1. Introduction

The motivation that realizing monolithic optoelectronic integrated circuits has spurred tremendous efforts to explore an efficient Si-based light source operating at room temperature [1–5]. Since the electron-hole interaction can be much intensified in low-dimensional structure, Si nanostructure, regarded as promising luminescent materials to overcome the insufficient light efficiency from bulk Si, has greatly stimulated theoretical and experimental studies [2–4,6,7]. So far, various methods have been used to fabricate Si nanocrystals with well-controlled size below 5 nm, and it has been demonstrated that the high density of Si nanocrystals with good surface passivation play a crucial role in enhancing PL efficiency as well as improving the emission stability [8]. Owing to possessing the low diffusivity of Si atoms, silicon nitride is always subjected to a thermal irradiation process at high temperatures in an attempt to fabricate high density of smaller Si nanostructure [9]. On the other hand, as one of the available wide-bandgap silicon compounds, silicon nitride has a low potential barrier for carrier injection. Therefore, SiN-based LEDs have been extensively studied by fabricating a highly efficient SiNx luminescent layer, ameliorating device structure as well as reducing the tunneling barrier with the aim of improving the electroluminescence (EL) quantum efficiency [10–13]. Among the different approaches to obtain SiN-based LED, multilayer structure is the one that has attracted the most attention to date due to its advantageous for fabricating size-controllable Si nanocrystals and improving radiative recombination efficiency as well as better control carrier injection through the structure [1,14–16]. In the previous work, we systematically studied the effects of potential barrier layers on the EL characteristics of SiN-based multilayer LEDs. By modulating the Si/N ratio of the barrier layer, the enhancement and tunable EL from the SiN-based LEDs was obtained, which was attributed to the variation in the band offset between the Si-rich SiN well layer and the N-rich SiN barrier layer [15].

Here, based on our previous work on as-deposited SiN-based multilayer structure [15], pulsed laser crystallization technique was carried out to induce Si nanocrystals in the Si-rich SiN well layers, which can effectively avoid the high temperature damages in the device fabrication because of its low-temperature fabrication process due to the short pulse length and optical absorption depth. The cross-sectional TEM analysis reveals that grain size of Si nanocrystals embedded in the Si-rich SiN sublayers is independent of the laser fluence, while the grain density remarkably increases with the laser fluence. The EL measurements show that a remarkably enhanced EL intensity of more than two times can be obtained in the devices by employing the laser crystallized multilayer instead of the as-deposited multilayer as luminescent active layer. Moreover, the devices also exhibit lower turn-on voltage and the external quantum efficiency is found to be obviously higher than that of the device without laser irradiation.

2. Experimental

Si-rich SiN/N-rich SiN multilayers were fabricated on the ITO (50 Ω/□) glass substrates in a computer-controlled plasma enhanced chemical vapor deposition system, as described in our previous works [15]. The Si-rich SiN well layers with thickness of 4 nm were fabricated at the silane-to-ammonia ratio R of 50%, while the N-rich SiN barriers sublayer with 3 nm thickness were prepared at the R of 12%, which had shown the maximized EL signal among the devices with different barrier layers. The prepared Si-rich SiN/N-rich SiN multilayer structures were then irradiated in air ambient by a KrF excimer laser with a wavelength of 248 nm and pulse duration of 30 ns. Only a single pulse was employed and the laser fluence was in the range of 0.3-0.9 J/cm². The corresponding laser spot area is 5×3mm². A 1.5-mm diameter Al circle thin film was evaporated onto the surface and used as the cathode in the LEDs. The J-V characteristics of the devices are measured at room temperature by using an HP4156C semiconductor parameter analyzer. The EL spectra measurements were performed on a Perkin-Elmer LS50B fluorescence spectrophotometer at room temperature. The multilayer microstructure was characterized by Raman scattering technique and transmission electron microscopy (TEM) with a JEM4000EX microscope.

3. Results and discussion

Figure 1 displays the Raman spectra of the samples. For the as-deposited sample, only a weak and broad band peaked at 476 cm⁻¹ can be observed in the Raman spectrum. However, after the samples were irradiated by a laser fluence of 396 mJ/cm², the Raman spectrum resembles the typical features of a-Si vibration modes. It can be decomposed into four Gaussian components, located at around 150 cm⁻¹, 280cm⁻¹, 360cm⁻¹ and 480cm⁻¹, which are attributed to transverse acoustic (TA), longitude acoustic (LA), longitude optical (LO) and transverse optical (TO) phonon frequencies, respectively. These observations strongly indicate that the a-Si clusters emerge in the laser-irradiated sample. By increasing the laser fluence up to 475 mJ/cm², a sharp peak around 515 cm⁻¹, arising from the first-order Raman Scattering of Si phonons, appears in the spectrum, which convincingly demonstrates the formation of Si nanocrystals in the multilayer structure when the laser fluence is larger than the threshold value due to the laser crystallization of a-Si clusters embedded in a-SiN_x layers. With increasing the laser fluence up to 554 mJ/cm², the intensity of the a-Si-like band rapidly decreases. By decomposing the Raman spectra into two Gaussian profiles center at around 480 and 520 cm⁻¹, respectively, the crystalline volume ratio of the samples can be quantified [17]. It is found that the crystalline fraction increase monotonously from 23% to 45% in the samples with the laser fluence from 475 to 554 mJ/cm². However, the Raman peak position of crystallized samples shows no any recognizable shift upon such laser irradiation conditions, which strongly indicates the grain size remains unchanged. According to the bond polarizability model [18], the grain size for all the crystallized samples is estimated to be about 3 nm. Thus, the increasing crystallinity of the samples can be ascribing to the increasing density of Si nanocrystal, which is further confirmed by the following HRTEM images shown in Fig. 2 .

 

Fig. 1 Raman spectra of multilayer samples irradiated at different laser fluences.

Download Full Size | PDF

 

Fig. 2 Cross-section TEM micrograph of a multilayer irradiated at (a) 554 mJ/cm², HRTEM micrograph of a multilayer irradiated at (b) 475 mJ/cm² and (c) 554 mJ/cm².

Download Full Size | PDF

Figure 2 presents the cross-section TEM micrograph of the multilayer structures irradiated at different laser fluence. From the Fig. 2(a), it is clear that the flat and abrupt interface between a-SiN_x and a-SiN_y sublayers is still reserved after laser irradiation. Furthermore, the well separated Si nanocrystals in the a-SiN_x well layer can be found in the image of laser-irradiated multilayer, as shown in Fig. 2(b)-(c). The grain size with about 3 nm is smaller than the initial a-SiN_x well layer thickness, which is due to the constrained crystallization effect resulted from the interface between the a-SiN_x and the a-SiN_y. On the other hand, the rather low diffusivity of Si atoms in silicon nitride matrix also restrict the growth of Si nanocrystals. Comparing the Fig. 2(b) with Fig. 2(c), it is interesting to find that the grain density of the sample laser-irradiated at 554 mJ/cm² is obviously higher than that irradiated at a laser fluence of 475 mJ/cm², while the grain size is independent of the laser fluence at all. This behavior is completely different from that observed in the case of a-Si/SiN_x multilayer where high laser fluence would result in an increasing lateral size of grain, as a consequence, decreasing the grain density [1].

Figure 3(a) displays the EL spectra of device after laser-irradiated at 554 mJ/cm². The EL spectrum becomes clearly detectable at a low turn-on voltage of 5V, which is reduced compared with the device without laser irradiation. The EL intensity rapidly increases by increasing the V_bias. At the same time, the EL peaks is found to slightly blueshift from 560 nm to 525 nm. It is interesting to note that the maximal EL signal obtained at the high V_bias is very similar to the PL spectrum measured in the same device, which indicates both PL and EL originate from the same luminescent center. For three-dimensional confined Si nanocrystals, the energy gap E can be expressed as E(eV)=1.16+11.8/d² due to the quantum confinement effect, where d (nm) is the nanocrystal size [19]. In our case, according to equation, Si nanocrystal size corresponding to 525 nm PL band is 3.1 nm, which is agree well with the result from HRTEM measurements. Thus, the origin of the PL and EL can be attributed to the electron-hole pair recombination in the Si nanocrystals dispersed in the Si-rich SiN well layer. Therefore, the blue-shift of EL spectra at the high V_bias can be understood because the Si nanocrystal with small size that emit the light with short wavelength can only be effectively excited under the high V_bias due to the large splitting energy level according to the quantum size effect. From the inset of Fig. 3, one can see that there is a linear relationship between the integrated EL intensity and the injected current, implying that the EL is from the bipolar recombination of electron-hole pairs at Si nanocrystals. Figure 3(b) gives the current-voltage (I-V) characteristics of the devices measured at room temperature under the forward bias conditions. It is obvious that there are significant differences in the I-V characteristics between the devices with and without laser irradiation. Under the same forward V_bias the devices after laser-irradiated at 554 mJ/cm² have a much higher current density than that of device without laser irradiation. This result indicates that high carrier-injection efficiency can be achieved in the devices with properly laser irradiation energy. From the inset of Fig. 3, it is also found that the I-V data can be well fitted based on the In(I)-E⁻¹ relation which is known to describe trap-assisted tunneling (TAT) behavior characterized by the equation of $I\propto \exp \left(-\frac{8\pi \sqrt{2q{m}^{*}}}{3hE}{\Phi }_t^{3/2}\right)$, where h is Planck’s constant, m* is the electron effective mass, Φ_t is the trap energy below the conduction band and E is the applied electric field between the anode and cathode electrodes [20]. This indicates that the TAT behavior is dominant in the carrier transport process in the laser-irradiated device, which can be ascribe to the defect states generated during the laser treatment process [21].

 

Fig. 3 (a) Room temperature EL spectra of the device after irradiated at a laser fluence of 554 mJ/cm², the circle line represents the PL from the same device excited with the 325 nm line from a He-Cd laser. (b) The characteristics of the current -voltage (I-V) of the devices with and without laser irradiation under the forward bias condition, respectively. Inset shows the trap-assisted tunneling (TAT) plots and the dependence of integrated EL intensity on the injected current through the device laser-irradiated at 554 mJ/cm².

Download Full Size | PDF

Figure 4(a) presents the EL spectra from the devices with and without laser irradiation, which is performed at the forward bias of 9 V. It is found that all the EL peak energy is nearly located at around 525 nm. However, it should be noted that for the devices irradiated at a laser fluence over 475 mJ/cm², the EL peak energy blueshifts with the applied V_bias as shown in Fig. 3. This behavior is completely different from the devices irradiated at a laser fluence below 396 mJ/cm² where the EL peak energy is independent of the applied V_bias [22]. The behavior of the EL independent the applied V_bias was also observed in the Si-rich nitride/Si superlattice structures and suggested to result from the recombination of electron-hole pairs at amorphous Si clusters [23]. It is interesting to find that under the same electric excited levels the EL intensity is greatly enhanced with increasing the laser fluence. In the device after laser-irradiated at 554 mJ/cm², the integrated EL intensity is found to be significantly enhanced by more than two times as compared to that of device without laser irradiation as indicated in Fig. 4(b). It is also found that the integrated intensity is proportional to the injection current, which implies that the EL intensity is controlled by the injected electron-hole pairs. We notice that the turn-on voltage for EL from the device after laser-irradiated at 554 mJ/cm² can be reduced down to 5 V, smaller than that (6 V) of the device without laser irradiation, as is shown in Fig. 5 . By taking account of the EL intensity-to-current ratio, the external quantum efficiency (EQE) of the devices can be evaluated. Figure 5 demonstrates that under the same electric excited levels the EQE of the devices remarkably increases with increasing the laser fluence from 396 to 554 mJ/cm², which is similar to that of EL intensity from the devices. As is demonstrated in Fig. 1 and Fig. 2, the increasing laser influence would improve the crystallinity of the samples via increasing the number density of Si nanocrystal. Since the electron-hole interaction can be much intensified in Si nanostructure with diameters less than the Bohr radius of exciton(~5 nm) owing to the quantum-size effects, the increasing density of Si nanocrystal acted as luminescent centers undoubtedly improves the radiative recombination probability, leading to an increase in EL efficiency. In fact, Si nanocrystals have a remarkably higher electrical pumping cross section than defect in silicon nitride [23,24], therefore, the nonradiative recombination arising from defects would be suppressed to some extent in the a-SiN_x well layers composing of Si nanocrystals. Moreover, the improved crystallinity as a consequence of the increasing density of Si nanocrystal ameliorates the structure order of Si-Si network in the a-SiN_x well layers, also reducing the nonradiative recombination probability in the samples. Therefore, it is very reasonable that the EL efficiency behavior with the laser fluence exhibits a similar tendency to that of the Si nanocrystal density dependent the laser fluence. Thus, by using the multilayer structure followed by subsequently laser irradiation, not only the output optical power but also the EQE of the devices can be significantly enhanced.

 

Fig. 4 (a) Room temperature EL spectra of the devices after irradiated at different laser fluence. (b) The integrated EL intensity and injected current density from the device vs the laser fluence.

Download Full Size | PDF

 

Fig. 5 (a) Turn-on voltage for EL and (b) EL intensity-to-current of the devices plotted as a function of laser fluence, respectively.

Download Full Size | PDF

4. Conculsion

In summary, we studied the influence of laser irradiation on the microstructure as well as EL properties of SiN-based nanometric multilayers. It is interesting to find that the grain size of Si nanocrystals embedded in the Si-rich SiN sublayers is independent of the laser fluence, while the grain density can be well controlled by the laser fluence. The EL intensity from the device can be significantly enhanced by employing the laser crystallized multilayer instead of the as-deposited multilayer as luminescent active layer. Moreover, the turn-on voltage for the device after laser-irradiated at 554 mJ/cm² can be reduced down to 5 V and the external quantum efficiency is found to be improved by more than 40%. The present advancement opens up the possibility of developing efficient LEDs based on the SiN-based multilayers structure combining with laser crystallization technique.