Experimental results and discussion

The structural features of InGaN–GaN MQWs are very important for device performance of LEDs, for instance, indium composition and interface. Therefore, the XRD measurement was performed on the a-plane MQWs sample to examine the well width and identify the indium composition. Figure 4.3.1 shows typical -2 scans for the InGaN–GaN MQWs grown on the a-plane GaN template. Three clear satellite peaks could be found which indicated good crystal quality of the MQWs. The structural parameters of the QW’s well width could be calculated by analyzing the angular positions of the main satellite peaks and the growth time of wells and barriers.

barrier thickness was estimated to be 45 and 20 nm, respectively, for the MQWs grown on the a-plane GaN template.

In figure. 4.3.2(a), the room-temperature EL spectra of the a-plane LED was measured at drive currents ranging from 1 to 100 mA. The spectra was measured under pulsed bias (the pulse duty cycle = 10%) to prevent red shift of emission peak due to the heating effect [42]. The EL peak emission first exhibited blue shift with the current increasing and then became constant at 506 nm, which differed from that reported by Detchprohm et al. [44]. They reported there was a red shift in the dominant peak as the drive current increased. The heating effect due to CW operation accounts for the apparent red shift in wavelength. In order to identify the quantity of emission shift, we plot the EL peak as a function of pumping current as shown in figure 4.3.2(b). An initial blue shift (~127 meV) was observed when the drive current was increased from 1 to 20 mA. No shift in the emission wavelength was observed above 20 mA. Since the QCSE is absent in the a-plane MQWs, the blue shift induced by screening effect is not considerable. Hence, we attributed the initial blue shift with the increase in drive current to the band filling effect as a result of high inhomogeneous indium incorporation [42, 43]. The inset shows the image of a-plane LED chip lightened at 20-mA drive current which emitted pure green light.

Figure 4.3.3 reveals the variation of output power and EQE as a function of drive

current. The inset shows the current-voltage (I-V) characteristic under forward bias.

The I-V curve of the LED exhibits a turn-on voltage between 2-3 V. The forward voltage at 20-mA drive current was 3.42 V and the differential series resistance was estimated to be 24 . The output power increased linearly as the current increased.

The output power was 55 W at 20 mA and 240 W at 100 mA, respectively. The EQE first increased as the drive current increased until the maximum value reached 0.1% at a drive current of 9 mA. After that, it decreased gradually as the drive current increased. Compared with the previous report [44], the EQE of our green LEDs exhibited higher value at the same current density of 12.7 A/cm². There could be two reasons accounting for the efficiency enhancement. One is the in situ SiNx nanomask is employed to reduce the defect density [45]. The other is the thermal effect is eliminated as a result of pulse operation. The degree of polarization of EL intensity of

our devices was analyzed by rotating a polarizer between the polarization angle of 0

and 360 as shown in figure 4.3.4. The polarization ratio is defined as  = (ImaxImin)/

(ImaxImin), where Imax is the intensity of light with polarization perpendicular to the -axis and is the intensity of light with polarization parallel to the c-axis. From figure 4.3.4, the degree of polarization was estimated to be about 67.4%. Compared with the UV a-plane LEDs ( =28.7%) [40], the EL polarization degree of green LEDs has a 2.34-fold increase. A similar behavior was observed by Koyama et al. [46]. They

reported that the PL anisotropic polarization in nonpolar LEDs will gradually increase with the red shift of the emission peak, which could attribute to the large valence band splitting of the InxGa1-xN wells of high x. Consequently, for our green LEDs, the higher polarization ratio compared with that of UV a-plane LEDs could be due to the higher indium incorporation in MQWs.

Fig. 5.3.1 XRD -2 scans for the InGaN–GaN MQWs grown on the a-plane GaN template

Fig. 5.3.2 (a) Room-temperature EL spectra with pulse bias for a-plane green LED. (b) EL peak emission shift as a function of pulse current. Inset shows the image of a-plane LED chip lightened at 20-mA drive current.

Fig. 5.3.3 Output power and EQE as a function of drive current. Inset shows the I-V characteristic of green LED under forward bias

Fig. 5.3.4 Variation of EL intensity with angular orientation of the polarizer at 20-mA operation current.

5.4 Summary

In conclusion, we have demonstrated a-plane green InGaN–GaN LEDs grown on r-plane sapphire and investigated their EL characteristics. A series of experiments showed that high indium composition is allowed to incorporate into a-plane GaN which encourage the development of green LEDs along the nonpolar direction.

Although the QCSE was eliminated in a-plane green LEDs, the EL spectra still exhibited blue shift with the current increasing due to band-filling effect as a result of inhomogeneous indium incorporation. The EL polarization anisotropy was observed clearly in a-plane green LEDs and the polarization degree had a 2.34 times increase compared with that of UV a-plane LEDs, which could be attributed to large valence band splitting induced by high indium concentration in the wells.