Chapter 4 Results and discussion
4.4 Vicissitudes of two shapes of InN dots
In Fig.4-4-1, we display the variations of dot densities for flat-top and dome-shaped InN islands only for coverages larger than 2 MLs. As discussed earlier, at this growth coverage region the total dot density remains virtually unchanged. However, we do observe an opposite tendency of variations of dot densites concerning the flat-top and dome-shaped islands. As can be seen in the figure, at a coverage of 2 ML the dot densities of these two dots are about the same, lying at 2.4x107 and 3.1x107cm-3 for flat-top and dome-shaped dots, respectively; as the coverge grows the dot density of dome-shaped island tends to increase, whereas that of flat-top island tends to decrease. One can notice that when the coverage reaches 12 ML, rather few flat-top islands can be found on the surface of sample. The resulted densities of flat-top and dome-shaped islands are 4.0x106 and 4.4x107cm-3, nearly an order of difference in magnitude. The above finding suggests that the formation of dome-shaped islands seem energetically more favorable under the circumstance of high growth coverages.
Bimodal distribution has also been observed in Ge/Si and InAs/GaAs systems. The relevant shape transitions have also been discussed intensively.
The illustrative example is Ge/Si system. The shap transitions in Ge/Si have been attributed to either stress relaxation or intermixing between island and substrate. For Ge/Si system, one is pyramidal island, having the characteristics of square-based islands bounded by {105} facets; the other is dome-shaped island with structures with large number of facets that look rounded at lower resolutions[19]. Both of the islands are type of coherent island. For these
3400nm3 for InAs/GaAs, respectively. When the dot volume is small, most of the dots are in the form of so-called pyramid (or hut) shape with facet contact angle to the underlying substrate of 25° and aspect ratio of 0.11[20]. Once the dot volume exceeds the critical volume, these coherent dots begin to transform from low-aspect-ratio pyramidal islands into high-aspect-ratio dome-shaped ones because of the relaxation of large stress in the island. However, different story may occur for shape transition in InN/GaN dot system. The primary difference between InN/GaN system and InAs/GaAs or Ge/Si is that the InN dots are more likely dislocated whereas they are coherent for InAs/GaAs or Ge/Si[21].
As mentioned earlier, for InN island growth on GaN the dome-shaped island becomes predominant as the growth coverage beyond 8ML. Because of high island volume associated with both types of islands, ~40,000-8,700,000 nm3, approximately 100 times higher than the critical volume of coherent islands that observed in either Ge/Si and InAs/GaAs material system, it is considered that both of the bimodal islands in InN/GaN are highly dislocated. This is ascribed to the characteristic of high density of dislocation density,
~109-1010cm-3, of GaN template employed for our InN dot growth, stemming from the heteroepitaxial growth of GaN on highly lattice-mismatched sapphire substrate (12% mismatch). Consequently, there exhibits one dislocation on the surface of GaN buffer layer for every 100-300 nm in distance. Since the base diameters of our InN islands are around 274 and 227 nm for flat-top and dome, respectively, supportive of containing dislocation for most InN islands in our samples. This reason accounts well for the hypothesis of dislocated InN islands here. The above argument is in good agreement with recent observation by J. G.
with GaN pure edge threading dislocations (TDs), which act as nucleation site for InN island growth[18].
Fig. 4-4-3 shows surface-volume plot for both types of flat-top and dome-shaped InN islands, which is an energetically relevant measure of size and shape. For the cases in InAs/ GaAs or Ge/Si system, the scatter plot reveals two distinct shapes, which has its own slope of surface to volume value [18]. However, we observed quite different results in InN islands /GaN system.
The corresponding surface-to-volume plots in InN varies almost continuously for both types of dome-shaped and flat-top InN islands. Comparing high coverage region, steeper slopes are attained at low coverages. Generally, the mean slopes of dome-shaped islands are higher than that of flat-top ones throughout the entire volume region. It is worth to mention that at low volume region, the slop of flat-top is deceased continuously with the increasing growth coverage and moves gradually toward the values close to that of dome-shaped islands. At volumes higher than ~3x106 nm3, the slopes of the flat-top islands appear to have values almost the same as that of dome-shaped islands, which seems to imply that flat-top island changes its shape with the increasing island volume and finally transform itself into a shape of dome-shaped ones. The above phenomenon explains why the density of dome-shaped island is increased with the coverage, accompanied simultaneously with the decreasing of the density of flat-top island, while total density of InN islands can still maintain about the same for growth coverages > 2ML. From energy point of view, the change of growing facets on the island leads to changes of surface energy and island volume which can result in the shape transition of island.
Exemplary result is Ge islands on Si. For this case, it has shown that the hut is
volume shape transition occurs which make hut transform into dome-shaped island. Nevertheless, the critical volume provides a necessary but not sufficient condition for shape transition. Since the shape transition is a first-order transition, there appears an energy barrier to trap the island in a matastable configuration.
The above finding is further confirmed by the contact angle-coverage plot, shown in Fig. 4-4-2. In this figure, the contact angles of entire InN islands also distribute bimodally at low coverages. The lower contact angle ~7o, corresponds to the type of flat-top island and higher contact angle ~21o, dome-shaped one. At 2 ML, the mean contact angles are ~7o and 21o for flat-top and dome-shaped islands, respectively. The contact angles shift monotonously towards ~10o and ~28o when the coverage reaches 4 ML. At 6 ML, an additional peak ~20o, appears, a value virtually the same as the mean contact angle of dome-shaped islands at 2 ML. The plane view of AFM image and line profile are shown in Fig. 4-4-4. As the coverage higher than 6ML, it is found that the count of mean contact angle (11o) for flat top drop considerably.
Its intensity becomes about 10 times lower than that of dome-shaped islands and reaches approximately zero for coverages exceeding 8 ML. Concerning the evolutions of contact angles for the other two existed structures, we can found the mean contact angle of domes-shaped stablizes at ~35o for coverages beyond 6 ML and at the same time the contact angle of newly developed island, which we believed is the type of flat-top islands, moves gradually toward to higher angles ~22o, similar to the case of evolution of dome-shaped islands at low coverages and finally completely merged into main stream of distribution of contact angles of dome-shaped islands at 12 ML. The contact angles of
respectively. Normally, the chemical potential of an island decreases continuously with growing size, due to the smaller surface/volume ratio.
Assuming that the surface energy is the same for every allowed facet, the island energy is described as[27]
-Vα α V
E= 2/3 4/3 , (3) The chemical potential per volume can be written as
-α
From the simulation, the higher aspect ratio will have lower energy with larger dot volume. As a result, material diffuses from smaller to larger islands via either coarsening, Ostwald ripening, or shape transition process. For coarsening or Ostwald ripening, some island continue to grow while others shrink and disappear. For a given volume, its equilibrium shape of island, for sure, has a lowest free energy. The flat-dome islands, which have the contact angle of ~20o, almost have the same volume of dome-shape ones with contact angle ~35o at each coverage. The scatter plot of contact angle of InN dots as a function of their volume is shown in Fig. 4-4-5. The contact angles of both types of dots increase with volumes, and the contact angles of dome-shape saturate around ~35o at volume of ~3x106nm3. The contact angles of flat-top shown in Fig.4-4-2 grow from 70 to 35o which is identified as dome shape which is respect in the dot density of flat-top decrease with coverage. Since the energy of island will have lower energy in high aspect ratio over the critical volume, the dome-shape of dots will tends to from at large volume. We deduce that there is a shape transition between flat-dome and dome-shape, which have the same volume, for example ~3.3x106nm3 for flat-dome and 3.2x106nm3 for dome at 6ML, but reducing the island base area and increasing the height. The
existence of one additional contact angle at 6 ML, together with a clear transition of slope of surface-to-volume plot at a volume of ~ 3x106nm3 and fact of fast vanishing in dot density of flat-top island suggests that the flat-top islands are hardly to sustained themselves at volumes greater than ~3x106 nm3, indicating of energy in favor of dome-shaped island at this high volume region.
0 2 4 6 8 10 12 0
1x10
72x10
73x10
74x10
75x10
76x10
7dot density (c m
-2)
coverage(ML)
Fig. 4-4-1. Dot density of two types of dots vs. coverage
0
Fig.4-4-2 tilt angle distribution vs. coverages
0.0 4.0x106 8.0x106 1.2x107 1.6x107
Fig. 4-4-3. Scatter plots with different coverage.
20o
Fig.4-4-4 Plan view and profile of AFM images of three groups of dots.
0.0 4.0x106 8.0x106 1.2x107 0
10 20 30 40 50
contact angle(degree)
Volume(nm3)
Fig. 4-4-5. scatter plot of contact angle of InN dots as a function of volume.
Table 4-4-1 Dot density of flat top and dome shapes
Flat top dome total
1.1ML 1.1×107 1.54×107 2.64×107 1.23ML 2.05×107 1.4×107 3.45×107 1.5ML 1.15×107 1.65×107 2.80×107 2ML 2.40×107 3.10×107 5.50×107 3ML 1.40×107 3.15×107 4.55×107 4ML 1.65×107 3.33×107 4.98×107 6ML 1.97×107 3.13×107 5.09×107 8ML 7.00×106 3.86×107 4.56×107 12ML 4.00×106 4.40×107 4.80×107
4.5 PL results
For light-emitting devices, the size of low-band-gap quantum dots embedded in the active layer is a matter related to the device luminescence efficiency. The use of a smaller dot structure will certainly result in a better carrier confinement, and quantum effects to improve the quantum efficiency.
In this section, we carried out photoluminescence measurements at 13K to investigate the optical properties of our MOVPE-grown InN dot samples with different coverages. The results are shown in Fig. 4-5-1.
As can be seen in the figure, the PL intensity is, as expected, increased with the increasing coverage and no signal can be detected for samples grown with coverages lower than 2 ML due to the detection limit of our PL system.
One can notice that for 2, 3, 4, 6ML samples there are two peaks located at
~0.73eV and 0.77eV in the PL spectra, which is contributed to deep acceptor and near band edge transition, respectively. As the coverage is increased to 10 and 12ML, the higher energy peak eventually predominates the spectra. The resulted FWHM is found to decrease from 99 to 60 meV as the growth coverage is increased from 3ML to 12ML, respectively. It is interesting to note that for those InN dot samples with coverage <6ML, whose mean dot heights are less than 35 nm, blueshifts in PL peak energies should be observed because of quantum size effects. For example, a blueshift of 270meV has been observed for capped InN QD ensemble of mean height of 6.2nm. Nonetheless, no apparent peak shift is observed. Since all of our InN dot samples studied here are uncapped, the nearly invariance of PL peak energy is considered highly probably related to the accumulation of large number of surface electrons in our InN islands, owing to extraordinary high density of surface
states existed inherently to this type of film. Under this circumstance, the surface electrons which spreading spatially separated will lower the transition energy to counterbalance the blueshift of energy caused by quantum size effect.
0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 0.00
1.50x105 3.00x105 4.50x105 6.00x105
In tensity(a.u.)
Energy(eV)
3ML 4ML 6ML 8ML 12ML
Fig. 4-5-1. PL spectra of InN dot vs. coverage ML low energy High energy
3 0.731 0.778
4 0.729 0.778
6 0.729 0.773
8 0.728 0.761
12 0.728 0.76
Chapter 5 Conclusions
We have studied the evolution of InN islands on GaN grown by MOVPE at 650oC in terms of morphological shape, height, diameter, aspect ratio, emission peak energy and it’s FWHM. Experimental results indicate that bimodal growth feature was taking place in our InN island/GaN system. There are two types of InN nanodots, assigned as flat-top and dome-shaped islands, exhibit characteristics of the hexagonal shape base with a flat plateau on top and a circular base with a dome on top.
It is found that the diameters of our InN islands tend to increase sharply from 133 to 428 nm for flat-top island and 143 to 332 nm for dome-shaped island for 1 to 3 ML coverage at a lateral growth rate of ~100 nm/ML. Then, the growth became slow from 428 to 667 nm and 332 to 509 nm in the range of 3-12 ML at a growth rate of ~20 nm/ML, which is nearly 5 times slower in growth for two coverage regions. Such a faster lateral growth rate at lower coverages is considered to relate with the characteristics of InN nanodots grown on the top of dislocation, extending from the interface between GaN buffer layer and sapphire substrate. The reduction of the total energy of island makes the lateral growth more favorably at initial growth stage. However, once the critical volume, is reached, at ~3ML, energy barrier starts building up around the edge of islands. It impedes not only the further attachment of surface adatoms to the edge of island but also directs the island growth to grow preferentially in the vertical direction, causing more and more flat-top islands to transform into dome-shaped islands. As a consequence, at 12 ML nearly all of the InN islands become dome-shaped islands, their respective densities are 4.4 x107 and 4.0x106cm-3.
The transformation of flat-top InN islands into dome-shaped ones can also be revealed by its surface-to-volume plot. The slope of flat-top is found to decrease continuously with the increasing growth coverage and moves gradually toward the values close to that of dome-shaped islands and finally reaches a value almost the same as that of dome-shaped islands as the island volume
>~3x106 nm3.
More clear evidence regarding the shape transition can be observed in the plot of contact angle as a function of coverage.
At the beginning, there exhibits two groups of contact angles peaked at 7
oand 21
o, corresponding to flat-top and dome-shaped islands, respectively. The mean peak values of dome-shape islands tend to move gradually toward higher values with the increase of coverage and finally stabilize at ~11
oand ~35
o. It is interesting to note that at coverage of 6 MLs one additional group appears at ~21
owhich turns to increase with its contact angle with increasing coverage and seems to merge completely into group of dome-shape islands at higher coverages.
Since the island having higher contact angle possesses lower
formation energy, we believe the additional group comes highly
probably from flat-top islands. The plot of surface-to-volume ratio
further confirms the argument of shape transition of flat-top to
dome-shaped island at higher island volume.
Similar to the case ofevolution of dome-shaped island mentioned previously, the contact angles of these islands shift gradually toward to higher values and merge completely into main stream of dome-shaped islands at 12 ML.
In summary, we have investigated comprehensively the evolution of structural parameters of bimodal InN islands. The existence of one additional contact angle at 6 ML, together with a clear transition of slope of surface-to-volume plot at a volume of ~ 3x106nm3 and the fact of fast vanishing in dot density of flat-top islands all suggest that the flat-top islands hardly to sustain themselves at volumes greater than 3x106 nm3. Eventually all InN islands are transformed into dome-shaped island, indicating that dome-shaped island is energetically favored at this high volume and hence high coverage region.
Reference:
[1] S. Nakamura, M. Senoh, and T. Mudai, Jpn. J. Appl. Phys., Part 2 32, L8 (1993).
[2] S. Nakamura, M. senoh, S. Nagahama, N. Iwasa, T. Yamada, T.
Matsushita, H. Kiuoku, and Y. Sugimoto, Jpn. J. Appl. Phys., Part 2
35, L74 (1996).
[3] V. W. L. Chin, T. L. Tansley, and T. S. Osotchan, J. Appl. Phys. 75, 7365 (1994).
[4] A. Yamamoto, M. Tsujino, M. Ohkubo, and A. Hashimoto, Sol.
Energy Mater: Sol. Cells 35, 53 (1994).
[5] V. Yu. Davydov, A. A. Klochikhin, R. P. Seisyan, V. V. Emtsev, S. V.
Ivanov, F. Bechstedt, J. Furthmuller, H. Harima, A. V. Mudryi, J.
Aderhold, O. Semchinova, and J. Graul, Phys. Stat. Solidi(b), 229(3), R1 (2002).
[6] J. Wu, W. Walukiewicz, K. M.Yu, J. W. Ager, E. E. Haller, H. Lu, W.
J. Schaff, Y. Saito, and Y. Nanishi, Appl. Phys. Lett. 80, 3967 (2002).
[7] R. Intartaglia, B. Maleyre, S. Ruffenach, O. Briot, T. Taliercio, and B.
Gil, Appl. Phys. Lett. 86, 142104 (2005).
[8] W. C. Ke, C. P. Fu, C. Y. Chen, L. Lee, C. S. Ku, W. C. Chou, W. H.
Chang, M. C. Lee, and W. K. chen, Appl. Phys. Lett. 88, 191913 (2006).
[9] W.C. Ke, H.Y. Huang, C.S.Ku, K.H. Yen, L. Lww, W.K. Chen,W.C.
Chou, M.C.Lee, W.H.Chen, W.J. Lin, Y.C. Chen, and Y.T.Cherng, Jpn. J. Appl. Phys., 43, L780 (2004).
[10] A. F. Wright, J. Appl. Phys., 82, 2933 (1997).
[11] Z. G. Qian, W. Z. Shen, H. Ogawa, and Q. X. Guo, J.Phys.: Condens.
[14] B. Bansal, M. R. Gokhale, A.Bhattacharya, and B. M. Arora, J. Appl.
Phys. 101, 094303(2007).
[15] C. Adlmann, B. Daudin, R. A. Oliver, G. A. D. Briggs, and R. E.
Rudd, Phys. Rev. B, 70, 125427(2004).
[16] T. I. Kamins, G. M. Ribeiro, D. A. A. Ohlberg, and R. S. Williams, J. Appl. Phys., 85, 2933 (1999).
[17] G. M. Ribeiro, A. M. Bratkovski, T. I. Kamins, D. A.A. Ohlberg, R. S.
Williams, Science, 279, 353.
[18] J. G. Lozano. A. M. Sanchez, R. Garcia, and D. Gonzalez, D. Araujo, S. Ruffenach, and O. Briot, Appl. Phys. Lett. 87, 263104 (2005).
[19] G. M. Ribeiro, T. I. Kamins, D. A. A. ohlberg, and R. S. Williams, Phys. Rev. B, 58, 3533(1998).
[20] Y. Zhang, and J. Drucker, J. Appl. Phys., 93, 9583 (2003).
[21] T. KEHAGIAS, A. Delimitis, P. Komninou, E. Iliopoulos, E. Dimakis, A. Georgakilas, and G. Nouet, Appl. Phys. Lett. 86, 151905 (2005).
[22] A. L. Barabasi, Appl. Phys. Lett. 70, 2565 (1997).
[23] H. T. Johnson, and L. B. Freund, J. Appl. Phys., 81, 6081(1997).
[24] M. Hammar, F. K. LeGoues, J. Tersoff, M. C. Reuter, R. M. Tromp, Surf. Sci., 349, 129(1996).
1835(1997).
[26] R. E. Rudd, G. A. D. Briggs, A. P. Sutton, G. Medeiros-Ribeiro, and R.
S. Williams, Phys. Rev. Let. 90, 146101 (2003).
[27] F. M. Ross, J. Tersoff, and R. M. Tromp, Phys. Rev. Lett. 80, 984 (1998).
[28] G. Costantini, A. Rastelli, C. Manzano, R. Songmuang, O. G. Schmidt, K, Kern, and H. Von Kanel, Appl. Phys. Lett. 85, 5673 (2004).
[29] A. F. Wright, and J. S. Nelson, Phys. Rev. B, 51, 7866(1995).