Analysis of defect selective passivation process

Wet chemical etching is commonly used for estimating defects density and defect types. In order to reveal the etched pits from dislocation in GaN and look for any consistency among the various chemical etches, we have used hot H₃PO₄ and molten KOH as defect etchants in GaN, which produce hexagonal-shaped etch pits. By varying the time and temperature, the etching process produces a pitted surface that clearly reveals the size and density of the pits that we need to know. Fig. 4.3 shows that the properties of etch pits depend upon the solution temperature and etched time. When immersing specimen in solution at long time and high temperature, there are often several pits clustered together. Moreover, chemicals overetch the GaN plate.

Surface morphology of samples was characterized by scanning electron microscopy (SEM) and atomic force microscopy (AFM). The SEM image of the GaN sample etched by H₃PO₄ for 2 min at 240 °C is shown in Fig. 4.4[(a),(b)]. The etch pits, with a density of about 9.58×10⁵ cm^-2, are of hexagonal shape and their size ranges from 1.50 to 1.88μm in diameter.

Etch pits might be related with threading screw dislocations or nanopipes (open core dislocations)[1,2].

36

Figure 4.5[(a),(b)] shows that the surface morphology of the GaN sample grown by the same run immersed in molten KOH for 2min30sec at 300°C. The size of the etch pits is about 1μm, and the density of etched pits is 10⁶ cm^-2. The EPD of the sample etched by KOH is ten

times higher than that found for the H₃PO₄-etched sample. Three different shapes of the etch pits were identified with three types of defects originated from screw-, edge-, and mixed-type TDs, respectively. It has been realized that dislocation etch pits etched by molten KOH to identify the origin and mechanism of etch pits in GaN layer[3].

Fig. 4.6(a) indicates that the the defect pits density of GaN with native pits is approximately 10⁵~10⁶ cm^-2. Then the Fig. 4.3(b) shows that a pit is filled with nanospheres.

Figs. 4.1[(a)-(d)] Schematic diagrams of defect selective passivation process.

37

Figs. 4.2[(a)-(c)] Schematic diagrams of simplified defect selective passivation process

Fig. 4.3 Optical microscopy image of GaN surface etched at varied time and temperature.

(a) (c)

NPs

(b)

38

Fig. 4.4 SEM image of GaN surface etched by H3PO4. Low (a) and high (b) magnification images.

Fig. 4.5 SEM image of GaN surface etched by molten KOH. Low (a) and high (b) magnification images.

Figs. 4.6[(a),(b)] SEM image of GaN surface spin-coated nanospheres. Low (a) and high (b) magnification images.

39

4.3 Analysis of DSP-LED performance

4.3.1 Optical properties analysis of LED grown on DSP epilayer

Photoluminescence(PL)

A LED without DSP process, a LED coated 40nm silica nanospheres on GaN defect pits, and a LED coated 100nm silica nanospheres on GaN pits denoted as sample A, B, C, respectively. The LED structure with 2μm n-GaN, 12pairs of InGaN/GaN multi-quantum wells(MQWs), and 30nm p-GaN were grown on the template. The room temperature photoluminescence spectra at 10mW pumping power are shown as Fig. 4.7. The MQWs emission wavelength ranges about at 440nm. The intensity of sample B and C are two or three times than that of sample A. From the result, we think that the crystal quality of GaN epilayer grown with DSP process has been effectively improved.

400 420 440 460 480 500

40

Cathodoluminescent (CL)

The optical characteristic is investigated by SEM cross section and cathodoluminescent images as shown in Figs. 4.8[(a)-(f)]. These images are taken by simply switching detection mode from scattering electron detection to cathodoluminescent detection under the same magnification condition and thus have one to one location correspondence. Comparing to the cross section CL image of different samples, A, B and C, respectively, the DSP methed minimizes the propagation of threading dislocations.

CL plane-view images as observed in Figs. 4.9[(a)-(c)]. The dark spots are corresponding with dislocations. We estimate the density of dark points on different samples, 8.24×10⁷ cm^-2 at sample A, 5.24×10⁶ cm^-2 at sample B and 2.94×10⁶ cm^-2 at sample C. It clearly reveals that the nanospheres effectively block the propagation of threading dislocations but also reduce the density of defect pits. The threading dislocation defects are the strong non-radiative recombination centers[4]. The significant increase in CL intensity demonstrates that the loss of excited carriers due to non-radiative recombination is greatly reduced in the defect passivated layer as a result of reduction in TD density.

4.3.2 Electrical properties analysis of LED grown on DSP epilayer

LED chips with size of 300×300 μm² were fabricated from the defect selective passivation epi-wafer. The electrical characteristics are compared with a reference-LED going through the

41

same fabrication process except for the defect passivation structure. The light-current (L-I) and voltage-current (V-I) curves of LEDs are shown in Fig. 4.11. Furthermore, the optical power was collected by an integrating sphere to make sure that radiation in all directions was collected. The L-I curve in Fig. 4.11 indicates that the optical output power of DSP-LEDs are much higher than that of reference-LED. The silica masks not only block the propagation of threading dislocation but also act as light scattering sites which reduce light trapped by total internal refection (TIR) to enhance light extraction efficiency. Between the reference-LED and DSP-LED passivated by 40nm SiO₂ nanospheres, the output power of DSP LED exhibits 46%, 39%, 25% and 18% enhancement at 50mA, 100mA, 150mA and 200mA, respectively.

Meanwhile, the output power exhibits 45%, 38%, 25% and 17% enhancement at respectively 50mA, 100mA, 150mA and 200mA, when comparing reference-LED and DSP-LED passivated by 100nm SiO₂ nanospheres.

Fig. 4.12 shows the reverse voltage versus current characteristics of DSP-LEDs and ref-LED. The leakages come from the non-radiative recombination of screw type dislocations.

As observed in Fig. 4.12, we believe that the defect passivation process does not introduce more screw defects[5]. As the above mentioned, the defect selective passivation method can reduce threading dislocations and improve GaN epilayer quality.

42

Sample A (a) (d)

Sample B (b) (e)

Sample C (c) (f)

Figs. 4.8[(a)-(f)] [(a)-(c)] SEM and [(d)-(f)] CL cross section image of the defect selective passivatied epi-wafer under the same magnification.

43

(a) Sample A

(b) Sample B

(c) Sample C

Figs. 4.9[(a)-(c)] Plane-view CL images of difference samples.

44

(a)

(b)

Figs. 4.10[(a)-(b)] CL spectrum of (a) plane-view and (b) cross section LED structure.

350 400 450 500 550 600

410 420 430 440 450 460 470 0

340 360 380 400 420 440 460 480 500 0

45

Fig. 4.11 L-I and V-I curve of DSP-LEDs and ref-LED.

Fig. 4.12 Leakage current of DSP-LEDs and ref-LED under reverse bias.

0 50 100 150 200 250

46

4.4 Reference

[1] S.K. Hong*,1, B.J. Kim, H.S. Park, Y. Park, S.Y. Yoon, T.I. Kim, J. Cryst. Growth, 191 (1998).

[2] S. K. Hong, T. Yao, B. J. Kim, S. Y. Yoon, and T. I. Kim, Appl. Phys. Lett. 77, 82 (2000).

[3]L. Lu, Z. Y. Gao, B. Shen, F. J. Xu, S. Huang, Z. L. Miao, Y. Hao, Z. J. Yang, G. Y.

Zhang, X. P. Zhang, J. Xu, and D. P. Yu, J. Appl. Phys, 104, 123525 (2008).

[4] T. Hino, S. Tomiya, T. Miyajima, K. Yanashima, S. Hashimoto, and M. Ikeda, Appl. Phys.

Lett., 76, 3421 (2000)

[5] J. W. P. Hsu, M. J. Manfra, R. J. Molnar, B. Heying, and J. S. Speck, Appl. Phys. Lett. 81, 79 (2002)

47

Chapter 5 Conclusion & Future work

In summary, we successfully simplify the process of defect selective passivation technique by choosing the GaN wafer with native pits and spin-coating silica nanospheres on GaN surface. This technique can effectively not only block the propagation of threading dislocations but also improve the crystal quality of GaN epilayer. Eventually, we use the GaN epilayer dealt with defect selective passivation technique to fabricate high efficient LED.

There are some further studies listed in the following:

1. Exploring additional chemical solution that is complementary to KOH in defect selective etching to increase the coverage rate of defect selective passivation. Same passivation process needs to be investigated before sending the wafer back to MOCVD for GaN epitaxial growth.

2. To measure the beam profile of LED at the operating current is the other further work. Try to change the size of nanospheres if the divergent angle of the LED is identified.