Layered Growth of Semiconducting Materials

Semiconducting alloys can be formed one atomic layer at a time in the process of epitaxial growth. The aim of this process is to enable sharp interfaces to be formed between one type of alloy and the next e.g. GaAs and AlAs, and thus create structures which may confine electrons and exibit 2-dimensional behaviour. It also provides a way of growing high quality bulk material, although there are other faster and more economic methods, and of introducing dopants into materials at precisely controlled levels.

Molecular beam epitaxy (MBE)

Molecular beam epitaxy (MBE) is basically a sophisticated form of vacuum evapouration. Molecular beams of the constituent elements are generated from sources and travel without scattering to a substrate where they combine to form an epitaxial film. In solid source MBE, material is evapourated from solid ingots by heating or with an electron beam. The rate of growth depends on the flux of material in the molecular beams which can be controlled by the evapouration rate and, most importantly, switched on and off with shutters in a fraction of the time required to grow one monolayer (ML). Typical growth rates are 1 ML per second, or 1 micron per hour, which is equivalent to a pressure of 10^-6mbar arriving at the substrate in the molecular beams. Great trouble is taken to ensure that neglible quantities of impurity atoms are introduced into the material: substrates are carefully prepared and cleaned; only ultra pure sources are used; the reaction chamber is evacuated to <10^-11mbar and the walls of the chamber cooled with liquid nitrogen. Even so the highest mobility layers are only grown after an extended run when the machinary has completely cleaned itself.

The basic principle of epitaxial growth is that atoms on a clean surface are free to move around until they find a correct position in the crystal lattice to bond. Growth occurs at the step edges formed since at an edge an atom experiences more binding forces than on the free surface (see diagram). In practice there will be more than one nucleation site on a surface and so growth is by the spreading of islands. In high quality material these islands will be large with height differences of less than a ML. The mobility of an atom on the surface will be greater at higher substrate temperature resulting in smoother interfaces, but higher temperatures also lead to a lower "sticking coefficient" and more migration of atoms within the layers already grown. In practice the beams do not contain individual atoms, but molecular species like As₂ or As₄. These are cracked at the surface and the efficiency of the process is also temperature dependent. Clearly there will be a compromise temperature to acheive the best results. For GaAs this is ~600^o C, while for SiGe alloys a much higher temperature is used, up to 950^o C.

In the growth of III-V alloys (e.g. GaAs) group V atoms (As) will only attach themselves to sites adjoining group III (Ga) atoms, whereas the group III atoms will stick to any site. Thus to produce high quality material an excess group V flux is maintained. The whole range of alloys Ga_1-xAl_xAs can be grown by introducing a fraction x of aluminum into the group III flux. Similarly the material can be doped by adding atoms like silicon either within the flux or by interupting the growth to deposit a single delta-layer.

During the MBE process, growth can be monitored in situ by a number of methods:

Reflection high energy electron diffraction (RHEED), using forward scattering at grazing angle, which shows a maximum when there is a completed monolayer and a minimum when there is a partial layer as this produces more scattering;
Low energy electron diffraction (LEED), takes place in backscattering geometry and can be used to study surface morphology, but not during growth;
Auger electron spectroscopy (AES), records the type of atoms present;
Modulated beam mass spectrometry (MBMS), allows the chemical species and reaction kienetics to be studied.

Post growth characterisation may be by a wide variety of techniques including electrical transport, pholtoluminescence, electron microscopy, secondary ion mass spectrommety ( SIMS ) or a host of other surface sensitive methods.

Chemical vapour deposition (CVD)

The growth process in MOCVD (metal-organic CVD, also known as MOVPE metal-organic vapour phase epitaxy) is similar to MBE, but the atoms are carried in gaseous form to the substrate. GaAlAs growth is achieved by using a mixture of hydrogen as a carrier gas and organometallic precursors such as trimethyl galium and/or trimethyl aluminium together with arsine. The growth rate can be 10 times greater than in MBE, the process does not require ultra high vacuum and it can be scaled up from research to production of commercial devices relatively easily. However, the preparation of the gaseous mixtures has to be very carefully controlled so that as yet it is unclear which technique will eventully dominate. One advantage MOVPE has over MBE is in the ability to grow phosphorous containing alloys, once phosphorous has been introduced into an MBE chamber it is almost impossible to grom anything else! One disadvantage is that in situ monitoring is more difficult.

A very promissing hybrid technique is gas source MBE (GSMBE, sometimes called chemical beam epitaxy CBE) where material is carried into a high vacuum reaction chamber as a gaseous compound, typically the hydride or a metalorganic.

Links to growth centres:

Imperial College, IRC for Semiconducting Materials
Clarendon Laboratory, Oxford MOVPE growth of antimonide based structures
Sheffield EPSRC Central Facility for III-V Semiconductors MBE and MOCVD growth
University of Nottingham, Nitride MBE growth
Cavendish Laboratory, Cambridge
Defence Evaluation and Research Agency
Silicon-Germanium growth - Cambridge Engineering page

C.T. Foxon and B.A. Joyce, in Growth and Characteristics of Semiconductors, Eds R.A. Stradling and P.C. Klipstein, (Adam Hilger/IOP, Bristol, 1990) p.35

M.J. Kelly, Low-Dimensional Semiconductors, (Oxford, 1995) Ch.2

Last updated 03/02/97 by D.R. Leadley