Light-driven rotary molecular motors on gold nanoparticles

Article abstract of DOI:10.1002/chem.200800814

We report the synthesis of unidirectional light-driven rotary molecular motors based on chiral overcrowded alkenes and their immobilisation on the surface of gold nanoparticles through two anchors. Using a combination of ¹H and ¹³C N

Full text of DOI:10.1002/chem.200800814

DOI: 10.1002/chem.200800814
Light-Driven Rotary Molecular Motors on Gold Nanoparticles
Michael M. Pollard, Matthijs K. J. ter Wiel, Richard A. van Delden, Javier Vicario,
Nagatoshi Koumura, Coenraad R. van den Brom, Auke Meetsma, and Ben L. Feringa*[a]
Abstract: We report the synthesis of
unidirectional light-driven rotary mo-
lecular motors based on chiral over-
crowded alkenes and their immobilisa-
tion on the surface of gold nanoparti-
cles through two anchors. Using a com-
bination of 1H and 13C NMR, UV/Vis
and CD spectroscopy, we show that
these motors preserve their photo-
chemical and thermal behaviour after
they have been attached to gold nano-
particles. Furthermore, we describe the
Taken together, these data support the
conclusion that these motors maintain
their unidirectional rotary cycle when
grafted to the surface of small (ca.
2 nm) gold nanoparticles. Thus, contin-
uous irradiation of the system under
appropriate conditions leads to unidir-
ectional rotation of the upper half of
the molecules relative to the entire
nanoparticle.
2
synthesis of H- and 13C-labelled deriv-
atives that were used to verify the uni-
directionality of the rotary cycle of
these motors both in solution and
while grafted to gold nanoparticles.
Keywords: circular dichroism
·
molecular devices · nanostructures ·
nanotechnology · photochemistry
Introduction
characterized on surfaces, however, are much more
scarce.[3a,b,11,13]
Controlled manipulation of matter on the nanoscale by
means of molecular machines has been identified as one of
the most challenging frontiers of nanotechnology.[1] To be
able to do this, chemists must develop molecular systems
that can perform mechanical tasks in an environment that is
continuously overwhelmed by Brownian motion.[2,3] In an
effort to gain control over motion at the molecular scale,
several designs of linear and rotary molecular motors have
been reported.[3–9] While most of these operate exclusively
in solution, it is anticipated that anchoring molecular motors
to surfaces is an essential step toward transmitting their ac-
tions into work upon their surrounding environment.
Indeed, anchoring linear motors to surfaces has enabled the
fabrication of more complex molecular devices that can per-
form mechanical work.[10] Synthetic rotary molecular motors
Biological systems use rotary molecular motors in a di-
verse array of key cellular functions including cellular trans-
location, ion pumping, and ATP biosynthesis.[12] In all of
these capacities, one “stator” half of the biomotor is immo-
bilized within a cell membrane, so that the cell can harness
the rotary motion of the “rotor”.
The designs of these complex natural systems inspired us
to explore the effect of immobilizing light-driven rotary mo-
lecular motors on surfaces. We found that the rotation of
the upper half relative to the lower half of the molecule in
solution could be converted to absolute rotation of the
upper half relative to a gold nanoparticle.[13] We report the
synthesis and characterisation of motors on small (ca. 2 nm)
gold colloidal particles. Using UV, CD and NMR spectros-
copy, we compared each of the photochemical and thermal
steps of the rotary cycle of the light-driven rotary motors at-
tached to the surface of gold nanoparticles to the analogous
steps of the parent molecule operating in solution. From this
comparison, we show that the photochemical and thermal
steps of the unidirectional rotary cycle of the motor are pre-
served when the molecules are grafted onto the nanoparti-
cles.
[a] Prof. Dr. M. M. Pollard, Dr. M. K. J. ter Wiel, Dr. R. A. van Delden,
Dr. J. Vicario, Dr. N. Koumura, Dr. C. R. van den Brom,
Dr. A. Meetsma, Prof. Dr. B. L. Feringa
Department of Organic and Molecular Inorganic Chemistry
Stratingh Institute for Chemistry, University of Groningen
Nijenborgh 4, 9747 AG Groningen (The Netherlands)
Fax : (+31)50-363-4296
Molecular design: One of the most promising designs of
rotary molecular motors is based on overcrowded alkenes
(exemplified by 1 in Figure 1 and 3 in Figure 2). As shown
E-mail: b.l.feringa@rug.nl
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.200800814.
11610
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11610 – 11622

FULL PAPER
methyl group and the naphthalene ring slip past the aromat-
ic moieties of the lower half. This rate-limiting step regener-
ates the stable isomer in which the methyl substituent is in a
pseudo-axial position, and thus 1808 rotation of the upper
half relative to the lower half has occurred. Repeated photo-
isomerisation followed by helix inversion results in continual
unidirectional rotation. Upon constant irradiation at appro-
priate temperatures that allow facile thermal helix inversion,
the upper half of the molecule undergoes continual unidirec-
tional rotation relative to the lower half. The speed of rota-
tion can be adjusted by replacing the sulfur atoms in 3 with
oxygen and carbon.[6b,8]
Attaching the motor appropriately to a solid support
transforms the relative rotary motion of the upper half with
respect to the lower half into absolute rotary motion relative
to the support. The use of gold colloids as solid support is
attractive because their synthesis is relatively straightfor-
ward[15] and well precedented, and because photochemical
reactions of molecules in monolayers on gold nanoparticles
are also known.[16,17] Moreover, the motor-decorated nano-
particles can be easily studied by UV/Vis and CD spectros-
copy. This is essential for following each step of the rotary
cycle and evaluating the effect, if any, that immobilisation
has on either the photochemical or thermal steps in the
rotary cycle.
Figure 1. Translation of unidirectional rotary motion in solution into ab-
solute rotary motion is achieved by grafting the stator half of the motor
to a gold nanoparticle.
In the design, it was essential that the motor molecule be
attached to the nanoparticle through two points to prevent
Brownian rotation around single bonds. The tethers should
be long enough to allow sufficient conformational flexibility
and, more importantly, to permit photoisomerisation of the
motor with minimal interference/quenching by the gold
nanoparticle. For these reasons an eight-carbon chain was
chosen.[17] Moreover, the choice of a longer tether would
avoid overcrowding of the motor molecules around the gold
core of the particle, which could influence both the photo-
chemical and thermal steps of the rotary cycle.
To facilitate complete characterisation of the different
states of the molecule through its four-step rotary cycle, the
motor was designed to be similar in structure to reported
slow motor molecules.[8] The reported half-life for the rate-
limiting thermal isomerisation step in the rotary cycle of 1
at room temperature is 215 h,[6] which suggests that the un-
stable isomers of 3 could be easily characterized at room
temperature. We anticipated that using CD spectroscopy
would make it possible to evaluate the photoequilibria and
rates of thermal isomerisation.
Figure 2. Rotary cycle of 3, 4 and 5, with illustrative Newman projections
along the central olefinic bond.[14]
in Figure 2, in the stable isomers of 3 a single stereogenic
centre bearing a methyl group dictates the direction of rota-
tion. This methyl group adopts a pseudo-axial position to
minimize steric strain with the adjacent aromatic moiety in
the lower half. The motor functions as follows: absorption
of a photon (energy input) leads to cis!trans isomerisation
of the central overcrowded alkene, the axis of rotation of
the motor. This isomerisation leaves the molecule in a high-
energy state in which the methyl group occupies a pseudo-
equatorial position, where it experiences steric crowding
with the lower half of the molecule. A thermodynamically
favourable helix inversion relieves the strain, as both the
Retrosynthesis: There are three major aspects to consider in
the retrosynthesis of our motor-functionalized gold nanopar-
ticle (Scheme 1). First, the motor molecule needs two teth-
ers for attachment to the gold nanoparticle. Since each
tether contains an oxygen-sensitive thiol group, it is must be
attached at a late stage of the synthesis, preferably on a de-
protected form of model motor molecule 5. Second, we an-
ticipated that the key step in the synthetic sequence would
be the Barton–Kellogg coupling reaction,[18,19] because this
method introduces the steric strain gradually over three con-
Chem. Eur. J. 2008, 14, 11610 – 11622
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11611

B. L. Feringa et al.
Scheme 1. Retrosynthesis of surface-bound molecular motor (2’R)-(M)-4.
secutive steps starting from high-energy intermediates. The
precursors required for the Barton–Kellogg reaction are the
known hydrazone 6 and a thioketone of lower half 7. Al-
though the synthesis of this thioketone initially appears rela-
tively easy, since it involves only well-established manipula-
tions of aromatic functional groups, it requires an 1,2,3-sub-
stitution pattern in one of its building blocks.
Finally, enantiomerically pure material is needed to dem-
onstrate unidirectionality of rotary motion of the motor
anchored to the gold nanoparticle. The motor molecule
must be resolved by chiral HPLC at some stage of the syn-
thesis. This resolution is preferably performed at a late stage
of the synthesis to minimize the number of transformations
that must be performed on enantiomerically pure material.
We envisioned that the retrosynthesis shown in Scheme 1
addresses the key criteria mentioned above for a functional
motor on the surface of a nanoparticle.
Scheme 2. Initial synthesis of the thioxanthone lower half with two
methoxyl groups amenable to further functionalisation.
yielded thioketone 7 in 83% yield as a dark green solid. In
summary, this approach required ten chemical transforma-
tions to obtain thioketone 7 in 27% overall yield.
Despite the fact that this multistep sequence included sev-
eral difficult purifications, as well as steps with modest
yields, it was suitable to initially obtain moderate amounts
of parent motor 5. We subsequently sought a shorter, more
efficient route to provide a larger amount of material for
study, in particular for use in the synthesis of the desymme-
trized 13C-labelled motor. A more convenient route was de-
veloped which employs the Snieckus anionic equivalent of
the Friedel–Crafts cyclisation (Scheme 3).[23]
In the first step, 3-methoxybenzoic acid (17) was convert-
ed to amide 18 via its acid chloride followed by treatment
with diethylamine in the presence of triethylamine.[24] The
ortho-lithiation of amide 18 was described previously by
Beak and Brown, and proceeded well with a complex of sec-
butyllithium and N,N,N’,N’-tetramethylethylenediamine
(TMEDA).[25] In situ treatment of lithium species 19 with di-
sulfide 16[26] gave thioether 20 in 84% yield. Subsequent
treatment of this amide with lithium diisopropylamide
Results and Discussion
Synthesis of parent motor 5: Synthesis of thioketone 7, re-
quired for the lower half, is deceptively challenging because
of the 1,2,3-substitution pattern in one of the necessary
arene building blocks. Synthesis of the lower half of the mol-
ecule started with 3-methoxy-2-nitrobenzoic acid (8), which
was reduced to aniline 9 in a hydrogen atmosphere with pal-
ladium on carbon (Scheme 2). A Sandmeyer-type reaction[20]
sequence provided an impure sample of 2-iodo-3-methoxy-
benzoic acid (11) which was purified after esterification with
methyl iodide to give methyl ester 10. Saponification regen-
erated acid 11, which was then subjected to classical Ull-
mann conditions[21] with copper powder and potassium car-
bonate in DMF at high temperatures in the presence of 2-
methoxybenzenethiol to give aryl thioether 14. As purifica-
tion of the carboxylic acid product was complicated by for-
mation of side products, methyl iodide was added to the
cooled reaction mixture to allow isolation of methyl ester 13
in acceptable yields. The acid functionality was restored by
saponification of the ester with lithium hydroxide, which
was followed by ring closure in polyphosphoric acid (PPA)
to give ketone 15. In analogy with a procedure used by Vasi-
liu and co-workers on a ketone of similar structure,[22] treat-
ment of 15 with phosphorus pentasulfide in toluene at reflux
Scheme 3. Improved route towards lower-half ketone 16 by using the
Snieckus anionic Friedel–Crafts cyclisation.[23]
11612
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11610 – 11622

Molecular Motors on Gold Nanoparticles
FULL PAPER
(LDA) resulted in cyclisation to afford the key ketone 15 in
high yield.[27] This new synthetic route conveniently provides
ketone 15 in only three steps in an overall yield of 71%.
Thioketone 7 was then used in the key step in the synthe-
sis of the motor molecule (Scheme 4): introduction of the
in the upper half of the molecule, and of the methoxyl sub-
stituents in the lower half. Since the parent motor molecule
5 used in this study appears as two pairs of identical isomers
in the four-step cycle, it was necessary to desymmetrize the
lower half by introducing an additional spectroscopic
handle, here a OCD3 moiety to replace one of the OCH3
groups. In contrast to desymmetrizing the lower half by in-
troduction of an additional functional group, this isotopic
substitution has the advantage that the photochemical and
thermal properties of the motor should be essentially unal-
tered in comparison to those of the parent molecule. How-
ever, the methoxyl groups in 5 are further from the shielding
anisotropy of the naphthalene moiety than they are in 3,
and thus the anticipated difference in their chemical shifts
in the cis and trans isomers is reduced. Additionally, the dif-
ference in the absorption of the methoxy signals of the
stable and unstable forms was expected to be negligible.
Therefore, the unidirectionality of the rotary cycle must
be demonstrated by comparing the extent of photoconver-
sion of molecules to the unstable isomer with the extent of
isomerisation of the overcrowded alkene. Both can be deter-
Scheme 4. Final steps to parent motor molecule 5 by a Barton–Kellogg
reaction sequence, followed by resolution by HPLC on a chiral stationary
phase.
1
mined by H NMR spectroscopy; the extent of photoconver-
sion of stable to unstable isomer can be measured by deter-
mining the absorptions from the methyl substituents at the
stereogenic centre, while the extent of isomerisation of the
central olefin can be measured by means of the change in
the position of the absorption from the OMe group in the
lower half of the molecule. If the ratio of stable to unstable
matches that of the material that underwent double-bond
isomerisation, the process is unidirectional. Thus, the next
stage of the project was to synthesize an analogue of 5 in
which one of the two CH3 moieties is replaced by a CD3
group, as in alkene 38.
The synthesis of 38 benefits from the approach used in
the shorter, higher yield synthesis of thioxanthene lower
half 15 (Scheme 3). Clearly, the diastereomers must be sepa-
rated prior to introduction of the CD3 group, preferably late
in the synthesis to minimize undesired photoisomerisation.
To allow subsequent introduction of this CD3 group in the
last step, the motor molecule was desymmetrized by replac-
ing one of the two methoxyl groups with a methoxymethyl
ether (MOM) protecting group.
central olefin through
a Barton–Kellogg reaction se-
quence.[18,19] Although several protocols for the formation of
sterically overcrowded alkenes have been developed,[28] ad-
vantages of the Barton–Kellogg reaction include the possi-
bility to selectively form unsymmetrical alkenes by using
two distinct building blocks, and the gradual increase in
steric strain in the molecule through several highly exother-
mic steps. These steps begin with a 1,3-dipolar cycloaddition,
followed by ring contraction from a five- to a three-mem-
bered ring with extrusion of N2, and finally extrusion of
sulfur to give the olefin.
Oxidation of hydrazone 6 to diazo compound 21 was per-
formed with silver oxide and few drops of a saturated solu-
tion of potassium hydroxide in methanol.[6] The reaction
mixture containing 21 was then filtered[29] at 08C into a
cooled reaction flask and then treated immediately with thio-
ketone 7. These compounds react in a 1,3-dipolar cycloaddi-
tion to give a thiadiazoline, which is unstable and spontane-
ously extrudes N2 to give episulfide 22 in 47% yield. Reduc-
tive desulfurisation of 22 with copper powder in refluxing p-
xylene gave alkene 5 in 95% yield. As alkene 5 was ob-
tained as a racemate, HPLC on a chiral stationary phase
was performed to resolve a sample to obtain enantiomeri-
cally pure (2’R)-(M)-5 (configuration determined by X-ray
diffraction).
Synthesis of the desymmetrized lower-half ketone
(Scheme 5) begins with treating MOM-protected phenol[30]
23 with n-butyllithium, followed by elemental sulfur to give
thiophenol 24 in 79% yield. This thiophenol was then oxi-
Synthesis of desymmetrized motor by deuteration for stud-
ies in solution: In previous studies, unidirectionality of the
rotary cycle of related second-generation molecular motors
(such as 3, Figure 2) was confirmed by characterizing each
of the stable and unstable isomers in the rotary cycle by
1H NMR spectroscopy.[4] This was made possible by the
changes in absorption of the stereogenic methyl substituent
Scheme 5. Synthesis of dissymmetric thioxanthone lower half 27.
Chem. Eur. J. 2008, 14, 11610 – 11622
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11613

B. L. Feringa et al.
dized to disulfide 25 by using the same procedure as used to
prepare 16.[26] The lithium salt of 18 was then treated with
disulfide 25 to provide amide 26, which was then cyclized by
the Snieckus protocol to yield dissymmetric ketone 27.
The MOM group was hydrolyzed under acidic conditions,
and the corresponding phenol 28 then reprotected with ben-
zoyl chloride or acetyl chloride to give ester 29 or 30 in 78
or 50% yield, respectively (Scheme 6). The ketone moieties
Similar difficulties were encountered with the cis and
trans isomers of acetyl-protected alkene 36. Here, cis-36[32]
proved to be the more soluble of the two diastereoisomers.
Unfortunately, the maximum enrichment that could be ob-
tained of either isomer of the alkenes reflected a diastereo-
meric purity of about 80%, so these isomerically enriched
cis- and trans-alkenes 36 were used in subsequent synthesis
and experiments. For trans-36 a sample was obtained con-
taining trans-36 and cis-36 in a ratio of 83:17, and for cis-36
a sample with cis-36:trans-36 ratio of 80:20 was used. The
ester moieties on both isomers were removed by reduction
with NaBH4 to provide acceptable yields of cis- and trans-
37. Upon deprotonation by K2CO3, the phenol groups in
both stereoisomers reacted readily with [D3]methyl iodide in
DMF to give deuterium-substituted molecular motors cis-38
and trans-38. Since isomerically enriched samples were used
for the synthesis, trans-38 contained 17% of the cis isomer,
and cis-38 20% of the trans isomer. The assignment of the
configuration (cis or trans) of the diastereoisomers of 38
were confirmed by comparison of their 1H NMR spectra
with that of 5.
Synthesis of enantiopure motors on gold nanoparticles
((2’R)-(M)-4): The aliphatic tethers required to anchor the
motor molecule to the surface of the nanoparticle were in-
troduced in a six-step procedure, which was followed by
enantioresolution of intermediate 43 by HPLC on a chiral
stationary phase[33] (Scheme 7). The surprising robustness of
motor molecule 5 allowed the methoxyl groups to be depro-
tected cleanly by treatment with boron tribromide to give
bis-phenol 39.[34]
2
Scheme 6. Synthesis of H-labelled molecular motor 38.
Alkylation of 39 with two equivalents of monotosylate 40
gave diol 41 in 86% yield.[35,36] Conversion of both of the
terminal alcohol moieties of 41 to the corresponding tosy-
lates to give 42 was followed by their displacement by potas-
sium thioacetate to provide 43 in 79% yield over the two
steps.
were converted to the corresponding thioketones by heating
with P2S5 in toluene to give 31 and 32 in 87 and 62% yield,
respectively. As both thioketones degraded slowly, they
were directly treated with diazo compound 21 to give episul-
fides 33 and 34 as a mixture of cis and trans isomers. In con-
trast to the majority of struc-
turally related episulfides,[4,5,6,8]
desulfurisation of 33 and 34
did not proceed readily with
copper powder in refluxing tol-
uene. However, they were ef-
fectively desulfurized by tri-
phenylphosphine to give 35
and 36, respectively, as a mix-
ture of diastereomers.
The cis and trans isomers of
benzoyl-protected alkene 35
were inseparable by flash chro-
matography. The only method
we found for separation of
these diastereomers employed
chiral HPLC, which was im-
practical to perform on a prep-
arative scale.[31]
Scheme 7. Functionalisation of the motor molecule and anchoring onto the surface of gold colloidal particles.
11614
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11610 – 11622

Molecular Motors on Gold Nanoparticles
FULL PAPER
Fortunately, at this advanced stage of the synthesis, the
enantiomers of 43 could be separated by preparative HPLC
on a chiral stationary phase. By comparison of the CD spec-
tra of the first-eluting enantiomer of 43 and (2’R)-(M)-5, the
absolute configurations of the enantiomers of 43 were as-
signed and the synthesis continued with (2’R)-(M)-43. De-
protection of the thiol by treatment with sodium methoxide
in a mixture of THF and methanol gave dithiol (2’R)-(M)-
44, which was immediately used in the Brust–Schiffrin
method for the preparation of small gold nanoparticles.[37]
Despite the small scale on which this preparation was per-
formed, the reaction proceeded analogously to those de-
scribed in the literature to give motor-substituted colloids
(2’R)-(M)-4.[38] These colloids were purified by twice induc-
ing their precipitation from toluene by adding methanol.
Figure 3. UV/Vis spectra of (2’R)-(M)-5 (black, solid), (2’R)-(M)-4
(black, dashed) and the difference spectrum of (2’R)-(M)-4 versus the
UV/Vis spectrum of gold nanoparticles protected with octanethiol (grey).
1
Their H NMR spectrum in [D8]toluene showed very broad
signals, and no resonances associated with motor molecules
that were free in solution were observed.
motors after grafting to the gold nanoparticles, desymmetri-
sation of the motor molecule functionalized with eight-
carbon “leg” groups could in principle be achieved by deut-
eration at C1 of the leg (the CH2 group of the phenol
ethers). Our efforts to this end were complicated by the fact
that instead of clean singlets (as is observed for OCH3 in 5),
the two hydrogen atoms of the OCH2 group are diastereo-
topic, and thus they show a complex ABXY pattern in
which the signals of the stable and the unstable isomers
The UV/Vis spectrum of free motor 5 in solution was
compared with that of its gold-attached counterpart (2’R)-
(M)-4 (Figure 3). The UV/Vis spectrum of nanoparticles
protected with octanethiol[15,39] was used as a control to
allow us to compensate for the absorption spectrum of the
gold core. Thus, the UV/Vis spectrum of nanoparticles pro-
tected with octanethiol was subtracted from that of (2’R)-
(M)-4 to approximate the UV/Vis spectrum of the motor
molecules alone (while bound to gold). The subtracted spec-
trum and the spectrum of 5 are an approximate superposi-
tion of one another. Some differences between the spectra
may be expected, since the use of different thiols as ligands
would lead to small differences in the environment of the
gold core.[40]
1
overlap. This made the use of H NMR unattractive. As a
viable alternative, we introduced a 13C isotopic label at
OCH2 of the linker moiety. Ultimately, this allowed us to
verify that the amount of cis–trans isomerisation matched
the degree of conversion of stable isomer to unstable
isomer, consistent with a unidirectional cycle.[13]
Synthesis of 13C-labelled motor 55 required to demon-
strate the unidirectional rotary cycle on the nanoparticle
(vide supra) started from motor diphenol 39 (Scheme 8).
Since it was not obvious which pairs of diastereomers of the
synthetic intermediates could be separated, we believed it
Synthesis of isotopically labelled motors grafted to gold
nanoparticles: To demonstrate unidirectional rotation of
Scheme 8. Preparation of 13C-labelled motor 55 for verification of unidirectionality on the nanoparticle. The black dots denote carbon sites which are en-
riched to 50% 13C.
Chem. Eur. J. 2008, 14, 11610 – 11622
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11615

B. L. Feringa et al.
wise to separate the isomers at an earlier stage to avoid a
potentially problematic separation in one of the final steps.
Additionally, early separation has the practical advantage of
making more material available to screen different phenol
protecting groups to maximize the chromatographic separa-
tion of the diastereomers. The drawback of this approach is
that it necessitates handling the material for many synthetic
steps under light-reduced conditions to avoid photoisomeri-
sation.
Brust–Schiffren procedure.[15] These colloids were purified
by precipitation from toluene/methanol (2ꢁ), followed by
size exclusion chromatography on Sephadex LH-20 with
CHCl3/MeOH as eluent.[13]
Structural characterisation of model motor 5: To verify that
its structure was analogous to those of reported motor mole-
cules derived from overcrowded alkenes, the X-ray structure
of 5 was determined (Figure 4a) on crystals of the first-
Screening a series of monoprotected derivatives of 39 re-
vealed that introduction of a single hydrophobic triisopro-
pylsilyl (TIPS) group as in 45 gave improved chromato-
graphic separation (Scheme 8).[41] The configuration of the
overcrowded alkene was assigned in analogy with the trends
in chemical shifts[42] observed for the isomers of 38 and was
confirmed by COSY and NOESY spectroscopy.
However, attempts to alkylate trans-45 under a variety of
reaction conditions[43] led to scrambling of the TIPS protect-
ing group between the cis and trans positions and resulted in
mixtures of diastereomers of cis and trans mono- and dialky-
lated products. The presence of dialkylated (and dihydroxy)
product in the mixture excludes simple intramolecular mi-
gration of the TIPS group as the only mechanism for this
unexpected behaviour. To overcome this scrambling, the re-
maining free phenoxyl group of a mixture of cis- and trans-
45 was protected by treatment with N,N-dimethylcarbamoyl
chloride in pyridine to give carbamate trans-46 and cis-46.
The presence of the dimethyl carbamate increased the dif-
ference in Rf of diastereoisomers trans-46 and cis-46 suffi-
ciently to allow their separation by flash chromatography.
Thus, the most expedient route to pure desymmetrized ma-
terial was to carbamoylate the cis/trans mixture of 45 and
then separate the mixture of isomers. Synthesis of the de-
symmetrized 13C-labelled motor was continued with cis-46.
The TIPS group was cleaved upon treatment with TBAF in
90% yield to give trans-47. This phenolic compound was al-
kylated with (50%-13C1/50%-13C8)-8-bromooctan-1-ol[44] (48)
to give trans-49. The 13C-labelled C-8 tether was installed
first to provide an extremely sensitive way to detect the
presence of the other diastereomer. This approach proved
useful to ensure the purity of the trans diastereomer at
every step of the synthesis.[45]
Figure 4. a) Pluto drawing of (2’R)-(M)-5 based on the X-ray structure,
seen perpendicular to the central alkene. b) Numbering of the molecule.
c) Newman projection along the C9=C1 bond.
eluted fraction from the enantioseparation of (2’R)-(M)-5
and (2’S)-(P)-5 (Figure 2) by HPLC on a chiral stationary
phase. This confirmed that the methyl substituent (2’ax) at
the stereogenic center adopts a pseudo-axial conformation
to minimize the steric crowding around the central olefinic
bond (C1=C9). Additionally, the substituents attached to
the central olefin adopt an anti-folded conformation in anal-
ogy to reported second-generation molecular motors and re-
lated overcrowded alkenes.[6,49] Deviations from planarity
are slight (Figure 4c). Ultimately these steric interactions
force the molecule to adopt a helical structure, wherein the
sign of the helix is dictated by the configuration at the ste-
reogenic center.
The carbamate protecting group was removed with
LiAlH4 in THF to give phenoxy compound 50 in 78% yield,
which was then alkylated with (isotopically unenriched) 8-
bromooctan-1-ol[46] (51) in DMF with Cs2CO3 to give 52 in
83% yield. This diol was transformed to diiodide 53 in 89%
yield by treatment with triphenoxyphosphonium iodide in
DMF.[47] The diiodide was then treated with KSAc at 508C
to afford bis-thioacetate 54 in 82% yield. Dithioester 54 was
deprotected with saturated NH3 in MeOH, which smoothly
afforded dithiol 55.[48] As before, this deprotection was per-
formed immediately prior to preparation of the gold colloids
to avoid oxidation of the thiol moieties by atmospheric
oxygen.
Since the X-ray structure was obtained on enantiomeri-
cally pure material, the Flack refinement[50] [x=0.01(5)] of
the crystallographic data allowed the absolute configuration
of the first-eluting fraction to be determined as (2’R)-(M)-
5.[51,52]
Gold colloids: transmission electron microscopy (TEM) and
dynamic light scattering (DLS): Sample substrates for analy-
sis by TEM were prepared by drop-casting dilute solutions
of nanoparticles (2’R)-(M)-4 in toluene onto an amorphous
carbon film (Figure 5). Analysis of 1246 particles gave a
mean diameter of 2.01ꢀ0.3 nm.
Comparison of the CD spectrum of a solution of (2’R)-
(M)-4 in toluene with the spectroscopic data obtained for
parent compound (2’R)-(M)-5 gives the molar quantity of
As with (2’R)-(M)-4, the gold colloids decorated with mo-
lecular motor with 13C-labelled legs 55 were prepared by the
11616
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11610 – 11622

Molecular Motors on Gold Nanoparticles
FULL PAPER
the molecule standing on the particle (approximate height
2.1 nm[56]) in addition to the diameter of the gold core
(2.0 nm) would be about 6.2 nm, very close to the observed
value.
Analysis of the rotary cycle: step 1 in solution: Using first
1
UV/Vis and CD spectroscopy and then H NMR spectrosco-
py, we show that the rotary cycle of (2’R)-(M)-5 is analogous
to those of reported rotary molecular motors (Scheme 9).
The data obtained from these measurements were then
compared with analogous data from motor-decorated gold
nanoparticles (2’R)-(M)-4.
Step 1 of the rotary cycle of 5 monitored by UV/Vis and CD:
First, a solution of (2’R)-(M)-5 in toluene (1.69ꢁ10ꢁ5 m) was
irradiated with polychromatic light (T=08C, lꢂ280 nm),
and the changes in the UV/Vis and CD spectra were moni-
tored until no further change occurred, that is, the photosta-
tionary state was reached (Figure 6). The major absorption
at 283 nm in the CD spectrum changed from positive (De=
+92.6) to negative (De=ꢁ43.9), indicative of inversion of
the helicity of the molecule. This is consistent with isomeri-
sation of (2’R)-(M)-5 to (2’R)-(P)-5. In the UV/Vis spec-
trum, a decrease in the intensity of the long-wavelength ab-
sorption was also consistent with the formation of (2’R)-(P)-
5.[57] The clear isosbestic point(s) in both the CD (at l=
305 nm) and the UV/Vis spectra (at l=291, 299, 305,
321 nm) indicated clean photochemical conversion from
(2’R)-(M)-5 to (2’R)-(P)-5. From the relative absorptions of
(2’R)-(M)-5 and (2’R)-(P)-5 in their UV/Vis spectra, the
ideal wavelength for formation of the unstable isomer was
determined to be about 365 nm. Irradiation of the sample at
this wavelength (mercury line, filter 365ꢀ10 nm) indeed re-
sulted in an improved photostationary state consisting of
94% (2’R)-(P)-5 and 6% (2’R)-(M)-5 (as was determined
by 1H NMR spectroscopy in [D8]toluene). This improved
photostationary state was also supported by an increase in
the intensity of the major CD band at 283 nm (De=ꢁ59.6)
and decrease in the intensity of the absorption at longer
wavelength at 347 nm (De=4000). The photostationary state
ratios determined by 1H NMR spectroscopy (see below)
were used to calculate the CD spectrum of pure (2’R)-(P)-5.
Figure 5. a) and b) Representative transmission electron micrograph of a
drop-cast film of (2’R)-(M)-4. c) Size distribution obtained from digital
analysis of 1246 nanoparticles.
motor chromophores per gram of nanoparticle. Together
with the known particle size (TEM: d=2.01 nm) and the
density of gold (19.3 gcmꢁ3)[53] the number of overcrowded
alkenes per nanoparticle was calculated to be approximately
26ꢀ2.[54] Since a core diameter of 2.01 nm corresponds to
251 gold atoms, the overall formula of monolayer-protected
gold cluster is approximately Au251{[SACHTNUGRTNEG(NU CH2)8O]2C27H18S2}26.
To determine the size of the functionalized gold colloid
particles, DLS measurements were performed at 30.08C and
l=633.3 nm on a solution containing 1 mg of nanoparticles
in 1 mL of toluene. Prior to the measurement, the samples
were centrifuged for 5 min at 3000 rpm to remove any inter-
fering dust particles from the scattering volume. The intensi-
ty autocorrelation functions obtained by DLS were analyzed
with CONTIN.[55] The intensity-mean hydrodynamic diame-
ter of the particles was determined to be 6.3 nm. Since the
hydrodynamic diameter includes the organic dithiols 44 li-
gated to the core, this value is close to the expected one:
1
Step 1: photoisomerisation of 5 monitored by H NMR spec-
troscopy: Formation of (2’R)-(P)-5 was confirmed by
1H NMR
spectroscopy
([D8]toluene) on
a
separate
(racemic)
sample. Three
changes in the NMR spectrum
are indicative for conversion
from stable 5 to unstable 5.
The doublet for the methyl
group adjacent to the stereo-
genic center shifted downfield
from d=0.54 ppm in stable 5,
in which it occupies a pseudo-
axial position, to d=0.90 ppm
Scheme 9. Step 1 of the rotary cycle of motors in solution and on the nanoparticles.
Chem. Eur. J. 2008, 14, 11610 – 11622
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11617

B. L. Feringa et al.
UV/Vis and CD spectra of (2’R)-(M)-4 before irradiation
(solid black) and the lꢂ280 nm photostationary state
(PSSꢂ280 nm; dashed black line, reached after irradiation for
30 min) show similar changes to their counterparts in solu-
tion (Figure 7). Analogous to 5 in solution, sign inversion of
Figure 6. a) UV/Vis spectra of (2’R)-(M)-5 (solid), photostationary state
(hn=280 nm; dashed) and photostationary state (hn=365 nm; dotted) in
toluene. b) CD spectra of pure (2’R)-(M)-5 (solid), PSS280 nm (dashed),
PSS365 nm (dotted) and the calculated CD spectrum of (2’R)-(P)-5 (thick
solid, see text for details) recorded in toluene.
in unstable 5. This shift is due to the increased deshielding
experienced by the pseudo-equatorial methyl group in the
unstable isomer because of its proximity to the lower-half
arene moiety and confirms the conformational change from
a pseudo-equatorial to a pseudo-axial orientation. The sig-
nals from the aliphatic protons in the upper half changed
from d=4.26 (m, 1H), 3.58 (m, 1H) and 2.87 ppm (dd, 1H)
for stable 5 to d=3.03 (overlapping m, 2H ) and 2.27 ppm
(m, 1H). The shift of the methoxyl protons is small and the
signals are poorly resolved. Most of the aromatic signals
overlap; the only clear change in the aromatic region is an
upfield shift of the doublet at d=7.88 ppm for stable 5 to
d=7.73 ppm for unstable 5. Comparison of the relative inte-
grations of the signals from the stereogenic methyl groups in
the 1H NMR spectrum showed that the photostationary
state when irradiated with UV light with l>280 nm con-
tained 91% unstable 5 and 9% stable 5.
Figure 7. a) UV/Vis spectra of (2’R)-(M)-4 (solid), PSSꢂ280 nm (dashed)
and PSS365 nm (dotted) containing predominantly (2’R)-(P)-4 in toluene.
All spectra are adjusted for molar concentration of chromophores. b) CD
spectra of pure (2’R)-(M)-4 (solid), PSSꢂ280 nm (dashed), and PSS365 nm
(dotted) in toluene; all spectra are adjusted for molar concentration of
chromophores for comparison with (2’R)-(M)-5.
the major CD absorption at 283 nm from positive (De=
+91.0) to negative (Deꢁ46.4) is indicative of inversion of
molecular helicity from (2’R)-(M)-4 to (2’R)-(P)-4. In the UV/
Vis spectrum a decrease in the intensity of the absorption at
351 nm from e=15.2ꢁ103 to 12.4ꢁ103 dmꢁ1 molꢁ1 cmꢁ1 was
also consistent with formation of (2’R)-(P)-4. The clear iso-
sbestic point(s) in both the CD spectrum at 305 nm and in
the UV/Vis spectrum at 307 and 321 nm indicate clean pho-
tochemical conversion from (2’R)-(M)-4 to (2’R)-(P)-4. As
was observed for parent compound (2’R)-(M)-5 in solution,
irradiation at 365 nm resulted in more complete photocon-
version of (2’R)-(M)-4 to (2’R)-(P)-4 with concomitant
changes in both its UV/Vis and CD spectra.
Analysis of the rotary cycle: step 1 on nanoparticles: Using
UV/Vis and CD spectroscopies we show that the photo-
chemical step in the rotary cycle of (2’R)-(M)-4 is analogous
to what we observed with (2’R)-(M)-5 (Scheme 9).
The composition of the PSS in step 1 comprising (2’R)-
(M)-4 and (2’R)-(P)-4 was estimated by comparing the CD
spectra of the motor grafted on nanoparticles with the CD
spectra of parent structure (2’R)-(M)-5, for which the PSS
was determined by both HPLC and H NMR spectroscopy.
Examination of the CD spectra show nearly identical ab-
sorptions at 317 nm in the CD spectrum of the photostation-
Step 1: photoisomerisation of (2’R)-4 monitored by UV/Vis
and CD spectroscopy: The photoisomerisation step of
motors grafted to gold nanoparticles were studied in toluene
solution at room temperature, in analogy with (2’R)-(M)-5.
A 1.04ꢁ10ꢁ5 m solution (chromophore concentration)[58] of
(2’R)-(M)-4 in toluene was irradiated at lꢂ280 nm. The
1
11618
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11610 – 11622

Molecular Motors on Gold Nanoparticles
FULL PAPER
ary state at lꢂ280 nm of the nanoparticle solution, for ex-
ample, De=++22.2 at l=317 nm for the PSS of (2’R)-4
versus De=20.7 at l=317 nm for the PSS of (2’R)-5. This
resemblance indicates that the photoequilibrium of the
motor mounted on the Au nanoparticle is very similar to
that of the free molecule in solution, although some influ-
ence of the different chromophore environments (in solu-
tion or on Au) on the CD spectra cannot be exclud-
ed.[13,17e,f,59] Moreover, we note that the irradiation times re-
quired for surface motor chromophores 4 to reach the PSS
are more than five times longer than those for an analogous
concentration of motor chromophore 5 in solution.
Analysis of the rotary cycle: step 2 in solution: As for
step 1, we first used UV/Vis and CD spectroscopy, and then
1H NMR spectroscopy, to show that the rotary cycle of
(2’R)-(M)-5 is analogous to those of reported rotary molecu-
lar motors (Scheme 10). The data obtained from these mea-
ACHTUNGTRENNUNGsurements were then used for comparison with the analo-
gous data from the motor-decorated gold nanoparticles
(2’R)-(M)-4.
Step 2 of the rotary cycle of (2’R)-5 monitored by UV/Vis
and CD spectroscopy: After (2’R)-(M)-5 was irradiated to
its PSS365 nm to generate predominantly (2’R)-(P)-5, it was
heated at 708C for 2 h, which led to regeneration of the
original CD spectrum observed before irradiation. This ob-
servation is consistent with complete conversion of (2’R)-
(P)-5 to stable (2’R)-(M)-5, as a result of the anticipated
thermal helix inversion.[6] The kinetics of this thermal helix
inversion were examined by CD spectroscopic monitoring of
the change in CD intensity at 317 nm with time (Figure 8).
From the change in CD with time at various temperatures
(50, 60, 70, 808C), the rate constant kt was determined and
subsequently used to calculate the Gibbs free energy of acti-
vation (D°G8=94ꢀ1 kJmolꢁ1), enthalpy of activation
(D°H8=27.2ꢀ0.2 kJmolꢁ1) and entropy of activation
(D°S8=ꢁ227ꢀ4 kJmolꢁ1) by using the Eyring equation.
From these parameters it was extrapolated that (2’R)-(P)-5
has a half-life of 1.55 h at room temperature (258C).
Figure 8. a) CD absorption of irradiated samples containing (2’R)-(M)-5
(black) and (2’R)-(M)-4 (grey) in toluene at 808C, monitored at 316 nm,
plotted with respect to time. b) Eyring plot for determination of the
Gibbs free energy of activation for helix inversion of (2’R)-(P)-4 to
(2’R)-(M)-4 (grey, dashed) and (2’R)-(M)-5 to (2’R)-(P)-5 (black, solid).
R2 >0.999 for both plots.
1
and clean and complete regeneration of the H NMR spec-
trum of stable 5.
Analysis of the rotary cycle: step 2 on nanoparticles: Using
UV/Vis and CD spectroscopy we show that the thermo-
chemical step in the rotary cycle of (2’R)-(M)-4 is analogous
to that observed for (2’R)-(M)-5 (Scheme 10).
Step 2 of the rotary cycle of (2’R)-4 monitored by CD and
UV/Vis spectroscopy: Heating a sample of (2’R)-4 at its PSS
to 508C resulted in restoration of the original CD and UV/
Vis spectra. This observation is consistent with quantitative
conversion of (2’R)-(P)-4 to stable (2’R)-(M)-4 by the antici-
pated thermal helix inversion while grafted to the nanoparti-
cles. As was done for 5, the kinetics of this thermal helix in-
version were examined by CD spectroscopy by monitoring
the change in CD intensity at
1
Step 2 of the rotary cycle of 5 monitored by H NMR spec-
troscopy: Warming a sample of 5 at its PSS in [D8]toluene
(containing predominantly unstable 5) to 608C for 2 h led to
disappearance of the absorptions attributed to unstable 5,
317 nm with time at various
temperatures (T=50, 60, 70,
808C). From this data, the rate
constant kt was determined
and subsequently used to cal-
culate the Gibbs free energy of
activation
(D°G8=96ꢀ
2 kJmolꢁ1), enthalpy of activa-
tion (D°H8=37ꢀ1 kJmolꢁ1)
and entropy of activation
Scheme 10. Step 2 of the rotary cycle of motors in solution and on gold nanoparticles.
(D°S8=200ꢀ4 kJmolꢁ1)
by
Chem. Eur. J. 2008, 14, 11610 – 11622
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11619

B. L. Feringa et al.
using the Eyring equation. This corresponds to a half-life of
3.3 h at room temperature (258C) for conversion of (2’R)-
(P)-4 to (2’R)-(M)-4. This energy barrier is slightly higher
than that found for (2’R)-(P)-5 and results in an approxi-
mate twofold increase of the half-life for thermal helix in-
version at room temperature. One possible origin of this
slight change in behaviour is the restricted flexibility of the
sterically overcrowded alkene when mounted on a gold sur-
face. At elevated temperatures this effect becomes less pro-
nounced, and at 808C the half-lives of the thermal helix in-
version are very similar: 704 and 725 s for (2’R)-(P)-5 and
(2’R)-(P)-4, respectively.
Overnight heating of the photostationary-state sample at
708C resulted in complete conversion of unstable trans-38
and unstable cis-38 to stable trans-38 and stable cis-38, re-
spectively. The ratio of stable trans-38 and stable cis-38 after
irradiation and subsequent heating was determined by
1H NMR spectroscopy to be 75:25, which is consistent with
isomerisation of the unstable isomers to the corresponding
stable isomers. The portion of trans isomer (75%) originates
from the sum of the unstable trans-38 (73%), which isomer-
izes to the stable form during heating, and the stable isomer
already present at the photoequilibrium (an additional 2%).
Similarly, the 25% of the cis isomer originates from the sum
of the unstable cis-38 (18%), which isomerizes to the stable
form during heating, and the stable isomer already present
in the PSS (an additional 7%).
1H NMR spectroscopic analysis of deuterated motor 38 to
confirm the unidirectional cycle: The photo- and thermal
1
isomerisation processes of 38 were monitored by H NMR
The same irradiation experiment was performed on a
sample composed of 83% stable trans-38 and 17% stable
cis-38. After irradiation (lꢂ280 nm, T=08C) a mixture was
obtained containing unstable cis-38, unstable trans-38, stable
trans-38 and stable cis-38 in a ratio of 76:15:7:2. Heating
(T=708C) of this sample also resulted in complete conver-
sion of unstable cis-38 and unstable trans-38 to stable cis-38
and stable trans-38, which were obtained in a ratio of 78:22.
By means of these two experiments the full rotary cycle
was demonstrated (Figure 9). Either stable isomer of 38 can
be converted in 91% selectivity in a photochemical step fol-
lowed by a thermal step resulting in the formation of the
other stable isomer of 38. During this process, the respective
intermediate unstable isomers of 38 are readily detected,
which confirms the unidirectionality of the rotary process
around the central double bond. Since the difference be-
tween a protonated or a deuterated methoxy substituent is
not significant in these types of cis!trans photo-isomerisa-
tion processes, the unidirectionality of rotation of isotopical-
ly labelled 38 also demon-
spectroscopy to verify that its rotary cycle is unidirectional
(Figure 9). Comparison of the relative integrals of the dia-
stereomers by 1H NMR spectroscopy confirmed that the
cycle is unidirectional, in analogy with related motor mole-
cules.[6,8,60]
Irradiation (lꢂ280 nm, T=08C) of a sample containing
stable cis-38 and stable trans-38 (in 80:20 ratio) in
[D8]toluene led to formation of a photostationary state con-
taining a 91:9 mixture of “total unstable isomers”:“total
stable isomers”, as expected from the analogous experi-
ments performed on non-deuterated motor 5. The mixture
comprised unstable trans-38, unstable cis-38, stable cis-38
and stable trans-38 in a ratio of 73:18:7:2, respectively. This
ratio of products is consistent with establishment of two sep-
arate photo-equilibria: one between unstable trans-38 and
stable cis-38 (in a ratio of 73:7, close to 91:9) and the other
between unstable cis-38 and stable trans-38 (in a ratio of
18:2, also close to 91:9).
strates that model motor 38
can undergo unidirectional ro-
tation around the central
olefin.
Previously, we reported a de-
tailed study employing 13C-la-
belled motor 55 grafted to
nanoparticles and 1H and
13C NMR spectroscopy to dem-
onstrate that the rotary cycle
of the motor attached to the
gold colloids (2’R)-(M)-4 is
unidirectional.[13] We found
that the extent of cis!trans
isomerisation was identical to
the extent of isomerisation of
stable to unstable isomer.[13]
Figure 9. Rotary cycle of deuterated model compound 38, focussing on the first photochemical and thermal
steps for clarity.
11620
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11610 – 11622

Molecular Motors on Gold Nanoparticles
FULL PAPER
[6] a) N. Koumura, E. M. Geertsema, A. Meetsma, B. L. Feringa, J. Am.
Chem. Soc. 2000, 122, 12005; b) N. Koumura, E. M. Geertsema,
M. B. van Gelder, A. Meetsma, B. L. Feringa, J. Am. Chem. Soc.
2002, 124, 5037.
[7] Motor systems driven by chemical transformations: a) S. P. Fletcher,
F. Dumur, M. M. Pollard, B. L. Feringa, Science 2005, 310, 80; b) Y.
Lin, B. J. Dahl, B. P. Branchaud, Tetrahedron Lett. 2005, 46, 8359–
8362; c) B. J. Dahl, B. P. Branchaud, Org. Lett. 2006, 8, 5841; d) T. R.
Kelly, H. De Silva, R. A. Silva, Nature 1999, 401, 150; e) T. R. Kelly,
X. Cai, F. Damkaci, S. B. Panicker, B. Tu, S. M. Bushell, I. Cornella,
M. J. Piggott, R. Salives, M. Cavero, Y. Zhao, S. Jasmin J. Am.
Chem. Soc. 2007, 129, 376.
Conclusion
We have reported the synthesis of a deuterium-labelled mo-
lecular motor and a 13C-labelled molecular motor. These iso-
topically labelled molecules were essential to demonstrate
that the unidirectional rotary cycle of a light-driven rotary
molecular motor was preserved when grafted to the surface
of a gold nanoparticle. Thus, we have shown that rotation of
the upper half relative to the lower half is effectively trans-
lated to absolute rotation of the “rotor” with respect to the
gold nanoparticle.
[8] M. M. Pollard, M. Klok, D. Pijper, B. L. Feringa, Adv. Funct. Mater.
2007, 17, 718.
Although the motors are each anchored onto the surface
through two tethers, the molecules have essentially the same
photophysical behaviour as the model motors that function
in solution. Comparison of the UV/Vis and CD spectroscop-
ic data obtained for free molecular motors in solution
((2’R)-(M)-5) and when anchored to gold nanoparticles
((2’R)-(M)-4) indicate that the photoequilibria are similar,
while the thermal isomerisation step is retarded slightly
[9] M. M. Pollard, M. Lubomska, P. Rudolf, B. L. Feringa, Angew.
Chem. 2007, 119, 1278; Angew. Chem. Int. Ed. 2007, 46, 1300.
[10] For selected examples, see a) J. Berna, D. A. Leigh, M. Lubomska,
S. M. Mendoza, E. M. Pꢃrez, P. Rudolf, G. Teobaldi, F. Zerbetto,
Nat. Mater. 2005, 4, 704; b) T. D. Nguyen, H.-R. Tseng, P. C. Celes-
tre, A. H. Flood, Y. Liu, J. F. Stoddart, J. I. Zink, Proc. Natl. Acad.
Sci. USA 2005, 102, 10029; c) Y. Liu, A. H. Flood, P. A. Bonvallet,
S. A. Vignon, B. H. Northrop, H.-R. Tseng, J. O. Jeppesen, T. J.
Huang, B. Brough, M. Baller, S. Magonov, S. D. Solares, W. A. God-
dard, C.-M. Ho, J. F. Stoddart, J. Am. Chem. Soc. 2005, 127, 9745.
[11] N. H. Katsonis, M. Lubomska, M. M. Pollard, B. L. Feringa, P.
Rudolf, Prog. Surf. Sci. 2007, 82, 407.
[12] Molecular Motors, (Ed.: M. Schliwa), Wiley-VCH, Weinheim, Ger-
many, 2003.
[13] For a preliminary report on this work, see: R. A. van Delden,
M. K. J. ter Wiel, M. M. Pollard, J. Vicario, N. Koumura, B. L. Ferin-
ga, Nature 2005, 437, 1337.
1
below 808C. Using H and 13C NMR spectroscopy on isotop-
ically labelled compounds, we were able to show unequivo-
cally that unidirectionality of rotation is preserved for sur-
face-anchored motor molecules. We anticipate that this
work is a promising step toward harnessing the unique mo-
lecular-scale controlled rotary motion generated by these
motors.[9]
[14] Three-dimensional models for 3 through its rotary cycle can be
found in ref. [6].
[15] a) M. Brust, M. Walker, D. Behell, D. J. Schiffrin, R. Whyman, J.
Chem. Soc. Chem. Commun. 1994, 801; b) M.-C. Daniel, D. Astruc,
Chem. Rev. 2004, 104, 293.
Experimental Section
[16] P. V. Kamat, J. Phys. Chem. B 2002, 106, 7729.
Experimental procedures and characterisation for all new compounds, in-
1
cluding copies of their H and 13C NMR spectra and TEM data for (2’R)-
[17] a) J. Zhang, J. K. Whitesell, M. A. Fox, Chem. Mater. 2001, 13, 2323;
b) J. Zhang, J. K. Whitesell, M. A. Fox, J. Phys. Chem. A 2003, 107,
6051; c) T. Kudernac, S. J. van der Molen, B. J. van Wees, B. L. Ferin-
ga, Chem. Commun. 2006, 3597; d) A. Manna, P-L. Chen, H. Akiya-
ma, T-X. Wei, K. Tamada, W. Knoll, Chem. Mater. 2003, 15, 20;
e) J. H. Hu, F. Liu, K. Kittredge, J. K. Whitesell, M. A. Fox, J. Am.
Chem. Soc. 2001, 123, 1464; f) S. D. Evans, S. R. Johnson, H. Rings-
dorf, L. M. Williams, H. Wolf, Langmuir 1998, 14, 6436; g) M. O.
Wolf, M. A. Fox, Langmuir 1996, 12, 955.
(M)-4, are provided in the Supporting Information.
Acknowledgements
[18] a) D. H. R. Barton, B. J. Willis, J. Chem. Soc. Perkin Trans. 1 1972,
305; b) R. M. Kellogg, S. Wassenaar, J. Buter, Tetrahedron Lett.
1970, 11, 4689; c) R. M. Kellogg, S. Wassenaar, Tetrahedron Lett.
1970, 11, 1987; d) R. M. Kellogg, J. Org. Chem. 1972, 37, 4045;
e) D. H. R. Barton, B. J. Willis, J. Chem. Soc. Chem. Commun. 1970,
1225; f) R. M. Kellogg, Tetrahedron 1976, 32, 2165; g) J. Buter, S.
Wassenaar, R. M. Kellogg, J. Org. Chem. 1972, 37, 4045.
[19] The Barton–Kellogg reaction was first developed by Staudinger and
Siegwart: a) H. Staudinger, J. Siegwart, Helv. Chim. Acta 1920, 3,
833. The reaction was later explored by A. Schçnberg et al.; for
leading references see: b) A. Schçnberg, D. Cernik, W. Urban, Ber.
Dtsch. Chem. Ges. B 1931, 64, 2577; c) A. Schçnberg, E. Frese,
Chem. Ber. 1968, 101, 701.
Financial support from the Netherlands Organisation for Scientific Re-
search (NWO-CW) through a NWO-CW Top-grant, Spinoza grant from
NWO and a Nanoned grant is gratefully acknowledged. J. Vicario thanks
the Departamento de Educaciꢂn, Universidades e Investigaciꢂn del Go-
bierno Vasco for a postdoctoral fellowship.
[1] F. Diederich, Angew. Chem. 2007, 119, 68; Angew. Chem. Int. Ed.
2007, 46, 68.
[2] R. D. Astumian, Phys. Chem. Chem. Phys. 2007, 9, 5067.
[3] For progress on molecular motors at work, see a) W. R. Browne,
B. L. Feringa, Nature Nanotech. 2006, 1, 25; b) for a comprehensive
review, see: E. R. Kay, D. A. Leigh, F. Zerbetto, Angew. Chem.
2007, 119, 72; Angew. Chem. Int. Ed. 2007, 46, 72; c) recent review:
B. L. Feringa, J. Org. Chem. 2007, 72, 6635; d) V. Balzani, A. Credi,
M. Venturi, ChemPhysChem 2008, 9, 202.
[4] a) N. Koumura, R. W. J. Zijlstra, R. A. van Delden, N. Harada, B. L.
Feringa, Nature 1999, 401, 152; b) M. K. J. ter Wiel, R. A. van Del-
den, A. Meetsma, B. L. Feringa, J. Am. Chem. Soc. 2005, 127, 14208.
[5] a) M. K. J. ter Wiel, R. A. van Delden, A. Meetsma, B. L. Feringa, J.
Am. Chem. Soc. 2003, 125, 15076; b) T. Fujita, S. Kuwahara, N.
Harada, Eur. J. Org. Chem. 2005, 4533; c) S. Kuwahara, T. Fujita, N.
Harada, Eur. J. Org. Chem. 2005, 4544.
[20] H. H. Hodgson, Chem. Rev. 1947, 40, 251.
[21] J. Lindley, Tetrahedron 1984, 40, 1433.
[22] G. Vasiliu, N. Rasanu, O. Maior, Rev. Chim. 1968, 19, 561, Chem.
Abstr. 1969, 71, 38739.
[23] a) R. J. Mills, N. J. Taylor, V. Snieckus, J. Org. Chem. 1989, 54, 4372;
b) M. C. Whisler, S. MacNeil, V. Snieckus, P. Beak, Angew. Chem.
2004, 116, 2256; Angew. Chem. Int. Ed. 2004, 43, 2206.
[24] This compound has been reported in the literature, but no experi-
mental details were provided: P. Grammaticakis, Bull. Soc. Chim.
Fr. 1964, 924.
Chem. Eur. J. 2008, 14, 11610 – 11622
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11621

B. L. Feringa et al.
[25] P. Beak, R. A. Brown, J. Org. Chem. 1982, 47, 34.
to the unstable form of the opposite diastereomer, which was not
evident from the 1H NMR spectra of the precursor. These inter-
mediates were surprisingly sensitive to light.
[26] Disulfides 17 and 25 were prepared by a procedure of N. A. Nourel-
din, M. Caldwell, J. Hendry, D. G. Lee, Synthesis 1998, 1587. Disul-
fide 17 has been reported previously: L. Gattermann, Ber. Dtsch.
Chem. Ges. 1899, 32, 1136.
[27] For leading references on this procedure, see: a) J. M. Fu, B. P.
Zhao, M. J. Sharp, V. Snieckus, J. Org. Chem. 1991, 56, 1683; b) M.
Gray, B. J. Chapell, N. J. Taylor, V. Snieckus, Angew. Chem. 1996,
108, 1609; Angew. Chem. Int. Ed. Engl. 1996, 35, 1558; c) O. B. Fam-
ilioni, I. Ionica, J. F. Bower, V. Snieckus, Synlett 1997, 1081.
[28] a) J. E. McMurry, Chem. Rev. 1989, 89, 1513; b) A. Furstner, B. Bog-
danovic, Angew. Chem. 1996, 108, 2582; Angew. Chem. Int. Ed.
Engl. 1996, 35, 2442.
[46] B. Maurer, A. Grieder, Helv. Chim. Acta 1977, 60, 1155.
[47] The iodide was used here because it led to a much cleaner and
faster reaction.
[48] When sodium methoxide was used in distilled, degassed methanol, it
was found that ca. 10% had nevertheless been oxidized to the disul-
fide. No disulfide was observed when the ammonia solution was
used.
[49] I. Agranat, P. U. Biederman, J. J. Stezowski, Eur. J. Org. Chem.
2001, 15–34.
[50] a) H. D. Flack, Acta Crystallogr. Sect. A 1983, 39, 876–881; b) H. D.
Flack, G. Bernardinelli, Acta Crystallogr. Sect. A 1999, 55, 908–915.
[51] Crystal data for (2’R)-(M)-4: C29H24O2S2; M=468.64; colorless, par-
allelpiped; monoclinic; space group P21 (no. 4); a=8.9625(6), b=
22.785(2), c=11.6435(8) ꢄ; b=90.127(1)8; V=2377.7(3) ꢄ3; Z=4;
1calcd =1.309 g·cmꢁ3; T=110(2) K; m=2.49 cmꢁ1; number of reflec-
tions: 11841; number of refined parameters: 786; final agreement
factors: wR(F2)=0.0863, R(F)=0.0420, GoF=1.036. CCDC-693600
contains the supplementary crystallographic data for this paper.
These data can be obtained free of charge from The Cambridge
Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/
cif.
[52] This is the first example described in the literature in which the ab-
solute configuration of a second-generation motor molecule has
been determined by a method other than CD spectroscopy.
[53] CRC Handbook of Chemistry and Physics (Ed. D. R. Lide), CRC
Press, Boca Raton, 2003–2004, 84, pp. 4–59. We acknowledge that
this value may differ (ꢀ10%) from the density of gold in the nano-
particle: O. D. Hꢅberlen, S.-C. Chung, M. Stener, N. Rçsch; J.
Chem. Phys. 1997, 106, 5189.
[29] This diazo compound is unstable and begins to undergo side reac-
tions upon warming to room temperature.
[30] 1H and 13C NMR spectroscopic data: R. S. E. Conn, A. V. Lovell, S.
Karaday, L. M. Weinstock, J. Org. Chem. 1986, 51, 4710. First prepa-
ration: Breslauer, Pictet, Ber. Dtsch. Chem. Ges. 1907, 40, 3785.
[31] This method necessitated the use of such an apolar eluent (hep-
ACHTUNGTRENNUNGtane:iPrOH 98:2) that the low solubility of 35 rendered this ap-
proach prohibitively time consuming and wasteful of solvent.
[32] This assignment is based on the position of the acetyl moiety rela-
tive to the naphthalene moiety in the upper half.
[33] HPLC on a Chiralcel AD column as stationary phase and n-hep-
A
enantiomers were 12.7/14.9 min.
[34] The relatively low solubility of 39 in common organic solvents made
it hard to purify.
[35] Tosylate 40 was prepared by the reaction of tosyl chloride with 1,8-
octanediol. Experimental details for its synthesis are given in the
Supporting Information.
[36] S. K. Nayak, Synthesis 2000, 11, 1575.
[37] a) M. Brust, M. Walker, D. Bethell, D. J. Schiffrin, R. Whyman, J.
Chem. Soc. Chem. Commun. 1994, 801; b) M. Brust, J. Fink, D. Be-
thell, D. J. Schiffrin, C. J. Kiely, J. Chem. Soc. Chem. Commun. 1995,
1655. M.-C. Daniel, D. Astruc, Chem. Rev. 2004, 104, 293.
[38] For the sake of clarity, only one motor is depicted, whereas TEM
and DLS data suggest the presence of an average of 26 motors per
nanoparticle.
[54] Allowing for a possible error in molar CD measurements of ꢀ5%.
[55] S. W. Provencher, Comput. Phys. Commun. 1982, 27, 229.
[56] Calculated by using a structure of 44 minimized by MM2.
[57] A reduction in the intensity of the long-wavelength absorption was
observed upon formation of the unstable isomer of 2.[6]
[58] Calculated by comparing the intensity of the CD absorption at
317 nm.
[39] B. S. Zelakiewicz, A. C. de Dios, Y.Y. Tong, J. Am. Chem. Soc. 2003,
125, 18.
[59] A number of authors have reported CD spectra of metal nanoparti-
cles protected with chiral nonracemic ligands, but none have de-
scribed any effects of macroscopic anisotropy. Selected examples:
a) T. G. Schaaff, R. L. Whetten, J. Phys. Chem. B 2000, 104, 2630;
b) C. Gautier, T. Bꢆrgi, J. Am. Chem. Soc. 2008, 130, 7077; c) M.
Tamura, H. Fujihara, J. Am. Chem. Soc. 2003, 125, 15742; d) H.
Yao, K. Miki, N. Nishida, A. Sasaki, K. Kimura, J. Am. Chem. Soc.
2005, 127, 15536.
[60] The composition of the photostationary state mixture was deter-
mined by comparison of the relative integrals of the absorptions of
the stable and unstable isomers, specifically by using protons of the
methyl substituent at the stereogenic center (d=0.54 in the stable
isomer vs. 0.90 ppm in the unstable isomer) and the aromatic pro-
tons at d=7.6–7.9 ppm to compare the ratio of stable to unstable,
and the absorptions of the methoxyl groups to evaluate cis:trans
ratios.
[40] The absorption of gold is quite sensitive to its environment: K. Wa-
tanabe, D. Menzel, N. Nilius, H.-J. Freund, Chem. Rev. 2006, 106,
4301. A. Moores, F. Goettmann, New J. Chem. 2006, 30, 1121.
[41] Preparative HPLC was still required to access preparative amounts
of diastereomerically pure material.
[42] Protons on the aromatic lower half positioned on the same side as
the naphthalene moiety are shielded and their signals are shifted up-
field.
[43] Typical aprotic conditions were explored, including Cs2CO3 in DMF
with 8-bromooctan-1-ol or 8-iodooctan-1-ol. Amine bases such as
diisopropylethylamine were also employed to no apparent advant-
age.
[44] Compound 48 was prepared from K13CN and 7-bromooctan-1-ol in
five steps. See Supporting Information for synthetic procedures.
[45] The synthesis was performed initially by installing the 13C-labelled
linker second. However, when this was performed it was later found
that a small amount (ca. 5%) of the material had photoisomerized
Received: April 30, 2008
Revised: July 31, 2008
Published online: November 13, 2008
11622
www.chemeurj.org
ꢀ 2008 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2008, 14, 11610 – 11622