(9a) Design and Control Considerations for Scale-up of a Cigs Inline Co-Evaporative Physical Vapor Deposition Process

Cu(InGa)Se2 (CIGS) is quickly emerging as the most credible absorber material for economical large-scale manufacture of next-generation thin film solar cells. Rapid advances have been made in improving the efficiency of CIGS thin film based solar cells (up to 18% [1]), but only at the laboratory scale. Since the key quality variables that influence final solar cell efficiency are film thickness and composition (Cu ratio and Ga ratio), the overall operational objective for a successful commercial process is robust and tight control of the mean values of these two variables as well as their uniformity across large-area substrate, for long deposition times. (For example, thickness variation of more than ±10% around the mean value is unacceptable). With scale-up proving to be much more difficult than expected [2], most candidate commercial technologies remain at the pre-manufacturing development stage. In this presentation, we show how successful scale-up involves problems in both design and control that must be considered simultaneously; and that unless the process is designed with explicit due consideration for the desired final film quality specifications, successful scale-up to achieve desired control and overall process performance will be very difficult.

The particular manufacturing technique adopted at the Institute of Energy Conversion (IEC) is the co-evaporative physical vapor deposition (PVD) of CIGS thin-films onto a flexible moving substrate using a roll-to-roll processing scheme (Fig. 1). The film is deposited by thermal evaporation from a series of elemental sources located sequentially in a vacuum deposition chamber. More specifically, controlled amounts of copper, gallium, and indium are delivered to the substrate, while selenium vapor, provided in excess using sparger pipes, is built into the film as determined by the stoichiometry. Ideally, the final film quality variables (measured by X-ray fluorescence sensor) should be controlled in a cascade manner by (i) an outer multivariable controller which manipulates the set-points for the elemental source effusion rates, and (ii) inner effusion rate regulatory controllers which achieve these desired set-points by manipulating power to each individual heating source. However, due to the unreliability and operational difficulties of direct effusion rate measurement techniques (such as atomic absorption spectroscopy), the desired effusion rates are achieved indirectly by controlling the individual source temperatures. The source temperatures required to achieve the desired effusion rates are determined using a simplified effusion model [3], which relates melt-temperature to effusion rate.

It has been customary to regard the scale-up of PVD processes strictly as a problem that can be handled exclusively by control, ignoring the possibility that process design assumptions may no longer apply at the commercial scale. For specific system under study, we have identified two primary issues (relatively insignificant for a pilot-scale process) that must be resolved for a successful commercial-scale development: (i) melt-temperature gradients, and (ii) the reduction of melt-level with time. Considering how these issues arise from elemental source design, and considering their impact on process controllability, the scale-up issues can be divided into the following two components: (i) The design issue ? where an elemental source is designed such that melt-temperature gradients are minimized, thus reducing product quality variations, and (ii) The control issue ? where the mean values of the final film quality variables are robustly controlled such that the desired set-points are achieved while simultaneously rejecting the disturbances introduced by the melt-level reduction.

In this presentation, we will focus mainly on the design issue. First, we will discuss in detail how the two main issues mentioned above are related to the design and operation of the elemental sources, and how they influence the film quality control. Second, we will present experimental results that quantify the temperature gradients present in an elemental source-boat, and the effect of these gradients on film thickness uniformity under normal operating conditions. Finally, we will present two design modifications that resolve the issue of thickness and composition non-uniformity over wider substrates. We will also discuss the effect of process design parameters such as source-to-substrate distance and beam collimation parameter (which determines angular flux distribution). These process parameters collectively play a significant role in deciding not only the elemental source design but also the overall process performance, especially material utilization efficiency.

References:

[1] Contreras, M.A., Egaas, B., Ramanathan, K., Hiltner, J., Swartzlander, A., Hasson, F. and Noufi, R. ?Progress toward 20% efficiency in Cu(In,Ga)Se2 polycrystalline thin-film solar cells?. Prog. Photovolt: Res. Appl. 7, 311-316 (1999).

[2] Birkmire, R.W., ?Compound polycrystalline solar cells: Recent progress and Y2K perspective?. Sol. Energy Mater. Sol. Cells, 65(1), 17-28, (2001).

[3] Junker, S.T., ?Modeling and control of a continuous co-evaporative physical vapor deposition process for production of thin-film copper indium gallium diselenide photovoltaic cells?. Ph.D thesis, Univeristy of Delaware (2003).