Using the SMS method to design concentrators has produced new families during the last decade . Among them, several families (RR , RR-RRIF [30,36], XR , XXf ], SMTS , DSMTS , RXI , RXIf , RXI-RX , TIR-R ) have been suggested for PV applications.
The nomenclature used for referring to the different designs (excluding SMTS and DSMTS) is as follows: each concentrator is named with a succession of letters indicating the order and type of incidence of the optical surfaces that the sun-ray encounters on its way to the cell. The following symbols are used: R=refraction, X=reflection, I=total internal reflection. The sub-index F is added to X and I of these mirrors coincidence with the flow line).
In the case of high concentration systems (i. e. RRIF, XR, RXI and TIR – R), these devices have been designed with rotational symmetry and conceived for small cells as miniconcentrators (concentrator entry aperture diameter = 3085 mm). In order to make modules, the concentrators are truncated as squares for tessellation. In these designs, the acceptance angle remains unchanged after truncation, and then the acceptance angle – concentration product is reduced by a factor 2/n = 64%. This reduction would be smaller if hexagonal concentrators were used but square tessellation is usually considered more aesthetic because it is free from edge effects.
The interest in miniconcentrators comes from the fact that small cell technology is very close to that of the LEDs, which is very well developed and highly automated. As an example, highly efficient 1 mm2 GaAs cells for 1000- sun operation have been demonstrated . The power to be dissipated is around 0.7 W. However, the Luxeon LED of Lumileds has a high-flux blue chip, also 1 mm2, and it was designed to dissipate 1.3 W.
We will review the aforementioned high-concentration SMS devices, assuming that the aperture is square and the cell’s active area is circular (unless otherwise specified). As will be seen later, apart from the TIR-R, these designs have aimed to provide excellent acceptance angle-concentration products but the uniformity of the irradiance distribution of the cell is not good. This is because their design focused on it as a bundle-coupling problem and not as a prescribed irradiance problem (see section 13.1.7). The TIR-R is the first SMS designed which has approached both problems.
The RR concentrator for PVs is formed by one primary lens with a flat entry surface and a continuous or Fresnel exit surface, and a refractive secondary that encapsulates the solar cell (see figure 13.12(a)). In contrast to the conventional Fresnel lens and non-imaging secondary concentrators, the primary and secondary are designed simultaneously, leading to a better concentration- acceptance angle product without compromising the compactness. It achieves в = 45° with good optical efficiency (around 85%), and for a90 = 1°, it can get about Cg = 1500 x for a square aperture and circular cell, although the irradiance homogeneity is very poor. A modification of this device has been proposed in the framework of the Hisicon EU project for front-contacted silicon solar cells , designed and manufactured by LETI, which is the project leader. This silicon cell concept performs well for much higher concentration levels than the back- contacted cells (and, of course, better than the two-side contacted cells). The grid-lines in the Hisicon cells are aluminum prisms (which contact the p+ and n+ emitters, alternatively), acting as a linear cone concentrator that concentrates Cg = 1.52x in the cross-sectional dimension of the prisms (see figure 13.13). The modified RR device consists in a squared primary with a secondary element which has a refractive rotational symmetric top surface that is crossed with two linear flow-line TIR mirror (see figure 13.12(b)). Then, in the cross section normal to the prisms, the secondary coincides with an RR concentrator with 2D concentration of Cg = 12 x, while in the cross section parallel to the prisms it coincides with an RRIF concentrator. The flow-line mirrors IF have linear symmetry perpendicular to the grid-lines providing the RR with an additional concentration Cg = 2.08 x as the grid-lines in this dimension. Therefore, the cell is rectangular (1:2.08 aspect ratio), the grid-lines being parallel to the shorter rectangle side. The geometrical concentration (defined according to section 184.108.40.206 using the cell’s active area, i. e. that of the naked silicon) is (12 x 2.08) x (12 x 1.52) = 455x for the square aperture and rectangular cell, and achieves a design acceptance angle of a9o = 1.8°.
The XR concentrator is composed by a (non-parabolic) aspheric primary mirror and a refractive secondary that encapsulates the solar cell (see figure 13.14). As in the RR design, and in contrast to the classical parabola plus non-imaging secondary configuration, the simultaneous design of the two optical surfaces of the XR enables it to be very compact (H/D & 0.25, compared to H/D & 1-1.5 for classical systems), without sacrificing performance. The XR can even get isotropic illumination of the receiver (в = 90°), if necessary. For в = 70°, n = 1.41, a squared entry aperture and round cell, and a90 = 1°, the XR can get Cg = 3100x for a square aperture and circular cell. Silver mirrors (reflectivity &93%) would lead to an optical efficiency around 87-88%. The illumination homogeneity is still not good (although better than in the RR).
The XR has been suggested for space applications with a glass secondary, where the increase in glass thickness with respect to conventional space modules improves the cell shielding. Note that, as the glass diameter is small, the weight of the glass (which is critical) is not necessarily increased.
The RXI concentrator, shown in figure 13.15, was used for PV applications with single-junction GaAs cells in the framework of the Hercules EU project . This device can be also very compact (H/D & 0.3) and has the feature that the front surface is used twice: once as refractive one and the other as a (totally internally) reflective one. The prototypes in the Hercules project were manufactured by low-cost injection moulding, and silver evaporation. The geometrical characteristics are a circular entry aperture (40 mm diameter) and a 1 mm2 square cell active area. The measurements showed that a90 = 1.6° and ^0pt = 86% at Cg = 1197x. The measured photocurrent density of the
Figure 13.12. (a) The RR concentrator. (b) The secondary of the RR-RRIf concentrator (supporting, not optically active, elements are also shown).
concentrator was 20.6 mA cm-2, and the cell was being illuminated up to в = 70° (as indicated in 220.127.116.11, all these numbers exclude the 4.7% inactive area of the front mirror). The illumination homogeneity was poor, and the cell fill factor was only FF = 0.77 (with uniform flash illumination, FF = 0.85 was obtained).
The RXI needs mirror coatings. The combination of low cost, high reflectivity and durability of the mirror coating is difficult to achieve, and has still to be proved. In this respect, the new mirror technology marketed by 3M , which consists in a dielectric interferential multiplayer structure based on giant birefringent optics (GBO) is very promising. This product is sold as films and
Figure 13.13. Concentrating grid-lines in the front-contacted solar cells of the Hisicon EU project
is (surprisingly) low cost, ultra-highly reflective (above 98% in the GaAs-useful spectrum has been proved) and potentially durable (due to the absence of metals). The availability of new technology for concentrator development may encourage designers to focus on mirror-based solutions.
The RXI also has some additional practical problems. First, the front surface is not flat, so it could possibly accumulate dust, hence impeding easy alignment by gluing the concentrator units to flat glass (which would also provide good outdoor resistance and UV filtering). Second, the TIR is not protected, and its reflectivity (theoretically 100%) decreases significantly when absorbing particles (like dust) accumulates when exposed outdoors.
The TIR-R concentrator, shown in figure 13.16, solves the problems of the RXI and also does not need mirror coatings. This device is composed of a TIR
lens primary and a refractive secondary that encapsulates the solar cell. The TIR lens was invented by Fresnel  as a revolutionary optics for lighthouses at the beginning of the 19th century. More recently, during the last decades, TIR lenses have proven to be excellent devices for illumination applications , although the method used for their design was not suitable for getting good PV performance. The use of the SMS method on the TIR-R configuration has permitted to achieve high concentration-acceptance angle product to be achieved, good illumination uniformity and also sufficiently big aspheric facets (this is important for keeping a high efficiency, due to possible vertex rounding when manufactured).
The SMS TIR-R can achieve a geometrical concentration around Cg = 2300x, for a90 = 1°, squared entry aperture and circular cell active area. The cell is illuminated up to в = 70°. The compactness is H/D & 0.4-0.5. The optical efficiency can be in the nopt = 80-83% range, if the draft angle of the nearly-vertical facets is 2° and the radius of the vertices is around 20 ^m.
The irradiance uniformity of the TIR-R designs is significantly improved over that of previously presented designs. It is because the edge-ray assignation was used to solve the bundle-coupling problem and the prescribed-irradiance problem for normal incidence rays. This strategy could be also applied to any of the SMS designs presented. As an example, a design for Cg = 1600 x (squared entry aperture and circular cell active area) achieves a90 = 1.2°, and peak irradiance for the sun at normal incidence about 1900 suns (assuming 850 W m-2 for the irradiance at the entry aperture), and when the sun is off centre, the irradiance peak increases further, up to about 3000 suns.
The TIR-R concentrator is being developed for Cg = 1250 x on GaAs cells in the framework of the Inflatcon and Hamlet EU projects leaded by Isofoton, and for Cg = 300 x back-point-contacted silicon solar cells by Sunpower Corporation.
In the case of Isofoton’s development, the module will consist on
miniconcentrator units with a squared aperture about 10 cm2 (i. e. around 1000 concentrators per m2), illuminating III-V cells with an acceptance angle over 1.2°. The manufacturing process will be highly automated. The most surprising feature is that the module thickness is around 2 cm, similar to a framed flat module. Isofoton aims to get over 20% efficient modules with GaAs cells and over 25% efficient modules with tandem cells, and to reach the goal of total cost of 2.5 €/Wp in medium-large generation plants.