Optical and electrical properties of alkaline-doped and As-alloyed amorphous selenium films

Abstract

Electrical and optical properties Cs-doped a-Se_0.95As_0.05 (stabilized a-Se that has been alloyed with As) have been investigated. As expected there was no electron paramagnetic resonance signal on Cs-doped films or bulk samples, which put the spin-active defect concentrations below 10¹⁵ cm⁻³. The Cs-addition to a-Se_0.95As_0.05, leads to the n-type doping of a-Se_0.95As_0.05 in the sense that the undoped material has μ_hτ_h>> μ_eτ_e whereas the alkaline doped material has μ_eτ_e>> μ_hτ_h. The Cs addition also leads to a reduction of the refractive index n and a reduction of the glass transition temperature T_g, and affects the temporal relaxation behavior of a-Se film thickness after annealing and sequential quenching. We have measured the refractive index dispersion, n(λ) versus λ, bandgap (E_g) and Urbach width (ΔE) for undoped and Cs-doped films at room temperature and at a temperature just below the glass transition temperature. The photoluminescence (PL) experiments confirm earlier experiments that the PL emission is a broad emission spectrum with a significant Stoke’s shift following roughly the ~ E_g/2 empirical rule. The present work confirms that Cs-doped and As-stabilized a-Se is n-type.

Appendix

The transmission spectrum T(λ) through a nonuniform filmof a non-uniform thin film with average thickness d and thickness variation over the field of illumination Δd may be expressed as [27,28,29,30],

$$ T(\lambda ) = \left( {\frac{1}{{\varphi_{2} -\varphi_{1}}}} \right) \int\limits_{{\varphi_{1} }}^{{\varphi_{2} }} {\frac{A(\lambda )X(\lambda )}{{B(\lambda ) - C(\lambda )X(\lambda )\cos (\varphi ) + D(\lambda )X(\lambda )^{2} }}} d\varphi $$

(4)

where λ is a light wavelength, φ = 4πnd/λ, φ₁ = 4πn(d − Δd)/λ and φ₂ = 4πn(d + Δd)/λ, n(λ) is the wavelength dependent refractive index. The functions A(λ), B(λ), C(λ) and D(λ) are defined as

$$ A\left( \lambda \right) = 1 6n\left( \lambda \right)^{ 2} s, $$

(5)

$$ B\left( \lambda \right) \, = \, \left[ {n\left( \lambda \right) + 1\left] {^{ 3} } \right[n\left( \lambda \right) + s^{ 2} } \right], $$

(6)

$$ C\left( \lambda \right) \, = { 2}\left[ {n\left( \lambda \right)^{ 2} - 1} \right]\left[ {n\left( \lambda \right)^{ 2} - s^{ 2} } \right]{\text{ and}} $$

(7)

$$ D\left( \lambda \right) \, = \, \left[ {n\left( \lambda \right) - 1\left] {^{ 3} } \right[n\left( \lambda \right) - s^{ 2} } \right] $$

(8)

where s is the refractive index of substrate which is assumed to be independent of λ, and X(λ) as

$$ X\left( \lambda \right) \, = { \exp }\left[ { - \alpha \left( \lambda \right)d} \right] $$

(9)

where α(λ) is an optical absorption coefficient.

To fit Eq. (4) to experimental data we need to choose the appropriate values of d and Δd as well as appropriate functions n(λ) and α(λ), i.e. α(E), where E is photon energy related to λ through E = hc/λ. One-pole Sellmeier equation has been used for the dispersion of refractive index n(λ) as in Eq. (1).

The absorption coefficient α(E) was chosen as a modified Urbach approximation that includes a second order term in the argument,

$$ \alpha (E) = \exp \left[ {\frac{{E - E_{g} }}{\Delta E} - a\left( {E - E_{g} } \right)^{2} } \right] $$

(10)

where E is photon energy, E_g is optical band gap energy, ΔE is Urbach width (reciprocal of the Urbach slope), and the parameter a represents an additional correction.

Overall, the model has eight adjustable parameters d, Δd, n₀, B_s, C_s, E_g, ΔE and a which allows Eq. (4) to fit experimental data nearly perfectly as shown in Fig. 5.