physica

status

phys. stat. sol. (b) 245, No. 5, 903 906 (2008) / DOI 10.1002/pssb.200778610

pss
solidi
www.pss-b.com

Electron irradiation and the equilibrium of open core dislocations in gallium nitride
M. Hawkridge*, 1, D. Cherns2, and Z. Liliental-Weber1
1 2

basic solid state physics

Lawrence Berkeley National Laboratory, 1 Cyclotron Road MS 62R0209, Berkeley, CA 94720, USA H. H. Wills Physics Laboratory, University of Bristol, Tyndalls Avenue, Bristol BS8 1TL, UK

Received 8 September 2007, revised 28 February 2008, accepted 28 February 2008 Published online 15 April 2008 PACS 61.72.Ff, 61.72.Qq, 61.72.uj, 61.72.Yx, 61.80.Fe
*

Corresponding author: e-mail [email protected], Phone: +1 510 486 6943, Fax: +1 510 486 4995

Under electron irradiation, nanopipes in GaN are found to evolve into the so-called bamboo structure and eventually into a chain of voids [Pailloux et al., Appl. Phys. Lett. 86, 131908 (2005)]. Here, the driving mechanism for this morphological evolution is examined using transmission and scanning transmission electron microscopy studies of undoped GaN grown by hydride vapor phase epitaxy. Irradiation at varying beam energies shows that morphological evolution occurs below the

threshold for direct knock-on damage, and that voids migrate towards the beam. It is proposed that these observations can be explained by the stimulated diffusion of point defects. In areas relatively unexposed to electrons, a survey of core character, diameter and depth into the layer suggests that a similar process occurs by post-growth annealing, i.e. due to point defects mobile at the growth temperature. The implications of these results for understanding dislocation core structures are assessed.

2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1 Introduction Several studies using transmission and scanning transmission electron microscopy (TEM and STEM, respectively) have indicated that impurities and dopants play a crucial role in determining the morphology of open core dislocations in GaN ([13] and references therein). However, Pailloux et al. [4] found that prolonged exposure to 300 kV electrons in a TEM caused nanopipes in undoped GaN to evolve into a so-called bamboo structure, where the lateral surfaces are partially constricted at periodic intervals, eventually pinching-off to form a string of voids configuration. The implication is that the core morphologies observed in cross-section may be the result of electron exposure during observation, rather than by a consequence of growth. However, the distribution of core morphologies before the exposure to electrons was unknown and the mechanism driving the instabilities was not fully determined. In this paper, we examine the equilibrium of open core dislocations in GaN using STEM and high-voltage TEM. By varying the beam energy and examining the projected movement of open cores, it is shown that diffusion of point defects either in the bulk or on the surfaces causes the instabilities that drive morphological evolution during irradia-

tion. These results are applied to the distribution of the diameter, depth and morphology of open cores in unexposed specimens, and strong evidence is found to suggest that thermal diffusion of point defects during growth plays a role in determining the morphology of open core dislocations. 2 Experimental Undoped GaN layers up to 55 m thick grown on (0001) sapphire by hydride vapor phase epitaxy (HVPE) were prepared for TEM analysis using mechanical polishing followed by low energy (<4 keV), low angle (<6) ion milling. Cross-sectional TEM experiments were performed in a Philips EM430 at Bristol, capable of providing accelerating voltages up to 300 kV in 50 kV steps and tilts up to = 60. The same 100 m condenser aperture was used throughout. During observation, the flux of electrons was less than 1019 cm2s1. To irradiate open cores, the beam was condensed to 100 nm diameter giving a flux of 1022 cm2s1 for 15 minutes. Plan-view specimens were examined using SuperSTEM 1, at the Daresbury labs, UK (probe size 1 , beam energy 100 keV, specimen flux 1025 cm2s1). All experiments were conducted at room temperature.
2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

pss
status solidi

904

physica

M. Hawkridge et al.: Electron irradiation and equilibrium of open core dislocations in GaN

3 Results and discussion 3.1 Electron irradiation processes Irradiation at varying beam energies showed that morphological evolution occurs below the threshold for knock-on damage. Morphological evolution of nanopipes into voids was observed for beam energies as low as 50 keV, well below the knock-on threshold of 185 keV [5]. Figure 1 shows one such nanopipe before and after irradiation at 100 keV. Following irradiation, the nanopipe walls are constricted at periodic intervals so that it resembles a bamboo structure. Outside the irradiated area, there is little change in the nanopipe morphology. The projected motion of irradiated cores shows that they move opposite to the beam direction. Images were taken at = 45 angles to the surface of the foil both before and after irradiation, with the irradiation occurring parallel to one of the imaging directions. This allowed the projected motion of the cores to be mapped. Figure 2 shows the projected motion of a series of voids isolated from a dislocation. The circle of lighter contrast identifies where sputtering has occurred from the bottom surface of the foil. Outside the irradiated area and before irradiation (data not shown, but distribution similar), the voids are spread about a mean position perpendicular to both beam directions. Following irradiation, the distribution of void positions has a tail away from the circle of lighter contrast parallel to the irradiation direction, i.e. opposite to the beam direction. In addition, the voids have ripened by an Ostwald process. The observations made in cross-section TEM can assist in explaining observations of nanopipe core collapse made in plan-view STEM. Figure 3(a) shows a nanopipe imaged end-on in high angle annular dark field (HAADF) before prolonged exposure to the beam. The core is clearly visible in the GaN matrix. After several minutes of exposure to the beam, the nanopipe walls are difficult to identify and the core begins to show signs of crystalline material with the

Figure 2 Projection of voids irradiated at 250 keV. Insets show beam directions relative to the foil. a) = + 45, b) = 45. The histograms beneath each image show the distribution of voids outside and inside (*) the irradiated area. Bin width = 10 nm.

same crystallography as the surrounding matrix. Eventually, the nanopipe is hardly recognizable (Fig. 3(b)). If the nanopipe were to become unstable and constrict as observed in cross-section, there would be a movement of crystalline material into the core, consistent with what is observed. Lateral migration perpendicular to the beam, if it occurred, would add to this effect. It should be noted that there is no indication of amorphous material in the core following exposure to the beam. The results indicate that knock-on damage is not significant to the process by which morphological evolution occurs. It may be the case that, above the knock-on threshold, there is some contribution to this process by atomic displacement in the bulk. However, this effect is expected to be minimal [6]. Hydrogen-mediated knock-on effects [7] are not expected in bulk hexagonal GaN [5]. There is no contribution by thermally stimulated point defects during irradiation. Thermal heating by the beam is negligible (0.2 K) [8] and the diffusion length at room temperature for the irradiation period is l 1027 cm using values determined by Ambacher et al. [9] for the temperature dependant diffusion coefficient, i.e. the point defects

Figure 1 (Left) Nanopipe before irradiation. (Right) After irradiation at 100 keV. The nanopipe has become constricted in the irradiated area. The contrast is somewhat obscured in the after image due to the build-up of surface contamination in the condensed beam during irradiation. However, the contrast-enhanced insets show the lateral constrictions more clearly (arrowed).
2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Figure 3 HAADF STEM images of a nanopipe collapse in plan view after a) 0 min and b) 10 min exposure. The insets show a representation of the core condition in cross-section.
www.pss-b.com

Original Paper
phys. stat. sol. (b) 245, No. 5 (2008) 905

are effectively immobile during irradiation due to thermal mobility alone. The projected motion of voids indicates that stimulated diffusion is the driving mechanism for the morphological evolution. The momentum transfer from the beam to the specimen is strongly peaked in the forward direction. As a result, atoms generally move in the direction of the beam and vacancies move opposite to the beam. The net result is that voids move opposite to the beam direction, consistent with what is observed. However, sputtered material is expected to re-deposit as amorphous material on the opposing surfaces of the void. This is inconsistent with the observations of core collapse in plan-view STEM. Indeed, the collapsing core retained a strong crystalline character and it is highly unlikely that any re-deposited material will crystallize at room temperature. The remaining mechanism consistent with the projected motion of irradiated voids is stimulated diffusion. This result is consistent with the Ostwald ripening of the voids and observations of enhanced self-diffusion under electron irradiation by Mera et al. [5]. The contribution of surface and bulk diffusion is not yet clear, but surface diffusion is far more likely as the energy barrier is lower. 3.2 Post-growth processes A survey of open core structures in specimens relatively unexposed to electrons showed a correlation between core character, depth and diameter. Open cores were classified into three categories determined by the level of constriction: none detectable (nanopipe), partial (bamboo) and full (voids) constriction at regular intervals. Core diameter was determined from micrographs by measuring the unconstricted width at several points in cross-section. The depth in the layer was measured from the surface or substrate interface where visible in micrographs, and estimated otherwise during observation in the TEM. Depth is defined here as the midpoint of the structure. The survey results are shown in Fig. 4. There is a clear correlation between diameter and character, with voids being more frequent at larger diameters and nanopipes more frequent at smaller diameters. Similarly, nanopipes are more frequent near the layer sur-

face and vice versa for voids. Bamboo structures show no strong correlation to depth or diameter. The survey data indicates that the evolution process observed during irradiation may also be occurring during growth due to thermally stimulated point defects. Energy calculations by Pailloux et al. [4] imply that larger radii nanopipes should be unstable under irradiation. The correlation between core diameter and character is consistent this implication, but the structures surveyed were not exposed to electron dosages significant enough to cause morphological evolution. Alternatively, it may be the case that thermally stimulated point defects are causing the evolution during growth. This process is analogous to annealing and is most likely at the temperatures used in HVPE growth (diffusion length l 100 at 1000 C [9]). The correlation between core character and depth is consistent with this implication. During growth, the region nearest the substrate will have been at the growth temperature for the longest period, with the time spent at the growth temperature decreasing towards the newly grown material at the surface. Hence, if point defects are mobile at the growth temperature, newly formed material will have been exposed to smaller doses of point defects compared to material near the substrate. Assuming that this is the case, these results imply that increasing dosage of thermally stimulated point defects has the same effect as increasing dosage of irradiation stimulated point defects. Together, these results imply that the periodic core structures observed in cross-section may be the result of a post-growth instability, rather than an as-grown process involving impurities at the growth surface. Post-growth instabilities should therefore be considered when discussing the formation of periodic open core defects. That bamboo structures show no significant correlations may indicate that the evolution from nanopipe to voids occurs quickly during growth. 4 Conclusion The equilibrium of open core dislocations in HVPE grown GaN layers were studied by TEM and STEM methods. The projected motion of open cores irradiated at various beam energies showed that the morphological evolution of nanopipes occurs due to the stimulated diffusion of point defects. A survey of core types in relatively unexposed specimens showed that a similar morphological evolution may be occurring post-growth due to thermally stimulated point defects. The key implication is that periodic open core morphologies may not form due to an as-grown process as previously thought.
Acknowledgements This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, Materials Sciences and Engineering Division, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231 and the Office of Naval Research (Colin Wood) under grant No. N0001403-1-0579. We are also grateful to D. C. Look and R. Molnar for the provision of samples, and to A. Bleloch for use of the SuperStem.
2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Figure 4 Statistical data from survey of core character, diameter and depth within the GaN layer.
www.pss-b.com

pss
status solidi

906

physica

M. Hawkridge et al.: Electron irradiation and equilibrium of open core dislocations in GaN

References
[1] D. Cherns, J. Phys.: Condens. Matter 12(49), 10205 (2000). [2] D. Cherns, Y. Q. Wang, R. Liu, and F. A. Ponce, Appl. Phys. Lett. 81, 4541 (2002). [3] M. Hawkridge and D. Cherns, Appl. Phys. Lett. 87, 221903 (2005). [4] F. Pailloux, J. Colin, J. F. Barbot, and J. Grilhe, Appl. Phys. Lett. 86, 131908 (2005).

[5] Y. Mera, K. Suzuki, and K. Maeda, Physica B 340, 488 (2003). [6] Y. Ohno, S. Takeda, T. Ichihashi, and S. Iijima, J. Appl. Phys. 99, 126107 (2006). [7] G. M. Bond, I. M. Robertson, F. M. Zeides, and H. K. Birnbaum, Philos. Mag. A 55, 669 (1987). [8] Calculated using classical thermal conduction theory. [9] O. Ambacher, F. Freudenberg, R. Dimitrov, H. Angerer, and M. Stutzmann, Jpn. J. Appl. Phys. 37, 2416 (1998).

2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

www.pss-b.com