betaFeedback

WorksPhysics and Astronomy

Rationalizing charge transfer: a single-molecule case study across three distinct surfaces

Abstract

Organic molecules are versatile building blocks for the formation of complex low-dimensional nanostructures on surfaces. Recently, interest has grown in translating on-surface synthetic protocols from noble metal surfaces to insulating surfaces, which offer technological relevance, and to magnetic or superconducting surfaces, which offer new routes to correlated phases. It is important, therefore, to understand how these alternative surfaces may alter the properties of adsorbates through processes like charge transfer and screening. Evidence for charge transfer is often obtained experimentally via scanning tunneling microscopy (STM) and spectroscopy (STS), and theoretically via density functional theory (DFT). In this work, we systematically compare the properties of isolated macrocyclic emeraldine (Me6_6) molecules on three different surfaces: metallic Au(111), a semiconducting ferromagnetic monolayer of FeCl2_2, and a cleaved superconducting surface of bulk 2H-NbSe2_2. We show that STS and DFT alone are insufficient to quantify charge transfer for these cases. Instead, we complement these methods with Kelvin probe force microscopy (KPFM) measurements and model Hamiltonian calculations to demonstrate that on Au(111) and FeCl2_2, Me6_6 molecules donate a single electron to the substrate, whereas on NbSe2_2 two electrons are transferred from molecule to substrate. Our work provides insight into the interplay of charge transfer, screening, and molecule-substrate coupling, and develops experimental and theoretical methods for comparing the properties of adsorbates across different surfaces for future studies.

Molecules adsorbed on atomically well-defined metal surfaces provide versatile building blocks for the bottom-up fabrication of complex low-dimensional quantum materials with tailored structural and electronic properties. By exploiting a delicate interplay between intermolecular interactions, metal–organic coordination, and on-surface covalent chemistry, molecular precursors can be assembled into well-defined one-dimensional chains, two-dimensional networks, and extended molecular architectures with atomic-scale precision.[1, 2]

Recent developments in on-surface synthesis of organic and metal-organic materials have unlocked new classes of magnetic materials. [3, 4] The ability to stabilize previously unobtainable radical products opened the field of π\pi-magnetism[5, 6, 7, 8] with recent developments including the synthesis of spin-chains,[9, 10, 11] while developments in 2D metal-organic frameworks have also resulted in promising magnetic properties including ferromagnetism,[12, 13] and rich magnetic phases mediated by coupling with radical ligands.[14] Despite some recent progress,[15, 16, 17, 18, 19, 20, 21] the necessary catalytic contribution from metal surfaces in the formation of many of these products has limited this field predominantly to noble metal surfaces. The ability to translate these on-surface protocols to insulating surfaces would enhance their technological prospects, and the ability to combine these materials with magnetic or superconducting surfaces may lead to new correlated states.

Control of the properties of this new class of magnetic materials via their electron population is also highly desirable. In the context of organic and metal-organic materials on surfaces, one unavoidable factor affecting electron population is charge transfer. Depending on the energy level alignment between electronic states of adsorbates and surfaces, electron transfer may occur upon adsorption. For a single-molecule on a metal surface, this depends upon the molecule’s ionization potential (IP), electron affinity (EA), and the surface’s work function (Φ\Phi) as illustrated schematically in Figure 1e-Figure 1h. In particular, charge transfer between adsorbed molecules and the substrate, governed by their energy-level alignment and interfacial hybridization, can modify molecular orbital occupations, stabilize charged molecular states, and profoundly influence the electronic properties of the molecular assemblies,[22, 23] or induce open-shell character.[18, 24, 25, 26, 27, 28] Beyond the single-molecule level, substrate-induced charge redistribution and electronic screening also modify intermolecular interactions, influencing both the formation and the collective behavior of molecular assemblies and enabling properties absent in the isolated building blocks, such as localized magnetic moments and collective magnetic or conducting phases.[12, 15, 16, 17] Understanding and controlling the interplay between molecular structure, intermolecular electronic coupling, and molecule–substrate charge transfer is therefore essential for establishing surface-supported molecular architectures as a versatile platform for engineering low-dimensional quantum materials with tailored functionalities.

Image not extracted

Figure 1. Charge transfer from Me6_6 molecules to Au(111), FeCl2_2/Au(111), and NbSe2_2 substrates. a Chemical structure of Me6_6. b–d Illustration of charge transfer from Me6_6 to each of the three substrates Au(111), ML FeCl2_2/Au(111), and NbSe2_2 respectively. e Energy-level diagram for the frontier orbitals of gas-phase Me6_6. f–h Energy-level diagrams for the frontier orbitals of Me6_6 on each of the three substrates demonstrating single electron charge transfer from molecule to substrate on Au(111) and FeCl2_2, leading to the gas-phase HOMO splitting into SOMO and SUMO orbitals separated in energy by the Coulomb repulsion, UU, and two electron transfer from molecule to substrate on NbSe2_2. Here, Φ\Phi represents the respective substrate’s work function.

Here, we systematically investigate charge transfer by studying macrocyclic emeraldine (Me6\mathrm{Me}_6; see chemical structure in Figure 1a) molecules on three different surfaces - metallic Au(111)\mathrm{Au}(111), a ferromagnetic semiconducting monolayer (ML) of FeCl2/Au(111)\mathrm{FeCl}_2/\mathrm{Au}(111),[29, 30] and a cleaved bulk crystal of superconducting 2H\mathrm{2H}-NbSe2\mathrm{NbSe}_2 (herein referred to simply as NbSe2\mathrm{NbSe}_2). Via scanning tunneling microscopy (STM) and spectroscopy (STS), non-contact atomic force microscopy (ncAFM), and Kelvin probe force microscopy (KPFM) supported by density functional theory (DFT) and model Hamiltonian calculations, we show that Me6\mathrm{Me}_6 molecules donate one electron to the surface via charge transfer on Au(111)\mathrm{Au}(111) and FeCl2\mathrm{FeCl}_2, and two electrons via charge transfer on NbSe2\mathrm{NbSe}_2 due to its larger work function (Figure 1b-Figure 1d). This leads to half-filling of the gas-phase HOMO on Au(111)\mathrm{Au}(111) and FeCl2\mathrm{FeCl}_2 (Figure 1b,Figure 1e-Figure 1f), causing it to split into SOMO and SUMO orbitals separated in energy by the Coulomb repulsion, UU, and to complete depopulation of the gas-phase HOMO on NbSe2\mathrm{NbSe}_2. Our work demonstrates the ability to control the macrocycle’s spin and charge by energy-level alignment dictated by substrate work function, and provides experimental and theoretical tools for rationalizing and quantifying charge transfer on different surfaces.

Results and discussion

Sample preparation and structural properties

To compare the properties of Me6\mathrm{Me}_6 molecules on both Au(111)\mathrm{Au}(111) and FeCl2\mathrm{FeCl}_2, we prepared samples with 0.5 ML coverage of FeCl2/Au(111)\mathrm{FeCl}_2/\mathrm{Au}(111) by sublimating stoichiometric FeCl2\mathrm{FeCl}_2 powder onto a clean Au(111)\mathrm{Au}(111) crystal held at ∼150∘C\sim150^\circ\mathrm{C}.[29, 30] We then sublimated the Me6\mathrm{Me}_6 molecules at ∼320∘C\sim320^\circ\mathrm{C} onto the half-FeCl2\mathrm{FeCl}_2, half-Au(111)\mathrm{Au}(111) samples held at slightly below room temperature (RT). To prepare NbSe2\mathrm{NbSe}_2 samples, we cleaved the crystal under vacuum conditions and subsequently annealed it to ∼100∘C\sim100^\circ\mathrm{C} in UHV. We then deposited the Me6\mathrm{Me}_6 molecules using the same sublimation conditions as above with the NbSe2NbSe_2 substrate held at RT.

After preparation, the samples were characterized using STM. Typical STM overview images of the two samples are shown in Figure 2a,Figure 2b. In panel a, the left half of the image shows a ML FeCl2FeCl_2 terrace and the right half of the image shows two bare Au(111)Au(111) terraces. The molecules preferentially adsorbed on the Au(111)Au(111) regions, but at the slightly cool substrate temperature used for preparation some molecules can also be found on the FeCl2FeCl_2 terraces. For the most part, the molecules remain isolated from one another after adsorption (with some molecular clusters also visible on the FeCl2FeCl_2 terrace). Similarly in panel b, the molecules appear to adsorb and remain isolated from one another on the NbSe2NbSe_2 substrate. Herein, we focus exclusively on the properties of isolated molecules.

Image not extracted

Figure 2. Sample preparations and structure of Me6_6. a,b Constant-current metal-tip STM overview images showing Me6_6 molecules adsorbed on terraces of both FeCl2_2 and Au(111) (Vb=1.6V_{\mathrm{b}} = 1.6 V, It=3I_{\mathrm{t}} = 3 pA), and on NbSe2_2 (Vb=1V_{\mathrm{b}} = 1 V, It=20I_{\mathrm{t}} = 20 pA) respectively. c Constant-height CO-tip ncAFM image of an isolated Me6_6 molecule on Au(111). d Simulation of ncAFM image of Me6_6 based on DFT optimized structure of Me6_6/Au(111) using the ProbeParticle method. Panels c,d scalebars: 5 Å. e Side-view of DFT optimized structure of Me6_6/Au(111). f Experimental measurement of Me6_6 adsorption height on Au(111) based on Δf(Δz)\Delta f(\Delta z) spectra acquired on Me6_6 (blue) and bare Au(111) (yellow), respectively, at the sites shown in panel c. Dashed lines indicate cubic fits used to accurately determine Δz\Delta z corresponding to Δfmin⁡\Delta f_{\min}.

On the Au(111)Au(111) and FeCl2FeCl_2 sample, we prepared a CO-terminated tip and performed ncAFM measurements on Me6Me_6 molecules on both substrates. The image on Au(111)Au(111) is shown in Figure 2c. The molecule has a three-fold symmetric appearance with greatest intensity at three half-ring features and weaker intensity at the center of the molecule. Using DFT, we calculated the optimized structure of Me6Me_6 molecules on Au(111)Au(111). A side view of the optimized structure on Au(111)Au(111) is shown in Figure 2e. The molecules adopt a non-planar adsorption configuration with three of the six benzene rings of the molecule tilted out of plane making the molecule three-fold symmetric. Based on the optimized geometry on Au(111)Au(111), we used the ProbeParticle model to predict the appearance of the Me6Me_6 molecules in ncAFM images.[31] As shown in Figure 2d, the simulated image strongly resembles the experimental image. Further agreement between theory and experiment was obtained by experimentally measuring the molecule’s adsorption height via CO-tip Δf(Δz)\Delta f(\Delta z) spectroscopy - as shown in Figure 2f. By comparing the relative height of the Δf(Δz)\Delta f(\Delta z) minimum on Au(111)Au(111) and on the molecule, we estimate that the tilted molecular ring is 4.2 Å above the Au(111)Au(111) surface, closely matching the calculated value of 4.4 Å (panel e).

Similar DFT calculations and CO-tip measurements were performed for Me6Me_6 molecules on FeCl2FeCl_2. Here, qualitatively identical ncAFM images were obtained (see Figure S3) and a slightly larger adsorption height was measured - in agreement with DFT calculations (Figure S1). On NbSe2_2, we were unable to prepare a CO-tip but similar appearance (though lower quality) was observed in an ncAFM image acquired with a Se-passivated tip (see Figure S4). DFT also confirms a similar adsorption height for the molecule in this case (Figure S1). A comparison of experimental measurements with DFT calculations therefore suggests similar structural properties across the three surfaces with Me6_6 molecules physisorbed in each case.

Electronic properties of Me6_6

The DFT calculations of the molecule on each of the three substrates also provided a first hint about their electronic properties. By calculating the projected molecular density of states (DOS) of the molecules on each of the three substrates, we found that the gas-phase highest-occupied molecular orbital (HOMO) of Me6_6 appears to be aligned with the surface Fermi level for Au(111) and FeCl2_2, and slightly to the unoccupied side of the Fermi level for NbSe2_2 (Figure S5). These calculations therefore suggest the possibility of charge transfer from molecule to substrate.

To test for the possibility of charge transfer, we experimentally probed the electronic properties of the Me6_6 molecules on each substrate via dI/dVdI/dV STS. A typical dI/dVdI/dV spectrum for an isolated Me6_6 molecule on Au(111) is shown in Figure 3a. This curve features four distinct peaks - one large peak at ∼−0.6\sim-0.6 V, two shoulders at ∼±0.2\sim\pm0.2 V, and a narrow resonance at zero bias (shown in higher resolution in Figure S6). We examined the spatial distribution of these four features by acquiring dI/dVdI/dV maps at each bias of interest, as shown in the insets of Figure 3a. The feature at ∼−0.6\sim-0.6 V exhibited three homogeneous rectangular lobes across the Me6_6 molecule whereas the three features closest to the Fermi energy (-0.2, 0, and 0.2 V) all shared a similar appearance featuring six smaller circular lobes.

Image not extracted

Figure 3. Differential conductance STS measurements of isolated Me₆ molecules on Au(111), FeCl₂, and NbSe₂. a-c STS measurements acquired on isolated Me₆ molecules (blue curves) and substrate references (grey curves) on Au(111), NbSe₂, and FeCl₂, respectively. Setpoints: Vb=700V_\mathrm{b} = 700 mV, It=400I_\mathrm{t} = 400 pA (a), Vb=1.4V_\mathrm{b} = 1.4 V, It=100I_\mathrm{t} = 100 pA (b), Vb=−1.4V_\mathrm{b} = -1.4 V, It=340I_\mathrm{t} = 340 pA (c). Panel c, left inset: KPFM Δf(Vb)\Delta f(V_\mathrm{b}) spectrum simultaneously acquired with the same panel’s dI/dV\mathrm{d}I/\mathrm{d}V spectrum. All other insets: Constant-height metal-tip dI/dV\mathrm{d}I/\mathrm{d}V maps acquired on Me₆ molecules at the specified bias voltages on each surface. d-f Highest occupied molecular orbitals (HOMOs) of Me₆ corresponding to positive ion resonances (PIRs), calculated in the gas-phase using the optimized geometry of Me₆ on Au(111) (isosurface: 0.02 e A˚−30.02\ e\ \mathrm{\AA}^{-3}). The PIR-1 and PIR-2 are degenerate. g-h Simulated ss-tip dI/dV\mathrm{d}I/\mathrm{d}V maps of the first PIRs of gas-phase Me₆ with on-surface optimized geometry associated with the orbitals shown in d-f. All scalebars: 5 Å.

For Me6_6 molecules on NbSe2_2, dI/dVdI/dV spectra displayed two pronounced peaks, as shown in Figure 3b. The spatial distribution of these two features are shown in the dI/dVdI/dV maps in the insets of Figure 3b. The feature at ∼−0.8\sim-0.8 V displays three rectangular lobes and the feature at ∼0.5\sim0.5 V displays six smaller circular lobes - similar to the two distinct appearances observed on Au(111).

Finally, for Me6_6 molecules on FeCl2_2, dI/dVdI/dV spectra featured two pronounced peaks. The first peak was a tall narrow resonance at ∼ ⁣−1.2\sim\!-1.2 V, and the second a shorter peak at ∼ ⁣0.35\sim\!0.35 V as shown in Figure 3c. The spatial resolution of the feature at 0.35 V shown in the dI/dVdI/dV map in the right inset of panel c is reminiscent of the near-Fermi character of the features of Me6_6 molecules on Au(111), with six circular lobes visible. As shown in the left inset of this panel, the tall resonance at ∼ ⁣−1.2\sim\!-1.2 V is accompanied by a discontinuity in the simultaneously acquired KPFM parabola measuring Δf(Vb)\Delta f(V_{\mathrm{b}}). From this observation, [32, 33] combined with ring-like intensity in dI/dVdI/dV maps revealed in the vicinity of this feature (see Figure S7) [18, 34, 35, 36] and linear shifting of the dI/dVdI/dV peak with varying tip-sample distance (Figure S8), [16, 36] we conclude that this feature is not an intrinsic molecular state but instead a charging feature that arises due to the double-barrier tunnel junction (where the first barrier is the usual tip-sample vacuum barrier, and the second created by the insulating FeCl2_2 region which separates the Me6_6 molecule from the underlying metal surface - discussed in greater detail in Section S5).

To interpret the STS results of Figure 3, we calculated the expected appearance of different dI/dVdI/dV features of Me6_6 molecules. To achieve this, we used DFT to calculate the molecular orbitals of gas-phase Me6_6 (but with the optimized on-surface geometry), and used these to simulate dI/dVdI/dV maps associated with ion resonances. Note that we recently showed that direct comparison with DFT-calculated molecular orbitals can be potentially unreliable, [37] but in this case these maps are qualitatively identical to calculated Dyson orbitals [38] (see comparison in Section S6). Molecular orbitals corresponding to the first positive ion resonances (PIR; occupied states) of gas-phase Me6_6 are shown in Figure 3d-Figure 3f, with the PIR-1 and PIR-2 orbitals found to be degenerate. Based on simulated dI/dVdI/dV maps using orbitals, we find that the gas-phase first PIR should exhibit six circular lobes, whereas the PIR-1/-2 should exhibit three larger rectangular lobes, as shown in Figure 3g,Figure 3h.

The simulated gas-phase PIR strongly resembles the first negative ion resonance (NIR; unoccupied state) observed at positive bias experimentally on each of the three surfaces. This provides strong evidence of charge transfer from molecule to substrate in each case. On Au(111), the -0.20 V and 0.20 V maps in Figure 3a both exhibit six circular lobes of intensity which resemble the gas-phase PIR in panel h, and the -0.58 V map in panel a resembles the gas-phase PIR-1/-2 in panel g. The observation of features with similar spatial intensity above and below the Fermi level points to a molecular charge of +1 (single electron charge transfer from molecule to surface), leading to a splitting of the gas-phase HOMO into SOMO and SUMO states, separated in energy by Coulomb repulsion, UU (as illustrated in Figure 1f). This hypothesis also naturally explains the additional sharp peak observed at zero bias in Figure 3a with a similar spatial distribution observed in the corresponding dI/dVdI/dV map to the -0.20 V and 0.20 V maps. The half-filled HOMO leads to the molecular hosting an unpaired electron and hence a spin 12\frac{1}{2}. On the metallic Au(111) surface, this spin can be Kondo screened by the itinerant substrate electrons to produce the well-established signature of a zero-bias peak observed in Figure 3a.[39] Using a Fano function convolved with the Fermi-Dirac distribution to fit this zero bias peak (see higher resolution spectrum in Figure S6),[40] we estimate a Kondo temperature of TK=14.5T_{\mathrm{K}} = 14.5 K from its half-width at half-maximum.

Similarly, there is a strong resemblance between the simulated dI/dVdI/dV maps and the NbSe2_2 experimental data. The -0.77 V map in Figure 3b resembles the gas-phase PIR-1/-2 in panel g, and the 0.45 V map in panel b resembles the gas-phase PIR in panel h. The observation of the gas-phase PIR feature exclusively at positive bias in this case points to a larger Me6_6 molecular charge of +2 on NbSe2_2, with the gas-phase HOMO becoming completely empty as a result of two-electron charge transfer from molecule to substrate (as illustrated in Figure 1h).

For Me6_6 on FeCl2_2, the observed state at 0.35 V in Figure 3c has spatial distribution resembling the gas-phase PIR, pointing to charge transfer from molecule to substrate. Because of the charging peak, however, no state could be observed at negative bias which prevents confident attribution of the molecular charge state as either +1+1 or +2+2.

From comparison of simulated and experimental dI/dV\mathrm{d}I/\mathrm{d}V maps, we attribute the molecular charge state of Me6\mathrm{Me}_{6} on Au(111)\mathrm{Au}(111) as +1+1, and of Me6\mathrm{Me}_{6} on NbSe2\mathrm{NbSe}_{2} as +2+2 but the case of Me6\mathrm{Me}_{6} on FeCl2\mathrm{FeCl}_{2} remains ambiguous.

Substrate work function and molecular charge

The origin of the different molecular charge states of Me6\mathrm{Me}_{6} on the different substrates must be related to charge transfer between molecule and respective substrate. One of the most important substrate properties which governs the magnitude of this charge transfer is its work function, as illustrated in Figure 1f-Figure 1h. The work function of Au(111)\mathrm{Au}(111) and NbSe2\mathrm{NbSe}_{2} are known to be ∼5.25\sim5.25 [41, 42] and 5.95.9 eV, [43] respectively. The larger work function of NbSe2\mathrm{NbSe}_{2} than Au(111)\mathrm{Au}(111) is consistent with the hypothesis of greater electron transfer from molecule to substrate leading to a molecular charge of +2+2 on this surface (compare Figure 1f and Figure 1h).

To estimate the work function, Φ\Phi, of FeCl2\mathrm{FeCl}_{2} we performed further KPFM measurements. The contact potential difference, VCPDV_{\mathrm{CPD}}, obtained from the maximum of Δf(Vb)\Delta f(V_{\mathrm{b}}) parabolas acquired in KPFM measurements is a measure of the difference between tip and sample work functions (Φt\Phi_{\mathrm{t}} and Φs\Phi_{\mathrm{s}} respectively):

VCPD=Φs−Φte,(1)V_{\mathrm{CPD}} = \frac{\Phi_{\mathrm{s}} - \Phi_{\mathrm{t}}}{e}, \tag*{(1)}

where ee is the charge of an electron. By measuring VCPDV_{\mathrm{CPD}} on Au(111)\mathrm{Au}(111) and on FeCl2\mathrm{FeCl}_{2} using the same tip, we can therefore estimate ΦFeCl2\Phi_{\mathrm{FeCl}_{2}}. KPFM measurements of FeCl2\mathrm{FeCl}_{2} and Au(111)\mathrm{Au}(111) acquired using the same tip are shown in Figure 4a-Figure 4b. From these measurements, we find a value of VCPDV_{\mathrm{CPD}} on FeCl2\mathrm{FeCl}_{2} that is ∼100\sim100 mV larger than on Au(111)\mathrm{Au}(111). This implies ΦFeCl2=ΦAu(111)+0.1≈5.35\Phi_{\mathrm{FeCl}_{2}} = \Phi_{\mathrm{Au}(111)} + 0.1 \approx5.35 eV (although we found some evidence for variations in this value across the surface - see discussion in Section S8). Therefore the FeCl2\mathrm{FeCl}_{2} substrate has a work function in between those of Au(111)\mathrm{Au}(111) and NbSe2\mathrm{NbSe}_{2}.

Image not extracted

Figure 4. Electrostatic comparison of Au(111) and FeCl2. a Constant-current STM image showing FeCl2 and Au(111) (Vb=1.6V_b = 1.6 V, It=3I_t = 3 pA). b CO-tip KPFM measurements acquired on Au(111) and FeCl2 at the sites indicated by colored markers in a. White dashed lines indicate parabolic fits from which the maxima, VCPDV_{\mathrm{CPD}}, were extracted, indicated by dotted vertical lines, allowing the difference in work function (ΔΦ\Delta\Phi) to be determined. c Illustration of the image charge effect for Me6_6 on Au(111), and on FeCl2/Au(111) showing adsorption heights, hh, screening distances, dd, molecular charges, QMQ_M, image charges, QimageQ_{\mathrm{image}}, tip sample distance, z0+Δzz_0 + \Delta z, and the model parameter associated with charge distribution on the tip, α\alpha, for the two different scenarios. d-e Tip-height dependent CO-tip KPFM measurements acquired on isolated Me6_6 molecules on Au(111) and FeCl2. Dashed lines indicate parabolic fits from which the maxima, VCPDV_{\mathrm{CPD}}, were extracted. Progressively lighter colors represent progressively larger tip-sample distances. f Plot showing extracted Δbf\Delta b_f for Me6_6/Au(111) and Me6_6/FeCl2 using Eqs. (2) and (3) (more details in Section S7). Solid lines indicate fits to Eq. (3) from which α\alpha and z0z_0 were determined from the blue fit to the Me6_6/Au(111), and these parameters used to provide an estimate of molecular charge QMQ_M for Me6_6/FeCl2 from the purple and black fits.

To gain greater insight into the charge properties of Me6\mathrm{Me}_{6} molecules on FeCl2\mathrm{FeCl}_{2}, we also performed further KPFM measurements on the molecules on Au(111)\mathrm{Au}(111) and FeCl2\mathrm{FeCl}_{2} over a range of tip-sample distances, Δz\Delta z, using the same tip condition. These measurements revealed a downward trend of VCPDV_{\mathrm{CPD}} values for both Me6/Au(111)\mathrm{Me}_{6}/\mathrm{Au}(111) and Me6/FeCl2\mathrm{Me}_{6}/\mathrm{FeCl}_{2} with decreasing Δz\Delta z, with the trend slightly faster for the molecules on FeCl2\mathrm{FeCl}_{2} - see Figure S13. This downward trend is a strong indication of positive charge on the molecule on both substrates.[44, 45] Note that for Me6/NbSe2\mathrm{Me}_{6}/\mathrm{NbSe}_{2}, we were unable to perform KPFM measurements with the same tip conditions. KPFM measurements acquired with a different tip, however, also show this downward trend that supports the hypothesis of positive molecular charge (see Figure S4). It is tempting to attribute the more rapid decrease of VCPDV_{\mathrm{CPD}} for Me6/FeCl2\mathrm{Me}_{6}/\mathrm{FeCl}_{2} compared to Me6/Au(111)\mathrm{Me}_{6}/\mathrm{Au}(111) to a larger positive charge of the molecule on Me6/FeCl2\mathrm{Me}_{6}/\mathrm{FeCl}_{2}. It is necessary, however, to also consider the screening environments of both charges. Arguably, the charged Me6\mathrm{Me}_{6} molecules are more strongly screened in direct proximity to metallic Au(111)\mathrm{Au}(111) than when they have an insulating FeCl2\mathrm{FeCl}_{2} layer between them and the Au(111)\mathrm{Au}(111).

To quantitatively compare the Au(111)\mathrm{Au}(111) and FeCl2\mathrm{FeCl}_{2} KPFM results, we developed a model which accounts for both the molecular charge and difference in screening lengths. This model is discussed in detail in Section S7. In brief, we measured Δf(V,Δz)\Delta f(V,\Delta z) curves like those shown in Figure 4d,Figure 4e, on Me6/Au(111)\mathrm{Me}_{6}/\mathrm{Au}(111), Au(111)\mathrm{Au}(111), Me6/FeCl2\mathrm{Me}_{6}/\mathrm{FeCl}_{2}, and FeCl2\mathrm{FeCl}_{2} while maintaining the same tip conditions. From the parabolic Δf(V)\Delta f(V) curves at each Δz\Delta z value, we performed fits of the form:

Δf(V)=af(V−VCPD∞)2+bf(V−VCPD∞)+cf,(2)\Delta f(V) = a_f \left(V - V_{\mathrm{CPD}}^{\infty}\right)^2 + b_f \left(V - V_{\mathrm{CPD}}^{\infty}\right) + c_f, \tag*{(2)}

where afa_f, bfb_f, and cfc_f are fitting parameters and VCPD∞V_{\mathrm{CPD}}^{\infty} represents the asymptotic value of the VCPDV_{\mathrm{CPD}} as Δz→∞\Delta z \to\infty (further details in Section S7). The most important of these fitting parameters is the coefficient of the linear term, bfb_f, which can be related to physical quantities such as the molecular charge, QMQ_{\mathrm{M}} (in dimensionless units of elementary charge), and the effective image charge distance, 2d2d, via:

Δbf(Δz)=bf(Me6/subs.)−bf(subs.)=−αQM(1(Δz−z0)3−1(Δz−z0+2d)3)(3)\Delta b_f(\Delta z) = b_f(\mathrm{Me}_6/\mathrm{subs.}) - b_f(\mathrm{subs.}) = -\alpha Q_{\mathrm{M}}\left(\frac{1}{(\Delta z-z_0)^3} - \frac{1}{(\Delta z-z_0+2d)^3}\right) \tag*{(3)}

where α\alpha is a parameter related to the condition of the tip (see Section S7) and z0z_0 represents the tip-sample distance at Δz=0\Delta z = 0. Note that Δbf\Delta b_f, representing the difference between bfb_f fit to the measurements on Me6\mathrm{Me}_6 on a particular surface, bf(Me6/subs.)b_f(\mathrm{Me}_6/\mathrm{subs.}), and the reference measurement on that surface, bf(subs.)b_f(\mathrm{subs.}), was computed to isolate the influence of the Me6\mathrm{Me}_6 molecules themselves from contributions from the respective surfaces. On both surfaces, Δz=0\Delta z = 0 was defined by retracting the tip 250 pm from the minimum of the measured Δf(Δz)\Delta f(\Delta z) spectrum on an Me6\mathrm{Me}_6 molecule on that surface (Figure S13). This model is illustrated schematically in Figure 4c.

Given that we could confidently attribute the molecular charge of Me6/Au(111)\mathrm{Me}_6/\mathrm{Au}(111) via the aforementioned dI/dVdI/dV measurements, our approach was to first fit the data acquired on Au(111)\mathrm{Au}(111) to Eq. (3) assigning a molecular charge of QM=+1Q_{\mathrm{M}} = +1, and choosing dd based on the adsorption height of Me6/Au(111)\mathrm{Me}_6/\mathrm{Au}(111) (3.5 Å, Figure 2e) and the estimated leaking of metallic states into vacuum (1.1 Å): d=3.5−1.1=2.4d = 3.5 - 1.1 = 2.4 Å.[46, 47] By doing so, we could fit the unknown parameter related to the condition of the tip, α\alpha, and the tip-sample distance at Δz=0\Delta z = 0, z0z_0. The resultant fit is the blue curve in Figure 4f.

Because the same tip was used for the measurements on FeCl2\mathrm{FeCl}_2 starting at an equivalent tip-sample distance, we used the same values of α\alpha and z0z_0 extracted from the fitted Me6/Au(111)\mathrm{Me}_6/\mathrm{Au}(111) data to then fit the Me6/FeCl2\mathrm{Me}_6/\mathrm{FeCl}_2 data using Eq. (3) in order to estimate the molecular charge on this surface, QMQ_{\mathrm{M}}. To do this, we imposed physical upper and lower bounds on the value dd (see Section S7), and then performed two fits at fixed values of QM=+1Q_{\mathrm{M}} = +1 and QM=+2Q_{\mathrm{M}} = +2. The curve associated with QM=+1Q_{\mathrm{M}} = +1 is shown in purple in Figure 4f. From this fit, we extract a realistic value of d=7.2±0.4d = 7.2 \pm0.4 Å. By contrast, the fit with QM=+2Q_{\mathrm{M}} = +2 could not converge and dd became saturated at its lower-bound of 3.7 Å (corresponding to the Me6_6/FeCl2_2 adsorption height; Figure S1). Alternative methods in which we estimated the value of dd and left QMQ_{\mathrm{M}} as a fitting parameter also support the claim of a +1+1 charge of Me6_6 on this surface (see Figure S15). These KPFM results provide strong evidence that the charge of Me6_6/FeCl2_2 is +1+1, arising from a single electron charge transfer from molecule to substrate.

Modeling charge transfer

The DFT calculations in Figure S5 provided first theoretical hints about our interpretation of charge transfer. However, this approach has some limitations. Firstly, conventional density functional theory (DFT) calculations employing local or semilocal exchange–correlation functionals systematically underestimate the fundamental electronic gap of molecules and solids.[48, 49, 50, 51] Secondly, though the DOS peak corresponding to the HOMO appears to be almost entirely shifted to the unoccupied side of the Fermi energy on NbSe2_2, it remains somewhat pinned to the Fermi energy, making it very difficult to quantify the charge transfer. Additionally, DFT poorly captures the bandgap of FeCl2_2 (see Section S3) relative to experimental observations,[29, 30] possibly resulting in physically unrealistic charge transfer directly between Me6_6 and FeCl2_2 rather than with the underlying Au(111). Finally, though not unprecedented,[17, 23, 52, 53] the two electron charge transfer from molecule to substrate on NbSe2_2 is somewhat surprising given the modest work function difference of ∼0.65\sim0.65 eV between Au(111) and NbSe2_2 - a finding that merits further examination.

To rationalize the influence of the surface on adsorbed Me6_6 molecules and provide greater insight into the nature of the charge transfer, we employ a model Hamiltonian that incorporates both the static polarization of the molecular electronic structure induced by a conductive substrate, and the dynamic exchange of electrons between molecule and surface:

H^=H^mol+ΔH^im+H^surf+H^hyb,(4)\hat{H} = \hat{H}_{\mathrm{mol}} + \Delta\hat{H}_{\mathrm{im}} + \hat{H}_{\mathrm{surf}} + \hat{H}_{\mathrm{hyb}}, \tag*{(4)}

where H^mol\hat{H}_{\mathrm{mol}} is the many-electron Hamiltonian of the isolated, gas phase molecule obtained from quantum chemistry calculations, ΔH^im\Delta\hat{H}_{\mathrm{im}} describes the renormalization of the molecular electronic structure due to image-charge screening by a conductive surface, H^surf\hat{H}_{\mathrm{surf}} represents the electronic states of the surface, and H^hyb\hat{H}_{\mathrm{hyb}} accounts for electronic coupling between molecule and surface. See Methods for further details of each of these terms.

Within this framework, the first term describes the molecular states of Me6_6 and the second is responsible for the renormalization of these states in the presence of a conductive surface. As illustrated in Figure 5a, fluctuations in charge density between an Me6_6 molecule in proximity with a conductive surface induce mirror charges in the surface. The electrostatic interaction between the molecular charge and its induced image charge causes renormalization of its electronic spectrum: effectively lowering the energy required to add or remove an electron from the molecule and screening electron-electron interactions. [47, 54] Within this model, we set half the effective distance between a molecular charge and its image to be 2.4 Å, as was the case for the KPFM model of Me6_6/Au(111).

Image not extracted

Figure 5. Modelling the influence of a substrate on Me6_6. a Illustration of the model Hamiltonian. The image model causes renormalization of the frontier orbitals due to the image charge effect. The coupling between the frontier orbitals of Me6_6 to a conductive surface is approximated by a 1D metallic chain. VV represents molecule-surface hybridization, tt electron hopping within the 1D metallic chain, and Φ\Phi surface work function. At values of Φ\Phi which promote molecule to surface charge transfer, the charge of the molecule is screened by its image charge in the surface. b Molecular charge of Me6_6 as a function of ΔΦ\Delta\Phi in both the gas-phase and with image-charge renormalization included. ΔΦ=0\Delta\Phi= 0 is defined as a situation in which the Fermi energy is exactly halfway between the PIR and NIR. Vertical lines indicate estimated positions of the work function of each of the three surfaces.

The second two terms in the model Hamiltonian then describe a conductive surface and its coupling with a molecule. As shown schematically in Figure 5a, we simplify the conductive surface as a 1D metallic tight-binding chain (see Methods). [55] Using this full model Hamiltonian, we are able to quantitatively describe changes in Me6_6’s molecular charge as a function of change in substrate work function. As can be seen in the blue curve Figure 5b, a charge state of +1 is extremely stable for gas-phase Me6_6 molecules, requiring a change of work function of greater than 3 eV before a second electron can be removed. When image charge corrections are included, however, the charge +1 plateau becomes much shorter with only ∼\sim0.5 eV change in work function required to remove a second electron. This model therefore provides strong quantitative agreement with the experimental findings on Au(111) and NbSe2_2, where a change in work function of ∼\sim0.65 eV was enough to transition from a charge of +1 to a charge of +2. This theory shows the crucial role of image charge-induced electronic renormalization in governing charge transfer processes.

The small change in work function of only ∼0.1\sim0.1 eV between Au(111) and FeCl2_2 also supports the attribution of a charge state of +1+1 on both surfaces within our model Hamiltonian.

Although our model does not capture the reduction in the image charge effect likely to be caused by the insulating FeCl2_2 monolayer, as alluded to in the previous section via a larger dd, we suggest that such a reduction in the image charge effect is likely to push the model towards the gas phase curve - making a two electron charge transfer scenario even less likely for FeCl2_2.

Throughout this work we are talking about integer electron charge transfer and orbital occupation. This is appropriate in the limit of weak physisorption, which the optimized structures and measured adsorption heights (Sections S1-S2) suggest is the case for Me6_6 on all three surfaces. Within the model Hamiltonian, this is represented as a relatively small hopping term in H^hyb\hat{H}_{\mathrm{hyb}} that results in steep transitions between charge states as shown in Figure 5b. A larger molecule-substrate interaction would reduce the gradient of the transition between charge states and potentially allow for fractional occupation in the case of a strongly hybridized molecule-substrate systems.

The KPFM measurements and the model Hamiltonian approach both suggest a single electron charge transfer of Me6_6 to FeCl2_2/Au(111). As in the case of Me6_6/Au(111), this implies the presence of an unpaired electron in the molecule on FeCl2_2, and hence a local magnetic moment. On Au(111), we were able to directly observe a manifestation of this local magnetic moment via the presence of a Kondo peak. On FeCl2_2, however, we were unable to observe evidence of magnetism. While this is beyond the scope of this work, the interaction of this local moment with the ferromagnetic FeCl2_2 surface may be an interesting avenue for future investigation.

Conclusion

In this work, we have systematically compared the properties of a single molecule on a metal surface, an insulating ferromagnetic surface, and a superconducting surface, providing insight into the role of charge transfer and screening in governing molecular properties. This work provides experimental and theoretical tools for quantifying and rationalizing charge transfer, and comparing adsorbate properties across different surfaces. Namely, we established methodology for quantifying molecular charge states and demonstrate the crucial role of image-charge screening in determining electron transfer at molecule–surface interfaces employing Kelvin Probe force spectroscopy. We show that a relatively small change in substrate work function (∼0.65\sim0.65 eV) can induce a transition between distinct integer molecular charge states, providing a route toward engineering molecular electronic and magnetic properties through substrate selection. This work will be valuable as molecular systems are increasingly prepared on more diverse surfaces of both fundamental and technological interest.

Methods

Samples preparation

The Au(111) crystal was prepared by several cycles of sputtering with Ar+\mathrm{Ar}^{+} ions and subsequent annealing at ∼500 ∘C\sim500\,^\circ\mathrm{C} in ultra-high vacuum (UHV). The 2D FeCl2\mathrm{FeCl}_{2} was prepared using stoichiometric powder commercially obtained from Sigma-Aldrich with a purity of 98%. The powder was sublimated in UHV using a Knudsen cell at ∼430 ∘C\sim430\,^\circ\mathrm{C} for 90 seconds onto the Au(111) sample held at ∼150 ∘C\sim150\,^\circ\mathrm{C}, resulting in a coverage of approximately 0.5 ML. The Me6_6 molecules were then sublimated in UHV using a Knudsen cell at ∼320 ∘C\sim320\,^\circ\mathrm{C} for 60 seconds with the sample cooled slightly below room temperature (RT), resulting in a coverage of <0.1<0.1 ML.

The 2H2\mathrm{H}-NbSe2\mathrm{NbSe}_{2} crystal was cleaved in vacuum and then annealed in UHV to ∼100 ∘C\sim100\,^\circ\mathrm{C}.

The Me6\mathrm{Me}_{6} molecules were then sublimated using the same conditions as above with the NbSe2\mathrm{NbSe}_{2} crystal kept at RT during deposition.

Scanning probe microscopy measurements

The majority of scanning probe microscopy (SPM) measurements were performed in a SPECS-JT microscope at a measuring temperature of 4.4 K using a Kolibri sensor (f0∼1f_{0} \sim1 MHz, Q∼160kQ \sim160\mathrm{k}, k∼1080k \sim1080 kN/m, 50 pm amplitude modulation). NcAFM measurements were performed by first functionalizing the tip with a carbon-monoxide (CO) molecule unless otherwise specified. KPFM measurements in Figure 4b were performed with the same Kolibri sensor functionalized with CO using an oscillation amplitude of 200 pm. These measurements were taken at a height defined by 1.1 nm above the minimum of a Δf(Δz)\Delta f(\Delta z) measurement performed on FeCl2\mathrm{FeCl}_{2}. A lock-in method was used for both STS and dI/dVdI/dV maps using Vmod=5V_{\mathrm{mod}} = 5 mV, fmod=723f_{\mathrm{mod}} = 723 Hz.

The constant-height ncAFM image in Figure 2c, and KPFM measurements in Figure 4d-Figure 4f and Section S7 were performed in a Createc LT-STM at a measuring temperature of 4.4 K using a qPlus sensor (f0∼30f_{0} \sim30 kHz, Q∼80kQ \sim80\mathrm{k}, k∼1.8k \sim1.8 kN/m, 50 pm amplitude modulation). The ncAFM and KPFM measurements were performed with a CO-functionalized tip. For the KPFM measurements in Figure 4d-Figure 4f (and Section S7), an oscillation amplitude of 200 pm was used and the tip-sample distance corresponding to Δz=0\Delta z = 0 was defined by retracting the tip 250 pm above the minimum of a Δf(Δz)\Delta f(\Delta z) measurement performed on an Me6\mathrm{Me}_{6} molecule on the respective substrate (see Figure S13).

DFT and Quantum Chemistry Calculations

The geometries of all molecules were optimized in their respective ground states using density functional theory (DFT) as implemented in the FHI-AIMS software package. [56] The hybrid PBE0 functional[57] was employed for these calculations, with the Tkatchenko-Scheffler[58] method incorporated to account for van der Waals interactions. To construct the molecular Hamiltonians employed in the models, we generate the one- and two-body integrals using the ORCA quantum chemistry software, [59] based on the closed-shell DFT orbitals obtained with the PBE0 exchange-correlation functional. [57] The model of the molecular Hamiltonian coupled to a discrete chain is solved at DMRG level of theory with the iTensor package in Julia, which provides a tensor network-based implementation of DMRG suitable for lattice models.

Model Hamiltonian

Metal–Molecule Interface

To describe charge transfer at a metal–molecule interface, we employ an effective Hamiltonian that incorporates both the static polarization of the molecular electronic structure induced by the metal substrate and the dynamic exchange of electrons between the molecule and the metallic electrode. The total Hamiltonian is written as

H^=H^mol+ΔH^im+H^metal+H^hyb,(5)\hat{H} = \hat{H}_{\mathrm{mol}} + \Delta\hat{H}_{\mathrm{im}} + \hat{H}_{\mathrm{metal}} + \hat{H}_{\mathrm{hyb}}, \tag*{(5)}

where H^mol\hat{H}_{\mathrm{mol}} is the many-electron Hamiltonian of the isolated molecule obtained from quantum chemistry calculations, ΔH^im\Delta\hat{H}_{\mathrm{im}} describes the renormalization of the molecular electronic structure due to image-charge screening by the metal, H^metal\hat{H}_{\mathrm{metal}} represents the electronic states of the metallic electrode, and H^hyb\hat{H}_{\mathrm{hyb}} accounts for the electronic coupling between the molecule and the metal.

Within this framework, the first two terms determine the effective electronic structure of the adsorbed molecule, while the latter two govern charge exchange across the interface.

Image-Charge Renormalization

When a molecule is brought into proximity with a metallic surface, fluctuations in its charge density induce mirror charges in the metal. The electrostatic interaction between the molecule and the metal ular charge and its induced image charge lowers the energy required to add or remove an electron from the molecule, leading to a renormalization of its electronic spectrum. As a consequence, occupied molecular orbitals are shifted upward in energy, unoccupied orbitals are shifted downward, and the HOMO–LUMO gap is reduced relative to the gas-phase molecule. This effect is incorporated through the image-charge correction

ΔH^im=−12Vim(N^−N0)2,(6)\Delta\hat{H}_{\mathrm{im}} = -\frac{1}{2}V_{\mathrm{im}}(\hat{N}-N_{0})^{2}, \tag*{(6)}

where N^\hat{N} is the molecular particle-number operator, N0N_{0} is the electron number of the neutral molecule, and VimV_{\mathrm{im}} denotes the image-charge interaction strength, which depends on the molecule–surface separation and the dielectric response of the substrate. After normal ordering, this correction modifies both the one-electron and two-electron components of the molecular Hamiltonian, thereby accounting for the screening of electron–electron interactions by the metallic environment. We set Vim=1/(4∣z−z0∣)V_{\mathrm{im}} = 1/(4|z-z_{0}|) using the adsorption height of the molecule zz and the image plane position z0z_{0} to match the experimental findings on Au(111) such that ∣z−z0∣=2.4|z-z_{0}| = 2.4 Å.[47]

Metal Hamiltonian and Hybridization

Following the image-charge renormalization, the molecule is coupled to the metallic electrode to describe charge transfer across the interface. The metal is represented by a one-dimensional tight-binding chain,

H^metal=∑iεic^i†c^i−t∑⟨i,j⟩(c^i†c^j+c^j†c^i),(7)\hat{H}_{\mathrm{metal}} = \sum_{i} \varepsilon_{i}\hat{c}_{i}^{\dagger}\hat{c}_{i} - t\sum_{\langle i,j\rangle}\left(\hat{c}_{i}^{\dagger}\hat{c}_{j}+\hat{c}_{j}^{\dagger}\hat{c}_{i}\right), \tag*{(7)}

where c^i†\hat{c}_{i}^{\dagger} (c^i\hat{c}_{i}) creates (annihilates) an electron on site ii, εi\varepsilon_{i} is the on-site energy, and tt is the nearest-neighbor hopping parameter that determines the bandwidth of the metallic states. We used t=1t = 1 eV and varied εi\varepsilon_{i} to create effective changes in work function, ΔΦ\Delta\Phi, as illustrated in Figure 5.

The interaction between the molecule and the metal is described by the hybridization Hamiltonian,

H^hyb=∑m,iVmi(a^m†c^i+c^i†a^m),(8)\hat{H}_{\mathrm{hyb}} = \sum_{m,i} V_{mi} \left(\hat{a}_{m}^{\dagger}\hat{c}_{i} + \hat{c}_{i}^{\dagger}\hat{a}_{m}\right), \tag*{(8)}

where VmiV_{mi} denotes the coupling between molecular orbital mm and metal site ii. This term enables electron tunneling between the molecule and the electrode and is responsible for charge redistribution, level broadening, and partial occupation of molecular orbitals. We used Vmi=0.1V_{mi} = 0.1 eV.

The complete Hamiltonian therefore separates the two dominant physical processes governing metal–molecule interfaces. The image-charge term provides a static renormalization of the molecular electronic structure arising from polarization of the metallic substrate, whereas the hybridization term describes the dynamic exchange of electrons between the molecule and the metal. This decomposition allows the influence of electrostatic screening and electronic coupling on charge transfer to be analyzed independently while remaining within a unified many-body framework.

Vertical lines in Figure 5 indicate estimated positions of the work function of each of the three surfaces. Given the confident attribution of a charge state of +1 for Me6_6/Au(111) from dII/dVV measurements, we chose the Au(111) vertical line to be aligned with the center of the +1 charge plateau. Then, using the known and measured work function differences between Au(111), FeCl2_2, and NbSe2_2, we added the lines for these two other surfaces.

Supporting Information Available

Measured and calculated adsorption heights, ncAFM measurements on FeCl2_2 and NbSe2_2 surfaces, DFT calculations, supporting STS measurements of Kondo feature on Au(111) and charging signatures on FeCl2_2, simulated dII/dVV maps and Dyson orbitals, detailed KPFM model with supporting measurements, evidence for local variations in FeCl2_2 work function.

Acknowledgements

The authors acknowledge support from GAČR 25-17866X, GAČR 23-06781M, GAČR 23-05486S, and the CzechNanoLab Research Infrastructure, supported by MEYS CR (LM2023051).

We also acknowledge financial support from the TERAFIT project (CZ.02.01.01/00/22_008/0004594).

B.L. acknowledges support from GAČR 25-16632I and from the European Union’s Horizon Europe research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 101203634. Z.S. was supported by ERC-CZ program (project LL2101) from Ministry of Education Youth and Sports (MEYS) and by the project Advanced Functional Nanorobots (reg. No. CZ.02.1.01/0.0/0.0/15_003/0000444 financed by the EFRR). The authors would also like to thank Dr. Benjamin Mallada and Dr. Alejandro Jiménez Martín for experimental support and advice.

References

Quantum Dot by Noncontact Atomic-Force Microscopy. Phys. Rev. Lett. 2005, 94, 056802.

References

  1. [1]Barth, J. V.; Costantini, G.; Kern, K. Engineering atomic and molecular nanostructures at surfaces. Nature 2005, 437, 671–679.pubmed.ncbi.nlm.nih.gov/16193042
  2. [2]Bartels, L. Tailoring molecular layers at metal surfaces. Nature Chemistry 2010, 2, 87–95.DOI
  3. [3]Song, S.; Su, J.; Telychko, M.; Li, J.; Li, G.; Li, Y.; Su, C.; Wu, J.; Lu, J. On-surface synthesis of graphene nanostructures with π-magnetism. Chem. Soc. Rev. 2021, 50, 3238–3262.DOI
  4. [4]Hou, R.; Zhang, C.; Xu, L.; Ding, Y.; Xu, W. Construction of metal–organic nanostructures and their structural transformations on metal surfaces. Phys. Chem. Chem. Phys. 2025, 27, 8635–8655.DOI
  5. [5]Pavliček, N.; Mistry, A.; Majzik, Z.; Moll, N.; Meyer, G.; Fox, D. J.; Gross, L. Synthesis and characterization of triangulene. Nature Nanotech 2017, 12, 308–311.DOI
  6. [6]Mishra, S.; Beyer, D.; Eimre, K.; Kezilebieke, S.; Berger, R.; Gröning, O.; Pignedoli, C. A.; Müllen, K.; Liljeroth, P.; Ruffieux, P.; Feng, X.; Fasel, R. Topological frustration induces unconventional magnetism in a nanographene. Nat. Nanotechnol. 2020, 15, 22–28.DOI
  7. [7]Mishra, S.; Vilas-Varela, M.; Ronč ević, I.; Paschke, F.; Albrecht, F.; Gross, L.; Peña, D. Synthesis and Characterization of a π-Extended Clar’s Goblet. J. Am. Chem. Soc. 2025, 147, 39067–39071.DOI
  8. [8]Frezza, F.; Matěj, A.; Sánchez-Grande, A.; Carrera, M.; Mutombo, P.; Kumar, M.; Curiel, D.; Jelínek, P. On-Surface Synthesis of a Radical 2D Supramolecular Organic Framework. J. Am. Chem. Soc. 2024, 146, 3531–3538.DOI
  9. [9]Su, X.; Ding, Z.; Hong, Y.; Ke, N.; Yan, K.; Li, C.; Jiang, Y.-F.; Yu, P. Fabrication of spin-1/2 Heisenberg antiferromagnetic chains via combined on-surface synthesis and reduction for spinon detection. Nat. Synth 2025, 4, 694–701.DOI
  10. [10]Sun, K.; Cao, N.; Silveira, O. J.; Fumega, A. O.; Hanindita, F.; Ito, S.; Lado, J. L.; Liljeroth, P.; Foster, A. S.; Kawai, S. On-surface synthesis of Heisenberg spin-1/2 antiferromagnetic molecular chains. Science Advances 2025, 11, eads1641.DOI
  11. [11]Zhao, C.; Yang, L.; Henriques, J. C. G.; Ferri-Cortés, M.; Catarina, G.; Pignedoli, C. A.; Ma, J.; Feng, X.; Ruffieux, P.; Fernández-Rossier, J.; Fasel, R. Spin excitations in nanographene-based antiferromagnetic spin-1/2 Heisenberg chains. Nat. Mater. 2025, 24, 722–727.
  12. [12]Lobo-Checa, J.; Hernández-López, L.; Otrokov, M. M.; Piquero-Zulaica, I.; Candia, A. E.; Gargiani, P.; Serrate, D.; Delgado, F.; Valdivares, M.; Cerdá, J.; Arnau, A.; Bartolomé, F. Ferromagnetism on an atom-thick & extended 2D metal-organic coordination network. Nat Commun 2024, 15, 1858.
  13. [13]Isufi Neziri, E.; Hensky, C.; Le, H. Q.; Radillo Ochoa, D.; Cebrat, A.; Parschau, M.; Ernst, K.-H.; Wäckerlin, C. 2D Metal-organic Ferromagnets. Advanced Science 2025, 12, 2415266.
  14. [14]Frezza, F.; Kumar, M.; Sánchez-Grande, A.; Soler-Polo, D.; Carrera, M.; Stetsovych, O.; Mutombo, P.; Curiel, D.; Jelínek, P. On-Surface Synthesis of a Large-Scale 2D MOF with Competing π–d Ferromagnetic/Antiferromagnetic Order. J. Am. Chem. Soc. 2025, 147, 19575–19582.DOI
  15. [15]Kinikar, A.; Xiang, F.; Palomino-Ruiz, L.; Lu, L.-S.; Dong, C.; Gu, Y.; Darawish, R.; Ammerman, E.; Gröning, O.; Müllen, K.; Fasel, R.; Robinson, J. A.; Ruffieux, P.; Schuler, B.; Borin Barin, G. Atomic-Scale Imaging of Transferred Graphene Nanoribbons for Nanoelectronic Device Integration. ACS Appl. Nano Mater. 2025, 8, 16457–16464.DOI
  16. [16]Lowe, B.; Field, B.; Hellerstedt, J.; Ceddia, J.; Nourse, H. L.; Powell, B. J.; Med-hekar, N. V.; Schiffrin, A. Local gate control of Mott metal-insulator transition in a 2D metal-organic framework. Nat Commun 2024, 15, 3559.DOI
  17. [17]Domínguez-Celorrio, A.; Edens, L.; Sanz, S.; Vilas-Varela, M.; Martinez-Castro, J.; Peña, D.; Langlais, V.; Frederiksen, T.; Pascual, J. I.; Serrate, D. Systematic modulation of charge and spin in graphene nanoribbons on MgO. Nat Commun 2025, 16, 5632.
  18. [18]Pawlak, R.; Liu, J.-C.; Li, C.; Hess, R.; Chen, H.; Drechsel, C.; Zhou, P.; Liu, X.; Häner, R.; Aschauer, U.; Glatzel, T.; Decurtins, S.; Loss, D.; Klinovaja, J.; Liu, S.-X.; Wulfhekel, W.; Meyer, E. Toward Gate-Tunable Topological Superconductivity in a Supramolecular Electron Spin Lattice. Nano Lett. 2025, 25, 15206–15214.DOI
  19. [19]Yan, L.; Silveira, O. J.; Alldritt, B.; Kezilebieke, S.; Foster, A. S.; Liljeroth, P. Two-Dimensional Metal–Organic Framework on Superconducting NbSe₂. ACS Nano 2021, 15, 17813–17819.DOI
  20. [20]Edens, L.; Romero-Lara, F.; Sai, T.; Biswas, K.; Vilas-Varela, M.; Frederiksen, T.; Peña, D.; Schulz, F.; Pascual, J. I. Spin and Charge Control of Topological End States in Chiral Graphene Nanoribbons on a 2D Ferromagnet. Advanced Materials 2025, e10753.DOI
  21. [21]Vaxevani, K.; Ortuzar, J.; Trivini, S.; Monninger, G.; Wang, D.; Vilas-Varela, M.; Gómez-Rodrigo, L.; Peña, D.; Pascual, J. I. Parity-dependent coupling of molecular spin chains to a superconductor. Nature 2026, 658, 617–622.
  22. [22]Lu, X.; Grobis, M.; Khoo, K. H.; Louie, S. G.; Crommie, M. F. Charge transfer and screening in individual molecules on metal substrates: A scanning tunneling spectroscopy and theoretical study. Physical Review B 2004, 70, 115418.
  23. [23]Heimel, G.; Duhm, S.; Salzmann, I.; Gerlach, A.; Strozecka, A.; Niederhausen, J.; Bürker, C.; Hosokai, T.; Fernandez-Torrente, I.; Schulze, G.; Winkler, S.; Wilke, A.; Schlesinger, R.; Frisch, J.; Bröker, B.; Vollmer, A.; Detlefs, B.; Pflaum, J.; Kera, S.; Franke, K. J. et al. Charged and metallic molecular monolayers through surface-induced aromatic stabilization. Nature Chemistry 2013, 5, 187–194.DOI
  24. [24]Wang, T.; Berdonces-Layunta, A.; Friedrich, N.; Vilas-Varela, M.; Calupitan, J. P.; Pascual, J. I.; Peña, D.; Casanova, D.; Corso, M.; de Oteyza, D. G. Aza-Triangulene: On-Surface Synthesis and Electronic and Magnetic Properties. J. Am. Chem. Soc. 2022, 144, 4522–4529.DOI
  25. [25]Calupitan, J. P.; Berdonces-Layunta, A.; Aguilar-Galindo, F.; Vilas-Varela, M.; Peña, D.; Casanova, D.; Corso, M.; de Oteyza, D. G.; Wang, T. Emergence of π-Magnetism in Fused Aza-Triangulenes: Symmetry and Charge Transfer Effects. Nano Lett. 2023, 23, 9832–9840.DOI
  26. [26]Žonda, M.; Stetsovych, O.; Korytár, R.; Ternes, M.; Temirov, R.; Raccanelli, A.; Tautz, F. S.; Jelínek, P.; Novotný, T.; Švec, M. Resolving Ambiguity of the Kondo Temperature Determination in Mechanically Tunable Single-Molecule Kondo Systems. J. Phys. Chem. Lett. 2021, 12, 6320–6325.
  27. [27]Esat, T.; Borodin, D.; Oh, J.; Heinrich, A. J.; Tautz, F. S.; Bae, Y.; Temirov, R. A quantum sensor for atomic-scale electric and magnetic fields. Nat. Nanotechnol. 2024, 19, 1466–1471.DOI
  28. [28]Li, D.; Kumar, M.; Stetsovych, O.; Lowe, B.; Sengupta, R.; Hayashi, H.; Maeda, H.; Jelínek, P.; Kawai, S. Spin State in Au Porphyrins Modulated by Charge Transfer on Au(111). J. Am. Chem. Soc. 2026, 148, 13863–13870.DOI
  29. [29]Aguirre, A.; Pinar Solé, A.; Soler Polo, D.; González-Orellana, C.; Thakur, A.; Ortuzar, J.; Stesovych, O.; Kumar, M.; Peña-Díaz, M.; Weber, A.; Tallarida, M.; Dai, J.; Dreiser, J.; Muntwiler, M.; Rogero, C.; Pascual, J. I.; Jelínek, P.; Ilyn, M.; Corso, M. Ferromagnetic Order in 2D Layers of Transition Metal Dichlorides. Advanced Materials 2024, 36, 2402723.DOI
  30. [30]Candia, A. E.; Peláez-Sifonte, E.; Thakur, A. J.; Hadjadj, S. E.; Kerschbaumer, S.; Saunot, A.; Corso, M.; Ilyn, M.; Lobo-Checa, J.; Rogero, C.; Serrate, D. Atomic scale demonstration of ferromagnetism in a single layer FeCl2 on Au(111). 2026; 10.48550/arXiv.2605.22783.arxiv.org/abs/2605.22783
  31. [31]Krejčí, O.; Hapala, P.; Ondráček, M.; Jelínek, P. Principles and simulations of high-resolution STM imaging with a flexible tip apex. Phys. Rev. B 2017, 95, 045407.DOI
  32. [32]Gross, L.; Mohn, F.; Liljeroth, P.; Repp, J.; Giessibl, F. J.; Meyer, G. Measuring the Charge State of an Adatom with Noncontact Atomic Force Microscopy. Science 2009, 324, 1428–1431.DOI
  33. [33]Stomp, R.; Miyahara, Y.; Schaer, S.; Sun, Q.; Guo, H.; Grutter, P.; Studenikin, S.; Poole, P.; Sachrajda, A. Detection of Single-Electron Charging in an Individual InAs Quantum Dot by Noncontact Atomic-Force Microscopy. Phys. Rev. Lett. 2005, 94, 056802.pubmed.ncbi.nlm.nih.gov/15783674
  34. [34]Pörtner, M.; Wei, Y.; Riss, A.; Seufert, K.; Garnica, M.; Barth, J. V.; Seitsonen, A. P.; Diekhöner, L.; Auwärter, W. Charge State Control of F16CoPc on h-BN/Cu(111). Advanced Materials Interfaces 2020, 7, 2000080.DOI
  35. [35]Yan, L.; Silveira, O. J.; Alldritt, B.; Krejčí, O.; Foster, A. S.; Liljeroth, P. Synthesis and Local Probe Gating of a Monolayer Metal-Organic Framework. Advanced Functional Materials 2021, 31, 2100519.DOI
  36. [36]Kumar, D.; Krull, C.; Yin, Y.; Medhekar, N. V.; Schiffrin, A. Electric Field Control of Molecular Charge State in a Single-Component 2D Organic Nanoarray. ACS Nano 2019, 13, 11882–11890.DOI
  37. [37]Kumar, M.; Soler-Polo, D.; Lozano, M.; Monino, E.; Veis, L.; Jelinek, P. Multireference Theory of Scanning Tunneling Spectroscopy Beyond One-Electron Molecular Orbitals: Can We Image Molecular Orbitals? J. Am. Chem. Soc. 2025, 147, 24993–25003.
  38. [38]Ortiz, J. V. Dyson-orbital concepts for description of electrons in molecules. J. Chem. Phys. 2020, 153, 070902.DOI
  39. [39]Ternes, M.; Heinrich, A. J.; Schneider, W.-D. Spectroscopic manifestations of the Kondo effect on single adatoms. J. Phys.: Condens. Matter 2008, 21, 053001.DOI
  40. [40]Nagaoka, K.; Jamneala, T.; Grobis, M.; Crommie, M. F. Temperature Dependence of a Single Kondo Impurity. Phys. Rev. Lett. 2002, 88, 077205.pubmed.ncbi.nlm.nih.gov/11863936
  41. [41]Rusu, P. C.; Brocks, G. Surface Dipoles and Work Functions of Alkanethiolates and Fluorinated Alkanethiolates on Au(111). J. Phys. Chem. B 2006, 110, 22628–22634.pubmed.ncbi.nlm.nih.gov/17092010
  42. [42]Hansson, G. V.; Flodström, S. A. Photoemission study of the bulk and surface electronic structure of single crystals of gold. Phys. Rev. B 1978, 18, 1572–1585.DOI
  43. [43]Shimada, T.; Ohuchi, F. S.; Parkinson, B. A. Work Function and Photothreshold of Layered Metal Dichalcogenides. Jpn. J. Appl. Phys. 1994, 33, 2696.DOI
  44. [44]Mallada, B.; Gallardo, A.; Lamanec, M.; de la Torre, B.; Špirko, V.; Hobza, P.; Jelinek, P. Real-space imaging of anisotropic charge of σ-hole by means of Kelvin probe force microscopy. Science 2021, 374, 863–867.DOI
  45. [45]Gross, L.; Schuler, B.; Mohn, F.; Moll, N.; Pavliček, N.; Steurer, W.; Scivetti, I.; Kotsis, K.; Persson, M.; Meyer, G. Investigating atomic contrast in atomic force microscopy and Kelvin probe force microscopy on ionic systems using functionalized tips. Physical Review B 2014, 90, 155455.DOI
  46. [46]Lam, S. C.; Needs, R. J. First-principles calculations of the screening of electric fields at the aluminium (111) and (110) surfaces. J. Phys.: Condens. Matter 1993, 5, 2101.DOI
  47. [47]Neaton, J. B.; Hybertsen, M. S.; Louie, S. G. Renormalization of Molecular Electronic Levels at Metal-Molecule Interfaces. Phys. Rev. Lett. 2006, 97, 216405.DOI
  48. [48]Perdew, J. P.; Levy, M. Physical Content of the Exact Kohn-Sham Orbital Energies: Band Gaps and Derivative Discontinuities. Physical Review Letters 1983, 51, 1884–1887.DOI
  49. [49]Sham, L. J.; Schlüter, M. Density-Functional Theory of the Energy Gap. Physical Review Letters 1983, 51, 1888–1891.DOI
  50. [50]Hybertsen, M. S.; Louie, S. G. Electron correlation in semiconductors and insulators: Band gaps and quasiparticle energies. Physical Review B 1986, 34, 5390–5413.pubmed.ncbi.nlm.nih.gov/9940372
  51. [51]Onida, G.; Reining, L.; Rubio, A. Electronic excitations: density-functional versus many-body Green’s-function approaches. Reviews of Modern Physics 2002, 74, 601–659.DOI
  52. [52]Biswas, K.; Urgel, J. I.; Xu, K.; Ma, J.; Sánchez-Grande, A.; Mutombo, P.; Gallardo, A.; Lauwaet, K.; Mallada, B.; de la Torre, B.; Matěj, A.; Gallego, J. M.; Miranda, R.; Jelínek, P.; Feng, X.; Écija, D. On-Surface Synthesis of a Dicationic Diazabenzocoronene Derivative on the Au(111) Surface. Angewandte Chemie International Edition 2021, 60, 25551–25556.DOI
  53. [53]Jiang, S.; Neuman, T.; Bretel, R.; Boeglin, A.; Scheurer, F.; Le Moal, E.; Schull, G. Many-Body Description of STM-Induced Fluorescence of Charged Molecules. Phys. Rev. Lett. 2023, 130, 126202.DOI
  54. [54]Thygesen, K. S.; Rubio, A. Renormalization of Molecular Quasiparticle Levels at Metal-Molecule Interfaces: Trends across Binding Regimes. Phys. Rev. Lett. 2009, 102, 046802.DOI
  55. [55]Zuzak, R.; Kumar, M.; Stoica, O.; Soler-Polo, D.; Brabec, J.; Pernal, K.; Veis, L.; Blieck, R.; Echavarren, A. M.; Jelinek, P.; Godlewski, S. On-Surface Synthesis and Determination of the Open-Shell Singlet Ground State of Tridecacene**. Angewandte Chemie International Edition 2024, 63, e202317091.
  56. [56]Blum, V.; Gehrke, R.; Hanke, F.; Havü, P.; Havü, V.; Ren, X.; Reuter, K.; Scheffler, M. Ab initio molecular simulations with numeric atom-centered orbitals. Computer Physics Communications 2009, 180, 2175–2196.DOI
  57. [57]Adamo, C.; Barone, V. Toward reliable density functional methods without adjustable parameters: The PBE0 model. J. Chem. Phys. 1999, 110, 6158–6170.DOI
  58. [58]Tkatchenko, A.; Scheffler, M. Accurate Molecular Van Der Waals Interactions from Ground-State Electron Density and Free-Atom Reference Data. Phys. Rev. Lett. 2009, 102, 073005.pubmed.ncbi.nlm.nih.gov/19257665
  59. [59]Neese, F. The ORCA program system. WIREs Computational Molecular Science 2012, 2, 73–78.

Paper details

vlnc:W_az9edsmd
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents