Enhanced photoluminescence of InN dots by FME

Recently, a great deal of work has been devoted to elaboration of InN studies, because of its unique properties such as high theoretical maximum mobility, high peak drift velocity and low energy bandgap (∼0.69 eV) [1-3]. Consequently, the use of InN and its alloys with GaN and AlN can extend the bandgap of nitride-based LEDs from approximately 6.2 to 0.69 eV, making it very suitable for fabrication of light emitting devices covering wide range of emission wavelengths from ultraviolet to near infrared region. Additionally, rather interesting PL result is observed in InN film.

Our early study shows that the PL peak energy of InN dots exhibits almost no shift as the measured temperature changed from 10 to 300 K [4]. Such a characteristic would benefit to fabricating 1.3-1.55 μm laser diodes with high wavelength-stability, advantages for optical communications. Despite numerous literatures have been published on discussing its bandgap and relevant physical properties, few attentions have been paid on InN dot growth. Up to now, the InN dot growth is mainly conducted by way of conventional growth manners using either molecular beam epitaxy (MBE) [5-8] or metalorganic chemical vapor deposition (MOCVD) techniques [9,10]. Nevertheless, to our knowledge, no photoluminescence result has been reported yet for MBE-grown InN dot samples, due to possibly the rather low growth temperature associated with such type of dot growth feature. As for MOCVD, the InN dot prepared by conventional method has been reported firstly by Ruffenach et al. [11] They observed a strong blue shift in emission energy from encapsulated nano-scale InN dots, originating fundamentally from the carrier confinement in their quantum size structure. In our previous paper [4], we have demonstrated that the InN dots with good optical quality can be also prepared successfully by a so-called flow-rate modulation epitaxy (FME), modified from MOCVD growth technique, in

which the In and N gas sources were alternately introduced into the growth chamber.

In this study, we further investigate the dependence of optical properties, together with dot density and relevant surface morphology and structure as functions of growth temperature and amount of NH3 background flow which introduced in In source step for our InN dots grown by FME, profound effects were found in this study.

7-1 Experimental details

The InN dots used here were grown on 1 μm-thick GaN buffer layer/sapphire (0001) substrates by FME with six growth cycles. The gas flow sequence for one growth cycle basically consists of four steps: 20-sec TMIn step, 20-sec NH3 step, intervened with 10-sec nitrogen carrier gas purge in between (Fig. 7-1). The TMIn and NH3 flow-rate use here were 150 and 18000 sccm for In and N steps, respectively.

It is worth to notice that in the TMIn step, a small amount of NH3 of 1000 sccm flow rate, if not mentioned elsewhere, was also intentionally provided so as to suppress the reevaporation of In atom during the step. The TMIn step here is also referred as growth step, because of the deterministic nature of group III elements in III-V compound growth, and whereas the NH3 step as annealing step where abundant NH3

was supply to convert the left unreacted adsorbed In atoms in growth step and reorganize them into InN form. The other growth conditions of InN dots were described in Table 7-1. In this study, several runs of the InN dots grown by conventional MOCVD method were also performed as references. In these cases, the TMIn and NH3 flow rates during the dot preparations were kept at 150 and 10000 sccm, respectively, with a dot growth time of 2 min., exactly the same TMIn flow rate and the total growth duration of six-cycle TMIn growth steps that employed in our FME dot growth. The PL measurements were performed by using the 488-nm line of an argon-ion laser as an excitation source. The PL signals were analyzed by a 0.5-m monochomator and detected by an InGaAs photodiode with a cut-off wavelength at 2.05 µm. The surface structures were examined by a NT-MDT atomic force microscopy (AFM) system. Imaging was performed in air using noncontact tapping mode with the silicon cantilever.

Table 7-1: The detail growth conditions of InN dots grown by FME technique.

7-2 Growth temperature

It is no doubt that the growth temperature is one of the most important parameters in determining the epitaxial growth of film, including the dot growth, we thus perform numbers of FME and MOCVD InN dot growths at different growth temperatures from 550 to 730^oC. Figure 7-2 shows the resulted Arrhenius plot of InN dots density as a function of reciprocal temperature. As anticipated, the density of InN dots is very sensitive to the substrate temperature. Despite of different growth

methods, two regions with about the same dividing temperature ~700℃ were clearly observed. In the temperature range of 550-700℃, the corresponding dot density tends to decrease gradually from 1.5×10⁹ to 3.2×10⁸ cm^-2 for FME and 7.5×10⁹ to 5×10⁸ cm^-2 for MOCVD with the increasing temperature. Afterwards, both of them decline drastically and eventually become zero, i.e. no dots growth, for temperature beyond 730℃. The obtained average dot height and diameter are listed in Table 1.

By referring to the island nucleation mechanism proposed by Robison et al., [12]

we learn that the dot density at low growth temperatures is governed mainly by the diffusion capability of adatom, while that at high temperatures is determined by the adsorption energy of adatom to the adsorbed site. The respective characteristic equations are activation energy of surface adatoms and adsorption energy needed for them to adsorb the dot islands.

The derived values of adatom diffusion energy Ed and adsorption energy Ea from the dot density curves are 0.7 and 16.4 eV for FME and 1.3 and 12.3 eV for MOCVD, respectively. Comparing to MOCVD, the InN dot growth by FME appears to possess much lower diffusion activation energy, almost half of its opponent, and higher adatom adsorption energy. The low diffusion activation energy in FME suggests that the species of random-walk adatoms on the growing surface due to the feature of alternate injection scheme differ largely from conventional MOCVD. For MOCVD

InN growth, it is commonly accepted that the adatoms that migrate on the growing surface are most likely the In-Nmolecules or its complexes, formed favorably via heterogeneous reaction in continuous growth manner, which diffuse a short distance before their incorporation into the solid. On the other hand, for FME InN growth, if the growth conducted in In growth step is totally absent from reactive nitrogen species, highly mobile In adatoms are easily formed on the surface to control the deposition. Since in this series of experiments, one tenth of NH3 flux that used in MOCVD was selected as the background flow in FME during TMIn growth step, it is believed that the surface migration adatoms are combination of the individual In atoms and In-N molecules or it complexes. Such a compound effect will certainly make the surface adatoms to migrate a longer distance before they meet each other to form new islands or be caught by existing islands. This explains for the sparser dot density observed in FME-grown InN dot film than that by MOCVD at low temperatures.

In regard to the high temperature regime, as mentioned earlier there appears to have steeper functions of dot density on temperature. Two likely mechanisms may account for this. One is InN dissociation corresponding to a fast N escape from InN, the other involves fast In evaporation from the growing surface [13]. Although under vacuum environment the dissociation of InN (∼ 600℃) [14,15] occurs much earlier than that of In evaporation on InN surface (∼670℃), the onset of dissociation temperature in reality is dependent to a great extent on the given growth conditions, such as chamber pressure, surface plane of substrate, growth rate, and incoming TMIn and NH3 source fluxes employed during sample preparation. If the dissociation does govern the growth at high temperatures, starting from 700℃ the surface would be covered with lots of In droplets, caused by large quantities of N loss leaving exposed

In atoms on the surface to form droplets, which is expect to become less severe with increasing temperature because of the increasing active N supply simply due to the improved cracking efficiency of NH3. Since neither no In droplets at 700℃ nor reduction of droplet density at higher growth temperatures were observed, we speculate the evaporation of excessive In adatoms is the most probable reason that responsible for sharp decline in dot density in our samples at high temperatures. Such an argument is also confirmed by the fact of steep slope in In growth efficiency in this temperature regime, as depicted in the inset in Fig. 1. One can see in the figure, the In growth efficiency drop quickly for growth temperatures above 700℃.

The resulted 10-K PL spectra of these uncapped InN dots grown by both FME and MOCVD from 550 to 730℃ were displayed in Figs. 7-3(a) and (b), respectively, the peak energy and FWHM as functions of growth temperature were also presented in the figure. The data of InN dot sample grown by conventional MOCVD at 550℃

is not shown here because no measurable signal can be attained. As can be seen in the figure, the peak energy for MOCVD sample decreases linearly from 0.79 to 0.75 eV with raising temperature (see the inset), whereas that for FME is generally lower than MOCVD-grown samples at low growth temperatures and approaches approximately the same peak energy as the MOCVD sample grown at 730℃. Nonetheless, they all deviate enormously from the reported 10-K InN bandgap, ∼0.69 eV [16], reflecting strong Moss-Burstein effect [17], stemming primarily from high carrier concentration, existed in all of our samples. The character of lower PL energy associated with FME samples indicate better quality InN dot with lower background doping can be achieved by using FME, particularly at low growth temperatures. As far as FWHMs of PL spectra are concerned, as anticipated a too high or too low growth temperature

in the study would lead to a broader linewidth in the emission spectrum. The FWHM values lie in the range of 63 – 90 meV, with minimum values, 63 meV for FME and 71 meV for MOCVD, occurred at 600 and 700℃, respectively. Knowing that the PL linewidth is one of the key parameters in gauging the material qualities, the optimal growth temperature for InN dot growth by FME here is determined to be ∼600℃, considerably lower than that, 700℃, in conventional growth method.

7-3 NH3 flow rate in TMIn flow period

To verify the influences of NH3 background flow in TMIn step on InN dot growth by FME, we subsequently conducted another series of experiments at a temperature of 600℃. Figs. 7-4 shows the dot density and their AFM images prepared at different NH3 background flows varied from 0, 500, 1000, 5000 and 10000 sccm, respectively. It is observed that the FME-grown InN dots density increases gradually from 6×10⁸ cm^-2 to 1×10¹⁰ cm^-2, nearly sixteen-fold in value with increasing NH3

background flow during the growth step. The increasing dot density with NH3

background flow reflects clearly that the migration length of surface adatoms is decreased about by a factor of four, assuming migration length λ_c =N_s^−1/2 is valid [17], where Ns is the dot density of the islands and N_s^−1/2represents the mean distance between nearest-neighbor islands. We would like to point out here, when NH3=10,000 sccm background is used, which is exactly the same value utilized in our conventional InN dot growth, the FME deposition in this case is more like the conventional method except chopped by nitrogen purged step. Consequently, by tuning the NH3 background flow in FME InN dot depositions the growth here basically undergo growth conditions from a case that is totally free from active nitrogen ambient to strong N-rich growth conditions. Under N-rich conditions, the adsorbate-surface interactions are predominantly realized by In-N bonds. The diffusion barriers on these surfaces are thus mainly characterized by breaking the strong cation-anion bonds. As a consequence, high diffusion barriers and hence low adatom diffusion mobilities are resulted. On the contrary, under entirely NH3-free indium growth step in FME, the surface exhibits a metal-like character, because of weak binding energy of the In-In bond, the In adatoms move more freely on the

In-terminated surface, giving rise to a marked decrease in dot density in its InN dot film.

Not only the dot density is affected, the change of NH3 background flow in TMIn growth step also affects greatly the appearance of InN islands. Figures 7-5 show the dot height, diameter and aspect ratio, which defined as ratio of base diameter to height, as functions of the NH3 background flux. The typical line profiles and AFM images of single individual dots are presented as well in the inset. Except lens-like shape for sample grown at 10,000 sccm NH3 background, all the other FME-grown InN dots exhibit a disk-like shape, or more accurately, a truncated pyramid with a hexagonal base, with base diameter a few times larger than the height. Regardless of the viewpoint of aspect ratio, dot diameter or height, their curves all show two distinct regions in the figure, divided by a NH3 background flow of ∼1,000 sccm, which are assigned as N-rich and In-rich regions, respectively. Under N-rich growth environments, the aspect ratios are low, ranged from 3.4 to 5.1, close to a value of 2.8 for non-truncated hexagonal InN pyramid (α=65^o) [18]. In this regime, a slow increase in aspect ratio is observed with decreasing NH3 background; the same growth tendency is also observed in dot height and diameter. In contrast, under In-rich conditions an abrupt increase feature is observed for all of the aforementioned dot geometry parameters. The aspect ratio jumps to 7.0 when NH3 background drops completely to zero, indicating faster rising in lateral growth rate with respect to vertical growth rate for InN dots growth in this area.

Based on above observations, it can be concluded that the FME depositions under In-excess conditions here are thermodynamically controlled, where the surface adatoms are mainly the In adatoms that move on metallic surfaces. Higher adatom

mobility under more In-rich regime will make the surface adatoms traveling longer distances, hopping over the step edge barriers of islands to find their energetically most stable sites, which resulting in higher aspect ratios of the deposited InN nanodots in the films. On the other hand, the depositions under N-rich conditions here are limited mainly by kinetics, where N-terminated surfaces are built and are kinetically stable. These surfaces exhibit higher diffusion barriers and hence significantly reduce In diffusion lengths which drive the In atoms to compile uphill to form low aspect-ratio nanodots. Since faceting is energetically more favorable under more N-rich conditions, hexagonal InN pyramids with a clear facet angle α= 48^o do occur at a NH3 background flow of 10,000 sccm in present work.

The 10-K PL spectra of the above InN dots grown by FME and by MOCVD are illustrated in Fig. 7-6. The corresponding PL FWHM and peak energy against background flow are also drawn in the figure. The PL results seem coincide well with the results of dot structure parameters in Fig. 7-5, where N-rich and In-rich regions are allocated previously in accord with their growth behaviors. Not surprisingly, the PL linewidth decreases significantly from 152 to 64 meV as background NH3 flow is decreased from 10,000 to 1,000 sccm in N-rich growth region, and seems almost unchanged in the so-called In-rich region. The narrowest linewidth, ∼63meV, occurred at 500 sccm NH3 background flow, is the best value ever reported for InN nanodots.

Beside the variation of PL linewidth, the improvement of PL intensity in In-rich growth region is also astonishing. The PL intensities of InN dot samples grown here are generally higher than in N-rich region. The maximum PL intensity occurred at NH3 background flow of 500 sccm has a magnitude almost 50 times higher than dots prepared at 10,000 sccm background flow and 15 times higher than MOCVD-grown InN dot sample. Such an improvement in optical quality can be attributed at least

partially to the auto-surfactant effects of In adlayer under In-rich growth conditions, which enhance adatom mobilities, planarize the top surfaces of truncated pyramid islands, and might also lead to reductions of point defects, such as indium vacancies (VIn) and indium interstitials (Ini), and stacking defaults in the dots.

7-4 Conclusions of InN Nanodots growth by FME

In summary, comparisons between InN dots grown by FME and by MOCVD have been comprehensively performed in terms of the optical and morphological structure properties. In comparison with MOCVD, the FME InN dots growth in the temperature range of 550-700℃ appears to have lower adatom diffusion activation energy (0.7 vs. 1.3 eV) and lower dot density, which gives rise to better optical quality InN dot samples in this region, owing to the enhanced adatom migration mobility in such type of deposition scheme. More interesting results were observed for FME InN dot samples grown with different NH3 background flows. The samples prepared under low NH3 backgrounds (<1,000 sccm), hence In-rich growth conditions, generally exhibit narrower and more intense photoluminescence signals than ones prepared under N-rich conditions (>1,000 sccm). For instance, the PL intensity of InN dots grown under a NH3 background flow of 500 sccm is increased by nearly a factor of 50 as compared to that under a NH3 background flow of 10,000 sccm, where the growth is conducted in the regime of N-rich condition.

References

[1] V. W. L. Chin, T. L. Tansley, and T. Osotchan, J. Appl. Phys. 75, 7365 (1994).

[2] B. E. Foutz, S. K. O’Leary, M. S. Shur, L. F. Eastman, J. Appl. Phys. 85, 7727 (1999).

[3] Ashraful Ghani Bhuiyan, Akihiro Hashimoto, and Akio Yamamoto, J. Appl. Phys.

94, 2779 (2003).

[4] W. C. Ke, C. P. Fu, C. Y. Chen, L. Lee, C. S. Ku, W. C. Chou, W.-H. Chang, M. C.

Lee, W. K. Chen, W. J. Lin and Y. C. Cheng, Appl. Phys. Lett. 88, 191913 (2006).

[5] A. Yoshikawa, N. Hashimoto, N. Kikukawa, S. B. Che, and Y. Ishitani, Appl. Phys.

Lett. 86, 153115 (2005).

[10] J. G. Lozano, A. M. Sánchez, R. García, D. González, D. Araújo, S. Ruffenach and O. Briot, Appl. Phys. Lett. 87, 263104 (2005).

[11] S. Ruffenach, B. Maleyre, O. Briot, and B. Gil, Phys. Stat. Sol.(c) 2, 826 (2005).

[12] V. N. E. Robinson and J. L. Robins: Thin Solid Films 20 (1974) 155.

[13] Pierre Ruterana, Martin Albrecht, Jörg Neugebauer, in Nitride Semiconductors Handbook on Materials and Devices, edited by WILEY-VCH Verlag GmbH (Germany, 2003), p. 247.

[14] Ming-Chih Lee, Heng-Ching Lin, Yung-Chung Pan, Chen-Ke Shu, Jehn Ou, Wen-Hsiung Chen, and Wei-Kuouka Chen, Appl. Phys. Lett. 73, 2606 (1998).

[15] W. L. Chen, R. L. Gunshor, J. Han, K. Higashimine, and N. Otsuka, MRS Internet J. Nitride semicond. Res. 551, W3.30 (2000).

[16] J. Wu, W. Walukiewicz, W. Shan, K. M. Yu, J. W. Ager III, S. X. Li, E. E. Haller, Hai Lu, and William J. Schaff, J. Appl. Phys. 94, 4457 (2003).

[17] Walukiewicz, W.; Li, S.X.; Wu, J.; Yu, K.M.; Ager III, J.W.; Haller, E.E.; Lu, Hai;

Schaff, William, J. Cryst. Growth 269, 119 (2004).

[18] Ken-ichi Shiramine, Tomohiko Itoh, Shunichi Muto, Tamotsu Kozaki, Seichi

Sato, J. Cryst. Growth 242, 332 (2002).

[19] E. Dimakis, E. Iliopoulos, K. Tsagaraki, Th. Kehagias, Ph. Komninou, and A.

Georgakilas, J. Appl. Phys. 97, 113520 (2005).

Fig. 7-1 The typical gas flow sequence of the formation of InN dots for FME and conventional MOCVD.

flow rate (sccm)

NH3

10000 18000

Time 0

5000

TMIn 150

0

NH3

TMIn 150

10000

# FME

# MOCVD r0

0

Table 7-2: The average height and diameter of InN dots grown at different growth temperatures from 550 to 730℃.

550 ℃ 600 ℃ 650 ℃ 700 ℃ 715 ℃ 730 ℃

MOCVD (nm) 13/102 30/154 33/207 36/281 71/413 80/457 FME (nm) 16/151 39/202 35/258 52/226 62/203 84/191

*H/D = Height/Diameter Method

Temperature H/D

1.00 1.05 1.10 1.15 1.20 1.25 10

10

10

10

10

1.00 1.05 1.10 1.15 1.20 1.25 10¹¹

10¹² 10¹³ 10¹⁴

MOCVD