Tailoring the electronic configurations of YPc₂ on Cu(111): decoupling strategies for molecular spin qubit platforms†

Soyoung Oh^abc, Franklin H. Cho^ad, We-hyo Soe^ad, Jisoo Yu^ab, Hong Thi Bui^ab, Lukas Spree^ad, Caroline Hommel^ad, Won-Jun Jang^ad, Soo-hyon Phark^ad, Luciano Colazzo^ad, Christoph Wolf*^ad and Fabio Donati*^ab
aCenter for Quantum Nanoscience, Institute for Basic Science (IBS), Seoul, 03760, Republic of Korea. E-mail: wolf.christoph@qns.science; donati.fabio@qns.science
^bDepartment of Physics, Ewha Womans University, Seoul, 03760, Republic of Korea
^cThe Clarendon Laboratory, Department of Physics, University of Oxford, Oxford OX1 3PU, UK
^dEwha Womans University, Seoul, 03760, Republic of Korea

First published on 25th July 2025

---

---

1. Introduction

Molecular spin qubits offer inherent discrete energy levels to realize scalable qubit platforms.^1,2 This molecule-based architecture allows for qubit design through chemical approaches, enabling precise tailoring of the electronic structure, molecular spacing, and spin noise from the environment.^3,4 Best molecular spin qubits exhibit a long spin relaxation time (T₁) of tens of ms and a spin coherence time (T₂) of hundreds of μs.^5–7 While the molecular spins in magnetically diluted bulk and solution states can be conveniently measured using ensemble-averaging techniques, difficulties arise in addressing single qubits individually.¹

Conversely, molecular spins adsorbed on a surface provide a viable way to read out such molecules at the atomic scale using scanning probe techniques.^8,9 Moreover, the capability of molecular spins to self-assemble on single crystal surfaces^10,11 makes them stable and tunable building blocks for nanofabricating quantum devices.¹² Both molecule–surface and intermolecular interactions significantly impact the molecular spin states; hence, the investigation of these properties is essential to achieve precise control of the molecular spin configurations. For example, molecular packing and adsorption sites on the surface can regulate the charge transfer and electronic structure.^13,14 Similarly, the electronic structure can also change depending on whether the molecules are in direct contact with a metal substrate or buffered by a diamagnetic layer.¹⁵ The latter strategy can additionally serve to reduce the coupling between the molecular spin and the substrate,¹⁶ which is known to be a source of decoherence. For both the buffer layer and the molecular spins, a flat structure and an appropriate symmetry of the molecule are desirable properties for a regular assembly of the molecular layer.^16,17

Metal phthalocyanine double-decker (MPc₂) is one of the prevalent platforms of molecular spin qubits on a surface with the aforementioned properties.^18–21 The flat geometry of MPc₂ facilitates great structural stability, which contributes to the flat adsorption and self-assembly on metal substrates. Furthermore, MPc₂ is robust during thermal sublimation in ultra-high vacuum (UHV).^8,22 When a lanthanide metal (e.g., Tb, Dy, or Er) is used,²³ such molecules exhibit very long T₁ and T₂,^24,25 as well as single-molecule magnet behavior with enhanced magnetic stability.^26,27 However, lanthanide phthalocyanine double-deckers consist of two exchange-coupled spin systems; one is the high-spin from 4f electrons localized on the lanthanide metal ion, and the other is the spin S = 1/2 π-radical delocalized over the two phthalocyanine (Pc) ligands,²² resulting in a complex magnetic structure with large spin multiplicity that can cause qubit leakage at high temperatures.^6,28 In addition, coherently controllable electron spin states of such phthalocyanine double-deckers are primarily limited to the radical spin transition due to the strong magnetic anisotropy of most lanthanides, making them less ideal as molecular qubit candidates.¹⁸ A more suitable system can be created by replacing the lanthanides with diamagnetic ions. In particular, yttrium phthalocyanine double-decker (YPc₂) has a diamagnetic trivalent yttrium ion that stabilizes a single, quasi-isotropic unpaired electron in the Pc ligands,^29,30 endowing a simple and clear understanding of the spin dynamics of a nearly-free electron. Due to its low spin–orbit coupling and weak hyperfine interaction with the nuclear spins in the molecule, YPc₂ exhibits a single and narrow electron spin resonance (ESR) spectrum in dense powder,³⁰ single crystals,¹⁸ and few-nm thick molecular films.³¹ In addition, it shows spin nutations up to room temperature and sizable magnetic coupling between molecules in a magnetically dense phase,³⁰ indicating its potential as a molecular spin qubit in realistic applications. Revealing its spin properties in the form of a self-assembled film on a surface requires the use of a surface-sensitive ensemble^31–33 or a scanning tunnelling microscope (STM) combined with an ESR setup.^31–33 However, it is crucial to first assess the feasibility of such characterization by investigating the morphology and electronic properties of YPc₂ on single crystal surfaces, which can serve as a support for the molecular film in these types of setups.

Here, we investigate the morphology and electronic configuration of YPc₂ deposited directly on a Cu(111) surface with a few-layer thick zinc phthalocyanine (ZnPc) decoupling layer inserted between Cu(111) and YPc₂. Specifically, we chose Cu(111) in view of characterizing these molecules using our recently developed surface-sensitive ensemble ESR setup,³¹ whose functioning is presently optimized for this type of surface, while we used ZnPc as a diamagnetic spacer to tune the hybridization between YPc₂ and the Cu surface. In all cases, we observe flat adsorption of the YPc₂, with the Pc plane parallel to the surface. On Cu(111), the mismatch between the four-fold symmetry of the molecule and the global three-fold symmetry of the surface is reflected in the misalignment between the molecular lattice axis and the crystal lattice orientation. When only a single monolayer (ML) of ZnPc is grown on Cu(111), the subsequently deposited YPc₂ frequently embeds within the ZnPc layer. However, no intermixing is found when YPc₂ is deposited on 2 or more MLs of ZnPc. Scanning tunneling spectroscopy (STS) reveals a large gap of more than 2.3 eV between the highest occupied and the lowest unoccupied molecular orbitals (HOMO/LUMO) when YPc₂ is directly adsorbed on the Cu(111) surface and embedded into the ZnPc layer. Conversely, when adsorbed on top of another YPc₂ layer or on top of 2 MLs of ZnPc, the HOMO splits into singly occupied and singly unoccupied molecular orbitals (SOMO/SUMO), while the gap is largely reduced. Combining these results with density functional theory (DFT), we find that when YPc₂ is directly adsorbed on Cu(111), the hybridization between molecular orbitals and surface electrons quenches the unpaired electron, while a pristine electronic configuration with S = 1/2 is attained when YPc₂ is decoupled from the metal using the ZnPc layer. Our results highlight the potential of YPc₂/ZnPc heterostructures as a robust molecular spin qubit architecture, contingent upon the ability to control its electronic configuration through a molecular decoupling layer.

2. Experimental and computational methods

Samples were prepared and measured using a home-built and a commercial STM (PanScan Freedom, RHK Technology). The former is equipped with an Auger electron spectrometer and was used to characterize the growth and composition of the molecules upon deposition on the Cu(111) surface. The latter was used to investigate the electronic structure of the surface-adsorbed molecules. In both machines, the preparation and the STM chamber are connected, enabling the transfer of samples in UHV. This ensures that the sample surface is unchanged while transferring, allowing reproducible results. The Cu(111) surface is cleaned in the preparation chamber (base pressure ∼1 × 10⁻⁹ torr) by several cycles of Ar⁺ sputtering (pressure 5 × 10⁻⁶ torr, energy 1 keV) followed by thermal annealing at around 673 K. The quality and (111) surface termination of the substrate were confirmed by STM measurements before molecule deposition. The YPc₂ molecules were synthesized in-house following a solvothermal route.³⁴ In the process, 2.685 g 1,2-dicyanobenzene (8 eq.), 0.844 g yttrium acetate monohydrate (1 eq.), and 1.588 g 1,8-diazabicyclo[5.4.0]undec-7-ene (4 eq.) were mixed in 30 mL ethanol and heated in a Teflon lined autoclave at 190 °C for 12 hours. The mixture was allowed to cool down to room temperature over the course of 24 hours. The crude product was washed with n-hexane and diethylether on a glass filter frit. The resulting powder was mixed with 0.5666 g of periodic acid (H₅IO₆) and stirred overnight in 400 mL of a 1:1 mixture of chloroform and methanol to oxidize anionic YPc₂ species. Filtration and washing with 1:1 CHCl₃/MeOH over a glass filter frit were employed to remove insoluble byproducts like free base Pc. The solution was then reduced in volume under vacuum and dried with 30 g of activated alumina. Purification of the product was facilitated by column chromatography on deactivated alumina with 9:1 CHCl₃/MeOH and collecting the first green band eluting from the column. The identity of the target compound was confirmed by UV-Vis spectroscopy and mass spectrometry. The ZnPc molecules were purchased from Sigma-Aldrich (97% purity). Both compounds were thermally sublimated on the sample kept at room temperature from a heated crucible. For the home-built STM, the deposition of YPc₂ is performed using a home-built evaporator at 723–733 K, where the temperature of the crucible was measured using a K-type thermocouple wire positioned inside the crucible. For the RHK STM, we use a Kentax 3-cell thermal evaporator (TCE-BSC) at 633–653 K. Due to the different distances from the sample, the deposition temperatures in the two systems were adjusted to obtain a sublimation rate of ∼0.01 ML min⁻¹. For the ZnPc, we set the sublimation temperature at 623 K to obtain a rate of 0.1 ML min⁻¹.

All STM and STS measurements were performed at 10 K using the RHK STM at a base pressure <3 × 10⁻¹⁰ torr, except for the data in Fig. 1(a), which was acquired using the home-built STM operating at 120 K with a base pressure <4 × 10⁻¹⁰ torr. Images were taken in a constant current mode, while differential conductance (dI/dV) spectra were obtained using a lock-in amplifier in a constant height mode. The voltage bias is swept with added AC bias modulation of 20∼50 mV. The dI/dV maps were measured with a modulation of 50 mV turned on in the constant current mode. We scanned at a speed of 1∼2 nm s⁻¹, allowing the tip to stay over the spatial region of a single pixel long enough to obtain the current signal with an integration time of 100∼200 ms. As a result, we acquired the topographic information and electronic configuration of the molecules at the same time. All STM images are filtered using WSxM software.⁴⁵

Ab initio calculations were performed using plane-wave DFT with pseudopotentials as implemented in Quantum Espresso (V 7.3). We used PAW pseudopotentials from the PSlibrary (V 1.0.0).³⁵ Exchange–correlation was treated using the PBE generalized gradient approximation.³⁶ The cutoff for the basis was set to 60 Ry (600 Ry for the charge density), dispersive forces were treated with the revised VV10 method and only the Gamma point was used in the integration of the Brillouin zone. Slab systems were built based on 3 MLs of Cu(111), up to 4 ZnPc molecules and a single YPc₂ molecule, padded by ∼2 nm of vacuum in the z-direction. We used 8 × 7 lateral supercells, which separate adsorbed YPc₂ molecules by at least 3 Angstroms (the closest ligand to ligand distance). We tested the convergence of all relevant quantities with respect to the k-point sampling and found a 1 × 1 × 1 grid sufficient for the slab calculations (see Fig. S3†). Molecules in a vacuum were calculated in a 3 × 3 × 3 nm vacuum box. All systems were relaxed until forces were lower than 0.02 eV per Angstrom and energy differences were below 0.001 eV before calculating the electronic ground state. To calculate the ionization potential (electron affinity) we use the Delta SCF method by adding or removing one electron whilst the molecules are frozen at the geometry of the neutral system. To decouple the charged cells from their periodic images, we used the Markov–Payne method.³⁷

3. Results and discussion

3.1. YPc₂ on Cu(111)

The top view of the molecular structure of YPc₂ is shown in Fig. 1(a). The top Pc is represented by blue-colored nitrogen and black-colored carbon atoms, while the bottom Pc is indicated by red-colored nitrogen and pink-colored carbon atoms. Each Pc has four reflection planes, rotated by 45° relative to each other. We marked one representative symmetry plane for each Pc, M₁ for the top Pc and M₂ for the bottom Pc. The yttrium ion serves as a bridge between the twisted Pc ligands. As a result, YPc₂ exhibits 4-fold D_4d symmetry.

We first characterized the orientation of isolated YPc₂ on Cu(111) (Fig. 1(b)) and how they self-assemble into molecular patches on Cu(111) (Fig. 1(e) and (f)) with respect to the 〈111〉 surface lattice vectors using STM. We determined the latter using dislocation line defects generated by poking the STM tip into the surface, while we took the azimuthal rotational angle of M₁ with respect to the horizontal scan direction of the STM images. As shown in Fig. 1(c), isolated YPc₂ molecules do not adsorb randomly on the surface but rather align along specific angles spaced at 30° intervals. These preferred M₁ angles correspond to the three 〈111〉 surface lattice vectors; see Fig. 1(d). This result is in line with the previous study of metal-free Pc on Au(111).

Increasing the coverage of YPc₂ leads to the formation of molecular patches with two types of self-assembled domains. We find one domain showing molecules in checkerboard patches without a visible internal lobe structure (Fig. 1(e)), while the other domain has molecules showing a visible internal lobe structure (Fig. 1(f)). In line with the previous work,²² we label these two domains as phase I and II, respectively. The more abundant phase I is characterized by a dark molecular center with no discernible internal structure and molecular lobes of the top Pc overlapping with nearby ones. Conversely, for phase II, the molecular lobes of the top Pc avoid overlapping with each other, resulting in a side-to-side self-assembled arrangement. We determine the orientation of the molecular patches by comparing the molecular unit lattice vectors C₁ and C₂ with respect to the 〈111〉 surface lattice vectors, with focus on the axis C₁ that corresponds to the direction of the maximal elongation of the patch. As shown in Fig. 1(g), the distribution of the azimuthal rotational angle is peaked at specific angles deviating by approximately ±10–15° from each 〈111〉 surface lattice vector, which is schematically displayed in Fig. 1(h).

To investigate the electronic properties of isolated YPc₂ on Cu(111), we perform point dI/dV spectroscopy and dI/dV maps. For the molecule shown in Fig. 2(a), both dI/dV spectra at the molecular lobe and center sites show two clear peaks, one below and the other above the Fermi level, indicating the HOMO and LUMO, respectively (Fig. 2(b)). As marked by the vertical lines, the HOMO peak appears at the same energy (−0.7 eV) at both the lobe and center, while the LUMO peak at the molecule center (1.65 eV) is slightly shifted towards higher energy compared to that at the lobe (1.55 eV). As a consequence, the HOMO–LUMO gap at the center site (2.35 eV) is slightly larger than that at the lobe (2.25 eV). This feature is observed in the self-assembled YPc₂ molecular patch with a slight variation in peak position depending on the sites, as shown in Fig. S1.† The gap difference between the lobe and center suggests that the largest contribution to the conductance is due to the electronic states at the Pc ligands, with a smaller contribution from the center yttrium ion.³⁸ The electronic structure of an isolated YPc₂ on Cu(111) is quite different from that of YPc₂ and TbPc₂ on Au(111), which show 4 peaks and a much narrower gap.^22,39 This result indicates that YPc₂ on Cu(111) has different electronic properties compared to YPc₂ on Au(111), which was shown to preserve the unpaired spin in the molecular ligands.²² In addition, no Kondo resonance was detected at the Fermi level for YPc₂ on Cu(111) down to 10 K, which may either indicate the quenching of the unpaired spin or a Kondo screening occurring at lower temperatures.

The differential conductance map of isolated YPc₂ on Cu(111) at the HOMO and LUMO energy is shown in Fig. 2(c) and (d). Both maps of HOMO and LUMO are 4-fold symmetric, with their more pronounced minima rotated by 45°. The dI/dV maps disclose a high density of states at the Pc ring (lobe) site, confirming the smaller contribution of the center ion. Similar patterns and a large size gap of over 2 eV in electronic structures have been observed in several vanadyl-based metal–organic complexes with no spin radical on the ligand, such as VOPc on TiOPc/Ag(100)¹⁷ and Au(111),⁴⁰ and calculated for VOTTDPz on Au(111).⁴¹

As discussed above, the different electronic structures between YPc₂ on Cu(111) and Au(111), together with the absence of the Kondo feature on the Cu(111) point towards the quenching of the unpaired spin, possibly due to the proximity of the more reactive Cu(111) surface. To validate this conclusion, we performed dI/dV spectroscopy on a YPc₂ molecule adsorbed on a patch of YPc₂ on Cu(111), see Fig. 3(a). Unlike the YPc₂ directly adsorbed on Cu(111), the dI/dV spectrum of the YPc₂ on top of a YPc₂ patch (Fig. 3(b)) shows four distinct peaks, whose maxima are marked with vertical lines at: (1) −1.4 eV, (2) −0.75 eV, (3) 0.50 eV, (4) and 1.70 eV. In addition, the electronic gap of YPc₂ on YPc₂/Cu(111), i.e., the difference between (2) and (3), is 1.25 eV, which is remarkably smaller than that of YPc₂ on Cu(111). The number of peaks and narrower gap are reminiscent of the electronic configuration on Au(111), for which the spin is not quenched. Different from YPc₂/Au(111), however, we do not observe any Kondo features on YPc₂ on YPc₂/Cu(111). For the latter, this might be due to a Kondo temperature that is much lower than the experimental temperature, possibly due to the presence of the 1^st YPc₂ layer that screens the interaction with the Cu(111) conduction electrons.

3.2. YPc₂/ZnPc heterostructures on Cu(111)

The results of the previous section suggest that YPc₂ on Cu(111) acts as a decoupling layer for the molecules adsorbed on top of it. Nevertheless, there are some shortcomings of using YPc₂ as the decoupling layer. Firstly, in STM measurements, we notice that the YPc₂ molecules on top of YPc₂/Cu(111) exhibit high mobility and are easily displaced by the tip, complicating the acquisition of stable and precise data. Secondly, in view of characterizing these molecules using ensemble-averaging measurements, it is generally more convenient to introduce a decoupling layer made of a different molecular species,^17,42 to avoid averaging over the decoupling layer containing the same metal ion but with potentially different magnetic behaviors. To circumvent these issues, we chose to use ZnPc as the diamagnetic buffer layer to decouple YPc₂ from the Cu(111) surface. Due to its planar structure, ZnPc is known to generate flat adsorption and self-assemble on the (111) metal substrates,⁴³ providing a suitable template for flat adsorption of YPc₂. Before addressing the properties of YPc₂ on ZnPc/Cu(111), we first characterize the adsorption and electronic properties of 1 ML of ZnPc on Cu(111). Fig. 4(a) shows the self-assembled layer of ZnPc with two types of self-assembly domains labeled as A and B. Black molecular schematics of ZnPc show the two arrangements of molecules in the layer, where the molecules in one domain are rotated by about 60° with respect to the molecules in the other domain. These different alignments reflect the symmetry of the Cu(111) substrate. A ZnPc molecule consists of a combination of two bright lobes marked with a blue line and two dark lobes marked with a green line forming a cross. The dI/dV spectrum of ZnPc (Fig. 4(e)) presents three broad features marked with the vertical lines, two below and one above the Fermi level. This electronic structure is similar to ZnPc on Ag(111);⁴³ however, what we observe are much broader features and a wider gap of 1.20 eV. This suggests that a stronger hybridization is present between ZnPc and Cu(111) compared to Ag(111). The LUMO is partially occupied near the Fermi level, indicating fractional charge transfer from Cu(111) and a slight metal character in conductance. The hybridization with the substrate is reduced by increasing the number of molecular layers, as shown in Fig. S2.†

To verify whether ZnPc can act as an adequate decoupling layer on Cu(111), we start investigating the structure of YPc₂ when deposited on top of 1 ML of ZnPc on Cu(111). To this extent, we first focus on determining the possible intermixing of the two molecules in the heterostructure, which is a common issue in ligands with very close structures. Fig. 4(b) shows the most common type of YPc₂ that is observed after deposition on the single layer of ZnPc. One can see both the self-assembled monolayer of ZnPc as well as YPc₂ molecules. The latter can be distinguished by their apparent height, encoded as a brighter color in our STM image. The YPc₂ molecules that are well separated from each other show an internal lobe structure, while molecules that are found in close proximity do not, indicating that the YPc₂–YPc₂ interaction affects their own electronic structure.

The apparent height of the YPc₂/ZnPc heterostructure, however, differs from that of the isolated YPc₂ on Cu(111). A comparison of the line profiles shown in Fig. 4(c) indicates a prominence of 1.5 Å with respect to the ZnPc base layer, whose height is 2.2 Å from the Cu(111) surface. Conversely, the height of YPc₂ on Cu(111) is 3.7 Å, which closely corresponds to the sum of the two previous height values (Fig. 4(c)). This result strongly points towards the possibility that YPc₂ molecules are physically embedded in the ZnPc layer rather than adsorbed on top of it, as sketched in Fig. 4(d). To further verify this conjecture, we measured the dI/dV spectrum of the YPc₂/ZnPc heterostructure (Fig. 4(e)). Similar to YPc₂ on Cu(111), the dI/dV spectrum shows two prominent HOMO–LUMO peaks, with the related electronic gap depending on the molecule type, which is found to be 2.2 eV for the YPc₂ molecules without an inner lobe (close to one another) structure and 2.7 eV for the molecule with an inner lobe structure (surrounded by ZnPc). While the former shows a HOMO–LUMO gap very similar to that of the YPc₂ on Cu(111), for the latter, the gap is larger, possibly due to the interaction with the surrounding ZnPc and/or modification of the dielectric environment. The similarity of the dI/dV spectra with those of YPc₂ on Cu(111), together with the comparison between the apparent heights in Fig. 4(c), allows us to conclude that YPc₂ is embedded in the ZnPc layer. The cause of this phenomenon can be ascribed to the similarity between the YPc₂ and ZnPc structures, which can facilitate the intermixing between the two molecular layers at a finite temperature, and to the larger adsorption energy of YPc₂ that favors the anchoring of this molecule to the Cu(111) surface, as shown by DFT in section 3.3.

The visibility of the inner molecular structures of YPc₂ is evidence that the top Pc ligands of YPc₂ are sticking out from the ZnPc decoupling layer into which the YPc₂ molecules are embedded. When two YPc₂ molecules are close to each other, the overlap of the molecular lobes of the top Pc modifies the distribution of the molecular orbitals, altering the appearance in STM images. This effect is similar to that found for the phase I and phase II YPc₂ on Cu(111) (Fig. 1(e) and (f)), for which the inner structure is only imaged properly when intermolecular interactions are weak (phase II in Fig. S1†).

To prevent molecular intermixing, we realized another YPc₂/ZnPc heterostructure with a thicker decoupling ZnPc layer. For this system, due to the absence of regions of exposed Cu(111), the thickness of the ZnPc layer could not be determined by STM images; therefore, we estimated it based on the deposition time and temperature using our previous experiment as a calibration, which gives an average thickness of 2.5 MLs. In this situation, we expect the coexistence of regions with 2 and 3 MLs of ZnPc, denoted in the following as 2–3 MLs. The STM image of YPc₂ deposited on self-assembled 2–3 MLs of ZnPc is shown in Fig. 5(a). The thicker decoupling ZnPc layer also grows flat on Cu(111), with a similar pattern to the 1 ML ZnPc case. On top of it, YPc₂ grows either along the dislocation lines of the ZnPc layer, forming linear molecular chains, or as isolated molecules. Similar to the previous observations, the isolated molecule displays an internal structure while the molecular chain does not. In addition, isolated molecules show a more pronounced two-fold symmetry, suggesting that the original four-fold symmetry is lowered due to the interaction with the underneath ZnPc (Fig. 5(b)). On this thicker ZnPc layer, however, the apparent height of YPc₂ is identical to that observed for the same molecule on bare Cu(111) (Fig. 5(c)), suggesting adsorption on top rather than embedding into the ZnPc layer.

3.3. Density functional theory

Using DFT calculations, we rationalize the change of charge and spin states of YPc₂ on Cu(111) and ZnPc/Cu(111). We first estimate the energy level alignment of the system by placing the respective ionization potential (IP) and electron affinity (EA) levels of ZnPc and YPc₂ relative to the work function of the Cu(111) substrate (see Fig. 6). Our calculations indicate that ZnPc in a vacuum is non-magnetic (Fig. 7) and its relatively large band gap of 2.5 eV and its IP = 5.073 eV place its electronic states well below the Fermi edge of the substrate (4.9 eV), indicating that ZnPc/Cu(111) will most likely retain the non-magnetic neutral charge state of ZnPc in a vacuum. We note that the deposition of a single layer of ZnPc on Cu(111) leads to a reduction in the work function of about 0.2 eV, which is in line with the general trend observed with ZnPc on other metallic surfaces.⁴⁴ Since the ZnPc molecule remains almost perfectly planar, the reduction in the work function is not a consequence of the molecular dipole. YPc₂ has a gap of 1.3 eV in a vacuum and its IP = 4.875 eV places it closer to the Fermi edge of the substrate, indicating that YPc₂/Cu(111) is most likely subject to charge fluctuations and spin quenching.

Our gas-phase calculations of YPc₂ indicate that it has an electronic configuration of Y: [Kr] 5s⁰ 4d⁰ Pc²¹; i.e., a non-magnetic trivalent Y³⁺ and a single unpaired electron delocalized over the Pc₂ ligand, which is split in a SOMO/SUMO state (see Fig. 7(b)). To understand the loss of the unpaired spin when deposited directly onto Cu(111) it is instructive to compare the respective projected density of states (PDOS) plots in Fig. 7. In a vacuum, the ligand is fully polarized. When deposited on Cu(111) the ligand loses all polarization and becomes strongly broadened, without significant charge transfer between Cu and YPc₂, as shown in Fig. S4.† These pieces of evidence indicate hybridization with the metal substrate as the main mechanism for the loss of spin polarization. Furthermore, this broadening is stronger for the lower Pc ring, i.e., the ring in proximity to the Cu(111) surface. When deposited on a free-standing layer of ZnPc, YPc₂ recovers its delocalized unpaired electron. This can be seen as the limit of a sufficiently thick ZnPc decoupling layer where YPc₂ is decoupled from Cu(111). The strong hybridization and resulting broadening can be an indication for the suppression of Kondo; however, more detailed calculations and experiments are required to assess with certainty the presence or absence of Kondo in this system.

The intermixing of YPc₂ in a single layer of ZnPc seems to be driven by more than just adsorption energy differences as DFT predicts a slightly larger adsorption energy for a single YPc₂ molecule (−4.1 eV) when compared to a single ZnPc molecule (−3.5 eV), calculated at low coverage, and might be driven by the formation of a densely packed layer (Table 1).

---

4. Conclusions

In this study, we identified the conditions to preserve the unpaired electron in YPc2 when adsorbed on a surface and built a well-behaved molecular spin architecture. We selected ZnPc as the decoupling layer due to its single-Pc-based planar adsorption and diamagnetic properties. As only a few MLs of ZnPc are sufficient to decouple YPc2 molecules from the metal substrate, this system is suitable to be addressed by STM related magnetic characterization studies such as ESR-STM.³³ In addition, the use of Cu(111) as the substrate opens up the possibility to investigate YPc2 with a recently developed surface-sensitive ensemble ESR,³¹ for which molecular films are typically grown on top of single crystal Cu(111)/Al₂O₃ microstrip resonators. Magnetically diluting YPc2 into suitable molecular matrices may potentially allow characterizing the spin dynamics of their spins down to the limit of ultra-thin molecular films using pulsed ESR.

Finally, we identified the hybridization between YPc2 and the metal substrate as the mechanism responsible for spin quenching. Further investigation of the spin properties of YPc2 on other metals such as Ag and Au will allow understanding this mechanism in more depth and design robust molecular spin architectures. Finally, diamagnetic spacers made of suitable molecular structures can also enable adjusting the distance of molecular spins,¹⁷ opening a way to the optimization not only of the molecule–substrate interaction but also of the intermolecular spin–spin coupling.

Author contributions

The experiment was conceived by FD and FHC. SO, WS, LC, and FD performed the STM characterization. The STM data were analysed by SO under the supervision of WS and FD. FHC, JY, and HB supported the early stage of the experiments. The home-made STM setup was designed and realized by WJ, SP and HTB. LS and CH synthesized the YPc₂ molecules. The manuscript was written by SO, CW, and FD and finalized through the contributions of all authors. All authors have given approval to the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data for this article, including STM images and dI/dV spectra, are available at Zenodo at https://doi.org/10.5281/zenodo.16413730.

Acknowledgements

This work was supported by the Institute for Basic Science (Grant No. IBS-R027-D1). The authors are grateful to Andreas J. Heinrich for valuable discussions and Seorhin Choi for experimental support.

References

1. A. Gaita-Arino, F. Luis, S. Hill and E. Coronado, Nat. Chem., 2019, 11, 301–309 CrossRef CAS PubMed .
2. J. Tejada, E. M. Chudnovsky, E. D. Barco, J. M. Hernandez and T. P. Spiller, Nanotechnology, 2001, 12, 181 CrossRef CAS .
3. F. Troiani, V. Bellini, A. Candini, G. Lorusso and M. Affronte, Nanotechnology, 2010, 21, 274009 CrossRef CAS PubMed .
4. K. Bader, M. Winkler and J. van Slageren, Chem. Commun., 2016, 52, 3623–3626 RSC .
5. A. Ardavan, O. Rival, J. J. Morton, S. J. Blundell, A. M. Tyryshkin, G. A. Timco and R. E. Winpenny, Phys. Rev. Lett., 2007, 98, 057201 Search PubMed .
6. K. Bader, D. Dengler, S. Lenz, B. Endeward, S. D. Jiang, P. Neugebauer and J. van Slageren, Nat. Commun., 2014, 5, 5304 Search PubMed .
7. J. M. Zadrozny, J. Niklas, O. G. Poluektov and D. E. Freedman, ACS Cent. Sci., 2015, 1, 488–492 CrossRef CAS PubMed .
8. Y. Zhang, P. Guan, H. Isshiki, M. Chen, M. Yamashita and T. Komeda, Nano Res., 2010, 3, 604–611 CrossRef CAS .
9. P. Willke, T. Bilgeri, X. Zhang, Y. Wang, C. Wolf, H. Aubin, A. Heinrich and T. Choi, ACS Nano, 2021, 15, 17959–17965 CrossRef CAS PubMed .
10. R. Otero, J. M. Gallego, A. L. de Parga, N. Martin and R. Miranda, Adv. Mater., 2011, 23, 5148–5176 CrossRef CAS PubMed .
11. Z. Li, Y. Li and C. Yin, Polymers, 2023, 15, 4176 CrossRef CAS PubMed .
12. E. Moreno-Pineda and W. Wernsdorfer, Photonic Quantum Technologies: Science and Applications, 2023, vol. 1, pp. 269–304 Search PubMed .
13. L. C. de Camargo, M. Briganti, F. S. Santana, D. Stinghen, R. R. Ribeiro, G. G. Nunes, J. F. Soares, E. Salvadori, M. Chiesa and S. Benci, Angew. Chem., 2021, 133, 2620–2625 CrossRef .
14. M. Briganti, G. Serrano, L. Poggini, A. L. Sorrentino, B. Cortigiani, L. C. de Camargo, J. F. Soares, A. Motta, A. Caneschi, M. Mannini, F. Totti and R. Sessoli, Nano Lett., 2022, 22, 8626–8632 CrossRef CAS PubMed .
15. M. Hollerer, D. Lüftner, P. Hurdax, T. Ules, S. Soubatch, F. S. Tautz, G. Koller, P. Puschnig, M. Sterrer and M. G. Ramsey, ACS Nano, 2017, 11, 6252–6260 CrossRef CAS PubMed .
16. I. Cimatti, L. Bondì, G. Serrano, L. Malavolti, B. Cortigiani, E. Velez-Fort, D. Betto, A. Ouerghi, N. B. Brookes, S. Loth, M. Mannini, F. Totti and R. Sessoli, Nanoscale Horiz., 2019, 4, 1202–1210 RSC .
17. K. Noh, L. Colazzo, C. Urdaniz, J. Lee, D. Krylov, P. Devi, A. Doll, A. J. Heinrich, C. Wolf, F. Donati and Y. Bae, Nanoscale Horiz., 2023, 8, 624–631 RSC .
18. D. Komijani, A. Ghirri, C. Bonizzoni, S. Klyatskaya, E. Moreno-Pineda, M. Ruben, A. Soncini, M. Affronte and S. Hill, Phys. Rev. Mater., 2018, 2, 024405 CrossRef CAS .
19. T. Frauhammer, H. Chen, T. Balashov, G. Derenbach, S. Klyatskaya, E. Moreno-Pineda, M. Ruben and W. Wulfhekel, Phys. Rev. Lett., 2021, 127, 123201 CrossRef CAS PubMed .
20. Y. Zhang, P. Liao, J. Kan, C. Yin, N. Li, J. Liu, Q. Chen, Y. Wang, W. Chen, G. Q. Xu, J. Jiang, R. Berndt and K. Wu, Phys. Chem. Chem. Phys., 2015, 17, 27019–27026 RSC .
21. E. Moreno-Pineda, C. Godfrin, F. Balestro, W. Wernsdorfer and M. Ruben, Chem. Soc. Rev., 2018, 47, 501–513 RSC .
22. R. Barhoumi, A. Amokrane, S. Klyatskaya, M. Boero, M. Ruben and J.-P. Bucher, Nanoscale, 2019, 11, 21167–21179 RSC .
23. A. Candini, D. Klar, S. Marocchi, V. Corradini, R. Biagi, V. De Renzi, U. del Pennino, F. Troiani, V. Bellini, S. Klyatskaya, M. Ruben, K. Kummer, N. B. Brookes, H. Huang, A. Soncini, H. Wende and M. Affronte, Sci. Rep., 2016, 6, 21740 CrossRef CAS PubMed .
24. F. Branzoli, M. Filibian, P. Carretta, S. Klyatskaya and M. Ruben, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 79, 220404 CrossRef .
25. F. Branzoli, P. Carretta, M. Filibian, M. J. Graf, S. Klyatskaya, M. Ruben, F. Coneri and P. Dhakal, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 134401 CrossRef .
26. M. Studniarek, C. Wäckerlin, A. Singha, R. Baltic, K. Diller, F. Donati, S. Rusponi, H. Brune, Y. Lan, S. Klyatskaya, M. Ruben, A. Seitsonen and J. Dreiser, Adv. Sci., 2019, 6, 1901736 CrossRef CAS PubMed .
27. Z. Deng, S. Rauschenbach, S. Stepanow, S. Klyatskaya, M. Ruben and K. Kern, Phys. Scr., 2015, 90, 098003 CrossRef .
28. J. Siewert, R. Fazio, G. M. Palma and E. Sciacca, J. Low Temp. Phys., 2000, 118, 795–804 CrossRef CAS .
29. F. Branzoli, P. Carretta, M. Filibian, S. Klyatskaya and M. Ruben, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 83, 174419 CrossRef .
30. A. K. Boudalis, J.-E. Olivares-Peña, E. Moreno-Pineda, A. Fediai, W. Wenzel, P. Turek and M. Ruben, Chem. Commun., 2021, 57, 11505–11508 RSC .
31. F. H. Cho, J. Park, S. Oh, J. Yu, Y. Jeong, L. Colazzo, L. Spree, C. Hommel, A. Ardavan, G. Boero and F. Donati, Rev. Sci. Instrum., 2024, 95, 063904 CrossRef CAS PubMed .
32. J. Rocker, D. Cornu, E. Kieseritzky, A. Seiler, O. Bondarchuk, W. Hansel-Ziegler, T. Risse and H. J. Freund, Rev. Sci. Instrum., 2014, 85, 083903 Search PubMed .
33. S. Baumann, W. Paul, T. Choi, C. P. Lutz, A. Ardavan and A. J. Heinrich, Science, 2015, 350, 417–420 CrossRef CAS PubMed .
34. L. Ruan, J. Tong, G. Qin, L. Zhou, X. Jiao and X. Zhang, Eur. J. Inorg. Chem., 2020, 2020, 2112–2117 CrossRef CAS .
35. A. Dal Corso, Comput. Mater. Sci., 2014, 95, 337–350 CrossRef CAS .
36. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 Search PubMed .
37. G. Makov and M. C. Payne, Phys. Rev. B: Condens. Matter Mater. Phys., 1995, 51, 4014–4022 Search PubMed .
38. M. Capsoni, A. Schiffrin, K. A. Cochrane, C. G. Wang, T. Roussy, A. Q. Shaw, W. Ji and S. A. Burke, J. Phys. Chem. C, 2017, 121, 23574–23581 CrossRef CAS .
39. A. Amokrane, S. Klyatskaya, M. Boero, M. Ruben and J.-P. Bucher, ACS Nano, 2017, 11, 10750–10760 CrossRef CAS PubMed .
40. D. E. Barlow and K. W. Hipps, J. Phys. Chem. B, 2000, 104, 5993–6000 CrossRef CAS .
41. J. Hou, Y. Wang, K. Eguchi, C. Nanjo, T. Takaoka, Y. Sainoo, R. Arafune, K. Awaga and T. Komeda, Commun. Chem., 2020, 3, 36 CrossRef CAS PubMed .
42. J. H. Lee, C. Urdaniz, S. Reale, K. J. Noh, D. Krylov, A. Doll, L. Colazzo, Y. J. Bae, C. Wolf and F. Donati, Phys. Rev. B, 2024, 109, 235427 CrossRef CAS .
43. A. Benhnia, S. Watanabe, R. Tuerhong, M. Nakaya, J. Onoe and J.-P. Bucher, Nanomaterials, 2021, 11, 1618 CrossRef CAS PubMed .
44. Y. L. Huang, E. Wruss, D. A. Egger, S. Kera, N. Ueno, W. A. Saidi, T. Bucko, A. T. S. Wee and E. Zojer, Molecules, 2014, 19, 2969–2992 CrossRef PubMed .
45. I. Horcas, R. Fernández, J. M. Gómez-Rodríguez, J. Colchero, J. Gómez-Herrero and A. M. Baró, Rev. Sci. Instrum., 2007, 78, 013705 CrossRef CAS PubMed .

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5nr02516h

This journal is © The Royal Society of Chemistry 2025