Molecular Solar Thermal Batteries through Combination of Magnetic Nanoparticle Catalysts and Tailored Norbornadiene Photoswitches

Article abstract of DOI:10.1002/chem.202005427

Cobalt catalysts are immobilized on the surface of iron oxide nanoparticles for the preparation of highly active quasi-homogeneous catalysts toward an efficient release of photochemically stored energy in norbornadiene-based photoswitches. The facile sepa

Full text of DOI:10.1002/chem.202005427

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
&
Energy Storage |Hot Paper|
Molecular Solar Thermal Batteries through Combination of
Magnetic Nanoparticle Catalysts and Tailored Norbornadiene
Photoswitches
Patrick Lorenz+, Tobias Luchs+, and Andreas Hirsch*[a]
Abstract: Cobalt catalysts are immobilized on the surface of
iron oxide nanoparticles for the preparation of highly active
quasi-homogeneous catalysts toward an efficient release of
photochemically stored energy in norbornadiene-based pho-
toswitches. The facile separation of the iron oxide nanoparti-
cles through exploitation of the intrinsic magnetic properties
of this material enables efficient cyclization of energy stor-
age and release. Through the transition from cobalt (II) sal-
phen to cobalt porphyrins, a 22.6-fold increase in the catalyt-
ic efficiency of the QC-NBD back-conversion is achieved,
with an initial TOF of up to 3.64 sÀ1 and excellent TON of
over 3305. In addition, a series of novel “push–pull” function-
alized norbornadiene derivatives is prepared, featuring excel-
lent absorption properties with maxima up to 366 nm, quan-
tum yields around 70%, high energy storage capacities of
up to 98.0 kJmolÀ1, and outstanding thermal stability with
t1/2 (258C) over 100 days. Finally, the energy storage poten-
tial of these molecular solar thermal (MOST) systems is har-
nessed in a heat release experiment. This demonstrates the
potential of norbornadiene-based photoswitches in combi-
nation with efficient magnetic catalysts for the generation of
environmentally benign process heat.
Introduction
ture goods makes up a significant share of the overall industri-
al energy demand, and to a large extent, low to medium heat
is required.[3] Even though industrial process heat and domes-
tic heating account for a large portion of the total energy con-
sumption, they still rely heavily on fossil fuels. Although still in
their infancy, green solutions for heating applications such as
In search of renewable energy sources, the concept of harvest-
ing and storing solar energy in a unimolecular device has re-
cently received increasing attention.[1] Such molecular solar
thermal (MOST) energy storage systems have a distinct advant-
age over common solar-based energy systems, which is that
they are not affected by the mismatch between energy availa-
bility and energy demand. Within these systems, the harvested
sunlight is used to drive a photoisomerization from a low-
energy isomer to a metastable energy-rich counterpart. There-
by, the energy is stored in chemical bonds and later released
on demand as thermal energy (Scheme 1).[2] Conceivable appli-
cations for these solar thermal batteries include not only do-
mestic heating, but also the generation of industrial process
heat. The heat used directly in industrial processes to manufac-
[a] P. Lorenz,+ T. Luchs,+ Prof. Dr. A. Hirsch
Department of Chemistry and Pharmacy
Institute of Organic Chemistry
Friedrich-Alexander-Universität Erlangen-Nürnberg
Nikolaus-Fiebiger-Strasse 10, 91058 Erlangen (Germany)
E-mail: andreas.hirsch@fau.de
[+] These authors contributed equally to this work.
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.202005427.
ꢀ 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.
Scheme 1. Schematic representation of a solar thermal battery based on the
norbornadiene (NBD)/quadricyclane (QC) photoswitch.
Chem. Eur. J. 2021, 27, 4993 –5002
4993 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
solar thermal collectors and photovoltaic heating are under in-
tensive investigation and are already in application.[4] To bridge
the gap between energy demand and the availability of solar
energy, thermal energy storage is required. MOST systems are
ideal candidates to meet the high and continuous demand of
industrial process heat and domestic heating owing to their
lack of transformation losses, long storage time, and fast
energy release.
In this paper we provide the tools that allow this enormous
potential to be harnessed. On the one hand, we have designed
new and much more efficient magnetic catalysts. On the other
hand, we have synthesized a variety of novel NBD–QC inter-
conversion couples and performed detailed investigations re-
garding their MOST potential. By studying the kinetics of these
magnetic NP-based catalysts, for the first time we were able to
show that the second generation, based on cobalt porphyrins
as the active catalyst, is far more reactive than the previously
reported cobalt salphen [Fe3O4-Cat1], which is impressively
shown by a 22.6-fold increase in reactivity. This new catalyst
generation in combination with the outstanding photophysical
properties of the novel NBD–QC couples (lonset up to 450 nm
and Fꢀ70%) paves the way toward an efficient MOST system
utilizing the sunlight-driven isomerization of NBD to QC and a
magnetic NP-anchored quasi-homogenous catalyst. Finally, we
have investigated the heat release capacity of the nanoparti-
cle-catalyzed back-conversion of QC1 to NBD1 in a 1m solu-
tion, yielding a temperature increase to 39.58C in a glass vial
without insulation. This demonstrates that the generation of
low to medium heat for industrial processes as well as domes-
tic heating is a field of application for this MOST concept.
Several photoswitches have been investigated with regard
to their MOST potential.[5] Despite the diversity of potential
candidates, the key requirements for a MOST system are
always the same: 1) good absorption properties of the parent
molecule across the solar spectrum; 2) little competing absorp-
tion of the metastable photoisomer; 3) high quantum yield of
the photoisomerization; 4) long half-life of the metastable
energy-rich isomer; 5) high energy storage density; and
6) good cyclability.[6]
The best studied[7] and arguably the most promising among
MOST candidates is the interconversion system of norborna-
diene (NBD) and quadricyclane (QC), shown in Scheme 1. The
parent system already combines a huge energy density with
the long half-life of QC.[8] Moreover, it has been shown that by
introducing substituents into the carbon framework of NBD, it
is possible to redshift the absorption into the visible region of
the spectrum, and at the same time, increase the quantum
yield.[9] However, these beneficial photophysical properties
come at the cost of an increased molecular weight and there-
fore lower energy storage capacities.[10] Optimization of one
characteristic of any NBD–QC couple always affects the remain-
ing characteristics as well, rendering the design of the ideal
NBD–QC interconversion system a challenging and yet un-
solved task. Nevertheless, by performing detailed and system-
atic investigations, the group of Moth-Poulsen in particular has
made considerable progress recently.[11] Next to a tailor-made
NBD–QC couple, an efficient and well-designed catalyst is of
paramount importance for a practical MOST system.[6a,12] The
catalyst must enable on-demand energy release with a suffi-
cient turnover number (TON) and turnover frequency (TOF).
Furthermore, the formation of side products must be ruled
out, and the catalyst should be readily separable from the
solar fuel to allow easy cyclization of the process. The latter re-
quirement is best met by heterogeneous catalysts; however, in
many cases they are inferior to homogeneous catalysts in
terms of reactivity and selectivity.[13] By using high-surface-area
materials such as magnetic iron oxide nanoparticles (NPs) as
the support for a molecular catalyst, the best of both worlds
can be achieved. Through the deposition of self-assembled
monolayers, the material’s properties can be modified to facili-
tate the encapsulation of apolar molecules,[14] interchangeable
dispersibility,[15] drug delivery,[16] or the design of quasi-homo-
geneous catalysts.[17] Recently, we showed that by anchoring a
CoSalphen catalyst onto magnetic Fe3O4 NPs, it is possible to
adapt this concept for the design of an easily cyclable MOST
system.[18] Despite this very successful proof of concept, the
huge potential of combining magnetic NP catalysts with the
NBD/QC interconversion in a MOST system has, so far, not
nearly been exhausted.
Results and Discussion
We opted to utilize commercially available Fe3O4 nanoparticles
with a specific surface area of 100 m2 gÀ1 (derived from BET
measurements) as the metal oxide core for the immobilization
of catalytically active functional molecules. Previously, we re-
ported that Fe3O4 nanoparticles, functionalized with a cobalt
(II) salphen moiety carrying a carboxylic acid (Cat1) as anchor
group, act as an effective magnetic catalyst for the exothermic
back-conversion of QC to NBD.[18] On the basis of these results,
we improved the nanoparticle functionalization process and
expanded it to cobalt tetraphenylporphyrin-based catalysts,
which are known to facilitate the QC to NBD interconversion
efficiently.[19] Cat2 is an asymmetric A3B porphyrin with a single
carboxylic acid as anchoring group, whereas Cat3 is a symmet-
ric A4 porphyrin carrying four carboxylic acid groups for an-
choring on metal oxide surfaces (Scheme 2). For the prepara-
tion of the magnetic catalyst, Fe3O4 nanoparticles were treated
with 10 wt% of the respective cobalt complex (Cat1, Cat2, or
Cat3) in isopropanol (Scheme 2). The resulting dispersions
were subsequently irradiated by tip sonication for 30 min and
then stirred under ambient conditions for 48 h. After centrifu-
gation, the reduced color in the supernatant indicated immobi-
lization of the cobalt complexes on the NPs. The resulting NP
hybrids [Fe3O4-Cat1–3] were submitted to multiple washing
steps consisting of redispersion in isopropanol and centrifuga-
tion to remove any unbound ligands. Finally, a washing step
with separation induced by an external magnet instead of cen-
trifugation was conducted. After oven drying, the catalyst load-
ing was determined by elemental analysis. Loadings of
7.9 wt% catalyst for [Fe3O4-Cat1], 3.9 wt% for [Fe3O4-Cat2],
and 7.4 wt% for [Fe3O4-Cat3] were determined (Supporting In-
formation).
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
4994 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
Scheme 2. Functionalization of iron oxide nanoparticles with three different catalytically active molecules Cat1, Cat2, and Cat3. a) Isopropanol, 30 min sonica-
tion, 48 h rt.
To establish the best possible energy storage and release
system, next to this new generation of magnetic NP-based cat-
alysts, we also synthesized NBD–QC couples with exceptional
half-lives, redshifted absorption profiles, outstanding quantum
yields, and high energy storage capacity. For this purpose, we
synthesized a library of NBD derivatives and investigated them
regarding their potential as MOST candidates. Considering our
encouraging experience with the previously published inter-
conversion couple NBD1/QC1,[18] we planned to synthesize
structurally similar NBD derivatives, but with superior photo-
physical properties. One way to redshift the absorption profile
of such derivatives, which has been used widely by the group
of Moth-Poulsen, is the amplification of the push–pull
effect.[10,20] Using this knowledge, we decided to attach elec-
tron-donating groups (EDGs) to the phenyl ring of NBD1. The
synthesis of derivatives NBD2–5 was achieved through two dif-
ferent strategies as depicted in Scheme 3. The NBD derivatives
NBD2–4 were synthesized in an effective transition-metal-free
Corey–Fuchs reaction from aldehyde precursors toward a
push–pull alkyne equipped with a carboxylic ester and a suita-
bly substituted phenyl ring (pathway A). The alkyne precursors
of NBD2–4 were synthesized in good overall yields, and the
alkyne precursor of NBD1 is also commercially available. These
alkynes were then further subjected to a [4+2] Diels–Alder re-
action with cyclopentadiene under microwave irradiation at
1808C to yield the desired norbornadiene derivatives NBD1–4.
NBD5 was synthesized according to pathway B: iodination of
N,N-dimethylaniline gave 4-iodo-N,N-dimethylaniline, which
was reacted with methyl propiolate in a Negishi reaction. This
alkyne was then subjected to a Diels–Alder reaction with cyclo-
pentadiene to give the desired product NBD5. To test our new
magnetic catalysts toward sterically more demanding deriva-
tives, we designed and synthesized NBD6 (Scheme 3, path-
way C). Because of the six methyl groups in combination with
the two ester moieties, the inner carbon framework of NBD6
and the corresponding QC6 is sterically well shielded. This
could potentially hamper the coordination of QC6 to the
active site of the catalyst and therefore slow down or even
prevent the back-reaction. Besides the steric hindrance, NBD6
also differs from the other NBD derivatives by the mechanism
underlying the photoisomerization. As mentioned above,
NBD1–5 are all push–pull substituted NBD derivatives, whereas
NBD6 can be regarded a so-called charge-transfer NBD. These
NBD derivatives distinguish themselves by having one elec-
tron-rich and one electron-poor double bond.[6a] Owing to the
close proximity of the two double bonds in NBD and the re-
sulting through-space interaction, the characteristic charge-
transfer band occurs in the UV/Vis region.[6a,21] Irradiation of
this band results in the formation of the corresponding QC de-
rivatives.[6a,21a]
With this library of new NBD derivatives at hand, the next
step was to test them regarding their potential as MOST candi-
dates. Whether or not an NBD/QC couple is suitable for MOST
applications depends on several criteria, which will be deter-
mined and discussed in the following section. As already men-
tioned, one of the key requirements for a MOST system is a
good spectral overlap between the absorption spectrum of
the parent molecule and the solar spectrum. Figure 1 shows
the photoisomerization of NBD1–6 (blue line) toward their cor-
responding QC derivatives (red line) followed by UV/Vis spec-
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
4995 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
Scheme 3. Synthetic pathways toward the functionalized NBD derivatives NBD1–6. Pathway A relies on the Corey–Fuchs reaction to prepare the alkyne spe-
cies, which is required for the Diels–Alder reaction with Cp towards the final push–pull NBD derivatives NBD1–4. Pathway B involves a Negishi coupling to
obtain the desired alkyne precursor, followed by a Diels–Alder reaction. Pathway C involves methylation of pentamethylcyclopentadiene followed by a Diels–
Alder cycloaddition to yield the charge-transfer derivative NBD6.
troscopy. Therefore, a quartz glass cuvette was charged with a
solution of the respective parent NBD1–4, NBD6 in acetoni-
trile, or NBD5 in benzene, and irradiated at 310 nm and
365 nm, respectively, until no further change in the absorption
profile could be detected, indicating full conversion. (Since
photoisomerization of NBD5 to QC5 in polar solvents is more
complicated than for the other NBD/QC couples, less polar sol-
vents such as toluene and benzene were chosen for this reac-
tion). As expected, a significant redshift of the absorption pro-
file compared with unfunctionalized NBD (lonset <267 nm)[10]
can be observed for all derivatives. For the push–pull NBD de-
rivatives NBD1–5, this redshift of the low-energy absorption
maximum is in line with the strength of the electron-donating
group attached to the phenyl ring (NBD1<NBD4ꢀNBD3<
NBD2<NBD5), with only NBD3 falling out of this sequence
owing to the substituents being bound in meta position. In
case of the donor–acceptor NBD6, the low-energy absorption
band is centered around 311 nm and assigned to the charac-
teristic charge transfer between the electron-rich and the elec-
tron-poor double bonds of donor–acceptor NBDs.[6a,21a] The ex-
istence of at least one isosbestic point in all of the monitored
photoisomerizations indicates a clean conversion of NBD1–6
to the corresponding QCs. This was corroborated by NMR ex-
periments, which not only confirmed the absence of any side
products, but also showed full conversion in all cases (see Sup-
porting Information). Upon comparing the absorption profiles
of the NBDs (blue line) with the corresponding QCs (red line),
a significant blueshift upon transformation can be observed
for all derivatives, enabling selective photoisomerization of the
NBDs.
With an absorption maximum at 366 nm (e=
17490mÀ1 cmÀ1) and an absorption onset of 450 nm, NBD5 al-
ready matches the high-energy portion of the solar spectrum
quite well. This encouraged us to test the photoisomerization
of NBD5 to QC5 directly driven by sunlight. For this purpose, a
quartz glass cuvette was filled with NBD5 (5.0 mg) dissolved in
deuterated toluene (3.0 mL) and exposed to direct sunlight on
the windowsill. Already after five minutes, the color of the so-
lution had changed from yellow to colorless, indicating isomer-
1
ization to QC5. An aliquot was taken for H NMR analysis, and
from the integral ratio of the signals a NBD5/QC5 ratio of
5:95% was estimated. After another five minutes of exposure
to sunlight the conversion to QC5 was complete, and remarka-
bly, no side products were observed (for more details see Sup-
porting Information). This impressively illustrates the outstand-
ing photophysical properties of this NBD/QC couple.
Apart from the overlap with the solar spectrum and the ex-
tinction coefficients, the quantum yield of the respective pho-
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
4996 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
Supporting Information). The obtained values for the quantum
yields vary between 51% for NBD5 and 82% for NBD6. How-
ever, both values have to be treated carefully; owing to the
low extinction coefficient of NBD6, quite a concentrated solu-
tion was necessary to guarantee an optically thick solution. In
the case of NBD5 a concentration dependence of the quantum
yield was observed whereby higher concentrations gave
higher quantum yields; however, a saturation of the obtained
quantum yield was detected at the reported value of 51% (for
more details see Supporting Information).
Another important characteristic of every NBD/QC couple is
its energy storage density. The enthalpies of the thermal back-
reaction were determined by performing differential scanning
calorimetry (DSC) experiments with QC2–6 (Figures S1–S11,
Supporting Information). As an example, Figure 2 depicts the
obtained DSC curves of QC2 (top) and QC5 (bottom). In every
case a strong exothermic peak was observed in the first heat-
ing cycle, which can be assigned to the thermal back-reaction.
The absence of this peak in the second heating cycle proves
that the transformation of QC to NBD was completed within
the first heating cycle. In the case of the NBD5/QC5 couple,
two peaks can be observed in the second heating cycle, which
are assigned to phase transformations of NBD5. The absence
of any side reactions was proven by 1H NMR analysis after
completion of the DSC experiments (for more details see Sup-
porting Information). The measured enthalpies vary between
75.0 and 98.0 kJmolÀ1, and in combination with the molecular
weight of the respective NBD/QC couple these correspond to
energy storage densities between 249.7 and 367.5 kJkgÀ1 in
the neat systems.
Figure 1. UV/Vis absorption spectra of NBD1–6 during photoisomerization.
NBD1–4 and NBD6 were irradiated at 310 nm in acetonitrile, whereas NBD5
was irradiated at 365 nm in benzene. The characteristic absorption features
of the NBD moieties (blue line) show a rapid decline during the irradiation
process. Distinct isosbestic points in each measurement indicate a clean
conversion of each NBD to its energy-rich isomer. The absorption profiles of
the resulting QCs QC1–6 (red-line) are significantly blueshifted relative to
their NBD forms.
As well as the amount of energy that can be stored within
the metastable QC derivatives, their kinetic stability is also of
paramount importance as it determines the timespan over
which the energy can be stored. The thermal half-life of QC de-
rivatives at room temperature can vary considerably between
a few hours and years.[10,20,23] To determine the thermal half-life
of QC2,4,5, we prepared stock solutions of the respective
NBDs in deuterated tetrachloroethane (NBD2,4) or deuterated
toluene (NBD5) and irradiated them with 310 and 365 nm
light, respectively, until 1H NMR indicated full conversion. By
toisomerization is needed to further quantify the energy stor-
age efficiency. The quantum yields of the reactions NBD1–6 to
QC1–6 were determined by using potassium ferrioxalate as
chemical actinometer,[22] and are presented in Table 1. For this
purpose, NBD1–4 and NBD6 were dissolved in acetonitrile and
irradiated with 310 nm light, whereas NBD5 was dissolved in
toluene and irradiated at 365 nm (for more information see
Table 1. Summary of the optical and thermodynamic properties of the interconversion couples NBD1–6/QC1–6. t1/2 (258C) determined in deuterated tetra-
chloroethane (QC1, QC2 and QC4) or in deuterated toluene (QC5). The optical parameters lmax, lonset, and emax were determined from UV/Vis spectra in ace-
tonitrile (NBD1–4 and NBD6) or toluene (NBD5). DH as determined from the average of at least two DSC measurements. Quantum yields for the photoiso-
merization in acetonitrile (NBD1–4 and NBD6) or toluene (NBD5) determined from at least two independent measurements, according to a literature-
known method.[22]
[a]
NBD
Derivative
MW
[gmolÀ1
t1/2 (258C)
[days]
lmax
lonset
emax
[mÀ1 cmÀ1
DH
[kJmolÀ1
DH
[kJkgÀ1
Fisom
]
[nm]
[nm]
]
]
]
[%]
[b]
NBD1
NBD2
NBD3
NBD4
NBD5
NBD6
240.30
270.33
300.35
296.41
269.34
292.38
450[b]
156
/
611
103
/
291
316
299
300
366
311
359
380
364
370
450
357
6460
10780
7500
8680
17490
420
88.3Æ0.2[b]
87.4Æ0.4
75.0Æ0.2
77.3Æ0.2
98.0Æ0.4
/
367.5Æ0.8
323.3Æ1.5
249.7Æ0.7
260.8Æ0.7
363.9Æ1.5
/
71
68
73
59
51[c]
82[d]
[a] lonset is defined as the wavelength with log(e)=2. [b] Values adapted from ref. [18]. [c] The quantum yield of NBD5 increased in line with the concentra-
tion, reaching a saturation at 51% (for more details see Supporting Information). [d] Owing to the low extinction coefficient of NBD6, a quite concentrated
solution was necessary to obtain an optically thick solution.
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
4997 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
back-conversion of the metastable QC isomers to their NBD
form. For a well-performing MOST device, a controlled and fast
release of the photochemically stored energy is of paramount
importance. Further, a facile removal of the applied catalytic
materials is crucial to ensure a functioning cycle of solar-driven
energy storage followed by on-demand energy release, trig-
gered by the catalyst. Owing to their intrinsic magnetic proper-
ties, the nanoparticle hybrids [Fe3O4-Cat1–3] allow easy purifi-
cation through the application of an external magnet. Addi-
tionally, we investigated these nanoparticle hybrids on their
catalytic activity for the back-conversion of QC to NBD. For this
purpose, NBD1 (8.66”10À2 mmol) dissolved in CHCl3 was irra-
diated (310 nm) for approximately 2 h until complete conver-
sion to QC1 was achieved. This was verified by peak integra-
tion of HPLC measurements. The resulting solution of QC1 was
then treated with 0.3 mg of the magnetic nanoparticle cata-
lysts [Fe3O4-Cat1–3]. After addition of the catalyst particles,
the reaction mixtures were mixed on a vibrating plate, and
after set time intervals, the catalyst particles were sedimented
with an external magnetic field and aliquots of the mixture
were taken for HPLC analysis. These investigations revealed a
very slow QC–NBD back-conversion rate for the first-generation
catalyst particles [Fe3O4-Cat 1]. Under the applied reaction
conditions, a QC1–NBD1 conversion of 5% was reached after
3 h. As described previously, an increase in the concentration
of [Fe3O4-Cat1] led to complete conversion.[18]
As evident from Figure 3, a tremendous increase in catalyst
performance was observed upon using the second generation
of porphyrin-based catalysts [Fe3O4-Cat2–3]. Both nanoparticle
hybrids displayed a very high initial activity compared with
[Fe3O4-Cat1], with [Fe3O4-Cat2] facilitating complete conver-
sion of QC1 to NBD1 after only 40 min (Figure 3).
However, after separation of the catalyst particles [Fe3O4-
Cat2], the supernatant displayed a yellow color, which indicat-
ed desorption of the porphyrin ligand Cat2 from the nanopar-
Figure 2. a) DSC measurement of QC2. The peak during the first heating
cycle is strongly exothermic and can be assigned to the back-conversion of
QC2 to NBD2 corresponding to a heat release of 87.4 kJmolÀ1. b) DSC mea-
surement of QC5. The peak in the first heating cycle is exothermic and can
be assigned to the back-conversion of QC5 to NBD5 corresponding to a
heat release of 98.0 kJmolÀ1. The two other peaks in the second heating are
assigned to phase transformations after the rapid cooling process.
following the thermal back-reaction of the thus-prepared QCs
directly by NMR, it was possible to estimate the rate constants
k of each QC derivative at different temperatures. These pa-
rameters in combination with the Eyring equation enabled us
to calculate DH##
and DS##
(see Supporting Informa-
thermal
thermal
tion). Furthermore, through extrapolation of the data, we de-
termined the half-life of QC2,4,5 at 258C (Table 1). With t1/2
(258C) between 103 and 611 days, all the QCs reported here
are suitable for long-term energy storage. Compared with the
photochemical characteristics of other literature-known NBD
derivatives, our compounds are highly competitive; in particu-
lar, NBD5 combines an exceptional absorption profile with
high thermal stability.[11a]
Figure 3. Comparison of the catalytic effectivity of the three magnetic nano-
particle catalyst systems [Fe3O4-Cat1–3] in the back-conversion of QC1 to
NBD1 in chloroform. The conversion of QC1 to NBD1 was followed by HPLC
analysis.
After an in-depth analysis of the optical and thermodynamic
properties of each NBD/QC couple we focused on the catalytic
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
4998 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
ticle surface. UV/Vis analysis later suggested the presence of
the dication of the free base porphyrin ligand Cat2 (Support-
ing Information). Thus, the high excess of QC1/NBD1 led to
desorption and degradation of [Fe3O4-Cat2]. In contrast,
[Fe3O4-Cat3] also displayed a high initial activity and resulted
in a QC1–NBD1 back-conversion of 75% after 3 h, without
showing any signs of desorption from the surface. Through
this combination of high activity and stability against desorp-
tion we deemed [Fe3O4-Cat3] to be the most suitable catalyst
candidate. The initial turnover frequency (TOF) after 5 min was
calculated at 1.15 sÀ1. This corresponds to a 22.6-fold increase
in activity compared with the first-generation catalyst [Fe3O4-
Cat1] reported previously [TOF (5 min): 0.0508 sÀ1].[18]
We decided to utilize [Fe3O4-Cat3] as the catalyst system for
all further investigations on the library of QC molecules QC1–
6, owing to the very stable anchoring and extremely high ac-
tivity compared with the other catalyst candidates. Therefore,
solutions of each NBD derivative were irradiated with a UV-LED
setup either at 310 nm in chloroform for NBD1–4 and NBD6 or
at 365 nm in toluene for NBD5. The samples were irradiated
until complete conversion to the metastable QC form was veri-
fied by HPLC analysis for the NBD-QC couple 1 or NMR analysis
for the NBD-QC couples 2–6. After the successful photoisome-
rization, QC2–6 (1.44”10À2 mmol) dissolved in the respective
solvent (0.5 mL) was treated with the magnetic nanoparticle
catalyst [Fe3O4-Cat3] (0.05 mg) dispersed in the corresponding
solvent (50 mL). The resulting dispersions were then mixed on
a vibrating plate. After set time intervals, the nanoparticle cata-
lyst was sedimented by using an external magnetic field and
the supernatant was removed for analysis. For each sample,
the conversion ratio of QC to NBD was monitored by compar-
ing the integrals from the NMR analysis (conversion values for
NBD1 were determined by HPLC and taken from Figure 3). The
results were then plotted versus the exposure time of each
sample to the nanoparticle catalyst (Figure 4). In addition to
the depicted time intervals, the QC-NBD conversion was also
recorded after an exposure time of 3 h.
Figure 4. Catalytic back-conversion of QC1–5 to NBD1–5, mediated by
[Fe3O4-Cat3] (0.03 mol% active catalyst). The conversion of QC1–4 to
NBD1–4 took place in chloroform and was followed by HPLC for NBD1 and
1
by H NMR analysis for NBD 2–4, whereas the conversion of QC5 to NBD5
1
took place in deuterated toluene and was followed by H NMR analysis.
tion was made by Maruyama et al. using methyl-decorated QC
derivatives.[24]
To quantify the catalytic activity of the magnetic nanoparti-
cle catalyst [Fe3O4-Cat3] with each QC derivative, we calculat-
ed the turnover frequency (TOF) after 5 min and the turnover
number (TON) of the nanoparticle catalyst after a maximum re-
action time of 3 h. The TON is calculated as the moles of con-
verted analyte divided by the moles of catalyst used, whereas
TOF is defined as TON per second. The interconversion couple
NBD4-QC4 displayed the highest initial TOF of 3.64 sÀ1 with a
minimum TON of 3305 after complete back-conversion of QC4
to NBD4. We observed the second highest TOF value of
2.62 sÀ1 for the back-conversion of QC2 to NBD2 with a very
high TON value of 3123 after a reaction time of 3 h. The back-
conversion of QC3 to NBD3 displayed a slightly reduced initial
TOF of 2.11 sÀ1; however, after reaction times longer than
10 min, a reduced activity was detected, resulting in a TON of
2412 after 3 h exposure to the catalyst particles. QC1, featuring
an unmodified phenyl ring as donor group, showed, on the
one hand, a reduced initial TOF of 1.15 sÀ1, but on the other
hand, a more stable conversion rate. Despite the lower initial
activity compared with QC3, this resulted in a higher TON of
2496 after a reaction time of 3 h. The lowest QC to NBD back-
conversion activity was measured for QC5, with a moderate in-
itial TOF of 0.95 sÀ1. This initial activity rapidly declined, result-
ing in the lowest TON of 641 after 3 h exposure to the nano-
particle catalyst [Fe3O4-Cat3]. As discussed above, this lowered
activity can be explained either by solvent effects or by inhibi-
Figure 4 displays some key differences between the NBD de-
rivatives. Most notable was the difference in catalytic QC-NBD
back-conversion activity that resulted from the chemical modi-
fications to the electron-donating phenyl substituent. In gener-
al, the presence of these electron-donating groups increased
the back-conversion activity of QC2–4 substantially.
In QC5 however, we observed a reduced back-conversion
activity. On the one hand, this can be attributed to the change
of the solvent. On the other hand, this reduced activity may be
a result of coordination of the amine groups of QC5 to the
cobalt centers in [Fe3O4-Cat3]. Hence, the addition of the
labile QC bonds to the cobalt center is hampered, resulting in
a reduced effectivity of [Fe3O4-Cat3] for back-conversion of
QC5. This correlates well with the observed initial activity,
which is reduced significantly after reaction times longer than
5 min (Figure 4). QC6 did not show any activity in the nanopar-
ticle-mediated back-conversion of the metastable QC form to
NBD6. We attribute this inactivity to the increased steric bulk,
which inhibits coordination to the surface-bound catalytic cen-
ters in the magnetic catalyst [Fe3O4-Cat3]. A similar observa-
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
4999 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
tion of the catalyst through coordination of the amine group.
After catalytic back-conversion and magnetic separation of the
catalyst particles, NBD1–5 could undergo further quantitative
steps of photoisomerization and catalytic back-conversion. For
NBD5, we additionally verified that the photoisomerization
quantum yield does not change after treatment with the mag-
netic nanoparticle catalyst [Fe3O4-Cat3].
version couple is a special challenge for the catalytic back-con-
version because the QC isomer is relatively stable and has a
higher activation barrier for the back-conversion than the vast
majority of substituted derivatives. Despite this, the nanoparti-
cle catalyst [Fe3O4-Cat3] facilitated complete conversion of QC
to NBD in both chloroform and acetonitrile as solvent. Howev-
er, we could observe a solvent-dependent difference in the
speed of the back-conversion. In acetonitrile the back-conver-
sion was significantly slower, which is probably a result of the
stronger competing coordination of acetonitrile compared
with chloroform to the cobalt catalyst,[26] as well as an influ-
ence of the solvent on the isomerization itself. The latter trend
was also observed for several substituted derivatives of NBD
by Moth-Poulsen and co-workers.[27] In general, less polar sol-
vents appear to be more suitable for the fast and on-demand
release of the photochemically stored energy.
In particular, the unsubstituted parent NBD-QC interconver-
sion couple holds great potential for an efficient MOST device.
Its low molecular weight, combined with the fact that both
NBD and QC are pumpable liquids, enables the usage of neat
NBD/QC, and therefore, very high gravimetric energy densities.
These benefits are to some extent hampered by the fact that
the parent NBD does not absorb light above 267 nm.[10] There-
fore, a triplet photosensitizer is necessary to facilitate the light-
induced isomerization of unsubstituted NBD to the energy-rich
QC.[8a,25] During this work, we opted to utilize acetophenone as
a photosensitizer. Generally, a solution of NBD and acetophe-
none in acetonitrile was irradiated with a mercury vapor lamp.
The progress of the photoisomerization was monitored by
GC/MS measurements. After a conversion of more than 80%
was achieved, QC was collected by fractioned distillation. For
investigation of the details of the catalyzed back-conversion of
QC to NBD, purified QC (20 mL) was dissolved in either chloro-
form or acetonitrile and then treated with the nanoparticle cat-
alyst [Fe3O4-Cat3] (2.0 mg) (Figure 5).
Translation of the stored energy into a practically usable
heat release is of paramount importance for the application of
MOST devices. On a theoretical level, the achievable tempera-
ture rise can be calculated by factoring in the key properties of
the storage molecule and solvent through Equation (1).[2]
cMWDH
DT
1
cMWCp
Cp solv1solv
NBD
In this case, MW and c are the molar weight of the norbor-
nadiene derivative and the concentration, and DH is the mea-
sured energy storage capacity of the respective QC in JgÀ1. For
Cp NBD, the heat capacity of NBD, we assumed the value known
for the unsubstituted parent norbornadiene (1.66 JgÀ1 KÀ1). In
the case of NBD1, the energy storage capacity is 367 JgÀ1. This
translates to a theoretical maximum temperature rise of 2218C
without solvent. In a 1m solution in chloroform this capacity is
reduced to a theoretical temperature increase of 48.58C.
Reaching these theoretical maxima under experimental condi-
tions requires a high degree of engineering to avoid losses
from insufficient insulation. Through utilization of a UHV
design and a fixed bed catalyst, Moth-Poulsen and co-workers
have recently demonstrated that these high theoretical values
can be reached with sufficient insulation.[12]
As described previously, these solutions were then mixed on
a vibrating plate, and after set intervals the nanoparticle cata-
lyst was removed by an external magnet and aliquots of the
solution were taken for analysis by GC/MS. In addition to its
limited optical properties, the unsubstituted NBD-QC intercon-
In this work, we demonstrate the heat release potential of
10 mL of a 1m solution of the interconversion couple NBD1-
QC1 in chloroform under ambient conditions in a standard
glass vial without any thermal insulation (Figure 6).
Upon addition of the solution of QC1 to 50 mg of the mag-
netic nanoparticle catalyst [Fe3O4-Cat3], an immediate temper-
ature increase was measured. Within 129 s, the temperature of
the solution increased from 248C to its peak at 39.58C. After
reaching its peak the temperature stayed above 308C for sev-
eral minutes. This converts to a total temperature increase of
15.58C, which could be achieved without thermal insulation in
a 1m solution of the interconversion couple NBD1-QC1
through catalysis by the magnetic nanoparticle hybrid [Fe3O4-
Cat3].
After the temperature fell below 308C the nanoparticle cata-
lyst was separated from the solution by application of an ex-
ternal magnetic field and the isomerization progress was moni-
Figure 5. Catalytic back-conversion of unsubstituted quadricyclane to nor-
bornadiene induced by [Fe3O4-Cat3] nanoparticles, followed by GCMS analy-
sis.
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
5000 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
and highly efficient magnetic catalysts [Fe3O4-Cat2] and
[Fe3O4-Cat3]. Desorption and degradation were major issues
with [Fe3O4-Cat2], but this problem was solved through multi-
ple carboxylate anchors in [Fe3O4-Cat3]. With this second gen-
eration of magnetic NP-based catalysts, we achieved a 22.6-
fold increase in catalyst performance compared with the first
catalyst generation. [Fe3O4-Cat3] was tested toward all NBD/
QC couples reported herein, and showed excellent per-
formance with an initial TOF up to 3.64 sÀ1 and a TON of more
than 3305. However, a solvent dependency of the catalyst per-
formance was observed, whereby a more polar and coordinat-
ing solvent seemed to hamper the catalyst performance. Fur-
thermore, we tested the applicability of [Fe3O4-Cat3] com-
bined with NBD1/QC1 in a heat release experiment to model
its MOST potential. Remarkably, a temperature increment of
15.58C was reached without any thermal insulation. This dem-
onstrates that an NBD/QC-based MOST system together with
magnetic Fe3O4 NP catalysts is, in principle, suitable for applica-
tions such as the generation of industrial process heat or do-
mestic heating.
Acknowledgements
Figure 6. Time-dependent temperature profile of the conversion of a macro-
scopic amount of QC1 in CHCl3 (1m, 10 mL) to NBD1 catalyzed by [Fe3O4-
Cat3] nanoparticles.
We gratefully thank the German Research Council (DFG) for
funding through project 391585168 “Photochemisch und mag-
netochemisch ausgelçste Speicherung/ Freisetzung von Sonn-
energie in gespannten organischen Verbindungen”. Additional-
ly, we would like to thank the Bavarian Collaborative Research
Project Solar Technologies go Hybrid (SolTech), the Cluster of
Excellence “Engineering of Advanced Materials” (EAM), funded
by DFG, the Graduate School Molecular Science (GSMS), and
the Graduate School Advanced Materials and Processes
(GSAMP) for financial support. Open access funding enabled
and organized by Projekt DEAL.
tored by NMR spectroscopy. The 1H NMR analysis showed com-
plete conversion of the 1m solution of QC1 to NBD1 within
the reported timespan alongside an effective and facile separa-
tion of the catalyst.
Conclusions
In summary, we have developed a highly efficient MOST bat-
tery through the combination of NBD photoswitches and a
cobalt porphyrin catalyst anchored to magnetic NPs. A stable
binding to the nanoparticles surface is achieved through multi-
ple carboxylate anchors, preventing desorption of the active
species. For the first time, a detailed analysis of the per-
formance of these novel magnetic NP catalysts was conducted.
The high catalytic activity enabled a fast and on-demand re-
lease of the photochemically stored energy together with
facile magnetic separation of the catalyst particles. In particu-
lar, we have significantly improved on our previous work on
the combination of a magnetic cobalt salphene catalyst and
the interconversion couple NBD1/QC1.[18] On the one hand,
the interconversion couple NBD1/QC1 was complemented by
a library of NBD/QC derivatives carrying electron-donating
groups on the phenyl ring. This led to a significant improve-
ment in their photophysical properties, culminating in facile
sunlight-driven conversion of NBD5 to QC5. On the other
hand, detailed quantitative experiments on the photochemical
properties and catalyst performance were conducted. For the
latter, we additionally anchored the two cobalt porphyrins
Cat2 and Cat3 onto Fe3O4 NPs, thereby generating two new
Conflict of interest
The authors declare no conflict of interest.
Keywords: magnetic
norbornadiene · process heat · solar energy storage
catalysts
·
nanoparticles
·
[1] a) P. Lorenz, A. Hirsch, Chem. Eur. J. 2020, 26, 5220–5230; b) F. Waidhas,
M. Jevric, L. Fromm, M. Bertram, A. Gçrling, K. Moth-Poulsen, O. Brum-
mel, J. Libuda, Nano Energy 2019, 63, 103872; c) M. Mansø, A. U. Peters-
en, K. Moth-Poulsen, M. B. Nielsen, Org. Biomol. Chem. 2020, 18, 2113–
2119; d) F. Waidhas, M. Jevric, M. Bosch, T. Yang, E. Franz, Z. Liu, J. Bach-
mann, K. Moth-Poulsen, O. Brummel, J. Libuda, J. Mater. Chem. A 2020,
8, 15658–15664; e) M. Bertram, F. Waidhas, M. Jevric, L. Fromm, C.
Schuschke, M. Kastenmeier, A. Gçrling, K. Moth-Poulsen, O. Brummel, J.
Libuda, J. Chem. Phys. 2020, 152, 044708; f) K. Ishiba, M.-a. Morikawa, C.
Chikara, T. Yamada, K. Iwase, M. Kawakita, N. Kimizuka, Angew. Chem.
Int. Ed. 2015, 54, 1532–1536; Angew. Chem. 2015, 127, 1552–1556.
´
[2] K. Moth-Poulsen, D. Coso, K. Bçrjesson, N. Vinokurov, S. K. Meier, A. Ma-
jumdar, K. P. C. Vollhardt, R. A. Segalman, Energy Environ. Sci. 2012, 5,
8534–8537.
[3] a) S. Hess in Renewable heating and cooling. Technologies and applica-
tions, Woodhead Publishing series in energy, number 89, Ed. G. Stryi-Hipp,
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
5001 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH

Full Paper
doi.org/10.1002/chem.202005427
Chemistry—A European Journal
Elsevier, Amsterdam, 2016, pp. 41–66; b) S. Kalogirou, Appl. Energy
2003, 76, 337–361.
Vivod, T. Luchs, T. Rejek, L. Portilla, L. Müller, H. Dietrich, A. Hirsch, D.
Zahn, M. Halik, Adv. Funct. Mater. 2019, 29, 1805742.
[4] a) C. A. McMillan, M. Ruth, Appl. Energy 2019, 239, 1077–1090; b) C. A.
Schoeneberger, C. A. McMillan, P. Kurup, S. Akar, R. Margolis, E. Masanet,
Energy 2020, 206, 118083.
[5] a) C.-L. Sun, C. Wang, R. Boulatov, ChemPhotoChem 2019, 3, 268–283;
b) A. Lennartson, A. Roffey, K. Moth-Poulsen, Tetrahedron Lett. 2015, 56,
1457–1465.
[6] a) Z. Yoshida, J. Photochem. 1985, 29, 27–40; b) T. J. Kucharski, Y. Tian, S.
Akbulatov, R. Boulatov, Energy Environ. Sci. 2011, 4, 4449–4472.
[7] a) A. D. Dubonosov, V. A. Bren, V. A. Chernoivanov, Russ. Chem. Rev.
2002, 71, 917–927; b) V. A. Bren’, A. D. Dubonosov, V. I. Minkin, V. A.
Chernoivanov, Russ. Chem. Rev. 1991, 60, 451–469.
[8] a) G. S. Hammond, N. J. Turro, A. Fischer, J. Am. Chem. Soc. 1961, 83,
4674–4675; b) X. An, Y. Xie, Thermochim. Acta 1993, 220, 17–25.
[9] a) K. Hirao, A. Ando, T. Hamada, O. Yonemitsu, J. Chem. Soc. Chem.
Commun. 1984, 300–302; b) V. I. Minkin, V. A. Breit, V. A. Chernoivanov,
[15] L. Zeininger, L. M. S. Stiegler, L. Portilla, M. Halik, A. Hirsch, ChemistryOp-
en 2018, 7, 282–287.
[16] a) S. Klein, T. Luchs, A. Leng, L. V. R. Distel, W. Neuhuber, A. Hirsch, Bioen-
gineering 2020, 7, 126; b) C. K. Kim, P. Ghosh, C. Pagliuca, Z.-J. Zhu, S.
Menichetti, V. M. Rotello, J. Am. Chem. Soc. 2009, 131, 1360–1361.
[17] a) A.-H. Lu, E. L. Salabas, F. Schüth, Angew. Chem. Int. Ed. 2007, 46,
1222–1244; Angew. Chem. 2007, 119, 1242–1266; b) R. B. N. Baig, R. S.
Varma, Chem. Commun. 2013, 49, 752–770.
[18] T. Luchs, P. Lorenz, A. Hirsch, ChemPhotoChem 2020, 4, 52–58.
[19] a) J. Manassen, J. Catal. 1970, 18, 38–45; b) H. Wilson, R. Rinker, J. Catal.
1976, 42, 268–274; c) M. Chen, H. Feder, J. Catal. 1978, 55, 105–107;
d) K. Maruyama, H. Tamiaki, Chem. Lett. 1982, 11, 1699–1702.
[20] M. Jevric, A. U. Petersen, M. Mansø, S. Kumar Singh, Z. Wang, A. Dreos,
C. Sumby, M. B. Nielsen, K. Bçrjesson, P. Erhart, K. Moth-Poulsen, Chem.
Eur. J. 2018, 24, 12767–12772.
A. D. Dubonosov, S. V. Galichev, Mol. Cryst. Liq. Cryst. 1994, 246, 151– [21] a) R. Hoffmann, Acc. Chem. Res. 1971, 4, 1–9; b) M. B. Robin, N. A. Kue-
154; c) V. Gray, A. Lennartson, P. Ratanalert, K. Bçrjesson, K. Moth-Poul-
sen, Chem. Commun. 2014, 50, 5330–5332.
[10] M. Quant, A. Lennartson, A. Dreos, M. Kuisma, P. Erhart, K. Bçrjesson, K.
Moth-Poulsen, Chem. Eur. J. 2016, 22, 13265–13274.
[11] a) J. Orrego-Hernµndez, A. Dreos, K. Moth-Poulsen, Acc. Chem. Res.
2020, 53, 1478–1487; b) M. Mansø, A. U. Petersen, Z. Wang, P. Erhart,
M. B. Nielsen, K. Moth-Poulsen, Nat. Commun. 2018, 9, 1945; c) M.
Kuisma, A. Lundin, K. Moth-Poulsen, P. Hyldgaard, P. Erhart, ChemSu-
sChem 2016, 9, 1786–1794.
bler, J. Chem. Phys. 1966, 44, 2664–2671.
[22] C. G. Hatchard, C. A. Parker, Proc. R. Soc. London Ser. A 1956, 235, 518–
536.
[23] A. Dreos, Z. Wang, J. Udmark, A. Strçm, P. Erhart, K. Bçrjesson, M. B.
Nielsen, K. Moth-Poulsen, Adv. Energy Mater. 2018, 8, 1703401.
[24] K. Maruyama, H. Tamiaki, J. Org. Chem. 1986, 51, 602–606.
[25] H. Taoda, K. Hayakawa, K. Kawase, J. Chem. Eng. Jpn. 1987, 20, 335–338.
[26] a) R. Díaz-Torres, S. Alvarez, Dalton Trans. 2011, 40, 10742–10750; b) S.
Alvarez, Chem. Eur. J. 2020, 26, 4350–4377.
[12] Z. Wang, A. Roffey, R. Losantos, A. Lennartson, M. Jevric, A. U. Petersen,
M. Quant, A. Dreos, X. Wen, D. Sampedro, K. Bçrjesson, K. Moth-Poul-
sen, Energy Environ. Sci. 2019, 12, 187–193.
[27] M. Quant, A. Hamrin, A. Lennartson, P. Erhart, K. Moth-Poulsen, J. Phys.
Chem. C 2019, 123, 7081–7087.
[13] D. J. Cole-Hamilton, R. P. Tooze, Catalyst separation, recovery and recy-
cling. Chemistry and process design, Catalysis by Metal Complexes,
Vol. 30, Eds. D. J. Cole-Hamilton, R. P. Tooze, Springer, Dordrecht, 2006,
pp. 1–8.
[14] a) T. Luchs, M. Sarcletti, L. Zeininger, L. Portilla, C. Fischer, S. Harder, M.
Halik, A. Hirsch, Chem. Eur. J. 2018, 24, 13589–13595; b) M. Sarcletti, D.
Manuscript received: December 21, 2020
Revised manuscript received: January 14, 2021
Accepted manuscript online: January 15, 2021
Version of record online: February 22, 2021
Chem. Eur. J. 2021, 27, 4993 –5002
www.chemeurj.org
5002 ꢀ 2021 The Authors. Chemistry - A European Journal published by Wiley-VCH GmbH