Electronic Characterization of a Charge-Transfer Complex Monolayer on Graphene
Avijit Kumar, Kaustuv Banerjee, Mikko M. Ervasti, Shawulienu Kezilebieke, Marc Dvorak, Patrick Rinke, Ari Harju, Peter Liljeroth
EElectronic Characterization of a Charge-Transfer ComplexMonolayer on Graphene
Avijit Kumar,
1, 2, ∗ Kaustuv Banerjee, Mikko M. Ervasti, ShawulienuKezilebieke, Marc Dvorak, Patrick Rinke, Ari Harju,
2, 3 and Peter Liljeroth † School of Basic Sciences, Indian Institute ofTechnology Bhubaneswar, Jatni, 752050 Khurda, India Department of Applied Physics, Aalto University, FI-00076 Aalto, Finland Varian Medical Systems Finland, FI-00270 Helsinki, Finland a r X i v : . [ c ond - m a t . m e s - h a ll ] F e b bstract Organic charge-transfer complexes (CTCs) formed by strong electron acceptor and strong elec-tron donor molecules are known to exhibit exotic effects such as superconductivity and chargedensity waves. We present a low-temperature scanning tunneling microscopy and spectroscopy(LT-STM/STS) study of a two-dimensional (2D) monolayer CTC of tetrathiafulvalene (TTF)and fluorinated tetracyanoquinodimethane (F TCNQ), self-assembled on the surface of oxygen-intercalated epitaxial graphene on Ir(111) (G/O/Ir(111)). We confirm the formation of the charge-transfer complex by d I /d V spectroscopy and direct imaging of the singly-occupied molecular or-bitals. High-resolution spectroscopy reveals a gap at zero bias, suggesting the formation of acorrelated ground state at low temperatures. These results point to new opportunities to realizeand study correlated ground states in charge-transfer complex monolayers on weakly interactingsurfaces. Organic charge-transfer complexes (CTCs) formed by electron-donor and -acceptormolecules are an intriguing and broad class of materials that can exhibit phenomena re-lated to strong electron correlations and electron-phonon coupling such as charge and spindensity waves, Mott metal-insulator transitions, charge ordering, spin-liquid phases, and su-perconductivity [1–6]. In bulk CTC crystals, donor and acceptor molecules typically stackin rows that maximize π − π electronic overlap along the rows only [7]. This anisotropyin the overlap results in pseudo one-dimensional electronic dispersion, providing a suitableplatform to investigate low-dimensional, as well as low-energy, physics.Despite the broad spectrum of intriguing physical phenomena that have been reportedin bulk CTCs, their two-dimensional (2D) films have been much less studied [8–15]. Inparticular, the studies have been confined to metal substrates, which strongly interact withthe molecular layer and mask the intrinsic electronic properties of the CTCs.The CTC formed out of tetrathiafulvalene (TTF) and tetracyanoquinodimethane (TCNQ)molecules is an archetypal example of a CTC. It possesses the highest bulk conductivityreported so far in a CTC and has been studied in detail [1, 7, 16, 17]. Another widely studiedsystem is formed by the Bechgaard salts consisting of small, planar organic molecules actingas an electron donor combined with an electron accepting small inorganic molecule. These ∗ [email protected] † peter.liljeroth@aalto.fi GaCl monolayeron Ag(111) [11, 15]. On the other hand, the substrate interaction can completely dominatethe low-energy electronic properties. On Au(111), TTF-TCNQ molecular states of the CTChybridize with the metal states to form dispersive interface states [8]. Further, the unpairedelectron of TCNQ molecules on the Au(111) surface exhibits the many-body Kondo effect dueto screening by the substrate conduction electrons [9]. Thus, the electronic properties of aCTC, especially close to the Fermi energy, can be strongly perturbed by the metal substrate,prohibiting the study of intrinsic electronic properties of CTC. Therefore, preparing 2Dfilms of CTCs on weakly interacting substrates is extremely desirable. Epitaxial graphenegrown on Ir(111) has been shown to decouple the adsorbate layer from the underlying metalsubstrate allowing investigation of intrinsic electronic properties of the adsorbate layers[18, 19].Here, we a present low-temperature scanning tunneling microscopy (LT-STM) study ofa 2D CTC of TTF and fluorinated TCNQ (F TCNQ) self-assembled on the surface ofoxygen-intercalated epitaxial graphene on Ir(111) (G/O/Ir(111)). Sequential deposition ofthe molecules on this surface leads to the formation of rotationally identical domains ofCTCs with alternating rows of TTF and F TCNQ lying parallel to the surface. The frontiermolecular orbitals of the molecular species in the CTC, as found from scanning tunnelingspectroscopy (STS), indicate charge transfer between TTF and F TCNQ molecules. High-resolution tunneling spectra exhibit a dip at Fermi Fermi energy closing at a temperatureof 20 K that may be attributed to the formation of a correlated ground state in the CTCmonolayer. These results open up interesting avenues for the investigation of exotic groundstates of the CTC in the monolayer limit.
RESULTS AND DISCUSSION
Figure 1 describes the assembly and structure of the TTF-F TCNQ CTC on a G/O/Ir(111)surface. The sample preparation is described in detail in the Methods section. Briefly, we3row a near monolayer coverage of graphene on Ir(111) by a combination of temperatureprogrammed growth (TPG) and chemical vapour deposition (CVD), as described previously[20–22], followed by oxygen intercalation to electronically decouple graphene from the un-derlying substrate [23]. Finally, the molecules are deposited at low temperatures ( ≈
100 K),followed by annealing at room temperature for 15-45 mins to allow the formation of highlyordered CTC islands.Figure 1a shows an STM topography image of oxygen intercalated graphene on Ir(111).The surface contains the periodic moir´e pattern of a G/Ir(111) surface with a periodicityof 25.4 ˚A. The additional superstructure visible on the surface is due to patches of (2 × I /d V -spectroscopy of the surface showing a phonon gap of ∼
160 mV [24, 25] (see Sup-porting Information (SI) Fig. S1a). Oxygen intercalation also results in strong p -dopingof graphene by ∼ ∼ I /d V spectra at high bias with the feedback loop on − here, the field-emission resonances allow to estimate the substrate work function [27–30](See SI Fig. S1b).Figure 1b shows an STM topograph of large islands of ordered CTC assembled on aG/O/Ir(111) surface. The long-range ordering is the result of the post-deposition room-temperature annealing; directly after the low-temperature deposition, we observe disorderedislands on the surface (see SI Fig. S2). The CTC islands grow across the step edges incarpet-like fashion [31, 32] and contain various domains rotated with respect to each other.Analysis of several images reveals a total of six domain orientations rotated w.r.t. each otherin multiples of 30 ◦ . Figure 1c shows a zoomed-in STM image to identify arrangement ofTTF and F TCNQ molecules within the CTC islands. As evident from the STM image,there are two different rows of molecules: one is composed of TTF and the other of F TCNQmolecules. Rows of TTF and F TCNQ are lying alternately on the surface. The molecularstructure obtained from density functional theory (DFT) calculations (see below) has beenoverlaid on the STM image for clarity. The molecular rows are found to be at an angle of ± ◦ w.r.t. graphene’s zigzag direction for each domain. The unit cell of the CTC is shown bya parallelogram with lattice parameters a = 18.5 ( ± b = 9.5 ( ± θ = 56 ( ± ◦ .This is the most common phase we observe for this stoichiometry ((F TCNQ) (TTF) ) of the4 IG. 1. Assembly and structure of the CTC on oxygen-intercalated graphene. (a) STM topographyimage of oxygen intercalated graphene on Ir(111). The additional superstructure apart from themoir´e is due to reconstruction of subsurface oxygen. Scale bar is 3 nm. Imaging parameters: 1.2nA and 10 mV. (b) Few large islands of CTC on the G/O/Ir(111) surface showing various domainsand the domain boundaries. Scale bar is 30 nm. Imaging parameters: 0.4 pA and 0.75 V. (c) Azoomed-in STM image of the CTC shows the arrangement of TTF and F TCNQ molecules. Eachmolecule forms a row next to the row of the other molecule. A molecular structure along with aunit cell is overlaid to elucidate the molecular arrangement within the unit cell. Scale bar is 2 nm.Imaging parameters: ∼ molecules. At a slightly different stoichiometry ((F TCNQ) x (TTF) − x ), we have observed a5heckerboard phase of the CTC where only F TCNQ rows are present and TTF moleculesare dispersed across in a checkerboard fashion (see SI Fig. S3).In order to further elucidate the structure of the molecular layer, we carried out a broadstructural search for different possible geometries using DFT (see Methods for details). Weperformed full structural relaxations of 300 CTC monolayers sampled by varying intermolec-ular distance, bond angles, and alignment with respect to the underlying graphene. Theinitial structures are systematically generated but done “by hand” without any input frommachine learning or structure search algorithms [33, 34]. After relaxation, the structuresare sorted by formation energy. One of the low energy conformations closely matches theexperimental structure both in terms of the unit cell dimensions ( a = 17.78 ˚A, b = 8.89 ˚A, θ = 60 ◦ ) and the relative orientation w.r.t. the graphene lattice (13.89 ◦ ). A DFT simulatedSTM image (at Fermi energy) is shown in Fig. 1d for the optimized geometry; it closelyresembles the STM image shown in Fig. 1c.We have also looked at the assembly of single component F TCNQ and TTF layers onthe G/O/Ir(111) surface. A sub-monolayer coverage of F TCNQ molecules forms chain-like structures (in contrast to non-planar adsorption on the G/Ir(111) surface [35]). On theother hand, TTF molecules tend to assemble in a close-packed geometry on the G/O/Ir(111)surface. The assembly of F TCNQ and TTF molecules is shown in SI Figs. S4 and S5.Figure 2 shows the experimental verification of charge transfer between TTF andF TCNQ molecules in the CTC by d I /d V spectroscopy and STM imaging. Fig. 2a compareslong-range d I /d V spectra recorded on F TCNQ molecules in single component chains tothose recorded in the CTC. The spectrum on the molecule in the chain shows a resonancecorresponding to the lowest unoccupied molecular orbital (LUMO) at 0.64 V without anyfeatures at negative bias. This indicates that the F TCNQ molecules on G/O/Ir(111) areneutral, in contrast to F TCNQ molecules on a G/Ir(111) surface, where they are chargedat lower sites of the moir´e pattern [35]. This difference is likely due to the increased workfunction of graphene due to oxygen intercalation. The spectrum recorded on a F TCNQmolecule in the CTC, on the other hand, shows two peaks at -0.44 V and 1.2 V. Fig. 2b com-pares the d I /d V spectrum on TTF molecules from the pristine assembly on a G/O/Ir(111)surface to that of TTF molecules from the CTC. Here, d I /d V spectrum on TTF moleculeshows a peak at -0.8 V, corresponding to the highest occupied molecular orbital (HOMO)of a neutral TTF molecule. Despite the high work function of the surface ( ∼ IG. 2. Charge transfer across the molecules. (a) Long range d I /d V -spectra on F TCNQ moleculesin a single-component chain on the G/O/Ir(111) surface (red line) and on the F TCNQ sites inthe CTC (black line). (b) Long range d I /d V -spectra on TTF molecules in a single componentassembly on G/O/Ir(111) (blue line) and on the TTF sites in the CTC (black line). (c) Biasdependent STM images of the CTC at the sample biases indicated in the figure. Size of each imageis 4 . × . . TTF molecules stay neutral. In the CTC, the spectrum on TTF molecules shows two peaksat -0.9 V and 0.95 V (similar to the two peaks on an F TCNQ molecule). The assignmentof these peaks is done on the basis of images recorded at sample biases at -0.5 and 0.8 V.The image at 0.8 V shows a relatively prominent TTF HOMO, while the image at -0.5 Vshows a relatively prominent F TCNQ LUMO [35] (see Fig. 2c). Electron transfer fromdonor TTF to acceptor F TCNQ molecules results in splitting of the TTF HOMO (-0.8 eVpeak) into singly occupied (SOMO, -0.95 V peak) and singly unoccupied molecular orbitals(SUMO, 0.95 V peak). Similarly, the F TCNQ LUMO (0.64 V peak) splits into SOMO(-0.44 V peak) and SUMO (1.2 V peak) after accepting an electron. Consequently, the TTFmolecule acquires a positive charge while F TCNQ molecules become negatively charged inthe CTC. The charge transfer between the molecules is also supported by DFT calculations,and based on Hirshfeld charge analysis [36] it amounts to ∼ ∼ ∼
100 meV),indicating that the coupling is quite weak.
FIG. 3. Short-range d I /d V -spectroscopy on the CTC. (a) Short-range d I /d V -spectroscopy onthe TTF and F TCNQ sites in the CTC show a dip at zero bias. (b) Magnetic field dependentd I /d V -spectra on a TTF site in the CTC shows that the shape and size of the zero-bias dip doesnot change with magnetic field up to 10 T. (c) Temperature-dependent d I /d V -spectra on a TTFsite in the CTC show that the dip is washed away with increasing temperature and the asymmetricbackground is also decreased at higher temperatures. (d) Temperature dependence of the zero-biasconductance (ZBC, normalized at the d I /d V at bias of 20 mV) shows saturation at 15-20 K. Interestingly, high-resolution d I /d V spectra on both molecules contains a dip close to zerobias which has pronounced asymmetry on TTF sites as is shown in Fig. 3a. Each spectrumshows a prominent dip at zero bias along with an asymmetry which is more pronounced onthe TTF sites. To investigate its origin, we have examined its dependence on temperature8nd on out-of-plane magnetic field. Care was taken to record these spectra on the samemolecule and with the same microscopic tip apex. Fig. 3b shows magnetic field dependentd I /d V spectra on the TTF sites of the CTC lattice in the range of 0 to 10 T. There is nomeasurable change in either the shape and size of the dip, or the observed asymmetry up tomagnetic field of 10 T. On the contrary, a clear temperature dependence is observed fromFig. 3c, which shows the temperature-dependent d I /d V -spectroscopy recorded on TTF sitesof the CTC from 2.7 K to 20 K (data on the F TCNQ site is shown in the SI Fig. S7). Theasymmetric dip is most prominent at the lowest temperature of 2.7 K. The dip amplitudedecreases with increasing temperature and at 20 K only a step at zero bias remains. Thetemperature dependence of the zero bias conductance (ZBC) extracted from these spectraclearly exhibits the saturation of the ZBC at temperatures between 15-20 K. This change inthe ZBC indicates the presence of a low-temperature correlated state, which we discuss inmore detail below.The temperature dependent spectroscopy measurements suggest that the signal can bedeconvoluted into a symmetric dip and a step; of these, only the latter remains visible at20 K. The deconvolution of a spectrum measured on a TTF site in the CTC is shown inFig. 4a. The entire spectrum (note the wider bias range here compared to Fig. 3) can bewell fitted (details of the fittings are described in the Methods section) by a sum of twoFano lineshapes [37]. Fig. 4b summarizes the temperature dependence of the half-widthhalf-maximum (HWHM) of the two Fano lineshapes used to fit the spectra on the TTFsite. The HWHM of the Fano lineshape corresponding to the dip at zero bias (Fano-1)shows a clear scaling with temperature. On the other hand, the HWHM of the step-likeFano lineshape (Fano-2) has a weaker temperature dependence. While the Fano lineshapeis taken here as a phenomenological description of the measured spectra, the choice is notcompletely arbitrary, as it typically arises in situations where there are two interfering tun-neling pathways present. For example, it is widely observed on Kondo impurities, wherethe interference occurs between a direct tip-sample tunneling and tunneling path via theKondo impurity [38–41]. In fact, a spectral shape combining a step-like Fano lineshape witha smaller energy gap-like feature - very similar to our measurements - has been observedon the heavy fermion compound URu Si [42]. There, the spectral response was explainedby a combination of Kondo screening of the uranium f -electrons and the gap-like featureresulting from a transition to a hidden order phase at low temperatures.9 IG. 4. Deconvoluting the low-bias features of the d I /d V spectra. (a) Short-range d I /d V -spectrumon on TTF molecules. The curve has been fitted with sum of two Fano functions: Fano-1 (brokenblack line) represents the central dip and Fano-2 (red line) represents the step. Final fit is indicatedby blue line. (b) Temperature-dependent evolution of HWHM extracted from the two Fano function(Fano-1: left, Fano-2: right) from the fits. (c) Short-range d I /d V -spectroscopy on CTC islands,recorded on a F TCNQ molecule showing steps at energies ∼ ∼
31 (arrow 2), ∼
31 (arrow 2), ∼
35 (arrow 3) and ∼
52 meV (arrow 4). (d) Temperature-dependent evolution ofthe steps at ∼ ∼
52 meV (Step-4: right).
Intriguingly, the d I /d V -spectra recorded on the F TCNQ molecules of the CTC (Fig. 4c- the bias range is again wider than in Fig. 3a) show additional step-like features at higherbiases, viz . at ± ±
35 and ±
52 mV. These steps can be attributed to inelastic electrontunneling processes similar to molecular vibrations of negatively charged F TCNQ molecules[9, 43]. The tunneling electrons can excite a molecular vibration once the sample biasmatches the energy of the corresponding vibrational mode [44–46]. The inelastic processcorresponds to opening of an additional tunneling channel and a sudden increase in thetunneling conductance. To corroborate this picture, we assess the phonon modes for theCTC monolayer with DFT (details in the Methods). There is good agreement between theenergies of the measured steps and the calculated energies of certain CTC phonon modes10ith a high electron-phonon coupling strength. Additionally, the calculated modes withstrong coupling strength near the energies of the inelastic steps are dominated by F TCNQvibrations (see SI Fig. S8 for details). This is consistent with our experiments, where we seethe inelastic steps only on the F TCNQ sites of the CTC.Although DFT calculations indicate presence of intermolecular phonon modes with en-ergies of a few mV, the temperature dependence of the dip close to zero bias does not fitwith thermally broadened inelastic steps. If we force a fit with an inelastic step to the data(feature marked with “1” in Fig. 4b), the position of the fitted step would be strongly tem-perature dependent (Fig. 4d, black symbols), which is not expected for inelastic features.The zero bias feature also washes out more quickly with temperature than what would beexpected for a vibrational transition, and at 15 K or above, only an asymmetric step re-mains, supporting the notion that it is a result of a gap closing transition. This is illustratedin Fig. S7b, which shows simulated temperature dependence assuming an inelastic step withthe experimental parameters fitted at T = 2 . TCNQ rows. This providesfurther evidence of the presence of CDW/Peierls ground state in the TTF-F TCNQ CTCmonolayer at low temperatures causing a gap in the density of states at the Fermi energy.
CONCLUSIONS
In conclusion, we have synthesized a monolayer of charge-transfer complex TTF-F TCNQon a weakly interacting epitaxial graphene substrate, and have investigated its intrinsicelectronic properties. TTF and F TCNQ molecules assemble into close-packed islands withalternating rows of TTF and F TCNQ molecules in a 1:1 stoichiometry. Low-temperatureSTM and STS measurements confirm the formation of a charge-transfer complex with d I /d V spectra consistent with the presence of TTF cations and F TCNQ anions. High-resolutionspectroscopy at low-temperatures and high magnetic fields show formation of a correlatedground state related to a CDW or Peirls instability with a transition temperature of 15-20 K. This work demonstrates CTC monolayers as intriguing example of two-dimensionalmaterials and opens up new avenues for exploring the low-energy, correlated physics oflow-dimensional charge-transfer complexes.
METHODS
Sample preparation.
The experiments were carried out in ultra-high vacuum (UHV),low-temperature scanning tunneling microscopes (STMs) (Createc LT-STM and UnisokuUSM-1300). Both STMs are equipped with a preparation chamber and operate at a base12ressure lower than 1 × − mbar. The sample was prepared by depositing F TCNQ andTTF molecules sequentially on an oxygen-intercalated graphene on Ir(111) substrate. TheIr(111) surface was cleaned by repeated cycles of sputtering using Ne ions at energy 1.5 kVand annealing at 900 ◦ C in an oxygen environment, followed by flashing to 1300 ◦ C. Epitax-ial graphene was grown using ethylene gas with a combination of temperature programmedgrowth (TPG) and chemical vapour deposition (CVD) steps to achieve a nearly full mono-layer coverage of graphene [20–22, 52]. In the TPG step, the cleaned Ir(111) substrate wasexposed to the ethylene gas for one minute at a pressure of 1 × − mbar followed by heatingthe substrate to 1300 ◦ C. The CVD step was carried out at this temperature by exposing thesubstrate to ethylene gas at 3 × − mbar for 60 s. This gives nearly a monolayer coverageof graphene on Ir(111) (G/Ir(111)). Oxygen intercalation of G/Ir(111) (G/O/Ir(111)) wascarried out by exposure of 9 × L oxygen at 225 ◦ C as reported by [23].The charge-transfer complex (CTC) was synthesized by first depositing ∼ TCNQ molecules on a G/O/Ir(111) surface at low substrate temperature ( ≈
100 K),followed by deposition of a similar amount of TTF molecules at a similar substrate temper-ature. This resulted in disordered islands of CTC on the surface. The sample was annealedat room temperature for 15-45 mins. to allow the formation of highly ordered CTC islands.While F TCNQ molecules were evaporated using a Knudsen cell heated to 92 ◦ C, TTFmolecules were evaporated from a home-made evaporator kept at temperature 23 ◦ C. Thedeposited amounts of the two molecules were adjusted to 1:1 stoichiometry (each of themat less than a half monolayer coverage). Subsequently, the sample was transferred into thelow-temperature STM housed within the same UHV system.
STM measurements.
The STM experiments were carried out at a temperature of 4.2K unless otherwise stated. Temperature-dependent measurements were carried out in theCreatec STM, while magnetic field dependent measurements were carried out in the UnisokuSTM. For the measurements at 2.7 K, the LHe cryostat of the STM was pumped, whilemeasurements at temperature higher than 4.2 K was achieved by heating the STM bya Zener diode installed on the STM scanner. To avoid any ambiguity, the temperature-dependent measurements were carried out on the same F TCNQ and TTF molecules of theCTC assembly using the same tip. Similar precautions were taken for the magnetic fieldmeasurements as well, where the same molecules and the tip was used for the full-rangeof the magnetic field sweep. STM measurements were carried out using mechanically cut13t/Ir tips. d I /d V -spectroscopy was performed using a standard lock-in technique, wherea voltage modulation with amplitude of 10-15 mV and 1-2 mV signal has been used forlong-range and short-range spectroscopies, respectively. WSxM [53] and Gwyddion ( http://gwyddion.net/ )[54] software were used to process all the STM images. Fitting of the d I /d V -spectra. We use two Fano lineshape functions to fit the short-ranged I /d V spectrum in Fig. 4a. The Fano lineshape function is: f Fano ( (cid:15) ) = A ( q + (cid:15) − (cid:15) Γ ) (cid:15) − (cid:15) Γ ) + c where A is the prefactor, (cid:15) is the energy, (cid:15) is offset from zero, Γ is the half width athalf maximum, q is the Fano parameter, and c is a constant background term. We firstfit the step-like Fano lineshape to capture the step of the spectrum (Fano-2) by excludingthe central dip during the fitting. Further, we subtract the step-like Fano fit (red linein Fig. 4a) from the spectrum to get a central dip which is fitted again using a dip-likeFano lineshape (Fano-1). The fitting process is repeated for all the recorded spectra atthe indicated temperatures to extract HWHM for the two Fano lineshapes as function oftemperature.To fit the temperature dependence of the pair of four step features seen in Fig. 4c (fouron each side of zero bias), we use a series of symmetric Fermi-Dirac distribution functionsas function of energy, (cid:15) : f step ( (cid:15) ) = (cid:88) i =1 (cid:0) f + F D + f − F D (cid:1) + s(cid:15) + c = (cid:88) i =1 (cid:32) a + i
11 + e (cid:15) + (cid:15)ikBT + a − i (cid:32) −
11 + e (cid:15) − (cid:15)ikBT (cid:33)(cid:33) + s(cid:15) + c where i is the step number, a + i is the amplitude of the i th step for (cid:15) > a − i is theamplitude of the corresponding step at (cid:15) < (cid:15) i is the position of the i th step, k B is theBoltzmann constant, T is the temperature, s is the slope, and c is a constant backgroundterm.To correct for the lock-in modulation voltage (V m ) we use the broadening function: f V m ( (cid:15) ) = 2 π (cid:60) (cid:112) V m − (cid:15) V m (cid:60) is the real part of a complex number. To account for thermal broadening dueto the temperature (T) of the tip, we use the derivative of the Fermi-Dirac distribution: f T ( (cid:15) ) = ∂∂(cid:15) (cid:18)
11 + e (cid:15)kBT (cid:19) Finally, the simulated LDOS is obtained by convolving these functions i.e. either, f Fanototal ( (cid:15) ) = f Fano ∗ f V m ∗ f T or, f steptotal ( (cid:15) ) = f step ∗ f V m ∗ f T The simulated LDOS is fitted to the experimental d I /d V spectra to obtain the intrinsiclinewidth Γ in the first case and the step-positions in the second case. DFT calculations.
Density functional theory calculations are performed with the full po-tential, all-electron, numeric atom-centered orbital code FHI-AIMS [55–58]. We use the stan-dard FHI-AIMS ‘light’ pre-constructed basis sets of numeric atomic orbitals. Supercell cal-culations are performed with a 8 × k -point sampling. We use the Perdew-Burke-Ernzerhof (PBE) generalized gradient approximation to the exchange-correlation functional[59]. Van der Waals interactions are included with the pair-wise Tkatchenko-Scheffler cor-rection [60]. Atomic forces are relaxed to less than 10 − eV/˚A. Vibrations are calculatedwith the finite difference method. Electron-phonon coupling constants are based on theelectronic friction approach [61, 62]. In pursuit of open materials science [63], the DFT re-laxed geometry of the monolayer is available in the NOvel MAterials Discovery (NOMAD)repository [64]. ACKNOWLEDGEMENTS
We thank Jose Lado for discussions. This research made use of the Aalto NanomicroscopyCenter (Aalto NMC) facilities and was supported by the European Research Council (ERC-2017-AdG no. 788185 “Artificial Designer Materials”) and Academy of Finland (Academyprofessor funding nos. 318995 and 320555, postdoctoral researcher nos. 309975 and 316347).15e gratefully acknowledge high performance computing resources from the Aalto Science-ITproject and the CSC-IT Center for Science, Finland. [1] D. J´erome, “Organic conductors: From charge density wave to TTF-TCNQ to superconduct-ing (TMTSF) PF ,” Chem. Rev., vol. 104, no. 11, pp. 5565–5592, 2004.[2] T. Enoki and A. Miyazaki, “Magnetic TTF-based charge-transfer complexes,” Chem. Rev.,vol. 104, no. 11, pp. 5449–5478, 2004.[3] H. Seo, C. Hotta, and H. Fukuyama, “Toward systematic understanding of diversity of elec-tronic properties in low-dimensional molecular solids,” Chem. Rev., vol. 104, no. 11, pp. 5005–5036, 2004.[4] B. J. Powell and R. H. McKenzie, “Strong electronic correlations in superconducting organiccharge transfer salts,” J. Phys.: Condens. Matt., vol. 18, no. 45, p. R827, 2006.[5] R. T. Clay and S. Mazumdar, “From charge- and spin-ordering to superconductivity in theorganic charge-transfer solids,” Phys. Rep., vol. 788, no. 38, pp. 1–89, 2019.[6] J. Zhang, W. Xu, P. Sheng, G. Zhao, and D. Zhu, “Organic donor-acceptor complexes as novelorganic semiconductors,” Acc. Chem. Res., vol. 50, no. 7, pp. 1654–1662, 2017.[7] M. Sing, U. Schwingenschl¨ogl, R. Claessen, P. Blaha, J. M. P. Carmelo, L. M. Martelo,P. D. Sacramento, M. Dressel, and C. S. Jacobsen, “Electronic structure of the quasi-one-dimensional organic conductor TTF-TCNQ,” Phys. Rev. B, vol. 68, p. 125111, Sept. 2003.[8] N. Gonzalez-Lakunza, I. Fern´andez-Torrente, K. J. Franke, N. Lorente, A. Arnau, and J. I.Pascual, “Formation of dispersive hybrid bands at an organic-metal interface,” Phys. Rev.Lett., vol. 100, p. 156805, Apr. 2008.[9] I. Fern´andez-Torrente, K. J. Franke, and J. I. Pascual, “Vibrational Kondo effect in pureorganic charge-transfer assemblies,” Phys. Rev. Lett., vol. 101, p. 217203, Nov. 2008.[10] F. J¨ackel, U. G. E. Perera, V. Iancu, K.-F. Braun, N. Koch, J. P. Rabe, and S.-W. Hla,“Investigating molecular charge transfer complexes with a low temperature scanning tunnelingmicroscope,” Phys. Rev. Lett., vol. 100, p. 126102, Mar. 2008.[11] K. Clark, A. Hassanien, S. Khan, K.-F. Braun, H. Tanaka, and S.-W. Hla, “Superconductivityin just four pairs of (BETS) GaCl molecules,” Nat. Nanotechnol., vol. 5, p. 261, Mar. 2010.
12] G. A. Rojas, P. Ganesh, S. J. Kelly, B. G. Sumpter, J. A. Schlueter, and P. Maksymovych,“Ionic disproportionation of charge transfer salt driven by surface epitaxy,” J. Phys. Chem.C, vol. 117, no. 38, pp. 19402–19408, 2013.[13] S. Jeon, P. W. Doak, B. G. Sumpter, P. Ganesh, and P. Maksymovych, “Thermodynamiccontrol of two-dimensional molecular ionic nanostructures on metal surfaces,” ACS Nano,vol. 10, pp. 7821–7829, Aug. 2016.[14] J. Rodr´ıguez-Fern´andez, M. Robledo, K. Lauwaet, A. Mart´ın-Jim´enez, B. Cirera, F. Calleja,S. D´ıaz-Tendero, M. Alcam´ı, L. Floreano, M. Dom´ınguez-Rivera, A. L. V´azquez de Parga,D. ´Ecija, J. M. Gallego, R. Miranda, F. Mart´ın, and R. Otero, “Tuning intermolecular chargetransfer in donor-acceptor two-dimensional crystals on metal surfaces,” J. Phys. Chem. C,vol. 121, pp. 23505–23510, Oct. 2017.[15] A. Hassanien, B. Zhou, H. Tanaka, A. Miyazaki, M. Tokumoto, A. Kobayashi, E. Zupanic, andI. Musevic, “Epitaxial growth of insulating and superconducting monolayers of (BETS) GaCl on Ag(111),” Phys. Stat. Sol. B, vol. 252, pp. 2574–2579, NOV 2015.[16] T. Nishiguchi, M. Kageshima, N. Ara-Kato, and A. Kawazu, “Behavior of charge densitywaves in a one-dimensional organic conductor visualized by scanning tunneling microscopy,”Phys. Rev. Lett., vol. 81, pp. 3187–3190, Oct 1998.[17] Z. Z. Wang, J. C. Girard, C. Pasquier, D. J´erome, and K. Bechgaard, “Scanning tunnelingmicroscopy in TTF-TCNQ: Phase and amplitude modulated charge density waves,” Phys.Rev. B, vol. 67, p. 121401, Mar 2003.[18] A. Kumar, K. Banerjee, and P. Liljeroth, “Molecular assembly on two-dimensional materials,”Nanotechnology, vol. 28, p. 082001, Jan. 2017.[19] A. Kumar, K. Banerjee, A. S. Foster, and P. Liljeroth, “Two-dimensional band structure inhoneycomb metal-organic frameworks,” Nano Lett., vol. 18, pp. 5596–5602, Sept. 2018.[20] A. T. N’Diaye, J. Coraux, T. N. Plasa, C. Busse, and T. Michely, “Structure of epitaxialgraphene on Ir(111),” New J. Phys., vol. 10, no. 4, p. 043033, 2008.[21] J. Coraux, A. T. N’Diaye, M. Engler, C. Busse, D. Wall, N. Buckanie, F.-J. Meyer zu Hering-dorf, R. van Gastel, B. Poelsema, and T. Michely, “Growth of graphene on Ir(111),” New J.Phys., vol. 11, no. 2, p. 023006, 2009.[22] S. K. H¨am¨al¨ainen, M. P. Boneschanscher, P. H. Jacobse, I. Swart, K. Pussi, W. Moritz,J. Lahtinen, P. Liljeroth, and J. Sainio, “Structure and local variations of the graphene moir´e n Ir(111),” Phys. Rev. B, vol. 88, p. 201406, Nov. 2013.[23] A. J. Martinez-Galera, U. A. Schroder, F. Huttmann, W. Jolie, F. Craes, C. Busse, V. Caciuc,N. Atodiresei, S. Blugel, and T. Michely, “Oxygen orders differently under graphene: newsuperstructures on Ir(111),” Nanoscale, vol. 8, no. 4, pp. 1932–1943, 2016.[24] Y. Zhang, V. W. Brar, F. Wang, C. Girit, Y. Yayon, M. Panlasigui, A. Zettl, and M. F.Crommie, “Giant phonon-induced conductance in scanning tunnelling spectroscopy of gate-tunable graphene,” Nat. Phys., vol. 4, p. 627, July 2008.[25] J. Halle, N. N´eel, M. Fonin, M. Brandbyge, and J. Kr¨oger, “Understanding and engineeringphonon-mediated tunneling into graphene on metal surfaces,” Nano Lett., vol. 18, no. 9,pp. 5697–5701, 2018.[26] S. Ulstrup, M. Andersen, M. Bianchi, L. Barreto, B. Hammer, L. Hornekær, and P. Hofmann,“Sequential oxygen and alkali intercalation of epitaxial graphene on Ir(111): enhanced many-body effects and formation of pn-interfaces,” 2D Materials, vol. 1, no. 2, p. 025002, 2014.[27] G. Binnig, K. H. Frank, H. Fuchs, N. Garcia, B. Reihl, H. Rohrer, F. Salvan, and A. R.Williams, “Tunneling spectroscopy and inverse photoemission: Image and field states,” Phys.Rev. Lett., vol. 55, pp. 991–994, Aug 1985.[28] C. L. Lin, S. M. Lu, W. B. Su, H. T. Shih, B. F. Wu, Y. D. Yao, C. S. Chang, and T. T.Tsong, “Manifestation of work function difference in high order Gundlach oscillation,” Phys.Rev. Lett., vol. 99, p. 216103, Nov. 2007.[29] F. Schulz, R. Drost, S. K. H¨am¨al¨ainen, and P. Liljeroth, “Templated self-assembly and localdoping of molecules on epitaxial hexagonal boron nitride,” ACS Nano, vol. 7, pp. 11121–11128,Dec. 2013.[30] F. Schulz, R. Drost, S. K. H¨am¨al¨ainen, T. Demonchaux, A. P. Seitsonen, and P. Liljeroth,“Epitaxial hexagonal boron nitride on Ir(111): A work function template,” Phys. Rev. B,vol. 89, p. 235429, Jun 2014.[31] K. Banerjee, A. Kumar, F. Federici Canova, S. Kezilebieke, A. S. Foster, and P. Liljeroth,“Flexible self-assembled molecular templates on graphene,” J. Phys. Chem. C, vol. 120, no. 16,pp. 8772–8780, 2016.[32] L. Yan, O. J. Silveira, B. Alldritt, O. Krejˇc´ı, A. S. Foster, and P. Liljeroth, “Synthesis andgating the charge state of a monolayer metal-organic framework,” 2020.
33] A. T. Egger, L. H¨ormann, A. Jeindl, M. Scherbela, V. Obersteiner, M. Todorovi´c, P. Rinke,and O. T. Hofmann, “Charge transfer into organic thin films: A deeper insight throughmachine-learning-assisted structure search,” Adv. Sci., vol. 7, no. 15, p. 2000992.[34] J. J¨arvi, P. Rinke, and M. Todorovi´c, “Detecting stable adsorbates of (1s)-camphor on cu(111)with bayesian optimization,” Beilstein J. Nanotechnol., vol. 11, pp. 1577–1589, 2020.[35] A. Kumar, K. Banerjee, M. Dvorak, F. Schulz, A. Harju, P. Rinke, and P. Liljeroth, “Charge-transfer-driven nonplanar adsorption of F TCNQ molecules on epitaxial graphene,” ACSNano, vol. 11, pp. 4960–4968, May 2017.[36] F. L. Hirshfeld, “Bonded-atom fragments for describing molecular charge densities,” Theoret.Chim. Acta, vol. 44, no. 2, pp. 129–138, 1977.[37] U. Fano, “Effects of configuration interaction on intensities and phase shifts,” Phys. Rev.,vol. 124, no. 6, p. 1866, 1961.[38] J. Li, W.-D. Schneider, R. Berndt, and B. Delley, “Kondo scattering observed at a singlemagnetic impurity,” Phys. Rev. Lett., vol. 80, no. 13, p. 2893, 1998.[39] V. Madhavan, W. Chen, T. Jamneala, M. Crommie, and N. Wingreen, “Tunneling into asingle magnetic atom: spectroscopic evidence of the Kondo resonance,” Science, vol. 280,no. 5363, pp. 567–569, 1998.[40] K. Nagaoka, T. Jamneala, M. Grobis, and M. Crommie, “Temperature dependence of a singleKondo impurity,” Phys. Rev. Lett., vol. 88, no. 7, p. 077205, 2002.[41] M. Ternes, “Spin excitations and correlations in scanning tunneling spectroscopy,” New J.Phys., vol. 17, p. 063016, jun 2015.[42] P. Aynajian, E. H. da Silva Neto, C. V. Parker, Y. Huang, A. Pasupathy, J. Mydosh,and A. Yazdani, “Visualizing the formation of the Kondo lattice and the hidden order inURu ,” Proc. Nat. Acad. Sci., vol. 107, no. 23, pp. 10383–10388, 2010.[43] M. Garnica, F. Calleja, A. L. V´azquez de Parga, and R. Miranda, “Mapping spin distributionsin electron acceptor molecules adsorbed on nanostructured graphene by the Kondo effect,”Surf. Sci., vol. 630, pp. 356–360, Dec. 2014.[44] N. Lorente, M. Persson, L. J. Lauhon, and W. Ho, “Symmetry selection rules for vibrationallyinelastic tunneling,” Phys. Rev. Lett., vol. 86, pp. 2593–2596, Mar 2001.[45] B. de la Torre, M. ˇSvec, G. Foti, O. c. v. Krejˇc´ı, P. Hapala, A. Garcia-Lekue, T. Frederiksen,R. Zboˇril, A. Arnau, H. V´azquez, and P. Jel´ınek, “Submolecular resolution by variation of the nelastic electron tunneling spectroscopy amplitude and its relation to the AFM/STM signal,”Phys. Rev. Lett., vol. 119, p. 166001, Oct 2017.[46] S. You, J.-T. L¨u, J. Guo, and Y. Jiang, “Recent advances in inelastic electron tunnelingspectroscopy,” Adv. Phys.: X, vol. 2, no. 3, pp. 907–936, 2017.[47] M. Ternes, “Probing magnetic excitations and correlations in single and coupled spin systemswith scanning tunneling spectroscopy,” Prog. Surf. Sci., vol. 92, no. 1, pp. 83 – 115, 2017.[48] L. Fritz and M. Vojta, “The physics of Kondo impurities in graphene,” Rep. Prog. Phys.,vol. 76, p. 032501, feb 2013.[49] J.-H. Chen, L. Li, W. G. Cullen, E. D. Williams, and M. S. Fuhrer, “Tunable Kondo effect ingraphene with defects,” Nat. Phys., vol. 7, pp. 535–538, Apr. 2011.[50] Y. Jiang, P.-W. Lo, D. May, G. Li, G.-Y. Guo, F. B. Anders, T. Taniguchi, K. Watan-abe, J. Mao, and E. Y. Andrei, “Inducing Kondo screening of vacancy magnetic moments ingraphene with gating and local curvature,” Nat. Commun., vol. 9, p. 2349, June 2018.[51] E. Abrahams, J. S´olyom, and F. Woynarovich, “The Landau theory of phase transitions inTTF-TCNQ (tetrathiafulvalene-tetracyanoquinodimethane),” Phys. Rev. B, vol. 16, pp. 5238–5249, Dec 1977.[52] C. Busse, P. Lazi´c, R. Djemour, J. Coraux, T. Gerber, N. Atodiresei, V. Caciuc, R. Brako,A. T. N’Diaye, S. Bl¨ugel, J. Zegenhagen, and T. Michely, “Graphene on Ir(111): Physisorptionwith chemical modulation,” Phys. Rev. Lett., vol. 107, p. 036101, July 2011.[53] I. Horcas, R. Fern´andez, J. M. G´omez-Rodr´ıguez, J. Colchero, J. G´omez-Herrero, and A. M.Baro, “WSXM: A software for scanning probe microscopy and a tool for nanotechnology,”Rev. Sci. Instrum., vol. 78, p. 013705, 2007.[54] D. Neˇcas and P. Klapetek, “Gwyddion: an open-source software for SPM data analysis,” Cent.Eur. J. Phys., vol. 10, pp. 181–188, 2012.[55] V. Blum, R. Gehrke, F. Hanke, P. Havu, V. Havu, X. Ren, K. Reuter, and M. Scheffler, “Abinitio molecular simulations with numeric atom-centered orbitals,” Comput. Phys. Commun.,vol. 180, no. 11, pp. 2175–2196, 2009.[56] V. Havu, V. Blum, P. Havu, and M. Scheffler, “Efficient integration for all-electron electronicstructure calculation using numeric basis functions,” J. Comput. Phys., vol. 228, no. 22,pp. 8367–8379, 2009.
57] X. Ren, P. Rinke, V. Blum, J. Wieferink, A. Tkatchenko, S. Andrea, K. Reuter, V. Blum, andM. Scheffler, “Resolution-of-identity approach to Hartree-Fock, Hybrid Density Functionals,RPA, MP2, and GW with numeric atom-centered orbital basis functions,” New J. Phys.,vol. 14, p. 053020, 2012.[58] S. V. Levchenko, X. Ren, J. Wieferink, R. Johanni, P. Rinke, V. Blum, and M. Scheffler,“Hybrid functionals for large periodic systems in an all-electron, numeric atom-centered basisframework,” Comput. Phys. Commun., vol. 192, p. 60, 2015.[59] J. P. Perdew, K. Burke, and M. Ernzerhof, “Generalized gradient approximation made simple,”Phys. Rev. Lett., vol. 77, pp. 3865–3868, Oct 1996.[60] A. Tkatchenko and M. Scheffler, “Accurate molecular van der Waals interactions from ground-state electron density and free-atom reference data,” Phys. Rev. Lett., vol. 102, p. 073005,Feb 2009.[61] M. Askerka, R. J. Maurer, V. S. Batista, and J. C. Tully, “Role of tensorial electronic frictionin energy transfer at metal surfaces,” Phys. Rev. Lett., vol. 116, p. 217601, May 2016.[62] R. J. Maurer, M. Askerka, V. S. Batista, and J. C. Tully, “Ab initio tensorial electronic frictionfor molecules on metal surfaces: Nonadiabatic vibrational relaxation,” Phys. Rev. B, vol. 94,p. 115432, Sep 2016.[63] L. Himanen, A. Geurts, A. S. Foster, and P. Rinke, “Data-driven materials science: Status,challenges, and perspectives,” Adv. Sci., vol. 6, no. 21, p. 1900808, 2019.[64] To avoid the generation of multiple DOIs, submission to NOMAD repository will follow in-clusion of reviewer’s comments. upplementary Information:Electronic Characterization of a Charge-TransferComplex Monolayer on Graphene S1 xygen-intercalated Graphene on Ir(111) FIG. S5. (a) Short range d I /d V spectroscopy of an oxygen-intercalated graphene on Ir(111)(G/O/Ir(111)) surface shows a phonon gap of ±
80 mV, which indicates that intercalated grapheneis significantly decoupled [24]. The setpoint is 10 pA at 0.5 V. (b) Field emission resonance (FER)spectrum recorded on the surface has the first peak at ∼ ∼ S2 isordered Islands of CTC FIG. S6. (a) An STM image of disordered islands of CTC on a G/O/Ir(111) surface. Imagingparameters: 0.8 pA and 1 V. The scale bar is 30 nm. (b) A zoomed-in STM image of a small areafrom panel (a) shows the arrangement of molecules in partially ordered and disordered islands. Thebright features in the island correspond to either one or two TTF molecules. Imaging parameters:0.8 pA and 1 V. The scale bar is 10 nm. S3 hecker-Board Phase of CTC FIG. S7. An STM topography image of the checkerboard phase of the CTC on a G/Ir(111) surface.The box indicates the boundary of two domains of the checkerboard phase. The boundary resemblesthe most common phase with stripes of TTF and F TCNQ molecules. Imaging parameters: 0.9pA and 1.4 V. The scale bar is 3 nm. S4 ssembly of F TCNQ Molecules on G/O/Ir(111)
FIG. S8. (a) Zig-zag arrangement of F TCNQ in F TCNQ molecular chains on a G/O/Ir(111)surface. Molecular structure of F TCNQ has been overlaid for clarity. Imaging parameters: 2 pAand 1.7 V. Scale bar is 2 nm. (b) A DFT simulated lowest unoccupied molecular orbital (LUMO)of F TCNQ. S5 ssembly of TTF Molecules on G/O/Ir(111) FIG. S9. (a) An STM image of a small TTF island shows an oblique unit cell of the TTF assemblywith unit cell of 8.0 ± ◦ . The unit cell and molecular structure of TTF havebeen overlaid for clarity. Imaging parameters: 0.7 pA and 0.6 V. Scale bar is 1 nm. (b) A DFTsimulated highest occupied molecular orbital (HOMO) of TTF. S6 lectronic Band Structure of CTC With or Without Graphene FIG. S10. (a) Electronic band structure (left panel) and DOS (right panel) of CTC for its spin-upconfiguration. The band structure in the spin-down configuration looks very similar to that of thespin-up configuration. (b) A zoomed-in plot of the band structure from panel (a). X and Y arepoints along the reciprocal lattice vectors. S7 emperature-Dependent Spectra – Experiment and Simulated FIG. S11. (a) Temperature-dependent d I /d V -spectra recorded on a F TCNQ site in the CTC.Similarly to the TTF site, the dip vanishes with increasing temperature with relatively weak asym-metry still present at 20 K. (b) Simulated temperature dependence of the d I /d V spectra assumingthat the central dip would correspond to inelastic steps, where we have used the parameters ex-tracted from the fitting of the IETS spectrum at 2.7 K. Further, each spectrum has been subjectedto thermal and modulation broadening using V amp = 2 mV for all spectra. S8 olecular Vibrations of CTC FIG. S12. (a) Total DOS of the calculated vibrational modes of CTC as function of energy. (b)Vibrations in the CTC corresponding to the vibrational energy mode of 30 meV. (c) Vibrations inthe CTC corresponding to the vibrational energy mode of 53 meV. S9 ssembly of CTC on G/Ir(111) FIG. S13. (a) An STM image showing a CTC assembly on graphene on a Ir(111) (G/Ir(111))surface. Similar to that on the G/O/Ir surface, the CTC has alternate rows of TTF and F TCNQmolecules. Imaging parameters: 10 pA and 0.1 V. Scale bar is 2 nm. (b) Short range d I /d V spectrum recorded on a TTF molecule in the CTC island shows a dip at zero sample bias. Thewidth of the zero-bias dip is comparable to that of a CTC on G/O/Ir(111) surface. A spectrumrecorded on a G/Ir(111) surface is presented as a reference. S10 harge Density Waves in CTC
FIG. S14. (a) STM topography image of CTC at imaging parameters: 5 pA and -100 mV. Thescale bar is 3 nm. (b) A contrast-optimized version of the topography in panel (a) shows periodictopography modulations. (c,d) Two-dimensional fast-Fourier transform (2D-FFT) of panel (a)shows features corresponding to the CTC lattice (red circles), spots due to the underlying graphenemoir´e (blue circles), and charge density wave modulations (white circles). The scale bar is 1 nm − ..