Solar-power development requires pragmatic input

Much promising work is going on today in developing new photovoltaic materials.

Jan 1st, 2004
Th 141684

Much promising work is going on today in developing new photovoltaic materials. But to drive these promising new materials to industrial reality, the "materials-research-type" thinking that has come up with these innovations must also be complemented by more pragmatic inputs.

Solar-cell products, mostly modules based on crystalline silicon (Si) wafer technology, are currently commercially available. Although these products technically perform very well, they suffer from high production costs, resulting in final cost/kilowatt-hour ratios that limit their economically viable applications. The cost of today's photovoltaic (PV)-generated electricity is on the order of five to ten times that of traditional electricity production. Nevertheless, the PV market has grown at an average rate of close to 35% per year over the past five years, and larger markets become available to PV as the resulting cost/kilowatt-hour is reduced.

The primary expensive production steps for Si-based photovoltaics are the purification process of the Si material itself, the crystal growth and wafer cutting, and the assembly of cells into interconnected modules for practical use. Considerable ongoing research is aimed at either cost reduction of these Si technologies or the development of alternative material systems from which potentially cheaper solar cells can result.

One of these is the thin-film solar cell, which attempts to replace Si wafers with thin films of semiconductors deposited onto low-cost substrates. In principle, the material quantities needed for thin films are extremely small, the films are grown by large-area coating processes, and the cell interconnections are formed using laser-scribing techniques. On the other hand, thin-film materials systems are far more complex than doped crystalline Si wafers.

At the laboratory level several thin-film systems have achieved energy conversion efficiencies far in excess of the needed commercial requirements. The necessary efforts currently focus more on up-scale, long-term stability, and low-cost production schemes than they do on improved efficiency. Various materials are candidates for thin-film solar cells but we will limit our interest here to just one: devices based on the copper indium diselenide base compound alloyed with gallium and/or sulphur—Cu(In,Ga)(S,Se)2—referred to as CIGS.

CIGS-based modules

Research on these devices began in the early 1970s and the progress made in the subsequent decades brought the technology to a point at which a number of industrialization attempts are ongoing. Products are commercially available, but production is modest, as production costs are still a challenge. The goal is the production of square-meter-sized modules at $50 to $100 per square meter and 12% to 15% efficiency (that is, 120 to 150 W/m2). This would result in kilowatt-hour costs two to four times less than today's Si products.


CIGS-based thin-film photovoltaic modules are fabricated from a monolithically series-interconnected sequence of p-n heterojunction "cells" on a low-cost substrate.
Click here to enlarge image

A CIGS-based module is a monolithic series-interconnected sequence of p-n heterojunction "cells" on a low-cost substrate (see figure). Each cell comprises a molybdenum (Mo) back contact, a p-type Cu(In,Ga)(S,Se)2 photon absorber layer, and an n-type transparent conducting oxide (TCO) window layer, often aluminum (Al)-doped zinc oxide (ZnO). Unfortunately, "complex" interfacial aspects require the additional use of buffer layers prior to the deposition of this TCO, and, even more of a problem, the best material and process to date is cadmium sulphide (CdS) grown by a chemical bath deposition (CBD) technique.

What are the large-scale low-cost potentials and limitations of this approach? The substrate is soda-lime glass (SLG), usually 3 mm thick. The Mo back contact (<1 µm) is DC sputtered onto the SLG and although specific properties are required, no particular large-area difficulties are expected. This Mo layer is then patterned (or scribed) for back contact isolation as shown in the figure. For this, satisfactory laser techniques have been developed. The next step is the deposition of a 1.5- to 2-µm-thick CIGS layer. The growth of this complex layer with the appropriate electronic properties is nontrivial and the various approaches to this problem constitute the majority of the efforts within the research community.

Basically, the best materials are produced by "questionable" techniques, while the best techniques produce materials of "questionable" quality. Nevertheless, depositions on large-area substrates (up to 60 × 120 cm2) have resulted in modules exhibiting efficiencies on the order of 12%. The commercial products of today, as well as the best laboratory devices, use the (CBD) CdS buffer layer. First, the use of Cd is not desirable in a product with environmental pretensions; second, (although it is a question of opinion) the CBD is a "wet step" and can be considered problematic in the context of the vacuum processing of the other layers. Much research is funded today with the objective of a "dry Cd-free" buffer-layer development. A second buffer layer of pure ZnO is applied (RF-sputtered) and at some point, before or after this pure ZnO layer, the second patterning (the interconnect) is performed. This is still often a mechanical scribing process, although the use of a selective laser technique (must remove the CIGS without removing the underlying Mo) would be preferable. At last, the Al:ZnO can be deposited by RF or DC sputtering. Following this, the third and last patterning step (mechanical or laser) disconnects the cells from one another and the active "circuit" is finished.

The product then further requires a glass cover encapsulation. Although technology from the Si modules may here be used, packaging issues arise relative to the long-term stability of these thin-film products.

On the small scale, thin-film PV works well. Larger scale has been demonstrated without excessive technical performance loss in relation to the laboratory level, but at what cost structure? Most probably, thin-film PV will be part of our future lives. The current material systems and structures may find a solid market place, but progress must be made toward the "ideal" solar cell, based on low-cost, abundant, and nontoxic materials as well as simple and high-throughput deposition methods. The long-term objective is for thin-film PV to become as cheap as, or cheaper than, traditional bulk electrical-power generation.

JOHN KESSLER is a professor at the University of Nantes, Laboratory of Solid State Physics for Electronics, 2 rue de la Houssinière, BP 92 208, 44322 Nantes Cedex 3, France; e-mail: john.kessler@physique.univ-nantes.fr.

More in Research