Room Temperature Spin‐Dependent Transport in 2D Hofmann‐Type Single‐Layer Network

Abstract

Room‐temperature magnetoresistance effect is reported on a 2D molecular‐based magnetic system using only one magnetic electrode, unlike the typical spin‐valve systems with two magnetic electrodes. Charge transport is measured using a scanning tunneling microscope on a molecular monolayer of a Hofmann‐type network consisting of a 2D [Pt(CN)₄Co]_x system. The layer is grown on a gold substrate by prior deposition of 4‐(ethyldisulfaneyl)pyridine (EtS‐Spy) so that the generated 4‐pyridinethiyl radical (pyS), anchored to gold through S, coordinates the Co^II ions through the pyridine. The formation of the 2D [Pt(CN)₄Co]_x layer is verified by the presence of only bridging cyanide bands using vibrational spectroscopy. Employing a 4‐mercaptopyridine‐functionalized magnetic nickel tip, the reversal of the nickel magnetization direction results in the shutdown of the peak corresponding to the transport through the path formed by Ni‐Spy‐Co‐pyS‐Au, around 10⁻⁴ G₀ of conductance. This assignment is confirmed by flicker noise analysis and Non‐Equilibrium Green's‐functions‐based density functional theory calculations, indicating that the 10⁻⁴ G₀ conductance feature corresponds to through‐bond charge transport while the one observed at 2·10⁻⁵ G₀ involves intermolecular contacts. This effect has been previously reported for magnetic molecules; however, its extension to 2D systems introduces an essential capability for applications in new spintronic devices.

Room‐temperature magnetoresistance effect in a 2D magnetic Hofmann‐type network has been found using break junction scanning tunneling microscopy technique. A [Pt(CN)₄Co]_x system is grown on a gold substrate using 4‐mercaptopyridine ligands as anchoring ligands with a magnetic nickel tip. The reversal of the tip magnetization results in the shutdown of the transport through the molecular‐based layer.

1. Introduction

To have spintronic devices with technological applications, magnetoresistance stands as the key fundamental property.^{[ 1 ]} There are widely established layered inorganic materials showcasing fantastic magnetoresistance properties exploited in disk heads through spin‐transfer (or orbit) torque memories, e.g., spin valves, to spintronic memristors.^{[ 2 ]} Spin valves have been extensively employed in technological applications.^{[ 3} , ^{4 ]} They are commonly fabricated by sandwiching a non‐magnetic layer between two ferromagnetic electrodes. Parallel or antiparallel alignment of the magnetization of the two magnetic layers results in a significant resistance change. Molecular alternatives hold promise due to their considerable tunability via chemical alterations and the large chemical diversity of molecular systems. The first proposals for molecular devices were focused on mimicking spin‐valve layered structures. In the early 2000s, many theoretical studies proposed spin‐dependent transport devices by sandwiching an organic molecular layer between magnetic contacts.^{[ 5} , ⁶ , ^{7 ]} This was experimentally confirmed in 2011 by Schmaus et al. with a hydrogen phtalocyanine sandwiched with two magnetic cobalt electrode.^{[ 8 ]} Also, in 2015 by single‐molecule junctions incorporating Fe‐terephthalic acid‐Fe under a magnetic field either parallel or perpendicular to the transport direction.^{[ 9 ]} Following previous work, recent advancements have showcased a handful of molecular spintronic devices exhibiting spin‐dependent transport at room temperature. Notably, Naaman et al. have presented extensive work using chiral systems.^{[ 10} , ¹¹ , ¹² , ^{13 ]} In their studies, the devices with only one magnetic electrode reversing its magnetization, induces a substantial change in electrical transport. The latter has been coined as the chiral‐induced spin selectivity (CISS) effect, which has been lately demonstrated on many chiral diamagnetic systems ranging from chiral perovskites to biomolecules like DNA and peptides.^{[ 14 ]} Dramatic changes have also been achieved in single‐molecule junctions using spin‐crossover systems based on triazole‐pyridine transition metal complexes [M(tzpy)₂(NCX)₂],^{[ 15} , ^{16 ]} where the magnetoresistance extension can be controlled upon different metal substitutions. Adittionally, other family of magnetic systems, Co^II and Cu^II metalloporphyrins, show a moderate effect because frontier orbitals are mainly localized in the porphyrin ligands.^{[ 17} , ^{18 ]} The above chiral or magnetic single‐molecule junctions present magnetoresistance without an external magnetic field and with only one magnetic electrode; the molecular backbone bridges between a non‐magnetic metal electrode, typically a noble metal like Au or Pt, and an ex situ magnetized ferromagnetic metal electrode, such as Ni or Co.^{[ 19 ]} The inversion of the magnetization direction of the magnetic electrode along the molecular axis leads to a substantial alteration in conductance through the molecular junction. In this scenario, using magnetic molecules, the remarkable observed spin‐dependent effects stem from a delicate interplay between the paramagnetic molecule's and the metal substrate's electronic structures, which bears significant spin‐orbit effects. This results in a net spin interface effect which is the cause of the observed magnetoresistance.^{[ 20} , ²¹ , ²² , ^{23 ]}

The aim of this article is to create 2D monolayers of molecular nature displaying perfect magnetoresistance at room temperature. Hence, we have grown a 2D single layer on a gold substrate that was previously functionalized with a 4‐pyridinethiyl radical (pyS· or ·Spy). This monolayer, generated by self‐assembling of Co^II and [Pt^II(CN)₄]²⁻ building blocks, would ideally afford a square‐grid framework where the Co^II and Pt^II cations are alternately linked by cyanide bridging ligands, forming Co─NC─Pt bonds (see Figure 1 ). Such a single molecular layer is attached to the substrate through the cobalt cation coordination to the nitrogen of the 4‐pyridinethiyl radical bound to the surface through the S atom. This structure is analogous to the host structure of Hofmann‐type clathrates,^{[ 24} , ²⁵ , ^{26 ]} which have been prepared as bulk powders and as multilayer thin films, especially of spin‐crossover Fe^II compounds.^{[ 27} , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ^{34 ]} The results obtained with magnetic molecules show that two fundamental requirements for magnetoresistance at room temperature are the presence of orbitals with unpaired electrons near the Fermi level of the electrodes and a relatively large interaction with the gold levels.^{[ 15} , ^{16 ]} The Co^II cation has been selected since we have previously seen, in mononuclear complexes, that Fe^II and Co^II systems with octahedral coordination are the most efficient to present magnetoresistance because the t_2g beta orbitals are half‐occupied and close to the Fermi level of the electrodes, but also to avoid sample oxidation processes when using Fe^II cations.^{[ 15} , ^{16 ]} Spin‐dependent charge transport will then be evaluated using the STM BJ technique^{[ 35} , ^{36 ]} (Figure 1a). For example, equivalent magnetic complexes of Ni^II and Mn^II do not exhibit magnetoresistance, the former because the symmetry of the alpha e_g orbitals does not combine well with anchoring ligands and gold, and the latter because its orbitals are very far from the Fermi level energy.

Figure 1.

a) Schematic representation of the studied magnetoresistance STM break junction of a 2D [Pt(CN)Co]_x layer attached to the gold substrate through the Au‐Spy‐Co sequence and through the 4‐pyridinethiyl radicals under a functionalized magnetic Ni tip electrode with the same radical. Light gray, pink, yellow, red, blue, dark gray, and white spheres represent platinum, cobalt, sulfur, oxygen, nitrogen, carbon, and hydrogen atoms, respectively, of the 2D molecular‐based layer. Electrode gold and nickel atoms are represented by golden yellow and brown colors, respectively. b) Schematic representation of the procedure to grow the {Co^II(pyS)₂Pt(CN)₄} Hofmann‐type host monolayer; the disulphide (EtS‐PyrT) to deposit PyrT molecules on the gold surface (1); the formation of the [Pt(CN)Co]_x layer (2); and finally, the completion of the axial coordination of Co^II with a mercaptopyrimidine (pyrT, 3) replacing the solvent EtOH molecules.

2. Results and Discussion

2.1. Growing 2D Hofmann‐Type Molecular Layers

The procedure for growing a monolayer of the Hofmann‐type network follows previous protocols used to prepare multilayer thin films of similar materials.^{[ 32} , ³³ , ³⁴ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ^{44 ]} To avoid sample oxidation processes, we use Co^II systems. Among the metals of group 10, Pt^II cations were selected, as this metal enables a unit cell size that resembles that of gold substrate supercells compared to equivalent Pd^II and Ni^II systems, facilitating the interface with the metal substrate. Although the Pt‐C and Pd‐C distances are similar, an analysis of the unit cell sizes and orientations indicates a better match for a 2D [Pt(CN)Co]_x lattice, where the Co atoms are aligned with the gold atoms to facilitate bridging coordination of the pyS to the two metals. Overall, the synthesis of the {Co^II(pyS)₂Pt(CN)₄} Hofmann‐type network is based on the successive exposure of the functionalized substrate to solutions of the corresponding building blocks, with intermediate cleaning steps with pure solvent. Several different temperatures (from the original −60 °C reported by Mallouk and coworkers^{[ 38 ]} to room temperature,^{[ 44 ]}) concentrations (typical values of 5–100 mM), and cleaning and immersion times are reported in the literature.^{[ 32} , ³³ , ³⁴ , ³⁷ , ^{43 ]} As previously described, the first step is the overnight reaction of the gold substrate with the EtS‐Spy disulfide (5 mM in CH₂Cl₂). This reaction time is longer than previously reported; however, shorter times may result in an incomplete formation of the monolayer.^{[ 45 ]} This procedure is more convenient than the direct reaction with 4‐mercaptopyridine since the homolytic cleavage of the S─S bond in the disulfide releases EtS· ethylthiyl radical groups on the surface, which space the pyS·4‐pyridinethiyl radicals on the gold surface. The monolayer deposition process can be easily followed by ellipsometry (see more details in Figure S1, Supporting Information), scoring an average height of 4.5 Å, which is compatible with a slightly tilted pyS (Figure 1b, left panel). Then, it is exposed to 5 mM Co(BF₄)_2·6H₂O in ethanol for 5 minutes and rinsed with the same solvent for 30 seconds at 4 °C. This step coordinates the Co^II ion to the nitrogen atom of the pyS radicals. The operation is repeated for K₂[Pt(CN)₄] in ethanol and cleaned for 1 min, and hence, the tetracyanoplatinate ion interconnects the different Co^II centers (Figure 1b, central panel). Although Sakaida et al.^{[ 44} , ^{46 ]} reported that 2D Hofmann‐type network could be grown at room temperature, at 4 °C, a good compromise was found by ellipsometry between the coordination of the different chemical species and the desorption ratio under ethanol. By ellipsometry, we estimated a height of 11.5 Å for the mentioned step, consistent with ethanol molecules working as axial ligands. Finally, the substrate is exposed to a 5 mM solution of 4‐mercaptopyridine in 19 mL of CH₂Cl₂ and 1 mL of ethanol for 10 minutes to substitute the axial ethanol. In this way, the solubility of 4‐mercaptopyridine in CH₂Cl₂ is improved while maintaining a low ethanol concentration. Later, the excess of 4‐mercaptopyridine is removed by rinsing the substrate for 30 s in pure CH₂Cl₂. The 16.7 Å height estimated by the ellipsometric measurements agrees with the expected height for the fully‐grown Hofmann‐type monolayer (Figure 1b, right panel). This layer thickness has also been confirmed by Atomic Force Microscopy (AFM, see Figure S2, Supporting Information). For charge transport measurements using the STM‐BJ technique, better performances are obtained without performing this third step and by functionalizing the STM tip with 4‐mercaptopyridine.^{[ 18 ]}

To check this point, the Infrared Reflection‐Absorption Spectroscopy (IRRAS) spectrum of the monolayer was compared to the Fourier‐Transform Infrared Spectroscopy (FT‐IR) of the equivalent bulk compound {Co(pyS₂Et)₂[Pt(CN)₄]} (Figure S3, Supporting Information). As determined via Powder X‐ray diffraction (PXRD, Figure S3, Supporting Information down), our bulk powder sample of {Co(pyS₂Et)₂[Pt(CN)₄]} is isostructural to the previously reported {Fe(pyS₂Et)₂[Pt(CN)₄]}.^{[ 43 ]} The comparison aims to demonstrate that the deposited layer has formed a 2D [Pt(CN)Co]_x network, indicating that the CN groups act as bridging ligands between the two metals and confirming the formation of the 2D layer. The vibration of the cyanide group is extremely sensitive to its coordination state; hence, a stretching above 2150 cm⁻¹ indicates that it is bridging two metals. As visible in Figure S3 (Supporting Information) middle, the monolayer spectrum has a single peak at 2179 cm⁻¹ while the {Co(pyS₂Et)₂[Pt(CN)₄]} bulk powder has one at 2166 cm⁻¹, and no significant signals appear below 2150 cm⁻¹. This confirms the formation of the [CoPt(CN)₄]_x monolayer via Co‐NC‐Pt coordination. Furthermore, X‐ray photoelectron spectroscopy (XPS) is also consistent with the expected for the monolayer deposited on gold (see Figure S4, Supporting Information).

2.2. Spin‐Dependent Charge Transport Properties

STM‐BJ is a well‐established technique that can be used to access the conductance of molecules when they are trapped individually in the STM‐based nanogap.^{[ 47} , ^{48 ]} One of the most common ways for creating single‐molecule junctions is the dynamic STM‐BJ‘s tapping approach, developed by Tao and collaborators, in which the tip is crashed into and driven out of contact with a substrate covered in molecules.^{[ 35 ]} Such an approach is not suitable for a periodic monolayer as the one studied here because the tip might disrupt the supramolecular structure during the pulling cycle due to the induced mechanical stress.^{[ 49 ]} An alternative STM‐BJ modality is its static “blinking” approach,^{[ 50} , ^{51 ]} which is based on the detection of spontaneous junction formation in a preset nanoscale gap between the two STM electrodes. In the blinking approach, instead of crashing the tip electrode against the substrate electrode, the tip is fixed at a constant distance from the surface (interelectrode distance) imposed by a prefixed setpoint tunneling current. Once the system has reached a state of mechanical stability, the current feedback control is switched off, and the current is continuously monitored over time [I(t)].^{[ 50 ]} Eventually, when a stochastic connection between the individual molecule and the two junction electrodes happens, a sudden increase (or “blink”) in the monitored current is detected until the molecule stochastically detaches. Consequently, the detected current returns to the initial setpoint value, thereby generating a telegraphic‐like signal. Hundreds of blinks are commonly collected to perform statistical analysis and determine the average molecular conductance.

To characterize the conductance of the studied supramolecular ensemble, STM BJ‐based blinking current measurements were performed using a non‐magnetic Au STM tip first (See Supporting Information, Figure S5, Supporting Information). As previously stated, the tip was functionalized with 4‐mercaptopyridine rather than undergoing complete coordination of Co^II with the 4‐mercaptopyridine ligand in the monolayer. In addition to measuring the complete {Co^II(Spy)(EtOH)Pt(CN)₄} monolayer, control experiments were conducted on the incomplete monolayer system lacking the tetracyanoplatinate backbone. This allowed characterizing the conductance of the Spy‐Co‐pyS pathway (Figure 2a). Furthermore, an incomplete supramolecular ensemble was also measured lacking the inclusion of either the Co or the Pt complex [see (Spy···pyS) in Figure 2a], to characterize a potential contact between the pyridines. The current measurement results show two single‐molecule conductance regimes. On the one hand, a high conductance regime, characterized by a mean value around 9.0·10⁻⁵ G₀, corresponds to the charge transport through the Spy‐Co‐pyS pathway. On the other hand, a low‐conductance regime, characterized by a mean value of ca. 1.9·10⁻⁵ G₀, corresponds to the intermolecular Spy···pyS contact. The assignment of each regime to a specific moiety will be corroborated later in this section by flicker noise analysis.

Figure 2.

a) 1D linear histograms of the conductance measurements, using the blinking approach and a gold tip, for the complete {Co^II(pyS)(EtOH)Pt(CN)₄} monolayer (blue), incomplete layer without adding the K₂[Pt(CN)₄] in the growth process (orange), and the case lacking any metal to facilitate the Spy···pyS contact (purple). Current is injected from the tip to the surface, refer to Figure S5 (Supporting Information), for the results of the complete monolayer in which the current was injected from the surface to the tip. b,c) 1D linear histograms of the conductance measurements, using the blinking approach, for the complete {Co^II(pyS)(EtOH)Pt(CN)₄} monolayer obtained with a magnetized Ni tip. Current is injected from the tip to the surface and (b) from the surface to the tip (c), respectively. The green and red traces correspond to the up and down directions of the magnetic moment, respectively.

To characterize the conductance of the studied supramolecular ensemble, STM BJ‐based blinking current measurements were performed using a non‐magnetic Au STM tip first (See Supporting Information, Figure S5, Supporting Information). As previously stated, the tip was functionalized with 4‐mercaptopyridine rather than undergoing complete coordination of Co^II with the 4‐mercaptopyridine ligand in the monolayer. In addition to measuring the complete {Co^II(Spy)(EtOH)Pt(CN)₄} monolayer, control experiments were conducted on the incomplete monolayer system lacking the tetracyanoplatinate backbone. This allowed characterizing the conductance of the Spy‐Co‐pyS pathway (Figure 2a). Furthermore, an incomplete supramolecular ensemble was also measured lacking the inclusion of either the Co or the Pt complex [see (Spy···pyS) in Figure 2a], to characterize a potential contact between the pyridines. The current measurement results show two single‐molecule conductance regimes. On the one hand, a high conductance regime, characterized by a mean value around 9.0·10⁻⁵ G₀, corresponds to the charge transport through the Spy‐Co‐pyS pathway. On the other hand, a low‐conductance regime, characterized by a mean value of ca. 1.9·10⁻⁵ G₀, corresponds to the intermolecular Spy···pyS contact. The assignment of each regime to a specific moiety will be corroborated later in this section by flicker noise analysis.

As previously reported,^{[ 15} , ^{16 ]} we conduct spin‐dependent conductance experiments using a ferromagnetic tip in our STM BJ platform. In this work, we conduct analogous experiments using the blinking approach with opposite directions (up and down) of the Ni tip magnetization. The results obtained align with those reported in previous works. Upon injection of the current from the substrate under down magnetization of the Ni tip, a conductance peak appears around 9·10⁻⁵ G₀ (see green histogram in Figure 2c), which is also consistent with that characterized for the case using the STM Au tip (see Figure 2a). However, when the magnetization of the tip is reversed (up direction), the 9·10⁻⁵ G₀ conductance feature disappears, and only the conductance peak assigned to the low‐conductance Spy···pyS contact is present. The conductance of this Spy···pyS contact has been shown to be insensitive to tip magnetization, as it is already present under non‐polarized currents (see Supporting Information), similar to what has been reported for non‐magnetoresistive electron pathways in magnetic metalloporphyrin systems.^{[ 17 ]} Nonetheless, when the supramolecular ensembles are measured under injection from the tip (see green and red histograms in Figure 2b), the conductance values show no variation for the two magnetization directions of the Ni tip. These results demonstrate the fundamental role of the magnetic monolayer in contact with the near‐Fermi levels of the Au substrate. The concept of spinterface has been introduced to describe the modifications in the electronic structure of metallic surfaces when interacting with magnetic molecular systems.^{[ 21} , ²² , ^{23 ]} Additionally, it is worth mentioning the studies that show magnetoresistance due to the spin‐crossover state change in compounds.^{[ 52} , ⁵³ , ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ , ⁵⁸ , ⁵⁹ , ⁶⁰ , ⁶¹ , ⁶² , ⁶³ , ⁶⁴ , ^{65 ]}

The evolution of junction conductance (via the blinking event) during the chemical contact of the supramolecular ensemble with both electrodes provides valuable insights into the structural nature of the studied monolayer. A standard phenomenology observed in many blinking events, as illustrated in Figure 3a, is that well after the junction has formed (blue region), transitory current drops close to the detection limit (see arrows) are likely to occur before a sequential and stepped final decay is observed (red and yellow regions in Figure 3a).^{[ 66} , ^{67 ]} Both the transitory drops and the final stepped decay can be attributed to the structural collapse of the supramolecular ensemble. The former results from transient disconnections between the ensemble and the electrodes. The latter are a consequence of a sequential disconnection that exposes Spy···pyS contacts due to an incomplete layer. As indicated by the integer conductance values, ca 4·10⁻⁵ G₀ and 2·10⁻⁵ G₀ (red and yellow regions, respectively in Figure 3a) during junction collapse, multiple Spy···pyS contacts in parallel can form,^{[ 68 ]} up to two in this case, based on our previous characterization (Figure 2) given that the conductance of Spy···pyS is equal to 2·10⁻⁵ G₀. We speculate that the structural collapse of the supramolecular ensemble, which causes both phenomena, is driven by heat generation in the metal layer during charge transport caused by electron‐phonon interactions.^{[ 69 ]} Given the timescales of the blinking events, this heating may be substantial enough to damage the ensemble, leaving only the pyridine groups in contact with the electrodes. In the right part of Figure 3a, for example, there is an intact shorter blink event. Note that only the longest left trace shows junction collapse with multiple steps in the current decay, likely due to prolonged charge transport causing structural damage to the clathrate. We rule out the possibility of a scenario where the tip stochastically connects and disconnects between the Co and free pyridines. This is because the stepped decay via Spy···pyS contacts represents an irreversible stage, from which the current signal never recovers to the higher levels associated with the complete supramolecular ensemble, around 9·10⁻⁵ G₀.

Figure 3.

a) Conductance versus time during one of the blinks for the {Co^II(pyS)(EtOH)Pt(CN)₄} monolayer. Blue, red, and yellow regions correspond to junctions formed with the network, double and single contacts between Spy···pyS groups, respectively. 2D histograms of normalized noise (NP/G) versus G b) for the low‐conductance peak through weak interactions between pyridinethiyl radicals on the gold electrodes and c) high‐conductance peak through the strong Spy─Co─pyS bonds, including the metal‐cyanide layer. The dotted contours correspond to bivariate Gaussian surface distributions fitted to the experimental data.

2.3. Flicker Noise Analysis

Further analysis based on power spectral density (PSD) flicker noise analysis has been conducted, providing additional evidence for the coexistence of junctions through weak interactions between the 4‐pyridinethiyl radicals and those with the metal‐cyanide layer between the two pyS anchoring ligands. In this type of fast‐temporal noise analysis,^{[ 70 ]} the current signal of individual single‐molecule junctions (blinking trace with subtracted background) is analyzed in the frequency domain. The noise power (NP) is extracted by integrating the PSD in the 100–1000 Hz range, following the specifications of our experimental setup, in line with previous research.^{[ 71 ]}

PSD can be employed to extract detailed information about the structure and associated transport characteristics of single‐molecule junctions at room temperature.^{[ 72 ]} A dominant noise source in the junction is the atomic motion on the metal electrode surface, which results in fluctuations in the molecule‐electrode coupling and, thus, in the conductance. As has been widely reported,^{[ 71} , ⁷³ , ^{74 ]} the noise power (NP) shows a power law dependence on the single‐molecule conductance G, as NP ∝ Gⁿ, with a scaling exponent n close to 1 when the charge transport is purely through strong bond interactions, as covalent or coordination bonds referred as through‐bond transport. The n scaling exponent increases to 2 when there is a tunnel junction and, thus, when the charge transport pathway implies weak intermolecular interactions, usually referred to as through‐space transport. We normalized the single‐molecule conductance (NP /G) obtained for each blink and represented them in front of the single‐molecule conductance to build 2D maps, as shown in Figure 3b,c. To analyze the correlation between overall NP and single‐molecule conductance, the value of n is extracted for which NP/G and G are not correlated, in accordance with the conditions under which Pearson's correlation coefficient is equal to zero.

We find that the NP values scale as G^1.80 and G^1.15 for the pyS···pyS and Spy‐Co‐pyS junctions, respectively. This implies that in the first case, a through‐space coupling mediates the charge transport. This leads us to reinforce our hypothesis that the observed phenomenology may be due possibly to weak π interaction between 4‐pyridinethiyl radical molecules deposited on the gold substrate and those functionalizing the Au electrode,^{[ 75 ]} which is a consequence of incomplete metal layer formation. The incomplete layer facilitates a strong through‐space charge transport, as evidenced by values approaching 2. Conversely, the reduced n of 1.15 associated with the G_HC is consistent with a through‐bond‐dominated charge transport. This further supports our hypothesis about forming a complete junction with a metal‐cyanide layer coordinated with the pyridine molecules between the two Au electrodes, stabilizing the junction and reducing the correlation between single‐molecule conductance and noise.

2.4. Theoretical Study

Transport properties were studied using full‐consistent non‐equilibrium Green's functions (NEGF) using DFT methods^{[ 76} , ^{77 ]} to calculate the system with semi‐infinite electrodes as implemented in the ATK 2023.09 code.^{[ 78 ]} Different functionals have been used, as it is well known that generalized gradient approximation (GGA) functionals, by underestimating the band gap, yield too high transmission values. The meta‐GGA functional r²SCAN (Figure S7, Supporting Information see Supporting Information) has been tested,^{[ 79 ]} and the calculated transmission values are relatively large. In a previous study with Co^II metalloporphyrins a very good reproduction of the experimental conductance values was obtained using a PBE+U functional^{[ 80} , ^{81 ]} with a value of U = 4 eV for the 3d cobalt orbitals.^{[ 18 ]} To generate a periodic system in the xy plane it is necessary to find a match between the gold cell on the Au(111) face and the {Co^II(pyS)₂Pt(CN)₄} layer (see Figure 4 ). The calculated transmission at the Fermi level, 8.2·10⁻⁵, matches very well with the experimental conductance value (8.93·10⁻⁵ G₀) shown in Figure 2. This fact suggests that the primary transport mechanism is likely due to coherent tunnelling, as evidenced by agreement with the Landauer approach implemented in the NEGF‐DFT method.^{[ 76} , ^{77 ]} The main difference in comparison with the r²SCAN results is that the levels corresponding to the empty t_2g Co^II orbital practically overlap the Fermi level, resulting in a huge transmission value at such energies. The PBE+U method shifts such levels to higher energy; thus, the transport is mainly due to the occupied beta t_2g Co^II orbitals (see transmission eigenstate and pathways in Figure S8, Supporting Information). The magnetoresistance effect in magnetic systems presents the same problem as the chiral‐induced selectivity effect,^{[ 10} , ¹¹ , ¹² , ¹³ , ^{14 ]} namely that it cannot be reproduced at the theoretical level using state‐of‐the‐art methodologies, such as NEGF‐DFT. In these cases, this method uses a nickel electrode, which provides the same transmission for both magnetizations of the magnetic electrode.

Figure 4.

a) Structural periodic model employed in the DFT calculations. b) DFT transmission curves for the two spin channels (blue‐up, orange‐down) were calculated with DFT+U (U_Co 4 eV) (see Figure S7, Supporting Information for comparison with r²SCAN results). The dashed line indicates the highest transmission of the beta channels at the Fermi level.

3. Conclusion

An ideal room‐temperature magnetoresistance effect has been observed in a 2D molecular‐based magnetic system. This behavior is equivalent to that previously found by some of us in molecular Fe^II and Co^II magnetic complexes.^{[ 15} , ^{16 ]} To the best of our knowledge, there are no reported studies of a molecular‐based 2D material showing perfect magnetoresistance at room temperature.^{[ 82} , ^{83 ]} Many studies on magnetoresistance have indeed been conducted for non‐molecular 2D systems, such as Fe₃GeTe₂ or Fe₃GaTe₂ ternary transition metal chalcogenides, among other magnetic 2D materials (CrI₃, CrSBr…).^{[ 84} , ^{85 ]} Still, these are entirely different devices from our molecular‐based system, since they are in some cases grown using molecular beam epitaxy or chemical vapor deposition techniques and usually prepared as thin films, usually through easy exfoliation, to be placed the electrodes without anchoring ligands.^{[ 82} , ^{83 ]} Furthermore, in most cases, these studies generally operate at low temperatures, except for the two Fe/Te systems mentioned above, which exhibit magnetic order at room temperature.^{[ 85 ]} Additionally, magnetoresistive devices utilizing such 2D inorganic magnetic materials are based on the change in orientation of the layers (anisotropic magnetoresistance) or spin‐valve devices, which feature two magnetic electrodes separated by a non‐magnetic insulating or semiconducting layer (giant or tunnelling magnetoresistance, respectively).^{[ 82} , ^{83 ]} In our case, 100% magnetoresistance is achieved with only one magnetic electrode, namely the STM tip, due to the electronic structure of the Co^II cations in the [Pt(CN)Co]_x layer. This makes a difference with the spin‐crossover magnetoresistive devices based on single molecules, which require a change in the ground state (high‐spin ↔ low‐spin) of the molecule caused by an external stimulus (temperature, light, electric field), and a net difference in the transport properties between both spin states.^{[ 64} , ^{65 ]}

In our study, the transport properties were measured using a STM BJ configuration on a molecular single layer of a typical Hofmann clathrate host compound, specifically a [Pt(CN)Co]_x layer. This single layer was carefully grown on a gold substrate after depositing an Et‐S‐S‐py disulfide, dissociating the S─S bond and ensuring that 4‐pyridinethiyl radical units were available to coordinate with Co^II through the pyridine, while the ethylthiyl radical group acted as spacers. For the selection of the cation for the square planar coordination we performed an analysis of the periodicity of the Au(111) surface and the single‐layer to ensure that the two periodicities matched. This would facilitate the coordination of the thiols of the mercaptopyridines with the hollow positions on the surface and, simultaneously, of the nitrogen atoms of the pyridines with the Co^II cations of the single‐layer. The linear Co···Co distance in the monolayer is around 10.4 Å for Pd and Pt (for Ni, it is 9.9 Å), and it corresponds to one of the distances between hollow centers of gold, which are the most stable positions for thiol coordination. Thus, in principle, Pt^II or Pd^II cations would be the two best options according to this criterion of periodicity in the interface. This layer growth was also verified using ellipsometry, AFM, and XPS. The formation of the 2D [Pt(CN)₄Co]_x grid layer was also confirmed by the presence of bridging CN⁻ ligands, as verified by vibrational IRRAS spectroscopy.

Using a 4‐mercaptopyridine‐functionalized nickel magnetic tip, we observed that the inversion of the nickel magnetization caused the disappearance of the peak corresponding to the transport pathway formed by Spy‐Co‐pyS, around 10⁻⁴ G₀ that appears only for the up magnetization. Flicker noise analysis confirmed that this peak corresponds to a through‐bond situation, in contrast to the peak at 2·10⁻⁵ G_0, which is attributed to a pathway involving intermolecular contacts, a result also observed in measurements without the [Pt(CN)Co]_x layer. NEGF‐DFT calculations also confirm the expected transport through the Spy‐Co‐pyS transport pathway. The key ingredient for the magnetoresistance effect is the role of the occupied beta t_2g orbital levels of the Co^II cations, which appear very close to the Fermi level of gold and are responsible for electron transport. While this effect has been previously reported for magnetic molecules, its extension to 2D systems introduces significant potential for device applications. At a theoretical level, it is not possible to reproduce the magnetoresistance effect with NEGF‐DFT calculation, as in the chiral systems and the chiral‐induced spin selectivity effect,^{[ 10} , ^{11 ]} the same values are obtained for the two magnetizations of the magnetic electrode. One of the limitations of this methodology is the exclusion of the effect of excited states on transport, particularly given that the experiments are conducted at room temperature. In these cases, the concept of spinterface and the role of the spin‐orbit coupling of the substrate that generates in materials such as gold a spin polarization of the surface has been introduced.^{[ 15} , ^{16 ]} The interaction of the magnetic orbitals of the molecule with gold is different depending on the orientation of the spin of the molecule aligned with the field generated by the tip.

The ability to control and manipulate magnetic properties at the molecular level in a 2D system could lead to innovations in data storage, sensors, and other technologies that rely on precise magnetic control. The stability, compared to devices based on molecules, is essential for the reliable performance of the system in real‐world applications. In summary, our findings demonstrate a perfect magnetoresistance effect in a 2D molecular‐based magnetic system. This breakthrough paves the way for future research and development in spintronics and its applications in electronic devices.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Aguilar M. R., Martín‐Rodríguez A., Gómez‐Coca S., et al. “Room Temperature Spin‐Dependent Transport in 2D Hofmann‐Type Single‐Layer Network.” Small 21, no. 50 (2025): e10193. 10.1002/smll.202510193

Data Availability Statement

The data supporting the findings of this study are available from the corresponding author upon reasonable request.