Measured surface loss from luminescent solar concentrator waveguides
Michael G. Debije1*, Paul P. C. Verbunt1, Brenda C. Rowan2, Bryce S. Richards2 and Theo L. Hoeks1
1. Polymer Technology, Department Chemical Engineering and Chemistry, Eindhoven University of
Technology, 5600 MB Eindhoven, NL
2. School of Engineering and Physical Sciences, Heriot-Watt University, Edinburgh, Eh14 4AS, United
* author for correspondence: firstname.lastname@example.org
The surface and edge emissions from dye-filled and dye-topped polycarbonate and PMMA
luminescent solar concentrators (LSCs) were measured. We demonstrate that about 40-50% of the
absorbed light energy (and 50-70% of the photons) is lost through the top and bottom surfaces of the
filled waveguide. In most cases the escape cone losses are greater at the top than the bottom surface.
OCIS codes: 350.6050, 350.5500, 310.2785, 260.2510, 230.7390, 160.5470
Luminescent solar concentrators (LSC) are attractive as possible replacements for large-area
silicon-based solar panels in the built environment, owing to their promise of lower production costs
Applied Optics (2008) 47(36), 6763–-6768
and suitability for building-integrated photovoltaics. Research on the LSC slowed in the mid-1980’s
due to two main factors. The first was a concern over the stability of the dyes used. At that time,
lifetimes of fluorescent dye materials were measured in days to weeks under solar illumination . In
the last quarter century there has been a tremendous improvement in the performance and lifetimes of
fluorescent dye molecules and these materials now have lifetimes of years to decades . In addition,
materials such as quantum dots  and rare earth materials  offer alternatives which may provide
improved stability. As we believe this concern over dye stability and efficiency has been mostly
alleviated, we now focus on the second major factor: losses from the waveguide itself.
In the past, most research on losses in the LSC has focused on minimizing the internal losses
due to dye re-absorptions (see for example ). Aside from emphasis for producing optically smooth
surfaces  and the use of higher or variable refractive index waveguides , the losses through the
surfaces (top or top and bottom) of the waveguides have been accepted as unavoidable. Most
theoretical simulations of the LSC describe the fluorescent dye system as an isotropic absorber and
emitter, and it becomes a simple calculation to determines that roughly 25% of light energy will be lost
via the ‘escape cone’ of the assumed n=1.5 waveguide) [8-10]. However, the dyes in general are
neither isotropic absorbers nor emitters, but are functional dipoles with more directed absorption and
emission (see for example [11-15]). Also, the majority of the input light is typically incident upon the
waveguide from one surface alone (i.e. the top surface). In such a situation, the dye molecules lying
with their dipoles directed parallel to the incoming electric field of the sunlight will tend to absorb
more light than those aligned perpendicular to the incoming light E-field . In addition, the dye
system will emit in a similar direction, resulting in an emission with an oblong profile rather than
spherical, with a significant fraction of emitted light directed towards the surface [14,15]. In addition
to this initial loss, there will be further surface losses from light re-absorbed and re-emitted by
subsequent dye molecules, a result of the limited Stokes shift of the dyes, a feature rarely considered in
theoretical calculations (for example, ). In our laboratory, we have been researching various
methods to reduce or eliminate these surface losses, including the application of organic semi-
transparent reflectors  and the unidirectional alignment of the emitting dye molecules .
The purpose of this paper is the direct measurement of the amount of light emitted from the
surface of two waveguide configurations: one with dye material dispersed throughout the bulk of the
sample, and the other with thin dye-doped layers on the surface of clear waveguides. We offer an
explanation for the differences in top/bottom and total emissions noted at high and low doping
A number of filled waveguides were produced by Sabic Innovative Plastics (Bergen op Zoom,
NL) by injection molding of polycarbonate (PC) (n=1.586 at 587 nm) with various concentrations of
the fluorescent dye Lumogen® F Red305 (hereafter referred to as Red305 - BASF) or
polymethylmethacrylate (PMMA) (n=1.49) with one concentration of Red305 into plates 50 x 50 x 3
mm. Surface-topped waveguides were produced by spin casting solutions of Red305 in PC from THF
or Red305 in a pentaacrylate (Polysciences, Inc.)/MMA (Aldrich) 80:20 blend containing 1%
photoinitiator (Irgacure 184, Ciba) on top of either clear PC (Sabic Innovative Plastics) or PMMA
(Plano Plastics) plates at 1000 rpm for 30 seconds. The pentaacrylate systems were subsequently
exposed to ultraviolet light in a N2 atmosphere to crosslink the system. The thicknesses of the dye
layers were measured by a Zoomsurf 3D interferometer (Fogale), and were about 3 µm for the PC
layers, and 15-30 µm for the pentaacrylate/MMA layers.
The absorbance for all samples was determined using a Shimadzu UV-3102
spectrophotometer: the reported optical density (OD) values in this paper refer to the absorbance at the
peak of the main absorption band.
Bottom surface emissions (that is, emission from the surface opposite the incident light source)
were determined by placing the 5 x 5 cm waveguides against the entry port of an integrating sphere
equipped with a SLMS LED 1050 light detection array (Labsphere), with a blank waveguide used as
the reference – see Figure 1 for a depiction of the experimental setup. The samples were exposed to the
light of a 300 W solar simulator with filters to approximate the 1.5AM (global) solar spectrum (Lot-
Oriel). To reduce the background spectra from the source light, a stack of cholesteric filters centered at
670, 710, and 750 nm were placed between the source and the sample to filter out these longer
wavelengths. The illumination area was limited to an approximately 2 cm diameter spot in the
Top surface loss measurements (that is, from the surface being illuminated by the light source)
were derived from emission data taken using an Autronic DMS 703 (Melchers GmbH) together with a
CCD-Spect-2 array detector (CCD-Camera). The LSC samples were placed in a custom-made sample
holder and exposed to a roughly uniform light source located at a distance of about 11 cm. Output
spectra were recorded for surface emissions from 30º to 70º from the normal of the waveguide surface
for both sides of the waveguide. See Figure 2 for a depiction of the experimental setup. The integrated
outputs were determined for both ‘top’ and ‘bottom’ sides of the waveguide, and a ratio of top/bottom
emission was obtained. This ratio was compared to the absolute measurements taken for the ‘bottom’
surface from the integrating sphere, and from this the ‘top’ emission was calculated.
Edge emissions from the waveguides were determined by placing the samples on a horizontal
stage with only one edge of the waveguide entering the integrating sphere. Illumination was over the
whole surface by the solar simulator, and the quantity of energy absorbed was determined from the
measured absorption spectrum of the waveguide and the emission spectra of the light source.
A representative absorption and edge emission spectra for a dye filled waveguide is shown
below in Figure 3. Emission from the edge of all the samples were recorded, and the total power
output of all the samples as a function of absorbance are shown in Figure 4.
Filled waveguide surface emissions were determined by placing the 5 x 5 cm waveguides
against the opening of the integrating sphere, and illuminating with a solar simulator. Using a blank
waveguide, we could determine the amount of light transmitted and reflected by the waveguide itself.
Absorption was essentially zero through this narrow width of the waveguide, and the reflection from
PMMA was measured to be around 8%, and from PC about 11.5% in the visible range.
Samples containing Red305 dye were then placed in the measurement position, and the results
recorded. Representative spectra of the blank and a filled waveguide may be seen in Figure 5a. By
subtracting the results of the blank measurement from the results of a filled waveguide, we obtained the
spectra seen in Figure 5b. The fraction of light emitted from the surface was determined by integrating
the emission spectra (the light emitted into the sphere from the surface of the waveguide, or the positive
region of the curves in Figure 5b, roughly 600nm out to our cutoff of 750 nm, depicted as region II)
and dividing by the light absorbed by the dye (the negative region of the spectra of Figure 5b, roughly
400 - 600 nm and depicted as region I). The integrations are seen in Figure 6.
Measurements of the surface-topped samples were done in a similar manner, with the dye layer
situated on the top side of the waveguide, facing the light source. For all but the lowest absorbance, the
emissions measured with the dye layer on top and with the dye layer at the back were within 5%.
Measurements using the Autronic system (described previously in the Experimental section and
Figure 2) were used to establish the ratio of top to bottom emission from the waveguide surfaces. Once
the ratio was determined, the absolute emission from the top surface was calculated given this ratio and
the measured value of bottom emission. The Table gives numerical loss ratios for the filled
polycarbonate samples; the ratios of the other samples are depicted in Figure 7.
Several samples were measured for output from both faces by repositioning of the waveguide
with respect to the incoming light in the Autronic setup, and the differences were determined to be less
We may derive some general conclusions regarding the edge emission from waveguides made
from PMMA and PC in bulk and as thin films on clear waveguides by considering Figure 4. The edge
emissions are essentially identical for bulk and thin-film waveguides of PC and PMMA up to OD ~
0.75 (about 85% absorption at the peak). At higher OD values, the filled waveguides demonstrate
higher efficiencies. This can be expected owing to the limited solubility of the Red305 dye in the thin
(3 µm for polycarbonate, ~20 µm for the pentaacrylate/MMA blend) dye layers. At the high
concentrations necessary to achieve >85% absorption, the dye materials start to self-associate, resulting
in the formation of crystallites and the quenching of fluorescence.
In looking at the Table and Figure 6, it becomes apparent that there is a considerable loss of
light from the surfaces of the filled waveguides, and this loss is constant at around 40% except in the
lightly doped waveguide. This translates to a loss of 52% of the photons. At low optical densities, the
energy losses are greater than 50%, which translates into 71% photon loss.
The differences in light emission between the top and bottom surfaces illustrated in Figure 7
may be understood by considering that light absorption through the sample will show a significant
gradient, with a greater fraction of light absorbed near the top surface in a heavily dye-doped
waveguide. This will result in an emission gradient, with more of the light being emitted near the top
surface; thus, more light will escape from the top, as that light emitted towards the bottom has a
significantly increased chance of being re-absorbed before escape. In the lightly doped waveguide,
more light will penetrate deeper into the waveguide before absorption, leading to a relative increase in
light escape fraction out the bottom surface, as light traveling towards the top surface will have an
increased chance of re-absorption.
One may calculate the emission profile of an isotropic collection of dye illuminated from an
arbitrary direction [using 14, 15]; see Figure 8. Using this profile, we may determine the fraction of the
emission light that will be directed into the waveguiding mode for any given input light angle, and the
fraction directed so as to escape the surface of the waveguide, unless otherwise reabsorbed and
redirected by subsequent emission. The results of this calculation are seen in Figure 9. This latter
Figure may be used to simulate the effect of solar variation during the day as the sun crosses the sky.
In the extreme cases of incoming light directed at 75° with respect to the normal of the waveguide
surface, there is a calculated reduction of surface-emitted light of 25%, but in absolute terms is still
50% of the emitted photons.
For the case of light incident perpendicular to the waveguide surface, simulated results suggest
65% of the photons initially emitted by the dyes will be directed towards the top and bottom in such a
way as to permit their escape from the waveguide if not reabsorbed. Our measured values tally
between 52 and 71%, depending on the optical density. This suggests there is a considerable amount of
reabsorption of surface-directed light in the high optical density waveguides, and that there is very little
in the low optical density waveguides.
Surface losses are somewhat reduced in the surface-coated waveguide systems (Figure 6).
However, the internal losses of the thin waveguide systems are much higher, and the edge output at
higher absorbancies is also reduced (Figure 4). This is at least partly due to the reduced dye-dye
spacing necessary in these thin layers (3 to 30 µm as opposed to 3 mm for the filled waveguides),
which result in enhanced intermolecular interactions and consequent increased internal light losses.
A series of simulations using the RAYLENE software  give some insight into the
measurements. Calculations were made for the number of interactions each photon emitted by the dye
undergoes in the filled PC samples before escaping the waveguide or being lost. For an OD of 4
(99.99% absorption at the peak) each emitted ray undergoes an average 1.8 dye encounters. At an OD
of 1, this becomes about 1.5 encounters, while at 0.08 it is about 0.4 encounters (see Figure 10). Thus,
it is clear to see at lower ODs, there is a factor four less encounters per ray in escaping the waveguide,
which corresponds to an increased surface losses in the low absorbance region of Figure 6 compared to
the higher absorbance region.
The significant energy losses through the surface of the waveguide have been accepted as a
matter of course in the past, with more than 20% leaving the top surface and 40% overall (equivalent to
more than 50% of the photons). Losses through the bottom may be reduced by the application of
mirrors or scatterers, for example, but a fraction of this light will itself be redirected out the top surface
again. Our lab has set up several research programs designed to reduce or eliminate these surface
losses, and to promote enhancement of edge emissions and the overall efficiency of the LSC system.
Among the features being considered for reduction of these surface losses are aligned dye arrays which
will restrict the light emission of the dyes into more defined directions rather than the more random
emission found in isotropic dye systems. A second feature is the selectively-reflecting chiral nematic
(cholesteric) liquid crystal layer (similar in function to the inorganic ‘hot mirrors’ described in ).
The purpose of this latter system is to allow sunlight to reach the fluorescent dyes, but to reflect the
dye-emitted light back into the waveguide, preventing this emitted light from leaving the surface. By
reducing surface and internal losses through selection of higher-quality waveguide host materials and
optimizing the waveguide dimensions, the output of the luminescent solar concentrator could be
significantly increased, an important step in making the LSC a viable competitor to the silicon solar
The surface energy losses from luminescent solar concentrator waveguides is on the order of
40% to more than 50%, translating into a photon loss of 50% to 70%, a considerable amount of light.
The losses from the top surface are greater than the losses from the bottom for more heavily-doped
waveguides. These findings accentuate the need to develop ways of reducing these losses in LSC
devices in order to make them more viable competitors of the standard silicon photovoltaic panel.
The authors would like to acknowledge the support of Altran Technologies Netherlands, and in
particular thank Roel Moonen of Altran for his assistance in the measurements, and Tom Akkermans
from the TU Eindhoven for help in sample preparation. MD would like to acknowledge the support of
STW VIDI grant 07940. BSR and BCR acknowledge the support of EPSRC grant EP/F02763X/1.
 A. M. Hermann, “Luminescent solar concentrators - a review,” Sol. Energy 29, 323-329 (1982).
 I. Baumberg, O. Berezin, A. Drabkin, B. Gorelik, L. Kogan, M. Voskobojnik and Zaidman,
“Effect of polymer matrix on photo-stability of photo-luminescent
dyes in multi-layer polymeric structures,” Polym. Degrad. Stab. 73, 403-410 (2001).
 K. Barnham, J. L. Marques and J. Hassard, “Quantum-dot concentrator and thermodynamic
model for the global redshift,” Appl. Phys. Lett. 76, 1197-1199 (2000).
 R.Reisfeld, “Future technological applications of rare-earth-doped materials,” J. Less Comm.
Mater. 93, 243-251 (1983).
 Bailey, S. T., G. E.Lokey, M. S. Hanes, J. D. M. Shearer, J. B. McLafferty, G. T. Beaumont, T.
T. Baseler, J. M. Layhue, D. R. Broussard, Y.-Z. Zhang
and B. P. Wittmershaus, “Optimized excitation energy transfer in a three-dye
luminescent solar concentrator,” Solar Energ. Mater. Sol. Cell 91, 67-75 (2007).
 A. A. Earp, G. B. Smith, P. D. Swift and J. Franklin, “Maximizing the light output of a
luminescent solar collector,” Sol. Energy 76, 655-667 (2004).
 A. Cutolo, L. Carlomusto, F. Reale and I. Rendina, “Tapered and inhomogeneous dielectric
light concentrators,” Opt. Laser Technol. 21. 193-197 (1989).
 W. A. Shurcliff and R. C. Jones, “The trapping of fluorescent light produced within objects of
high geometrical symmetry,” J. Opt. Soc. Am. 39, 912-196 (1949).
 J. S. Batchelder, A. H. Zewail and T. Cole, “Luminescent solar concentrators. 1: Theory of
operation and techniques for performance evaluation,” Appl. Optics 18, 3090-3110 (1979).
 M. Carrascosa, S. Unamuno and F. Aguilo-Lopez, “Monte Carlo simulation of the performance
of PMMA luminescent solar collectors,” Appl. Optics 22, 3236-3241 (1983).
 E. W. Thulstrup, J. Michl and J. H. Eggers, “Polarization spectra in stretched polymer sheets. II.
Separation,” J. Phys. Chem. 74, 3868-3878 (1970).
 E. W. Thulstrup and J. Michl, “A critical comparison of methods for analysis of linear
dichroism of solutes in stretched polymers,” J. Phys. Chem. 84, 82-93 (1980).
 L. V. Natarajan, F. M. Stein and R. E. Blankenship, “Linear dichroism and fluorescence
polarization of diphenyl polyenes in stretched polyethylene films,” Chem. Phys. Lett. 95, 525-
 C. Sánchez, B. Villacampa, R. Cases, R. Alcalá, C. Martínez, L. Oriol, and M. Piñol, “Polarized
photoluminescence and order parameters of ‘in situ’ photopolymerized liquid crystal films,” J.
Appl. Phys. 87, 274-279 (2000).
 M. van Gurp and Y. Levine, “Determination of transition moment directions in molecules of
low symmetry using polarized fluorescence. I. Theory”, J. Chem. Phys. 90, 4095-4102 (1989).
 F. L. Arbeloa, V. Martínez Martínez, T. Arbeloa and I. López Arbeloa, “Photoresponse and
anisotropy of rhodamine dye intercalated in ordered clay layered films,” J. Photochem.
Photobiol. C 8, 85-108 (2007).
 J. Roncali and F. Garnier, “Photon-transport properties of luminescent solar concentrators:
analysis and optimization,” Appl. Optics 23, 2809-2817 (1984).
 M. G. Debije, R. H. L. van der Blom, D. J. Broer and C. W. M. Bastiaansen, “Using selectively-
reflecting organic mirrors to improve light output from a luminescent solar concentrator,”
World Renewable Energy Conference IX, Florence, Italy (2006).
 M. G. Debije, D. J. Broer and C. W. M. Bastiaansen, “Effect of dye alignment on the output of
a luminescent solar concentrator,” 22nd European Photovoltaic Solar Energy Conference,
Milan, Italy, 87-89 (2007).
 B. S. Richards and K. R. McIntosh, "Overcoming the poor short wavelength spectral response
of CdS/CdTe photovoltaic modules via the luminescence down-shifting: ray tracing
simulations,” Prog. Photovoltaics Res. Appl., 15, 27-34 (2006).
 B.S. Richards, A. Shalav and R.P. Corkish, A low escape-cone loss luminescent concentrator,
th EC Photovoltaic Solar Energy Conference, Paris, France, 113-116 (2004).
Rotates 30 → 70 º
Rotates 0 → 180º
0 0.51 1.52 2.53 3.54 4.55
Edge Output (mW)
15 Download full-text
350400 450500550600650700 750800
350 400450500550 600 650 700750800