Skip to main content

Advanced Gratting for Thin Films Solar Cell

Final Report Summary - AGATHA (Advanced Gratting for Thin Films Solar Cell)

Executive Summary:
See enclosed "Final report.pdf"

Efforts developed by the European consortium during the AGATHA project has been dedicated to the production of textured substrates and to the fabrication of solar cells and mini-modules of the different photovoltaic technologies on these substrates. Simulation tasks have been carried out to assess the benefit of new substrates on photovoltaic performances of solar cells.
• Simulation activity
Based on material properties obtained in the first period of the project from optical measurements carried out by the different partners, simulation activities have been used to describe the optical behaviour of solar cells and predict the influence of textured substrate. As the consortium has no experience in term of optical simulation for the CIGS technology, a particular attention has been paid to the definition of the model. A 3D optical modelling of the CIGS solar cell taking into account roughness of all layers has been developed and fits very well the experimental measurements. For Si thin film technologies, models were already available at TUD.
As textured substrates finally used in the AGATHA project have been fabricated alternatively by the European consortium, the design of the features originally forecasted in the project was not achievable. Simulations conducted in the first part of the project were thus useless and new simulations have been conducted with the new substrates. A first step of substrate characterization and modelling has been necessary before simulation of solar cells can be performed.
Last, optical simulation of different photovoltaic technologies (CIGS and a-Si:H/µc-Si:H double junction) have been performed on their respectively available textured substrates. Valuable insights on light trapping properties as function of texturation parameters have been obtained.

• Substrates fabrication
As no substrate has been received from the Indian consortium during the whole AGATHA project, European consortium decided to fabricate their own textured substrates. Two kinds of substrates have been employed. First, TUD developed randomly textured substrates obtained with chemical etching of sacrificial layer and transfer to glass of the features. On top of these substrates, a TCO layer is deposited and an additional etching step is performed to provide the small scale texturation. These substrates, used for Si-based technologies, are later called chemically etched substrates and are compared with reference substrates consisting in flat glass with etched TCO.
In parallel, CEA fabricated periodically textured substrates with the use of SiO2 microspheres of different sizes buried into glass substrates and covered with Mo back electrode. No small scale texturation has been brought because it has been demonstrated by numerical simulation that it would not have a positive impact on photogenerated current. These substrates, called microspheres textured substrates have been used for the CIGS technology and are compared with Mo coated flat reference glass.
• Solar cells implementation
Solar cells and mini-modules with ultrathin absorber layers have been deposited on these new substrates. Concerning Si-based technologies, a-Si:H and µc-Si:H single junction solar cells have been deposited on chemically etched substrates to assess the optical gain using MST. On the same type of substrates, a-Si:H/µc-Si:H tandem mini-modules have been fabricated. For CIGS technology, two deposition routes (1-stage and 3-stages coevaporation processes) have been first compared on flat reference substrates. The use microspheres textured substrates to increase current for ultrathin solar cells has been successfully tested for the 1-stage deposition route.

Project Context and Objectives:
See enclosed "Final report.pdf"

3.2 Project objectives for the period
3.2.1 Overview
During the last period of the AGATHA project (M54 – M72), objectives do not exactly fit with the Annex I of the Grant Agreement for two main reasons:
• The lack of textured substrates received by the consortium and consequently related delays in the previous period
• The modification of objectives of the project validated during the mid-term review.
The main goal of AGATHA project lies in the proof of concept of fabricating efficient thin film solar cells from different technologies (a-Si:H, CIGS, µc-Si:H and tandem a-Si:H/µc-Si:H) with ultrathin absorber layers on modulated surface texture (MST). To achieve these final goals at the end of the project, two new objectives have been added in the period (n.1 and n.2) and one objective has been modified from Annex I of Grant Agreement (R7.2). The 3 other objectives (5.1 5.2 and 5.3) are identical to Annex I. The list of objectives, achievements during the project and documents in which they are reported are given in the table below.
# Objective within the period Reported in Achievement
n.1 Fabrication of MST on glass substrates for Si-based thin films solar cells and modules D05.01
- µm-scale textured glass substrates fabricated by etching of a sacrificial ITO layer on top of a flat Corning XG glass
- MST obtained with HCl etching of the ZnO layer deposited on top of the textured glass
n.2 Fabrication of MST on glass substrates for CIGS solar cells Final Report
Section - Textured glass substrates fabricated by burying SiO2 microspheres into soda-lime glass.
- No modulation with Mo NPs: simulation show the lack of interest of this approach
(MS02) a-Si:H solar cell on MST with 150 nm thick absorber and JSC=14 mA-cm-2 Final report
Section - 12.0 for a 150 nm thick a-Si:H solar cell on textured substrate
- Solar cells with 150 nm absorber layer on textured substrate perform better than solar cells with 250 nm and 350 nm absorber layers on textured substrates
5.2 CIGS solar cell on MST with 600 nm thick absorber and JSC=30 mA-cm-2 Final report
Section - 26.9 for a 570 nm thick CIGS solar cell on textured substrate (25.3 for the reference flat substrate)
- Current gain up to 4.1 and efficiency gain up to 5% relative for thicker samples
5.3 µc-Si:H solar cell on MST with 1000 nm thick absorber and JSC=24 mA-cm-2 D05.03 - ~22 for a 1000 nm thick µc-Si:H solar cell on textured substrate
- MST always underperform reference flat substrates with
(MS04) a-Si:H(150nm)/µc-Si:H(1600 nm) tandem mini-modules of 10x10 cm2 on MST D07.02 - 4 tandem mini-modules have been deposited on MST along with 1 on reference flat glass + etched TCO back contact
- Mini-modules on MST slightly underperform the reference one.
Table Objectives of the AGATHA project for its last period.
3.2.2 Follow-up of previous review (if applicable)
No comments have been made on the mid-term report submitted on the 29th of April, 2015.

Project Results:
See enclosed "Final report.pdf"

3.3 Work progress and achievements during the period
3.3.1 Progress overview and contribution to the research field WP01: Design of optimal modulated surface texture and glass embossing process using modelling (TUD)
The twin goals of the WP1 are (i) to determine the optimum design of modulated surface texture for obtaining maximal current enhancement in the three thin-film solar cells technologies developed in the AGATHA project (namely a-Si, CIGS, μc-Si) and (ii) to develop a model for hot embossing process for design, simulation and optimization of the embossing operations. In this final report only results related to the design of modulated surface textures are presented (carried out at TUD), since the modelling for hot embossing process was supposed to be carried out by Indian partners of AGATHA.
The work of WP01 carried out at TUD was divided in 5 tasks.
T01.01 (Calibration of 2-D optical model for simulations of periodic surface structures) consisted of:
• A thorough characterisation of materials used in the three thin-film solar cell technologies developed in AGATHA across the three indicated cell manufacturers in the project (CEA, FZJ and TD), in order to obtain reliable material properties to be used in the optical simulations carried out in subsequent tasks. The possibility to measure and analyse the same materials at different institutions (CEA, FZJ and TUD) resulted in better and more trustworthy characterization, which in turn ensured that results of optical simulations could be fully trusted.
• Calibration of the employed optical simulation software to ensure that simulations could correctly predict the real behaviour of manufactured devices:
o For silicon-based devices, results published previous to the beginning of AGATHA by TUD showed that simulations agree with measurements of real devices.
o For CIGS solar cells, one of the major achievement of this WP was to demonstrate (for the first time) the calibration of a 3-D rigorous optical simulator for CIGS devices. In particular, the parameters which are fundamental to achieve correct simulations results of CIGS architectures were clearly identified (optical properties of materials, correct thickness of layers, implementation of natural roughness resulting from deposition of materials, determination of minimum simulation domain to achieve randomness of the interface textures). The results of this calibration were presented at an international conference dedicated to photovoltaics (PV) and were included in a manuscript which has been published in an international peer-reviewed journal.
o A peculiarity of AGATHA was the development of asymmetric structure to enhance the absorption in different thin-film solar cell technologies. For this reason, calibration of the simulation software was required to ensure that the asymmetric diffraction of light due to the presence of asymmetrically textured interfaces could be correctly modelled. Results of this phase confirmed the behaviour predicted by theory and the findings of work previously published by other groups. This calibration, alongside results of following tasks of this WP, were presented (in the form of a poster) at an international conference dedicated to PV and published in the proceedings of the same conference.

The goal of T01.02 (Optimal design of asymmetric periodic surface structures) was to determine the optimal dimensions of periodic asymmetric gratings for the three thin-film solar cell technologies investigated in AGATHA:
• For amorphous silicon (a-Si:H), optimal period and height of the grating (1 μm and 300 nm, respectively) were obtained from studies previously published by TUD. The optimal grating angle (which indicates the asymmetry) of the periodic textures was found to be dependent on the thickness of the absorber layer. However, the main result was that the optical performance of a-Si:H architectures on asymmetric periodic structures is similar for a wide range of grating angles, and that asymmetric gratings clearly outperform their symmetric counterpart.
• Similar conclusion can be drawn from the analysis of micro-crystalline silicon (µc-Si:H) structures with respect to the asymmetry. Once again the simulated performance of µc-Si:H devices on asymmetrically textured substrates was higher than for symmetrically textured ones, and the optimal angle depended on the thickness of the absorber material. On the other hand, since µc-Si:H has different optical properties than a-Si:H, period and height of the grating needed to be different. In particular, a grating period of 2 μm and a grating height of 900 nm were selected. The dimension of the period was limited by the indicated manufacturing capabilities of AGATHA’s Indian partners. The height value was selected considering its impact on both optical and electrical performance of a real device (i.e. high enough for good light in-coupling, not too high to avoid issues with deposition). The results of both a-Si:H and µc-Si:H architectures on asymmetric gratings, together with the calibration of the software for asymmetric diffraction of light, have been presented at an international conference dedicated to PV and were included in the proceedings of the same conference.
• For CIGS solar cells results are significantly different tan for silicon-based devices. In fact, from simulation it clearly emerges that asymmetry of the gratings does not provide any advantage over symmetrical ones. This is due to the fact that CIGS is a material with high absorptivity, hence light-trapping approaches are not very effective unless ultra-thin absorbers are employed. Direct consequence of this is that the most important aspect of light management becomes the in-coupling of light into the CIGS layer, which can be achieved by reducing primary reflectance of the solar cell and by using more transparent materials at the front side. Additionally, the presence of molybdenum as back metal contact showed to be another major source of optical losses in CIGS architectures. Results of this phase were not directly published, but were of paramount importance for a better understanding of the optical behaviour of CIGS devices. This accumulated knowledge was instrumental for the publication of a paper in an international peer-review journal, where also the software calibration described for T01.01 was included.
T01.03 (Simulations of modulated surface-textured structures) and T01.04 (Optical modelling of a-Si:H, µc-Si:H and CIGS solar cells on substrates with modulated surface texture) both dealt with the optimization of modulated surface textures to be applied to the three thin-film solar cell technologies investigated in AGATHA:
• For both a-Si:H and µc-Si:H architectures, modulation was achieved by superimposing the typical texture of chemically etched aluminium-doped zinc oxide to the (larger) asymmetric periodic gratings optimized in the previous phase of the project (T01.02). Results showed that the implementation of the modulated surface texture (MST) results in an optical performance further enhanced with respect to devices in which only periodic gratings are included. For solar cells with 150-nm thick a-Si:H absorbers, the best result achieved was an implied photocurrent density (Jph) of 17.3 mA/cm2, a value slightly higher than the goal of AGATHA (17.0 mA/cm2). For µc-Si:H architectures with 1 μm-thick absorbers the best result was a Jph of 26.4 mA/cm2, a value significantly higher the goal set at the beginning of the project (24.0 mA/cm2).
• In CIGS architectures, modulation of the substrate was to be achieved by including small molybdenum nanoparticles on top of the textured molybdenum back reflector, to take advantage of surface plasmonic effects taking place at the textured interface between metal (molybdenum) and semiconductor (CIGS). However, simulation results showed how the added presence of nanoparticles of molybdenum only increase the absorption of light in the metal, and no apparent increase of absorption in the CIGS layer was observed. The conclusions drawn from this set of simulations confirms what was observed in the previous phase of AGATHA (T01.02) in which the presence of molybdenum at the back side of a CIGS device is a major source of optical losses for this type of solar cells. The best achieved value of Jph for architecture with a 600 nm absorber was 29.8 mA/cm2, very close to the 30.0 mA/cm2 goal of AGATHA.
In conclusion, for WP01 all tasks to be conducted at TUD (T01.01 T01.02 T01.03 and T01.04) were completed in time. Results of simulations showed that both silicon- and CIGS-based devices have the potential to improve their efficiency to the level of AGATHA’s goals. The work conducted in WP01 was presented in two international conferences dedicated to PV and resulted in the publication of two articles (one in an international peer-reviewed journal, the other in the proceedings on one international PV conference).
Use of resources
TUD did not claim additional resources for this WP during the second period. The use of resources corresponds to the Grant Agreement Annex I. WP03: Conformal TCO and Mo deposition and texturation (MD)
Tasks of WP03 for the European consortium were planned in the first part of the project (M39 – M54) and no delay has been encountered in this WP. However, the mid-term report stated the failure in obtaining large (> 30 nm diameter) Mo nanoparticles (NPs) but some indications in the last experiments suggested that larger NPs were obtained (see Fig. 3.2.17 of the mid-term report).
Transmission electron microscopy (TEM) images have been carried out at CEA to assess the existence of bigger NPs as suggested by mass spectra measurements (see deliverable D03.03). New NPs have been deposited directly on TEM grids and 5 deposition conditions supposed to provide the biggest NPs have been observed. Two images of such NPs at different magnifications are depicted in Fig. These pictures reveal that the big features measured by mass spectra consist are made of “Mo chains” consisting in small (10 – 20 nm) NPs aggregated together. Same kind of Mo chains have been observed for all conditions showing mass spectra particles > 20 nm.

Figure TEM images of Mo NPs deposited on TEM grids.
During the AGATHA project, objective of depositing 300 nm diameter NPs has not been achieved. The biggest Mo NPs are 15 nm diameter and bigger features have been obtained with “Mo-chains”. However, TUD demonstrated with numerical simulations that the presence of Mo NPs in the range 10 – 300 nm would not lead to increased photocurrent and even would be detrimental since it would largely increase optical losses in the Mo back contact. Thus, this objective failure is not impacting at all the AGATHA project and it demonstrated again the importance of numerical simulations in the implementation of new structures.
Use of resources
• MD did not claim additional resources for this WP during the second period of the project.
• CEA was initially not involved in this WP but due to its participation during the whole project as characterization support of MD, it finally claimed 1.62 PM for the TEM analysis (see form C). WP05: Implementation of modulated structure on various thin film technologies (TUD)
WP05 was initially planned to end in August 2015 (M60). However, due to the lack of substrates provided by the Indian consortium, additional tasks have been implemented in WP05 to fabricate these missing textured substrates. These tasks have been carried out by TUD and CEA for silicon solar cells and CIGS solar cells respectively. As a consequence, solar cells fabrication with ultrathin absorber layers have been postponed and delayed until the end of the project (August 2016, M72). Thus, very recent results unpublished in D05.01 and D05.02 are described thereafter.
a. a-Si pin junction technology POC
In this section the manufacturing procedure and the results achieved for single-junction (hydrogenated) amorphous silicon (a-Si:H) solar cells are presented.
i. Textured substrates fabrication

Figure Procedure employed to manufacture modulated surface textured substrates for a-Si:H solar cells 1) A sacrificial layer (ITO) is wet-chemically etched to create large crater-shaped features on the glass substrate. 2) ZnO:Al is deposited on top of the textured substrates to act as front TCO. 3) The surface of the TCO undergoes another wet chemical etching step, to create smaller features. 4) The MST substrates which includes the front TCO, is complete.
Since the substrates designed and optimized in WP01 could not be manufactured by the Indian partners, a new type of modulated surface texture (MST) needed to be developed. For a-Si:H the modulation was achieved as follows (see also Fig.
1. Large features (3-5 μm) were manufactured directly on glass, by wet chemical etching of a sacrificial tin-doped indium oxide (In2O3:Sn, ITO).
2. After a layer of aluminium-doped zinc oxide (ZnO:Al, AZO) was deposited, to act as front transparent conductive oxide (fTCO), the sample was once again wet-chemically etched to create a texture with smaller features (~600 nm lateral size). The superposition of this small features onto the previously achieved μm-scale texture results in an MST substrate.
The process of wet-chemical etching of ITO and AZO was previously optimized at TUD for μc-Si:H and tandem devices, while it is expected to be less effective for a-Si:H solar cells.
ii. Device fabrication
On top of the substrate manufactured as described above, a-Si:H devices were manufactured following the indications of the simulation phase (WP01):
• p- and n-doped nc-SiOx were employed in place of p- and n-doped a-Si:H layers, owing to the higher transparency of silicon oxide.
• a TCO (AZO) was deposited between the n-doped layer and the metallic back contact, to reduce plasmonic losses.
In Fig. (a) a scheme of the structure of the device is shown. The thickness of all supporting layer was chosen based on the results of the simulations and on the previous experience of TUD in the manufacturing of a-Si:H single-junction devices. Three different absorber thicknesses were considered: 350 nm (state-of-the-art), 250 nm, and 150 nm (which is the goal of AGATHA), as can be observed in Fig. (b).

Figure (a) Scheme of the manufactured a-Si:H devices.(b) Picture of the rear side of the realized a-Si:H solar cells, with different a-Si:H layer thicknesses (150 nm, 250 nm, 350 nm).
iii. Results
The manufactured cells were characterized with external quantum efficiency (EQE) and current-voltage (J-V) measurements. Results are summarized in table and were obtained by averaging the measured values of current, voltage, fill factor and efficiency of all 30 cells deposited for each absorber thickness.
Thickness 150 nm 250 nm 350 nm
Jph (mA/cm2) 12.0 13.3 13.7
VOC (V) 0.84 0.84 0.83
FF (-) 0.62 0.56 0.53
η [%] 6.25 6.26 6.06
Table Summary of measured external parameters of a-Si:H cells with different absorber thickness.
It can be observed that the voltage remains constant for all thickness, while the current density increases with thickness and the fill factor decreases for larger values of the thickness. The obtained values for the 150 nm thick sample are lower than the goals of AGATHA (Jph = 14 mA/cm2, VOC = 0.9 V).
iv. Use of resources
The results of this task (T05.01) were significantly delayed (12 months) with respect to the original plan. This happened due to the incapability of AGATHA’s Indian partners to provide asymmetrically and periodically textured substrates in time. Despite this, the use of resources was not increased and corresponds to the Grant Agreement Annex I.

b. CIGS based pin junction technology POC
As described in deliverable D05.02 and in the mid-term report, ultrathin CIGS solar cells have be deposited on textured substrates provided by TUD in the first period of the project. However, these first attempts were unsuccessful because scale and aspect ratio of texturation did not match with requirements for light trapping in CIGS solar cells. Thus, fabrication of new substrates with feature sizes close to the optimized values found in WP01 has been carried out at CEA and is described below. Additionally, improvement of the deposition process for ultrathin CIGS solar cells has been performed and is detailed in a second section. Optical performances of textured substrates for ultrathin CIGS solar cells is then highlighted.

i. Textured substrates fabrication
The fabrication of the textured glass substrates is based on the deposition of a monolayer of silica spheres on a soda-lime glass (SLG) substrate followed by an annealing step to burry these spheres into the glass. In this study, 3 different sphere sizes have been used (diameter of 0.5 µm, 1.0 µm and 2.5 µm) to obtain textured substrates with different feature sizes (called S-0.5 S-1.0 and S-2.5 respectively).
An enhanced Langmuir-Blodgett technique oriented to very large planar or non-planar substrates (called Boostream) is used is used to deposit this monolayer of spheres as described in reference . This method allows homogeneous deposition on large surface (> 1 m2) at low cost (CEA estimation for spheres deposition is 3 €-m-2). The periodicity of the texturation is linked to the shape and dimension of particles as well as the deposition technique. During Langmuir-Blodgett process, spheres automatically arrange in a hexagonal close-packed lattice during transfer from solvent to substrate. This periodicity is well controlled for 0.5 µm and 1.0 µm spheres but becomes less pronounced for 2.5 µm spheres.
After spheres deposition, a short etching step in SF6 + CHF3 + O2 gaz at 50 mTorr is used in order to remove organic residual solvents used during process deposition. Substrates and spheres are then annealed in a lamp furnace at 800°C for 30 s (parameters are varied in the 20°C/10s range depending on the sphere size) under inert Ar atmosphere to obtain substrates with two-thirds buried spheres. Scanning electron microscopy images of S-1.0 textured substrate is depicted in Fig. (a). For all substrates, height of the pattern is about 1/3 of the sphere diameter and period of the texturation can be approximated by the particle diameter. Aspect ratio (defined as pattern height on the period) is thus close to 1/3. These values are close to those use by TUD for their simulations in WP01.

Figure (a) SEM images of S-1.0 textured substrates. (b) Structure of the solar cell deposited on textured substrates.
These textured substrates are then integrated in the structure depicted in Fig. b. A 50 nm SiO2 diffusion barrier is deposited with E-beam evaporation in an Oerlikon Univex chamber and in-situ densified by ion beam bombardment on top of the buried microspheres. A typical Mo/CIGS/CdS/ZnO/ZnO:Al solar cell is then deposited simultaneously on a flat SLG + SiO2 diffusion barrier reference substrate and on a textured substrate in order to assess the optical gain allowed by texturation.
For each type of textured substrate (S-0.5 S-1.0 and S-2.5) with their respective flat reference substrates, 3 CIGS thicknesses have been used (targets of 600 nm, 1100 nm and 2000nm) and 2 routes of CIGS co-evaporation have been tested (1-stage and 3-stages co-evaporation processes, see next section). The list of the samples is given in table along with the exact thickness of the CIGS absorber and the Ga/(Ga+In) (GGI) ratio measured by X-Ray fluorescence.
On each sample, 9 solar cells (0.5x1 cm2) have been defined.

Table list of CIGS solar cells.
ii. CIGS coevaporation route
Two co-evaporation routes have been tested for CIGS deposition in the AGATHA project; both of them have been performed in an Alliance Concept EVA450 deposition chamber. The first one is a classical 3-stages deposition process implying a composition (GGI = [Ga]/([In]+[Ga])) and bandgap gradient in the depth of the absorber layer. Another deposition scheme has been tested as well, which consists in co-evaporating simultaneously at constant fluxes Cu, In, Ga and Se on Mo coated substrates at 550°C. This process, called 1-stage process, is supposed to give lower efficiencies than the 3-stages process but presents other advantages: an easier control of the deposition thickness and a better assessment of the influence of absorber thickness on solar cells properties. Indeed, in the case of 3-stages process, composition gradient are not scalable with the absorber thickness. For a fixed global GGI in the absorber, varying the thickness will modify slopes of the composition gradient and/or GGI at the interfaces and notch (position of minimum bandgap). Additionally, is has been argued in reference that 1-stage deposition process could decrease CIGS deposition cost by almost a factor 2.
The photovoltaic properties of all the cells (list in table fabricated on flat reference substrate with 3-stages and 1-stage co-evaporation processes as function of absorber thickness are depicted in Fig. In the 3-stages case, the power conversion efficiency (PCE) of CIGS solar cells continuously decreases with absorber thickness (see Fig. (a)). The best efficiencies (12.9% due to the absence of Na) are obtained for the 1800 nm thick samples while efficiencies in the range 5%-7.5% are obtained for solar cells with 550 nm thick absorbers. A different behavior is found for the 1-stage co-evaporation process: efficiencies for the ultrathin samples are comparable or slightly better than in the case of 3-stages process. At intermediate thickness, both processes are still comparable but a saturation of PCE with absorber thickness is obtained at ~ 1100 nm. As a consequence, PCE for 1-stage process is notably lower for solar cells with nominal thicknesses (~ 2000 nm) than for 3-stages process. This behavior have been attributed to influence of parasitic resistances as shown in Fig. (b): series resistance have similar values for ultrathin samples but significantly increase with absorber thickness for the 1-stage process, leading to a limited efficiency at nominal thickness. Shunt resistances are similar for both processes except for ultrathin samples where 3-stage process reveal slightly lower values, which is reflected in the open-circuit voltage (VOC) and thus PCE.

Figure (a) PCE and (b) parasitic resistance of solar cells deposited on flat reference substrates (list of samples in table
These results confirm that the 3-stages process gives better efficiency for CIGS solar cells with nominal absorber thickness (ThCIGS ~2000 nm). However, as far as ultrathin absorber are targeted, 1-stage deposition process are very promising since they exhibit comparable or better efficiencies at reduced deposition cost. Thus, only 1-stage co-evaporation route has been used for the study on textured substrates.
iii. Current enhancement on textured substrates
All 1-stage co-evaporated CIGS layers presented in the previous section on reference flat substrates have been deposited simultaneously on textured substrates (S-0.5 S-1.0 and S-2.5). Adhesion of CIGS layer with ThCIGS = 1100 nm and ThCIGS = 1960 nm on S-0.5 texured substrate was not good enough and delamination occurred (visual and SEM observations), these samples have thus been removed from the graph. A comparison of photovoltaic properties (short-circuit current JSC and PCE) of solar cells fabricated on textured and reference substrates as function of absorber thickness is depicted in Fig.

Figure (a) JSC and (b) PCE comparison of 1-stage co-evaporated CIGS solar cells on textured substrates and flat reference substrates as function of CIGS thickness (ThCIGS)
S-2.5 texturation leads to JSC increase for thick (ThCIGS = 1640 nm) and intermediate (ThCIGS = 910 nm) CIGS (gain of 4.1 mA-cm-2 and 1.3 ma-cm-2 for ThCIGS1640/S-2.5 and ThCIGS910/S-2.5 respectively). S-1.0 texturation outperforms reference flat substrate only for intermediate CIGS thickness (current gain of 1.3 mA-cm-2 for ThCIGS1130/S-1.0) and the only texturation improving current for ultrathin CIGS absorber is the smallest one S-0.5 (current gain of 2.1 mA-cm-2 for ThCIGS570/S-0.5). For all other cases, texturation does not improve JSC in CIGS solar cells. These promising results in term of current enhancement are not perfectly reflected into PCE improvement (Fig. (b)). In two cases, texturation leads to better efficiencies: namely ThCIGS570/S-0.5 and ThCIGS1640/S-2.5 for which a relative increase in PCE of 2% and 5% relative are obtained. In the other cases (ThCIGS1130/S-1.0 and ThCIGS910/S-2.5) despite a better current, lower PCE is achieved. This result is attributed to an important degradation of the fill factor (FF) on textured substrate in these two sets of samples: a mean value of 40% is found against 60% for the flat reference substrates (in the samples with improved PCE, only a slight decrease in FF of about 2% absolute is observed) . Dark J-V curves show that RSh decrease from 5000 Ω-cm2 to 600 Ω-cm2 for ThCIGS1130/S-1.0 and from 4000 Ω-cm2 to 75 Ω-cm2 for ThCIGS910/S-2.5. Technological improvement of the co-evaporation process is still required to tackle these technological issues and to fully benefit the effect of substrates texturation on CIGS solar cells.
EQE measurements for samples with increased JSC due to texturation are shown in Fig. All curves show the same behavior: the maximum value of EQE at short wavelength is not modified by texturation while EQE is notably increased at long wavelengths were light absorption in CIGS is insufficient. The EQE gain at 1100 nm can exceed 40% for ultrathin CIGS deposited on 0.5 µm textured substrates (ThCIGS570/S-0.5). This improvement, only in the red part of the spectrum, as well as an unincreased maximum EQE are the hallmark of a current gain due to light trapping into the absorber rather than antireflection due to increased roughness.

Figure Comparison of EQE measurements of solar cells on textured substrates (continuous line) and flat reference substrates (dashed line) for which increase in JSC has been obtained.
In this study, we explored two routes for fabricating CIGS solar cells with ultrathin absorber layers. Additionally to the classical 3-stages co-evaporation route, we demonstrated that the 1-stage co-evaporation method can be chosen as far as ultrathin solar cells are targeted. The lower efficiency at nominal thickness due to higher RS is not observed for thinner samples and comparable or better performances are obtained for submicronic CIGS solar cells.
CIGS solar cells with ultrathin absorber thickness suffer from low JSC due to insufficient light absorption. We have used periodically textured glass substrates to improve light trapping into CIGS absorber layers and we have demonstrated that it allows to obtain current (up to 4.1 mA-cm-2, more than 6% relative) and efficiency (up to 5% relative) increase in our 1-stage co-evaporated solar cells.
iv. Objective achievement
A comparison between objectives described in the Description of Work and results obtained in the project is given in table This table shows that substrate texturation notably improves photovoltaic properties of ultrathin CIGS but is not sufficient to achieve quantitatively the objectives of the AGATHA project both in term of JSC and VOC.
• The lack of current is directly related to optical losses in the Mo back electrode. As demonstrated by TUD by optical simulations during the project (not shown), texturation of the back contact implies a strong increase in light absorption in the Mo and consequently current loss for the solar cell. Introducing a thin insulating (Al2O3 for instance) layer between Mo and CIGS can allow to quench optical losses in Mo and drastically increase JSC as demonstrated by TUD optical simulations. This solution is technically feasible but no time was found to implement it during the AGATHA project.
• The voltage deficit is well explained by the absence of Na in our CIGS layers: Na is well-known to improve VOC in CIGS solar cells . In this study, a SiO2 deposition barrier to prevent Na diffusion from the SLG has been used and thus no Na can be found in the CIGS layers. This strategy has been used for an easier comparison of flat reference substrates and textured substrates because of the unknown role of SiO2 microspheres towards Na diffusion. It is worth to notice that post-deposition treatments can be applied to both kind of solar cells (flat and textured substrates) to improve VOC but they have not been used because of missing time.
AGATHA objectives Flat ref. substrate Textured substrate (S-0.5)
Thickness 600 nm 570 nm 570 nm
JSC 30 25.3 26.9
VOC 690 mV 493 mV 506 mV
Table Comparison of objectives in the AGATHA project.
v. Use of resources
Description of tasks and results obtained in WP05 for the fabrication of CIGS solar cells on textured substrates has been summarized in the previous sections. These tasks have been largely delayed (12 months) compared to the original planning because of the lack of substrates provided by the Indian partners. Additionally, due to this absence of substrates, a considerable effort has been devoted by CEA to the substrates fabrication (development of a new process, optimization of the process and production of textured substrates for experimental plan of solar cells on textured substrates).
On another hand, CEA was supposed to work on performance, reliability, and lifetime evaluation of a-Si:H mini-modules produced by European partners in WP07. For the same reason (no textured substrates provided by Indian partners), fabrication of mini-modules on textured substrates has been delayed to the last month of the project (see deliverable D07.02) and no time remained for the advanced module testing. The effort dedicated to this task at CEA has been transferred to the fabrication of textured substrates for CIGS technology.
As described in Form C, CEA transferred the 16 PM effort of T07.03 (mini-modules testing) to T05.02 (fabrication of textured substates for CIGS based technology POC).

c. µc-Si pin junction technology POC (FZJ)
In this work package, microcrystalline Si single junction thin-film solar cells were prepared on textured glass substrates. In the first part, textured glasses with different morphologies are used as substrates. It is demonstrated that the topography of textured glass substrate has a significant influence on the light trapping effect of thin-film silicon solar cells. In the second part, the as-deposited ZnO:Al films on textured glass were textured again by HCl solution.
The microcrystalline silicon p-doped and intrinsic (i-) layers, and the amorphous silicon n-doped layer were deposited on the front contact ZnO:Al films using plasma enhanced chemical vapor deposition (PECVD) in a 30 x 30 cm² system at 13.56 MHz. The microcrystalline silicon i-layer is around 1.1 µm. For some solar cells, 80 nm ZnO:Al film was sputtered at room temperature on the n-doped Si layer to serve as back reflector. 700 nm Ag were thermally evaporated through a shadow mask, which defined the active cell area. The solar cell area was 1 cm2. In addition to the µc-Si:H pin solar cells deposited on textured glass substrates, reference solar cells were fabricated on wet-chemically etched ZnO:Al front contacts which exhibit the state-of-the-art texture for light trapping in µc-Si:H thin-film solar cells.
The photovoltaic parameters of solar cells were measured by a double source Wacom solar simulator under standard test conditions (AM1.5 100 mW/cm² and 25°C). Fill factor (FF) and open-circuit voltage (Voc) were evaluated for the solar cells. The measurement of external quantum efficiency (EQE) was performed under 0 V bias. Short-circuit current density (Jsc) was calculated by the spectra integral of the product of EQE data and the AM 1.5 spectra. The reflectance (R) of solar cells was measured from glass side by a UV-Vis-NIR photo spectrometer (PerkinElmer Lambda 950) in the wavelength range of 300- 1100 nm with a step of 2 nm. The internal quantum efficiency (IQE) of the solar cells was calculated as IQE=EQE/(1-R).
(1) Textured glass prepared by ion beam etching
A series of µc-Si:H thin-film solar cells with different ZnO:Al front contact thickness (60 nm, 250 nm and 600 nm) were prepared on the textured glass substrates fabricated by Ar ion beam etching. Figure 1 a and Figure 1 b show two 15×15 µm2 AFM images which correspond to the 3d etching mask and the textured glass substrate, respectively. The etching mask was prepared by etching ZnO:Al film in 0.5w/w% HCl solution and the ion beam etchant was Ar. Comparing the two AFM images, it is seen that the small features in the etching mask are disappeared in the textured glass, and only large craters remain. The root mean square roughness is 124 nm for the etching mask and 96 nm for the textured glass substrate. The textured glass is smoother than the etching mask, due to slightly lower etch rate of the glass material compared to that of the ZnO:Al film. Moreover, since more surface atoms are involved in the ion beam collision cascade for crater top as compared with crater valley, the etch rate of material on the crater top is higher than the others.

Figure Atomic force microscopy (AFM) images (15×15 µm2) of the 3d etching mask (a), the textured glass (b). The etching mask was prepared by HCl etching, and the textured glass was prepared by Ar ion beam etching.
The back contacts for these solar cells are ZnO:Al/Ag. Table gives the short-circuit current density (Jsc), open-circuit voltage (Voc) and fill factor (FF) of µc-Si:H thin-film silicon solar cells prepared at the same deposition run on various front contacts. Figure 2 shows the external quantum efficiency (EQE), total cell reflectance (R) and internal quantum efficiency (IQE) of these µc-Si:H pin solar cells. For the front contact, magnetron sputtered ZnO:Al films with different thicknesses were investigated for application in µc-Si:H solar cells made on textured glass substrates. For comparison, a reference solar cell was fabricated on wet-chemically etched ZnO:Al front contacts, named reference.
When the ZnO:Al films are 250 nm and 600 nm thick, there is reduced external quantum efficiency in the short wavelength range (400 – 600 nm). Comparing the EQE with the cell reflectance, it is seen that the EQE is reduced at the wavelength where reflectance maximum occurs. For the solar cell with 60 nm thick ZnO:Al film, the maxima in reflectance are shifted to wavelength shorter than 400 nm, so the EQE doesn’t have significant loss in the relevant wavelength range. The above observation indicates that similar to the solar cells on flat glass substrate, there are interferences in the thin ZnO:Al layer due to the conformal growth. To avoid significant loss in reflectance, the thickness of ZnO:Al layer needs to be around 60 nm. Due to the decreased parasitic optical losses for the solar cell with the 60 nm thick front ZnO:Al layer the internal quantum efficiency (IQE) is increased in the short wavelength range (400 – 600 nm) compared with the reference solar cell. All solar cells on this series of textured glass exhibit reduced spectral response in the long wavelength range (600 – 1100 nm). Possible reason is the smoother texture of the textured glass as compared to the state-of-art wet-chemical-etched ZnO:Al substrates. As a result, poor light trapping in the long wavelength range leads to reduced short circuit current (Jsc).
Although the solar cell with 60 nm thick front contacts does not have significant loss in the short wavelength range, it must be considered that the fill factor (FF) remains low due to relatively high sheet resistance (see Table 1). The sheet resistance of a 60 nm ZnO:Al film is around 150 Ω/☐, while the value of the state-of-art wet-chemical-etched ZnO:Al substrate is only around 7 Ω/☐. More conductive TCO layers, for example tin-doped indium oxide (ITO) films were examined in a separate series. Two µc-Si:H thin-film silicon solar cells were deposited on the same textured glass substrates made by Ar ion beam etching of HF textured etching mask. One front contact is 60 nm ZnO:Al film, and the other front contact is 60 nm ITO covered by 10 nm ZnO:Al film. The fill factor was improved from 0.28 to 0.57 by replacing ZnO:Al with ITO (Table Although there is still a reduced FF as compared with reference solar cells, it is possible to further improve FF by reducing the cell width. In addition, it shall be noticed that with regard to the industrially most relevant tandem thin-film silicon solar cells the decrease in FF is less severe. As the Jsc is much lower in these devices, the impact of the sheet resistance on the FF decreases.

Figure External quantum efficiency (EQE), reflectance R, and the internal quantum efficiency (IQE) of µc-Si:H solar cells prepared on the textured glass substrates with different thick ZnO:Al layers and the state-of-art wet-chemically etched ZnO:Al film.
Textured glass substrates with ZnO:Al front contact thickness of state-of-art wet-chemical-etched ZnO:Al substrate
60 nm 250 nm 600 nm
Jsc [mA/cm2] 18.1 19.3 18.1 24.7
Voc [mV] 522 510 520 510
FF [%] 37 67 73 67
Table Short-circuit current density (Jsc), open-circuit voltage (Voc) and fill factor (FF) of µc-Si:H thin-film silicon solar cells deposited on textured glass substrates made by Ar ion beam etching of HCl textured etching mask. The front contact thicknesses are 60 nm, 250 nm and 600 nm. In addition, the data of a reference solar cell prepared on a state-of-art etched ZnO:Al front contact is shown.
60 nm ZnO:Al 60 nm ITO + 10 nm ZnO:Al
Jsc [mA/cm2] 22.1 22.1
Voc [mV] 482 482
FF [%] 28 57
Table Photovoltaic parameters of µc-Si:H thin-film silicon solar cells deposited on textured glass substrates made by Ar ion beam etching of HF textured etching mask. The front contacts are 60 nm ZnO:Al film and 60 nm ITO plus 10 nm ZnO:Al films.
(2) Different ion beam etching made textured glass substrate
A series of µc-Si:H solar cells were fabricated on three different types of textured glass substrates as well as the state-of-art wet-chemical-etched ZnO:Al substrate. The topographies of the different textured glass substrates are shown in Figure b, b, b, respectively. On the textured glass substrates, the front ZnO:Al films are about 60 nm thick. The back contacts for those solar cells are ZnO/Ag.
Figure a and Figure b show two 15×15 µm2 AFM images which correspond to the 3d etching mask and the textured glass substrate, respectively. The etching mask was prepared by etching ZnO:Al film in 1w/w% HF solution and the ion beam etchant is Ar. The roughness is 100 nm for the etching mask and 80 nm for the textured glass substrate. Besides small craters, inhomogeneous hillock structures are observed on the textured glass.

Figure Atomic force microscopy (AFM) images (15×15 µm2) of the 3d etching mask (a), the textured glass (b). The etching mask was prepared by HF etching, and the textured glass was prepared by Ar ion beam etching.
Figure a and Figure b show two 15×15 µm2 AFM images which correspond to the 3d etching mask and the textured glass substrate, respectively. The etching mask was prepared by etching ZnO:Al film in HCl solution and the ion beam etchant is CF4. The roughness is 108 nm for the etching mask and 157 nm for the textured glass substrate. The textured glass is rougher than the etching mask, because the etch rate of glass material by CF4 etchant is faster than that of ZnO:Al film.
Figure Atomic force microscopy (AFM) images (15×15 µm2) of the 3d etching mask (a), the textured glass (b). The etching mask was prepared by HCl etching, and the textured glass was prepared by CF4 ion beam etching.
The distribution of surface inclination angle of the above three different types of textured glass substrates is compared in Figure The state-of-art wet-chemical-etched ZnO:Al substrate is included in the figure as reference. It is seen that the percentage of large angle is significantly increased by modifying the preparation process of the textured glass. Especially for the textured glass made by CF4 reactive ion beam etching, the maximum of the distribution locates at approximately 20º, which is similar to the state-of-art wet-chemical-etched ZnO:Al substrate.

Figure Angle distribution of the surface texture of the textured glass substrates made by three different parameters and the state-of-art wet-chemical-etched ZnO:Al substrate (reference). The AFM images of the texture glass substrates are already shown in Figures 1.2(b) 1.3(b) and 1.4(b).

Figure External quantum efficiency (EQE) of µc-Si:H solar cells based on the textured glass substrates made by three different parameters and the state-of-art wet-chemical-etched ZnO:Al substrate (reference). The short-circuit current density values of these solar cells are noted in the figure.
The external quantum efficiency (EQE) and the short-circuit current density (Jsc) of these solar cells is given in Figure Comparing with the textured glass made by Ar ion beam etching of the HCl textured etching mask, the two modified textured glass substrates give much higher spectral response in the long wavelength range (600 – 1100 nm), resulting a similar Jsc as the state-of-art wet-chemical-etched ZnO:Al substrate. Hence, we conclude that the optimized textured glass texture was achieved using the ion beam etching of wet-chemical-textured etching mask.
(3) Solar cells on HCl textured ZnO:Al films on textured glass
In this part, a single experiment with wet-chemically etched ZnO:Al film on textured glass was performed. The textured glass substrate was the same as the one shown in Figure b. 800 nm ZnO:Al film was sputtered on the textured glass substrate. Then, the ZnO:Al film was etched in diluted HCl solution for 40 s. Microcrystalline silicon thin-film solar cell was prepared on this substrate. Back contacts for these two solar cells were ZnO:Al/Ag. Wet-chemically etched ZnO:Al film on flat glass was used as reference.

Figure External quantum efficiency (EQE) and cell reflectance (R) of the solar cell on HCl textured ZnO:Al on textured glass and flat glass (reference).
ZnO:Al texturation reference
Jsc [mA/cm2] 23 24.9
Voc [mV] 507 507
FF [%] 69 69
Table Short-circuit current density (Jsc), open-circuit voltage (Voc) and fill factor (FF) of μc-Si:H thin-film silicon solar cells deposited on wet-chemically etched ZnO:Al on textured glass and flat glass (reference).
The photovoltaic parameters of the two solar cells are shown in Table and the external quantum efficiency EQE and cell reflectance R are presented in Figure The open circuit voltage and fill factor of the two solar cells are the same, indicating that the silicon grows normally on the etched ZnO:Al film on textured glass. Even though the spectra response the solar cell on textured glass is slightly lower than reference solar cell, it still exhibit very high short circuit current and low cell reflectance. As a result, the efficiency of the single junction microcrystalline thin-film solar cell is higher than 8 %.
Use of ressources
The use of resources corresponds to the Grant Agreement Annex I. WP06: Characterization of texture optical functionality (FZJ)
All technical results within this Work Package have been presented into D06.02.
Use of ressources
The use of resources corresponds to the Grant Agreement Annex I. WP07: Global assessment on a-Si minimodules (3SUN)
All technical results within this Work Package have been presented into D06.02.
Use of ressources
The resources have been directed to WP07 to make all the efforts to be ready to use full size textured glasses. In case glasses would have been delivered at the very end of the project, we were been able to process them thanks to the tests performed during WP07. Different people within 3SUN have been involved in the tests from different areas to assess all the aspects related to the production of a full size modules. Experts in quality, encapsulation, lamination and deposition have been involved to fully cover and anticipate all the activities necessary to have the higher amount of data related to use of the new glass. Thanks to this work also the evaluation of the cost vs benefits has been performed in an accurate way to give at least a good indication to the possibility of using this kind of glass. This work can give also a baseline to define thresholds also for other process steps or materials changes. WP09: Project management (CEA)
Activities related to WP08 are described in section 3.5.1. WP09: Dissemination and exploitation of project results (CEA)
Activities related to WP09 are described in section 3.5.2. WP10: EU and India scientific and technical coordination (CEA)
Activities related to WP10 are described in section 3.5.3.

Potential Impact:
See enclosed "Final report.pdf"

Potential impact
Due to the fast changes in the photovoltaic industrial environment in past few years, the 3 years delay for the project start seriously affected the potential economic impacts of AGATHA project. Particularly, from 2010 to 2013 (official start date), thin films Si technologies had to face an ever harsher competition in term of efficiencies and only tandem devices are currently on the market. The lack of improvement in devices efficiencies along with over costs due to substrates fabrication make this approach hardly suitable at the industrial level in 2016. Concerning CIGS, the use of textured substrates can be considered only if a parasitic absorption in Mo can be faced and demonstration of low cost textured substrates fabrication is made.

Dissemination activities and exloitation of results
Dissemination activities, use of knowledge and exploitation activities have been carried out in WP09.
The goal of WP09 was to develop dissemination and exploitation activities of the European consortium during the AGATHA project. It was the main interface between the project and the academic as well as industrial outside world. Two tools were used to facilitate communication tasks during the whole duration of the project:
• a private groupware platform was available for all partners. All official documents (Grant Agreement, Consortium Agreement, Cooperation Agreement, Description of Work) as well as deliverables and presentations were regularly updated and directly downloadable.
• a public website with regular updates of the project public news, events and publications. It can be accessed at the web address:
Work in WP09 was performed by the Project Management Office mainly during the Steering committees which took place in each European Progress Meeting. The periodical steering committee meetings have been fully and usefully used for the sharing of information and particularly as far as exploitation of project results are concerned:
1. Status of IP protection, discussion of ideas which needed protection (only for European consortium)
2. 3 patents deposited by CEA about AGATHA activities
3. All forecasted publications were shared within the consortium before publication (information for all the partners)
Several papers have been published or submitted to international peer review journals specialized in photovoltaics physics and condensed matter physics during the project as well as presented in international conferences. The list of publication is given below:
Publications in per-reviewed journals:
1. W. Zhang, U. W. Paetzold, M. Meier, A. Gordijn, J. Hüpkes, T. Merdzhanova, Development of thin-film silicon solar cells on textured glass, Energy procedia 44 (2014) 151.
2. C. Onwudinanti, R. Vismara, O. Isabella, L. Grenet, F. Emieux, M. Zeman, Advanced light management based on periodic textures for Cu(In,Ga)Se2 thin-film solar cells, Optics Express 24(6), A693-A707 (2016)
3. R. Vismara, O. Isabella, M. Zeman, Optimization of thin-film silicon solar cells based on 1-D asymmetric periodic gratings, Proceedings of the 31st European Photovoltaic Solar Energy Conference and Exhibition, pp. 1176–1180 (2015).
4. L. Grenet, F. Emieux, O. Dellea, A. Gerthoffer, F. Roux, S. Perraud, Influence of coevaporation process for ultrathin CIGS solar cells and current enhancement with periodically textured glass substrates, submitted to Thin Solid Films
5. L. Grenet, F. Emieux, R. Vismara, O. Isabella & al. Optical analysis of co-evaporated CIGS solar cells on textured glass substrates, in preparation
Participation to Conferences:
• Oral presentations
1. Srinivas Rao Saranu, David Joyce, An advanced and low cost light trapping technology for thin film solar cells, International Conference on Plasma Surface Engineering in Garmisch-Partenkirchen, Germany (2014).
2. W. Zhang, U.W. Paetzold, M. Meier, A. Gordijn, J. Hüpkes, T. Merdzhanova, Development of thin-film silicon solar cells on textured glass, EMRS 2013 Spring Meeting, Strasbourg, France (2013)
3. R. Vismara, C. Onwudinanti, L. Grenet, F. Emieux, O. Isabella, M. Zeman, Optical Optimization of Cu(In,Ga)Se2 solar cells on periodic gratings, Light, Energy and the Environment Congress 2015 in Suzhou, China (2015).
• Posters
1. T. Merdzhanova, C. Zhang, U.W. Paetzold, A. Lambertz, V. Smirnov, A.Hoffmann J.Kirhhoff M. Meier, A. Gordijn, Adaptation of reflection and transmission of µc-SiO:H intermediate reflector in tandem solar cells, EMRS 2013 Spring Meeting, Strasbourg, France (2013)
2. R. Vismara, O. Isabella, M. Zeman, Optimization of thin-film silicon solar cells based on 1-D asymmetric periodic gratings, 31st European Photovoltaic Solar Energy Conference and Exhibition in Hamburg, Germany (2015)
Dissemination activities also included organization of specific events (summer schools and workshops) between European and Indian consortia as well as research exchange program. These events were originally included into the Indian tasks and budgeted by DST. Due to funding cuts by DST, India was not been able to organize and host these events. This modification of the DST funding was announced by Dr. Sanyal, coordinator of the Indian consortium during the Kick-off meeting in September, 2013.
Both consortia explored together alternative solutions of funding in order to be able to organize one joint event (workshop + summer school) during summer 2015. Thus, an application was submitted by Dr. Sanyal (CGCRI, India) and Dr. Merdzhanova (Forschungszentrum Jülich GmbH, Germany) to IGSTC (Indo-German Science and Technology Centre) Workshop Call 2014, but it was announced in June, 2015 that this event was not recommended for funding by IGSTC. No solution has been found for organizing dissemination events in the framework of AGATHA project.
In conclusion, no dissemination event (workshop or summer school) has been organized during the AGATHA project due to DST funding cuts.

List of Websites:
Contact: Dr. Louis Grenet – CEA, 17 rue des Martyrs 38054 Grenoble Cedex 09 France –