Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 8

40

5. Efficiency enhancer patent

This chapter is a seminal aspect of improving the performance of concentrator solar cells

and, at this stage, it represents a conceptual advance that now awaits experimental

realization. The reprint of the affiliated journal article yet to be published in Optics Letters

is attached as Appendix II.

5.1 Shading and series resistance trade off

The metal fingers architecture reflects the tradeoff between two opposite trends. Higher

metallization reduces the series resistance, but at the same time shade greater part of the

cell area wasting the light impinges on them. This trade off was quantified by Spectrolab

(who produced the cell tested in this thesis), the actual gridline spacing corresponds to the

minimum of the “total” curve below.

Fig 5-1: Simulation results [1] for contributions to efficiency loss in the design of a recent
generation of similar ultra-efficient multi-junction concentrator cells tailored so efficiency peaks at
~500 suns.

This tradeoff can be overcome by introducing an optical device that redirects all the light to

the cell’s active area within the busbars. Such a device was explored by our group in

collaboration with Prof. Naftali Eisenberg from Jerusalem College of Technology and is

registered under patent pending application [38].


41

5.2 Optical design strategy

The solar cell resides at the exit of a macro-concentrator. The all-dielectric micro-

concentrator (ADMC) would be sandwiched and optically coupled between them. The

notion for flux redistribution that completely obviates shading losses is illustrated

schematically in Fig 5-2. Nonimaging θ1/θ2 [20, 33] ADMCs predicated on total-internal-

reflection (TIR) represent solutions that, for given maximum input and output angles, θ1

and θ2, respectively, they: (a) provide the possibility of loss-less optics, (b) represent the

most compact device possible for any prescribed metallization coverage for a flat ADMC

entry, and (c) can yield V-troughs in many cases of practical interest (as developed below),

particularly amenable to accurate fabrication.

Figure 5-2: A section of the cell's metal grid (showing current flow in the emitter and grid) [21], and
the introduction of a nonimaging ADMC to redistribute impinging light and totally eliminate front
contact shadowing losses. In this illustration, the grid spacing remains unchanged. In the general
analysis below, the optic is also tailored to smaller grid spacings, i.e., to higher coverage fractions,
while still totally eliminating grid shading and introducing essentially zero optical loss.

The height of the metal fingers is commonly only a few μm: not a limiting factor in the

design of the micro-concentrator troughs analyzed below. The width of the metal fingers,

typically ~10 μm, can be increased without changing their height, so the upper range

metallization coverage considered below is equally tractable. Also, there is negligible

absorption in the ADMC due to short optical path length.


42

The angular extent of cell irradiation at the exit of a PV macro-concentrator, typically

~15°-30° [22-27], represents the input θ1 values for ADMC designs. Such seemingly low θ1

values are still sufficient for achieving the net flux levels of hundreds (even up to 2000)

suns at which the efficiency of the most advanced concentrator cells peaks [18, 21, 22, 28-32].

Our designs relate to the most common metal grid pattern of parallel strips. Each ADMC is

then trough-like (2D) with flat vertical faces in the orthogonal plane (Fig 5-2). Furthermore,

some high-flux macro-concentrators are now all-dielectric themselves [27, 33] (although

externally mirrored and not depending on TIR). Cell linear dimensions of order 1 mm

render such devices practical with relatively low mass per unit aperture area. (ADMCs that

completely eliminate metallization shadowing can also be tailored to other metal grid

patterns. Irregular and crossed grids would mandate 3D concentrators and more complex

fabrication procedures, but in principle are achievable.)

Fresnel reflective losses can be essentially zero because the anti-reflective coating on the

cell creates an effective n (index of refraction) close to that of the dielectric of the macro-

concentrator (and the ADMC would be fabricated from a material of comparable n).

Nevertheless, the ADMC design and its efficiency enhancement apply equally well to

macro-concentrators such as Fresnel lenses where light enters the cell from air. Since the

gridline spacing in most current high-efficiency concentrator cells is roughly 0.1 mm

independent of cell linear dimension [13, 18, 21, 22, 28-32, 34, 35], the results below

should be universally applicable.


43

5.3 Micro-concentrator contours

The ADMC is a θ1/θ2 nonimaging transformer [20,33] (Fig 5-3), which accepts all

rays incident up to ±θ1, concentrates them onto the absorber over angular range ±θ2. In its

complete untruncated form, it attains the corresponding thermodynamic limit to 2D flux

concentration

Cmax = sin(θ2)/sin(θ1) (5.1)

which is also the ratio of entry to exit width at no ray rejection. Its contour comprises (1) an

upper parabolic arc with its axis tilted at θ1 relative to the optic axis and its focus at the

absorber edge, and (2) a lower straight segment (Fig 5-3).

Figure 5-3: θ1/θ2 ADMC. Upper contour EDB is the arc of a parabola with focus at A' and axis
rotated θ1 relative to the optic axis. Lower section BA (red) is a straight line tilted at (θ 2 - θ1)/2.
Truncation to point D at angle θT reduces device depth and concentration. Sufficient truncation
yields a pure V-trough. The dielectric region is darkened (blue), and the metal fingers (yellow) in
contact with the solar cell surface fit comfortably between adjacent troughs. The actual truncated
ADMC here – one of the practical instances examined below - has entry D'D (~100 μm), θ1 = 30°,
θ2 = 55° and θT = 46°, with C = 1.50 and AR = 0.80.
44

The derivation of the contour begins at point A, (see figure 5-3). A ray which reaches there

at angle θ1 or lower with respect to the optical axis should be reflected at an angle no higher

than θ2 with respect to the optical axis. Thus the angle between the contour and the optical

axis at point A can be no greater than (θ2-θ1)/2. For the contour around A to be the lowest

achievable at any given infinitesimal distance from A, the highest allowed angle between

the contour and the optical axis should be adopted. Therefore the beginning of the lowest

possible contour has been derived. It is an infinitesimal strait segment which begins at point

A and tilted (θ2-θ1)/2 degrees away from the optical axis.

It is not a coincidence that this piece of contour would have also emerged if instead of

trying to achieve the lowest contour possible, the contour would have been tailored to direct

the imminent edge ray to corresponding exiting edge ray. Actually for every infinitesimal

contour segment, the very same arguments that lead to highest achievable angle of the

segment with the optic axis are equivalent to matching edge rays. Therefore up to a certain

point (point D in figure 5-3) the design coincides with nonimaging θ1/θ2 concentrator. The

certain point is where the contour line reaches the midperpendicular to the metal finger.

There it meets the matching contour from the following adjacent ADMC and terminates.

Thus the contour is a truncated θ1/θ2 concentrator.

In some instances, the truncation cuts into to the linear section, and the ADMC becomes a

pure V-trough, putatively the simplest contour to fabricate. Indeed, every V-trough is

rigorously a truncated version of some θ1/θ2 device. Hence V-troughs are subsumed in the

general analysis.

The expressions for ADMC aspect ratio AR (depth/entry) and C as functions of θ1, θ2

and θT follow from straightforward geometry:


45

 2( sinθ( 1 ) + s inθ( 2 ) ) s inθ( T )


 -1 θ1 ≤ θT ≤ θ 2
1 − c o s(θ 1 + θ T )



 (5 .2 )
C=  θ 2 − θ1 
 ta nθ( T ) + ta n 
  2  θ 2 ≤ θ T ≤ 9 0
 θ 2 − θ1 
 ta nθ( T ) − ta n 
  2 


 ( sin(θ 1 ) + sin(θ 2 ) ) cos(θ T )


 2( sin(θ ) + sin(θ ) ) sin(θ ) + cos(θ + θ ) − 1 θ1 ≤ θ T ≤ θ 2
 1 2 T 1 T

 (5.3)
AR = 
 1
θ 2 ≤ θ T ≤ 90
 θ −θ 
 tan 2 1  + tan(θ T )
  2 

Satisfying TIR for all incident rays additionally requires [27,28]

θ1 + θ2 < 180° - 2θc (5.4)

where θc = Sin-1(1/n) is the dielectric's critical angle. (TIR is respected in both the plane of

concentration and the orthogonal plane.) We bound θ2 by requiring negligible (e.g., < 1%)

Fresnel reflective losses, which is ~55° based on measurements for the type of multi-

junction concentrator cells with anti-reflective coatings currently in use [13]. Hence TIR is

satisfied for the entire parameter space considered here (θ1 up to ~30°, θ2 up to ~55° and n

= 1.5).

Fabrication tolerances militate against a perfect match of adjoining troughs, so real designs

must anticipate trough overlap that reduces concentration (Fig 5-4). Convolving reasonable
46

production tolerances and alignments with the actual width of the metal fingers would

increase the equivalent finger width for which designs are generated by approximately

20%, while preserving the complete elimination of metallization shadowing.

Figure 5-4: Illustration of designing for overlapping concentrator troughs toward accommodating
realistic fabrication tolerances.

The lowest concentration C of interest would be ~1.1. Concentration is increased when the

metal fingers are widened (the gridline spacing is lessened), to a coverage ratio of 1-(1/C)

(with C bounded by Eq 5.1).

With θ1 ordained by the macro-concentrator, there are two design degrees of freedom,

selected as C and θ2 (subject to the bounds noted above). The most practical designs are

deemed those with the lowest AR, preferably V-troughs. Representative results are plotted

in Fig 5-5 (and include Fig 5-3), a range of which includes ultra-compact V-troughs.

Figure 5-5: Micro-concentrator aspect ratio (AR) as a function of C at prescribed (macro-


concentrator-input) θ1, at the maximum θ2 = 55°. As C is decreased from its maximum value (Eq
5.1) by truncation, a point is reached (♦) below which the ADMC is a pure V-trough.
47

The micro-concentrator creates the added benefit of reduced gridline spacing, and hence

reduced Rs. The simulation results of Fig 5-1 indicate an associated maximum efficiency

augmentation of 7% (relative), which should improve with macro-concentration above 500

suns.

You might also like