Structural characterisation of MBE grown zinc-blende Ga1-xMnxN/GaAs(001) as a function of Ga flux. In: Microscopy of semiconducting materials: proceedings of

: Ga 1-x Mn x N films grown on semi-insulating GaAs(001) substrates at 680°C with fixed Mn flux and varied Ga flux demonstrated a transition from zinc-blende/wurtzite mixed phase growth for low Ga flux (N-rich conditions) to zinc-blende single phase growth with surface Ga droplets for high Ga flux (Ga-rich conditions). N-rich conditions were found favourable for Mn incorporation in GaN lattice. α-MnAs inclusions were identified extending into the GaAs buffer layer.


INTRODUCTION
III-V ferromagnetic semiconductors are of interest because of their potential application within spintronic device structures (Wolf et al 2001). Theoretical prediction of the Curie temperature for various semiconductors (Dietl et al 2000) suggests that a T C value above room temperature is possible for zinc-blende GaN containing 5 at% Mn and a hole concentration of 3.510 20 cm −3 . In view of the limited solid solubility of Mn in GaN, it becomes necessary to use non equilibrium growth techniques such as plasma-assisted molecular bean epitaxy (PAMBE) to establish appropriate conditions for the growth of uniform Ga 1-x Mn x N alloys. To date, high p-type Ga 1-x Mn x N layers with carrier concentrations exceeding 10 18 cm -3 have been obtained by PAMBE (Novikov et al 2004).
Earlier work on the growth of zinc-blende GaN suggests that exact control of the III:V ratio close to the stoichiometric condition allows the production of single phase zinc-blende epitaxial layers, whilst deviation to Ga or N-rich conditions reportedly produces mixed zinc-blende and wurtzite material Giehler et al 1995;Ruvimov et al 1997). More recently, various Mn-N or Ga-Mn-N precipitations have been reported for wurtzite GaN epilayers grown on sapphire substrates (e.g. Kuroda et al 2003 andNakayama et al 2003).
In this paper, the influence of the Ga:N ratio on the microstructural development of Ga 1- x Mn x N/GaAs(001) grown by PAMBE is assessed using a variety of complementary analytical techniques.

EXPERIMENTAL
Zinc-blende Ga 1-x Mn x N epilayers were grown on semi-insulating (001) oriented GaAs substrates at 680°C by PAMBE. Briefly, a GaAs buffer layer of thickness ~0.15µm was deposited to provide a clean surface for epitaxy. Following initiation of the N plasma, the Mn and N shutters were opened whilst the As shutter was closed. The Mn flux was fixed at a level of 1.010 -8 mbar while the Ga:N ratio was varied by changing the Ga flux from 7.510 -8 mbar to 1.2x10 -6 mbar. This Published in Springer Proceedings in Physics, Springer-Verlag Springer Proc. Phys. 107 (2005) pp 155-158 Microscopy of Semiconducting Materials Conference XIV, MSM 2005 corresponded to a transition from N-rich to Ga-rich conditions, with the latter being identified by the development of Ga droplets on the growth surface. An overall chamber pressure of 2-310 -5 mbar was maintained by a flow of N 2 . The growth conditions for the sample set are summarised in Table 1.
The bulk and fine scale defect microstructure of each sample was assessed. A Philips X-pert Diffractometer was initially used to assess the bulk crystal structure of the deposited epilayers. The complementary technique of reflection high energy electron diffraction (RHEED) using a modified Jeol 2000fx transmission electron microscope, with as-grown or HCl etched specimens mounted vertically, immediately beneath the projector lens, was then applied to appraise the sample near surface microstructure. Sample morphology was assessed using an FEI XL30 scanning electron microscope operated at 15-20kV. Samples for TEM investigation across the stoichiometric range were prepared in plan-view and cross-sectional geometries using sequential mechanical polishing and argon ion beam thinning. Samples were assessed using conventional diffraction contrast techniques using Jeol 2000fx and 4000fx instruments and energy dispersive X-ray (EDX) analysis using an Oxford Instruments ISIS system.

RESULTS AND DISCUSSION
The formation of zinc-blende Ga 1-x Mn x N was confirmed by XRD spectra obtained across the sample set. Variation in the full width at half maxima (FWHM) values for the 002 reflection across the stoichiometric range (Table 1) suggests that the layer structural quality becomes optimised for conditions of slightly Ga rich growth. However, no evidence for the presence of second phase wurtzite material was discerned for any of the spectra. As observed using SEM, the sample grown closest to ~1:1 stoichiometric conditions appears specular, indicative of a smooth surface. Samples grown under N-rich conditions appear to exhibit a slightly rougher surface, whilst samples grown under Ga-rich conditions showed increasing amounts of Ga droplets on the sample surface with increasing Ga flux.
RHEED patterns recorded along <110> projections for samples A, D and G are presented in Fig. 1(a-c). It is noted that clear, sharp spots was only obtained for the Ga-rich samples after removal of surface Ga droplets using boiling HCl. All the samples demonstrated the cubic structure with extra spots and/or streaks indicating varying degrees of mixed phase growth and stacking disorder on inclined {111} planes. In particular, a transition from mixed hexagonal/cubic (α/) phase growth for N-rich conditions to single phase cubic material for Ga-rich conditions was observed (as distinct from the previous indications of XRD).
By way of example, for sample A grown under N-rich conditions, dominant diffraction spots from both cubic and hexagonal material were identified (Fig. 1a). The indexing of Fig. 1a is clarified with reference to the schematic diagram of Fig. 1d which illustrates the orientation relationship between the two phases, with <110>  // <11 2 0>  and {111}  // {0001}  . It is noted that the extra spots due to the hexagonal phase became faint with increasing Ga flux, disappearing when the Ga:N ratio approached 1:1 stoichiometry (Fig. 1b).
For samples grown under N-rich conditions and ~1:1 stoichiometry, streaks preferentially aligned along one <111> direction were also observed, indicating the preferential alignment of planar defects (i.e. thin microtwins and stacking faults) inclined to the growth surface on just one set of {111} planes (Figs 1 a and b). Similar streaks were observed along both <111> directions for samples grown under Ga-rich conditions, again attributable to a high density of inclined planar defects (Fig. 1c). It is noted that samples grown under N-rich and nearly 1:1 stoichiometric conditions exhibited strong anisotropy in the distribution of planar defects, being present for just one <110> sample projection, whilst samples grown under Ga-rich conditions exhibited planar defects for both orthogonal <110> and <110> sample projections. This variation in the anisotropic distribution of planar defects suggests that this effect is associated with the transition from N-rich to Ga-rich growth, i.e. due to differences in III:V stoichiometry at the growth surface during the process of epilayer nucleation, rather than being due to slight vicinality of the substrate surface. In addition, the presence of streaks perpendicular to the shadow edge of samples grown under Ga-rich conditions (Fig. 1c), following HCl etching, are attributed to patches of relatively smooth surface. More precisely, however, the diffraction effect of streaks perpendicular to the growth surface is attributed to the material that is not perfectly flat, but with slight local misorientations combined with some degree of surface disorder (Cowley 1992).
Overall, the indication from these RHEED patterns together with XRD spectra and SEM observation is that nearly 1:1 stoichiometry (or slightly Ga-rich conditions) correspond to an optimised microstructure. Fig. 1e shows a centred dark field image formed from a diffraction spot attributed to only wurtzite Ga 1-x Mn x N, as distinct from a overlap of spots due to wurtzite Ga 1-x Mn x N and microtwin spots from the zinc-blende Ga 1-x Mn x N located at 1/3<111> positions. This indicates the localisation of small grains of wurtzite Ga 1-x Mn x N at the growth surface. However, the overlap from stacking fault streaks through the objective aperture, due to slight imaging beam convergence, also contributes to this dark field image, partially highlighting the stacking disorder on one set of {111} planes. Since selected area diffraction experiments provided no evidence for the presence of wurtzite domains through the bulk of the epilayer and no evidence was found for hexagonal phase material at the epilayer/substrate interfaces, the formation of wurtzite Ga 1-x Mn x N are attributed to a cool down effect at the end of growth whereby a slight change in surface stoichiometry might have occurred under Nrich conditions, allowing small grains of the more stable hexagonal phase to be established. The small volume fraction of these surface hexagonal grains explains why there were not detectable by XRD.
EDX measurements from the epilayers during TEM observation indicated a variation in the Mn content across the sample set, with a relatively uniform Mn content of ~3.3at% for sample A, peaking at a value of 40.3% for sample D, while the Mn content was below the detectability limit of EDX for samples grown under Ga rich conditions. This is consistent with reports of MBE grown wurtzite Ga 1-x Mn x N/sapphire which demonstrate that N-rich (and Mn-rich) conditions are required for the successful incorporation of Mn into the crystal lattice (Haider et al 2003;Kuroda et al 2003), as assessed using EDX and SIMS respectively.
By way of illustration, Fig. 2a presents a dark field image of Sample A, demonstrating the highly faulted nature of the epilayer, and pyramidal precipitates (arrowed) extending into the GaAs buffer layer. EDX measurements confirmed the presence of Mn and As within such inclusions (Fig.  2c), whilst associated selected area electron diffraction patterns (Fig. 2b) confirmed that the inclusions comprised α-MnAs. The indexing of Fig. 2b is clarified with reference to the schematic diagram of Fig. 2d. The orientation relationship here between α-MnAs and GaAs is given by <11 2 0> MnAs // <110> GaAs and {0001} MnAs // {111} GaAs . It is emphasised that such MnAs inclusions extending into the buffer layer were identified within all the samples with decreasing size upon transition to Ga-rich growth conditions. No evidence for Ga-Mn-N or Mn-N inclusions was found in these samples. In view of the very different levels of hardness of the epilayer and substrate, it is considered that voids present within the GaAs buffer layer as marked in Fig. 2a arise due to preferential ion beam milling of localised strain centres. However, some co-operative mechanism associated with MnAs precipitate formation during the process of growth might also be implicated in their initial formation.
In summary, N-rich conditions are required for the incorporation of Mn within Ga 1-x Mn x N, whilst slightly Ga-rich conditions are associated with optimised structural properties. All samples exhibited MnAs inclusions extending into the GaAs buffer layer, arising from the limited solid solubility of Mn in GaN.