This invention relates to a low cost metal substrates for CIGS thin-film solar cell . A copper indium gallium selenide solar cell (or CIGS cell) is a thin-film solar cell used to convert sunlight into electric power. CIGS is a I-III-VI2 compound semiconductor material composed of copper, indium, gallium, and selenium . The material is a solid solution of copper indium selenide (often abbreviated "CIS") and copper gallium selenide, with a chemical formula of CuIn _x Ga(i-x)Se2, where the value of x can vary from 1 (pure copper indium selenide) to 0 (pu re copper gallium selenide). The bandgap varies continuously with x from about 1.0 eV (for copper indium selenide) to about 1.7 eV (for copper gallium selenide) .

The efficiency of thin film Cu(In,Ga)Se2 (CIGS) has steadily increased in the past years and maximum conversion efficiency of 22.6 % for CIGS on rigid glass substrate was reported . These glass based solar cells are heavy and brittle, and produced by a batch process. With process modifications CIGS solar cells can also be fabricated on flexible substrates e.g . stainless steel or polyimide, which facilitates cost effective roll-to-roll production. The realization of high efficiency solar cells in a fu lly roll-to-roll compatible process based on common mild steel foils might allow to further reducing costs of solar modules. In recent years progress has been made using stainless steel foil as a su bstrate material. A certified efficiency of 17.7 % has been achieved on stainless steel using a low temperatu re CIGS deposition process and 17.6 % efficiency was achieved for CIGS grown on an enamelled mild steel substrate. Modu les with about ~16% efficiency are already available with stainless steel su bstrates.

The electrical performance of CIGS solar cells on mild steel substrates is known to deteriorate due to the diffusion of elements like Fe, Mn, Al etc. into the CIGS active layer during processing at high temperatures.

Many studies have been performed to mitigate this.

The object of this invention is to provide a low cost CIGS thin-film solar cell that does not suffer from electrical performance losses.

The object of this invention is also to provide a low cost flexible CIGS thin-film solar cell that can be produced in a roll-to-roll production.

Another object of the invention is to provide a method of producing said low cost flexible CIGS thin-film solar cells.

Another object of the invention is to provide a method of producing corrosion resistant low cost flexible CIGS thin-film solar cells.

Another object of the invention is to provide a low cost flexible CIGS thin-film solar with an efficiency of at least 15% when depositing the CIGS layer at 450 °C. Another object of the invention is to provide a low cost flexible CIGS thin-film solar with an efficiency of at least 14% when depositing the CIGS layer at 500 °C.

One or more of these objects is reached by the invention embodied in claim 1. Preferred embodiments are provided in the dependent claims 2 to 8.

The advantage of the solar cell according to the invention is that the mild steel substrate can be produced at low cost, and the nickel and chromium layer can be applied using electrodeposition in commercial coating lines. Consequently the coated mild steel substrate is relatively inexpensive. Also, the thermal expansion coefficient of mild steel is comparable to that of CIGS absorber layers (asteei = l l-13- 10 ^-5 K ^-1 , 10 ^-5 K ¹ at 20°C). The stability of the mild-steel substrate (thermally and chemically) is good, compared to that of polyimide which starts to degrade at temperatures above 450 °C. The coated mild steel substrate is also corrosion resistant. The formability of the steel substrate is also a big advantage because it can be formed into the desired shape easily or it can be used as a load bearing component in (e.g .) a roof structure. The nickel and chromium layer form an effective barrier to prevent the deterioration in electrical performance of CIGS solar cells due to the diffusion of elements like Fe, Mn, Al into the CIGS active layer during processing at high temperatu res and electrical performance losses are the resu lt. The Fe, Mn and Al from the mild steel substrate do not diffuse through the Ni-Cr barrier layer or at least do not pass the critical concentration limits. It is important that the nickel layer and the chromium layer after deposition onto the mild steel su bstrate are not su bjected to a high temperature annealing before the CIGS application process. If the mild steel substrate needs to be recrystallisation or recovery annealed, then this has to be done before applying the nickel layer and the chromium layer, because otherwise the impurity elements will diffuse into the nickel layer and the chromium layer and diffuse through the molybdenum back contact layer during the growing of the CIGS absorber layer and finally potentially end up in the CIGS absorber layer. The inventors found that it is important to keep the iron content in the CIGS absorber layer as low as possible, preferably below 20 ppm, more preferably below 7 ppm .

It is noted that the mild steel substrate may be provided on both sides with the same barrier layer. However, usually the other side (referred to as bottom side for convenience) often only needs protection against corrosion and there is no need to block the transfer of detrimental elements on the bottom side, so in most cases a conventional size nickel layer or zinc layer is sufficient for corrosion protection. Even an organic coating system such as a paint, or a lacquer or thermoplastic film could be used as coating of the bottom side. If left uncoated, the bottom side wou ld be prone to corrosion of the mild steel . It should be noted however, that any coating layer may be provided on the bottom side and the invention is not limited to any particular coating on the bottom side, as long as the barrier layer on the top side is according to the invention. In the barrier layer according to the invention the nickel layer thickness is between 0.25 and 5.5 μηι, and the chromium layer thickness is between 0.01 (10 nm) and 0.30 μηπ (300 nm). In the presence of a dielectric layer the Ni and Cr layer can be thinner than without a dielectric layer. A suitable minimum chromium layer thickness is 15 nm . A suitable maxim um chromium layer thickness is at most 200 nm, preferably at most 150 nm . A suitable minimum nickel layer thickness is 0.4 μηι . A suitable maximum nickel layer thickness is at most 3.5 μηι, preferably at most 2.5 μηι .

In a preferable embodiment the nickel layer thickness is between 1.75 and 2.5 μηπ and/or the chromium layer thickness is between 0.075 and 0.125 μηι . These layer thicknesses are particu larly suitable for CIGS process temperatu re between 450 - 500 °C and the modules are made using the shingling approach. For the monolithical approach there has to be a dielectric layer and therefore the nickel and chromium layers can be thinner, and the dielectric layers also prevents the movement of detrimental elements to the CIGS layer, depending on the natu re of the coating .

Using the Ni+Cr barrier layer the diffusion of detrimental elements from the substrate into the absorber could be blocked to below the detection limit of SIMS measurements and no additional defect levels could be observed using admittance spectroscopy.

The steel substrate can be of varying thickness from 25 m to 3 mm . The lower thickness forms a flexible solar module, whereas thicknesses of over 0.3 mm can take a rigid form or even be directly integrated to building elements, in which case an electrically insulating layer is applied . The steel grade itself is not critical, and any low carbon (LC), extra low-carbon (ELC) or ultra-low carbon (ULC) could be used . Suitable steels are the steels DCOl to DC07 as defined by EN 10130 : 2006, e.g . in table 2. Preferably the surface condition of the steels is bright (R _a ≤ 0.4 μηι, EN 10130 : 2006) or mirror-like (R _a ≤ 0.10 μηι, more preferably Ra≤ 0.08 μηι) to minimise the possible negative effects of surface roughness. For the sake of avoiding any misu nderstanding : the term mild steel (also referred to as plain-carbon steel or low carbon steel) specifically excludes stainless steel (aka inox).

The performance of the Ni+Cr barrier layer depends on the temperature used in the CIGS process. Generally these temperatures vary on the type of process used - between 250 - 700 °C. The higher the growing temperature and the longer the growing process of the CIGS absorber layer, the higher the risk of impurity diffusing through the barrier layer.

The standard back-electrode is a molybdenum layer and usually deposited with a direct current/magnetron sputtering process. The thickness of this layer is between 300 and 750 nm, and preferably about 600 nm .

The CIGS absorber layer can be deposited by different known processes such as printing nano particles; sputtering Cu, In & Ga followed by selenization; electroplating Cu, In & Ga and selenization; co-evaporation of Cu, In, Ga & Se (low & high temperature processes) and spraying pre-cursors and sinter to get the CIGS crystal.

A thin (~50-80 nm) CdS-layer is used to form a p-n junction layer. This layer is applied using Chemical Bath Deposition technique, which is a standard. Alternative Cd free layers such as ZnS are used nowadays, which are deposited using a vacuum process.

The transparent conductive oxide (TCO) layer can either be ZnO :AI (ZnO doped with Al) or indium tin oxide (InSnOx) or SnOx doped with F, which are applied by sputtering process. This layer varies between 300 - 800 nm and optionally a metallic grid is applied on top of them .

Optionally, a thin i-ZnO layer is deposited between the semiconductor layer and the

TCO layer to maximize the performance of the solar cell. The thin layer of intrinsic zinc oxide is preferably between 50 and 100 nm . The layer has a high electric resistivity and good optical transmittance. In addition, it is believed that another benefit of the i-ZnO layer is that it can protect the underlying semiconductor layer from sputter damage in the subsequent step of the fabrication process, where a transparent conducting oxide (TCO) is sputtered on top of the i-ZnO.

The optional dielectric layer between the molybdenum back contact layer and the barrier layer is usually sputtered or applied by means of solution processed methods. The dielectric layer can be metal oxides, nitrides or carbides such as alumina, silicon nitrides, titanium nitride or their respective carbides. Alu mina is preferably deposited by means of a sol-gel method because of the speed of this process, although it can also be a pplied by mea ns of the much slower sputtering method . The latter provides a purer and/or denser alumina layer but at the expense of production speed, and the performance of the sol-gel based alu mina layer is adequate. The total thickness of the dielectric layer is preferably at least 2000 nm . The thicker the layer, the higher ability of the layer to prevent migration of detrimental elements into the CIGS absorber layer and result in higher breakthrough voltage.

When all the layers are deposited, the modules ca n be made using the known Shingling approach or using the known monolithical approach. It is noted that for the monolithical approach the dielectric layer is needed between the barrier layer and Mo back electrode.

When the modules are made, they are encapsulated to protect them from moistu re and oxygen so that these modules last for many years.

According to a second aspect a method is provided to produce the low cost flexible CIGS thin-film solar cells according to the invention in claim 9. Preferred embodiments are provided in the dependent claims 10 to 15.

Figure 1 shows a schematic build-up of a solar cell according to the invention (not to scale) and the various layers are indicated . A mild steel substrate 1 is provided with a nickel layer 2 on top of the mild steel substrate and a chromium layer 3 on top of the nickel layer to form a coated mild steel substrate 4. The nickel layer and the chromium layer jointly are referred to as the barrier layer. On the barrier layer a molybdenum back contact layer 5 is provided, followed by a CIGS active layer (aka absorber layer) 6, an n-type semiconductor layer 7 such as a CdS layer (as depicted in the figure) or a ZnS layer and a TCO layer 8. Optionally, a thin i-ZnO layer is deposited between the semiconductor layer 7 and the TCO layer 8 to maximize the performance of the solar cell . The i-ZnO layer is not shown in Figure 1 and 2. The metal grid, such as a Ni/AI grid, and an a nti- reflection coating, such as Mg F2-layer, on top of the TCO are not drawn. Instead of the CdS layer 7 a ZnS layer 7 may be used . In Figure 2 the same build-up is shown with an optional dielectric layer 9 between the coated mild steel substrate 4 and the molybdenum back contact layer, to enable monolithical interconnection of cells, and with an optional TiN barrier layer 10. Within the context of this invention the optional dielectric layer and the optional TiN barrier layer, if present, a re part of the coated mild steel su bstrate. Figure 7 shows the same structure on the top surface of the mild steel substrate as in figure 1, and a further corrosion protective layer 11, such as a nickel layer, or any other suitable corrosion protective layer, such as a nickel and chromium layer such as the one on the barrier layer on the top surface. The metal layers making up the barrier layer on the top surface and the corrosion protective layer on the bottom surface of the mild steel substrate can be deposited simu ltaneously, i.e. a nickel layer may be deposited on both sides simu ltaneously, followed by other metal layers. In another embodiment the same structure on the top surface of the mild steel substrate as in figure 2 is combined with a fu rther corrosion protective layer 11 on the bottom su rface of the mild steel su bstrate, such as a nickel layer, or any other suitable corrosion protective layer, just as in Figure 7.

CIGS solar cells were grown onto different substrates at 450 °C. The results in terms of efficiency and consistency are shown in the table below. Several examples exceed the 15% threshold for efficiency. A measure for the consistency is given by the percentage of cells having a lower efficiency of between 5 and 10, and below 5%. If the achievable efficiency is over 15% and the number of cells below 10 and 5% is low, then there is a good consistency. The table clearly shows the superiority of the Ni+Cr layers, provided the Ni-layer is not su bjected to an annea ling treatment (samples 10- 13). The consistency of sam ples 12 and 13 is q uite good, but the overall efficiency and other cell parameters are not good enough. It shou ld be noted that the chromium layer in samples 8 and 9 were produced from different electrolytes, and that for the electrolyte of sam ple 9 no corrector was used because this contains iron, and this iron would end up in the coating layer. Instead 20 g/l KCI was added to the electrolyte.

CIGS solar cells were processed on different Ni+Cr coated mild steel substrates (see Table 1) . 600 nm thick Molybdenum back contacts were deposited by direct cu rrent (DC) sputtering . Subsequently, CIGS absorber layers were deposited by a multistage co- evaporation method at different nominal substrate temperatures of 450 °C and 500 °C. 450°C corresponds to the standard low temperature process. Devices were finished by a chemically deposited CdS buffer layer, sputtered i-ZnO/ZnO :AI window layer and electron- beam evaporated Ni/AI grids and covered with Mg F2 anti reflection coating .

In some cases TiN barrier layers were applied between the barrier layer and the molybdenum back contact. The TiN barrier layer is composed of a DC sputtered [Ar=21 seem, p=8.5* 10 ^"3 , Power = 2 W/cm ² ] 20 nm thick metallic Ti adhesion layer and a DC sputtered [Ar=21 seem, N = 10 seem, p=5* 10 ^"3 mbar, Power=3W/cm ² ] 250 nm TiN barrier layer. If a TiN barrier layer and a dielectric layer is used then the TiN barrier layer is applied first, and the dielectric layer is applied on top of the TiN barrier layer (See Figure 2).

Table 1 - CIGS deposition at 450 °C

^* Trivalent chromium Trilyte® electrolyte provided by Enthone.

" Trivalent chromium TriChrome® Plus electrolyte provided by Atotech.

" ^* DA = Diffusion annealing as a result of recrystallisation annealing of full-hard substrate.

""Deposited from standard Watts nickel bath (Nickel(II)sulphate (270 g/l), Nickel(II)Chloride (50 g/l), Boric Acid (30 g/l), wetting agent, deposition at 55 °C, CD ~5 A/dm ² . The chromium layers of samples 1 and 2 are thin layers used for packaging steels (TccT®) which are not 100% metallic but rather a mixture of chromium and chromium oxide. It should be noted that all used chromium electrolytes are based on trivalent chromium, and that no hexavalent chromium is used . Table 2 - CIGS deposition at 500 °C

5 BP+M F + Ni + Cr 2 - 5.7 100 93

6 BP+M F + Ni + Cr 2 7 10.5 72 43

8 BP+M F + Ni + Cr 2 100 14.8 48 15

9 BP+M F + Ni + Cr 2 100 14.9 10 0

10 DA after Ni-plating +Cr 2 7 10.6 75 11

Detrimental elements like Fe, Cr or Ni can introduce defects in the CIGS absorber layer. As a consequence the efficiency of the device decreases. A decreasing VOC (Voltage Open Circuit) is an indirect indication for detrimental elements like Fe, Ni or Cr present in the absorber layer. In the case without an additional TiN barrier layer between the barrier layer and the molybdenu m back contact layer 4 the VOC significantly drops. This effect is stronger with higher Fe impurity content within the absorber layer. Elemental depth profiles measured using SIMS measurements showed that at temperatures above 500 °C the Cr and Fe content is roughly 10 times higher in the case without an additional TiN barrier coating . Figure 3 shows SIMS Elemental cou nts within the absorber layer. The measured Fe, Ni and Cr concentrations are very low and for the low values very close to the detection limit of the used device.

Comparing the efficiencies at 450 and 500 °C deposition temperature the decrease in efficiency is low. The only outlier is sample 2 which performs well at 450 °C but is poor at 500 °C. It is believed that this strong decrease is the result of increased defect levels in the Cr layer, due to plating bath used . Figure 4a shows that a layer of 7 nm chromium is not able to prevent the migration of iron into the absorber. Figure 4b on the other hand shows that a 2 μηπ Nickel layer combined with a 100 nm chromium layer is able to prevent this to a large extent. The figure also shows a clear correlation with efficiency and VOC values, with decreasing values with increasing iron content in the absorber layer. This figure give support for the need to use a thicker Cr-layer to obtain good efficiency, high VOC and a good corrosion resistance.

Figure 5 shows the performance of samples 3 and 9 in terms of the fill factor (FF) and the Voltage Open Circuit (VOC) . The FF a nd the VOC values even increase after 500 hrs, indicative of the effective prevention of the diffusion of detrimental elements into the absorber layer. Results show that after 1000 h of accelerated stress tests the CIGS solar cells still provide 99% of their original power. The test conditions in Figure 5a and b are 1 Sun illumination, 80 °C, MPP (maximum power point). It is noted that CIGS based thin- film solar cells on meta l substrate have been produced according to the invention having an efficiency of 18%! (See Figure 6).

There is evidence that the surface roughness of the mild steel affects the performance of the solar cell . It is therefore preferable that the roughness of the mild steel is at least in the category of Bright Finish, more preferably in the category of Mirror Finish.