17. CIGS-Based Flexible Solar Cells

A CIGS solar cell is a multilayer device where the interfaces significantly affect the performance. The typical structure is: substrate (glass, 1–3 mm)/back contact (Mo 1 µm by sputtering)/absorber (CIGS, 1.8 µm by PED)/buffer layer (CdS, 70 nm, by chemical solution)/window layer (transparent conductive oxides, TCO, ZnO-ZnO:Al, by sputtering).

The Mo/CIGS interface is identified as the critical interface to improve the electrical performance of the solar cell, in particular of flexible devices during the bending.

The selection of suitable flexible (polymeric) substrates is based on three parameters: (a) temperature resistance (up to 300 °C); (b) thermal expansion coefficient (TCE) and (c) stress stability. Their surface morphology has to be comparable to the glass (roughness ≈6 nm).

Kapton is chosen as benchmark for testing the Mo growth, being cheap and easily available. Other polymeric substrates, such as polyimide (PI) or polyethylene naphthalate (Teonex) from Dupont, perfluoroalkoxy alkanes (a copolymer of tetrafluoroethylene, PFA), fluorinated ethylene propylene (FEP) and ethylene tetrafluoroethylene (ETFE), provided by Saint-Gobain, were analysed in the temperature range 165–260 °C. Besides, metallic foil (stainless steel, SS 430B) and carbon fibres were also tested as possible substrates.

Obviously, the flexible substrate requires a tailored substrate holder.

The film thickness and growth rate are the fundamental parameters to be monitored and controlled. The oscillating quartz microbalance, which is commonly used in physical vapour deposition (PVD) techniques, cannot be utilised with highly energetic pulsed beams. An alternative method, using an infrared (IR) pyrometer, can be used to measure the heat irradiated by the sample during the deposition. The total radiated power per unit area, called the photon radiant emittance, is constant at constant temperatures (Stefan-Boltzmann law), but it is subjected to interference phenomena when irradiated from a thin film, leading to changes of the radiant emittance not depending on temperature. By analysing the emittance profile, the interference fringes due to the CIGS thickness variation appear, enabling the measurement of growing film thickness [18]. Once the film reaches the desired value, a dedicated software moves the substrate.

Optical Emission Spectroscopy (OES) has been employed to monitor in real time the composition of the PED plasma plume. OES is based on the detection of the optical emission (excitations and ionizations) occurring in the plasma. Since the plasma consists of the elements composing the targets, the absolute intensity of the emission peaks gives information on the ablation rate, while the ratio between emission peaks from different elements reveals the plasma stoichiometry. Both the ablation rate and the stoichiometry transfer depend on the PED acceleration voltage and the gas flow.

To achieve the physical ablation of the CIGS target, it is necessary to heat it suddenly, preventing the thermal evaporation. Being CIGS an incongruent melting material, the slow heating leads to the target melting, resulting in the deposition of unwanted phases. In short, this is the reason why PED is so efficient to deposit complex and incongruent melting materials such as CIGS.

The energy power (i.e. e-beam acceleration voltage multiplied by electron current) is the main parameter for controlling the mechanisms of PED ablation (congruent/incongruent evaporation, plume distribution and expansion range, stoichiometry transfer, adatom energy, particulate density, etc.) [19]. The e-beam acceleration is nominally the voltage needed to fully charge the capacitance connected with the hollow cathode and is set by the operator in commercial PED sources. The electron current depends on the combination of charge voltage, source geometry (tube length, cathode aperture), trigger type, gas type and pressure. While the former parameters are constant during the deposition, the gas pressure inside the source changes due to the warming of the cathode and the dielectric tube. The evaluation of the current drop induced by the change of the local pressure and the way to keep constant the energy power transferred to the target is essential for long time depositions. This measurement, not provided by the commercial PED system, can be made by using a contactless toroidal coils, called Rogowski coils, placed around the dielectric tube.

The variation in time of the discharge current of the pulsed current (pulse duration ~100 ns), generates a variation of magnetic flux inside the Rogowski coil, proportional to the variation of the discharge current in time. The automatic feedback of the discharge current measurement stabilizes the signal, by remotely controlling the flux-meter injecting the gas into the chamber.

The deposition rate in the PED technique depends on many factors, among which the electrical parameters of the source, the material to ablate and, above all, the pulse repetition rate. For CIGS, a typical deposition rate value is 0.25 nm/pulse over a 3 square-inch substrate (implying about 10′ to deposit 1.6 μm thick CIGS layer). This process must be much faster for a pre-industrial development.

The easiest approach would be to increase the pulse frequency; since a single pulse has a duration of only 100–200 ns, the maximum frequency could be much higher than 20 Hz, that is the maximum repetition rate provided by the only 2 companies commercializing the PED sources (Organic Spintronic, Italy and Neocera, USA). While a faster electronic could be in principle easily implemented, the main problem is the presence of the dielectric tube guiding the electron beam from the PED body that cannot stand higher frequency for the rapid heating, leading to the melting of the tube.

An innovative solution is to drive the electron beam using a supersonic gas beam, instead of a tube [20]. Further advantages of the tubeless solution, are: (i) easier process (the tubes are the only parts that require periodic cleaning and substitution), (ii) removal of geometrical constraints, and (iii) reduction of contaminants, typically silicon and carbon.

Molybdenum (Mo) is generally used as back contact in TFSC to collect the photo-generated charges. While the deposition of Mo on rigid glass does not represent a major issue, the use of flexible substrates poses serious limitations to the solar cell performance. Mo films sputtered on flexible substrates are investigated by atomic force microscopy (AFM) and compared to the film grown on glass at room temperature (RT), as reference. Typically, films have a surface roughness of  approx. 2 nm and composed of grains with an average diameter of 60 nm. The roughness parameter H [21], which describes how the film morphology changes in the plane, is approx.  0.3, indicating that films are quite smooth over large area.

Mo films were firstly sputtered on kapton, by using the same experimental conditions used for glass (film thickness 500 nm, substrate temperature RT). Such films are rougher (≈6 nm) and composed of bigger grains with an average diameter of 160 nm. H isapprox. 0.04, indicating a jagged surface, in agreement to the roughness increases. In order to decrease it, kapton was heated at 120 °C during the deposition. Doing so, the surface roughness significantly decreases to approx 3 nm, but severe cracks appear in the film due to the permanent bending of the polymeric substrate generated by the heating. In order to remove such cracks, suitable sample holders for flexible substrates are designed, fabricated and successfully tested (Fig. 17.2). Other polymeric substrates (PFA, FEP, ETFE and PI) exhibiting smooth surfaces (roughness ranging from 0.5 to 4 nm) and temperatures resistance from 165 to 260 °C, were tested. Among them, preliminary results on adhesion and conductive properties of Mo films sputtered on FEP in the same experimental conditions (film thickness 500 nm, substrate kept at RT) indicate that it is a promising substrate for fabricating flexible CIGS-based TFSCs.

The exploitation of the Mo layer required the development of a radically new method for the substrate heating, based on the application of a DC electrical power directly through the Mo back contact of the cell, thus converting electrical energy into heat by Joule effect (Fig. 17.3). The very efficient heat transfer to the thin (<1 μm) Mo layer requires a low electrical power density (few W/cm²) to achieve the required deposition temperature, that is much lower compared to the traditional resistor- or lamp-based external heaters. Joule-heated CIGS based TFSCs shows comparable or slightly better PV properties than conventional heaters [22].

The real-time control of the CIGS film thickness is carried out by analysing the heat radiation emitted “in situ”. A pyrometer (model Marathon MM LT), located outside the chamber measures the IR radiative power emitted from the substrate surface through an IR-transparent ZnSe viewport. A LabView software calculates the T = f(t) function, stopping automatically the deposition when the CIGS thickness achieves the desired pre-set value.

The temperature is acquired and filtered every 200 ms and the average temperature value is calculated every 5 points (1 s), limiting the signal noise. Both 1st and 2nd derivative of the T versus time function are instantaneously calculated. When the first inflection point (2nd derivative = 0) is detected, the average deposition rate and the expected time for maximum peak and deposition end are calculated. These calculations are repeated when the maximum (corresponding to a CIGS thickness = 1018 nm) and the second inflection point are reached. Finally, when the curve reaches the minimum (roughly corresponding to the desired final thickness = 2036 nm), the software shuts down the PED source, automatically stopping the deposition process.

The system is very precise and it has been installed in all the deposition chambers currently in use at IMEM-CNR equipped with PED.

The deposition rate is recorded by analysing the temperature versus time, as shown in Fig. 17.4. To further increase the sensitivity in the calculation of the deposition rate, a real-time evaluation of the PED parameters by OES is employed, using a Hamamatsu mini-spectrometer (250–800 nm range). Due to the pulsed nature of PED process, 20 subsequent spectra taken every 300 ms are averaged (every 6 s). The dependence of the OES spectrum on the acceleration voltage is shown in Fig. 17.5. By increasing the voltage, the intensity of the In peak at 325 nm, hence the deposition rate, reaches a maximum at 16 kV, then slowly decreases for higher voltages. The ratio between Ga (417 nm) and In peak becomes constant (0.5) at voltages greater than 12 kV, while the ratio between the Cu (521 nm) and In peaks saturates at 0.30 for values exceeding 14 kV. These trends confirm that an optimal stoichiometry transfer is achieved at voltages higher than 14 kV, while the maximum ablation rate occurs at 16 kV. During the deposition process, when a change in the OES spectrum and/or a variation in the deposition rate are detected, the PED parameters (voltage and gas flow) can be varied in order to recover the optimal conditions. It worth noting that the gas flow is an essential parameter for controlling both the ablation rate and the stoichiometry transfer. The latter can be always obtained at low Ar flows (from 5 to 8 sccm) and in this range both Cu/In and Ga/In peak ratio are not affected by the gas flows.

By measuring the PED e-beam current and voltage with the Rogowsky coil, the beam energetic distribution (fast electrons > ablation, slow electrons > evaporation) can be unambiguously identified.

This information is important to design and realize a new PED source, with a more efficient and highly energetic electron beam, optimized for the ablation of the CIGS target.

Aiming to increase the PED deposition/growth rate (currently limited to 15–20 Hz pulses frequency), the innovative idea was first to use a tubeless supersonic beam instead of a ceramic tube to guide the pulsed electron beam. The main problem is that a micrometric nozzle is required for the generation of the supersonic beam that rapidly tents to be blocked.

Commercial, custom made molybdenum aperture, selected for its chemical inertia, i.e. a disc of about 10 mm diameter, thickness around 0.1 mm and nozzle aperture of 20, 40 and 50 μm, have been tested in different working conditions. As expected, the use of the 20 μm nozzle gives the best results in terms of electron beam propagation and minimum gas load; however, the argon pressure in the supersonic beam source as high as 1000–2000 mbar is necessary, with a pressure increase in the vacuum chamber from 10⁻⁶ mbar to 10⁻⁴–10⁻³ mbar. Clearly, the larger the nozzle, the more difficult the control of the beam.

During the ablation process at high voltage (>14 kV), in particular in presence of oxygen, the nozzle aperture only lasts a few minutes before obstructing, while using low power electron beams, no damaging of the Mo diaphragm has been observed, also in case of exposure to air.

Unfortunately, the threshold for the CIGS target ablation is 16 kV, (Fig. 17.6) and in these conditions the nozzle lifetime is limited to 30 min; after that, unavoidably the occlusion forms, due to massive erosion and material melting, irreversibly closing the aperture.

The same problem occurs by using a source made of quartz with a 50 μm nozzle. Again, its maximum lifetime is around 30–40 min, then the gas pressure sudden increases due to the formation of cracks around the nozzle.

Although the principle of the supersonic beam has been successfully demonstrated, this solution is considered not suitable for the production scale-up.

An alternative way to increasing PED deposition rate is to provide a more efficient control of the pulse generation to increase the pulsed frequency. The difficulty is mainly related to the necessity to manage a quick (transient of few ns) and fast (width of 100–200 ns) impulse of current at very high voltages (10–20 kV) and peak current (1000–2000 A). High voltage (HV) power supply can be only used up to 40 Hz, because of its high internal intrinsic resistance. A gas HV switch Thyratron is finally adopted, by far more complex to control, requiring different low/middle voltage supplies for proper working, but enabling a reliable and efficient control unit.

The radically new PED sources, fabricated in collaboration with Noivion s.r.l.,¹ show a good performance in the working conditions at frequencies up to 150 Hz (Fig. 17.7).

The installation of the new PED sources could not fit the existing chamber, therefore a new deposition system has been designed and built. Aiming at the pre-industrial scale-up, it hosts 4 sources, able to deposit solar cells as large as 160 × 160 mm², on both rigid (glass) and flexible substrates (Fig. 17.8).

The preliminary results confirms that, by adjusting the PED sources geometry, it is possible to homogeneously deposit on 160 × 160 mm², corresponding to the size of a conventional Si-based solar cells.

To design the radius of the wheels, an AFM combined with a mechanical buckling system (Fig. 17.10) is used to measure, in situ and in real time, the maximum bending curvature of CIGS-based TFSCs deposited on a 1 cm wide flexible stripe [23]. To determine the maximum curvature to which the device is irreversibly damaged, both topographical changes and electrical performances are tested during the deformation induced by the bending.

FEP bare stripes were first investigated to understand how the surface roughness changes with the angle subtended by the circular arc formed by the bended film (α). The roughness for α = 0° (flat stripe) is 10 nm, which increases linearly to 20 nm for α = 180° (maximum bending angle) due to the relatively large deformation of the polymeric stripe. In order to test different polymeric substrates in different conditions, PI (roughness 4 nm) was coated with 500 nm of Mo to test the back electrode behaviour versus bending, while ETFE was coated with ITO (indium tin oxide, 40 µm) and CIGS (4 µm) to test the behaviour of the CIGS solar cell. Mo film is invariant with respect to bending up to α = 70°, then the surface roughness increases suddenly from 4 to 6 nm and the electrical performances irreversibly change. CIGS films are morphologically invariant versus bending up to α = 180°, while the electrical resistance changes drastically after α = 70°.

On the base of these experimental results, a critical angle α_C of 70° is set as limit to preserve both mechanical and electrical properties of Mo and CIGS films. This angle corresponds to a wheel radius of 1.5 cm, but a safer roll-to-roll system with 5 cm wheel radius has been realized to make it suitable for any polymeric substrates.

An enabling signal on the HV unit, switches on/off the pulsed electron ablation. The growth of thin films determines the movement of the flexible substrate, by interconnecting the PED source, the reel-to-reel motors and the thickness monitoring system. All the operations are controlled by a LabView program and a friendly user screenshot visualizes the cell final length and the time required to perform the whole process, allowing to change the parameters during the operation.