Stable magnesium zinc oxide by reactive Co-Sputtering for CdTe-based solar cells

Highlights

• Combinatorial synthesis of MZO by reactive sputtering.

• Identification of optimal composition as a function of architecture and process.

• MZO stable with respect to time and processing with oxygen.

• High efficiency devices achieved in multiple laboratories.

Abstract

Magnesium zinc oxide (MZO) is a promising front contact material for CdTe solar cells. Due to its higher band gap than traditional CdS, MZO can reduce parasitic absorption to significantly increase short-circuit current density while also providing a benefit of conduction band offset tuning through Mg:Zn ratio optimization. MZO has been successfully implemented into CdTe devices, however its stability has been of concern. The MZO stability issue has been attributed to the presence of oxygen in the CdTe device processing ambient, leading to double-diode behavior (S-kink) in the current density-voltage curves. Here we report on MZO thin films deposited by reactive co-sputtering. The reactively co-sputtered MZO thin films have encouraging stability, show no significant variation in work function of the surface over a period of 6 months, as measured by Kelvin probe. Energy conversion efficiencies of around 16% have been achieved both with and without presence of oxygen in device processing ambients across multiple research facilities. These efficiencies should be possible to increase further by tuning of the thin film deposition and device processing parameters, especially through optimization of the back contact.

Introduction

Cadmium telluride (CdTe) thin film photovoltaic technology has become a commercial leader with over 25 GW installed capacity worldwide [1]. In recent years, the efficiency has been improved significantly to reach 22.1% [2], primarily through alloying CdTe with Se and substitution of CdS as a window layer with wider band gap materials, both of which allow for improved short circuit current density (J_sc), such that they are close to theoretical limits in champion devices [3,4]. Nevertheless, further increases in efficiency can be achieved through improvements in open circuit voltage (V_oc) and fill factor (FF), which is still well below its theoretical maximum. A key to achieving these improvements is higher carrier concentration and lifetime [5,6]. Significant progress has been made in those areas through group V doping and Se alloying, respectively [[7], [8], [9]]. However, to fully realize these benefits improved emitter layers are required [5]. The ideal emitter should be transparent, have appropriate conduction band position and doping levels, and it must be chemically compatible with both the materials and device processing [[10], [11], [12]].

Magnesium zinc oxide (Mg_xZn_1-xO, MZO) has been demonstrated to be a superior emitter to conventional CdS [10]. MZO is transparent (Eg > 3.3 eV) and its conduction band edge can be appropriately aligned with the CdTe absorber by tuning the degree of Mg incorporation, reducing interface recombination. Optimal performance requires flat or slightly positive conduction band offset [[10], [11], [12]]. In addition, in double heterostructures it has been shown that MZO passivates CdTe and results in higher carrier lifetime [13]. Despite these successes, concerns remain about the stability of MZO [14], and its use constraints the conditions used for subsequent device processing. In particular, it imposes a requirement that no oxygen may be used in subsequent processing, as its presence has been attributed to formation of an “S-kink” in the current density-voltage (J-V) curves [15,16]. Another unresolved challenge has been the achievement of stable doping levels in MZO.

To date the MZO films integrated into CdTe solar cells have been produced almost exclusively by sputtering using ceramic targets [3,[15], [16], [17]]. This limits exploration of composition to discrete values, and the optimal Mg/Zn value remains unclear and is likely to be a function of both the specific device architecture and processing steps employed. In addition, ceramic targets are expensive, have relatively low deposition rates, and suffer from target aging and poor utilization, so metal targets are desirable for MZO sputtering [18]. There have been several reports on combinatorial deposition of MZO [19,20], including co-sputtering from Mg metal and ZnO ceramic targets [21], but without integration in CdTe photovoltaic devices.

Here, we demonstrated a combinatorial reactive co-sputtering of MZO films from Mg and Zn targets for application as emitters in CdTe photovoltaic solar cell devices. This deposition approach allows the composition and thickness of MZO film to be varied continuously on one substrate by keeping it stationary during deposition. These combinatorial MZO libraries are then converted into photovoltaic devices by depositing CdTe on top of them, accelerating optimization of solar cells. Optimization of device performance led to device efficiencies of around 16% with oxygen present in the device processing ambient. We hypothesize that native point defects are the origin of MZO instability in other studies, and that the concentration of these defects is stabilized by sputtering in an oxygen ambient in this report. The robust nature of reactively sputtered MZO was validated by its use for successful CdTe device fabrication across multiple laboratories.