Non-Covalent Postfunctionalization of Dye Layers on TiO2 — A Tool for Enhancing Injection in Dye-Sensitized Solar Cells

Article abstract of DOI:10.1002/chem.202004928

We report on newly tailored dye layers, which were employed, on one hand, for covalent deposition and, on the other hand, for non-covalently post-functionalizing TiO₂ nanoparticle films. Our functionalization concept enabled intermixing a stabl

Full text of DOI:10.1002/chem.202004928

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
&
Dye-Sensitized Solar Cells |Hot Paper|
Non-Covalent Postfunctionalization of Dye Layers on TiO2 — A
Tool for Enhancing Injection in Dye-Sensitized Solar Cells
Tobias Luchs,[a] Anna Zieleniewska,[b] Andreas Kunzmann,[b] Peter R. Schol,[b] Dirk M. Guldi,*[b]
and Andreas Hirsch*[a]
Dedicated to Professor Luis Echegoyen on the occasion of his 70th anniversary
Abstract: We report on newly tailored dye layers, which
were employed, on one hand, for covalent deposition and,
on the other hand, for non-covalently post-functionalizing
TiO2 nanoparticle films. Our functionalization concept en-
abled intermixing a stable covalent attachment of a first
layer with a highly versatile and reversible hydrogen bond-
ing through the Hamilton receptor–cyanuric acid binding
motif as a second layer. Following this concept, we integrat-
ed step-by-step a first porphyrin layer and a second porphy-
rin/BODIPY layer. The individual building blocks and their
corresponding combinations were probed with regard to
their photophysical properties, and the most promising
combinations were implemented in dye-sensitized solar cells
(DSSCs). Relative to the first porphyrin layer adding the
second porphyrin/BODIPY layers increased the overall DSSC
efficiency by up to 43%.
Introduction
tionalization through covalent modifications has been ex-
plored. The most prominent examples rely on nucleophilic
substitutions[10] or click chemistry, especially the dipolar Huis-
gen 1,3-cycloaddition to couple azides and acetylenes.[11] In
contrast to covalent approaches, self-assembly based on non-
covalent functionalization supports a high degree of reversibili-
ty and versatility.[12] Controlled SAM formation based on cova-
lent as well as non-covalent interactions made a series of
mixed inorganic–organic nanoparticles with smoothly tunable
surface properties accessible.[13] The resulting nanoparticles
have, for example, been utilized in either removing contami-
nants from water[5b] or in radiation therapy.[14] The layer-by-
layer technique (LbL), which was first introduced by Decher, is
one prominent example of hierarchical architectures, which are
solely based on the utilization of non-covalent forces.[15] A sub-
strate, for instance, is alternately dipped into solutions of
either positively or negatively charged polymers. Hereby, the
layered architectures are stabilized through electrostatic inter-
actions. In recent years, this strategy was used for the forma-
tion of functional, nanosized architectures by layering tailored
molecular building blocks and charged nanoparticles.[16] When
using H-bonding, stronger binding and higher directionality
evolved. Important is the fact that solvents govern the
strength of H-bonding and, in turn, the precise control over
the reversibility. Similar to the LbL methodology, H-bonding
stands out for functionalizing nanomaterials.[17] One of the
most prominent examples is the Hamilton receptor/cyanuric
acid binding motif (Figure 1).[18] Its high binding constant in
the range from 103 to 106 LmolÀ1 is truly remarkable.[18a,19] Not
surprisingly, this binding motif enabled the preparation of mo-
lecular wires,[20] rotaxanes through a H-bonding template,[21] at-
tachment to carbon allotropes,[22] perylene–fullerene–porphyrin
The construction of hybrid architectures starting at nanosized
entities and molecular building blocks is a powerful tool to-
wards the development of advanced multifunctional materi-
als.[1] Leading examples have made it into applications such as
catalysis,[2] sensing,[3] drug delivery,[4] removal of water contami-
nants,[5] and dye sensitized solar cells[6] (DSSCs). Potential path-
ways in terms of realizing such hybrid architectures are mani-
fold and include among many others self-assembly via cova-
lent or non-covalent interactions. Covalent functionalization,
for example, has extensively been employed in the context of
highly ordered self-assembled monolayers (SAMs).[7] This ap-
proach has enabled the realization of many functional materi-
als like transistors[8] and DSSCs.[9] More recently, SAM post-func-
[a] T. Luchs, Prof. Dr. A. Hirsch
Chair of Organic Chemistry II, Department of Chemistry & Pharmacy
Friedrich-Alexander-Universität Erlangen-Nürnberg
Nikolaus-Fiebiger-Straße 10, 91058 Erlangen (Germany)
E-mail: andreas.hirsch@fau.de
[b] Dr. A. Zieleniewska, Dr. A. Kunzmann, P. R. Schol, Prof. Dr. D. M. Guldi
Chair of Physical Chemistry I, Department of Chemistry & Pharmacy
Friedrich-Alexander-Universität Erlangen
Egerlandstraße 3, 91058 Erlangen (Germany)
E-mail: dirk.guldi@fau.de
Supporting Information and the ORCID identification number(s) for the
author(s) of this article can be found under:
https://doi.org/10.1002/chem.202004928.
ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-
VCH GmbH. This is an open access article under the terms of the Creative
Commons Attribution Non-Commercial NoDerivs License, which permits
use and distribution in any medium, provided the original work is properly
cited, the use is non-commercial and no modifications or adaptations are
made.
Chem. Eur. J. 2021, 27, 5041 –5050
5041 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
Herein, we report on the controlled deposition of porphy-
rin–dipyrromethane boron difluoride (BODIPY) and porphyrin–
porphyrin bilayers onto the surface of doctor-bladed TiO2 films
via the combination of covalent and non-covalent self-assem-
bly (Scheme 1). Initially, several Hamilton receptor porphyrins
and/or porphyrin cyanurates featuring one or two carboxylates
as TiO2 anchors were prepared through multi-step organic syn-
theses. Furthermore, a series of BODIPY or porphyrin ligands
equipped with a complementary supramolecular recognition
motif was prepared. In a first step the supramolecular binding,
electrochemistry, and photophysical properties of these build-
ing blocks were studied in solution. Finally, the binding was
transferred from solution onto doctor-bladed TiO2-films. In the
resulting hierarchical architectures, a first layer, that is, a cova-
lently attached porphyrin, was combined with a second layer,
that is, a non-covalently attached porphyrin or BODIPY. Such a
sequential coating enables, on one hand, optimizing the DSSC
performance through complementing the absorption of a first
layer dye by a second layer dye and on the other hand a high
degree of versatility through the reversible H-bonding attach-
ment.
Figure 1. Schematic representation of the complementary hydrogen bond-
ing in the Hamilton receptor–cyanuric acid binding motif.
hybrids with a charge-transfer character[23] and even the at-
tachment of gold nanoparticles on polymeric surfaces.[24]
Our research group has taken the aforementioned as a start-
ing point and advanced it by utilizing the Hamilton receptor/
cyanuric acid binding motif to integrate tailored organic build-
ing blocks onto metal oxide nanoparticles.[25] Following such a
covalent/non-covalent strategy, reversible superstructures con-
sisting of gold and TiO2 nanoparticles were held together by
H-bonding.[26] Through the controlled surface immobilization Results and Discussion
of a Hamilton receptor featuring carboxylic acids our group
Synthesis and complexation
also succeeded in the non-covalent orthogonal assembly of a
porphyrin–cyanurate onto TiO2 surfaces. The latter is not only
redox but also photoactive and combines the benefits of the
first layer functionalization based on covalent surface attach-
ment with the control over the reversibility of the second layer
functionalization based on H-bonding.[27]
We opted for a (metallo)porphyrin as first-layer dye in the
DSSCs. As such, the (metallo)porphyrin should feature a suit-
able TiO2 anchor, on one hand, and a recognition motif for the
post-functionalization by means of H-bonding, on the other
hand. Therefore, the synthesis of a trans-AB2C (metallo)por-
Scheme 1. Schematic representation of the step-by-step functionalization of doctor-bladed TiO2 films with dye 1 and dye 2 by means of covalent and non-co-
valent methodologies, respectively.
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5042 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
Scheme 2. Synthesis of two different trans-AB2C porphyrins featuring anchor groups for TiO2 surfaces: a) 0.9 equiv. BF3 OEt2, 2.8 equiv. DDQ, CHCl3, 2 h;
b) 0.4 equiv. I2, 1.5 equiv. p-chloranil, dichloromethane (DCM), 5 min; c) 1.5 equiv. ZnOAc, THF, 2 h; and d) 3 equiv. TBAF, THF, 3 h.
phyrin carrying either a carboxylic acid ester (R1) or an iso-
phthalic acid ester derivative (R2) as protected anchors, was
conducted (Scheme 2). The (metallo)porphyrin synthesis was
carried out in two different ways. For method A, the condensa-
tion of mesityl dipyrromethane (1), 4-[2-(trimethylsilyl)ethinyl]-
benzaldehyde (2) and benzaldehyde derivative (3) was initiated
by addition of BF3·(OEt)2; after stirring for 2 h the reaction mix-
ture was treated with DDQ (2,3-dichloro-5,6-dicyano-1,4-benzo-
quinone). In this standard wet chemical approach, the desired
trans-configured porphyrin was isolated after a single-column
chromatographic step in yields of 24% (4a) and 14% (4b), re-
spectively.
porphyrins in a Sonogashira cross-coupling reaction. In a stan-
dard approach, iodo-Hamilton receptor 7 was coupled to
ethinyl (metallo)porphyrins 6a or 6b, catalyzed by Pd(PPh3)4,
PPh3, and NEt3 under inert conditions. The Hamilton receptor-
functionalized (metallo)porphyrins were successfully formed in
up to 57% yield (9a) after stirring at 808C for approximately
3 days. For the preparation of the cyanuric acid-modified (met-
allo)porphyrin 10a, microwave syntheses were explored. Ethin-
yl (metallo)porphyrin 6a and 8 were treated with catalytic
amounts of Pd(PPh3)2Cl2, CuI, and PPh3 (Scheme 3). After de-
gassing the mixture, it was heated to 1508C under microwave
irradiation. A closer look at the reaction monitoring indicated
that an almost complete conversion set in after 32 min. Cyanu-
ric acid-modified (metallo)porphyrin 10a was obtained in 40%
yield after column chromatography. Once again, the prepara-
tion of 10b was carried out under the previously described
standard conditions in a yield of 46%. Finally, deprotection of
the esters under basic conditions yielded the (metallo)porphy-
rin building blocks 9d and 10c for the first-layer dye. Depro-
tection of the monocarboxylate ester 9a to the corresponding
carboxylic acid, however, was not possible. The successful de-
protection of the esters was verified by means of high-resolu-
tion mass spectrometry and NMR spectroscopy. As second
layer dye, BODIPYs as well as (metallo)porphyrins were chosen.
First, the asymmetrical iodo-BODIPY (11) was treated with the
ethinyl Hamilton receptor 12 in a Sonogashira reaction to
obtain the Hamilton receptor-functionalized BODIPY 15 after a
single-column chromatographic step (Scheme 4). Cyanuric
acid-functionalized BODIPY 16 was also obtained in a Sonoga-
Procedure B was based on utilizing microwave irradiation. In
particular, iodine was added to a mixture of dipyrromethane 1
and the aldehyde building blocks 2 and 3 (Scheme 2). The mix-
ture was heated to 408C for 5 min using microwave irradiation.
This was followed by oxidation with p-chloranil. The desired
trans-AB2C porphyrin was obtained after a single column chro-
matographic step in yields of 22% (4a) and 21% (4b). Overall,
the microwave-based procedure B resulted in significantly
shorter reaction times and easier purification. After subsequent
metalation with zinc acetate the resulting (metallo)porphyrins
were subjected to desilyation using a 1m solution of TBAF (tet-
rabutylammonium fluoride) in THF.
Deprotected ethinyl (metallo)porphyrins 6a and 6b were
obtained in good yields. Next, an iodine-substituted Hamilton
receptor (7) and an iodine-substituted cyanurate (8) were pre-
pared according to literature procedures.[20a,28] These building
blocks were then coupled to the asymmetric ethinyl (metallo)-
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5043 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
Scheme 3. Synthetic pathway to different AB2C porphyrins carrying anchor groups for metal oxide surfaces and a Hamilton receptor or cyanuric acid group
for further post-functionalization via hydrogen bonding: a) Pd2(dba)3 (dba = dibenzylidenacetone), AsPh3, CuI, THF/NEt3, 808C, 3 days; b) Pd(PPh3)4, PPh3, CuI,
THF/NEt3, 808C, 3 days; c) Pd(PPh3)2Cl2, PPh3, CuI, THF/NEt3, 1508C, 32 min, and d) NaOH, THF, 24 h.
shira reaction of 11 and ethinyl cyanurate (13). Several chroma-
tographic steps including reverse-phase HPLC were, however,
necessary to purify the crude reaction mixture. Finally, a new
microwave-assisted procedure for preparation of the literature-
known cyanuric acid (metallo)porphyrin 17 was developed.
chiometry. To this end, the product of the chemical shift varia-
tion of the Hamilton receptor amide proton NH1 (red dot) and
the molar fraction of the Hamilton receptor (Host) were plot-
ted against the molar fraction of the host (Figures 2 and S25,
S26).
Ethinyl (metallo)porphyrin 14 and
8
were treated with
In all cases, a maximum at a mole fraction of 0.5 verified the
1:1 binding stoichiometry of the Hamilton receptor/cyanuric
acid couples.
Pd(PPh3)2Cl2 and CuI at 1508C to obtain 17 in 39% yield. This
reduced the reaction time from 3 days to 32 min. To validate
the complementary H-bonding capabilities, the specific bind-
ings of the Hamilton receptor/cyanuric acid couples 9a·17 and
Physicochemical investigations
1
10a·15 were investigated by means of H-NMR titrations. The
change in the chemical shifts of the amide protons (red dot
and purple dot) of the Hamilton receptor prompts to the suc-
cessful complexation of cyanuric acid through H-bonding (Fig-
ures 2 and S25, S26). The binding constants were obtained by
plotting the changes in chemical shift of the Hamilton receptor
amide protons versus the concentration of cyanuric acid.
These plots were then analyzed with the help of the calculator
bindfit from supramolecular.org to obtain the corresponding
binding constants. For 9a·17 a binding constant of 2.98”
105 LmolÀ1 and for 15·10c a binding constant of 1.74”
104 LmolÀ1 was determined. Furthermore, a Jobs plot analysis
of the titration data was employed to verify the binding stoi-
The absorption spectra of 9d, 16, and 9d·16 taken in chloro-
form are shown in Figure 3. For example, the absorption spec-
trum of 9d exhibits signatures typical for (metallo)porphyrins,
that is, a strong Soret-band absorption at 427 nm and weaker
Q-band absorptions at 554 and 599 nm. In contrast, the ab-
sorption spectrum of 16 is dominated by a high energy p!p*
transitions at 393 nm and a low energy spin-allowed p!p*
transition at 550 nm.[29] A comparison of the absorption spectra
taken for 9d·16 with those of 9d and 16 revealed only minor
changes. In particular, the Soret-band absorption of 9d broad-
ens and the low-energy p!p* transition of 16 red-shifts from
543 to 549 nm. A closer look at Figure 3b reveals that all the
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5044 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
Scheme 4. Synthesis of several dye molecules with Hamilton receptor or cyanuric acid groups for the supramolecular functionalization of the first dye layers:
a) Pd(PPh3)2Cl2, PPh3, CuI, THF/NEt3, RT, overnight; and b) Pd(PPh3)2Cl2, PPh3, CuI, THF/NET3, 1508C, 32 min.
absorption features further broaden after adsorption on the
surface of transparent TiO2 films. In solution, BODIPY-related
features dominate in the 500–600 nm range of the (metallo)-
porphyrin-based Q-band signatures in 9d·16. In the TiO2 films,
BODIPY-related transitions at 523 nm are only clearly discern-
able in 9d·16, but they do not dominate.
9d, 16, and 9d·16 were also studied using steady-state fluo-
rescence spectroscopy at room temperature (Figure S29). 9d
exhibits the characteristic fluorescence of (metallo)porphyrins,
namely maxima at 610 and 656 nm. For 16, we note an intense
fluorescence, which maximizes at 582 nm. Selective 506 nm
photoexcitation of BODIPY in 9d·16 shows exclusively the fluo-
rescence that relate to BODIPY (Figure S29a). These results sug-
gest that BODIPY in 9d·16 has little or no effect on the fluores-
cence of 9d. The overlapping fluorescence of 9d and 16
render, however, an unambiguous analysis rather difficult.
Analogous observations have been made with 10c, 15, and
10c·15 in the steady-state absorption and fluorescence mea-
surements (Figure S30).
trodes in the presence of the electrolyte. The experiments
were conducted for 9d, 16, and 9d·16 as well as for 10c, 15,
and 10c·15, in a solvent mixture of THF/acetonitrile (ACN) (1:1
v/v). Time-dependent excited-state features and kinetics there-
of were evaluated by means of multiwavelength analysis to
corroborate the mechanism, by which charges are injected
into TiO2 (vide infra).
Those experiments, which were conducted in solution, serve
as reference. In the solvent mixture, 430 nm photoexcitation of
either 9d or 10c commences with the formation of transient
absorption spectra, which include 457, 604, and 657 nm
minima and 461, 580, 635, and 1278 nm maxima, which domi-
nate the fs-TAS dynamics (Figures S31 and S34 in the Support-
ing Information). The 457 nm minimum is due to ground-state
bleaching of the Q-band absorption whereas the 603 and
657 nm minima are due to stimulated fluorescence. These
transform into triplet-excited state features located at 450 and
840 nm within 1380 ps.
In solution, no appreciable changes are observed neither in
9d·16 nor in 10c·15 (Figures S33 and S36). Similar to the
steady-state fluorescence experiments, we failed to gather evi-
dence for any energy transfer from the (metallo)porphyrins to
To elucidate the processes taking place upon photoexcita-
tion, transient absorption spectroscopy (TAS) measurements
were performed both in solution and on semiconducting elec-
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5045 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
Figure 3. a) Steady-state absorption spectra of 16, 9d, and 9d·16 (1:1 ratio)
in chloroform. b) Steady-state absorption spectra of TiO2·9d, TiO2·10c,
TiO2·9d·16, and TiO2·15·10c with electrolyte.
Figure 2. a) 1H NMR titration of Hamilton receptor 15 with several equiva-
lents of cyanuric acid porphyrin 10c (CDCl3/CD2Cl2, 400 MHz). b) Jobs plot of
the NMR titration data of complex 15·10c.
oxidation of, for example, 9d (Figure S44). Based on the multi-
wavelength analysis, the charge injection kinetics are best de-
scribed by three components.
BODIPY. Instead, the differential absorption spectra look exactly
like those recorded for (metallo)porphyrins 9d and 10c. Impor-
tantly, excitation of either 15 or 16 at 430 nm results in differ-
ential absorption spectra typical for BODIPYs, that is, minima
at 532 and 610 nm (Figures S32 and S35). In essence, the lack
of BODIPY-related features upon 430 nm photoexcitation of
9d·16 or 10c·15 is not a result of selective (metallo)porphyrin
excitation.
In all cases, an ultrafast component, deconvolution of which
is limited by the instrument response, is implied. The failure to
detect this component, when 610 nm is used as excitation
wavelength (Figures S41 and S43) rather than 430 nm, sug-
gests that the latter is likely to be charge injection from the
second singlet excited state (S2). Taking into consideration that
the transient absorption spectra upon 610 nm excitation fea-
tures contributions from the first singlet excited state (S1) and
the one-electron oxidized form of the (metallo)porphyrin, it is
fair to assume that upon 430 nm photoexcitation the majority
of charge injection evolves from the S2 state. The ultrafast
charge injection stemming from S2 is followed by two slower
components. The faster of them, that is, the second one, takes
around 5.6 and 2.47 ps in the cases of 9d and 10c, respective-
ly. As the same lifetimes are also discernible upon 610 nm exci-
tation it is very likely that this process is a charge injection
from the S1 state. The slower of them, that is, the third one
The picture changes once 9d, 16, and 9d·16 as well as 10c,
15, and 10c·15 are deposited on 9 mm-thick TiO2 films in the
presence of an electrolyte: TiO2·9d·16 and TiO2·10c·15. Upon
430 nm excitation of 9d and 10c, they both show the instan-
taneous formation of the one-electron oxidized form of the
(metallo)porphyrin (Figures S36 and S37). Most notable is the
lack of stimulated fluorescence, on one hand, and 538 and
850 nm maxima, on the other hand. The maxima are in sound
agreement with those detected upon spectro-electrochemical
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5046 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
takes 227 ps for 9d and 308 ps for 10c. Based on the literature
and also considering the red-shifts of the transient absorption
features in the 660 to 685 nm region, the third component is
assigned with confidence to charges that are injected into
non-mobile surface states, which undergo thermal relaxation
into the conduction band.[30]
but are rather related to a charge transfer. To gather a better
understanding, we took a closer look at the energetics (Fig-
ure S46). Three scenarios are taken into consideration. First, a
direct charge transfer from BODIPY to TiO2; second, a charge
transfer in terms of electrons from the (metallo)porphyrin to
BODIPY; third a charge transfer in terms of electrons from
BODIPY to the (metallo)porphyrin. Based on a previous report
by our research group,[27] we rule out anchoring of the cyanu-
rates to TiO2 and direct charge injection from BODIPY to
TiO2.[27] If a charge transfer from the (metallo)porphyrin to
BODIPY would take place, such mechanism should be opera-
tive in solution studies as well. However, we could not gather
any experimental evidence for this pathway. As such, we rule
out any charge transfer from the (metallo)porphyrin to BODIPY.
Upon closer inspection of the transient absorption spectra, sig-
natures of the one-electron oxidized form of BODIPY are noted
in the form of a broadened bleaching and an intensification of
the 700 nm features. Both of them appear right after the con-
clusion of charge injection from the (metallo)porphyrin to TiO2
(Figure S45).[31] Full deconvolution of the latter is, however, dif-
ficult due to the overlap with the signatures of the one-elec-
tron oxidized form of the (metallo)porphyrin. In summary, we
postulate for TiO2·9d·16 a charge transfer from BODIPY 16 to
(metallo)porphyrin 9d following the initial charge injection
from (metallo)porphyrin 9d into TiO2 as the modus operandi.
15·10c shows no pronounced changes relative to 10c film.
Changes were noted in the transient absorption spectra of
TiO2·9d·16 (Figure 4). In addition to the one-electron oxidized
form of the (metallo)porphyrin 9d broadening of the ground
state bleaching is observed at around 565 nm. The latter is in
sound agreement with the BODIPY signatures. Considering the
lack of comparable observations in the solution studies, the
evolving features are not a result of energy transfer processes
Application of orthogonally assembled bilayers in DSSC
devices
Following earlier work,[27] the photovoltaic characteristics were
investigated using a standard sandwich-type DSSC configura-
tion comprised of a 9 mm thick TiO2 layer as photoanode (Fig-
ures S48 and S49), a IÀ/I3À-based electrolyte, and a platinum
counter electrode— for more information please refer to the
Supporting Information. In contrast to previous studies by our
group,[27] we opted for a reverse ordering of the bilayer struc-
ture and used (metallo)porphyrins with different anchors to
bind to the semiconducting TiO2 (Scheme 1).
First, devices featuring monolayers of 10c (TiO2·10c) and bi-
layers of 10c·15 (TiO2·10c·15) were prepared and character-
ized. Current–voltage (J–V) assays were performed enabling
first insights into the effect of post-functionalization on the
device figures-of-merit, that is, efficiency (h), fill factor (FF),
short-circuit current (JSC), and open-circuit voltage (VOC)
(Figure 5, top and Table 1). Importantly, h increased from
0.14% for TiO2·10c to 0.16% for TiO2·10c·15 due to higher
JSCs, namely 0.42 versus 0.46 mAcmÀ2. We rationalize our J–V
observation based on either a direct charge injection from
BODIPY into TiO2 or a charge transfer from BODIPY to the
(metallo)porphyrin (vide infra). To proof or disproof these two
different pathways, IPCE (incident photon-to-current efficiency)
measurements were performed. By virtue of the lack of addi-
tional features in the IPCE spectra seen for TiO2·10c·15, a
direct charge injection from BODIPY into TiO2 is very unlikely
(Figure 5, bottom left). Consequently, the higher hs are in line
with the TAS measurements (vide supra) attributed to the se-
Figure 4. a) Differential absorption spectra with the depicted time delays
and b) 3D representation of differential absorption spectra (visible and near-
infrared) obtained upon femtosecond flash photolysis (430 nm) of
TiO2·9d·16 with electrolyte with several time delays between 0 and 5500 ps
at room temperature. c) Time absorption profiles of the spectra at 564, 665,
and 837 nm.
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5047 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
Figure 5. J–V assays (top) and IPCE measurements (bottom) of TiO2 devices: TiO2·10c and TiO2·10c·15 (left), TiO2·9a, TiO2·9a·16, and TiO2·9a·17 (middle), and
TiO2·9d, TiO2·9d·16, and TiO2·9d·17 (right).
of a cyanurate. Additionally, the post-functionalization includes
Table 1. Summary of the devices’ figures of merits.
BODIPY (16) as well as (metallo)porphyrin 17. When comparing
the figures-of-merit for TiO2·9a with TiO2·10c, a higher efficien-
cy of 0.2% for TiO2·9a rather than 0.14% for TiO2·10c stem
from different JSCs: 0.62 versus 0.42 mAcmÀ2. The post-func-
tionalization with 16 and 17 resulted, however, in lower hs of
0.1% and 0.19% for 9a·16 and 9a·17, respectively. Lower hs
are linked to lesser FFs, Vocs, and JSCs. The latter goes hand-in-
hand with weaker contributions in the Soret-band region at
around 430 nm in the IPCE measurements upon post-function-
alization. A closer inspection of the devices following the
measurements showed that the red-colored electrolyte turned
bright green as a consequence of post-functionalization. This is
taken as clear evidence for desorption from the semiconduct-
ing TiO2. From the aforementioned we conclude that changing
to a Hamilton receptor rather than a cyanurate affects h, but a
monocarboxylic ester anchor leads to quick and quantitative
desorption when post-functionalized.
Dye combination
Efficiency [%]
FF [%]
JSC [mAcmÀ2
]
VOC [V]
10c
15·10c
9a
9a·16
9a·17
9d
9d·16
9d·17
0.14
0.16
0.20
0.10
0.19
0.54
0.61
0.77
65
68
62
60
60
75
74
76
0.42
0.47
0.62
0.38
0.60
1.25
1.43
1.50
0.51
0.50
0.52
0.47
0.51
0.62
0.57
0.61
quence of charge injection from (metallo)porphyrin 10c into
TiO2 followed by charge transfer from BODIPY 15 to (metallo)-
porphyrin 10c. Such a sequence assists in rationalizing the
rather moderate increase in h and JSC. A moderate increase in
JSC suggests one or more of the following scenarios as bottle-
necks. First, the dicarboxylate anchor might potentially slow
down or even hinder charge injection. Second, the positioning
of the Hamilton receptor might shift the electron density away
from the semiconducting surface. Third, the use of BODIPY in
the post-functionalization might be inefficient in terms of
charge injection.
To remedy this shortcoming, TiO2·9d was prepared and
tested in combination with TiO2·16 and TiO2·17. 9d is still a
Hamilton receptor, but features unlike 9a a dicarboxylate
anchor. The figures-of-merit for TiO2·9d taken from the J–V
measurements show a Voc of 0.62 V compared to 0.52 V deter-
mined for 9b. This is attributed to a better surface coverage.
Of even greater importance is, however, a JSC of 1.25 mAcmÀ2
for 9d, which is with 0.62 mAcmÀ2 double than that seen for
TiO2·9a. Accordingly, h for TiO2·9d is 0.54%; please compare
that to a h of 0.2% for TiO2·9a. The main reason behind the JSC
changes is explained by the IPCE assays. Here, a broadening
and an increase regarding contributions in the Soret- and Q-
To investigate the aforementioned series of devices, featur-
ing monolayers of 9a (TiO2·9a) or bilayers of 9a·16 and 9a·17
(TiO2·9a·16 and TiO2·9a·17), were prepared and characterized
by means of J–V and IPCE assays. To this end, 9a was synthe-
tized with only a monocarboxylic ester anchor rather than a di-
carboxylate anchor like in 10c and a Hamilton receptor instead
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5048 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
band regions are noted for TiO2·9d relative to the observations
made for TiO2·9a (Figure 5 middle and right).
transfer process from 16 to 9d following the initial interfacial
charge injection from 9d into TiO2.
Such a broadening is associated with stronger interactions
with TiO2. In turn, charge injection is facilitated and JSC is in-
creased. Next, we turned to devices based on bilayers of 9d·16
and 9d·17 and their characterization. Most important is the
fact that h further increased from 0.54% for TiO2·9d to 0.61%
and 0.77% for TiO2·9d·16 and TiO2·9d·17, respectively. The
(metallo)porphyrin post-functionalization, that is, TiO2·9d·17,
outperforms the BODIPY post-functionalization, that is,
TiO2·9d·16, by 26%. The main cause is again an enhanced JSC,
most likely due to a charge transfer from 16 and 17 to 9d.
To summarize, all factors, that is, the use of a Hamilton re-
ceptor and its positioning, the post-functionalization as well as
the respective anchor are critical in terms of maximizing the
overall efficiencies. This highlights the necessity of a stable and
robust anchor as well as a tailored self-assembly to generate
high-performing devices. Building on the knowledge gathered
through this work, higher overall efficiencies could be targeted
through careful synthetic modifications of the first- and second
layer dyes. In particular, a direct attachment of the supramolec-
ular recognition motif and the carboxylate anchoring groups
to the porphyrin moiety, as well as addition of strong electron
donor groups, could lead to significantly higher efficiencies.
Experimental Section
Full experimental details regarding the synthesis and characteriza-
tion of all organic building blocks, electrochemical characterization,
optical spectroscopy, and the preparation of DSSC devices are
given in the Supporting Information.
Acknowledgements
T.L. and A.Z. contributed equally to this work. We gratefully
thank the Cluster of Excellence “Engineering of Advanced Ma-
terials”(EAM) funded by the German Research Council (DFG),
the Bavarian collaborative research project “Solar Technologies
go Hybrid (SolTech)”, the Graduate School “Molecular Science”
(GSMS) and the Graduate School “Advanced Materials and Pro-
cesses” (GSAMP) for financial support. Additionally, we ac-
knowledge the support of Dr. Fabian Lodermeyer. Open access
funding enabled and organized by Projekt DEAL.
Conflict of interest
Conclusions
The authors declare no conflict of interest.
In this work, we prepared a library of AB2C (metallo)porphyrins
featuring a stable anchoring group, on one hand, as well as a
versatile supramolecular recognition motif, on the other hand.
Additionally, we complemented the supramolecular recogni-
tion motif by designing corresponding porphyrins and BODI-
PYs. In a step-by-step fashion, we succeeded in binding the
anchor to TiO2 nanoparticles and in employing a second dye
layer via hydrogen bonding. In short, our coating strategy syn-
ergizes the benefits of a strong covalent anchoring of the first
dye layer, that is, a (metallo)porphyrin, with the versatility of a
non-covalently attaching a second dye layer, that is, either a
(metallo)porphyrin or a BODIPY. Our investigations were, how-
ever, not only limited to solution studies but were matched
with corresponding DSSC-based assays. Our studies demon-
strated the importance of integrating the Hamilton receptor
within the first dye layer and using a dicarboxylic acid anchor
to prevent desorption from the TiO2 nanoparticles. To this end,
highest DSSC efficiencies were noted when employing Hamil-
ton receptor porphyrin 9d and TiO2 to afford TiO2·9d. A non-
covalent post-functionalization with either cyanuric acid
BODIPY 16 or cyanuric acid (metallo)porphyrin 17 led to an in-
crease in DSSC efficiency. TiO2·9d·17, which turned out to be
most effective bilayer, improved the DSSC efficiency by 43%. A
comprehensive understanding of the underlying mechanics
came from transient absorption spectroscopy (TAS) measure-
ments with the (metallo)porphyrin-BODIPY couples 10c·15 and
9d·16. It is only in TiO2·9d·16 that the spectroscopic features
of the one-electron oxidized form of the (metallo)porphyrin 9d
and those of the one-electron oxidized BODIPY 16 were ob-
served. This suggests for TiO2·9d·16 an intramolecular charge-
Keywords: dyes · hierarchical · sequential functionalization ·
solar cells · titania
[1] a) J. Kao, K. Thorkelsson, P. Bai, B. J. Rancatore, T. Xu, Chem. Soc. Rev.
2013, 42, 2654–2678; b) C. Sanchez, B. Juliµn, P. Belleville, M. Popall, J.
Mater. Chem. 2005, 15, 3559–3592; c) J. Yang, C.-H. Lee, H.-J. Ko, J.-S.
Suh, H.-G. Yoon, K. Lee, Y.-M. Huh, S. Haam, Angew. Chem. Int. Ed. 2007,
46, 8836–8839; Angew. Chem. 2007, 119, 8992–8995; d) D. M. Guldi,
G. M. A. Rahman, N. Jux, D. Balbinot, U. Hartnagel, N. Tagmatarchis, M.
Prato, J. Am. Chem. Soc. 2005, 127, 9830–9838.
[2] a) Q. M. Kainz, O. Reiser, Acc. Chem. Res. 2014, 47, 667–677; b) T. Luchs,
P. Lorenz, A. Hirsch, ChemPhotoChem 2020, 4, 52–58; c) R. B. N. Baig,
R. S. Varma, Chem. Commun. 2013, 49, 752–770.
[3] a) Y. Kim, R. C. Johnson, J. T. Hupp, Nano Lett. 2001, 1, 165–167; b) L.
Qin, G. Zeng, C. Lai, D. Huang, P. Xu, C. Zhang, M. Cheng, X. Liu, S. Liu,
B. Li, H. Yi, Coord. Chem. Rev. 2018, 359, 1–31; c) L. Zeininger, S. Nagel-
berg, K. S. Harvey, S. Savagatrup, M. B. Herbert, K. Yoshinaga, J. A. Capo-
bianco, M. Kolle, T. M. Swager, ACS Cent. Sci. 2019, 5, 789–795.
[4] a) C. K. Kim, P. Ghosh, C. Pagliuca, Z.-J. Zhu, S. Menichetti, V. M. Rotello,
J. Am. Chem. Soc. 2009, 131, 1360–1361; b) S. S. Agasti, A. Chompoosor,
C. C. You, P. Ghosh, C. K. Kim, V. M. Rotello, J. Am. Chem. Soc. 2009, 131,
5728–5729.
[5] a) M. Sarcletti, D. Vivod, T. Luchs, T. Rejek, L. Portilla, L. Müller, H. Die-
trich, A. Hirsch, D. Zahn, M. Halik, Adv. Funct. Mater. 2019, 29, 1805742;
b) T. Luchs, M. Sarcletti, L. Zeininger, L. Portilla, C. Fischer, S. Harder, M.
Halik, A. Hirsch, Chem. Eur. J. 2018, 24, 13589–13595; c) X. Li, H. Lu, Y.
Zhang, F. He, Chem. Eng. J. 2017, 316, 893–902.
[6] B. O’Regan, M. Grätzel, Nature 1991, 353, 737–740.
[7] a) M. Halik, A. Hirsch, Adv. Mater. 2011, 23, 2689–2695; b) L. Portilla, M.
Halik, ACS Appl. Mater. Interfaces 2014, 6, 5977–5982; c) S. P. Pujari, L.
Scheres, A. T. Marcelis, H. Zuilhof, Angew. Chem. Int. Ed. 2014, 53, 6322–
6356; Angew. Chem. 2014, 126, 6438–6474; d) J. C. Love, L. A. Estroff,
J. K. Kriebel, R. G. Nuzzo, G. M. Whitesides, Chem. Rev. 2005, 105, 1103–
1170; e) P. J. Hotchkiss, S. C. Jones, S. A. Paniagua, A. Sharma, B. Kippe-
len, N. R. Armstrong, S. R. Marder, Acc. Chem. Res. 2012, 45, 337–346.
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5049 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202004928
Chemistry—A European Journal
[8] a) A. Jedaa, M. Salinas, C. M. Jäger, T. Clark, A. Ebel, A. Hirsch, M. Halik,
Appl. Phys. Lett. 2012, 100, 063302; b) M. Burkhardt, A. Jedaa, M. Novak,
A. Ebel, K. Voitchovsky, F. Stellacci, A. Hirsch, M. Halik, Adv. Mater. 2010,
22, 2525–2528.
J.-M. Lehn, Chem. Eur. J. 2002, 8, 1227–1244; d) A. Ciesielski, G. Schaeff-
er, A. Petitjean, J.-M. Lehn, P. Samorì, Angew. Chem. Int. Ed. 2009, 48,
2039–2043; Angew. Chem. 2009, 121, 2073–2077; e) G. Schaeffer, O.
Fuhr, D. Fenske, J.-M. Lehn, Chem. Eur. J. 2014, 20, 179–186.
[19] C. Dethlefs, J. Eckelmann, H. Kobarg, T. Weyrich, S. Brammer, C. Näther,
U. Lüning, Eur. J. Org. Chem. 2011, 2066–2074.
[20] a) F. Wessendorf, A. Hirsch, Tetrahedron 2008, 64, 11480–11489; b) F.
Wessendorf, B. Grimm, D. M. Guldi, A. Hirsch, J. Am. Chem. Soc. 2010,
132, 10786–10795.
[21] A. Tron, P. J. Thornton, B. Kauffmann, J. H. R. Tucker, N. D. McClenaghan,
Supramol. Chem. 2016, 28, 733–741.
[22] S. Bosch, L. Zeininger, F. Hauke, A. Hirsch, Chem. Eur. J. 2014, 20, 2537–
2541.
[23] B. Grimm, J. Schornbaum, H. Jasch, O. Trukhina, F. Wessendorf, A.
Hirsch, T. Torres, D. M. Guldi, Proc. Natl. Acad. Sci. USA 2012, 109,
15565–15571.
[24] a) W. H. Binder, C. Kluger, M. Josipovic, C. J. Straif, G. Friedbacher, Macro-
molecules 2006, 39, 8092–8101; b) W. H. Binder, C. Kluger, C. J. Straif, G.
Friedbacher, Macromolecules 2005, 38, 9405–9410.
[25] L. Zeininger, M. Klaumunzer, W. Peukert, A. Hirsch, Int. J. Mol. Sci. 2015,
16, 8186–8200.
[26] M. Ali, D. H. Hasençhrl, L. Zeininger, A. R. M. Müllner, H. Peterlik, A.
Hirsch, Helv. Chim. Acta 2019, 102, e1900015.
[27] L. Zeininger, F. Lodermeyer, R. D. Costa, D. M. Guldi, A. Hirsch, Chem.
Commun. 2016, 52, 8842–8845.
[28] <M. J. Plater, J. P. Sinclair, S. Aiken, T. Gelbrich, M. B. Hursthouse, Tetra-
hedron 2004, 60, 6385–6394.
[29] T. Lazarides, G. Charalambidis, A. Vuillamy, M. RØglier, E. Klontzas, G.
Froudakis, S. Kuhri, D. M. Guldi, A. G. Coutsolelos, Inorg. Chem. 2011, 50,
8926–8936.
[30] J. Jiang, J. R. Swierk, K. L. Materna, S. Hedstrçm, S. H. Lee, R. H. Crabtree,
C. A. Schmuttenmaer, V. S. Batista, G. W. Brudvig, J. Phys. Chem. C 2016,
120, 28971–28982.
[9] M. Grätzel, Acc. Chem. Res. 2009, 42, 1788–1798.
[10] G. E. Fryxell, P. C. Rieke, L. L. Wood, M. H. Engelhard, R. E. Williford, G. L.
Graff, A. A. Campbell, R. J. Wiacek, L. Lee, A. Halverson, Langmuir 1996,
12, 5064–5075.
[11] a) S. Mischler, S. Guerra, R. Deschenaux, Chem. Commun. 2012, 48,
2183–2185; b) S. H. Etschel, L. Portilla, J. Kirschner, M. Drost, F. Tu, H.
Marbach, R. R. Tykwinski, M. Halik, Angew. Chem. Int. Ed. 2015, 54,
9235–9238; Angew. Chem. 2015, 127, 9367–9370; c) C. Haensch, S.
Hoeppener, U. S. Schubert, Chem. Soc. Rev. 2010, 39, 2323–2334.
[12] a) S. Mann, W. Shenton, M. Li, S. Connolly, D. Fitzmaurice, Adv. Mater.
2000, 12, 147–150; b) A. B. Descalzo, R. Martínez-MµÇez, F. Sancenón, K.
Hoffmann, K. Rurack, Angew. Chem. Int. Ed. 2006, 45, 5924–5948;
Angew. Chem. 2006, 118, 6068–6093.
[13] a) L. Zeininger, S. Petzi, J. Schçnamsgruber, L. Portilla, M. Halik, A.
Hirsch, Chem. Eur. J. 2015, 21, 14030–14035; b) L. Zeininger, L. M. S.
Stiegler, L. Portilla, M. Halik, A. Hirsch, ChemistryOpen 2018, 7, 282–287;
c) L. M. S. Stiegler, T. Luchs, A. Hirsch, Chem. Eur. J. 2020, 26, 8483–
8498; d) T. Pellegrino, L. Manna, S. Kudera, T. Liedl, D. Koktysh, A. L.
Rogach, S. Keller, J. Rädler, G. Natile, W. J. Parak, Nano Lett. 2004, 4,
703–707.
[14] a) S. Klein, T. Luchs, A. Leng, L. V. R. Distel, W. Neuhuber, A. Hirsch, Bioen-
gineering 2020, 7, 126; b) S. Klein, L. M. S. Stiegler, C. Harreiss, L. V. R.
Distel, W. Neuhuber, E. Spiecker, A. Hirsch, C. Kryschi, ACS Appl. Bio
Mater. 2018, 1, 2002–2011.
[15] G. Decher, Science 1997, 277, 1232–1237.
[16] a) A. Burger, R. D. Costa, V. Lobaz, W. Peukert, D. M. Guldi, A. Hirsch,
Chem. Eur. J. 2015, 21, 5041–5054; b) A. Burger, A. Kunzmann, R. D.
Costa, R. Srikantharajah, W. Peukert, D. M. Guldi, A. Hirsch, Chem. Eur. J.
2018, 24, 7896–7905.
[17] a) D. J. Broer, C. M. W. Bastiaansen, M. G. Debije, A. P. H. J. Schenning,
Angew. Chem. Int. Ed. 2012, 51, 7102–7109; Angew. Chem. 2012, 124,
7210–7218; b) A. Ciesielski, C.-A. Palma, M. Bonini, P. Samorì, Adv. Mater.
2010, 22, 3506–3520; c) V. Georgakilas, J. N. Tiwari, K. C. Kemp, J. A.
Perman, A. B. Bourlinos, K. S. Kim, R. Zboril, Chem. Rev. 2016, 116, 5464–
5519.
[18] a) S. K. Chang, A. D. Hamilton, J. Am. Chem. Soc. 1988, 110, 1318–1319;
b) S. K. Chang, D. Van Engen, E. Fan, A. D. Hamilton, J. Am. Chem. Soc.
1991, 113, 7640–7645; c) V. Berl, M. Schmutz, M. J. Krische, R. G. Khoury,
[31] S. Aute, P. Maity, A. Das, H. N. Ghosh, New J. Chem. 2017, 41, 5215–
5224.
Manuscript received: November 12, 2020
Revised manuscript received: January 7, 2021
Accepted manuscript online: January 11, 2021
Version of record online: February 15, 2021
Chem. Eur. J. 2021, 27, 5041 –5050
www.chemeurj.org
5050 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH