KOH蚀刻对镓面极性n-GaN表面性质的影响

1. Introduction  

The performance of GaN-based devices is significantly influenced by the quality of the metallic  contacts used to make connection to power sources and/or other devices [1]. The electrical properties  of metal/n-GaN contacts depend on a wide variety of processing parameters, such as cleaning and  etching of the GaN surface [2, 3], the composition and thickness of the metallic contacting layers, and  the annealing temperature and atmosphere used to activate the contact. Because enhanced diffusion  couple metallization layers and annealing conditions have been identified to gain improved contact  performance, increased attention is also being given to the preparation of GaN surfaces prior to  metallization. For example, previous work has shown that ohmic contacts of lower resistance can be  obtained by making contact to plasma or wet-etched surfaces [4].

Detailed investigations are required to appraise the intimate relationship between the chemical  treatment employed, the resulting chemical/structural modifications to the GaN surface and the  electrical properties of the contacts. Whilst detailed understanding has been gained on the effects of  surface cleaning, for example HCl [5] or HF [6] treatments, much less is known about the  consequences of surface etching, for example KOH or plasma treatments, partly because of the dual  action of the removal of GaN material combined with surface cleaning. One approach to address this  problem is to compare cleaned and etched surfaces in order to assess the consequences of the removal  of GaN material on contact performance. Furthermore, in order to appraise the effects of GaN surface  treatment on the performance of electrical contacts, it is helpful to use metallic contacts that do not  alloy with GaN, such as Au [7] or Pt [8], thus avoiding further modification to the GaN material.

The present work studies the mechanisms that determine the barrier height by investigating the  effects of KOH etching on the properties of n-GaN surfaces and associated Au/n-GaN Schottky  contacts. To investigate the role of KOH etching beyond surface cleaning, the KOH treatment was  used in conjunction with an HCl clean (samples denoted as 'KOH treated') and compared with a  reference HCl clean (samples denoted as 'reference'). Hence, reference and KOH-treated n-GaN  surfaces have been investigated, initially with regard to sample surface composition, bonding states,  morphology and crystalline structure. Consideration has then given to the electrical properties of Au  Schottky contacts deposited onto these treated n-GaN surfaces, in order to appraise the respective  electron barrier heights and the extent of non-radiative recombination activity. In this context, surfaces  of Ga-polar GaN/sapphire grown by molecular beam epitaxy (MBE) and metal-organic chemical  vapour deposition (MOCVD) are also compared.  

2. Experimental  

The starting MBE GaN/sapphire wafer was 2.5 µm thick and exhibited an n-type carrier concentration  of 6.4 x 1016 cm.-3 The MOCVD GaN/sapphire (0001) wafer was 3 µm thick with a 0.9 µm thick Sidoped capping layer and an estimated n-type carrier concentration of 1 x 1017 cm.-3 Cleaved samples  were first sequentially degreased using a standard procedure of ultrasonic baths of lotoxane, methanol,  acetone, propanol and de-ionized water, for three minutes each. The subsequent KOH treatment  consisted of sample dipping in a 4 molar solution of KOH and de-ionized water for 1 minute at 60ºC,  followed by a 1 minute dip in de-ionized water, then blowing dry using N2. The cleaning process  consisted of dipping into a 37 % solution of HCl and de-ionised water for three minutes, again blowing dry using N2. The surface chemistry of the etched GaN samples was appraised by X-ray  photoelectron spectroscopy (XPS) using a VG scientific ESCALAB spectrometer using AlKα X-rays  at an anode voltage of 10 kV and a filament emission current of 20 mA. Samples were loaded into the  XPS chamber immediately after the final HCl cleaning stage. Surface morphology was investigated by  atomic force microscopy (AFM) using a Dimension AFM in tapping mode. The near-surface  crystallographic structure of samples was assessed by means of reflection high-energy-electron  diffraction (RHEED), performed within a modified Jeol 2000fx transmission electron microscope  (TEM), using the glancing-angle diffraction of electrons at the surface of specimens mounted  vertically just below the projector lens.

3. Results  

3.1. Characterization of n-GaN surfaces  

The acquired XPS survey spectra covering the 0-1100 eV binding energy range were dominated by N  and Ga peaks, with significant levels of C and traces of O and Cl, in all cases. The atomic weight  percentages (atom wt %) of the surface species were determined using peak areas and sensitivity  factors appropriate for N1s, Ga3d, C1s, O1s and Cl2p photoelectron lines, as summarized in Table 1.  These elemental ratios are comparative and do not represent absolute stoichiometry since the  photoionization cross-section and transmission function of the spectrometer for each element are not  taken into account. However, a distinct trend is revealed, with the KOH treatment resulting in an  increase of the N surface content and a decrease of the Ga surface content for the case of both MBE  and MOCVD samples, in agreement with [9] for the case of MOCVD-grown GaN. It is also evident,  following KOH treatment, that the surface C content decreased slightly for the case of the MBE-grown  GaN and significantly for the case of the MOCVD-grown sample. Variations in the O and Cl surface  content were also detected, but these values were below 1 %, and on the scale of the sensitivity of this  XPS experimental arrangement. No remnant K content was detected as a consequence of the KOH  treatment.

Fig1

4. Discussion  

The combined AFM, SE and RHEED investigations provide a clear description of the sample surface  morphology and near-surface microstructure, whilst I-V and EBIC characterization provide evidence  for changes in the contact electronic barrier height and the level of non-radiative recombination as a  consequence of KOH treatment. Furthermore, XPS data provide clues to changes in the surface  chemistry and by implication surface electronic structure, following etching, to help reconcile these observations.