layer transfer from the LUMO (lowest unoccupied molecular orbital) level of pentacene to the conduction band of the polysilicon floating-gate by tunneling through a thin SiO₂ layer. Although the electrons are minority carriers in the p-type pentacene active layer, the energy barrier (2.25 eV) for electron transport between the LUMO level of pentacene and the conduction band of SiO₂ is lower than that (4.85 eV) for holes transport between the HOMO (highest occupied molecular orbital) level and the valence band of SiO₂. Therefore, the above behavior is possibly explained by the fact that the electrons at the interface between the organic semiconductor and the SiO₂ layer are injected into the channel by high positive V_E. Moreover, due to the fact that the barrier (4.68 eV) between the poly-Si floating gate and the SiO₂ tunneling insulator in the case of erasing operations is slightly lower than the barrier (4.85 eV)^[38] between the HOMO level of pentacene and the poly-Si floating gate in the case of programing operations, it is easier for holes to be ejected from the poly-Si floating gate to the channel during erasing operations than from the channel to the poly-Si floating gate during programing operations. In contrast, the obvious shifts of transfer curves towards the positive and negative direction by P/E operations are matched correctly with those in the literature.^[39] As a result, a large memory window (ΔV_T) of 57.5 V is obtained at V_P/V_E of ±70 V in a time of 5 s. Figure 5(d) shows the values of V_T at different V_P/V_E voltages.

In the present ONFGM device, the drain current responses for program, read, erase, and read cycles, which are called dynamic memory switching behaviors, are also simulated at sequential and repeated gate voltages of −60, −20, 60, and −20 V in an interval time of 5 s as shown in Fig. 6(a). Meanwhile, the ON/OFF memory current ratio is calculated to be about 10³ as shown in Fig. 6(b).

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (a) Switching behavior for the ONFGM device with poly-Si floating gate (E: erase, P: program, ROn: read-on state, and ROff: read-off state), and (b) the drain currents of ON state and OFF state after erase and program.