Spin-Crossover in Supramolecular Iron(II)-2,6-bis(1H-Pyrazol-1-yl)pyridine Complexes: Toward Spin-State Switchable Single-Molecule Junctions.

Spin-crossover (SCO) active iron(II) complexes are an integral class of switchable and bistable molecular materials. Spin-state switching properties of the SCO complexes have been studied in the bulk and single-molecule levels to progress toward fabricating molecule-based switching and memory elements. Supramolecular SCO complexes featuring anchoring groups for metallic electrodes, for example, gold (Au), are ideal candidates to study spin-state switching at the single-molecule level. In this study, we report on the spin-state switching characteristics of supramolecular iron(II) complexes 1 and 2 composed of functional 4-([2,2'-bithiophen]-5-ylethynyl)-2,6-di(1H-pyrazol-1-yl)pyridine (L1) and 4-(2-(5-(5-hexylthiophen-2-yl)thiophen-2-yl)ethynyl)-2,6-di(1H-pyrazol-1-yl)pyridine (L2) ligands, respectively. Density functional theory (DFT) studies revealed stretching-induced spin-state switching in a molecular junction composed of complex 1, taken as a representative example, and gold electrodes. Single-molecule conductance traces revealed the unfavorable orientation of the complexes in the junctions to demonstrate the spin-state dependence of the conductance.

Introduction

Ever since its genesis dating back to the 1930s,[1] the spin-crossover (SCO) phenomenon[2−7] continually attracted the interest of chemists and physicists alike, and the device suitable characteristics of the SCO systems have been elucidated.[8−11] First-row transition metal complexes featuring a d4–d7 electronic configuration undergo SCO[12−19] upon application of an external stimulus, for example, temperature, light, or electrical field.[20] While bistable SCO systems showing a thermal hysteresis loop (ΔT) are suitable for fabricating molecule-based switching and memory elements,[21] single-molecule junctions composed of spin-state switchable molecules are useful for developing molecular spintronics elements.[22−30]

Studies on spin-state switching at the bulk and single-molecule levels represent two different paradigms. In the bulk state, intermolecular interactions play a significant role in the spin-state switching process,[31−36] and the switching is often induced by applying temperature as an external stimulus.[3] At the single-molecule level, intermolecular interactions cease to exist, and spin-state switching is achieved by applying a stimulus under isothermal conditions.[37−39] Charging of ligands,[37] electric field-induced distortion of a molecule featuring a large dipole moment,[38] and stretching a molecule in a single-molecule junction[39] are some methods used to induce SCO at the single-molecule level. Electron-induced spin-state switching at the single-molecule level in surface-bound thin films of SCO active molecules has also been demonstrated in scanning tunneling microscope (STM) junctions at 2 K.[40−44] The spin-state switching at the single-molecule level is accompanied by conductance switching, and high-spin (HS) junctions show a spin-filtering effect,[45,46] elucidating the utility of spin-switchable molecular junctions as molecular spintronics elements.

Supramolecular-SCO systems featuring anchoring groups for electrodes, for example, gold (Au) and single-layer graphene (SLG), are ideal systems to study spin-state switching at the single-molecule level.[24,47] Iron(II) complexes based on 2,6-bis(1H-pyrazol-1-yl)pyridine (BPP) ligand systems[48−50] are suitable systems to study the spin-state dependence of conductance switching at the single-molecule level. Such complexes are remarkable, because BPP ligands feature moderate ligand field strength for iron(II); therefore, SCO is facilitated.[48] Moreover, the pyridine and pyrazole rings of the BPP skeleton are amenable to functionalization, enabling the synthesis of functional SCO systems.[51] Consequently, a family of supramolecular BPP ligands featuring thioacetate (SAc), pyrene,[52−54] and fullerene (C60)[53,55,56] anchoring groups has been designed, and the SCO characteristics of the resultant supramolecular iron(II) complexes have been investigated. By preparing a graphene−SCO complex hybrid material, we have elucidated that the interface between the molecule and graphene surface modifies the SCO parameters, relative to the parameters observed in the bulk state.[57] Subsequently, we have demonstrated spin-state dependence of conductance characteristics at a single-molecule junction composed of SLG electrodes and a pyrene-tethered iron(II)–BPP complex.[24] In another direction, we have studied transport characteristics of SCO–gold nanoparticle (Au–NP) hybrid ensembles, demonstrating the utility of iron(II)–BPP complexes featuring sulfur-based anchoring groups for the development of SCO molecule-based spintronic architectures.[47]

Despite a few studies discussed in the foregoing paragraph, the utility of iron(II)–BPP complexes as a spin-switchable entity at single-molecule junctions remains largely unexplored. In an attempt to remedy this, we have designed a set of supramolecular iron(II)–BPP complexes in view of their implementation in single SCO molecule junctions. These complexes, 1 and 2 (Scheme ), are composed of 4-([2,2′-bithiophen]-5-ylethynyl)-2,6-di(1H-pyrazol-1-yl)pyridine (L1) and 4-(2-(5-(5-hexylthiophen-2-yl)thiophen-2-yl)ethynyl)-2,6-di(1H-pyrazol-1-yl)pyridine (L2), respectively. In the following sections, we report on the syntheses of the ligands and complexes, X-ray crystal structures of L2 and complex 2, and the SCO characteristics of complexes 1 and 2. We then present stretching-induced spin-state switching of 1 predicted by density functional theory (DFT) calculations and our attempts to study complexes 1 and 2 at a single-molecule level by connecting them between Au electrodes in a mechanically controllable break junction (MCBJ) device architecture.[58−61]

Scheme 1

Syntheses of Ligands L1 and L2 and Complexes 1 and 2 Discussed in This Study

Results and Discussion

Design and Syntheses of Ligands and Complexes

The ligands L1 and L2 feature bithiophene and hexyl-bithiophene, respectively, supramolecular functional addends tailored with the iron(II)-coordinating BPP motif via an ethynyl spacer. The ethynyl spacer between the BPP and bithiophenyl entities ensures planar ligand skeletons, which are useful for developing conductive SCO systems.[19,62−70] The presence of bithiophene groups might enable the anchoring of the complexes at single-molecule junctions. Such anchoring could be used to study the spin-state dependence of conductance at a molecular level, a progressive step toward the realization of SCO-based applications. Alkyl chain containing ligand 2 was designed to facilitate the formation of the X-ray quality of single crystals of the corresponding complex.

Ligands L1 and L2 were synthesized by performing the classical Sonogashira coupling between 4-iodo-2,6-di(1H-pyrazol-1-yl)pyridine (I-BPP)[71] and the corresponding ethynyl precursor 5-ethynyl-2,2′-bithiophene or 5-ethynyl-5′-hexyl-2,2′-bithiophene[72] as shown in Scheme . See Figures S1–S4 for the 1H and 13C NMR spectra of the ligands. Treatment of Fe(BF4)2·6H2O, dissolved in 1 mL of acetonitrile (ACN), with L1 in chloroform resulted in the formation of complex 1 as a precipitate. A similar reaction yielded a clear red-orange solution of complex 2; diffusion of diethyl ether (Et2O) into the solution at 4 °C resulted in the formation of X-ray quality single crystals of 2.

Complex 2 crystallized with lattice acetonitrile solvents, as discussed in the section, X-ray Crystal Structures of . However, drying of the complex under a high vacuum resulted in the removal of cocrystallized acetonitrile solvent molecules, as inferred from thermogravimetric (Figure S5) and elemental analyses. Keeping in mind that the entire SCO investigation was done on the lattice solvent-free complex, we used the same notation 2 for the initial and dried samples throughout the script to be more concise.

X-ray Crystal Structures of L2 and 2

X-ray structure analysis revealed the crystallization of L2 (Figure a and Table S1) in the P1 space group, belonging to the triclinic crystal system, with one unique molecule in the asymmetric unit. In the crystal lattice, the ligands form a one-dimensional (1D) chain along the crystallographic b axis through C–H···N hydrogen bond interactions between the BPP moieties (d(C10–H10···N1) = 2.51 Å, (∠(C10–H10···N1) = 174.41°), and the 1D chains are packed in the ab plane (Figure S6). In addition, the thiophene and BPP moieties stack via π–π interactions along the c axis with short intermolecular contact distances of 3.32–3.40 Å.

Figure 1

X-ray crystal structures of (a) L2 and (b) complex 2. In the case of 2, the lattice acetonitrile solvent and counter cations are omitted for clarity. Thermal ellipsoids are drawn at 50% probability.

Complex 2 crystallized (Figure b and Table 1) in the P1 space group, belonging to the triclinic crystal system. The asymmetric unit of 2 contains a cationic [Fe(L2)2]2+ unit, two BF4– anions, one CH3CN molecule, and residual electron densities. The residual electron densities were flattened by employing the SQUEEZE subroutine of PLATON,[73] and the flattened electron densities (24 e– per asymmetric unit) are assigned to a disordered CH3CN molecule (22 e–), pointing to the possible presence of two additional lattice CH3CN molecules in the unit cell. One of the peripheral hexyl chains of the [FeL2]2+ unit is disordered over two positions. The bond lengths, angles, and distortion indices, obtained using the software OctaDist,[74] of the [FeL2]2+ unit summarized in Table S2 indicate the low-spin (LS)state of the complex at 173 K, corroborating well with the LS state of the complex observed from the magnetometry study discussed in the next section.

A non-terpyridine-embrace[75] packing pattern was observed in the crystal lattice of 2, as shown in Figure S7c. This observation is contrary to some of the BPP-based iron(II) complexes following a terpyridine-embrace packing pattern in their crystal lattice.[75,76] Six types of intermolecular interactions contribute to the packing of molecules in the crystal lattice of 2; see the discussion in Section S4 of the Supporting Information for more details. When one considers the role of such intermolecular noncovalent interactions in ordering of the complex entities in the crystal lattice, the complexes reported in this study are termed as supramolecular complexes.

Spin-Crossover Characteristics of 1 and 2

Variable temperature magnetometry studies were carried out to shed light on the SCO characteristics of complexes 1 and 2. A gradual spin-state switching with T1/2 = 254 K was observed for 1. At 5 K, a χT value of 0.21 cm3 mol–1 K was observed, indicating a LS phase contaminated with a remnant HS fraction along with a contribution from temperature independent paramagnetic (TIP) susceptibility. The TIP susceptibility arises due to the mixing of wave functions of the excited 5T2g state with the ground 1A1g state, termed as the second-order Zeeman effect.[77] Upon heating, a gradual LS-to-HS switching occurred; χT = 3.9 cm3 mol–1 K obtained at 400 K, indicating the HS state of the complex at that temperature. A second heat–cool cycle (cycle 2; Figure S8) performed at a scan rate of 5 K min–1 yielded a similar χT versus T profile as that of cycle 1, indicating the reproducible nature of spin-state switching characteristics of 1.

The rather high χT value obtained for the HS state of 1 compared to the one predicted on the basis of the spin only contribution (χT = 3 cm3 mol–1 K) indicates the presence of significant spin–orbit coupling due to unquenched orbital angular momentum associated with the HS iron(II) center, as reported for a iron(II)–BPP complex.[78] The spin-only value is commonly used as a reference to assign the HS state of iron(II) complexes. However, such usage is not entirely correct because the 5T2g state features triply degenerate t2g orbitals with the occupancy of four (t2g4), giving rise to orbital angular momentum. However, in the case of LS iron(II) complexes, the t2g and eg orbitals are completely filled and empty, respectively; therefore, the orbital angular momentum is quenched.

For complex 2, a χT value of 0.027 cm3 mol–1 K was obtained at 5 K, indicating the LS state of the complex. The small nonzero χT values obtained for 2 below 100 K are attributed to the temperature independent paramagnetic (TIP) susceptibility. Heating caused the onset of gradual LS-to-HS switching around 150 K, which continued until 310 K. A stepwise SCO, featuring two distinct steps with midpoints of the steps located at 313 and 321 K, followed the gradual phase. An abrupt spin-state switching ensued the stepwise region, leading to the completion of the LS-to-HS switching around 380 K. The saturated nature of the χT versus T plot and χT of 3.16 cm3 mol–1 K at 400 K evidence the HS state of 2 at that temperature. Upon cooling, the HS-to-LS switching profile retraced the heating profile, except in the stepwise region, where a small hysteresis (ΔT1/2 = 2 K) was observed. Overall, T1/2 of 356 K and three different types of SCO, gradual, stepwise, and abrupt, were observed for 2 in the first heat–cool cycle. The spin-state switching characteristics of 2 remained similar in the subsequent (second) heat–cool cycle (Figure S9), elucidating the stable and reproducible SCO characteristic of 2.

To check for the role of an ordered molecular organization contributing to the occurrence of the stepwise and abrupt spin-state switching in the crystalline form of 2, another form of the complex, hereafter referred to as complex 2a, was prepared, and the spin-state switching characteristic of the complex was investigated. Spin-state switching of 2a proceeded in a gradual manner without any step-like features, as shown in Figure S10. At 5 K, a χT value of 0.073 cm3 mol–1 K was observed, indicating the LS state of the complex; the small χT value is attributed to the temperature independent paramagnetic susceptibility. Upon heating, a gradual LS-to-HS switching occurred; χT of 3.023 cm3 mol–1 K was obtained at 400 K. Though the observed χT value is in the range expected for a HS iron(II) center (S = 2), the nonsaturation of χT versus T plot around 400 K indicates the presence of a remnant LS fraction. When the χT value of 3.16 cm3 mol–1 K observed for 2 was considered as the pure HS value, a LS-to-HS conversion of ∼96% was estimated for 2a at 400 K. Remarkably, a scan rate independent ΔT1/2 value of 5 K was observed for 2a around 360 K.

Our attempts to shed light on the structural facets contributing to the occurrence of stepwise spin-state switching in complex 2 are not successful due to the cracking of the crystals above 300 K, most probably due to spontaneous lattice acetonitrile solvent loss. In the absence of experimental validations, we attribute the occurrence of stepwise SCO to temperature-dependent changes in the nearest neighbor intermolecular interactions during the switching process.[79] Such changes could have contributed to the abrupt switching of a small fraction of complexes at 313 and 321 K separated by a plateau region. The existence of such region is attributed to a thermally stable phase, albeit in a small temperature range, composed of a mixed LS/HS proportion.

Small- and Wide-Angle X-ray Scattering (SWAXS) Studies

Variable temperature SWAXS studies were performed to get insights into the molecular organization of 2 and 2a. The SWAXS studies revealed reversible changes in the medium and wide-angle ranges of the SWAXS patterns of 2 and 2a (Figure and Figure S11) upon temperature variation, attributed to spin-state switching-induced changes in the molecular organization. A comparison between the SWAXS patterns of 2 and the simulated pattern obtained from the SCXRD study reveals that the molecular organization in 2 is different relative to the organization in the single-crystal lattice of 2. Such difference is attributed to the absence of acetonitrile-based C–H···F and C–H···N hydrogen bonding and C–H···π interactions in the lattice of solvent-free 2, as against the presence of such interactions in the single-crystal lattice of 2.

Figure 3

Variable temperature SWAXS studies. SWAXS profiles of (a) 2 and (b) 2a. The green traces in (a) and (b) are the pattern simulated from the single-crystal X-ray structure of 2. Blue stars mark the differences between 2 and 2a patterns in the initial LS states. Temperature-dependent variations of the peak positions are indicated by black (LS) and red (HS) stars.

The SWAXS patterns of 2 and 2a have quite comparable intensity profiles in the small-angle region, except for the presence of additional reflections marked with blue stars in Figure a,b and for small peak shifts due to slightly different lattice sizes/geometries. The comparable nature of the SWAXS profiles indicates the crystalline nature of 2a with similar unit cell parameters as that of 2. This also means that the molecular self-organization in the lattices of complexes 2 and 2a is not identical but rather similar. Incidentally, the switching behavior is maintained for both complexes, although with different switching characteristics (Figures and S10). This underlines that subtle changes in intermolecular interactions between 2 and 2a can have a significant impact on the spin-state switching process. Such attribution is in line with a study by Murray and co-workers demonstrating remarkable changes in spin-state behavior associated with polymorphs of a dinuclear iron(II) complex.[77] Both SCO behaviors are nevertheless reversible, as revealed by the switchover between typical SWAXS patterns of LS and HS states and in line with the magnetometry studies.

Figure 2

Spin-state switching characteristics of 1 and 2. χT versus T plots of complexes 1 (T1/2 = 254 K) and 2 (T1/2 = 356 K). Experiments were performed in the settle mode under an applied DC field of 0.1 T; a scan rate of 1 K min–1 was employed.

Computational Studies: Stretching-Induced Spin-State Switching in Complex 1

The spin-state switching characteristics of complexes 1 and 2 detailed in the previous section present SCO as a macroscopic phenomenon governed by intermolecular interactions in a crystal lattice. Spin-state switching could also be induced at single-molecule junctions,[23,41] as discussed in the Introduction. In such junctions, intermolecular interactions play no role, and the switching is controlled by how a molecule interacts with an external stimulus, for example, stretching or electric field. To probe the utility of stretching as a stimulus inducing spin-state switching in iron(II)–BPP junctions, density functional theory (DFT) calculations have been performed on the basis of a model in which complex 1 is wired between Au electrodes (Figure a). Complex 1 is taken as a representative example, considering the similar nature of the molecular structures of 1 and 2.

Figure 4

(a) The spin-crossover single-molecule junction, showing the molecular structure of complex 1, and its anchoring to the pyramid apexes of the gold electrodes. (b) Mechanism of the SCO: a change in the Fe–N ligand distance r is induced by pulling the gold electrodes, which induces stretching of the compound, therefore reducing the crystal field splitting Δ. The different filling of the orbitals leads to a change in the ground state spin from S = 0 to S = 2. (c) Energy dependence E–Emin of the two spin configurations S = 0, 2 of complex 1 in a vacuum, as a function of the stretching distance measured between the outermost sulfur atoms, referred to the equilibrium distance.

The ligand field interaction of approximate Oh symmetry felt by the iron(II) ion splits the 5-fold degenerate energy spectrum of its 3d electronic shell into two well-separated eg and t2g levels by a ligand field splitting energy Δ, as shown schematically in Figure b. The energy (Δ) depends on the average distance r between the central iron(II) ion and its neighboring ligand atoms approximately as r–5 and therefore is decreased by stretching it. At the equilibrium distance of the molecule, Δ is larger than the exchange interaction (pairing energy) J among the six d electrons in the iron(II) ion. As a result, complex 1 is in a low-spin (LS), S = 0, ground state (Figure b). However, Δ decreases upon stretching the molecule and becomes eventually smaller than J, making the eg states accessible. These levels are then filled to maximize the spin multiplicity according to Hund’s first principle, resulting in a high-spin (HS) state, S = 2, above a certain molecular length (Figure c). It is found from our studies that the LS–HS transition occurs when the molecule is stretched by 1.5 Å, which is about 5% greater than the initial molecular length, as shown in Figure c. This elongation corresponds to an increase of the Fe–N distance by about 0.15 Å, a value similar to that found for similar complexes in graphene junctions.[24] The value is also in line with about a 0.2 Å increase in Fe–N bond lengths observed for HS iron(II)–BPP complexes relative to their LS counterparts.[78,80] The calculated small bond length variation between the LS and HS states of complex 1 confirms that the spin crossover can be triggered by pulling electrodes in an MCBJ setup.

Our analyses of the electronic structure of the isolated complex 1 in LS and HS states reveal that the low-energy molecular states are mostly conjugated across the whole molecule (Figure a, top and bottom). Similarly, most of the low-energy molecular states of the S = 0 complex linked to the gold electrodes are conjugated across the whole molecule (Figure b, top), except for an empty state located at about +0.7 eV that has a large weight at the Fe atom. Similarly, the low-energy spin-up states of S = 2 of the gold-linked molecule are also conjugated across the molecule (Figure b, bottom), except for an occupied state located at about −0.9 eV. In contrast, the low-energy spin-down states have all significant weight at the Fe atom. These electronic distributions have an impact on the electron transmission function across the molecule T(E) that we depict in Figure . The transmission function of the S = 0 complex T0(E) shows a broad large-weight peak A at the Fermi energy and two narrow peaks B and C at positive energies, where only peak C corresponds to a state with a large concentration at the iron atom (Figure a, left). We expect that the A peak will dominate the low-voltage electrical response of the junction whenever the complex stays in the S = 0 state and that the impact of the B peak will be apparent at voltages larger than V ∼ 2EB/e ∼ 0.7 V. In contrast, the S = 2 complex shows spin-polarized transport since T2↑(E) ≠ T2↓(E). The spin-up transmission function only shows the peak E at low energies and furthermore has a low transmission T2↑ of ∼10–4 (Figure a, right). In contrast, the spin-down transmission function displays two low-energy peaks G and H that have a much larger transmission T2↓ of ∼10–1 and are largely concentrated at the iron atom. We expect that these two peaks will give rise to a spin-down polarized current shooting up at voltages (V ∼ 2EG,H/e) of ∼0.6 and 0.9 V, respectively.

Figure 5

(a) Density of states (DOS) of isolated complex 1 in LS (top) and HS (bottom) states. (b) DOS of complex 1 anchored to the gold electrodes in LS (top) and HS (bottom) states. The DOS projected onto the iron atom and the rest of atoms in the complex are shown in red and blue lines, respectively. The black up and down arrows represent spin-up and spin-down electrons, respectively.

Figure 6

(a) Transmission function T(E) for spin S = 0 (left) and S = 2 (right). Capital letters label transmission peaks generated by specific molecular orbitals. (b) Distribution of the density of states of each of the states labeled in (a) along an axis running longitudinally along the molecule.

Figure a shows the IV characteristics of the S = 0 junction. We confirm here that the junction displays a large low-voltage conductance for voltages up to about 0.25 V, which is driven by the energy location of the A peak above. In contrast, we see in Figure b that the S = 2 junction should show a very low conductance up to 0.6 V, where the down-spin peak G enters the voltage window followed by a further conductance increase at 0.9 V due to the H spin-down peak. The voltage-dependent ratio I/I shown in Figure c quantifies this different current response. Remarkably, this ratio reaches values close to 103 at the lowest voltages. Figure d indicates that the current should show a very large degree of polarization at those voltages.

Figure 7

Electrical transport properties of the junction. Current (I) versus voltage (V) characteristic of complex 1 for the (a) LS and (b) HS states. (c) Magneto-conductance ratio I/I and (d) the degree of spin polarization of the current for the S = 2 state.

Our simulations have strong implications for MCBJ junctions. We predict that, at the initial states of a pulling cycle, the low voltage conductance should be large and spin-unpolarized. However, the low voltage conductance should show a steep drop at the threshold pulling distance even though the junction would not be broken. This should be manifested by a conductance increase around 0.6 V, where a strongly spin-polarized current is expected to set in.

Single-Molecule Transport Studies

Transport measurements were performed at room temperature using the mechanically controlled break junction (MCBJ) technique, which uses a three-point bending mechanism to break lithographically defined gold nanowires on a flexible substrate. The technique enables the user to stretch the wire until it breaks while monitoring the conductance of the junction.[81] To measure the electrical properties of the molecules, a solution containing them is drop cast onto the nanowire. The solvent evaporates, leaving the molecules on the gold surface. During breaking, the gold nanowire ultimately ruptures, and molecules can be contacted to form molecular junctions. In this way, the conductance of molecules can be monitored while separating the electrodes until the molecule loses contact. The process of stretching the junction and monitoring the conductance is called a breaking trace and can be repeated many times. In the current experiments, 0.1 mM solutions of complex 1 and 2 in dichloromethane (DCM) were used for drop-casting the molecules onto the MCBJ setup.

Figure a,b shows two-dimensional conductance versus electrode displacement plots displaying 10 000 consecutive breaking traces for complexes 1 and 2, respectively. In both plots, little evidence is present of a fully stretched molecule, which would appear as a horizontal area of high counts (that is, a plateau) extending up to the length of the molecule. Figure c,d displays the corresponding one-dimensional conductance histograms, scaled such that at a conductance of 1G0 the count/trace is fixed at 1 for direct comparison. Here, G0 is the conductance quantum, the conductance value corresponding to a single bridging gold atom. The experimental measurement of this value ensures the proper formation of an atomic contact, which after rupture, results in two atomically sharp electrodes. In Figure c,d, the red curves in the conductance histograms correspond to the measurements in (a) and (b). Especially in (c), a broad peak appears around 3 × 10–5G0. This value corresponds to an electrode separation distance of about 0.5 nm, which is close to the lateral dimension of the complexes and can thus be indicative of the complexes perpendicularly oriented in the junction as schematically shown in for complex 1 in Figure . However, a laterally oriented, fully stretched molecule is the preferred choice to stretch the molecule to trigger stretching-induced spin-crossover.

Figure 8

Two-dimensional conductance versus electrode displacement plot containing 10 000 consecutive breaking traces after deposition of (a) complex 1 and (b) complex 2. One-dimensional conductance histograms scaled such that the conductance at 1G0 is fixed at 1 count/trace for (c) complex 1 and (d) complex 2.

Figure 9

Tentative orientation of complex 1 in the molecular junction.

Complex 1 shows gradual SCO with T1/2 = 254 K. The fast precipitation of the complex from the reaction mixture prohibiting the orientation of the switching entities in an ordered manner could have caused the gradual SCO. The random arrangement of the molecules in the powder mitigates the effective communication of a switching event of a molecule to its nearest neighbors, thereby reducing the cooperativity. Complex 2 is one of the BPP-based complexes showing the above room temperature (RT) SCO.[76] The origin of such SCO characteristics associated with the complex is attributed to soft-intermolecular interactions operating between the switching entities. The lack of solvent molecules in the lattice of the sample used for magnetic measurement indicates the role of the intermolecular organization in governing T1/2 and the nature of the spin-state switching process. The gradual, stepwise, and abrupt events coupled with the SCO of 2 are a testimony of complexities associated with the spin-state switching process. The absence of steps and abruptness in the spin-state switching of complex 2a, relative to 2, elucidates the important role of intermolecular interactions in inducing stepwise and abrupt spin-state switching and the sensitive nature of SCO even to the slightest of changes. The similar molecular organization of 2 and 2a inferred from the SWAXS studies elucidates the role of alkyl chains facilitating molecular organization irrespective of the preparation method. Recently, we have utilized the self-assembly directing nature of the dodecyl alkyl chain to realize similar spin-state switching characteristics in the bulk and thin-film states.[82] Overall, complex 2 adds to the family of iron(II) complexes tethered with alkyl chains.[33,82−86] Such complexes show interesting structure magnetic property relationships, such as alkyl chain conformation-dependent interconversion between two distinct HS–LS states,[86] reverse SCO,[87] and tail length-dependent T1/2.[84]

From the transport front, stretching-induced spin-state switching of iron(II)–BPP systems in single-molecule junctions is yet to be demonstrated both theoretically and experimentally to the best of our knowledge. Our attempts to experimentally demonstrate conductance switching in single-molecule junctions were not successful due to the perpendicular orientation of the complexes, instead of the fully stretched orientation, in the junctions. The DFT studies predicting the stretching-induced spin-state switching of 1 and the related spin-state-dependent conductance modulation are encouraging to develop molecular spintronic modules on the basis of the prototypical iron(II)–BPP systems.

Conclusions

The spin-state switching characteristics of supramolecular iron(II) SCO complexes, 1 and 2, featuring organic semiconductor-like ligands L1 and L2 elucidate the sensitive nature of SCO to subtle changes in the molecular environment. Crucially, the SCO active nature of the complexes implies the optimal ligand field strength around iron(II) centers conducive to single-molecule transport studies aiming for the elucidation of spin-state-dependent conductance switching. Computational studies revealed that it is possible to induce SCO by stretching complex 1 in an MCBJ-like device setup. Such elucidation well augurs for the development of SCO-based spintronic elements based on the prototypical iron(II)–BPP systems. The presence of the perpendicularly oriented complexes in the junctions, instead of the expected fully stretched complexes, inferred from the single-molecule conductance traces indicates that the complexes reported in this study need to be tailored with anchoring groups with an affinity for gold electrodes to demonstrate stretching-induced spin-state switching.

Methods

Syntheses and characterization data of the ligands and complexes discussed in this study are presented in Section S1 of the Supporting Information.

The X-ray diffraction data collection of the ligand L2 was carried out on a Bruker PHOTON-III CPAD diffractometer using Mo Kα radiation (λ = 0.71073 Å) under a cold nitrogen stream at 120 K. The X-ray diffraction data collection of the complex 2 was carried out on a Bruker APEX II DUO Kappa-CCD diffractometer using Cu Kα radiation (λ = 1.54178 Å) under a cold nitrogen stream at 173 K. The cell refinement and data reduction were performed using SAINT in APEX3 (APEX3: SAINT-Plus and SADABS, Bruker AXS Inc., Madison, Wisconsin, USA, 2016). A semiempirical absorption correction was applied using SADABS in APEX3. The structures were solved by direct methods (SHELXT Version 2014/5)[88] and standard difference map techniques and refined using full-matrix least-squares procedures on F2. The refinement and all further calculations were carried out using SHELXL Version 2014/7.[89] Anisotropic refinement was applied to all non-hydrogen atoms. All hydrogen atoms were placed at the calculated positions and refined using a riding model. The residual electron densities corresponding to the disordered solvents in complex 2 were flattened by employing the SQUEEZE subroutine of PLATON.[73] One of the peripheral hexyl chains of the [Fe(L2)2]2+ unit in complex 2 was refined over two positions with constraints on the temperature parameters (EADP commands).

Magnetic measurements were performed on a MPMS-3 SQUID-VSM magnetometer (Quantum Design). The temperature-dependent magnetization was recorded at an applied DC field of 0.1 T. Temperature sweeping rates of 5 and 1 K min–1 were employed. Gelatin capsules were used as sample holders in the 5 to 400 K temperature range.

Small- and wide-angle X-ray scattering (SWAXS) patterns were obtained with a linear monochromatic Cu Kα1 beam (λ = 1.5405 Å). The beam was obtained using a sealed-tube generator equipped with a bent quartz monochromator and a curved Inel CPS 120 counter gas-filled detector. Periodicities of up to 70 Å can be measured, and the sample temperature was controlled to within ±0.01 °C from 5 to 200 °C. The sample was filled in sealed cells of adjustable paths, and the exposure time was 4 h.

Simulations

We have used the ab initio code SIESTA[90] to extract the properties of the single-molecule complex 1 discussed in the body of the manuscript. Simulations were performed for complex 1 in vacuum as well as for a full junction, where complex 1 was linked to gold electrodes. We used a double-ζ polarized basis set for all atoms in the simulation having radii on the order of 7 Bohr or larger. We also used fully tested pseudopotentials for all atomic species in the simulations. We used the conventional exchange-correlation GGA-PBE functional.[91]

Simulations of the whole junction were carried out so that each gold electrode was oriented along the (111) crystallographic direction. Each electrode was then terminated by a four atomic layer pyramid having atom numbers of 10–6–3–1. Complex 1 was then inserted in the junction. Simulations were performed for a range of distances between the pyramid tips of the two electrodes. The complex was allowed to relax for each distance (d) until the largest atomic force was smaller than 0.02 eV/angstrom, and then, we extracted the total energy of the junction, E(d). The final junction arrangement was chosen at the tip–tip distance (d0) that yielded the lowest energy. We found that E(d0) < E(d0) as expected.

All transport calculations were performed with the GOLLUM package,[92] where the junction Hamiltonians were read from SIESTA’s output at the equilibrium tip distance, d0.