Conductivty

Arhenius plot of conductivity data of amorphous silicon. The full triangles correspond to sputtered material [Brodsky-1972jncs], open symbols are from PE-CVD material (circles: [LeComber-1970prl, LeComber-1972jncs), squares: author's data)

The conductivity of the PE-CVD samples shows a very different behaviour with several different regions. Different aspects of charge transport have been discussed by LeComber in a detailed publication, concluding the following [LeComber-1970prl]:

• Between room temperature and 180°C (450 K), the characteristic is fairly linear; its slope corresponds to an activation energy around 0.8 eV. The behaviour suggests conduction by charge carriers that are thermally activated across the band gap. In this range amorphous silicon behaves quite like a crystalline semiconductor with a gap of forbidden states. With the additional measurement of an optical band gap of about 1.7 eV, we conclude that the conduction is that of an (almost) intrinsic semiconductor. LeComber concludes the same transport mechanism for their samples from room temperature down to 250 K, but probably their samples show some n-type behaviour with a Fermi level at 0.6 eV below the conduction band (see their nominally undoped samples in the doping diagram below).
• In the intermediate region between 250 and 200 K, free charge carriers in the bands become less probable, but some of the charge carries trapped into states close to the band edges may be excited and re-trapped. Conductivity will therefore depend on the location of the trapping levels. LeCombers model is based on a state a well defined energy 0.2 eV below the conduction band. The concept of a single state has later been abandoned in favour of a distribution of states whose density tails off expnentially into the bands [Tiedje-1981prl].
• The flatter part at temperatures below resembles the transport in the sputtered sample, only he density of deep trap states is much reduced in PE-CVD material. In this region thermal excitation does not take place into the bands any more, but only into neighbouring trap levels.
• At yet lower temperature, hopping into neighbouring states with higher energy becomes less and less probable. Eventually, it might become less probable than tunneling into states further away but at the same energy. This process is called variable range hopping with a temperature dependence as follows [Mott-1968jncs]: The quantities σ₀ and T₀ contain the density of defect states and the average distance to states that the carriers can hop into. The theory has been used for explaining a variety of experimental observations with sputtered samples like the one shown in the diagram, but there are also cases where the underlying approximations are stretched too far and the and the density of defects is predicted to be in the range of 10³¹ cm^-3 [Brodsky-1972jncs].