Research Collection

Journal Article

Luminescent solar concentrators based on melt-spun polymer

optical fibers

Author(s):
Jakubowski, Konrad; Huang, Chieh-Szu; Gooneie, Ali; Boesel, Luciano F.; Heuberger, Manfred; Hufenus,
Rudolf

Publication Date:
2020-04

Permanent Link:
https://doi.org/10.3929/ethz-b-000396538

Originally published in:

Materials & Design 189, http://doi.org/10.1016/j.matdes.2020.108518

Rights / License:
Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

This page was generated automatically upon download from the ETH Zurich Research Collection. For more
information please consult the Terms of use.

ETH Library
 Materials and Design 189 (2020) 108518

Contents lists available at ScienceDirect

Materials and Design

journal homepage: www.elsevier.com/locate/matdes

Luminescent solar concentrators based on melt-spun polymer

optical fibers
Konrad Jakubowski a,b, Chieh-Szu Huang c,d, Ali Gooneie a, Luciano F. Boesel c,
Manfred Heuberger a,b, Rudolf Hufenus a,⁎
a
Empa, Swiss Federal Laboratories for Materials Science and Technology, Laboratory of Advanced Fibers, Lerchenfeldstrasse 5, St Gallen 9014, Switzerland
b
Department of Materials, ETH Zurich, 8092 Zurich, Switzerland
c
Empa, Swiss Federal Laboratories for Materials Science and Technology, Laboratory of Biomimetic Membranes and Textiles, Lerchenfeldstrasse 5, St Gallen 9014, Switzerland
d
Laboratory of Inorganic Chemistry, ETH Zurich, 8092 Zurich, Switzerland

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Bi-component melt spinning was ex-

plored as a novel way to manufacture
fiber luminescent solar concentrators
(LSCs).
• Results revealed that solubility and dis-
persion of luminescent dye in polymers
play an important role in LSCs' perfor-
mance.
• Presented LSCs are a cost-effective solu-
tion to enhance the illumination-angle
dependence of solar cell power produc-
tion.
• Due to high industrialization potential
and overall performance, such fibers
may find their way into future
applications.

a r t i c l e i n f o a b s t r a c t

Article history: Luminescent solar concentrators (LSCs) collect incoming sunlight and direct it to a smaller-area photovoltaic cell.
Received 16 December 2019 In the presented work, form factor and illumination angle-dependent performance of LSCs consisting of bi-
Received in revised form 20 January 2020 component melt-spun fibers is demonstrated. Three thermoplastic polymers act as dispersing host material for
Accepted 21 January 2020
the luminescent dye Lumogen Red 305 (LR305). Molecular dynamics simulations provide numerical access to
Available online 22 January 2020
Hildebrand solubility parameters, which are an estimate for the mixing compatibility of dye with polymer matrix.
Keywords:
Actual emission intensity measurements from material samples are compared to Monte Carlo ray tracing simu-
Luminescent solar concentrators lations. Some samples show an increased absorption, which led to the hypothesis that there exist optically pas-
Polymer optical fibers sive dye aggregates if the dispersion is not optimal. The best-performing polymer/dye pair is identified and used
Energy harvesting to melt-spin fibers. Geometrically defined bundles of LSC fibers are studied in a scenario of white light illumina-
Melt spinning tion and variation of illumination-angle. This experiment simulates a theoretical daily course-of-sun illumination
in absence of atmospheric effects. We report optical conversion efficiencies of the prepared LSCs between 2% and
15%, depending on illumination angle and bundle geometry.
© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).

⁎ Corresponding author.
E-mail address: rudolf.hufenus@empa.ch (R. Hufenus).

https://doi.org/10.1016/j.matdes.2020.108518
0264-1275/© 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
2 K. Jakubowski et al. / Materials and Design 189 (2020) 108518
1. Introduction
Current photovoltaic (PV) cells exhibit efficiencies in the range 10%
to 40% [1]. To increase PV system power, large-area panels are required.
The costs scale essentially linearly with PV area [2], and the maximum
power achieved is often limited by the available system integration
area (e.g. roof top) [3,4]. The output power of planar solar cells depends
on the cosine of the illumination angle, manifesting itself as a midday
power peak, which has to be embraced by the power grid [5]. Lumines-
cent solar concentrators (LSCs) were first proposed in 1976 to enhance
the performance of PV solar cells. In practical use, a LSC must not only be
efficient in collecting photons at fixed illumination angle, but ideally
during the course of a whole day. Furthermore, the power gain by
LSCs should be cost-effective compared to the PV panel's price per
area. Polymer optical fibers (POFs) provide a suitable form factor to ad-
dress those requirements [6]. The favorable impact of the fiber geome-
try becomes apparent in terms of the LSC gain factor F(α, λ), which is
defined as the ratio of power generated by a solar cell with/without
LSC [7,8]:
P LSC
F ðα; λÞ ¼ ¼ ηopt ðα; λÞ  Gðα Þ  ηcell ðλÞ
P cell
where ηopt(α, λ) signifies the optical conversion efficiency [9,10]. We
note that for a fixed illumination, ηopt(λ) contains the optical attenua-
tion in the fiber. The factor ηcell(λ) is the wavelength-dependent effi-
ciency of the solar cell and G(α) is the angle-dependent geometrical
concentration factor, i.e. the ratio of the illumination angle dependent
collection surface (projected LSC shadow area), divided by the LSC out-
put area (out-coupling area facing photovoltaic cell).
Gðα Þ ¼ Acollection ðα Þ=Aoutput
For illumination at a normal angle, the G-factor of a fiber is maximal
and it is also proportional to the fiber aspect ratio, underlining the ad-
vantage of the length-scalable fiber form factor [11].
Synthetic fibers can be produced in large quantity, which speaks for
the economic benefit and feasibility of the proposed approach. Melt-
spinning is a widely used technique for production of textile fibers
[12], and we explore an adaptation for manufacturing fiber LSCs.
Namely, bi-component melt-spinning enables a continuous, single-
step production of polymer optical fibers (POFs) [13]. The classical
step-index POF consists of a light-conducting core, coaxially embedded
inside a lower-refractive-index cladding. In-situ spinning with the clad-
ding in place, elegantly removes the need for a solution coating process.
Bi-component spinning is solvent-free and represents a cost-effective
production method [14,15].
The most common polymer used in the design, production and re-
search of LSCs is polymethyl metacrylate (PMMA). Several other trans-
parent thermoplastics are available [16–18], which may offer a variation
of mechanical or optical properties. Luminescent light converting dyes
constitute a fast growing research and application field. Many different
luminophores are being studied as sunlight-converting additives, such
as organic dyes, nanocrystals or quantum dots [19–23]. Lumogen Red
305 (LR305) is a prominent example: this commercially available dye
has a close-to-unity quantum yield and an excellent thermal stability
up to 370 °C – a suitable additive for melt-spinning.
A careful selection of the polymer host-luminescent dye pairing is
crucial to help dispersion of luminescent molecules. The self-
absorption parameter [10,24] describes how the light emitted from a
dye molecule is reabsorbed by adjacent dye molecules along the light
path. Likewise, a low solubility of the dye in the host polymer melt
may lead to dye aggregation, which leads to suppressed light emission
via luminescence quenching [25,26]. While the dispersion of dye mole-
cules in liquid solvents is well understood [27–29], solubility of dye in
polymer melts has received less systematic attention.
In this study, some key material factors pertaining to the develop-
ment of melt-spun fiber LSCs are considered. Three polymers, namely
cycloolefin polymer (COP), polycarbonate (PC) and PMMA were evalu-
ated as possible host materials for the here fixed choice of LR305 dye. To
assess luminescent properties in these host polymers, we prepared dye-
doped polymer plates. Monte-Carlo ray tracing and molecular dynamics
simulations were used to link molecular properties to the plate geome-
try and results were validated experimentally. Prototype melt-spun bi-
component LSC fibers were manufactured using a doped COP core and
low-refractive index sheath material, namely terpolymer of tetrafluoro-
ethylene, hexafluoropropylene and vinylidene fluoride (THVP). The
illumination-angle dependent performance in absence of atmospheric
effects (such as scattering) was determined for defined bundles of
such prototype LSC fibers.
2. Materials and methods
2.1. Materials
Three polymers were tested as host materials for LR305 in the fiber
ð1Þ core: COP granulate (Zeonor 1020R, refractive index 1.53) was pur-
chased from Zeon Europe GmbH, Germany; PC granulate (Sabic Lexan
103R, refractive index 1.58) was purchased from Lenorplastics AG,
Switzerland; PMMA granulate (Plexiglas 7 N, refractive index 1.49)
was purchased from Evonik Industries AG, Switzerland. In addition, ter-
polymer of tetrafluoroethylene, hexafluoropropylene and vinylidene
fluoride granulate (THVP 2030GZ refractive index 1.35) was purchased
from 3 M Company, Germany, and used as a low-refractive-index, me-
chanically stable fiber sheath. The dye molecule LR305 was purchased
from BASF SE, Germany. All materials were used without further purifi-
ð2Þ cation. The chemical structures of the materials are shown in Fig. S1 in
the supporting information.
2.2. Melt-processing of materials
Polymer granulates were first dried overnight in a vacuum oven at
elevated temperatures. LR305 powder was then added to the dried
granulates of the fiber core polymer. To assure a uniform distribution
of the dye, the granulate/powder mixtures were tumbled for 5 h. To pre-
pare thin plates, the pre-mixed samples were melt-mixed at their re-
spective processing temperatures for 2 min in a torque rheometer
(HAAKE™ Rheomix OS Lab Mixer). The solidified extrudate collected
after melt-mixing was grinded, and compression molded into 1 mm-
thick plates with a hot press (Lindenberg Technics).
The fiber LSCs was melt-spun in a custom-made pilot plant de-
scribed elsewhere [30]. The sheath (cladding) material and the core
polymer melt were fed from two separate extruders. Metering pumps
transferred the melts into the spin pack, and the bi-component fiber
exited the spinneret into the quenching chamber, where it air-cooled
and solidified. Finally, the POF was taken up, drawn by heated godets,
and spooled onto a bobbin. Adapted godet speeds defined the optimal
tension for each fiber type. The nominal throughput was 150 m of con-
tinuous fiber LSC per minute. More details on melt-processing, with
exact processing temperatures and godet speeds, can be found in the
Supporting Information.
2.3. Optical characterization of plates
Absorption of the plates was measured using a UV–Vis spectropho-
tometer (Agilent Cary 4000 UV/Vis). Emission intensity was measured
from the plates using a spectrofluorometer (Horiba FluoroMax) with a
custom-made sample holder. Illumination and detection were set up
at perpendicular sample faces. Each measurement was repeated three
times.
 K. Jakubowski et al. / Materials and Design 189 (2020) 108518 3

2.4. Performance of melt-spun bi-component LSCs used in the experiment. For the presented analysis, the histograms
were fitted using Origin 2018.
To study optical behavior of individual fiber LSCs, an established
setup was used [31]. A glass optical fiber (Thorlabs M28L02) coupled 2.7. Molecular dynamics simulations
to a green LED light source (Thorlabs M505F3), illuminating the fiber
from the side, was translated along the fiber axis. At each position, the Fully atomistic molecular dynamics (MD) simulations were per-
emitted light at the fiber's tip was collected by another glass optical formed in order to calculate the Hildebrand solubility parameters of dif-
fiber (Thorlabs M59L01), which was coupled to a mini-spectrometer ferent molecules of interest from the cohesive energy density (CED)
(Hamamatsu C10083MD). [33–35]. Boxes containing 100 molecules were generated using the
Amorphous Cell module of the BIOVIA Materials Studio simulation soft-
ware [36]. The molecules were allowed to interact with each other using
2.5. Performance of LSC bundles
the COMPASS force field of the Forcite module of Materials Studio. Prior
to CED calculations, the simulation box was equilibrated in the NPT en-
Melt-spun fibers were combined into defined bundles (Fig. 5b): each
semble for 2–7 ns to reach equilibrium at 298 K and 1 atm.
bundle consisted of 200 fibers. To define their arrangement, the fibers
were inserted into transparent PMMA tubes (Rohm AG, Switzerland)
3. Results and discussion
of 5 mm inner and 7 mm outer diameter. The fibers covered 85% of
the cross-sectional area of the tube's lumen. The tubed bundles had
3.1. Luminescent properties of the dye-polymer systems
lengths of 20 mm, 40 mm and 80 mm. A calibrated photodetector
head with input optics 11 mm diameter diffusor window and spectral
In order to examine how different polymer hosts influence the lumi-
responsivity between 400 and 1000 nm (RW-3705-2, Gigahertz-Optik
nescence of LR305, the optical emission was measured from dye-doped
GmbH, Germany), attached to an optometer (P 9710, Gigahertz-Optik
polymer plates. PMMA, PC and COP plates were prepared with different
GmbH, Germany) was placed on a rotational stage. The sensitive surface
LR305 concentrations, as summarized in Table 1.
was masked with a black, non-transparent plate, leaving a circular
Absorbance spectra of three host materials (COP, PMMA and PC)
opening of diameter corresponding to that of the inner diameter of
with 0.05 wt% LR305 concentration were measured to determine the
the tubed fiber bunches. The rotational stage with the sensor resided in-
absorption coefficients for each wavelength required for ray-tracing
side a black box to prevent illumination by stray light. A Xenon short-arc
simulations described later in the paper, as well as potential self-
lamp (OSRAM XBO 450 W OFR (6000 K)) was used as a white light
absorption region of the LR305. Fig. 1 displays those results.
source, placed 3 m from the detector to approximate a plane wave
To quantify the emission strength of each of the host materials with
front, reducing square-distance intensity artefacts to b2%. Light inten-
different quantities of LR305, the emission spectra of all prepared plates
sity was measured with the masked photodetector and with perpendic-
were measured and comparatively shown in Fig. 2.
ular tubed fiber bundles on the circular opening. The detector in both
To avoid spectral overlap between emission spectra and excitation
configurations was rotated, and the intensity was recorded for a selec-
beam, the plates were excited by the fixed wavelength 450 nm. This
tion of illumination angles. Each measurement was repeated 3 times.
wavelength lies far from the expected emission and exhibits similar ab-
In addition to these detector measurements, a PV solar cell with dimen-
sorbance for LR305 in the three host materials (Fig. 1). The illumination
sions of 46 mm by 40 mm (Velleman SOL3N Polycrystalline Solar Mod-
was perpendicular to the plate's surface, and measured in transmission
ule 1 V, Conrad Electronic AG, Switzerland) was used for similar
at the same angle. As can be seen in Fig. 1, different plates exhibit a sim-
measurements. The solar cell was also masked with a black, non-
ilar absorbance at that wavelength. For each of the host materials, two
transparent foil, with a circular opening of diameter corresponding to
prominent emission peaks are visible (Fig. 2): namely at ≈590 nm
that of the inner diameter of the tubed fiber bunches. The spectral sen-
and at ≈640 nm. A third, significantly smaller peak is present at
sitivity of the PV solar cell is more closely representing a real application,
≈675 nm, which can be deconvoluted by means of more detailed
i.e. higher sensitivity in red and NIR. The mask was placed identically on
peak analysis. For PMMA and PC samples, both emission and absorption
the cell for all measurements to exclude variations of internal resistance
spectra seem red-shifted with respect to COP samples – this effect oc-
from the shadowed area. The illumination-angle dependent power was
curs due to the polarity of the host material [37]. It can also be observed,
measured, in the same setup as in the previous case, but now the max-
that the position of the first peak shifts towards longer wavelengths.
imum power point was followed on the I(V) curve. Each measurement
To gain additional insight, Fig. 3 displays the integral spectral area for
was repeated 3 times. More details about these measurements can be
all the measured thin-plate samples, calculated from the data in Fig. 2.
found in the SI.
We observe that the LR305 yields a higher emission intensity in the
COP matrix than in PMMA or PC. In addition, the position of the first
2.6. Monte-Carlo ray tracing simulations
Table 1
An open-source, Python-based ray-tracing software PVTRACE (ver- LR305 concentration in the prepared melt-processed thin plates.
sion 2.0.4) [32] was used to track random photons inside each of the
Sample LR305 concentration [wt%] LR305 concentration [μmol/cm3]
prepared polymer plates. The measured absorption and emission spec-
tra of LR305/polymer host systems were used as input data for these COP0 0.002 0.02
calculations. For the simulation, a point light source was placed inside COP1 0.01 0.09
COP2 0.05 0.47
the plate geometry next to the illuminated plate/air interface. The mo- COP3 0.07 0.66
lecular emission spectrum in the simulation corresponded to the previ- COP4 0.1 0.94
ously measured emission spectrum of LR305. For each simulation, COP5 0.25 2.34
250,000 random photons were generated. Each generated photon's ini- PMMA0 0.002 0.02
PMMA1 0.05 0.55
tial wavelength was randomly selected based on the defined emission
PMMA2 0.1 1.10
spectrum. The simulation accounts for photons that undergo different PMMA3 0.25 2.76
optical processes inside the geometry of the LSC plate (like self- PC0 0.002 0.02
absorption, re-emission, absorption by the polymer matrix, etc.). The PC1 0.05 0.56
resulting distribution of emitted photon wavelengths was statistically PC2 0.1 1.11
PC3 0.25 2.78
analyzed in a histogram (Fig. 4), which imitates the spectral analysis
4 K. Jakubowski et al. / Materials and Design 189 (2020) 108518

1.2
2.0 COP2 COP
PMMA1 PMMA
PC1 1.0
PC

Normalized intensity
1.5
Absorbance [a.u]

0.8

1.0 0.6

0.4
0.5

0.2

0.0
400 450 500 550 600 650
Wavelength [nm]
Fig. 1. Absorption spectra (with standard deviation) of COP2, PMMA1 and PC1 (all with
0.05 wt% LR305). The results indicate good absorption of blue, green and a small portion
of red light by LR305. (For interpretation of the references to colour in this figure legend,
the reader is referred to the web version of this article.)
peak apparently shifts towards longer wavelength with increasing dye
concentration (Fig. 2), which is the result of a marked self-absorbance
in this wavelength range (Fig. 1). As expected, the position of the second
and the third peak remain unchanged, in agreement with negligible
self-absorbance above 650 nm.
We used Monte Carlo ray-tracing simulations to connect the previ-
ously measured emission data to the light path in this plate geometry.
Normalized emission spectra of COP0, PMMA0 and PC0 (0.002 wt%
LR305 each) were used as reference emission spectra in the simulation,
since they were measured at minimal self-absorption (i.e. at low LR305
concentration). The absorption behavior of the plates with 0.05 wt%
LR305 (COP2, PMMA1 and PC1) was then simulated, assuming the pre-
viously measured absorption behavior (Fig. 1). Fig. 4 shows a compari-
son between simulated and measured emission spectra of these
samples.
The simulated emission spectra reproduce the peak positions and
peak shapes rather well. A comparison reveals, however, that the simu-
lation predicts a relatively stronger first emission peak. For quantification
of this effect, peak area ratios are calculated using the following relation:
Afirst peak
Rpeak area ¼
Asecond peak þ Athird peak
This parameter estimates the intensity of the first emission peak (i.e.
the one most affected by die self-absorption) relative to the other two
peaks (unaffected by self-absorption). The comparison of fitted
700 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
3
Concentration [Pmol/cm ]
Fig. 3. Normalized integrated emission intensities of the measured thin-plate samples.
simulated and experimental results is thus summarized in Table 2.
One can readily recognize a trend; namely, that theory predicts higher
peak ratios—an effect most prominent for the PC, and least prominent
for the COP sample.
Table 2 and Fig. 4 indicate the presence of additional attenuation ef-
fects not accounted for in the simulation; these effects depend on the
host material.
Collected results thus suggest there is an excess of self-absorption,
which is more pronounced in PMMA and PC samples. It is reasonable
to hypothesize that aggregation-induced quenching of the luminescent
dye molecules is the likely cause; it can explain the observed depen-
dence on the host polymer matrix.
To investigate this hypothesis, we performed molecular dynamics
(MD) simulations on pure COP and LR305 individually to compute
their respective Hildebrand solubility parameters, δ. This parameter
represents the cohesive energy density for each pure phase. The square
of the difference between them, Δ = (δA − δB)2, is a simple estimate of
the dispersion quality. A lower parameter Δ would indicate a better dis-
persion of the dye in the polymer matrix [38]. For the MD computation,
the structures of COP and LR305 were taken from previous publications
[39,40]. Hildebrand solubility parameters of pure PMMA and PC were
taken from literature [41]. The MD simulation is described in more de-
ð3Þ
tail in the Experimental Section. Table 3 summarizes the solubility pa-
rameters and Δ values.
The predicted best matrix for the dispersion of LR305 is indeed COP.
Therefore, we would expect less dye aggregates in COP than in PMMA or
PC [25,26].

Fig. 2. Luminescence spectra of a) COP plates, b) PMMA plates and c) PC plates with different LR305 concentrations (see: Table 1), measured perpendicular to the plates.
 K. Jakubowski et al. / Materials and Design 189 (2020) 108518
Fig. 4. Simulated (gray histogram) and measured (red curve) emission spectra of a) COP2 b) PMMA1 c) PC1 (all with 0.05 wt% LR305) in the plate geometry. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)
We think the systematically higher absorption of the first peak ob-
served in the previously simulated spectra (Fig. 4, Table 3) can be ex-
plained by aggregate formation, particularly since this 600 nm-peak
would be selectively affected by absorption from non-emitting dye ag-
gregates. The intensity ratio (“sim/exp” in Table 2) can thus be regarded
as a simple measure for dye-aggregate formation in the polymer matrix.
3.2. Properties of melt-spun, bi-component fiber LSCs
Based on previous results, a continuous bi-component fiber LSC was
produced on a pilot melt-spinning plant with COP as a core and THVP as
a cladding material. The production rate was 150 m per minute. The ob-
tained fibers exhibit a numerical aperture of 0.72. The LR305 dye con-
centration was 0.05 wt% in the core – this value was found to have the
highest emission intensity (Fig. 4). More details on the continuous
fiber production can be found in the Experimental section as well as in
the Supporting Information. Fig. 5a shows a cross-section of the studied
melt-spun bi-component fiber LSC – its core radius was 139 ± 1μm,
with an outer radius of 163 ± 2μm, amounting to a core portion of
73 ± 1vol%. This geometry allowed easy handling and characterization.
Melt-spinning provides excellent uniformity of produced fiber LSCs.
The hence obtained fiber LSC was studied with respect to the spectral
characteristics of the emitted light. Therefore, a point light source was
translated along the fiber axis, illuminating it from the side and the
light emitted at the tip was recorded using a fiber-coupled spectrome-
ter. The distance dependence of the emitted spectrum in Fig. 6a served
as basis to calculate a wavelength-dependent attenuation as displayed
in Fig. 6b, described in more detail the SI.
As seen in Fig. 6a, the spectrum maxima apparently shift towards
longer wavelength as the distance to the illumination point increases;
this is, again, the result of dye self-absorption. Light is lost along the
way in an exponential manner, which can be described by an attenua-
tion ratio [dB]. As seen in Fig. 6b, the attenuation spectrum resembles
the L305 absorption spectrum in COP (Fig. 1). Light in a wavelength
range below 615 nm is attenuated below measurement noise within a
traveling distance of 1 mm. On the other hand, the light in the range
above 615 nm can travel longer distances. For example, this can be
quantified in Fig. 6a (pink curve) where around 16% of the total light
can travel 160 mm to the fiber tip. Additional details about the setup
of the optical measurements are available in the SI.
Table 2
Peak area ratios (Rpeak area) for simulated and experimental results.
Simulation Experiment Ratio (sim/exp)
COP2 2.37 1.85
PMMA1 1.55 1.14
PC1 1.81 1.28
5
3.3. Characterization of multi-fiber bundle LSCs
Fiber bundles were prepared to test their light concentrating perfor-
mance with respect to the illumination angle. Bundles of three different
lengths were used, namely 20 mm, 40 mm and 80 mm (Fig. 5b). The
maximum length was chosen in accordance to the previously presented
attenuation measurements for a single fiber. To preserve a well-defined
orientation of the 200 fibers as part of a 5 mm diameter bundle, they
were held together by transparent PMMA tubes. Due to geometric con-
straints, a portion of ~62% of the tube area was covered with emitting
fiber cores. The photodetector used for intensity measurements was
masked with a fixed circular aperture matching the inner diameter of
the PMMA tubes. A Xe-arc lamp was used as an excitation light source;
it exhibits a quasi-white light illumination that approximates the solar
spectrum. A calibrated photodetector was rotated on a stage to vary
the relative direction of the incoming light and thus simulate different
sun positions on the sky with respect to the receiver, disregarding atmo-
spheric effects such as Rayleigh scattering. We adapt the following con-
vention: angle α = 0° for the illumination parallel to the surface of the
detector and α = 90° for the illumination perpendicular to the surface
of the detector. Fig. 7a compares the recorded intensities at different
bundle lengths as well as the masked bare sensor.
The highest power was measured for the bare photodetector (PD) at
perpendicular illumination (α = 90 °). We note that the measured in-
tensity exhibits a distinctively different angular dependence in presence
of the fiber bundles. Namely, at shallower illumination angles, the inten-
sity recorded for bundles is higher, as the incoming light is received by a
larger projected area of the LSC bundle, which is expressed by the fol-
lowing equation:
Acollection ðα Þ ¼ L∙2r∙ cosα þ π∙r 2 ∙ sinα ð4Þ
where L is the length of the cylinder and r is its radius. To study the
length-dependence of the bundle LSC efficiency from such angle-
dependent measurement, we split the incoming illumination into two
orthogonal components:
Iðα Þ ¼ A  cosα þ B  sinα ð5Þ
where I(α) is the angle dependent intensity, and A and B are the orthog-
onally convoluted contributions containing geometrical and optical
Table 3
Hildebrand solubility parameters of luminescent dye and polymer hosts.
Material Hildebrand solubility parameter [(J/cm3)0.5] Δ [J/cm3]
LR305 17.05 –
1.28 COP 15.47 2.50
1.36 PMMA 19.00 3.80
1.42 PC 19.90 8.12
6 K. Jakubowski et al. / Materials and Design 189 (2020) 108518

Fig. 5. a) Microscopic image of a cross-section of the prepared melt-spun bi-component fiber LSC. b) 20 mm, 40 mm and 80 mm PMMA tubes filled with fiber LSCs.

factors influencing the performance of bundle LSC. By fitting the curves
in Fig. 7a to Eq. (5), parameters A and B can be extracted. The hence ob-
tained values are plotted in Fig. 7b for different lengths of fiber bundles.
As expected, the portion A grows with increasing bundle length, due to
enhanced collection of lateral light by the LSC. We note that the growth
of A is sub-linear with bundle length. Fitting the observed asymptotic
behavior of this component reveals the useful LSC length of around
80 mm, as imposed by fiber attenuation, presented in Fig. 6b. This max-
imum effective length will depend on dye and agglomerate concentra-
tion. The portion B encompasses a light collection maximum for bare
detector, and sharply decreases with implementation of bundles. This
relates to the attenuated direct transmission of illuminating light
along the bundle axis, i.e. mainly in-coupling at the open fiber tip end.
To study angle-dependent optical conversion within the fiber, we
expand the analysis used in literature by, for example, R. Reisfeld et al.
[9], E.-H. Banaei et al. [10], S. Correia et al. [42] and I. Parola et al. [43],
and calculate the optical efficiency ηopt as a ratio between optical
power produced by bundle LSC, to power received from the illumina-
tion, taking into account the effect of varying illumination angle:
P out ðα Þ Iout ðα Þ∙Aoutput
ηopt ðα Þ ¼ ∙100% ¼ ∙100%
P in ðα Þ Iin ∙Acollection ðα Þ
where Pout(α) and Pin(α) are produced and received powers, respec-
tively, Iout(α) and Iin are output and incoming intensities, respectively,
and Aoutput and Acollection(α) are emitting and collecting areas, respec-
tively. In the case of cylindrical geometry, such as the bundle LSC pre-
sented, the collecting area is changing with the angle following
Eq. (4). Fig. 8 shows the calculated efficiencies for varying angles α, in
comparison with previously published references.
We note that the values obtained for bundle LSCs are comparable to
those reported in other studies [14,15,42–46], underlining the compet-
itiveness of the presented bi-component polymer optical fibers. The 80-
mm-long bundle, despite yielding the highest absolute intensities
(Fig. 7a), achieves the lowest optical efficiencies among studied config-
urations (Fig. 8), while the 20-mm-long bundle performs in the oppo-
site way. The increase of optical efficiency above 80° is presumably
caused by the non-ideal structure of a bundle – as luminescent fibers
do not cover the whole cross section, direct light outside the fiber
cores can reach the detector at α = 90°.
We now assess a more realistic scenario comprising the spectral re-
sponse of a polycrystalline silica (pSi) PV solar cell, i.e. additionally
benefitting from the conversion of blue into red light. As before, we dis-
regard atmospheric effect. Fig. 9 shows the maximal power generated
by the pSi cell, again without and with fiber bundles of different lengths,
using the same illumination setup as before.
The power is determined at the maximum power point of the I
(V) curve. The polycrystalline silica solar cells have increased sensitivity
around 760–800 nm. [47] Compared to the white-calibrated photode-
ð6Þ
tector, the light conversion by LR305 thus explains the excess of
power recorded (viz. Figs. 9 and 7a). This conversion gain is known in
the LSC literature [48].
For α = 0°, the F-factor (Eq. (1)) reaches values of 2.8, 4.3 and 5.3 for
bundle lengths of 20 mm, 40 mm, and 80 mm, respectively. To estimate
the theoretical performance of a LSC-enhanced PV solar cell at varying
angles (i.e. simulating different arrangements between the sun and

Fig. 6. a) Spectrum of light reaching the fiber LSC's tip for different distances between illumination point and emitting tip. b) Calculated attenuation of light as a function of wavelength. For
wavelengths b 615 nm it was calculated between 0 mm and 2 mm and for wavelengths N 615 nm attenuation was calculated between 0 mm and 160 mm.
 K. Jakubowski et al. / Materials and Design 189 (2020) 108518 7

Fig. 7. a) Intensity of light detected by a bare photodetector, as well as a photodetector in contact with LSC bundles of three different lengths. b) Calculated values of A and B deconvolution
parameters for bundles of different length.

cell's surface without considering atmospheric effects), one can inte-
grate the measured intensity of Fig. 9 over the full range of angles;
Table 4 summarizes the results.
We estimate that the pSi cell, enhanced with 40 mm and 80 mm
bundles, can be improved 13% and 26% respectively, compared to the
bare PV panel. A beneficial side effect is that power generation is now
distributed more evenly over the angles, as compared to the reference
pSi cell without LSC. This angular redistribution property is interesting
to mitigate the mid-day power peak generated by conventional photo-
voltaics with fixed orientation.
4. Conclusions
polymer melts. The here described methods are suitable to study
other polymer-dye systems. A prototype of an endless melt-spun
bi-component fiber LSC demonstrates the feasibility and up-scaling
capability of the proposed solution. Fiber-based LSCs could enhance
and complement the illumination-angle dependency of commercial
PV solar panels. More generally, the high values of the F-factors for
shallow angles suggest that fiber-based LSCs, in form of textiles or
bundles, could also find photovoltaic applications in more indirect il-
lumination scenarios. Our results suggest that there is a significant
potential in utilizing industrial melt-spinning as cost-effective
means for fiber-based LSC production.

Three transparent thermoplastic polymers, COP, PMMA and PC,
were evaluated as host matrices for LR305. COP was best suited as
host material for LR305, in terms of luminescent properties. Compar-
ison of measured spectra and ray tracing simulations revealed an ex-
cess absorption in plate model samples that we attribute to the
formation of dye aggregates in the polymer matrix. Molecular dy-
namics simulation were invoked to estimate the interactions be-
tween luminescent dye and polymer host in melt-processing,
providing a tool to predict dispersion of (luminescent) additives in
CRediT authorship contribution statement
Konrad Jakubowski: Conceptualization, Investigation, Validation,
Visualization, Formal analysis, Writing - original draft. Chieh-Szu
Huang: Investigation, Validation. Ali Gooneie: Methodology, Validation.
Luciano F. Boesel: Conceptualization, Validation. Manfred Heuberger:
Conceptualization, Validation, Writing - review & editing, Supervision.
Rudolf Hufenus: Conceptualization, Validation, Writing - review &
editing, Supervision.

100
20mm bundle 1.4
Unassisted cell 8

40mm bundle 20mm bundle power & F-factor

Normalized maximal power

80mm bundle 1.2 40mm bundle power & F-factor 7

Ref. 42 80mm bundle power & F-factor

Calculated F-factor

6
1.0
10
5
Ref. 46
0.8
ηopt [%]

4
0.6
Ref. 15
3
Ref. 14
1
0.4
2

Ref. 44
0.2 1
Ref. 43, Ref. 45
0.0 0
0.1 0 20 40 60 80
0 20 40 60 80 100
Angle D [o]
Angle D [ ] o

Fig. 9. Maximal power registered as a function of the illumination angle α (solid lines) and
Fig. 8. Angle-dependent optical efficiencies of the studied bundle LSCs. Reference values, values of the F-parameter (dashed lines), for a bare PV solar cell and an enhanced cell with
reported previously in the literature, are marked on the graph at α = 0°, as reported. LSC bundles of different lengths, in absence of atmospheric scattering effects.
8 K. Jakubowski et al. / Materials and Design 189 (2020) 108518
Table 4
Integral of the theoretical maximal power between illumination angles of 0° and 90° of the
LSC-assisted solar cell, in proportion to the unassisted cell.
Measurement Ratio between area under curves (Fig. 9) of
configuration LSC-assisted and unassisted cell
Unassisted solar cell – [15]
Cell with 20 mm fiber LSC bundle 0.92
Cell with 40 mm fiber LSC bundle 1.13
Cell with 80 mm fiber LSC bundle 1.26
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgements
The authors thank Martin Amberg and Urs Schütz for assistance with
the solar cell measurement setup, Mathias Lienhard and Benno Wüst for
assistance with polymer melt-processing, Dr. René Rossi for valuable
discussions, as well as Dr. Sergii Yakunin for assistance with
photoluminescent quantum yield measurements.
Data availability
The raw and processed data required to reproduce these findings
cannot be shared at this time as the data also forms part of an ongoing
study.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.matdes.2020.108518.
References
[1] The National Renewable Energy Laboratory, Best research-cell efficiency chart,
https://www.nrel.gov/pv/cell-efficiency.html 2019, Accessed date: 1 July 2019.
[2] P.G.V. Sampaio, M.O.A. Gonzalez, Photovoltaic solar energy: conceptual framework,
Renew. Sust. Energ. Rev. 74 (2017) 590–601, https://doi.org/10.1016/j.rser.2017.02.
081.
[3] F. Meinardi, A. Colombo, K.A. Velizhanin, R. Simonutti, M. Lorenzon, L. Beverina, R.
Viswanatha, V.I. Klimov, S. Brovelli, Large-area luminescent solar concentrators
based on stokes-shift-engineered nanocrystals in a mass-polymerized PMMA ma-
trix, Nat. Photonics 8 (2014) 392–399, https://doi.org/10.1038/nphoton.2014.54.
[4] M.G. Debije, P.P.C. Verbunt, Thirty years of luminescent solar concentrator research:
solar energy for the built environment, Adv. Energy Mater. 2 (2012) 12–35, https://
doi.org/10.1002/aenm.201100554.
[5] W. Tress, K. Domanski, B. Carlsen, A. Agarwalla, E.A. Alharbi, M. Graetzel, A. Hagfeldt,
Performance of perovskite solar cells under simulated temperature-illumination
real-world operating conditions, Nat. Energy 4 (2019) 568–574, https://doi.org/10.
1038/s41560-019-0400-8.
[6] C.-A. Bunge, T. Gries, M. Beckers, Polymer Optical Fibres, First, Elsevier, 2017https://
doi.org/10.1016/C2014-0-00562-X.
[7] S.F.H. Correia, P. Lima, E. Pecoraro, S. Ribeiro, P. André, R. Ferreira, L. Carlos, Scale up
the collection area of luminescent solar concentrators towards metre-length flexible
waveguiding photovoltaics, Prog. Photovolt. Res. Appl. 24 (2016) 1178–1193,
https://doi.org/10.1002/pip.
[8] R. Reisfeld, D. Shamrakov, C. Jorgensen, Photostable solar concentrators based on
fluorescent glass films, Sol. Energy Mater. Sol. Cells 33 (1994) 417–427, https://
doi.org/10.1016/0927-0248(94)90002-7.
[9] R. Reisfeld, New developments in luminescence for solar energy utilization, Opt.
Mater. (Amst). 32 (2010) 850–856, https://doi.org/10.1016/j.optmat.2010.04.034.
[10] E.-H. Banaei, A.F. Abouraddy, Design of a polymer optical fiber luminescent solar
concentrator, Prog. Photovolt. Res. Appl. 23 (2015) 403–416, https://doi.org/10.
1002/pip.
[11] K.R. McIntosh, N. Yamada, B.S. Richards, Theoretical comparison of cylindrical and
square-planar luminescent solar concentrators, Appl. Phys. B Lasers Opt. 88
(2007) 285–290, https://doi.org/10.1007/s00340-007-2705-8.
[12] R. Hufenus, Y. Yan, M. Dauner, D. Yao, T. Kikutani, Biocomponent fibers, Handb. Fi-
brous Mater, 2019.
[13] F.A. Reifler, R. Hufenus, M. Krehel, E. Zgraggen, R.M. Rossi, L.J. Scherer, Polymer op-
tical fibers for textile applications - bicomponent melt spinning from cyclic olefin
polymer and structural characteristics revealed by wide angle X-ray diffraction,
Polym. (United Kingdom). 55 (2014) 5695–5707, https://doi.org/10.1016/j.
polymer.2014.08.071.
[14] S.F.H. Correia, A.R. Frias, L. Fu, R. Rondão, E. Pecoraro, S.J.L. Ribeiro, P.S. André, R.A.S.
Ferreira, L.D. Carlos, Large-area tunable visible-to-near-infrared luminescent solar
concentrators, Adv. Sustain. Syst. 1800002 (1–9) (2018) 1800002, https://doi.org/
10.1002/adsu.201800002.
R.H. Inman, G.V. Shcherbatyuk, D. Medvedko, A. Gopinathan, S. Ghosh, Cylindrical
luminescent solar concentrators with near-infrared quantum dots, Opt. Express 19
(2011) 24308, https://doi.org/10.1364/OE.19.024308.
[16] G. Griffini, Host matrix materials for luminescent solar concentrators: recent
achievements and forthcoming challenges, Front. Mater. 6 (2019) https://doi.org/
10.3389/fmats.2019.00029.
[17] M. Zettl, O. Mayer, E. Klampaftis, B.S. Richards, Investigation of host polymers for lu-
minescent solar concentrators, Energy Technol 5 (2017) 1037–1044, https://doi.
org/10.1002/ente.201600498.
[18] Y. Li, X. Zhang, Y. Zhang, R. Dong, C.K. Luscombe, Review on the role of polymers in
luminescent solar concentrators, J. Polym. Sci. Part A Polym. Chem. 57 (2019)
201–215, https://doi.org/10.1002/pola.29192.
[19] K. Wu, H. Li, V.I. Klimov, Tandem luminescent solar concentrators based on
engineered quantum dots, Nat. Photonics 12 (2018) 105–110, https://doi.org/10.
1038/s41566-017-0070-7.
[20] A.R. Frias, M.A. Cardoso, A.R.N. Bastos, S.F.H. Correia, P.S. André, L.D. Carlos, V. de Z.
Bermudez, R.A.S. Ferreira, Transparent luminescent solar concentrators using Ln 3
+ -based ionosilicas towards photovoltaic windows, Energies 12 (2019) https://
doi.org/10.3390/en12030451.
[21] Y. Zhao, G.A. Meek, B.G. Levine, R.R. Lunt, Near-infrared harvesting transparent lumi-
nescent solar concentrators, Adv. Opt. Mater. 2 (2014) 606–611, https://doi.org/10.
1002/adom.201400103.
[22] C. Yang, R.R. Lunt, Limits of visibly transparent luminescent solar concentrators, Adv.
Opt. Mater. 5 (2017) 1–10, https://doi.org/10.1002/adom.201600851.
[23] A. Anand, M.L. Zaffalon, G. Gariano, A. Camellini, M. Gandini, R. Brescia, C. Capitani, F.
Bruni, V. Pinchetti, M. Zavelani-Rossi, F. Meinardi, S.A. Crooker, S. Brovelli, Evidence
for the band-edge exciton of CuInS 2 nanocrystals enables record efficient large-area
luminescent solar concentrators, Adv. Funct. Mater. (2019) 1906629, https://doi.
org/10.1002/adfm.201906629.
[24] L.R. Wilson, B.S. Richards, Measurement method for photoluminescent quantum
yields of fluorescent organic dyes in polymethyl methacrylate for luminescent solar
concentrators, Appl. Opt. 48 (2009) 212, https://doi.org/10.1364/AO.48.000212.
[25] Y. Huang, J. Xing, Q. Gong, L.-C. Chen, G. Liu, C. Yao, Z. Wang, H.-L. Zhang, Z. Chen, Q.
Zhang, Reducing aggregation caused quenching effect through co-assembly of PAH
chromophores and molecular barriers, Nat. Commun. 10 (2019) 169, https://doi.
org/10.1038/s41467-018-08092-y.
[26] R.F. Chen, J.R. Knutson, Mechanism of fluorescence concentration quenching of car-
boxyfluorescein in liposomes: energy transfer to nonfluorescent dimers, Anal.
Biochem. 172 (1988) 61–77, https://doi.org/10.1016/0003-2697(88)90412-5.
[27] H. Yoo, J. Yang, A. Yousef, M.R. Wasielewski, D. Kim, Excimer formation dynamics of
intramolecular π-stacked perylenediimides probed by single-molecule fluorescence
spectroscopy, J. Am. Chem. Soc. 132 (2010) 3939–3944, https://doi.org/10.1021/
ja910724x.
[28] R.O. Al-Kaysi, T. Sang Ahn, A.M. Müller, C.J. Bardeen, The photophysical properties of
chromophores at high (100 mM and above) concentrations in polymers and as neat
solids, Phys. Chem. Chem. Phys. 8 (2006) 3453–3459, https://doi.org/10.1039/
b605925b.
[29] L.R. Wilson, B.C. Rowan, N. Robertson, O. Moudam, A.C. Jones, B.S. Richards, Charac-
terization and Reduction of Reabsorption Losses in Luminescent Solar Concentra-
tors, Vol. 49, 2010 1651–1661.
[30] R. Hufenus, F.A. Reifler, K. Maniura-Weber, A. Spierings, M. Zinn, Biodegradable bi-
component fibers from renewable sources: melt-spinning of poly(lactic acid) and
poly[(3-hydroxybutyrate)-co-(3-hydroxyvalerate)], Macromol. Mater. Eng. 297
(2012) 75–84, https://doi.org/10.1002/mame.201100063.
[31] I. Parola, E. Arrospide, F. Recart, M. Illarramendi, G. Durana, N. Guarrotxena, O.
García, J. Zubia, Fabrication and characterization of polymer optical fibers doped
with perylene-derivatives for fluorescent lighting applications, Fibers 5 (2017) 28,
https://doi.org/10.3390/fib5030028.
[32] D.J. Farrell, D.J. Farrell, pvtrace: optical ray tracing for luminescent materials and
spectral converter photovoltaic devices, https://github.com/danieljfarrell/pvtrace
2019.
[33] A. Gooneie, P. Simonetti, K.A. Salmeia, S. Gaan, R. Hufenus, M.P. Heuberger, En-
hanced PET processing with organophosphorus additive: flame retardant products
with added-value for recycling, Polym. Degrad. Stab. 160 (2019) 218–228, https://
doi.org/10.1016/j.polymdegradstab.2018.12.028.
[34] K.A. Salmeia, A. Gooneie, P. Simonetti, R. Nazir, J.-P. Kaiser, A. Rippl, C. Hirsch, S.
Lehner, P. Rupper, R. Hufenus, S. Gaan, Comprehensive study on flame retardant
polyesters from phosphorus additives, Polym. Degrad. Stab. 155 (2018) 22–34,
https://doi.org/10.1016/j.polymdegradstab.2018.07.006.
[35] A. Gooneie, S. Schuschnigg, C. Holzer, A review of multiscale computational methods
in polymeric materials, Polymers (Basel) 9 (2017) https://doi.org/10.3390/
polym9010016.
[36] Accelrys Inc, BIOVIA materials studio, http://accelrys.com/, Accessed date: 20 July
2018.
[37] J.R. Lakowicz, Effects of solvents on fluorescence emission spectra, Princ. Fluoresc.
Spectrosc, Springer US, Boston, MA 1983, pp. 187–215, https://doi.org/10.1007/
978-1-4615-7658-7_7.
[38] L.H. Sperling, Introduction to Physical Polymer Science, John Wiley & Sons, Inc., Ho-
boken, NJ, USA, 2005https://doi.org/10.1002/0471757128.
 K. Jakubowski et al. / Materials and Design 189 (2020) 108518
[39] A.A. Leal, J.P. Best, D. Rentsch, J. Michler, R. Hufenus, Spectroscopic elucidation of
structure-property relations in filaments melt-spun from amorphous polymers,
Eur. Polym. J. 89 (2017) 78–87, https://doi.org/10.1016/j.eurpolymj.2017.02.009.
[40] K.A. Colby, J.J. Burdett, R.F. Frisbee, L. Zhu, R.J. Dillon, C.J. Bardeen, Electronic energy
migration on different time scales: concentration dependence of the time-resolved
anisotropy and fluorescence quenching of Lumogen red in poly(methyl methacry-
late), J. Phys. Chem. A 114 (2010) 3471–3482, https://doi.org/10.1021/jp910277j.
[41] A.F.M. Barton, CRC Handbook of Solubility Parameters and Other Cohesion Parame-
ters, Routledge, 2017https://doi.org/10.1201/9781315140575.
[42] S.F.H. Correia, P.P. Lima, P.S. André, M.R.S. Ferreira, L.A.D. Carlos, High-efficiency lu-
minescent solar concentrators for flexible waveguiding photovoltaics, Sol. Energy
Mater. Sol. Cells 138 (2015) 51–57, https://doi.org/10.1016/j.solmat.2015.02.032.
[43] I. Parola, M.A. Illarramendi, J. Zubia, E. Arrospide, G. Durana, N. Guarrotxena, O.
García, R. Evert, D. Zaremba, H.-H. Johannes, F. Recart, Polymer optical fibers
doped with organic materials as luminescent solar concentrators, Proc. SPIE - Int.
Soc. Opt. Eng. 10101 (2017) 101010Z, https://doi.org/10.1117/12.2252956.
[44] I. Parola, D. Zaremba, R. Evert, J. Kielhorn, F. Jakobs, M.A. Illarramendi, J. Zubia, W.
Kowalsky, H.H. Johannes, High performance fluorescent fiber solar concentrators
9
employing double-doped polymer optical fibers, Sol. Energy Mater. Sol. Cells 178
(2018) 20–28, https://doi.org/10.1016/j.solmat.2018.01.013.
[45] R. Rondão, A.R. Frias, S.F.H. Correia, L. Fu, V. De Zea Bermudez, P.S. André, R.A.S.
Ferreira, L.D. Carlos, High-performance near-infrared luminescent solar concentra-
tors, ACS Appl. Mater. Interfaces 9 (2017) 12540–12546, https://doi.org/10.1021/
acsami.7b02700.
[46] E.-H. Banaei, A.F. Abouraddy, Fiber luminescent solar concentrator with 5.7% conver-
sion efficiency, High Low Conc. Syst. Sol. Electr. Appl. VIII. VIII 2013, pp. 1952–1955,
https://doi.org/10.1117/12.2022601.
[47] A. Pourakbar Saffar, B. Deldadeh Barani, Thermal effects investigation on electrical
properties of silicon solar cells treated by laser irradiation, Int. J. Renew. Energy
Dev. 3 (2014) https://doi.org/10.14710/ijred.3.3.184-187.
[48] X. Huang, S. Han, W. Huang, X. Liu, Enhancing solar cell efficiency: the search for lu-
minescent materials as spectral converters, Chem. Soc. Rev. 42 (2013) 173–201,
https://doi.org/10.1039/C2CS35288E.