Pruebas realizadas en hoyos a diferentes profundidades y a diferentes

Prior to epitaxial growth of a high quality film, it is necessary to have a clean surface.5-8 Of the methods available, there are two principle categories: in situ and

ex situ treatments. While ex situ chemical treatments are simple to implement and have been reported to be successful in thinning the oxide layer that forms on a GaSb surface, contaminants are likely to persist on the wafer, either from the original oxide layer, or those introduced by solution processing. In the case of complete removal of the oxide layer, exposure to anything other than an inert atmosphere upon removal from solution will lead to further oxide formation. This assertion is supported by Kodama et al., who report that a 20-30 Å oxide layer remains after a CH3COOH:HNO3:HF etch, citing high surface reactivity as the cause.9

In situ techniques are preferable as they do not require exposure to the atmosphere after cleaning, prevent particulates alighting upon the substrate during movement inside or outside the growth system. Of these methods, thermal desorption has been widely used on GaSb,10-13 as well as other substrates.14-18 As a technique, it has the benefits that it lacks the requirement of any additional equipment and may be performed over a timescale of minutes. Baba et al. report, however, that carbon species may persist following thermal treatment of substrates, indicated by depletion of carrier concentration.19 Furthermore, it has been reported that thermal oxide desorption can result in roughening of a GaAs surface,20 suggesting that care must be taken with the less resilient antimonide. However, where a buffer layer is to be applied, it may be possible to recover the surface.

In the case of GaSb, substrate temperatures exceeding 300°C are sufficient to alter the stoichiometry and morphology of the (100) surface, although the oxide layer persists at higher temperatures.21 As such, thermal oxide removal under an overpressure of antimony may present a viable option, analogous to oxide removal

As a result of damage resulting from thermal cleaning, there has been investigation into other in situ cleaning, involving ion, radical or plasma processes, including H- plasma, H-gas and H-radical cleaning.19 Compared to thermal desorption alone, each of the hydrogen methods are cited as providing a cleaner surface, although radiation damage is evident in the sample exposed to plasma. Ion sputtering may also be employed as a cleaning technique, although this induces surface patterning and morphological changes.23,24 However, ion bombardment conducted at a glancing angle with respect to the substrate has demonstrated promise in application on GaAs, resulting in a flat surface.25

RHEED observations have resulted in reports detailing a number of surface reconstructions for the oxide-free GaSb (001) surface, with the following having been confirmed by STM studies: (1 x 5), (2 x 5), c(2 x 10), c(2 x 6) and (1 x 3).13,26 While there are conflicting reports as to the boundary conditions for the transitions between these reconstructions, there are two aspects on which there is consensus. The first is that the (n x 5) and c(2 x 10) reconstructions are observed in a more antimony-rich regime, whereas the c(2 x 6) and (1 x 3) result from a lower density of surface antimony species.* The second point is that transitions between these surface phases have strong dependence on both the incident antimony flux and surface temperature,27 although the experiments conducted in chapter 3 illustrate the difficulties that may be encountered with respect to accurate and reproducible determination of temperature.

Among III-V films, the (n x 5) reconstructions are thought to be unique to GaSb, due to the near equivalence of the unit cell size of gallium antimonide with the lattice parameter of trigonally-bonded Sb.13

Reconstructions are reported as occurring under different fluxes and substrate temperatures. These are likely to arise from the measurement inconsistencies endemic in inter-laboratory comparison, as discussed in chapter 3. It is not always clear how the V:III ratios have been determined, which increases difficulties associated with translating parameters from one growth system to another. It is generally considered that expressing material fluxes as a function of limiting growth rates instead of beam equivalent pressures provides a more consistent means of quantifying the supply of material to the substrate. Furthermore, the clarity of the RHEED image is also affected by the apparatus with which it is observed, the quality of the camera used for acquisition, and the presence of any adsorbed species on the interior of the phosphor-coated window.

GaSb has been grown on a number of substrates, including Si, GaAs and GaSb.2,28-33 The substrates used for growth each have associated benefits and drawbacks: while GaAs and Si are relatively inexpensive and can be produced with a lower density of inherent defects than GaSb, their lattice constants differ significantly (5.430Å and 5.653Å respectively) from that of GaSb, resulting in dislocations that propagate through the grown film.29

In terms of MBE growth, antimony has been described as a “nuisance” element, due to its low vapour pressure.1,2,32 There is a tendency of antimony atoms to aggregate, causing Sb vacancies and GaSb defects.2,30,34 This process is cited as the reason for

reported that epitaxial growth of GaSb, regardless of substrate, must be conducted under an excess of antimony.2 As an aside, it should be noted that n-type GaSb layers have been reportedly formed by liquid phase epitaxy (LPE) at temperatures > 400°C.2,36

When Ga-rich flux ratios are used, Lee et al. report a progressive decline in the intensity and clarity of the (1 x 3) RHEED pattern associated with an antimony- stabilised GaSb surface above 400°C, until there is no observable pattern.28 This suggests formation of an amorphous surface with gallium aggregation, which is supported by reports in the same publication of rough surfaces with a metallic lustre. While an antimony rich flux is required, Lee et al. refer in the same publication to the near minimum excess of antimony, reporting a decrease in PL efficiency for Sb4:Ga ratios above this value.

Upon sublimation, antimony is generally produced as Sb4. Epitaxial growth may proceed also with monomeric or dimeric Sb by means of a cracker, though the populations of the different species are often assumed from the cracker temperature regime or inferred from published data,36 rather than directly measured.27 The latter method is hindered by the difference in ion gauge sensitivity between the diatomic nitrogen most commonly used for calibration and antimony. Additionally, the hot filament common to all ion gauges can cause thermal cracking of both Sb2 and Sb4 species. It is reported that growth with Sb4 proceeds with a second-order dependency upon antimony flux.12 This is in common with growth of GaAs with As4 species, where the second-order characteristics are caused by a two-body dissociative process.15

As with reported observations of surface reconstructions, published growth conditions vary, especially with respect to measurements of antimony beam equivalent pressure and substrate temperature. The former results from the variation between experimental systems and the inclusion (or absence) of relative sensitivity factors in flux calculations, which is rarely stated explicitly in the literature. It has been demonstrated in chapter 3 that values resulting from temperature measurement may vary according to the method of determination. With these factors in mind, it was considered important to establish lab-dependent parameters, using published data only as a guide.

In the case of MBE growth undertaken in a chamber in which arsenic has been used, it is possible that some As incorporation will occur.37 This has been reported where arsenic has been deposited on a heat-cycled piece of apparatus, such as the antimony cracker. Upon heating, sublimation may effectively create a secondary As source. However, this may decrease as a function of time of operation of the antimony cracker and is less likely to be present at significant levels where the antimony cracker tip is kept at elevated temperatures when not in use.36

For GaSb growth temperatures of 550°C, which is the optimum recorded by Lee et al., and a group III limited growth rate of 1 µm h-1, the minimum excess in terms of Sb4:Ga flux is reported to be ca. 2.2:1. Lee et al. state that the free exciton transition at 810 meV was not observed in any PL experiments, suggesting that the optical quality of the films grown was relatively poor.

There are reports that native defect concentrations may be decreased by use of Sb2 in 36

undergo incorporation with first-order dependence and do not need the larger atomisation energy of Sb4 from the surface, thus allowing both a lower growth temperature and enhancing the reaction with surface gallium species. There is no significant difference in hole concentration or mobility between films grown using Sb and Sb2 species.38

Of the growth schemes for GaSb/GaAs and GaSb/GaSb reported the key facts are summarised below:

Substrate TSub (°C) Sbx:Ga Ratio Growth Rate Ref.

Te-GaSb Cr-GaAs 500-600 500-600 - - 0.6-2.5 µm h-1 0.6-2.5 µm h-1 32 - 500-600 - 0.6-1.5 µm h-1 2 GaAs (001) GaSb (001) 480-620 550 0.7-6.5:1 2.5:1 1.0 µm h-1 1.0 µm h-1 28 Cr-GaAs SI (001) 500 2:1 (x = 4) 0.5 µm h-1 31 GaSb 470-550 1.56-4.52:1 (x = 2,4) 0.8 µm h-1 37 GaSb (001) 380-550 1.2-4 (x = 1,2) 1.1 µm h-1 33

Ignoring the different methods that have been used to measure the conditions, particularly in the case of V:III ratios, the literature indicates that the oxide layer may be removed by thermal cleaning, although a stabilising antimony flux would be required to prevent degradation of the surface. Furthermore, it appears that successful growth must employ some excess of antimony, relative to the gallium flux supplied to the substrate.