Printing Cells onto Large-Area Flexible Substrates

The reasons for ascendancy of CIGS include efficiency and durability, but the main reason is that these cells can be manufactured, without vacuum equipment, on a roll-

to-roll mass basis similar to a modern printing press. The next figure shows one possible step-by-step prescription for CuIn1_xGaxSe2 CIGS cell manufacture that does not require vacuum equipment. (Many variations are possible, including those that avoid CdS and also those in which the substrate is simply aluminum foil.)

Figure 6.11, top to bottom, shows steps in one version of an ink-printing approach to CIGS solar cell formation. Substrates are either glass coated with Mo or a Mo foil, which can be thin and flexible. The “CIGS absorber layer” is formed in steps 2-5. The absorber is chemically CuIn1_xGaxSe2. The Cu, In, and Ga metal components are printed onto the cell in the form ofoxide nanoparticles suspended in liquid, that is, a water-based ink. This is inherently efficient with regard to the use of the expensive elements, and the formulation ofthe ink allows close control ofthe composition. The layer is reduced (the oxygen is removed) by heating in a reducing atmosphere of

Printing Cells onto Large-Area Flexible Substrates

Figure 6.11 Ink-printed nonvacuum approach [75] to fabricate CuIn1_xGaxSe2 solar cells on flexible

substrates.

nitrogen and hydrogen, so that oxygen leaves the film in the form of water vapor. The resulting metallic alloy layer is transformed to the semiconductor alloy by gas reaction at modest temperature with selenium using H2Se gas, producing the “CIGS absorber.” Speaking approximately, this is a P-type semiconductor of controlled bandgap and of thickness in the 1000 nm range. The thickness must be enough so that most of the light is absorbed. Again, speaking approximately, the device is a PN junction, and the next step, “Junction formation,” is accomplished by deposition of CdS (CBD, or chemical bath deposition). The transparent electrode to the N-side of the junction is ZnO (OMCVD, organometallic chemical vapor deposition). The highest efficiency these workers achieved was 13.6% using a Mo-coated glass substrate. A variation of this process that does not use Cd is described next.

This basic process is important because it uses the minimum amount of the expensive metals, because it avoids expensive high vacuum equipment, and because it can be scaled up to large areas. Think of printing a newspaper, how many square meters of paper is printed by a major newspaper each day. The maximum efficiency reported for this type of cell is nearly 20%. This value is close to the value 20.3% mentioned for polycrystalline silicon cells, which are not amenable to a similar large – scale nonvacuum fabrication. It appears that some form of this basic ink printing process is employed in CIGS manufacture by Nanosolar, Inc. in California, and ISET, and for CdTe by a new firm, Solexant.

The cell shown in cross section in Figure 6.12 was formed on 12.5 pm thick commercial polyimide. The Mo back contact of 1 pm was deposited by DC sputtering.

The CIGS layer was established following sequential evaporation of the metals Cu,

In, and Ga, followed by evaporation of Se, with a controlled temperature anneal. The CdS layer was then deposited in a chemical bath process. RF sputtering was used to apply the upper window layer described as I-ZnO/ZnO:Al of 300 nm thickness. Finally, Ni-Al contact grids for better current collection were applied by electron

Printing Cells onto Large-Area Flexible Substrates

Figure 6.12 Scanning electron microscope to top: polyimide (notshown), Mo back contact, cross-section image [76] of CIGS cell grown on CIGS absorber layer, CdS junction-forming layer, polyimide flexible substrate. Efficiency of 14.1% ZnO insulator layer, and ZnO:Al conductive was achieved in cells ofthis type. Layers, bottom window layer.

beam evaporation. This is a complicated and expensive process but led to the most efficient CIGS cells, 14.1%, ever prepared on a flexible polyimide substrate.

The traditional process for forming the PN junction, the use of a chemical bath to deposit a layer of CdS, is objectionable because Cd is toxic. An alternative means of making the CIGS PN junction without Cd has also been demonstrated [77] achieving 16% efficiency. In this method, the CIGS absorber layer is contacted by a CIGS:Zn junction-forming layer, Zn0.9Mg0.1O “insulator” layer, ITO (indium tin oxide) conductive window layer, and current collecting grid.

In this work [77], the CIGS absorber was deposited on Mo-coated glass by physical vapor deposition. The PN junction was formed by evaporating Zn onto the exposed surface of the CIGS, held at 300 °C. The zinc is believed to diffuse to a depth of about 50 nm in an annealing time about 5 min. This forms an internal PN homojunction within the CIGS. The Zn0.9Mg01O “insulator” layer was grown on the CIGS:Zn surface by cosputtering the oxides from separate sources with adjustable rfpower to establish the desired ratio Zn/Mg = 9. This ratio was found to adjust the conduction band position in the new layer to match that in the CIGS:Zn layer, to allow photoelectrons to pass out to the ITO electrode without scattering. The resulting physical interface was studied by transmission electron microscopy, TEM, and found to be epitaxial. The ITO transparent conductor about 100 nm thick was prepared by sputtering, followed by metal grid electrodes applied by evaporation.

Подпись: Figure 6.13 Current-voltage curve [78] of CIGS cell grown using a dry process and avoiding use of cadmium, on Mo-coated glass substrate. Efficiency of 16.2% was achieved in cells of this type after applying a MgF2

High-efficiency CIGS cells are compared in Figure 6.14. (The same group has recently reported a record efficiency for this type of cell as 19.9%.) These cells use a coevaporation method for the CIGS layer. The traditional wet process is used to form the PN junction. This is described as growing 50-60 nm CdS films on the CIGS layer

antireflection coating. Layers, bottom to top: glass, Mo back contact, CIGS absorber layer, CIGS:Zn junction-forming layer, Zn09Mg01O insulator layer, and ITO (indium tin oxide) conductive window layer.

Подпись: Figure 6.14 Quantum efficiency of CIGS solar cells [79]: solid curve, CIGS thickness 1 mm (17.2% efficiency); dashed curve, CIGS thickness 2.5 mm (18.7% efficiency). These cells benefit from a final antireflection coating of

100 nm MgF2 after depositing the 200 nm thick window layer of Al-doped ZnO, which exhibited a sheet resistance 65-70 V/square, and a Ni/Al grid to collect current.

by immersion for 13 min in a 60 °C bath composed of 1.5 mM CdSO4, 1.5 M NH4OH, and 75 mM thiourea.

The authors report nuances in the method of coevaporation of the absorber to accomplish bandgap grading, and also suggest that for CIGS layer thickness below 1 mm the deep-level density increases. For this reason, mobility and lifetime are reduced when the thickness is reduced to 0.5 mm. This specific process is not scalable, but demonstrates a high efficiency, 19.9%, in single-junction CIGS solar cells.

The European firm Avancis has produced 15.8% efficient CIGS modules using a more conventional linear process. It appears that the firms using the ink process have not won customers, and, in fact, the total production of the CIGS cells is small compared to the production of CdTe cells, which presently dominate the thin-film solar cell market.

6.4

Updated: October 27, 2015 — 12:10 pm