Room-Temperature Switching of Spin-State in Single-Molecule Junctions

The emerging of molecular spintronics offers a unique chance for the design of molecular devices with different spin-state, and the control of spin-state becomes essential for molecular spin switches. However, the intrinsic spin switching from low-spin to high-spin state is a temperature-dependent process with a small energy barrier that low temperature is required to maintain the low-spin state, and thus the room-temperature operation of single-molecule devices have not yet been achieved. Here, we investigated the single-molecule charge transport through a diamagnetic square planar nickel(II) porphyrin using the scanning tunneling microscope break-junction (STM-BJ) technique. The reversible single-molecule conductance switches are demonstrated by utilizing a coordination-induced spin-state switching to manipulate the spin state between S = 0 and S = 1 at room temperature. Furthermore, the different coordinated complexes could be distinguished from the conductance traces, which cannot be realized by the ensemble investigations such as NMR and UV-vis spectrums. The combined DFT calculations revealed that the conductance changes come from the different spin-states of the molecules varying the number of coordination ligands, suggesting coordination-induced spin-state switching provides a new way towards room-temperature molecular spintronics.


Introduction
Molecular spintronics is an emerging field that aims to exceed the limits of conventional electronics by utilizing the spin of the electrons, which has many advantages such as non-volatility, high storage density, low energy consumption, and high response speed. [1][2][3][4][5] As one of the prototype devices, molecular spin switches, especially intrinsic spin switches, which can be switched between at least two stable spin-states by external stimuli such as temperature, 6 pressure, 7 photons, 6,8 electric fields, 9 etc., have been considered as the essential step towards the development of molecular spintronics. [10][11][12][13] Recently, several types of single-molecule prototype devices have been demonstrated, such as voltage-triggered, 14 stretching-induced 15,16 or electron injection induced [17][18][19] spin switches. However, due to the difference in spin multiplicity and vibrational levels, the phase transition of intrinsic spin switch from low-spin (LS) to high-spin (HS) is an entropy-increasing process (ΔS > 0), suggesting that the transition is temperature-dependent, and the LS state is more stable at low temperature. 20,21 Thus, the low-temperature vacuum conditions are required to obatin a stable, low-spin initial state on an electrode surface, and the increasing of their operating temperature to room temperature remains challenging.
To realize room-temperature manipulation of spin-state in single-molecule devices, the interfacial interaction between molecule and electrode is another major issue. [22][23][24] After adsorption on electrode surfaces, the intrinsic spin switch functionality may be lost or changed (such as, a lower phase transition temperature), which brings difficulty for the molecular design. To address this issue, coordination-induced spin-state switching (CISSS), 25 which is insensitive to interfacial interaction, provides a promising strategy for the control of spin-state via the coordinated at room temperature.
The spin-states of some 3d metal ions 25-27 are sensitive to the coordination of the geometric structure, and thus the squar planar Ni-center diamagnetic low-spin state will be switched-on (S = 0 → S = 1) by controlling axial ligand coordination to a square pyramidal or octahedral configuration. 18,25,[28][29][30] On the other hand, previous studies have also demonstated that the charge transport through the metal porphyrin could determine the different elements and even charge states of the metal center, [31][32][33] suggesting the control of nickel porphyrin between different spin-state may offer a promising way towards single-molecule spin switches at room temperature via CISSS strategy.
In this work, we designed a square planar nickel(II) porphyrin derivative, 5,15-bis-(4methylthiophenyl)-10,20-bis-(2,3,4,5,6-pentafluorophenyl)-Ni(II) porphyrin (NiTPPF, Figure 1A), and investigated its spin-induced single-molecule conductance switching using STM-BJ technique at room temperature. Without axial ligands, NiTPPF exhibits a diamagnetic low-spin state (S = 0) with an empty " # $% # and a fully occupied & # orbital of the central nickel ion. One electron is transferred from the & # orbital to the " # $% # orbital leading to a paramagnetic high-spin species (S = 1) upon axial coordination with at least one 3, 5-Lutidine molecule ( Figure 1A). Compared with charge transport through the Zn(II)-analogue, a giant conductance switch was achieved by manipulating the spin-states, which is comparable with most of the single-molecule magnetoresistance systems requiring low temperature and ultra-high vacuum. Furthermore, we found that the five-and sixcoordination NiTPPF complexes can be identified by their single-molecule conductance, which cannot be done with the ensemble investigations such as NMR and UV-vis spectra. These spin manipulations are further confirmed by density functional theory calculation. derivatives are often used to study CISSS in references. 25, [28][29][30] Here, we designed a new porphyrin NiTPPF, which has two -SMe anchor groups at opposite ends of the molecule, to enable charge transport studies. The NMR spectroscopy and UV-vis spectra of NiTPPF and its mixture with the electron-rich auxiliary ligand 3, 5-Lutidine were collected at room temperature to monitor the axial coordination reaction process (Supplementary Figure 1). One index from low spin diamagnetic Ni(II) transitioning to a high-spin state (S = 1) is whether the chemical shift of the pyrrole proton to a low field with larger ppm value, giving rise to a five or six coordinated complex. 25,28-30 From their NMR spectroscopy, the chemical shifts of the pyrrole protons in the pure NiTPPF are located in ~ 8.8 and  Figure 3A), which is consistent with reference. 29 Although we cannot distinguish the five-and six-coordinated complexes from the NMR and UV-vis titration methods, both spectra demonstrate the CISSS in NiTPPF and 3, 5-Lutidine solutions.

Single-molecule conductance measurements.
To characterize the charge transport through NiTPPF and its axial coordinated derivative molecules, the conductance of single-molecule junctions was measured using the STM-BJ technique. [34][35][36][37] We measured three solutions which are 0.1 mM pure NiTPPF solution, 0.1 mM pure NiTPPF solution with 1.1 equivalent 3, 5-Lutidine (NiTPPF-1), and with 2.2 equivalent 3, 5-Lutidine (NiTPPF-2), respectively. For each solution, we repeatedly form and break gold-gold atom contacts. During the gold tip retraction, a plateau around G0 (2e 2 /h, 77.6 µS) representing a gold-gold atom contact can be observed first, followed by some other plateaus below G0. These additional plateaus indicate the formation of Au-molecule-Au junctions.
For each experiment, we collected thousands of conductance traces (3869, 2788, and 2377 traces for NiTPPF, NiTPPF-1, and NiTPPF-2, respectively) to analyze the 1D conductance histograms ( Figure 1B) and 2D conductance-displacement histograms ( Figure 1D, and Supplementary Figures 5,7). All complexes showed a prominent low conductance peak and a less prominent high conductance peak. 38 The measured plateaus lengths can be fitted with a Gaussian function. The plateau length of high conductance junctions is ca. 0.42 nm, which is nearly half-length of the whole molecule (ca. 0.92 nm), after adding a 0.5 nm gold-gold snap back distance to the stretching distance (Supplementary Figure 5), which suggests that this plateau occurs when one of the gold tips connects with the porphyrin ring. 38 The plateau length corresponding to the low-conductance junctions is ca.
1.4 nm. Hence, the corrected junction length is ca. 1.9 nm, which is in line with the sulfur-sulfur distance of NiTPPF. Therefore, the low conductance peak for each sample contains the core information for studying the CISSS-induced charge transport features. The most probable conductance value for each sample is obtained by fitting a Gaussian function to the conductance peak. As shown in Figure 1B, the low conductance of the isolated NiTPPF is 10 -5.37 G0 (0.331 nS).
The conductance shifts to 10 -5.66 G0 (0.170 nS) for NiTPPF-1, and further shifts to a lower value 10 -5.76 G0 (0.135 nS) for NiTPPF-2. We can define a parameter R, analogous to magnetoresistance effect, 39 to evaluate the efficiency of resistance switching rate in the CISSS behavior, defined using the following formula: where RHS, RLS, SHS, and SLS represent the resistance or conductance of high-spin and low-spin states, respectively. In this system, the value of R reached 145.8% with increasing the 3, 5-Lutidine to 2.2 equivalents. The magnitude of R in the present experiments is comparable with most of the singlemolecule magnetoresistance systems reported in the literature. 39,40 To evaluate the effect of spin manipulation on charge transport, we further measured the single- Benefiting from the stability of NiTPPF in the solution of CF3COOH at room temperature, [11c] the five-or six-coordinated NiTPPF can decomplexate to pure NiTPPF by adding excess CF3COOH to completely protonate the 3, 5-Lutidine. As a result, the high spin state (S = 1) can be switched back to low spin state (S = 0). Through adding F3COOH or 3, 5-Lutidine in the samples, we observed the conductance switching in a reversible way ( Figure 1C). Interestingly, three conductance plateaus can be observed in most single traces, which is in line with an ensemble chemical reaction mechanism. An algorithm (see in Supplementary information section V) was programmed to select the traces containing all three conductance.
We assume the conductance of the three components are 10 -5.  Table 1). When the time is extended to 6 ms (120 single shots), 20.6% (1011 in 4898) and 20.3% (1152 in 5658) of all traces remain, respectively. Three peaks can be observed in the 1D conductance histogram shown in Figure 2B and 2E, and those peaks were gradually covered as the screening conditions are relaxed (Supplementary Figure 11). In the three peaks, the highest conductance peak is centered around 10 -5.32 G0 (0.371 ns) for NiTPPF-1, and 10 -5.34 G0 (0.355 ns) for NiTPPF-2, which is well in agreement with the conductance of pure NiTPPF. Thus, the two lower conductance peaks corresponding to the five-or six-coordinated complexes ( Figure 1A and supplementary Figure 1). The centers of the 1D conductance histogram are around 10 -5.68 G0 (0.162 nS) and 10 -6.01 G0 (0.075 nS) for NiTPP-1, and 10 -5.68 G0 (0.162 nS) and 10 -6.04 G0 (0.071 nS) for NiTPP-2. Based on the above analysis (Supplementary Table 2), the resistance of the high-spin state can be enhanced up to 400% compared to the diamagnetic state.
From the single trace analysis, we can capture the dynamic processes of the axial-coordinated reaction, and distinguish the five-or six-coordinated complexes through STM-BJ technique, which cannot be distinguished from the ensemble investigations such as NMR and UV-vis spectrums. Theoretical calculation. To provide insight into the experimental results, theoretical simulations and analysis based on density functional theory, a tight-binding model, and quantum transport theory are carried out for the four molecules (ruffled-NiTPPF, planar-NiTPPF, five-coordinated NiTPPF-1L, six-coordinated NiTPPF-2L). 42,43 The DFT calculated spin-resolved projected density of states of the four complexes are summarized in Figure 3. Generally, the pure NiTPPF molecules exhibit a ruffled geometry in crystal lattices, while a planar geometry is possible in the gas-phase and the solvent. Hence, both of the geometries were examined ( Figure 3A and B). The calculated results show that ruffled and planar geometries are both in the low-spin ground states (S = 0), and the former is more energetically favorable than the latter by 20 meV lower total energy. Four d orbitals ( "% , %& , "& , & # ) of Ni atom are fully occupied, whereas the " # $% # orbital moves to a higher energy and becomes the LUMO (the lowest unoccupied molecular orbital) for the planar molecule.
Considering the axial coordination by one and two 3, 5-Lutidine, the ruffled porphyrin tends to be planar, with a high-spin ground state (S = 1).
The PDOS plotted in Figure 3C  Those very narrow and tiny peaks can be observed both in low-spin and high-spin states which originate from the localized orbitals on porphyrin rings and Ni atoms and rare weights on the -SMe anchor groups (Supplementary Figure 12, 13). Some orbitals are even completely silent, such as, the & # orbital of the planar-NiTPPF molecule, " # $% # orbital of spin-up for the NiTPPF-2L molecule.
In the inset of Figure 4B, these pronounced resonances are associated with the orbitals marked by stars in Figure 3. The Fano-resonance at 0.1eV of the planar-NiTPPF arises from the orbital containing large components of " # $% # , which is localized on the porphyrin ring and the Ni atom and can be rationalized using a tight-binding model (Supplementary Figure 16). The Fermi energy is close to the LUMO in all four molecules, suggesting that LUMOs are the dominated charge transport channels. The transmission coefficients of planar and ruffled NiTPPF near the Fermi energy level are larger than those of five-or six-coordinated NiTPPF complexes. Those features reveal the decreasing trend in conductance between the low-spin and high-spin states. We also calculated the Zn(II)-analogue system (Supplementary Figure 17, 18), and demonstrated that the small energy change in electron energy levels after axial coordination can slightly suppress charge transport.
However, this effect is weaker than that of the Ni system with spin manipulation, which is in good agreement with the experiments.

Conclusions
In summary, we demonstrate the reversible switching of spin-state in single-molecule junctions at room temperature using STM-BJ technique, and we have achieved a conductance difference of up to 400% for different spin-state using coordination-induced spin-state switching. We further proved that the single-molecule conductance is an effective method to study rapid time-scale reactions beyond the ensemble measurements. On the basis of this study, we suggest two possible ways to improve the device. First, from the theoretical calculation, the contribution of 3d orbitals of nickel ions to electron transmission is suppressed by the highly conjugated porphyrin ring. Hence, taking advantage of chemical tunability to eliminate the conjugation of the ligand and increase the contribution of the metal ion in the conductive channel will hopefully result in a device with a higher conductance difference. Second, the auxiliary ligand can be functionalized so that it can respond to more stimuli to develop a multifunctional device. For example, photon-driven coordination-induced spin-state switching (LD-CISSS) single-molecule electric device can be explored by introducing a photonactive ligand such as azopyridine derivatives. 25,28 Therefore, we confidently anticipate that the CISSS concept provides a promising route towards room-temperature single-molecule spintronics, which is of potential for the future development of spin-based molecular memory devices.

Methods
All reagents were purchased from TCI and used without further purification. The NMR spectrums were recorded on a Bruker 400 M instruments and Tetramethylsilane (TMS) was used as an internal standard.

Synthesis of 5-(Pentafluorophenyl)-dipyrromethane (DPMF)
DPMF was synthesized followed by a reported literature procedure. 38,44 A solution of Pentafluorobenaldehyde (3.92 g, 20 mmol) and an excess amount of pyrrole (50 mL) was degassed with N2 at room temperature. To this, was added trifluoroacetic acid (TFA) (0.17 mL, 2 mmol) and the reaction mixture was stirred for 30 min at room temperature. After the reaction, the mixture was treated with 0.1 M NaOH Solution and then extracted with ethyl acetate for three times.
The combined solution was concentrated and purified by column chromatography over silica gel using DCM: Hex = 1:2 as eluent to give product (1.51 g, 24%) as white solid. 1