Ligand exchange as a tool to improve quantum dot miscibility in polymer composite layers used as luminescent down-shifting layers for photovoltaic applications

Article abstract of DOI:10.1039/c6tc01261b

The efficiency of solar cells with a poor UV response can be improved by altering the incident light spectrum with luminescence down-shifting (LDS). Stable luminescent additives with a large Stokes shift, high photoluminescence quantum yield (PLQY) and a high absorption coefficient are required for this purpose. Quantum dots (QDs) are attractive candidates which meet most of the basic requirements for LDS additives. Herein, commercially available heavy metal free core shell CuInS₂/ZnS QDs are theoretically and experimentally evaluated as LDS additives. The small apolar dodecanethiol (DDT) ligands of the QDs are exchanged with thiol functional oligo-caprolactone ligands, via a ligand exchange process, to improve the QD compatibility with the UV curable resin matrix. Aggregation of the QDs is prevented to a large extent in the polymer-QD composite films and the material exhibits luminescence properties which are virtually identical to the luminescence of the QDs dispersed in chloroform. Raman spectroscopy and NMR spectroscopy are used to elucidate the ligand exchange and ligand binding processes. It is shown that well-dispersed CuInS₂/ZnS QDs are required with a near unity quantum yield to increase the efficiency of solar cells significantly especially if high performance inorganic solar cells are employed.

Full text of DOI:10.1039/c6tc01261b

Journal of
Materials Chemistry C
View Article Online
View Journal
PAPER
Ligand exchange as a tool to improve quantum
dot miscibility in polymer composite layers used
as luminescent down-shifting layers for
photovoltaic applications†
Cite this:DOI: 10.1039/c6tc01261b
Guy J. J. Draaisma,*ab Damien Reardon,a Albertus P. H. J. Schenning,b
Stefan C. J. Meskersb and Cees W. M. Bastiaansenbc
The efficiency of solar cells with a poor UV response can be improved by altering the incident light
spectrum with luminescence down-shifting (LDS). Stable luminescent additives with a large Stokes shift,
high photoluminescence quantum yield (PLQY) and a high absorption coefficient are required for this
purpose. Quantum dots (QDs) are attractive candidates which meet most of the basic requirements for
LDS additives. Herein, commercially available heavy metal free core shell CuInS2/ZnS QDs are theoretically
and experimentally evaluated as LDS additives. The small apolar dodecanethiol (DDT) ligands of the QDs
are exchanged with thiol functional oligo-caprolactone ligands, via a ligand exchange process, to improve
the QD compatibility with the UV curable resin matrix. Aggregation of the QDs is prevented to a large
extent in the polymer–QD composite films and the material exhibits luminescence properties which are
virtually identical to the luminescence of the QDs dispersed in chloroform. Raman spectroscopy and NMR
spectroscopy are used to elucidate the ligand exchange and ligand binding processes. It is shown that
well-dispersed CuInS2/ZnS QDs are required with a near unity quantum yield to increase the efficiency of
solar cells significantly especially if high performance inorganic solar cells are employed.
Received 27th March 2016,
Accepted 16th May 2016
DOI: 10.1039/c6tc01261b
www.rsc.org/MaterialsC
term photo-stable.1,3 Fluorescent organic dyes fulfill most of the
above described requirements. However, they have a small
1. Introduction
Photovoltaic (PV) cells can suffer from low external quantum Stokes shift and consequently exhibit a large spectral overlap
efficiency (EQE) in the UV spectral region. Luminescence down- which results in significant re-absorption losses.1 In addition
shifting (LDS) has been proposed to improve the spectral they suffer from poor photo-stability although this depends on
response by absorbing short wavelength UV light and emitting the type of dye and matrix.5,6 Rare earth metal chelates have high
at wavelengths where PV cells have high EQE.1 Besides fluor- PLQY but they have a narrow absorption band and an extremely
escent organic dyes, down-shifting additives such as rare earth low absorption coefficient and require ligands or absorbers that
metal chelates and quantum dots (QDs) were proposed.1–4 It was transfer the energy to the luminescent ion.1 QDs are inorganic
also shown that efficient down-shifting additives: (i) require semi-conductor nanocrystals with tunable optical bandgaps;
a near unity quantum yield, (ii) require a high absorbance they have a wide absorption band and high photo-stability.7,8
in the spectral region where the cell EQE is low, (iii) require a Depending on the type of semi-conductor, they suffer from a
large Stokes shift and minimal spectral overlap, (iv) emit in large spectral overlap and consequent re-absorption losses.1,9
the region where the cell EQE is high, (v) are transparent QDs typically consist of toxic heavy metal containing semi-
at wavelengths where the cell has high EQE and (vi) are long conductors such as CdSe and PdS. Recently less toxic heavy
metal free QDs based on for example InP and ZnS doped with
Mn or CuInS2 have been developed. As LDS additives QDs based
on InP are not attractive due to the limited Stokes shift and
spectral overlap.10 ZnS QDs doped with Mn and CuInS2 QDs
might be useful as down-shifting additives and they are already
evaluated for lighting applications.10,11 CuInS2 QDs are charac-
terized by a large Stokes shift of 1.0–1.4 eV and emission in the
500–700 nm region.10 A PLQY of 470% was reported previously
for core shell systems of CuInS2 with a ZnS shell.12–15 The wide
a DSM Ahead, P.O. Box 18, 6160 MD Geleen, The Netherlands.
E-mail: Guy.draaisma@dsm.com; Tel: +31 464764255
b Eindhoven University of Technology, Chemical Engineering and Chemistry,
Functional Organic Materials & Devices (SFD), P.O. Box 513, 5600 MB,
Eindhoven, The Netherlands
c School of Engineering and Materials Science, Queen Mary University of London,
Mile End Road, London E1 4NS, UK
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
c6tc01261b
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. C

View Article Online
Paper
Journal of Materials Chemistry C
absorption band and the large Stokes shift in combination with Under a slight exothermic reaction an off white precipitate
emission in the 500–700 nm spectral region, where most PV cells formed over 15 minutes. The mixture was kept stirring at room
have high EQE, make CuInS2/ZnS QDs attractive heavy metal free temperature for 20 minutes after the exothermic reaction. The
QD based LDS additives. Polymer–QD composites of various product was filtered off and was washed with ethyl acetate (40 mL)
types of QDs have been successfully prepared.4,16–20 However a and water (50 mL). The white product (4.36 g, 10.7 mmol) was
common issue is the aggregation of the QDs during processing isolated and dried at 40 1C under reduced pressure to constant
leading to a number of issues such as transmission losses due to mass with an isolated yield of 87%. 1H-NMR (300 MHz,
light scattering, a red shift of the emission spectrum, a drop of chloroform-d) d 3.63 (t, J = 6.6 Hz, 4H, –CH2–OH), 2.66 (t, J =
the PLQY and inhomogeneous light absorption.11,19,21,22 Aggre- 7.4 Hz, 4H, –S–S–CH2–), 1.75–1.48 (m, 8H, –CH2–CH2–OH,
gation of QDs and matrix compatibility are highly dependent on –CH2–CH2–S–), 1.44–1.18 (m, 30H, –CH2–CH2–CH2–) ppm.
the organic ligands that are grafted onto the surface of the QDs. 13C-NMR (300 MHz, chloroform-d) d 63.4 (–CH2–OH), 39.6
Introducing other types of ligands can be done by ligand exchange (–S–S–CH2–), 33.2 (–CH2–CH2–OH), 30.0–29.5 (–CH2–CH2–CH2–),
processes.23,24 Commercially available ligands such as 11-mercapto- 28.9 (–S–S–CH2–CH2–CH2–CH2–), 26.1 (–CH2–CH2–OH) ppm.
1-undecanol, thioctic acid, thiogycolic acid, 1-octadecanethiol and IR (powder) nmax 3424, 3313, 2917, 2848, 1469, 1347, 1056, 1037,
1-dodecanethiol are either very hydrophilic or very hydrophobic 1003, 975, 717, 606, 522 cmÀ1. Anal. calcd for C22H46O2S2: C 64.97,
and yield QDs which are either dispersible in polar or apolar H 11.40, N 0.0; found: C 63.6, H 10.8, N o0.1. HRMS (ESI)
media. However for miscibility with polymeric materials these m/z [M + Na]+ calcd for C22H46O2S2, 429.28; found: 429.2826.
ligands are not always satisfactory. Polymeric ligands that are
2.3 Ligand synthesis
intrinsically miscible with the host matrix were reported to
obtain well-dispersed QDs in polymeric materials.22
e-Caprolactone (4.24 g, 37 mmol) was added to the diol initiator
In this work, we describe a method for preparing a polymer (950 mg, 2.30 mmol). The mixture was heated under a nitrogen
heavy metal free CuInS2/ZnS QD luminescent composite, which atmosphere to 120 1C at which tin(II)octoate (3.5 mg, 8.6 mmol)
is prepared from commercially available QDs obtained from was added as a catalyst. The mixture was kept at 120 1C for
PlasmaChem GMBH and a UV curable aliphatic polyester- 1 hour for the ring opening polymerization of e-caprolactone.
urethane acrylate resin (Neoradt U-20-12H) obtained from Upon cooling to room temperature the crude ligand dimer
DSM. The small molecule ligand 1-dodecanethiol (DDT) is crystallized. The ligand dimer was dissolved in THF (100 mL) to
compared to a synthesized oligomeric ligand to improve the which hydrochloric acid (7 mL, 0.7 mmol) was added. Sodium
QD dispersibility in the matrix. Raman and NMR spectroscopy borohydride (250 mg, 6.6 mmol) was added in portions over
are used to elucidate the ligand exchange and ligand binding 30 minutes as a reducing agent for the disulfide linkage. The
processes. The luminescence properties of the QDs before mixture was allowed to react for 1 hour at room temperature
formulation and UV curing as well as of UV cured composite after the last addition. After the reaction a solvent switch to
film were evaluated. Moreover the obtained luminescent QD 2-methyl-THF was made and the product was worked up by
containing films were assessed as down-shifting layers for extraction with 0.1 M hydrochloric acid. The organic layer was
photovoltaic (PV) cells based on an experimental evaluation dried over sodium sulfate and purified by precipitation in a
and based on theoretical simulations.
methanol : water mixture (6 : 1). The off white product was dried
under reduced pressure to constant mass (3.41 g, 3.1 mmol)
and was obtained with an isolated yield of 66%. 1H-NMR (300 MHz,
chloroform-d) d 4.05 (t, J = 6.7 Hz, 18H, –CH2–O–CQO), 3.63
(t, J = 6.5 Hz, 2H, –CH2–OH), 2.51 (q, J = 7.4 Hz, 2H, –CH2–SH),
2.30 (t, J = 7.5 Hz, 18H, –O–CQO–CH2–), 1.94–1.48 (m, 42H,
–CH2–CH2–CH2–), 1.48–0.93 (m, 33H, –CH2–CH2–CH2) ppm.
13C-NMR (300 MHz, chloroform-d) d 174.0 (CQO), 64.9–64.4
(–CH2–O–CQO), 63.0 (–CH2–OH), 34.7–34.3 (O–CQO–CH2–),
32.7 (–CH2–CH2–OH), 29.8–28.6 (–CH2–CH2–CH2–), 26.4–24.8
(–CH2–S, –CH2–CH2–CH2–) ppm. IR (powder) nmax 3440, 3188,
2941, 2859, 1721, 1472, 1420, 1396, 1367, 1237, 1177, 1108, 1045,
961, 936, 842, 732, 586, 524 cmÀ1. Anal. calcd for C59H104O17S
(n = 8): C 63.41, H 9.38, N 0.0; found: C 62.9, H 9.0. N o0.1.
HRMS (ESI) m/z [M + Na]+ calcd for (n = 6) C47H84O13S, 911.55
found: 911.5510. Calcd for (n = 7) C53H94O15S, 1025.62 found:
1025.6186 and higher oligomers were detected.
2. Experimental section
2.1 Chemicals
All reagents were used as received without further purification
unless otherwise stated. A photo-curable aliphatic polyester-
urethane acrylate resin (NeoRadt U-20-12H) was provided by
DSM. CuInS2/ZnS core/shell QDs were purchased from PlasmaChem
(Berlin, Germany). Sodium iodide (NaI), 11-mercapto-1-undecanol,
e-caprolactone, tin(II)octoate, sodium sulfate anhydrous, hydrogen
peroxide 30% solution in water, sodium borohydride, lauryl acrylate,
2-methyl-THF, ethyl acetate, acetone, THF, sodium hydroxide,
glycerol, 0.1 M hydrochloric acid and dichloromethane were
obtained from Sigma-Aldrich.
2.2 Diol initiator synthesis
2.4 Ligand exchange
11-Mercapto-1-undecanol (5.02 g, 24.6 mmol) was dissolved in
ethyl acetate (100 mL) to which NaI (36.0 mg, 0.24 mmol) was A general procedure for the exchange of dodecanethiol (DDT)
added as a catalyst. To the stirring solution 30 wt% hydrogen ligands with thiol functional oligo-CL ligands for CuInS2/ZnS
peroxide in water (3 mL) was added as the oxidizing agent. QDs is described below. Excess ligands were removed prior to
J. Mater. Chem. C
This journal is ©The Royal Society of Chemistry 2016

View Article Online
Journal of Materials Chemistry C
Paper
the ligand exchange procedure by two consecutive precipita- were performed on a Jobin Yvon Horiba – LabRam confocal
tions of the DDT capped QDs. Precipitations were performed by Raman spectrometer equipped with an Olympus BX40 micro-
the addition of methanol to a QD dispersion in chloroform. scope. As a light source a 632 nm HeNe laser was used. Optical
The QD precipitate was collected by centrifugation. Typically and fluorescence microscopy was performed using an Olympus
QDs were dispersed in chloroform with a concentration of BX51 microscope equipped with a CCD camera (Olympus XC50)
30 mg mLÀ1 during the procedure. For the ligand exchange and an X-Cite series 120Q excitation source. TEM/STEM was
QDs (30 mg) with DDT ligands were dispersed in chloroform performed using an FEI Osiris TEM operating at an acceleration
(1 mL). In a separate vial thiol functional oligo-CL (250 mg) was voltage of 200 kV in the TEM bright field and STEM (HAADF)
dissolved in chloroform (1 mL) and was added as a whole to the mode. Ultrathin sections were obtained at À45 1C with an average
QD dispersion. The mixture was heated under an inert atmo- thickness of 60–70 nm. Elemental analysis for the elements C, H
sphere to reflux and the chloroform was allowed to evaporate. and N was carried out using a LECO Truspec CHN analyzer. Mass
The remaining melt of the thiol functional oligo-CL ligand with spectrometry flow injection analysis in THF was performed using
QDs was further heated to 130 1C for 4.5 hours. After the a Maxis HD Q TOF mass spectrometer equipped with an ESI
reaction time the mixture was cooled to 60 1C and acetone source operating in the positive ion mode.
(500 ml) was added, a transparent colored QD dispersion was
obtained. The dispersion was further cooled down to room
temperature. At room temperature the excess of the thiol
3. Results and discussion
functional oligo-CL was removed by consecutive precipitations
with methanol. The procedure was typically repeated three
times. The QD precipitate was collected by centrifugation and
dried under reduced pressure.
3.1 Quantum-dot absorption and emission characteristics
CuInS2/ZnS core/shell QDs with an emission wavelength at
610 nm were selected as down shifting additives since most
PV cells have high EQE at this wavelength. The UV-Vis absorp-
tion spectrum of the as received QDs was recorded and showed
that the QDs absorbed below a wavelength of 600 nm with a
decreasing absorption coefficient as a function of wavelength
(Fig. 1). The absorption coefficient of the QDs determined at
350 nm was 5.0 cmÀ1 g mgÀ1, the extinction coefficient was
obtained in this unit by expressing the QD concentration as
milligram QDs per gram liquid. QDs have high molar extinction
coefficients; however, expressing the extinction coefficient on a
2.5 Fabrication of down-shifting layers
QD dispersions in chloroform were mixed with the UV curable
aliphatic polyester-urethane acrylate resin by placing the sample
on a roller bench for 15 minutes. The homogenously mixed
formulation was applied between two glass microscope slides
separated by appropriate spacers to obtain the desired layer
thickness. The films were cured by irradiating the sample with
UV light (2.5 J cmÀ2), as a UV light source a Thomson UV MAG
was used.
2.6 Methods
UV-Vis absorption spectra were recorded using a Perkin Elmer
(Lambda 35) spectrophotometer equipped with an integrating
sphere. Absolute PLQY measurements were performed accord-
ing to the procedure described by Wilson et al.25 As an excita-
tion source a 405 nm laser with an output power of 5 mW was
used, emission spectra were recorded using an integrating
sphere (Labsphere LMS-100) equipped with a CCD detector
(Labsphere CDS-610). Data were acquired using the LightMtrX
software package. NMR analysis was performed on a 300 MHz
Bruker Avance 1 spectrometer. Dynamic light scattering (DLS)
measurements were performed in quartz cuvettes on a Malvern
Zetasizer Nano ZS in a 1731 backscatter configuration at 25 1C
in chloroform. The refractive index of the QDs was set at 2,
since the exact refractive index was unknown. EQE measure-
ments were performed on a set-up consisting of the following
components: a light source (Muller LXH-100), a monochromator
(Newport 74000), a pre-amplifier (Stanford research systems
SR570), and a lock-in amplifier (Stanford research systems
SR830). Data processing was done using the software package
Origin (OriginLab). The EQE was measured at an interval of
10 nm in the range of 350–1100 nm. Fluorescence lifetimes were
recorded using an Edinburgh Instruments Lifespec-PS equipped
with a PicoQuant PDL-800-B pulsed laser. Raman measurements
Fig. 1 UV-Vis absorption and photoluminescence emission spectra of
dilute solutions of DDT and oligo-CL capped CuInS2/ZnS QDs in chloro-
form and of cured composite films. The absorption and emission spectra
of QDs capped with DDT as well as with oligo-CL ligands are identical
which suggests that the ligand exchange did not influence the light
absorption and emission characteristics of the QDs. A red shift of 25 nm
and significant broadening of the emission peak was observed for com-
posite films prepared with DDT capped QDs. The broadening was attrib-
uted to the severe aggregation of the quantum dots and a consequent loss
of the quantum confinement. A small 10 nm red shift of the emission
spectrum and no peak broadening was observed for composite films
prepared with oligo-CL capped QDs.
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. C

View Article Online
Paper
Journal of Materials Chemistry C
mass basis is more relevant for down-shifting additives since
this is proportional to the required amount of the additive
in the LDS layer. According to the Beer–Lambert law, the
measured extinction coefficient of 5.0 cmÀ1 g mgÀ1 at 350 nm
corresponds to a required concentration of 40 mg gÀ1 in the
resin formulation for a 50 micron thick LDS layer with an optical
density of 1. The photoluminescence emission spectrum (Fig. 1)
was recorded with an integrating sphere coupled to a laser
excitation source with a wavelength of 405 nm.
The emission spectrum revealed a broad emission peak with
a full width half maximum (FWHM) of 4100 nm; which is
typical for CuInS2/ZnS QDs.15,26–28 Previously, the broad emis-
sion peak was attributed to defect related emission.12,15,26,29,30
Emission can result from multiple luminescence decay path-
ways with either slightly higher energy (band edge) or lower
energy (sub bandgap) states and explains the broad nature
of the emission peak.27,31 However, it should be noted that
no consensus has been reached about the exact decay
mechanism.27
Fig. 2 (A and B) Fluorescence microscopy and bright field TEM micro-
graphs of polymer–QD composites prepared with DDT capped QDs.
Aggregation of the QDs was observed in the case of DDT ligands. The
inset in image B shows a higher magnification TEM micrograph of
aggregated QDs in the polymeric film. (C and D) Fluorescence microscopy
and high angular annular dark field (HAADF) STEM micrographs of
polymer–QD composites prepared with QDs capped with oligo-CL ligands.
QDs capped with oligo-CL ligands were well distributed throughout the
film, HAADF-STEM with high magnification was required to visualize the
QDs, the QDs were present as small clusters in which the individual QDs
could be distinguished as shown in the inset of image D.
3.2 QD matrix compatibility and ligand exchange
The as received QDs were capped with apolar dodecanethiol
(DDT) ligands and were dispersible in solvents such as toluene
and chloroform. Upon addition of the QDs to these solvents
highly colored optically clear dispersions without light scatter-
ing were obtained. Unfortunately, DDT capped QDs were not
compatible with the resin formulation and formed large
agglomerates upon addition to the resin formulation (Fig. 2).
DDT capped QDs were not evaluated further for this reason.
A ligand exchange process was explored to improve the disper-
sion of the QDs in the resin formulation. Commercially avail-
able ligands such as 11-mercapto-1-undecanol and lipoic acid
were too polar and were as such also not soluble in the resin
formulation. Since a ligand with the desired properties was not
commercially available, a ligand compatible with the resin was
synthesized.
Fig. 3 Synthetic scheme of oligo-CL ligand synthesis.
3.3 Ligand design and synthesis
3.4 Ligand exchange, a NMR and Raman spectroscopy study
A suitable ligand needs to be miscible with the resin formula- Ligand exchange was performed in a melt of the oligo-CL ligand
tion, have an interaction with the QD surface and be able to at elevated temperature and at high concentration of the new
replace existing ligands. Oligomeric caprolactone (CL) was ligand to drive the equilibrium in favor of the new ligand,
ˆ
miscible with the resin formulation; however, as such an according to Le Chatelier’s Principle. Successful ligand exchange
oligomer has no affinity with the QD surface and is not suitable was achieved with conversions 480% on a molar basis, accord-
as a ligand to replace DDT. As a QD binding moiety a thiol end- ing to 1H-NMR spectroscopy (Fig. S1, ESI†). In the 1H-NMR
group was introduced to the oligomer. For the study an oligo- spectrum it was observed that the integral of the two aliphatic
CL ligand with a molecular weight (Mn) of B1100 g molÀ1 a-protons of the thiol end-group at 2.5 ppm did no longer match
was synthesized from 11-mercapto-1-undecanol, which was the theoretical amount after ligand exchange. A 2D carbon–
used as an initiator for the ring opening polymerization of proton (HSQC) correlation NMR revealed that the observed
e-caprolactone, in a three step synthetic approach (Fig. 3). The broad peak at 2.6–2.9 ppm was assigned to the thiol a protons
oligomeric ligand with an Mn of B1100 g molÀ1 was selected to of ligands that were reversibly bound to the QD surface (Fig. 5).
already drastically impact the properties of the QDs when only a The reversible nature of the ligand binding process explained
small fraction of the ligand would bind to the QDs. In addition why the integral of the peak at 2.5 ppm did not match the
the relatively high molecular weight ligands might reduce the theoretical integral, the signal originated only from the ligand
aggregation of the QDs by steric repulsion. The obtained fraction in solution. From the 1H-NMR spectra (Fig. S1, ESI†) the
oligomeric ligands were grafted onto the QD surface via a ligand fraction free in solution was estimated to be around 25%
ligand exchange procedure (Fig. 4).
while 75% of the ligands were bound to the QD surface in the
J. Mater. Chem. C
This journal is ©The Royal Society of Chemistry 2016

View Article Online
Journal of Materials Chemistry C
Paper
Fig. 4 Schematic representation of the ligand exchange process of DDT
ligands with oligo-CL ligands.
NMR sample diluted with CDCl3. In addition the higher resolu-
tion of the 2D spectrum, recorded at 600 MHz, revealed two
broad signals in the 2.6–2.9 ppm region and not one as pre-
viously observed in the 1H-NMR spectrum recorded at 300 MHz
(Fig. S1, ESI†). The lower intensity broad signal at 2.8 ppm was
attributed to DDT ligands and the higher intensity signal to
oligo-CL ligands. Solid-state NMR relaxometry was applied to
investigate the relaxation behavior of the oligo-CL ligand free in
CDCl3 solution as well as of QDs with the oligo-CL ligands
dispersed in CDCl3 (Fig. S2, ESI†). QDs with the oligo-CL ligands
dispersed in CDCl3 showed a complex relaxation behavior. The
decay could be fitted with a three component exponential
function; the multi-component relaxation behavior was attributed
to the heterogeneous mobility of ligands reversibly bound to the
QD surface. Besides NMR, the Raman spectroscopy of solid QD
samples after ligand exchange confirmed that the ligands were
bound via the thiol functionality to the QD surface based on the
absence of the characteristic Raman scattering peak of the thiol
Fig. 5 (A) Graphical representation of a QD after ligand exchange with
oligo-CL ligands and some remaining DDT ligands. (B) Proton–carbon 2D
(HSQC) correlation NMR spectrum of CuInS2/ZnS QDs after ligand
exchange with oligo-CL ligands. (C) Raman spectra of the oligo-CL ligand
with a thiol end-group, a ligand dimer and of QDs with oligo-CL ligands.
The oligo-CL ligand and the ligand dimer served as reference compounds
for, respectively, the thiol end-group and disulphide. The spectra were
background corrected and normalized on the carbonyl (CQO) signal at
1721 cmÀ1. In the inset the spectrum from 550–800 cmÀ1 is shown to
indicate the C–S signal more clearly.
(S–H) at 2576 cmÀ1 and disulfides (S–S) at 526 cmÀ1. Other prepared by UV curing. The films had a high transmission
characteristic Raman peaks of the ligand, like the carbon sulfur outside the QD absorption region and were not light scattering
bond (C–S) at 650–750 cmÀ1 and the ester carbonyl (CQO) at (Fig. S3, ESI†). The QD distribution was assessed with optical
1721 cmÀ1, were present (Fig. 5). It should be noted that the fluorescence microscopy as well as with scanning transmission
thiol (S–H) peak was completely absent in the Raman spectrum, electron microscopy (STEM). With optical fluorescence micro-
in contrast to solution state NMR which indicated 25% scopy a homogenous fluorescence was observed throughout the
free ligands in solution. It is hypothesized that the ligands sample (Fig. 2). STEM showed that the QDs were present as
coordinated to the QD surface in the dry state. A shift of the small clusters throughout the sample (Fig. 2). The slight
equilibrium between free ligands and bound ligands might be aggregation of QDs had almost no effect on the transmittance
induced during the preparation of the dry samples for Raman of the film since the clusters were smaller than 20 nm, and
spectroscopy. From the above NMR and Raman spectroscopy according to Raleigh’s law transmission losses in the visible
analysis it followed that 480% of the DDT ligands were light region were well below 0.2% for clusters this small. This
replaced with oligo-CL ligands and that the ligands were was also confirmed by the transmittance measurement in
reversibly bound via the thiol functionality to the QD surface. comparison to a film without QDs (Fig. S3, ESI†).22 The
The successful ligand exchange process resulted in drastically fluorescence properties of the composite resembled that of
improved dispersion of the QDs in the resin formulation. QDs the QD starting material, the absolute PLQY was unchanged
capped with oligo-CL ligands were easily dispersible in the at 15% while a 10 nm red-shift of the emission spectrum was
resin formulation and optically transparent formulations were observed (Fig. 1). The red shift was only observed in cured films
obtained and the agglomeration of the QDs was prevented also and not after ligand exchange, the small red-shift of the
after curing of the resin with UV light (Fig. 2).
emission spectrum was for this reason attributed to QD cluster-
ing in cured films as observed with STEM (Fig. 2). The absolute
PLQY of 15% of the batch of CuInS2/ZnS QDs used for the
3.5 Composite characterization
From the QD formulation with oligo-CL ligands in the resin evaluation was low compared to literature values where PLQY higher
optically transparent polymer–QD composites were successfully than 70% were frequently reported for CuInS2/ZnS QDs.12–15
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. C

View Article Online
Paper
Journal of Materials Chemistry C
We attributed the low PLQY to surface defects of the QDs which
resulted in an increased number of trapped states which were
prone to non-radiative recombination processes. Most impor-
tantly the ligand exchange process as well as the UV curing for
the preparation of the composite films did not negatively impact
the PLQY or cause a large shift of the emission spectrum. No
changes were observed in fluorescence lifetime measurements of
QD before and after ligand exchange as well as of the prepared
composite films (Fig. S4, ESI†). The lifetime decay plots were
identical and indicated that the luminescence process of the
QDs was not affected by the processing.
3.6 Experimental and theoretical evaluation of the LDS layers
on a PV cell
Fig. 7 EQE measurements and simulations of a mc-Si solar cell in
combination with a polymer–QD down-shifting layer. Due to the low
PLQY of the QD composite a reduction of the EQE was measured in the
region where the QDs absorbed. Good agreement between the simulated
EQE plot and the measurements was obtained. The measurements were
performed in duplicate (open and closed symbols).
The prepared luminescent composite films were evaluated as
LDS layers on top of a multi-crystalline silicon (mc-Si) solar cell
by recording the external quantum efficiency (EQE) of the cell
as a function of wavelength. Optical contact between the film
and the cell was obtained by the addition of glycerol (n B 1.47)
between the LDS layer and the cell. An illustration of the
QD-polymer composite on top of the PV cell is shown in
Fig. 6. The results of the EQE measurement were compared
to a blank measurement with a polymer film without QDs
(Fig. 7). The comparison was made in this way to correct for
additional reflections caused by the introduction of an extra
layer on top of the bare PV cell. Unfortunately the introduction
of the LDS layer resulted in a reduction of the cell’s EQE in the
region where the QDs absorbed light. This EQE reduction was
attributed to the low PLQY of 15% of the QDs and the already
relatively good UV performance of the blank cell.
used as input the EQE of the bare PV cell, the measured PLQY
(15%) and the absorption and emission spectrum of the LDS
layer. Re-absorption losses were calculated to be B2% based on
the spectral overlap of the absorption and emission spectrum
and out-coupling losses were calculated for a refractive index of
1.5 which corresponds to B13% loss according to Snell’s law
(Section S5, ESI†). The outcome of the theoretical simulation
matched closely the measured EQE of the mc-Si cell as illu-
strated in Fig. 7. It was concluded that the model was suitable
for predicting the effect of the QD-composite LDS layer on the
performance of a PV cell. The theoretical model was further
used as a predictive tool to evaluate the effect of the PLQY
(Fig. 8). The simulations showed that a PLQY 4 75% was
required to improve the UV performance of the evaluated
mc-Si cell and that a near unity PLQY was required to increase
the short circuit current of the cell. Various EQE curves
A theoretical model to simulate the effect of LDS on the
performance of the PV cell as outlined by Rothemund et al. has
been applied for theoretical evaluations.32,33 A brief description
of the theoretical model and related equations is given in the
Section S5, ESI.† A theoretical simulation was performed which
Fig. 8 Simulated EQE plots as a function of PLQY compared to the
measured EQE of a mc-Si cell without an LDS layer. From the simulation it
follows that a PLQY 4 0.75 is required to improve the UV response of the cell.
Fig. 6 Photograph of the polymer–QD composite placed on top of a
mc-Si solar cell. The photo has been taken under 366 nm UV illumination.
J. Mater. Chem. C
This journal is ©The Royal Society of Chemistry 2016

View Article Online
Journal of Materials Chemistry C
Paper
luminescence properties comparable to the starting material.
The introduction of an oligomeric caprolactone ligand via
ligand exchange was essential to improve the compatibility of
the QDs with the resin and reduced QD aggregation in the
polymer matrix. The performance of the composites as LDS layers
was poor due to the low PLQY of the QDs. A large discrepancy was
observed between the PLQY of CuInS2/ZnS QDs reported in the
literature (470%) and the results presented here (B15%). Experi-
mental results were in good agreement with theoretical simula-
tions and from the theoretical model it followed that a near unity
PLQY was required to have a positive effect on the integrated
current of the evaluated PV cells. The application of CuInS2/ZnS as
a down shifting additive for PV application remains limited as
long as the PLQY is not near unity.
Fig. 9 Absolute integrated current simulations for various cell types as a
function of the PLQY of the CuInS2/ZnS LDS layer.
Acknowledgements
obtained from literature sources, representing different PV cell
types, were evaluated with the model to get an indication of the
potential effect on the current of these cells, by simulating the
effect of a LDS layer with CuInS2/ZnS QDs.9,34–36 The effect on
the integrated current of the evaluated cell is shown as a function
of the PLQY of the LDS layer (Fig. 9). The data were obtained by a
full integration over the wavelength region of the cell’s absorbance,
as the input spectrum the AM 1.5g solar spectrum was used. The
theoretical simulations, as a function of the PLQY, showed that as
expected a low PLQY leads to a reduction of the integrated current
of the cell. Depending on the cell type, however, an increase of the
maximum current of the cell was possible with a PLQY 4 50%.
The absolute gain of the cells integrated current, with a unity
PLQY, can be up to 0.5% for the CdTe/CdS cell and 0.4% for the
evaluated CIGS cell. The evaluated perovskite and mc-Si cell could
only benefit marginally even when the PLQY was assumed to be
unity. From this evaluation it can be concluded that in general cell
types with a poor UV performance might benefit from LDS while
cells with a good UV response like recent record holding cells as
reported by Green et al. will not benefit at all.37
The authors would like to thank F. J. M. Colberts MSc for the
help with the EQE measurements, Dr J. Laven for the help with
Raman spectroscopy, Dr V. Litvinov for the help with the NMR
relaxometry experiments, Dr P. L. E. M. Pasmans for support
with the theoretical simulations and DSM for financial support.
References
1 E. Klampaftis, D. Ross, K. R. Mcintosh and B. S. Richards,
Sol. Energy Mater. Sol. Cells, 2009, 93, 1182–1194.
2 E. Klampaftis and B. S. Richards, Prog. Photovoltaics, 2011,
19, 345–351.
3 B. S. Richards, D. Alonso-Alvarez, D. Ross, E. Klampaftis,
K. R. McIntosh, S. Jia, P. Storiz and T. Stolz, Prog. Photo-
voltaics, 2014, 23, 479–497.
4 S. D. Hodgson, W. S. M. Brooks, A. J. Clayton, G. Kartopu,
V. Barrioz and S. J. C. Irvine, Nano Energy, 2014, 4, 1–6.
5 R. Kinderman, L. H. Slooff, A. R. Burgers, N. J. Bakker,
A. Bu¨chtemann, R. Danz and J. A. M. van Roosmalen, J. Sol.
Energy Eng., 2007, 129, 277–282.
Most cell types currently developed have already a fairly
good UV performance and will consequently not benefit from
LDS. It should however be noted that the potential of LDS is
also dependent on the input spectrum and might be more
beneficial for blue shifted spectra.
6 A. A. Earp, T. Rawling, J. B. Franklin and G. B. Smith,
Dyes Pigm., 2010, 84, 59–61.
7 W. G. J. H. M. Van Sark, A. Meijerink and R. E. I. Schropp,
InTech, 2011, 1–29.
Based on the outcome of the theoretical prediction it was
concluded that developing QDs with near unity PLQY is essential
for their application as LDS additives for PV applications. Increas-
ing the PLQY to the desired near unity values is challenging for
CuInS2/ZnS QDs, because of the high proportion of defect-based
nonradiative decay pathways.27 Currently the PLQY values for
CuInS2/ZnS QDs are still far from unity, despite the fact that large
improvements have been made in developing a synthetic route for
CuInS2/ZnS QDs with higher PLQY.
8 D. Bera, L. Qian, T. K. Tseng and P. H. Holloway, Materials,
2010, 3, 2260–2345.
9 W. G. J. H. M. Van Sark, A. Meijerink, R. E. I. Schropp,
J. A. M. Van Roosmalen and E. H. Lysen, Semiconductors,
2004, 38, 962–969.
10 M. J. Anc, N. L. Pickett, N. C. Gresty, J. A. Harris and
K. C. Mishra, ECS J. Solid State Sci. Technol., 2012, 2,
R3071–R3082.
11 C. Yoon, T. Kim, M. Shin, Y. Song, K. Shin, Y. Kim and
K. Lee, J. Mater. Chem. C, 2015, 3, 6908–6915.
12 W. Zhang, Q. Lou, W. Ji, J. Zhao and X. Zhong, Chem. Mater.,
2014, 26, 1204–1212.
4. Conclusions
Luminescent QD–polymer composites were prepared success- 13 L. Li, A. Pandey, D. J. Werder, B. P. Khanal, M. Pietryga and
fully via UV curing, and the process yielded a material with
V. I. Klimov, J. Am. Chem. Soc., 2011, 133, 1176–1179.
This journal is ©The Royal Society of Chemistry 2016
J. Mater. Chem. C

View Article Online
Paper
Journal of Materials Chemistry C
14 W. S. Song and H. Yang, Appl. Phys. Lett., 2012, 100, 0–4.
15 X. Wang, Z. Liang, X. Xu, N. Wang, J. Fang, J. Wang and
G. Xu, J. Alloys Compd., 2015, 640, 134–140.
26 W. Liu, Y. Zhang, J. Zhao, Y. Feng, D. Wang, T. Zhang,
W. Gao, H. Chu, J. Yin, Y. Wang, J. Zhao and W. W. Yu,
J. Lumin., 2015, 162, 191–196.
´
16 N. Tomczak, D. Janczewski, M. Han and G. J. Vancso, 27 A. D. P. Leach and J. E. Macdonald, J. Phys. Chem. Lett.,
Prog. Polym. Sci., 2009, 34, 393–430.
2016, 7, 572–583.
17 P. K. Khanna and N. Singh, J. Lumin., 2007, 127, 474–482.
18 A. G. Midgett, J. M. Luther, J. T. Stewart, D. K. Smith,
28 K. Qi, Y. Wang, R. Wang, D. Wua and G. Li, J. Mater. Chem.
C, 2016, 4, 1895–1899.
L. A. Padilha, V. I. Klimov, A. J. Nozik and M. C. Beard, 29 K. E. Knowles, H. D. Nelson, T. B. Kilburn and D. R.
Nano Lett., 2013, 13, 3078–3085. Gamelin, J. Am. Chem. Soc., 2015, 137, 13138–13147.
19 K. Matsuyama, Y. Maeda, T. Matsuda and T. Okuyama, 30 N. Droseros, K. Seintis, M. Fakis, S. Gardelis and A. G.
J. Supercrit. Fluids, 2015, 103, 83–89. Nassiopoulou, J. Lumin., 2015, 167, 333–338.
20 L. Chen, C. Zhang, Z. Du, H. Li, L. Zhang and W. Zou, 31 D. H. Jara, K. G. Stamplecoskie and P. V Kamat, J. Phys.
J. Appl. Polym. Sci., 2015, 132, 42667. Chem. Lett., 2016, 7, 1452–1459.
21 O. O. Matvienko, Y. N. Savin, A. S. Kryzhanovska, O. M. 32 R. Rothemund, S. Kreuzer, T. Umundum, G. Meinhardt,
Vovk, M. V. Dobrotvorska, N. V. Pogorelova and V. V.
Vashchenko, Thin Solid Films, 2013, 537, 226–230.
T. Fromherz and W. Jantsch, Energy Procedia, 2011, 10, 83–87.
33 R. Rothemund, Sol. Energy Mater. Sol. Cells, 2014, 120, 616–621.
¨
22 S. Ehlert, C. Stegelmeier, D. Pirner and S. Forster, Macro- 34 B. S. Richards and K. R. McIntosh, Prog. Photovoltaics, 2007,
molecules, 2015, 48, 5323–5327. 15, 27–34.
23 H. Kim, M. Suh, B.-H. Kwon, D. S. Jang, S. W. Kim and 35 R. Wuerz, A. Eicke, F. Kessler, S. Paetel, S. Efimenko and
D. Y. Jeon, J. Colloid Interface Sci., 2011, 363, 703–706. C. Schlegel, Sol. Energy Mater. Sol. Cells, 2012, 100, 132–137.
24 Y. Shen, R. Tan, M. Y. Gee and A. B. Greytak, ACS Nano, 36 D. Son, J. Im, H. Kim and N. Park, J. Phys. Chem. C, 2014,
2015, 9, 3345–3359. 118, 16567–16573.
25 L. R. Wilson and B. S. Richards, Appl. Opt., 2009, 48, 37 E. D. Dunlop, M. A. Green, K. Emery, Y. Hishikawa and
212–220.
W. Warta, Prog. Photovoltaics, 2015, 23, 1–9.
J. Mater. Chem. C
This journal is ©The Royal Society of Chemistry 2016