Activation kinetics of the As acceptor in HgCdTe

Abstract

The amphoteric model of As in HgCdTe is the basis of an investigation into how the transfer ${\hbox{A}\dot{\hbox{s}}_{\rm Hg} {\rightarrow} \hbox{As}^{\prime}_{\rm Te}}$ is achieved under Hg saturation and so obtain a method for calculating optimum annealing times for the transfer. It is concluded that Schaake’s assumption that the transfer, or activation, process is diffusion limited, rather than reaction-rate limited, is a better fit to experimental data. The identification of the Te self-diffusivity in HgCdTe as due to Te _i defects is considered to be incorrect. As a result, Schaake’s activation model can only underestimate the optimum anneal times for ${\hbox{As}^{\prime}_{\rm Te}}$ activation if the experimentally observed Te self-diffusivity is used. An equation based on experimental activation data is given that permits an estimate of the optimum anneal time to be obtained in ${x_{\rm Cd} \sim 0.3}$ HgCdTe layers.

Appendices

Appendix 1

Schaake’s numerical solution to the diffusion equation describing As′_Te activation [6] also provided the empirical relation valid for z  ≤ 0

$$ c_{\rm Te}(L,t)/c_{\rm Te}(\infty) = \exp(-0.0045 z^{3} - 0.0737z^{2} + 0.1387z - 0.0518) $$

(A1)

where z = ln β, c _Te(∞) and c _Te(L,t) are the As′_Te concentrations at the layer surface and at the layer/substrate interface after anneal time t, respectively. Schaake assumed that c _Te(0,t) = c _Te(∞), i.e. the surface concentration rapidly attains its equilibrium value, c _Te(∞). The activation level or efficiency is ≥  c _Te(L,t)/c _As or c _Te(L,t)/c _Te(∞) as under annealing at Hg saturation c _Te(∞) is essentially equal to c _As.

Appendix 2

A common feature in the II–VI semiconductors, such as ZnSe, CdS, CdSe, CdTe and MCT (x _Cd = 0.2), is that the anion tracer self-diffusivity is ∝ the reciprocal of the Zn, Cd or Hg partial pressure [24]. A straightforward interpretation of this dependence is that self-diffusion is due to the neutral anion interstitial. It has also been noted, however, that the reciprocal pressure dependence could also arise through an anion anti-site defect diffusing via alternating jumps into cation and anion vacancies [27]. The treatment in [27] is a general one and applied to the MCT case becomes:

• (i) The Te_Hg tracer jumps into an adjacent V_Te at a rate ∝ [V_Te];

• (ii) The Te_Te tracer now jumps into an adjacent V_Hg at a rate ∝ [V_Hg] so that the tracer Te_Hg has now made an overall jump to a nearest-neighbour site on the cation sub-lattice via consecutive jumps between the two sub-lattices;

• (iii) The overall jump rate will be determined by the slower of the rates in (i) and (ii);

• (iv) Assume that (i) is the slower and that the V_Te involved is the neutral charge state so that the overall jump rate of the Te_Hg tracer is ∝ P _Hg;

• (v) The diffusivity of the Te_Hg tracer is therefore ∝ P _Hg;

• (vi) The Te self-diffusivity due to Te_Hg is ∝ [Te_Hg]P _Hg;

• (vii) Assume that it is the neutral charge state of the Te_Hg that is mobile so that [Te_Hg]∝ 1/P ²_Hg and the Te self-diffusivity is then ∝ 1/P _Hg, as found experimentally.

There is no obvious way in which the observed 1/P _Hg dependence can be accounted for in terms of ionised native defects and be consistent with known electrical behaviour.