Organic Thin-Film Transistor Memory with Nanocrystal Carbon Dots

Organic thin-film transistor (OTFT) memory devices were fabricated with nanocrystal carbon (nc-C) dots incorporated into the pentacene/oxide interface in the active layer. The nc-C dots were arranged precisely in order on the OTFT channel region by a focused ion beam (FIB) technique using low-energy Ga ions and phenanthrene as the carbon source. The structural information of nc-C dot arrays was obtained by scanning ion microscopy (SIM) and atomic force microscopy (AFM). These images indicate that the nc-C dot array was successfully formed on the oxide layer. The density of the two-dimensional nc-C dots was 1.3-1.5×10 cm−2. The current-voltage (I-V ) characteristics showed that the OTFTs exhibit memory behavior upon the application of forward and reverse gate bias stresses. Depending on the polarity of gate bias, write and erase modes were induced, and a maximum threshold voltage shift ∆Vth of 0.97 V was obtained. [DOI: 10.1380/ejssnt.2010.250]


I. INTRODUCTION
High-performance devices based on organic semiconductors have been rapidly developed in recent years on the basis of the technology of silicon ultralarge-scale integration (ULSI) circuits.Among the most organic devices, organic thin-film transistors (OTFTs) have great potential for application in radio-frequency identification tags, electric papers and flexible transparent display devices [1].In previous reports, nonvolatile memory devices with nanocrystal Si, Ge and metal particles embedded into the oxide layer by several techniques, such as chemical vapor deposition (CVD) [2,3], ion implantation [4] and co-sputtering [5], have been proposed.We propose here an OTFT embedded with nanocrystal carbon (nc-C) for application to the next generation of nonvolatile memory on flexible display devices.Organic nonvolatile memory must be operated at a higher output current (>1 mA) and lower voltage (<10 V) and with a short write/erase time (<1 ms) comparable to that of current floating-gate type memories.Among several methods, the focused ion beam (FIB) is also a powerful tool for forming nanostructure in OTFTs [6,7].
Previously, we showed metal-oxide-semiconductor (MOS) devices with nc-C dots embedded on an oxide layer by the FIB technique [8].The memory effects of these devices were demonstrated to be due to the underlying the charge storage such as the quantum confinement effect and Coulomb charge effect, which prevented the charge from escaping [9].These mechanisms led to the development of nanocrystal-based transistors consisting of an organic material as the active or semiconductor layer for use in memory device applications for future display devices.
In this study, we formed nc-C dot arrays in the channel region of OTFTs based on pentacene, by FIB-CVD.Then, we measured the electrical characteristics of OTFTs with a bottom contact configuration embedded with nc-C dots in the pentacene/oxide interface in the active layer to achieve memory effects.Pentacene is one of the most promising material for OTFTs owing to its extremely high carrier mobility and atmospheric stability, and its use is expected to make organic devices competitive with standard amorphous silicon technology [10].The experimental evidence for the effect of traps on the charge storage characteristics in nc-C is presented.

II. EXPERIMENTAL
The OTFTs with a bottom source/drain contact configuration, as shown in Fig. 1, were fabricated as follows.The starting material was Sb-doped n + -type (<0.1 Ωcm), Czochralski (CZ)-grown 4-inch Si(100) wafers, cut into 1.5 cm×1.5 cm square chips.These chips were rinsed in deionized water, ethyl alcohol, acetone, and then methyl alcohol, and were cleaned by a standard RCA method.After cleaning, the samples were treated with dry oxygen ambient at 1000 • C for 20 min to form 40-nm-thick SiO 2 film.After that, a Au electrode was formed by vacuum deposition and photolithography methods on SiO 2 as the source/drain.The nc-C dots were embedded be- tween the defined source and drain regions (in the channel region) by FIB-CVD.As shown in Fig. 2, FIB-CVD was carried out using a Ga + ion beam operated at 30 kV.The source gas phenanthrene (C 14 H 10 ) was released from the gas gun, and the C 14 H 10 molecules were adsorbed on the substrate.The adsorbed molecules were dissociated by Ga + FIB irradiation, and carbon dots could be deposited.Raman spectroscopy measurement indicated that the amorphous carbon deposited using C 14 H 10 was diamond-like carbon [8].Finally, a 50-nm-thick pentacene layer was evaporated onto the top of the nc-C dots and Au electrodes by vacuum evaporation at a deposition rate of 4 Å/s under a pressure of 2.0×10 −6 Torr.A sample of OTFT without the nc-C dots was also prepared for comparison.Figure 3 shows the view of the pattern for the source/drain electrode with the defined channel length L = 20 µm and channel width W = 20 µm.
The nc-C dots on the SiO 2 film were observed directly by a scanning ion microscopy (SIM) with the FIB system.The morphology of the nc-C dots in the channel region was investigated by atomic force microscopy (AFM).Finally, the electrical characteristics of the OTFTs were measured in a shield box under atmospheric pressure at room temperature.

III. RESULTS AND DISCUSSION
The distribution of nc-C dots in the channel region was observed by SIM, as shown in Figs.4(a) and (b).The formation of nc-C dot arrays in the channel region was confirmed by top-view SIM with the FIB apparatus.We fabricated three types of OTFTs: without nc-C dots, and with dots distributed throughout the entire region (Fig. 4(a)) and in a half-region near the drain side (Fig. 4(b)).The two-dimensional density of the dots was 1.3×10 8 (half-region) and 1.5×10 8 cm −2 (whole region).The detailed morphology of nc-C dots in the channel region were observed using an AFM image.As shown in Fig. 4(c), the AFM image showed that the nc-C dots adhered to the SiO 2 layer and were uniformly distributed at regular intervals.The lateral size of the dots was approximately 100 nm and the height was about 10 nm.Thus, the dot was growing laterally owing to the wide scattering of C 14 H 10 molecules upon impact by Ga + ions.However, their height was kept to nanometer size.It is also considered that the interval between deposition and refresh times is important to control the nc-C dot shape, but is not fully optimum in this experiment.The size and spacing of dots are two of the parameters because these affect the performance of nonvolatile characteristics such as endurance and storage time [11].In our experiment, the spacing was set to 50 nm.Thus, the charge stored might remain in individual nc-C dots without carrier hopping from one dot to neighboring one.
Figure 5(a) shows the drain current-drain voltage (I ds -V d ) characteristics as a function of gate voltage (V g ) of the OTFTs without nc-C dots.The I ds -V d curves were obtained by sweeping V d from 0 to −10 V.With increasing negative gate bias (V g ), I ds increases.Thus, the device shows a p-type operation, i.e., holes are injected from the source electrode into the pentacene layer.The carrier mobility and on/off ratio were 0.1 cm 2 /Vs and 10 3 , respectively.The subthreshold characteristics, I 1/2 ds -V g , for the OTFT are shown in Fig. 5(b) before and after 5 min at charge injection into the channel at a constant applied voltage.The OTFT showed an enhancement mode operation, and no shift of the threshold voltage itself was observed.Thus, the holes injected are not effectively trapped at the pentacene/oxide interface.
Figure 6 shows the I ds -V d and I 1/2 ds -V g characteristics for the OTFTs with nc-C dots on the entire region in the channel area.Similar behavior as the I ds -V d characteristics in Fig. 5(a) was observed on the OTFT with nc-C dots in the initial state.The mobility of the OTFT with nc-C dots is almost the same as that of the OTFT without dots.In Fig. 6(b), the threshold voltage (V th ) of the OTFT before charge injection was −2.72 V. Here, V th was defined as the voltage at which the linear fit of I 1/2 ds -V g intercepts the gate voltage axis.Next, a forward bias gate voltage V g of +3 V is applied for 5 min to inject hot holes into nc-C dots (write operation).After injection, V th changed to −1.75 V, as shown in Fig. 6(b).Then, when reverse bias of V g = +10 V is applied for 5 min (erase operation), saturation drain current decreases and V th shifts again to −2.72 V.After charge injection, it is considered that a small number of hot holes injected accumulate into the potential well of the pentacene/nc-dot/oxide structure; accordingly, the inversion carrier density becomes modified and V th shift occurs.When reverse bias is applied, in contrast, charge emission from the nc-C dots arises.As a result, V th shifts again to the initial level.The maximum threshold voltage shift (∆V th ) was 0.97 V toward lower voltage after hole injection.For the emission process, the voltage shifts again to the initial level.Thus, the recovery process is subsequently fully completed after reverse bias is applied.Owing to the potential around the charge captured by the nc-C dots, the channel becomes difficult to form, which causes the shift in current flow, i.e., shift of V th .The total charge per dot, Q dot , can be estimated as follows: where C ox is the capacitance of SiO 2 , q is the elemental charge, and N dot is the 2-dimensional density.For the OTFT with nc-C dots in the whole region, the total number of charges per nanodot, n is estimated to be 3437.Previous reports showed the V th shift of 0.5-0.65 V for the dot density of 10  the same V th shift (0.76-0.9 V) is obtained in the case of 10 8 cm −2 .The difference is originated from the fact that in our device the discharge time for the dots charged is longer as compared to previous devices owing to relatively large spacing of 50 nm between the dots.
Figure 7 shows the I ds -V d and I 1/2 ds -V g characteristics for the OTFTs with nc-C dots distributed in the half-region near the drain side.It is expected that large charge trapping efficiency is obtained in the device of the dots distributed in the half region near the drain side if the electric field is localized near the drain edge and hot carriers are injected effectively in the nc-dots.
The behavior of the I ds -V d characteristics in the initial state is similar to those in Figs.5(a) and 6(a).In Fig. 7(b), V th of the OTFT before charge injection was −2.53 V. First, forward bias gate voltage V g = +3 V is applied for 5 min to inject holes into nc-C dots (write operation).V th changes to −1.77 V.Then, reverse bias of V g = +10 V is applied for 5 min (erase operation) and saturation drain current decreases, indicating charge emission from the nc-C dots.V th shifts again to −2.53 V.This corresponds to a threshold voltage shift (∆V th ) of 0.76 V toward lower voltage after hole injection into nc-C dots.∆V th of 0.76 V is slight smaller than the value of 0.97 V for the OTFT with nc-C dots in the whole region.This difference may be due not to the efficiency of hole trapping, but to the difference in the 2-dimensional density between them.Typical writing time is 50 s and no marked difference is observed in the configuration of the dots.After the emission process, the threshold voltage shifts again to the initial level.Erasing time is typically 80 s.First, V th shifts rapidly and then saturates to the initial level above 80 s.Thus, the recovery process is subsequently confirmed when forward bias is applied.Using eqs.( 1) and ( 2), the charge stored per dot is calculated to be 3078 for the memory window of 0.76 V.When forward bias is applied to the gate, the charges are efficiently injected into the pentacene layer from the Au electrode owing to the Au work function, which is close to the highest occupied molecular (HOMO) level of pentacene, thus creating an abundance of charges in the pentacene layer.This promotes the injection of charges from the pentacene layer to the nc-C dots and leads to the accumulation of charges in nc-C dots or a charge-trapping effect [11].When reverse bias is applied to the gate, the charges accumulated in the nc-C dots are ejected back again from the pentacene layer, and the threshold voltage returns to the initial state.Therefore, this indicates that for these write and erase operations, the accumulation of charges in the nc-C dots is responsible for the shift of threshold voltage upon the application of gate bias.

IV. CONCLUSIONS
In this study, we successfully fabricated an OTFT memory with nc-C dots incorporated into the pentacene/oxide interface by the FIB-CVD technique.The formation of nc-C dot arrays in the channel region was confirmed by SIM and AFM, respectively.The electrical characteristics exhibited a threshold voltage shift corresponding to a change in the total number of charges retained in the nc-C dots at the pentacene/oxide interface, which confirmed the occurrence of a memory effect.

FIG. 3 :
FIG. 3: Top view of pattern with source (left side) and drain (right side) electrodes on the gate oxide film.

FIG. 4 :FIG. 5 :
FIG. 4: SIM images of nc-C dots on the whole channel region (a) and nc-C dots on a half-region (b).AFM image of nc-C dots embedded in the channel region (c).

FIG. 6 :
FIG. 6: (a) Output (I ds -V d ) characteristics of OTFT with nc-C dots in whole of channel region.(b) Subthreshold (I 1/2 ds -Vg) characteristics after reverse and forward bias stresses.

FIG. 7 :
FIG. 7: (a) Output (I ds -V d ) characteristics of OTFT with nc-C dots in half-region near the drain side.(b) Subthreshold (I 1/2 ds -Vg) characteristics after reverse and forward bias stresses.