Prediction of Photovoltaic Cu(In,Ga)Se2 p-n Device Performance by forward Bias Electrochemical Analysis of Only the p-Type Cu(In,Ga)Se2 Films

Being able to predict the output power of solar cells by analysis of only the first semiconductor layer – commonly called the absorber layer – i.e. before a number of further deposition steps are carried out – would be a clear advantage both at academic research level and for process monitoring in industry. Since the absorber layer is normally the most difficult to fabricate, the hypothesis is that the power conversion efficiency of the solid-state device is dominated by its optoelectronic properties.

One option that allows extracting information on the absorber optoelectronic properties is by formation and interrogation of virtually reversible electrolyte (Schottky) junctions ¹. We have recently shown that the short circuit current density and thus efficiency of Cu(In,Ga)Se₂ solid-state devices can be predicted by measuring the electrochemical photocurrent density of the respective absorber layers in the presence of Eu³⁺ under illumination ². In such a system the Eu³⁺acts as a scavenger for the electrons generated on the p-type semiconductor upon irradiation with photons with energy greater than the semiconductor band-gap. However, this correlation was found complicated by surface non-ideality, leading in some cases to pronounced charge-carrier recombination. Therefore, here we explore an alternative relationship.

Similarly to p-n junctions, the electrical behavior vs bias of p-semiconductor/electrolyte junctions is a direct consequence of the energy-barriers for hole transfer ³. As such, the current-voltage dependence can be expressed by the diode equation ⁴, including the contribution of parasitic shunt paths (Eq. 1).

Eq. 1      J_D,L = J_Ph + J₀(exp(-eΔV/kT)-1) + ΔV/R_Sh

where J_D,L  are the dark and illuminated electrical current-density, J_Ph  is the photocurrent density ( = 0 in the dark and < 0 under illumination), J₀ is the reverse saturation current, e is the elementary charge, ΔV is the applied voltage, k is the Boltzmann constant, T is the absolute temperature and R_Sh is the shunt resistance of the solar cell.

The dark-current in reverse bias is related to the parasitic terms R_Sh and J₀, with R_Sh normally dominating. For good solid-state devices R_Sh should be as large as possible and J₀ as small as possible. From inspection of literature solid-state device data the J₀ term is more important in determining the overall efficiency of a device than R_Sh. However, J₀ cannot be distinguished from R_Shin reverse bias, while in forward bias it is possible.

Eu²⁺ ions are required to interrogate the junction in forward bias. Eu²⁺ acts as an electron donor in the system, revealing the ability of the semiconductor film to accept electrons when kept under forward bias. Due to the exponential term in Eq. 1, the higher this ability to accept electrons, the higher the expected reverse saturation current from such a semiconductor in a full device whilst the effect of R_Shis negligible.

In this work we demonstrate that the reverse saturation current of CuInSe₂ solar cell devices can be predicted by measuring their forward bias characteristics in the presence of Eu^2+/3+, allowing the prediction of final solid-state device efficiency.

To this end a series of CuInSe₂ absorber layers were prepared by intentional deviation from a physical vapour deposition routine and they were split into two. Half were completed into solar cell devices and the other half were tested electrochemically. The chosen absorber layers gave solid-state device power conversion efficiencies between 6 and 12.5% ⁵.

The electrochemical experiments consisted of voltammetric analyses in the dark with Eu^2+/3+species. The extraction of a term electrochemically-equivalent to the reverse saturation current density is obtained by linear extrapolation of the forward dark current density to voltages close to the open circuit voltage of the semiconductor-electrolyte junction in the dark.

Such electrochemically extracted saturation current is compared with the solid-state device saturation current. These findings open the way to a more reliable electrochemically-based estimation of thin film semiconductor properties with a clear link with solid-state device physics. A comparison between this proposed method and the method involving the measurement of the electrochemical photocurrent density will be made.

References

1. Peter, L. M., Semiconductor Electrochemistry. Encyclopedia of Life Support Systems: Oxford, 2010.

2. Colombara, D. et al. Electrochemistry Communications 2014, 48, 99-102.

3. Tan, M. X. et al. Progress in Inorganic Chemistry, John Wiley & Sons, Inc.: 2007; pp 21-144.

4. Shockley, W. Bell System Technical Journal 1949, 28, 435-489.

5. Depredurand, V. et al. PVSC Proceedings, 2011 37th IEEE, 19-24 June 2011; 2011; pp 000337-000342.

Acknowledgements

LPV, LEM, Nexcis team members and Prof. Laurence M. Peter are acknowledged for help and discussion.

Funding was provided by the European Commission through the "Scalenano" program (grant agreement nº 284486)/