1. Introduction

Short-wavelength semiconductor lasers have raised considerable attention due to the potential applications in high capacity data storage, display, and lighting [1]. ZnO, as a wide band gap (E_g = 3.37 eV) semiconductor material with a high exciton-binding energy of 60 meV, shows promising prospects in realizing the low-threshold laser at room temperature [2]. The lasing mechanisms of ZnO-based nanostructures can be mainly classified into two catalogues, i.e. the cavity-lasing and the random-lasing [3]. The random-lasing (RL) is particularly attractive compared to the cavity-lasing due to its board angular distribution that are suitable for lighting and display application, as well as its easy realization [4]. Since the discovery of the optically pumped RL in zinc oxide polycrystalline films [5], much attention has been focused on the investigation of ZnO-based RL [6, 7]. Compared with the optically pumped RL, the electrically pumped lasing is more desirable. In 2006, Leong et al. [8] demonstrated the first electrically pumped RL in heterostructural p–i–n junctions using patterned ZnO–SiO₂ nanocomposite film as the light-emitting layer. Later, Ma et al. presented a new strategy to realize electrically pumped RL through a Au/SiO_x/ZnO metal-insulator-semiconductor (MIS) structure on Si substrate [9]. Regarding the formation mechanism of RL in MIS structure, it is generally believed that the recurrent scattering of light among the grain boundaries of the ZnO polycrystalline film accounts for it [4, 9]. As a result, the light scattering capability of the ZnO layer, which is determined by its morphology, is crucial in realizing RL. For example, Tian et al. [10] have demonstrated that the light scattering in sol-gel derived ZnO film is better than the sputtered one, which is favorable for the formation of electrically pumped RL. As a viable alternative to realize the RL action, however, the insulating layer may also be designed properly with certain morphology to function as the light confining and scattering cavities.

Herein, the electrically pumped RL has been realized from FTO/porous insulator/n-ZnO/p⁺-Si devices. The porous morphology of the insulating layer was proved to account for the RL, which presents a new strategy to realize the electrically pumped RL in MIS devices. In addition, our devices show a remarkable negative differential resistance (NDR) effect under sufficiently high forward bias following a normal I-V characteristic. Although the NDR effect was already observed in some cases [11], the mechanism still remains an open issue. We think the glow discharge is responsible for the observed NDR effect in our devices.

2. Experiment

The ZnO films were fabricated on 1cm*1cm sized p⁺-Si (111) substrates via a simple hydrothermal method. Prior to the growth of ZnO films, thin ZnO seed layers with a thickness of around 30 nm were pre-deposited onto the substrates by magnetron sputtering. Then, the substrates were annealed at 550 °C in air for 1 h to improve the crystallinity of the seed layers. Afterwards, the substrates were put into a glass bottle filled with solution of zinc nitrate hexahydrate (Zn(NO₃)₂•6H₂O) and methenamine (C₆H₁₂N₄) in equal concentration of 0.05 M or 0.1 M. The reaction was kept at 95 °C for 2 h in an oven. After growth, the samples were rinsed with deionized water and dried. Finally, they were annealed at 400 °C in air for 1 h to promote the crystallization of the ZnO film.

The structure of the light-emitting devices is schematically illustrated in Fig. 1(a). Firstly, 150 nm thick Al electrode was deposited on the back of the Si substrates by magnetron sputtering as the bottom electrode, and the samples were rapid-annealed at 500 °C in N₂ atmosphere for 1 min to obtain ohmic contact between Al and Si. Then, a layer of transparent insulating material (Poly(methyl methacrylate) (PMMA) or spin on glass (SOG)) with a thickness of around 50 nm was spin-coated on the surface of the ZnO film. Subsequently, the samples were either annealed at 200 °C for 30 min (in the case of PMMA) or at 500 °C for 1 h (in the case of SOG) in air to solidify the insulating layer. At last, a piece of FTO glass was directly contacted with the top surface of the sample as the top transparent electrode.

 

Fig. 1 Schematic of the FTO/insulator/n-ZnO/p⁺-Si device. (b) Room temperature PL spectrum of the as-grown ZnO film. The inset shows the top-view SEM image of the ZnO film grown with the precursor concentration of 0.05 M. The scale bar is 1 μm.

Download Full Size | PDF

The morphology of the samples was characterized by scanning electron microscopy (SEM, KYKY-3200) and field-emission SEM (FESEM, Hitachi S-4800). The photoluminescence (PL) and electroluminescence (EL) measurements were carried out on a spectrometer (Edinburgh Instruments, FLS 920). For the PL measurement, a Xe lamp with the exciting wavelength of 300 nm was used as the excitation source. To acquire the EL spectra, the devices were applied with forward bias utilizing a Keithley 2400 source meter, with the positive voltage connected to the bottom Al electrode.

3. Results and discussions

The typical top-view SEM image of the as-grown ZnO film with the precursor concentration of 0.05 M is shown in the inset of Fig. 1(b). The film is composed of discrete ZnO nanocolumns with the diameter of around 200 nm and the height of 1.5 μm. Figure 1(b) shows the typical PL spectrum of the ZnO film. It is composed of three emission peaks, i.e. the ultraviolet (UV), blue and orange emission peaks. The UV emission can be attributed to the near-band-edge emission, while the orange emission may originate from the oxygen interstitial defect that is commonly observed in hydrothermal grown ZnO samples [12]. The origin of the blue peak is not clear yet, and it has also been observed in the PL spectra of ZnO nanoparticles grown by pulsed laser irradiation in solution [13]. Considering that we do not find this peak in the thermal evaporation grown ZnO nanowires that we reported before [14], we speculate that it might also be related with the defect induced by the oxygen-rich hydrothermal growth environment.

The EL spectra of the ZnO film with PMMA as the insulating layer are shown in Fig. 2. At low bias voltage, the spectrum is dominated by the yellow peak centered at round 575 nm, while the intensity of UV peak at 406 nm is relatively weak. We believe that this can be attributed to the spontaneous emission at the interface of the n-ZnO/p-Si heterojunction [15, 16]. It is noted that the EL spectra of our device are clearly different from its PL spectrum, which is due to the fact that EL is an interfacial process while PL probes the bulk property of the material [16]. With the further increase of the applied voltage, however, there is a sudden decrease of the current from above 200 mA to about 3 mA, nearly two orders of magnitude. At the same time, discrete sharp peaks with very narrow line-width (less than 0.1 nm) begin to emerge in the spectra. The spectra are very similar to RL, but considering that the spontaneous emission should also be observable in addition to the stimulated emission for a typical RL, they can hardly be regarded as RL. Moreover, the life-time of the emission is quite short. Its intensity will drop over an order of magnitude in a few seconds. Therefore, they are more like glow discharge spectra [17], and the magnitude of the current (several mA) is consistent with that of glow discharge. We have also changed the insulating layer from PMMA to SOG, and observed similar phenomenon. Hence, the glow discharge instead of the commonly observed RL occurred in our MIS devices.

 

Fig. 2 Room temperature EL spectra of the FTO/insulator/n-ZnO/p⁺-Si devices under different forward bias voltages with PMMA as the insulating layer.

Download Full Size | PDF

During the EL measurement, we observed the formation of a layer of white substance on FTO at the same time of the current drop. When the FTO glass was detached from the sample, it could be clearly seen that the white substance firmly attached on both the FTO glass and the sample. The typical top-view SEM images of the FTO glass and SiO_x-coated ZnO film before the EL measurement are shown in Figs. 3(a) and 3(b), respectively. Figures 3(c) and 3(d) are the top-view SEM images of the white substance on both the FTO glass and the ZnO film after the glow discharge when SOG was used as the insulating layer. It clearly shows that the white substance forms porous pattern on the FTO glass, and the feature size between walls is about several micrometers as shown in Fig. 3(c). In addition, Fig. 3(d) shows that the thin SiO_x(x<2) layer we have coated on the ZnO film also has changed to porous and rough structure after the glow discharge. Hence, we can conclude that the white pattern on the FTO glass was formed during the EL measurement. Considering the glow discharge phenomenon we have observed, the SiO_x porous pattern was most probably formed due to the sputtering of SiO_x onto the FTO glass from the insulating layer during the discharge process. Interestingly, the obtained SiO_x pattern was porous with numerous random cavities, which may formed due to the fact that the ZnO film we used was composed of discrete ZnO nanocolumns with high surface roughness, and the protrusion parts are more likely to be sputtered onto the FTO. Due to the unique morphology of the porous pattern, it is possible to use it as the random resonant cavities to realize RL.

 

Fig. 3 Typical top-view SEM images of (a) FTO glass and (b) SiO_x-coated ZnO film before the glow discharge, and of the SiO_x patterns attached on (c) the FTO glass and (d) the surface of the ZnO film after the glow discharge. The inset of (c) is an enlarged version of the same sample.

Download Full Size | PDF

To testify the above assumption, we prepared the lighting-emitting devices using the PMMA or SiO_x porous pattern attached FTO glass, which was formed during the previous EL measurements, to contact the ZnO films, and their EL property is investigated. Besides, device with bare FTO glass was also tested as a reference. In order to prevent the direct contact between ZnO nanocolumns and FTO through the porous insulating pattern, we used the ZnO film fabricated with the precursor concentration of 0.1 M in this case, which is composed of ZnO nanocolumns that merge together with each other to form a continuous film, and the corresponding plan-view SEM image is shown in the inset of Fig. 4(a). Figure 4(a) shows the EL spectra of the device when the bare FTO glass was used. The spectra are quite analogous with those shown in Fig. 2 under low bias voltage, and no RL could be observed with the increase of the driving voltage and current. This might be due to the fact that the ZnO layer used in this case is a continuous film with few grain boundaries, therefore its light scattering capacity is not sufficient enough to induce the RL. However, when we put exactly the same sample on the PMMA porous pattern attached FTO glass and measured its EL spectra, we observed discrete sharp peaks emerge in the UV area of the spectra at high bias voltage (Fig. 4(b)). In contrast to the situation shown in Fig. 2, there is no sudden current drop in this case. Besides, the spontaneous emission and the sharp peaks appear simultaneously, which suggests the observed spectra are RL instead of glow discharge. Moreover, we find that the position of the emission peaks varies from different runs of scanning during the EL measurements, which further verifies RL action. In the case of using SiO_x porous patternattached FTO glass, the device behaved in a similar way and the corresponding EL spectra are shown in Fig. 4(c). Moreover, we also put the SiO_x coated ZnO film onto the SiO_x porous pattern attached FTO glass and the corresponding EL spectra are shown in Fig. 4(d). It can be seen that the spectra also exhibit similar RL characteristics as those shown in Figs. 4(b) and 4(c). These results validate that the random cavities within the insulating pattern play an important role in the formation of RL.

 

Fig. 4 Room temperature EL spectra of the ZnO-based light-emitting devices under different forward bias voltages. The devices are prepared by placing bare ZnO film directly on (a) bare FTO glass, (b) PMMA porous pattern attached FTO glass, (c) SiO_x porous pattern attached FTO glass, and (d) by placing the SiO_x coated ZnO film on SiO_x porous pattern attached FTO glass. The inset of (a) is the plane-view SEM image of the ZnO film grown with the precursor concentration of 0.1 M. The scale bar is 1 μm.

Download Full Size | PDF

Based on the above experimental results, the whole light-emitting process of our devices can be understood according to the energy band alignment, as is shown in Fig. 5(a). Under the forward bias condition, the holes are injected from the p⁺-Si substrate into the ZnO layer, while the electron are mainly injected into the ZnO layer via the impact ionization process in the insulating layer under high electric field [18]. When the bias voltage is low, due to the large valance band offset between p⁺-Si and ZnO, the energy barrier is too large for the holes to be injected into the valance band of ZnO, Therefore, the holes are mainly injected into the defect energy level in the band gap of ZnO, and combine with the electrons in the conduction band to emit the visible light. With the increase of the bias voltage to a certain value, the holes possess enough energy to overcome the valance band offset between p⁺-Si and ZnO and can also be injected into the valance band of ZnO. In addition, the energy band of ZnO will severely bend upward at the ZnO/insulating layer interface, which results in a hole accumulation layer there. So the holes and electrons will combine at this interface leading to the UV emission. The UV emission can be partially confined in the closed-loop random cavities within the insulating layer through recurrent scattering and interference [9]. When the generation rate of photon with certain frequency overcomes its loss during the random walk process due to the scattering, lasing oscillation occurs and results in the sharp RL peaks in the EL spectra with high intensity. This process is schematically illustrated in Fig. 5(b). In previous study, the occurrence of RL was attributed to the light scattering among the ZnO grain boundaries that formed the closed-loop [4, 9]. In this case, however, it is the porous insulating layer that behaves as the random resonator cavities to facilitate the formation of RL, since RL could be observed only after the insulating layer became porous.

 

Fig. 5 (a) Schematic energy band alignment of the FTO/insulator/n-ZnO/p⁺-Si device under forward bias. (b) Schematic illustrates the formation mechanism of RL.

Download Full Size | PDF

The results point out that the RL can be realized in MIS structure by simply changing the morphology of the insulating layer. In fact, it is expected that the porous insulating pattern can be intentionally fabricated using the common photolithography method to replace the solid insulating film, which may become a viable manner to realize RL in the future.

4. Conclusion

In conclusion, electrically pumped RL has been realized in FTO/porous insulator/n-ZnO/p⁺-Si devices. It is believed that RL originates from the confining and recurrent scattering of light in the closed-loop random cavities within the insulating layer, which are formed due to the glow discharge. The glow discharge is also thought to be responsible for the observed NDR effect. The results here present a new strategy to realize the electrically pumped UV RL, and may provide more insights into its formation mechanism.