Field-induced Conductance Switching by Charge-state Alternation in Organometallic Single-Molecule Junctions
Charge transport through single molecules can be
influenced by the charge and spin states of redox-active metal
centres placed in the transport pathway. These molecular
intrinsic properties are usually addressed by varying the
molecule’s electrochemical and magnetic environment, a procedure
that requires complex
setups with multiple terminals. Here we show that oxidation and reduction of
organometallic compounds containing either Fe, Ru or Mo centres
can solely be triggered by the electric field applied to a
two-terminal molecular junction. Whereas all compounds exhibit
bias-dependent hysteresis, the Mo-containing compound additionally
shows an abrupt voltage-induced conductance switching, yielding
high-to-low current ratios exceeding 1000 at voltage stimuli of
less than 1.0 V. DFT calculations identify a localized,
redox-active molecular orbital that is weakly coupled to the
electrodes and closely aligned with the Fermi energy of the leads
because of the spin-polarised ground state unique to the Mo
centre. This situation opens an additional slow and incoherent
hopping channel for transport, triggering a transient charging
effect of the entire molecule and a strong hysteresis with
unprecedented high low-to-high current ratios.
Switching an electric signal from a low- to a high-current state
is one of the key elements in an electric circuit with
applications in signal processing, logic data manipulation or
storage. In current Si-based technology with device dimensions
approaching the sub-5nm range, it becomes increasingly difficult
to maintain large high-to-low ratios mainly because of leakage
currents. Therefore alternative switching mechanisms are needed. In single-molecule electronics, a variety of intrinsic
conductance-switching mechanismsMolen2010 () exist: Gating of
the molecular orbitals (MOs) by electrostaticSong2009 () or
electrochemical meansLiao2010 (), which requires a third
electrode, or modifying specific photoactive molecular structures,
e.g. by optically irradiating the molecule to form or break
bondsIrie2000 (); Dulic2003 (); Kronemeijer2008 (); vanderMolen2009 (). Also mechanically-induced changes in the molecule–metal
couplingQuek2009 () can lead to conductance alternations. Another trigger is the electric field inherently present in a
molecular transport junction: Conformational changes due to
interactions between the electric field and molecular dipoles were
demonstrated to alternate the conductance of single-molecule
junctionsBlum2005 (); Loertscher2006a (); Meded2009 () by up to a
Potentially more powerful mechanisms exploit intrinsic molecular
quantum phenomena related to spin and charge states. An early
example of addressing the spin state of a single
moleculeKahn1998 (); Sato2007 (); Baadji2009 () revealed Kondo
resonances using cobalt (Co) metal centres.Park2002 () More
recently, a spin cross-over was induced by an electric field in
iron (Fe)-based molecular nanoclustersPrins2011a () (with
high-to-low ratios of 2), and in a coupled spin pair of two
Co atomsWagner2013 () (with high-to-low ratios of 2 - 3). Regarding intrinsic charge states, Coulomb blockade peaks were
reported in ruthenium (Ru)-containing wiresKim2007 (), but not
confirmed in self-assembled monolayers.Luo2011 () On the
single-molecule level, Ru-based molecules showed
conformation-induced changes in the conductanceRuben2008 ()
rather than changes due to intrinsic redox mechanisms. Two Ru
metal centres in a photochromatic compound demonstrated reversible
light-induced conductance switching in ensemble
junctions.Meng2012 (); Meng2014 () In another study, the
importance of the copper (Cu) coordination on the conductance was
Placing Individual Metal Centres in the Transport Pathway
Earlier we studied dinuclear organometallic Fe compounds with various anchoring schemesSchwarz2014a (); Lissel2014 () and discovered indications of field-induced conductance switching in the case of weak molecule–metal coupling. Motivated by these findings, we have developed mononuclear organometallic compounds of the type (MeCOSCH-CC-)M(PP) (M = Fe; PP = 1,2 - bis(diethylphosphino)ethane: (1); M = Ru, Mo (Molybdenum); PP = 1,2 -bis(diphenylphosphino)ethane: (2), (3)) using weak thiol couplingReed1997 (); Zotti2010 () to preserve molecule-internal spin and charge degrees of freedom for the solid-state molecular junctions. Fe, Ru and Mo were chosen as metal centres. The synthetic strategy aims at placing the metal centres right in the transport pathway (Fig. 1a,b) to achieve an optimal influence on transport and maximum interaction with the electric field. We used identical acetylenic backbones to constrain the variable parameters to the metal centres and their ligand fields. To prevent dimerization, the sulphur end groups were acetyl-protected, with the protection groups being hydrolyzed in situ, forming the metal–molecule–metal junctions Au–1–Au, Au–2–Au, and Au–3–Au (Fig. 1c). For the Fe metal centre, we used the bidentate phosphine ligands depe (depe = 1,2-bis(diethylphosphino)ethane) and for the RuWuttke2014 () and Mo centres, dppe (dppe = 1,2-bis(diphenylphosphino)ethane) chelate ligands. To extend the molecular length to 2.5 nm (S–S distance), we chose phenylene spacers as conducting backbones. The synthesis of 2 was reproduced using a previously reported procedureTouchard1997 (); Wuttke2014 (), whereas new synthetic protocols were established for 1 and 3 (see SI for details).
Conductance Switching in Single-Molecule Junctions
First, we perform current–voltage – data acquisition by
repeatedly forming and breaking the junctionLoertscher2007 ()
(see SI). In the entire data sets, we find a substantial number
of curves ( 90 ) that all exhibit distinct features
differing from conventional non-linear molecular transport,
namely, curves with hysteretic behaviour. Here, the curves
acquired for sweeps from negative to positive bias are separated
in a given voltage range from those acquired in the opposite
direction. For the Au–1–Au junction, around 85
of the curves show hysteresis, for Au–2–Au 80 ,
and for Au–3–Au 95 . Fig. 1d shows 50
representative – curves taken at 50 K (see SI sections 10,
16-18 for statistics, sampling rate and temperature dependence). The hysteretic behaviour of the three compounds differs in the
voltage range and the transition between the two envelopes. Accordingly, we categorised the –s into two types: Type
I curves are found for all compounds and are characterised
by a small hysteresis that affects only a particular section of
the voltage (blue backgrounds in Fig. 1d), whereas the –s
for the low- and high-bias regimes are nominally identical. The
conductance gap (as defined by the onset in transport) is not
altered, and the transition between the curves is continuous. Type II curves are only found for the Mo compound and
differ from type I curves by an approximately 100x lower
current and an abrupt switching between two distinct curves,
accompanied by a hysteresis. Here, the conductance gaps change
substantially (e.g. from 0.15 V to 0.85 V). Fig. 1e summarises
the experiments schematically by providing also the sweep
directions. When analysing the occurrences of type I and
type II curves, we find that they depend on the junction
configuration: Type II curves are found just before
breaking the molecular junction and show a switching between two
distinct states (Fig. 2a). When extracting the maximum
high-to-low ratio in the hysteresis region and plotting it versus
the corresponding voltage (Fig. 2b), we find that type I
curves display a narrow energy distribution, whereas type
II curves seem to depend non-linearly on energy, with an
increasing ratio for increasing bias. The high-to-low current
ratios are 1.5 to 20 for type I switching consistently for
all compounds, and exceed 1000 for type II switching.
Coherent Tunneling and Decoherent Hopping Transport
In principle, several possible explanations exist for the hysteretic curves at smaller junction distances (type I) and the abrupt switching at larger junction distances (type II): As the ground state of the Mo compound 3 and some of the excited states of the Fe compound 1 and the Ru compound 2 are magnetic, a high-spin/low-spin (HS/LS) crossover comes to mind, given the observed switching in similar systems.Prins2011a (); Wagner2013 () There, the HS/LS crossover caused a drastic change in the electronic structure across the entire electron wavelength spectrum that accompanied the switching between two distinct electronic states. In our experimental data, however, we do not find such drastic changes, e.g. no change in the conductance gap for type I curves and almost identical curves for certain voltage regimes for both types. Moreover, the nuclei of the molecules would adapt to that and thereby preclude hysteresis simply because there is nothing that could cause a time delay. In contrast, an oxidation or reduction of the transition-metal compounds would enable a potential observation of hysteresis, as proposed theoretically.Galperin2005 (); Kuznetsov2007 (); Migliore2013 () Those authors argued that if the charging rate is similar to the bias sweeping rate, a time delay required for a memory effect could occur, making the charging visible in the – curves. Here, the conductance is governed by a coherent tunnelling channel mediated by a delocalized molecular orbital (MO) (“fast channel”), whereas hysteresis is related to charging of a localized MO in an electron-hopping channel (“slow channel”) (Fig. 3a). The probability of the localized MO to be occupied determines the respective conductance contributions from the two charging states at every bias increase.
To simulate transport through single molecules, these charging probabilities were implemented into a stochastic approach. For calculating – curves, we combine proposed algorithms Migliore2013 () with data from density functional theory (DFT) calculations for the transmission (defining coherent tunnelling) and the transfer integral, reorganization energy and driving force (describing electron hopping Kastlunger2013 (); Kastlunger2014 (); Kastlunger2015 ()). First, we demonstrate that by varying the ratio between the bias sweeping rate and the charging/hopping rate, the abrupt switching shown in Fig. 1c can be qualitatively reproduced for a simple two-MO tight-binding model. In Fig. 3b, we show – curves simulated for forward and reverse bias sweeps for various coupling strengths, , and fixed voltage sweeping rates, . For the strongest coupling, = 10 V, a statistical average of multiple switching events is observed, and as a consequence, the forward and backward sweeps fall together with the average of the two limiting – curves obtained from the integration of the transmission functions of the reduced and the oxidized systems (shown as orange and gray dotted lines in Fig. 3b, respectively). When lowering the coupling to = 10 V, averaging covers a smaller number of redox processes per integration time, , resulting in fringes. For / = 100 V, there is roughly one jump in each integration interval, and for the weakest coupling, / = 0.5 V, only a single jump happens during a full sweep. Going from the strongest to the weakest coupling step by step (Fig. 3b), the voltage range where both sweeps follow the lower curve for low bias and the upper one for high bias becomes larger because ever higher voltages are needed to increase the likelihood of jumps. In the wide range of couplings from / = 10 to 100 V, however, the forward and backward – curves still follow the same path, although our stochastic approach creates uncertainties or line-thickening for ratios of / smaller than 10 V. Only with / as low as 1 V an irreversible switching or a “lock-in” process does take place, where the forward sweep follows the lower limiting curve for the reduced state and the backward sweep the upper one for the oxidized state; a scenario that qualitatively explains the type II curves for the Mo junction. Fig. 3c displays the hopping rates for oxidation and reduction, showing that the lower the ratio, the later and cross the horizontal line defining the sweeping rate. determines the probability of charging the compounds, whereas the stability of this oxidized state depends inversely on . Therefore, as can be seen from the left-hand-side panel, the bias necessary for reaching the charged state is inversely proportional to the coupling strength. , in contrast, decreases with the coupling strength, which makes a reduction even at lower biases less likely, finally enabling the occurrence of a “lock-in” process during bias sweeps.
Let us now look at the electronic structures of the transport junctions as obtained from DFT calculations (Fig. 4). Energetic positions of the MO’s and their spatial distributions as well as the corresponding transmission functions are computed by a Nonequilibrium Green’s Function (NEGF) DFT formalismBrandbyge2002 (); Xue2002 (); Rocha2005 () with the GPAW codeMortensen2005 (); Enkovaara2010 (). Because the Mo compound 3 is the only compound among the three with a spin-polarised ground state, we show its MO eigenenergies and transmission functions for spin-up and spin-down separately (green curves in Fig. 4a). The magnetic property of the Mo system is the reason why a very localised MO with symmetry on the metal atoms (where is the transport direction) moves close to the Fermi level for one spin orientation (violet dots). In contrast, this MO lies far outside of any reasonable bias window for the Fe compound 1 and the Ru compound 2. As a high degree of localization of a MO results in a very weak coupling to the electrodes, this MO can be considered the “slow channel” for the Mo compound, while for the Fe and Ru compounds the HOMO-1 () plays this role. For all three systems, the “fast channel” is provided by the delocalized HOMO (Fig. 4b).
To calculate the hopping rates for the oxidation/reduction that govern the switching between neutral and charged compounds, we follow an approach developed earlier.Kastlunger2015 () Table 1 lists all relevant parameters for both charging states for all three systems at the equilibrium distance. For the Mo compound, it also gives those values at an elongation of the bonding distance between the anchor group and the electrodes by 0.5 Å on both sides according to the experimental findings that type II curves are found for elongated junctions just before rupture. The driving force , which is defined by the energy difference of the “slow-channel” MO and , is lower for Mo than for Fe and Ru by a factor of 2 - 3. Its transfer integral is two orders of magnitudes smaller and even three orders of magnitude smaller at the elongated distance. Because of the self-interaction problem of DFT, which becomes more severe for localized states, the calculations overestimate the spatial extension of the respective orbital and thereby also the transfer integral. Additionally, we have to account for the fact that the binding of the molecule to the metal surfaces is idealised in our DFT calculations, where we use perfectly planar Au(111) surfaces and symmetric bonding of the compounds at equilibrium distances. To account for these aspects, all calculated transfer integrals are consistently scaled down by a factor of 100 for the calculated – curves in the lower panels of Fig. 4c. In all panels, different sweeping rates are used for the forward and backward sweep, in agreement with the experimental situation (see SI). Whereas for the Au–1–Au and the Au–2–Au junctions, an elongated configuration reveals only a minor influence on the hysteresis and the functional behaviour, the Au–3–Au junction shows an abrupt transition at the weaker coupling conditions induced by elongation. This situation perfectly reproduces the experimental findings in terms of switching energy, relative current levels, type of hysteresis and drastic change in the conductance gap. Furthermore, DFT calculations can also reproduce the high-to-low current ratios, which are around 1.5 - 4.5 for type I hysteresis and around 200 for type II hysteresis with abrupt switching (Fig. 4c).
|Mo (+0.5 Å)||0.269||0.075||-0.013||0.062|
Conclusion and outlook
In summary, we have experimentally and theoretically investigated
the transport properties of organometallic molecules containing
Fe, Ru and Mo metal centres in their transport pathway. We find
hysteretic transport properties with continuous transitions for
all three transport junctions, and additionally an abrupt
switching for the Mo compound. Comprehensive DFT modelling,
taking into account bias-driven charging, indicates an
oxidation/reduction mechanism mediated by a weakly coupled,
localized MO that is unique to the Mo compound because of its
spin-polarized ground state. This MO gives rise to abrupt
switching with high-to-low current ratios of more than 1000,
outperforming all previously explored molecular-intrinsic
conductance-switching mechanisms, such as
magnetoresistance.Schmaus2011 () DFT combined with a
two-channel transport model qualitatively agrees with experiments
regarding the functional behaviour of the hysteresis. We
therefore conclude that intrinsic redox functionality is
maintained in weakly-coupled solid-state organometallic junctions,
remains accessible at feasible electric fields in a two-terminal
geometry, and can be controlled by tuning the voltage sweeping
rate in respect to the intrinsic oxidation and reduction rates. Moreover, by bias-induced charge-state alternations, a conductance
switching with technologically relevant high-to-low current ratios
exceeding 1000 at voltages of 1.0 V could be achieved in a
single-molecule building block. Even though technological
parameters, such as fatigue, switching speed, non-volatility etc.,
remain to be determined in real device geometries, such ultimately
scaled building blocks fulfill in principle the requirements for
future memory in terms of reasonably low
operational fields, speed, and large high-to-low current ratios.
- (1) Van der Molen, S.J. Liljeroth, P. Charge transport through molecular switches. J. Phys. 22, 133001 (2010).
- (2) Song, H. et al. Observation of molecular orbital gating. Nature 462, 1039–1043 (2009).
- (3) Liao, J. et al. Cyclic conductance switching in networks of redox-active molecular junctions. Nano Lett. 10, 759–764 (2010).
- (4) Irie, M. Diarylethenes for memories and switches. Chem. Rev. 100, 1685–1716 (2000).
- (5) Dulic, D. et al. One-way optoelectronic switching of photochromic molecules on gold. Phys. Rev. Lett. 91, 207402 (2003).
- (6) Kronemeijer, A.J. et al. Reversible conductance switching in molecular devices. Adv. Mat. 20, 1467–1473 (2008).
- (7) Van der Molen, S.J. et al. Light-controlled conductance switching of ordered metal–molecule-metal devices. Nano Lett. 9, 76–80 (2009).
- (8) Quek, S.Y. et al. Mechanically controlled binary conductance switching of a single-molecule junction. Nature Nanotech. 4, 230–234 (2009).
- (9) Blum, A.S. et al. Molecularly inherent voltage-controlled conductance switching. Nature Mat. 4, 167–172 (2005).
- (10) Lörtscher, E., Ciszek, J.W., Tour, J.M., Riel, H. Reversible and controllable switching of a single-molecule junction. Small 2, 973–977 (2006).
- (11) Meded, V., Bagrets, A., Arnold, A. Evers, F. Molecular switch controlled by pulsed bias voltages. Small 5, 2218–2223 (2009).
- (12) Kahn, O. Martinez, C. J. Spin-transition polymers: From molecular materials toward memory devices. Science 279, 44–48 (1998).
- (13) Sato, O., Tao, J. Zhand, Y.-Z. Control of magnetic properties through external stimuli. Angew. Chem. Int. Ed. 46, 2152–2187 (2007).
- (14) Baadji, N. et al. Electrostatic spin crossover effect in polar magnetic molecules. Nature Mat. 8, 813–817 (2009).
- (15) Park, J. et al. Coulomb blockade and the Kondo effect in single-atom transistors. Nature 417, 722–725 (2002).
- (16) Prins, F., Morabal-Capilla, M., Osorio, E.A., Coronado, E. van der Zant, H.S.J. Room-temperature electrical addressing of a bistable spin-crossover molecular system. Adv. Mat. 23, 545–1549 (2011).
- (17) Wagner, S. et al. Switching of a coupled spin pair in a single-molecule junction. Nature Nanotech. 8, 575–579 (2013).
- (18) Kim, B. et al. Temperature and length dependence of charge transport in redox-active molecular wires incorporating Ruthenium(II) bis(-arylacetylide) complexes. J. Phys. Chem. C 111, 7521–7526 (2007).
- (19) Luo, L. et al. Length and temperature dependent conduction of Ruthenium-containing redox-active molecular wires. J. Phys. Chem. C 115, 19955–19961 (2011).
- (20) Ruben, M. et al. Charge transport through a cardan-joint molecule. Small 4, 2229–2235 (2008).
- (21) Meng, F. et al. Photo-modulable molecular transport junctions based on organometallic molecular wires. Chem. Sci., 3, 3113–3118 (2012).
- (22) Meng, F. et al. Orthogonally modulated molecular transport junctions for resettable electronic logic gates. Nature Comm. 5, 3023 (2014).
- (23) Ponce, J. et al. Effect of metal complexation on the conductance of single-molecular wires measured at room temperature. J. Am. Chem. Soc. 136, 8314–8322 (2014).
- (24) Schwarz, F. et al. High-conductive organometallic molecular wires with delocalized electron systems strongly coupled to metal electrodes. Nano Lett. 14, 5932–5940 (2014).
- (25) Lissel, F. et al. Organometallic single-molecule electronics: Tuning electron transport through (diphosphine)FeCFe(diphosphine) building blocks by varying the molecule–metal anchoring scheme from coordinative to covalent. J. Am. Chem. Soc. 136, 14560–14569 (2014).
- (26) Reed, M.A., Zhou, C.; Muller, C.J., Burgin, T.P. Tour, J.M. Conductance of a molecular junction. Science 278, 252–254 (1997).
- (27) Zotti, L.A. et al. Revealing the role of anchoring groups in the electrical conduction through single-molecule junctions. Small 6, 1529–1535 (2010).
- (28) Wuttke, E. et al. Divinylphenylene- and ethynylvinylphenylene-bridged mono-, di- , and triruthenium complexes for covalent binding to gold electrodes. Organometallics 33, 4672–4686 (2014).
- (29) Touchard, D. et al. Vinylidene-, Alkynyl-, and trans- bis(alkynyl)ruthenium complexes. crystal structure of trans-[Ru(NH)(CC-Ph)(PhPCHCHPPh)]PF. Organometallics 16, 3640–3648 (1997).
- (30) Lörtscher, E., Weber, H.B. Riel, H. Statistical approach to investigating transport through single molecules. Phys. Rev. Lett. 98, 176807 (2007).
- (31) Galperin, M., Ratner, M. Nitzan, A. Hysteresis, switching, and negative differential resistance in molecular junctions: A polaron model. Nano Lett. 5, 125–130 (2005).
- (32) Kuznetsov, A.M. Negative differential resistance and switching behavior of redox-mediated tunnel contact. J. Chem. Phys 127, 084710 (2007).
- (33) Migliore, A. Nitzan, A. Irreversibility and hysteresis in redox molecular conduction junctions. J. Am. Chem. Soc. 135, 9420–9432 (2013).
- (34) Kastlunger, G. Stadler, R. Charge localization on a redox-active single-molecule junction and its influence on coherent electron transport. Phys. Rev. B 88, 035418 (2013).
- (35) Kastlunger, G. Stadler, R. Density functional theory based calculations of the transfer integral in a redox-active single-molecule junction. Phys. Rev. B 89, 115412 (2014).
- (36) Kastlunger, G. Stadler, R. A density functional theory based direct comparison of coherent tunnelling and electron hopping in redox-active single molecule junctions. Phys. Rev. B 91, 125410 (2015).
- (37) Brandbyge, M., Mozos, J.-L., Taylor, J. Stokbro, K. Density-functional method for nonequilibrium electron transport Phys. Rev. B 65, 165401 (2002).
- (38) Xue, Y., Datta, S. Ratner, M.A. First-principles based matrix Green’s function approach to molecular electronic devices: General formalism. Chem. Phys. 281, 151–170 (2002).
- (39) Rocha, A.R. et al.. Towards molecular spintronics. Nature Mater. 4, 335–339 (2005).
- (40) Mortensen, J.J., Hansen, L.B. Jacobsen, K.W. Real-space grid implementation of the projector augmented wave method. Phys. Rev. B 71, 035109 (2005).
- (41) Enkovaara, J. et al. Electronic structure calculations with GPAW: A real-space implementation of the projector augmented-wave method. J. Phys. 22, 253202 (2010).
- (42) Schmaus, S. et al. Giant magnetoresistance through a single molecule. Nature Nanotech. 6, 185–189 (2011).
We are grateful to M. Koch for support
with the synthesis of the end groups and to O. Blacque for
single-crystal X-ray diffraction. We also acknowledge G. Puebla-Hellmann, V. Schmidt, and F. Evers for scientific
discussions, and M. Tschudy, U. Drechsler and Ch. Rettner for
technical assistance. We thank W. Riess and B. Michel for
continuous support. Funding from the National Research Program
“Smart Materials” (NRP 62, grant 406240-126142) of the Swiss
National Science Foundation (SNSF) and the University of
Zürich is gratefully acknowledged. G.K. and R.S. are
currently supported by the Austrian Science Fund FWF, project
Nos. P22548 and P27272, and are deeply indebted to the Vienna
Scientific Cluster VSC, on whose computing facilities all DFT
calculations were performed (project No. 70174). In addition,
G.K. receives a grant co-sponsored by the Austrian Academy of
Science ÖAW, the Springer Verlag
and the Austrian Chemical Society GÖCH.
F. L. and G. K. made equal contributions
to this work and should therefore be considered joint first authors.
F. L., C. E., S. N. S., K. V., and
H. B. designed and synthesized the compounds. F. S., and E. L. set up and performed the experiments and the data analysis. G. K. and R. S. carried out the calculations. F. S., G. K.,
K. V., H. B., R. S. and E. L. wrote the paper. All authors
results and commented on the manuscript.
Supplementary information accompanies this
paper at www.nature.com/ naturenanotechnology. Reprints and
permission information is available online at
http://www.nature.com/reprints/. Correspondence and requests for
materials should be addressed to: firstname.lastname@example.org (H.B.),
email@example.com (K.V.) for chemistry,
firstname.lastname@example.org (R.S.) for DFT calculations, and
email@example.com (E.L.) for experimental work. CCDC-1040144
(for 1), CCDC-1040145 (for 2) and CCDC-1040146
(for 3) contain the supplementary crystallographic data
(excluding structure factors) for this paper. These data can be
obtained free of charge from The Cambridge Crystallographic Data
Centre via https://summary.ccdc.cam.ac.uk/structure-summary-form.
Competing financial interests
The authors declare no competing financial interests.
Chemical Synthesis. The synthetic steps and
full characterisation of all compounds can be found in the
Electron-beam-structured break junctions are mechanically actuated
in a three-point bending mechanism operated under
ultra-high-vacuum conditions (UHV; pressure 2
10 mbar) at 50 K. Molecules are deposited from a
highly diluted solution in dry tetrahydrofuran (THF; 4
10 m/L). Electrical characterization is carried out with a
Hewlett-Packard Semiconductor Parameter Analyzer HP4156B upon
repeated opening and closing of the molecular junction (more
details can be found in the supporting information).
Computational details. All calculations of transmission probabilities and – curves were performed within a NEGF-DFT framework with the GPAW code. We chose a linear combination of atomic orbitals (LCAO) on a Double Zeta level with polarisation functions (DZP) for the basis set and a Perdew-Burke-Ernzerhof (PBE) parametrisation for the exchange-correlation (XC) functional. The MO eigenenergies were calculated by decoupling the basis functions localised on the molecule from those of the surface states via a subdiagonalisation of the transport Hamiltonian.
For the redox process, we combine a recent formalism with a coherent tunnelling description based on NEGF-DFT for the calculation of the – characteristics of the reduced and oxidized states and a hopping description of the redox reaction based on Marcus theory. By calculating the bias-dependent reaction rates of oxidation and reduction, a probability can be determined that describes the system’s probability to be in one of the respective charge states after a given integration time . To simulate single – sweeps, we apply a stochastic approach, in which we trap the system into one distinct charge state in every step. By calculating the change of probability , defined by either or , between two time steps and , where , and comparing defined in this way with a random number between 0 and 1, we create a criterion for the switching between the two states. The overall current is then calculated from a mean value , averaging over all current values, with . More details can be found in the supplementary information.