High Concentration PhotoVoltaics HCPV
HCPV is the most common CPV systems with the best industrial maturity. HCPV concerns concentration factors higher than 150, which are mainly obtained using Fresnel lenses and to a lesser extent micro-dish or dense array receivers . All these technologies are called “imaging” (by definition they form the source image on the focal point) and consequently can concentrate only direct sunlight. As seen above, they are interesting only in sunny regions.
Fresnel lenses have the same optical function than conventional lenses, their particularity lies in the fact that all the material not needed to perform concentration has been removed (see fig. 1.8). This design makes lens more compact, which eases their implementation, reduces the con-centrator weight as well as wind pressure. More details on Fresnel lens concentrator can be found in this seminal paper .
Common HCPV concentration factors vary between 400 and 700, but values higher than 2000 have been reported . The world record, (in terms of module eﬃciency) is hold by Semprius with 35.5% eﬃciency for Cgeo = 1100 (ηopt ≈ 0.8) . It is worth noting that eq. (1.6) corollary is that sun tracking system must be extremely accurate to achieve good light concentration with such high concentration factors. In addition, HCPV requires active cooling to keep a reasonable temperature in the solar cells.
What should be understood is that HCPV goal is to achieve cost reduction through expensive semi-conductor material saving. Because concentrating optics is expensive (mainly tracker), targeted solar cells for HCPV are the most eﬃcient (III-V multi-junctions). High concentra-tion optics cost plus additional cost of highly eﬃcient cells (compared to c-Si or thin film) is counterbalanced by the material saving granted by HCPV. Rather, HCPV seems not adapted to concentration on “low cost” solar cells because the material saving induced cost reduction is not high enough.
Medium Concentration PhotoVoltaics MCPV
Medium CPV is a kind of trade oﬀ between high optical concentration and high industrial capacity (10 < Cgeo < 150). It is usually done on c-Si PV cells and requires less accurate sun tracker than HCPV. Even a one axis tracking system is sometimes suﬃcient . Popular optics for MCPV is Fresnel mirror, which are based on the same technology than Fresnel lens. Fresnel mirror enables to make thin a parabola cutting it in diﬀerent segments. Concentration factors are lower than parabolic compounds but industrialization is easier and cheaper. Figure 1.9 illustrates an example of MCPV: Achieving higher concentration requires other optics, such as prismatic lenses reaching Cgeo = 120 . Going lower in concentration level, we end up with low concentration photovoltaics.
Low Concentration PhotoVoltaics LCPV
Concentration factor lower than 10 belongs to low CPV. Generally LCPV does not require sun tracking system (or simple ones) and consequently are not expensive. Furthermore, the low concentration factor does not require heat dissipation systems. Compatible with low cost cells (thin films or c-Si), LCPV can be a way to increase module performances with a soft extra-cost. Diﬀerent technologies have been studied. Among them we can highlight:
• Prism Array Concentrator.
• Parabolic mirror.
• Non fluorescent flat plate concentrator.
• Luminescent solar concentrator.
Luminescent Solar Concentrator place in CPV landscape & consistency with low size PV cells
Working with low cost photovoltaic cells (thin films), concentrator must be low-cost to preserve the whole system competitiveness. In this respect, concentration optics should not comprise expensive sun-tracker, and consequently θacc should not be to low. Typically, a value of θacc ≈ 7.5◦ allows to track the sun only one time per hour. HCPV is therefore not adapted (see eq. (1.6)). Taking this line of thought further, any concentration system with θacc 6=π/2 concentrates light on a focal point, which can be an additional challenge if used with extremely low size PV cells. Superimposing micron-size focal points to micron-size solar cells taking into account Sun movement is challenging. As such, θacc = π/2 systems become even more interesting when PV cell has micron size. In addition, these systems can concentrate diﬀuse light, whose proportion exceeds 50% in temperate regions. Unfortunately, eq. (1.6) greatly limits the concentration factor.
Luminescent Solar Concentrator may be an alternative because this technology has numerous desired benefits (θacc = π/2, low-cost) but also may have concentration factors exceeding the n2 limit derived of eq. (1.6). For all these reasons, we have chosen to evaluate the light concentration potential on micro CIGS-based solar cell with a luminescent solar concentrator (part I). Part II of this work intends to study an innovative light concentration alternative inspired from LSC, in response to LSC fundamental limitations.
Lowering front loss: Photonic Band Stop & dye anisotropic emission
As previously written, front trapping is performed by TIR whose eﬃciency depends only on refractive indexes and photons angle. Enhancing TIR eﬃciency requires:
1. Increase Δn = nmat − nair.
2. Anisotropic dye emission to reduce emission in the escape cone.
Implementing the first solution implies an increase of Lext not mentioning diﬃculties to find a proper material which is still transparent and able to host dyes. Nevertheless clearly a trade-oﬀ must be found on refractive index between Lext and Lfront, which will be done later. The second solution would have been far more eﬃcient except that anisotropic dye molecules are hard to conceive and not as eﬃcient (PLQY, absorption and emission spectra). Some work on this field can be found in . A third solution is to take advantage of the fact that photon wavelength after dye emission is red shifted and use an optical filter designed to reflect luminescence. This concept was first proposed by U. Rau . Indeed, the incoming solar spectrum is broad-band but the dye action enables to diﬀerentiate three spectral regions illustrated in fig. 2.6.
Modeling: Algorithm & Flowchart
Following the LSC description adopted in this thesis, there are two main technical solutions to compute a LSC: ray trace and thermodynamic modeling. Thermodynamics-based calculation was not addressed in this thesis, but explanations can be found in A. Chatten and co-workers publications [68, 69], as well as a comparison between the two approaches , which concludes that both methods are accurate. A third way of modeling, in Finite-Diﬀerence Time-Domain (FDTD), has been investigated  in the case of LSC embedded in a photonic crystal (to reduce Self-Absorption). Although eﬃcient, it was done only in 2D.
Ray trace technique is a very common way of modeling LSC using a Monte Carlo algorithm [72, 73, 74]. A home-made code (LSC Vectorized Monte-Carlo Algorithm LSC-VMCA) has been written during this thesis in order to be able to modify it and obtain all the desired data. This section describes its principle, considered events and the inputs/outputs.
By definition a Monte-Carlo method is based on random events to determine a numerical nb. incoming photons value. In our case the numerical value will be the LSC optical eﬃciency ηopt = . nb. collected photons.
Derive η opt is interesting but do not help to understand the physics of LSC. Yet, knowing how the remaining photons (1 − ηopt) have been lost (which loss channels, which spectral reparti-tion) contributes to understand the value of ηopt. Monte-Carlo technique is really well suited for this task since no additional work is required, except putting flags on the correct code lines. Finally, since Monte Carlo is based on random events, statistics must be done to have accurate results. Statistical data are also available and will be used in this thesis.
This code runs in 3 dimensions and uses Cartesian coordinates. 105 photons coming at normal incidence will be tested to be in a statistical regime and are treated independently. The independence property enables to parallelize the code to run faster. The photons have a spectral density similar to the AM1.5G which can be found on this website  3. LSC-VMCA flowchart is shown in fig. 2.7 and the main blocks will be described.
Table of contents :
Abbreviations and acronyms list
1 Photovoltaics, one answer to energy challenge
1.1 Energy issues
1.1.1 Current situation
1.1.2 PV in the electricity mix
1.2 Photovoltaics: from principles to technology
1.2.1 PV basic principles
184.108.40.206 Photo-carriers generation
220.127.116.11 Charge carrier separation
18.104.22.168 Charge collection
1.2.2 PV technologies
1.3 State of the Art on photovoltaics light concentration
1.3.1 CPV: interests & features
1.3.2 Different concentration levels for different targets
22.214.171.124 High Concentration PhotoVoltaics HCPV
126.96.36.199 Medium Concentration PhotoVoltaics MCPV
188.8.131.52 Low Concentration PhotoVoltaics LCPV
184.108.40.206 Luminescent Solar Concentrator place in CPV landscape & consistency with low size PV cells
1.4 Big picture
I Luminescent solar concentrator based concentration
2 Basics of Luminescent Solar Concentrators
2.1 Luminescent Solar Concentrator: State of the art
2.1.2 Description via geometrical optics
2.1.3 Loss channels
2.1.4 Lowering front loss: Photonic Band Stop & dye anisotropic emission
2.1.5 Thermodynamics consideration
2.1.6 Record efficiencies
2.2 Modeling: Algorithm & Flowchart
2.2.1 Flowchart: global-description
2.2.2 Flowchart: quantitative description
2.3 Big picture
3 LSC Physics
3.1 Thermodynamics limit: ideal dye case
3.1.2 All-ideal system
3.1.3 Semi-ideal case: sensitivity analysis
220.127.116.11 Matrix absorption
18.104.22.168 Back loss
22.214.171.124 RPBS loss
126.96.36.199 PLQY loss
3.1.4 Non ideality comparison
3.1.5 Interdependencies of non ideality
3.2 Mismatch effect between dye and solar cell bandgap
3.3 Toward realistic LSC
3.3.1 Semi-ideal dye
3.3.2 Real dye
3.3.3 Realistic system
3.4 Semi-analytical formulation with MBEOs
3.4.1 Expressions derivation
3.4.3 Reducing MBEOs only to d
3.5 Big picture & LSC fundamental contradiction
4 CIGS-based micro solar cells coupled to Luminescent Solar Concentrator
4.1 Microcell array and LSC, a winning duo?
4.2 Prototype fabrication & LSC-VMCA validation – Generation 1
188.8.131.52 Microcell array fabrication
184.108.40.206 LSC fabrication
4.2.2 Coupling of generation 1 prototype
220.127.116.11 Base-Line case
18.104.22.168 Lowering side loss
22.214.171.124 Back reflection schemes
4.3 Air-gap LSC – Generation 2
126.96.36.199 Geometrical effects
4.3.2 Pillar-induced loss
4.4 Opal filter on LSC – Generation 2
4.4.2 Optical characterization
4.4.3 Conclusion & Perspectives
4.5 Big picture
Part I Conclusions
II Nanophotonic-based concentration
5 Nano-antenna for PV applications
5.1 Nano-antenna and spectral convertor coupling
5.2 Strategies to improve light absorption in a PV absorber
5.2.1 Back surface reflector and classical limit
5.2.2 Nanophotonic light trapping grating and photonic crystal
5.2.3 Nano-wire-based light trapping
5.2.4 Plasmonics-based light trapping
5.3 Optical nano-antenna based on Metal-Insulator-Metal geometry
5.3.2 Resonance mechanisms in a MIM structure
188.8.131.52 Fabry-Perot resonance
184.108.40.206 Plasmonic resonance
220.127.116.11.a Surface plasmon polaritons (SPP)
18.104.22.168.b Plasmonic resonance coupling
22.214.171.124 Guided-mode resonance
5.4 Rigorous Maxwell Constitutive Approximation
5.5 Big picture
6 Mono-Resonant Concentration Device (MRCD)
6.1 Optical design
6.1.2 Modeling: a multi-scale problem
126.96.36.199 Comprehensive approach
188.8.131.52 Statistical approach
184.108.40.206.a Global optimization & a particular case
220.127.116.11.b Antenna analysis
18.104.22.168.c Size and material robustness
22.214.171.124 Photon recycling
6.1.3 MRCD improvements
126.96.36.199 Reducing back loss: contacts removal
188.8.131.52 Reducing front (ext.) loss: High matrix refractive index (antireflection coating)
184.108.40.206 Combining high refractive index, no contact layers and ARC
6.2.1 pin junction reported on pyrex
6.2.2 Top metal patches preparation
6.2.3 Top metal patches & etching
6.2.4 Electrical connection
6.3 Electrical performances
6.3.1 Nano-diode fabrication
6.3.2 Current-Voltage measurements by AFM-CP
6.3.3 Passivation with polyphosphazen
220.127.116.11 Polyphosphazen formation
18.104.22.168 Polyphosphazen characterization
22.214.171.124.a In-situ characterization
126.96.36.199.b Others characterizations (XPS,EDX)
188.8.131.52 Polyphosphazen influence on surface recombination
184.108.40.206.a Luminescence measurements
6.4 Big picture
Part II Conclusions
General conclusions & Perspectives
Résumé en Français