Communications
Surface Chemistry
Angewandte
Chemie
International Edition: DOI: 10.1002/anie.201804110
German Edition:
DOI: 10.1002/ange.201804110
On-Surface Bottom-Up Synthesis of Azine Derivatives Displaying
Strong Acceptor Behavior
Nerea Ruiz del _rbol, Irene Palacio, Gonzalo Otero-Irurueta, Jos8 I. Mart&nez,
Pedro L. de Andr8s, Oleksander Stetsovych, Mar&a Moro-Lagares, Pingo Mutombo,
Martin Svec, Pavel Jel&nek, Albano Cossaro, Luca Floreano, Gary J. Ellis, Mar&a F. Llpez, and
Jos8 A. Mart&n-Gago*
Abstract: On-surface synthesis is an emerging approach to
obtain, in a single step, precisely defined chemical species that
cannot be obtained by other synthetic routes. The control of the
electronic structure of organic/metal interfaces is crucial for
defining the performance of many optoelectronic devices. A
facile on-surface chemistry route has now been used to
synthesize the strong electron-acceptor organic molecule
quinoneazine directly on a Cu(110) surface, via thermally
activated covalent coupling of para-aminophenol precursors.
The mechanism is described using a combination of in situ
surface characterization techniques and theoretical methods.
Owing to a strong surface-molecule interaction, the quinoneazine molecule accommodates 1.2 electrons at its carbonyl ends,
inducing an intramolecular charge redistribution and leading
to partial conjugation of the rings, conferring azo-character at
the nitrogen sites.
Organic heterostructures based on electron acceptor–donor
organic molecules on surfaces have become strategic materi[*] N. Ruiz del ^rbol, I. Palacio, J. I. Mart&nez, P. L. de Andr8s,
M. F. Lkpez, Prof. J. A. Mart&n-Gago
ESISNA Group, Materials Science Factory. Institute of Materials
Science of Madrid (ICMM-CSIC)
Sor Juana In8s de la Cruz 3, 28049 Madrid (Spain)
E-mail: gago@icmm.csic.es
G. Otero-Irurueta
Centre for Mechanical Technology and Automation (TEMA)
University of Aveiro
3810-193 Aveiro (Portugal)
O. Stetsovych, M. Moro-Lagares, P. Mutombo, M. Svec, P. Jel&nek
Institute of Physics, Academy of Sciences of the Czech Republic
Cukrovarnicka 10, 1862 53 Prague (Czech Republic)
A. Cossaro, L. Floreano
Laboratorio TASC, CNR-IOM
Basovizza SS-14, Km 163.5, 34149 Trieste (Italy)
G. J. Ellis
Polymer Physics Group
Institute of Polymer Science and Technology (ICTP-CSIC)
Juan de la Cierva 3, 28006 Madrid (Spain)
Supporting information and the ORCID identification number(s) for
the author(s) of this article can be found under:
https://doi.org/10.1002/anie.201804110.
T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co.
KGaA. This is an open access article under the terms of the Creative
Commons Attribution-NonCommercial License, which permits use,
distribution and reproduction in any medium, provided the original
work is properly cited and is not used for commercial purposes.
8582
als owing to their huge technological impact in fields such as
organic light-emitting diodes (OLEDs), organic field effect
transistors (OFETs), or solar cell devices, amongst others. In
electronic devices, organic layers are placed on metallic
surfaces for electrical contact, and the structure of the metal–
organic interface enormously affects the performance.[1] In
particular, some molecules promote charge transfer at the
interface with metal electrodes owing to their donor
(acceptor) nature, which may induce energy level realignment
that can be exploited to tune the transport properties of the
system.[1–5] Despite the potential impact of this approach, only
a few molecules have been shown to efficiently donate (or
accept) significant charge, this quality being related to the
presence of donor (acceptor) moieties in their structures.[6] A
typical example is tetracyano-p-quinodimethane (TCNQ),
a strong acceptor molecule that when deposited on Cu(100)
accommodates around 1.6 electrons whereby almost one
electron aromatizes the central hexagonal ring and the
remaining fraction of the charge is accommodated in one of
the peripheral nitrogen atoms of the cyano groups.[3]
On the other hand, on-surface synthesis can generate
unique molecules or extended molecular architectures that
have been rationally formed via alternative synthetic routes
to those available through solution-based chemistry.[7, 8] These
surface-stabilized species can lead to compounds that are
difficult or impossible to obtain via conventional synthetic
procedures.[9–12]
Herein, we combine both of the aforementioned features.
We show that the on-surface coupling reaction of two simple
and inexpensive para-aminophenol (p-Ap) molecules can be
employed to form a quinonoid-like derivative, quinoneazine
(QAz). This molecule has been theoretically proposed for
organic electrodes owing to their extreme redox voltages, and
it can also be employed as an intermediate in the preparation
of several chemically and biologically active compounds.[13, 14]
We show that, similar to the case of TCNQ molecules, the onsurface synthesized QAz molecule can accept about 1.2
electrons from the Cu surface through a strong interaction
with the surface atoms. We observe that the role of the surface
is two-fold. Firstly, it catalyzes the synthesis of QAz, a nonaromatic compound difficult to obtain by conventional synthesis routes.[15] Secondly, it stabilizes the molecular structure
by donating to the QAz more than one electron to form
a resonant structure, see Scheme 1, as we will discuss below.
We used a combination of several surface-science experimental techniques, including synchrotron radiation-based X-
T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2018, 57, 8582 –8586
Communications
Scheme 1. Azine-coupling of p-aminophenol precursors (p-Ap) to form
quinoneazine (QAz) via thermally induced reaction of two p-aminophenol molecules on Cu(110) surfaces.
ray photoemission spectroscopy (XPS) and near-edge X-ray
absorption fine structure (NEXAFS), scanning tunneling
microscopy (STM), and non-contact atomic force microscopy
(nc-AFM), supported by first-principles theoretical calculations (details are provided in the Supporting Information).
This large battery of tools allows us to obtain a full and
consistent picture of this unusual on-surface chemical reaction and the strong charge redistribution process occurring at
the organic–metal interface.
The starting point of our method is the evaporation in
ultra-high vacuum (UHV) of p-Ap molecules on an atomically clean Cu(110) surface. Depending on the surface
temperature during evaporation, we observe two distinct
cases: the adsorption at room temperature leads to individual
p-Ap molecules adsorbed on the surface (RT phase, Figure 1 a), and the evaporation onto the surface at 520 K (HT
phase, Figure 1 b), where molecular structures exhibiting
long-range ordered and a uniform, well-defined surface
morphology is found. The HT phase can also be obtained
by direct annealing of the RT phase at 520 K, displaying
equivalent results.
The size and shape of the circular features dispersed
across the surface observed in the STM image of Figure 1 a
correspond to single p-Ap molecules,[16] and the nc-AFM
image (in the inset) clearly shows the hexagonal carbon rings
confirming the presence of single isolated molecules. They
form locally a 4 X 4 superperiodicity. In contrast, in the STM
image in Figure 1 b the surface molecular species observed are
larger and elliptical in shape and of uniform size, with a bright
Figure 1. STM images (7.7 W 7.7 nm2). a) Constant-current image
recorded at RT with Vbias = + 1 V and Itunnel = 141 pA, where the single
bright spots are individual p-Ap molecules. The inset shows a nc-AFM
with a functionalized tip of some adsorbed p-Ap molecules, and
b) constant height image recorded at 5 K after deposition of p-AP at
520 K, where larger molecular species with clearly defined size, and
orientation are seen. (1 mV,5 pA). Inset: a LEED pattern exhibiting
a long-range order with a [(5, 1), (@1, 2)] symmetry with the substrate
(electron energy = 33 eV).
Angew. Chem. Int. Ed. 2018, 57, 8582 –8586
Angewandte
Chemie
feature in the center. Further, they are well-organized into
precisely aligned linear rows oriented at an angle of 2088 with
respect to the [001] surface direction (Supporting Information, Figure S3). The LEED pattern (see inset) indicates that
the molecular arrangement is commensurate with the substrate suggesting that the Cu(110) crystal termination plays
a fundamental role in the process of the formation of the HT
phase.
The chemical structure of the RT and HT phases can be
unequivocally followed in situ by high-resolution XPS and
NEXAFS (see the Supporting Information). Figure 2 shows
the XPS core-level peaks of the elements of the p-Ap
precursors, N 1s, O 1s, and C 1s, and their evolution with
coverage and temperature. The C 1s spectrum displays no
significant core-level shift with respect to the multilayer and
in all cases the components associated with the four sp2
carbons cannot be individually resolved, in agreement with
a recent study on hydroxycyanobenzene,[17] and indicating
Figure 2. a),b),c) N 1s, O 1s, and C 1s XPS spectra for a physisorbed
multilayer (green curve) self-assembled monolayer at RT (red curve, RT
phase) and at 520 K (blue curve, HT phase). The photon energies are,
500, 650, and 400 eV, respectively. d) NEXAFS spectra for C k-edge for
the same phases recorded at s-polarization (electric field vector
parallel to the surface) and p-polarization (electric field vector perpendicular to the surface).
T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8583
Communications
that the atoms of the carbon ring do not participate in the
surface chemical reaction. On the contrary, the N 1s and O 1s
peaks in the (sub)monolayer RT phase display large corelevel shifts (in opposite directions), indicating that the p-Ap
precursor adsorbs in its oxidized form (red spectra in
Figure 2). Both hydroxy and primary amine moieties of the
p-Ap molecules deprotonate to generate phenoxy and
secondary amine groups.[18, 19] However, when p-Ap is dosed
at high temperature (520 K, blue curve labeled HT in
Figure 2), a significant shift to lower binding energy of the
N 1s peak with respect to the RT layer indicates the full
oxidation of the amine terminations into their iminic
form.[20, 21]
The overall intensities of the C 1s, N 1s, and O 1s peaks
show no significant change upon heating, indicating that the
process does not affect the stoichiometry of the molecular
layer. However, an overall small shift of the C 1s spectrum
towards lower binding energy is observed, indicating that
carbon atoms have accepted charge.[21, 22] We may conclude
that a dimerization reaction between two adjacent p-Ap
molecules takes place via surface-mediated thermally induced
covalent coupling of the NH species, to generate an azine
bond leading to the QAz molecule.
To obtain further insights into the structure of the HT
phase and the role of the surface, we used NEXAFS
spectroscopy. By varying the surface orientation with respect
to the linear polarization of the incident light in NEXAFS, the
spectral response corresponding to the p*-symmetry components shows dichroic anisotropy (Figure 2 d).[23] The curve
that corresponds to the multilayer p-Ap presents no important variations with polarization. However, both RT and HT
phases show a strong enhancement of the p* part of the
spectra in p-polarization and quenching in s-polarization. This
dichroic behavior indicates that the carbon ring is oriented
parallel to the metal surface, with an average tilt of 588, in good
agreement with the STM images and calculations (Figure 3 a).
Moreover, the change of shape of the NEXAFS spectra
(overall quenching and shift to lower energy of the first
resonance) indicates a net charge transfer from the surface to
the lowest molecular orbital (LUMO) localized on the carbon
ring.
The nc-AFM images of the HT phase (Figure 3 b) show
intramolecular features with two bright protrusions that can
be assigned to two equivalent ring structures linked together
as a covalent dimer. Simulations of the nc-AFM images using
a probe particle model[24] are in close agreement with the
experimental data. Moreover, although an apparent chainlike structure is often seen in STM images, simulations rule
out any polymerization of the precursors, and the chain
appearance of the images is related to a modification of the
electronic properties of the last Cu layer owing to the strong
interaction with the QAz molecule (Supporting Information,
Figure S10).
DFT calculations show that QAz prefers to be stacked
along a direction forming a 2088 angle with respect to the [001]
crystallographic axis of the Cu substrate, in agreement with
the LEED measurements. The optimized ground-state structure shows both terminating oxygen atoms at nearly-on-top
positions of the Cu atoms, at an average perpendicular
8584
www.angewandte.org
Angewandte
Chemie
Figure 3. a) DFT optimized geometry of QAz molecules on Cu(110),
represented for only the first layer of the substrate. The main symmetry
axis of QAz (from one oxygen to the other presented in the structure)
are at 5288 with respect to the [110] direction of the substrate. The
oxygen atoms are bound to the Cu (110) at bridge positions. The
substrate directions are indicated in the Scheme and the unit cell
indicated in yellow. b) Frequency-shift (force) image acquired by ncAFM/STM with a functionalized probe at 5 K. c) AFM image simulation of optimized QAz/Cu(110) configuration using the probe particle
AFM model. d) Experimental STM image of QAz formed in the HT
phase recorded at RT. Vbias = + 1350 mV and Itunnel = 31 pA. e) Computed Keldish–Green STM image under the same experimental conditions as (d).
distance from the substrate of around 2.0 c. This configuration steers both N atoms to a symmetrical bridge site. Both
carbon rings are essentially symmetrical with a slight axial tilt
of 4.588 along the stacking direction, in good agreement with
the experimental NEXAFS value of about 588. A N@N bond
length of 1.31 c was obtained that is slightly longer than
values reported for gas-phase azobenzene,[25] or gas-phase
dimetacyano-azobenzene (DMC), which range from 1.27 to
1.29 c.[20, 26] Interestingly, the QAz molecule does not structurally deform on the surface, whereas in the case of TCNQ,
porphyrins, or other donor–acceptor blends, the active
moieties need to modify their structure to strongly bond to
the surface.[2–4]
STM provides further indirect proof for the azine linkage.
Firstly, no configurational isomers (cis–trans forms) typical of
azo-groups are found, not even at the island borders or steps,
where such molecular structures might be more easily
accommodated. Secondly, the STM images show a bright
protrusion in the center of the dimer, that is, at the N@N
position, which can be qualitatively attributed to a charge
density increase due to the formation of the covalent bond in
the QAz molecule. In an azo-coupled molecule, such as
azobenzene, two independent lobes are observed at the rings
with a depression in the center of the molecule.[25]
The net electronic charge transferred from the substrate
to the QAz molecule was calculated by integration of the
computed charge density difference (Figure 4 a), resulting in
a value of 1.21 e@ per molecule. The large amount of
electronic charge gained by the QAz molecule, resulting
from the very different electronegativities of the Cu surface
and the dimer (@2.07 eV in the gas phase), is redistributed
within the molecule according to Figure 4 b. This shows
T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2018, 57, 8582 –8586
Angewandte
Communications
Figure 4. a) Computed plane-averaged charge density difference, D1(r)
in qe b@1 units, for the optimized QAz/Cu(110) configuration along the
z-direction. Vertical red and blue lines indicate the average z-positions
for the topmost Cu layer and the flat-lying molecule. b) Surface chargedensity difference w.r.t. the gas-phase moiety, D1(r) in qe b@2 units,
color map for the optimized QAz/Cu(110) configuration in the xyplane at z = 0.6 b above the average z-position of the molecule (see
vertical dashed black line in (a)). c) Surface charge density, 1(r) in
qe b@2 units, color map in the xy-plane at z = 1 b above the average zposition of the molecule for: (top) the optimized QAz/Cu(110)
configuration, (middle) the optimized gas-phase azobisphenol molecule, and (bottom) the optimized quinoneazine molecule. For the
three cases, the computed Pauling bond orders for the most representative bonds are included.
a computed 2D color map of the surface charge density
difference, defined as D1(r) = 1mol/Cu(r)@[1mol(r) + 1Cu(r)],
where 1mol/Cu(r) is the spatial charge density of the QAz/
Cu(110) system, and 1mol(r) and 1Cu(r) the spatial charge
densities of the non-interacting QAz, and the Cu(110) with
the geometry they adopt at the interface. The red and blue
regions denote a loss or gain, respectively in the net charge
after the formation of the interface. These regions that
correspond to the maximum charge displacement are spatially
localized, mostly in three critical zones: 1) charge accumulation at the two terminating C@O moieties; 2) charge
depletion at the two C@N bonds and accumulation between
the N atoms; and 3) redistribution within the ring to the four
lateral C=C bonds.
Figure 4 a,b illustrates the significant charge reorganization at the interface and strong bonding with the surface, with
a strong depletion of electronic density located just above the
topmost Cu layer (see also the Supporting Information,
Figure S10). Figure 4 c shows surface charge density color
maps of the QAz and ABP molecules in the gas phase, and for
the optimized QAz molecule on a Cu(110) surface, where
a close similarity to the aforementioned azo-compound can
be observed, mainly at N and O sites as well lateral sites in the
C rings. Pauling bond-order analysis is a powerful strategy to
rationalize aromaticity on organic molecules.[27] To numeriAngew. Chem. Int. Ed. 2018, 57, 8582 –8586
Chemie
cally quantify the charge rearrangement within, each bond we
have determined the bond-order for the three cases of
Figure 4 c revealing that O@C, C@N and N@N bonds are
single, single, and double, respectively, for gas-phase ABP and
double, double, single, for QAz. Aromaticity in the C rings for
gas-phase ABP is also observed, but not for the gas-phase
QAz. Interestingly, the same Pauling analysis carried out on
the molecules at the QAz/Cu(110) interface shows a flip in the
bond orders of the molecular bonds with respect to QAz in
the gas phase, becoming effectively more similar to those
found for gas-phase ABP (Supporting Information, Table S1).
Furthermore, the interfacial interaction and significant charge
redistribution in the system generates a strong adsorbateinduced dipole at the surface with the negative end pointing
towards vacuum, hence increasing the work function.
In conclusion, we have shown that the on surfacesynthetized azine-coupled QAz molecule extracts about 1.2
extra electrons from the metal surface inducing a strong
intramolecular charge reorganization, which is accommodated throughout the molecule resulting in a partial recovery
of the aromatic character of the rings and inducing an azo-like
character at the N@N linkage, as shown by bond-order
analysis. The choice of the precursors determines the charge
transfer direction at the metal–organic interface, while the
metal surface plays a crucial role in mediating both the
synthesis and the charge transfer. Strong interaction between
the surface and the molecule beyond self-assembly is
necessary for an efficient charge transfer. Our work suggests
a promising and novel route to the synthesis of hybrid organometallic electrodes with predictable electronic properties, and
opens the door for the use of on-surface synthesis protocols to
fabricate # la carte donor or acceptor interfaces on organic
heterostructures.
Acknowledgements
Graphene Flagship Core2-Graphene-based disruptive technologies EU Horizon 2020 (785219), ERC-synergy program
(ERC-2013-SYG-610256 Nanocosmos), CTI-CSIC and Spanish MINECO (grants MAT2017-85089-C2-1-R, MAT201454231-C4-1-P, and MAT2014-54231-C4-4-P) are acknowledged. We are grateful to the Czech GACR funding (grant
17-24210Y) and LM2015087, Czech Academy of Sciences
through the Praemium Academiae award and FCT program
(IF/01054/2015).
Conflict of interest
The authors declare no conflict of interest.
Keywords: ab initio calculations · charge transfer ·
photoelectron spectroscopy · scanning probe microscopy ·
surface chemistry
How to cite: Angew. Chem. Int. Ed. 2018, 57, 8582 – 8586
Angew. Chem. 2018, 130, 8718 – 8722
T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8585
Angewandte
Communications
[1] R. Otero, A. L. V#zquez de Parga, J. M. Gallego, Surf. Sci. Rep.
2017, 72, 105 – 145.
[2] T. Meier, R. Pawlak, S. Kawai, Y. Geng, X. Liu, S. Decurtins, P.
Hapala, A. Baratoff, S. X. Liu, P. Jel&nek, et al., ACS Nano 2017,
11, 8413 – 8420.
[3] T. C. Tseng, C. Urban, Y. Wang, R. Otero, S. L. Tait, M. Alcam&,
D. Pcija, M. Trelka, J. M. Gallego, N. Lin, et al., Nat. Chem. 2010,
2, 374 – 379.
[4] G. Zamborlini, D. Lgftner, Z. Feng, B. Kollmann, P. Puschnig, C.
Dri, M. Panighel, G. Di Santo, A. Goldoni, G. Comelli, et al.,
Nat. Commun. 2017, 8, 1 – 7.
[5] M. Hollerer, D. Lgftner, P. Hurdax, T. Ules, S. Soubatch, F. S.
Tautz, G. Koller, P. Puschnig, M. Sterrer, M. G. Ramsey, ACS
Nano 2017, 11, 6252 – 6260.
[6] E. Goiri, P. Borghetti, A. El-Sayed, J. E. Ortega, D. G. De Oteyza, Adv. Mater. 2016, 28, 1340 – 1368.
[7] J. M8ndez, M. F. Llpez, J. A. Mart&n-Gago, Chem. Soc. Rev.
2011, 40, 4578.
[8] R. Lindner, A. Kghnle, ChemPhysChem 2015, 16, 1582 – 1592.
[9] G. Otero, G. Biddau, C. S#nchez-S#nchez, R. Caillard, M. F.
Llpez, C. Rogero, F. J. Palomares, N. Cabello, M. A. Basanta, J.
Ortega, et al., Nature 2008, 454, 865 – 868.
[10] J. R. Sanchez-Valencia, T. Dienel, O. Grçning, I. Shorubalko, A.
Mueller, M. Jansen, K. Amsharov, P. Ruffieux, R. Fasel, Nature
2014, 512, 61 – 64.
[11] O. Stetsovych, M. Švec, J. Vacek, J. V. Chocholoušov#, A.
Jančař&k, J. Ryb#ček, K. Kosmider, I. G. Star#, P. Jel&nek, I. Starý,
Nat. Chem. 2017, 9, 213 – 218.
[12] D. Toffoli, M. Stredansky, Z. Feng, G. Balducci, S. Furlan, M.
Stener, H. Ustunel, D. Cvetko, G. Kladnik, A. Morgante, et al.,
Chem. Sci. 2017, 8, 3789 – 3798.
[13] J. Safari, S. Gandomi-Ravandi, RSC Adv. 2014, 4, 46224 – 46249.
8586
www.angewandte.org
Chemie
[14] D. Tomerini, C. Gatti, C. Frayret, Phys. Chem. Chem. Phys. 2015,
17, 8604 – 8608.
[15] R. Willst-tter, M. Benz, Ber. Dtsch. Chem. Ges. 1907, 40, 1578 –
1584.
[16] G. Otero-Irurueta, J. I. Mart&nez, R. A. Bueno, F. J. Palomares,
H. J. Salavagione, M. K. Singh, J. M8ndez, G. J. Ellis, M. F.
Llpez, J. A. Mart&n-Gago, Surf. Sci. 2016, 646, 5 – 12.
[17] G. Olivieri, A. Cossaro, E. Capria, L. Benevoli, M. Coreno, M.
De Simone, K. C. Prince, G. Kladnik, D. Cvetko, B. Fraboni,
et al., J. Phys. Chem. C 2015, 119, 121 – 129.
[18] N. V. Richardson, P. Hofmann, Vacuum 1983, 33, 793 – 796.
[19] R. V. Plank, N. J. DiNardo, J. M. Vohs, Surf. Sci. 1995, 340,
L971 – L977.
[20] M. Piantek, J. Miguel, A. Krgger, C. Nav&o, M. Bernien, D. K.
Ball, K. Hermann, W. Kuch, J. Phys. Chem. C 2009, 113, 20307 –
20315.
[21] K. Diller, F. Klappenberger, M. Marschall, K. Hermann, A.
Nefedov, C. Wçll, J. V. Barth, J. Chem. Phys. 2012, 136, 014705.
[22] F. Bebensee, K. Svane, C. Bombis, F. Masini, S. Klyatskaya, F.
Besenbacher, M. Ruben, B. Hammer, T. Linderoth, Chem.
Commun. 2013, 49, 9308.
[23] J. Stçhr, D. A. Outka, Phys. Rev. B 1987, 36, 7891 – 7905.
[24] P. Hapala, G. Kichin, C. Wagner, F. S. Tautz, R. Temirov, P.
Jel&nek, Phys. Rev. B 2014, 90, 1 – 9.
[25] C. R. Crecca, A. E. Roitberg, J. Phys. Chem. A 2006, 110, 8188 –
8203.
[26] N. Henningsen, R. Rurali, K. J. Franke, I. Fern#ndez-Torrente,
J. I. Pascual, Appl. Phys. A 2008, 93, 241 – 246.
[27] L. Gross, F. Mohn, N. Moll, B. Schuler, A. Criado, E. Guitian, D.
Pena, A. Gourdon, G. Meyer, Science 2012, 337, 1326 – 1329.
Manuscript received: April 6, 2018
Version of record online: June 21, 2018
T 2018 The Authors. Published by Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2018, 57, 8582 –8586