Experimental Methods and Materials
3.0 Experimental methods and materials.
3.1
Introduction.
This chapter describes the experimental techniques and materials used in the fabrication and characterisation of silicon-doped lateral p-n junction structures formed on patterned (100) and (110) GaAs substrates. It is organised in the order in which the planar and nonplanar substrates are processed; chemical etching to create the required pre-growth morphology, followed by a descripition of the Molecular Beam Epitaxial growth method and planar and non planar substrate characterisation.
Patterned substrates were wet etched through a photoresist stripe mask, thus creating a series of ridge-mesas. When overgrown with silicon-doped GaAs, lateral p-n junctions were formed at the boundaries of the differently doped surfaces. Lateral junctions were processed into p-n diodes via photolithographically defined ohmic metallisation contacts and characterised using Current-Voltage (IV) and Capacitance-Voltage (CV) profiling methods [1- 2].
Planar substrates of orientation closest to the etched planes were co mounted with the patterned substrates during the growth runs. Si-doped layers grown on these were measured using the Van der Pauw Hall technique and an estimate of the carrier densities on either side of the lateral junctions derived from the measured carrier concentrations [1-2]. However, these figures could only be used as a guide, as Ga adatom migration effects in the vicinity of the junctions would probably have modified the true carrier concentrations [1]. Also, the etched sidewalls were not true singular surfaces, leading to increased dopant compensation and reduced carrier densities compared to the planar substrates.
3.2
Etching of the patterned GaAs substrates.
W et chemical etchants were used in the etching process, working on the
reduction-oxidation (redox) principle [3]. Firstly, a constituent of the etchant reacts with GaAs to form an oxide. This is reduced to an activated complex by
another component and dissolved from the semiconductor surface, thus exposing new material for the etchant to attack [4]. Two theories exit on why etchants preferentially attack particular surfaces. The conventional picture is that it attacks Ga terminated surfaces less strongly than As terminated ones, leading to the formation of (N11)A sidewalls on patterned (100) GaAs substrates [3-4]. Etching of this type is anisotropic due to the differing surface etch rates, with the angle of the etched surface being determined by the composition of the etchant. An alternative explanation proposed by our group is that the etchant dissolves the native oxide underneath the photoresist mask, exposing GaAs for the solution to etch [5]. The speed at which the oxide is removed controls the angle of the etched sidewall, with the GaAs etch rate being isotropic. The rate of oxide removal is determined by the ratio of the etchants’ components, thereby controlling the angle of the etched facet. Our results are consistent with the later process.
Semi-insulating (SI) (100) GaAs substrates were initially processed. The (311)A facet was chosen as the mesa sidewall, as it offered the requisite p-type conductivity when doped with silicon and its shallow angle to the (100) plane created relatively small shadowing effects and subsequent alignment problems during the metalising process [1-2]. Etching was carried out through a series of 300|im wide, 1.7mm pitch rectangular windows in a 1pm thick layer of Micropost S I 830 photoresist. These openings were defined using the photolithographic technique outlined in Section 3.4.2 and aligned along the O i l direction, in order to yield the (311)A sidewall. Initial trials used a H3P0 4 :H2 0 2:H2 0
(100:1.2:100) mixture, which produced an etched facet of 27° to the (100) plane, midway between the (311)A and (211)A singular surfaces [1]. This etchant yielded lower facet-flat intersections with a very rounded morphology, making p- n junction location and subsequent metallisation alignment very difficult. The upper boundary however possessed a sharp demarcation, allowing accurate metallisation. Only upper junctions were therefore characterised in the initial study [1].
All subsequent etching was carried out using a HCL:GH3COOH:K2Cr2 0 7
and (211)A surfaces (27°) on the (100) substrates, but generated sharp interfaces at the upper and lower facet-flat boundaries. This allowed accurate metallisation and characterisation of both types of lateral junction. Preferential etching at the substrate-photoresist mask boundary created a small (411 )A surface at the upper facet-flat intersection. In the case of the SI (110) substrates, the photoresist windows were aligned along one of the <001 > directions. Etching through these produced a curved facet, which for part of its length approximated a (100) surface [2]. Etching behaviour of this sort adds weight to the isotropic etching model, as curved surfaces are created when isotropic etching is done through a mask opening [4].
Substrate type, batch and substrate number, etchant type and etched depth for the patterned (100) and (110) substrates are listed in Table 3.1:
Substrate Type
Batch No.
Substrate No. Etchant Etched depth (100) P atti S a il m l, 2,3,4,5,6 H3P0 4 :H2 0 2:H2 0 17pm (100) Patt2 U5052,53,54, 55,56,57
HCLiCHaCOOHiKsCraOr
5pm (100) Patt3 Sa12m51,52,57, 58,59,60,61,62,63 HCL:CH3COOH:K2Cr207 5pm (110) Patt4 Sa12m88,89,92 HCL:CH3C00H:K2Cr207 5pm Table 3.1. Etching details of all device batches.3.3
Molecular Beam Epitaxy.
Epitaxy refers to the ordered growth of one crystal upon another crystal, such that the orientations of the crystals bear some well defined relationship to each other [6]. The term ‘epitaxy’ was constructed by Royer from the Greek phase, “arrangement on” [7]. There are many epitaxial growth techniques for GaAs and related compounds, including Chemical Vapour Deposition (CVD), Liquid Phase Epitaxy (LPE), Metal Organic Chemical Vapour Deposition (MOCVD) and Molecular Beam Epitaxy (MBE). The growth method used in this study was Molecular Beam Epitaxy, so we shall limit the discussion to that technique.
A schematic illustration of the MBE system is shown in Figure 3.1. The growth system is contained within a Ultra-High Vacuum enclosure and the effusion cells and substrate holder are encircled by liquid nitrogen cryoshields [8].
Rotating substrate holder and heater
Fluoresent Screen Electron Gun G aA s substrate M olecular Beam M echanical Shutter G a As Effusion Cell H e ate r Coil
Figure 3.1. Schematic of a Molecular Beam Epitaxy growth system.
The requirement for a very high vacuum environment is two-fold. Firstly, molecular beams are generated by heating source material within the effusion cells. These cells are called Knudsen cells, as they are designed so that the probability of collisions between the atoms of evaporated species is considerably lower than that of the atoms and cell walls [8]. Consequently, the flux of particles exiting the cell is jet-like and forms a well defined beam. If the atmosphere within the growth system contained large quantities of residual atoms or molecules, collisions between these atoms and those forming the beams would destroy the shape and symmetry of the beams impinging on the substrate surface. Controlled growth would not be possible, as the growth process required a certain number of atom or molecules to be delivered per unit time.
Secondly, contamination of the epitaxial layer from residual gases could affect or even change the conductivity of the doped material. This behaviour is very undesirable, as contamination levels would not be constant so reducing the
repeatability and uniformity of the grown material. Very high purity source materials are used to further minimise unwanted impurities. Before the commencement of growth, the chamber is given a long bake at high temperature to outgas any impurities absorbed into the walls [9].
Fluxes from the effusion cells impinge on the substrate from different directions. Their distributions across the substrate surface are therefore non- uniform, leading to compositional and thickness variations [9]. This is overcome by rotating the substrate during growth, the speed of rotation being related to the time required to deposit one monolayer (ML) of epitaxial material, typically, 2 to 3 ML per rotation. High speed mechanical shutters modulate the molecular beams, allowing the sequential growth of different materials [9]. The electron gun and fluorescent screen shown in Figure 3.1 form part of a Reflection High Energy Electron Diffraction (RHEED) measurement system. The desorption temperature of GaAs native oxides and epitaxial growth rate can be determined by this system, making it an valuable in-situ growth control aid [9].
The arsenic beam from a Knudsen cell is composed of As4 tetramers [9]. A
high temperature cracker at the front of the cell can be used to convert these into As2, though this was not done in our studies. MBE growth of GaAs therefore
consisted of surface interactions between atomic Ga and molecular AS4. A
model for the growth chemistry of the (100) GaAs surface is shown below;
Desorption Precursor State
Chemisorbed State As, molecule
2nd order reaction
Ga stabilised GaAs surface
Figure 3.2. (100) GaAs growth model from beams of Ga and AS4.
Incident AS4 molecules are absorbed into mobile precursor states. From these
states a pairwise dissociation of AS4 molecules occurs, with four As atoms being
chemisorbed onto adjacent Ga atoms and four desorbing as an AS4 tetramer.
This is a second order reaction because it requires the encounter and interaction 51
between two As molecules, so therefore the As sticking coefficient can never exceeds 0.5, even with an excess Ga surface population [9-11]. Sticking coefficients of Ga and Si are unity, as each supplied atom binds an arsenic atom [10]. The rate-limiting step for (100) GaAs growth is the As reaction probability and consequently the main variables in MBE growth are substrate temperature and As4:Ga flux ratio, both of which mediate the As precursor population i.e high
As4:Ga ratio and low growth temperature create a high As surface population [3].
Epitaxial growth on the (110) surface proceeds in a similar manner to that described above, but the As surface lifetime is very short [12]. To achieve good quality epitaxial films on the (110) surface, growth must be conducted under high As4:Ga flux ratios or low substrate temperatures, as the As incorporation
probability increases with surface population density [13]. Unfortunately, these growth conditions produced n-type GaAs when doped with silicon, contrary to the required p-type [13]. Therefore, conditions which resulted in (110) material of poor morphological quality had to be used when growing the p(1 lO)-n(IOO) lateral p-n junction structures. This was to have consequences in the electrical performance of the (110) lateral p-n junctions, which is discussed in Chapter 6.