Results and discussion

Figs. 2.5.1(a) and (b) show the temperature-dependent variations of PL spectral peak and normalized integrated PL intensity, respectively, of the three samples. From part (a), one can see the blue shift of sample HB when compared with samples HU and HW. Only sample HU generates a clear S-shape variation [14].

0 50 100 150 200 250 300 10^-2

10^-1 10⁰

Normalized Intensity (b)

Temperature (K)

HU HW HB 2.30

2.35 2.40 2.45 2.50 2.55

(a)

Photon Energy (eV)

HU HW HB

Fig. 2.5.1 (a) Temperature-dependent PL spectral peak positions, and (b) integrated PL intensities of the three samples

As shown in Fig. 2.5.1(b), the radiative efficiency of sample HB is higher than those of the other two samples. Such results are quite similar to what we have reported previously with the samples emitting violet photons [2]. Barrier-doped samples always result in higher optical quality. Fig. 2.5.2 shows the PL and DEDPLE spectra of the three samples at 10 K.

48 spectra of the three samples.

The eV values shown are the detection photon energies. peak positions as functions of excitation photon energy of the three samples.

The DEDPLE spectra were normalized to the level at 3.5 eV (GaN absorption peak).

One can see that in sample HU, the luminescence intensity level of the InGaN absorption band decreases with increasing detection photon energy. However, the opposite variation trend is observed in sample HW. Meanwhile, almost the same absorption spectra, when the detection photon energy is varied, are measured in sample HB. Such major differences imply the significant variations in sample nanostructure upon silicon doping of different conditions.

Fig. 2.5.3 shows the EEDPL spectral peak positions as functions of excitation photon energy for the three samples. In sample HU, the EEDPL spectral peak maintains almost constant

49

when carriers are excited in InGaN. That of sample HW increases with decreasing excitation energy. However, the opposite trend is observed in sample HB.

Figs. 2.5.4-6 show typical SSA images of samples HU, HW, and HB, respectively. In these SSA images, line scans were conducted along the shown white lines. Here, the line scan values 1 and 1.1, respectively, represent indium compositions of 0 and 60 % (estimated based on the assumption of a specimen thickness larger than 30 nm). Different colors stand for various ranges of indium composition, as shown in the legends.

Fig. 2.5.4. A typical strain-state analysis (SSA) image of the undoped InGaN/GaN quantum well sample HU.

The color legend indicates the estimated indium mole fractions (1.0 = 0 %, 1.1 = 60 %).

Fig. 2.5.5. A typical strain-state analysis (SSA) image of the well-doped InGaN/GaN quantum well sample HW.

The color legend is the same as that in Fig. 2.5.4.

Fig. 2.5.6 A typical strain-state analysis (SSA) image of the barrier-doped InGaN/GaN quantum well sample HB.

The color legend is the same as that in Fig. 2.5.4.

50

As shown in Fig. 2.5.4 for sample HU, although the QW interface is blurred, indium is basically confined within the well. Here, within the well a few spots of indium aggregation can be observed. From the line-scan results, one can observe quite a weak fluctuation in indium composition either along the well layer or in the growth direction. The fluctuation contrast (the difference between the maximum and minimum in the scan range) is around 0.03 along the QW layer.

Then, in Fig. 2.5.5 for sample HW, the QW is not as well shaped as sample HU. The indium composition fluctuations in both directions are relatively stronger in comparing with sample HU. In particular, more indium-aggregated clusters can be observed within the well layer. The fluctuation contrast is now around 0.045 along the QW layer. As shown in Fig.

2.5.6, the SSA image of sample HB shows quite a different nanostructure from the other two samples. Here, the QW layer becomes unclear. Instead, a distribution of clusters of different sizes and shapes exists. The indium composition fluctuation (the contrast is now around 0.075 along the QW layer) is much stronger than those of the other two samples, implying stronger carrier localization for effective recombination.

In Fig. 2.5.2(a), the DEDPLE signal intensity decreases with increasing detection photon energy in sample HU. This trend can be attributed to the relatively weaker potential fluctuation in this sample. With such an energy level distribution, photo-generated carriers in higher energy levels can easily transport to the absolute potential minimum for recombination within a certain region. Hence, when the detection photon energy is low, most carriers can contribute to photon emission. When the detection photon energy is higher, fewer carriers can actually recombine at this relatively higher energy level and hence the DEDPLE signal becomes weaker. In this situation, PL spectral peak is always located at the absolute potential minimum and is independent of the excitation level, as shown in Fig. 2.5.3. On the other hand, in sample HW the potential fluctuation becomes relatively stronger. In particular, more clusters are formed in the designated QW layers (see Fig. 2.5.5). In this situation, it requires certain amount of energy and hence is more difficult for carriers to transport from a local minimum to another of a deeper level. Therefore, the majority of photo-generated carriers can be trapped by local minima of relatively higher energy. When the detection photon energy is high, stronger DEDPLE signals are recorded, as shown in Fig. 2.5.2(b). In this case, as the excitation energy becomes lower in the EEDPL measurement, more carriers can actually be trapped in local potential minima of relatively higher levels such that the PL spectral peak energy increases with decreasing excitation energy, as shown in Fig. 2.5.3. Then, in sample HB, because of the strongly clustering structure with an island-by-island configuration (see Fig. 2.5.6), when carriers are generated at high InGaN energy levels, they can transport directly into individual potential minima without a cascading relaxation process. In this situation, carrier distributions after relaxation among shallow and deep potential minima can be quite even such that the DEDPLE signal intensity is almost independent of the detection photon energy. Also, as the excitation photon energy decreases, the local potential minima of relatively lower levels can collect more carriers and hence the EEDPL peak position decreases.

Carrier localization and quantum-confined Stark effect (QCSE) have been the two major mechanisms for explaining the S-shape behavior of temperature dependent PL spectral peak position. However, it is still unclear which one dominates under a certain condition.

With the results above, one can speculate that upon silicon doping in wells or barriers, the QCSE is reduced, due to strain relaxation and/or carrier screening, and the S-shape behavior disappears in samples HW and HB (see Fig. 2.5.1). This argument actually implies that the usually observed S-shape behavior can be dominated by the QCSE. On the other hand, carrier localization can be the key to the blue shift and radiative efficiency improvement, particularly in the barrier-doped sample. The similar nanostructures of samples HU and HW explain well their close PL spectral peak positions, particularly below 150 K, and their close integrated PL

51

intensity curves. The stronger carrier localization in sample HB does result in higher radiative efficiency. This result may imply that carrier localization is more effective in blue-shifting luminescence and improving radiative efficiency of a sample, when compared with the relaxation of QCSE.

2.5.3. Conclusions

In summary, we have compared the results of PL, DEDPLE, EEDPL, and SSA of three InGaN/GaN QW samples with un-doped, well-doped, and barrier-doped structures. The SSA images showed strongly clustering nanostructures in the barrier-doped sample and relatively weaker composition fluctuations in the undoped and well-doped samples. Such variations in nanostructure resulted in different carrier transport processes that well explained the DEDPLE and EEDPL observations. Also, the PL results provided us clues for speculating that the S-shape PL peak position behavior is dominated by the QCSE in an undoped InGaN/GaN QW structure. However, carrier localization is more effective in blue-shifting luminescence and improving radiative efficiency of a sample, when compared with the relaxation of QCSE.

References

1. K. Uchida, T. Tang, S. Goto, T. Mishima, A. Niwa, J. Gotoh, Appl. Phys. Lett. 74, 1153 (1999).

2. Y. C. Cheng, Cheng-Hua Tseng, Chen Hsu, Kung-Jen Ma, Shih-Wei Feng, En-Chiang Lin, C. C. Yang, and Jen-Inn Chyi, J. Electron. Mater. 32, 375 (2003).

3. Y. H. Cho, F. Fedler, R. J. Hauenstein, G. H. Park, J. J. Song, S. Keller, U. K. Mishra and S. P. DenBaars, J. Appl. Phys. 85, 3006 (1999).

4. M. S. Minsky, S. Chichibu, S. B. Fleischer, A. C. Abare, J. E. Bowers, E. L. Hu, S.

Keller, U. K. Mishra and S. P. DenBaar, Jpn. J. Appl. Phys. 37, L1362 (1998).

5. T. Deguchi, A. Shikanai, K. Torii, T. Sota, S. Chichibu and S. Nakamura, Appl. Phys.

Lett. 72, 3329 (1998).

6. T. Wang, H. Saeki, J. Bai, T. Shirahama, M. Lachab, S. Sakai and P. Eliseev, Appl. Phys.

Lett. 76, 1737 (2000).

7. J. Dalfors, J. P. Bergman, P. O. Holtz, B. E. Sernelius, B. Monemar, H. Amano and I.

Akasaki, Appl. Phys. Lett. 74, 3299 (1999).

8. S. Ruvimov, Z. Liliental-Weber, T. Suski, J. W. Ager III, J. Washburn, J. Krueger, C.

Kisielowski, E. R. Weber, H. Amano and I. Akasaki, Appl. Phys. Lett. 69, 990 (1996).

9. E. Oh, C. Sone, O. Nam, H. Park and Y. Park, Appl. Phys. Lett. 76, 3242 (2000).

10. M. Y. Ryu, P. W. Yu, E. J. Shin, J. I. Lee, S. K. Yu, E. Oh, O. H. Nam, C. S. Sone and Y.

J. Park, J. Appl. Phys. 89, 634 (2001).

11. C. K. Choi, Y. H. Kwon, B. D. Little, G. H. Gainer, J. J. Song, Y. C. Chang, S. Keller, U.

K. Mishra and S. P. DenBaars, Phys. Rev. B, 64, 245339-1 (2001).

12. I. Ho, G. B. Stringfellow, Appl. Phys. Lett. 69, 2701 (1996).

13. D. Gerthsen, B. Neubauer, A. Rosenauer, T. Stephan, H. Kalt, O. Schon and M. Heuken, Appl. Phys. Lett. 69, 2701 (1996).

14. Y. H. Cho, G. H. Gainer, A. J. Fischer, J. J. Song, S, Keller, U. K. Mishra, and S. P.

DenBaars, Appl. Phys. Lett. 73, 1370 (1998).

52

2.6 IMPROVEMENTS OF InGaN/GaN QUANTUM WELL