The Backside of Graphene: Manipulating Adsorption by Intercalation

Oct 16, 2013 - Daniela Dombrowski , Wouter Jolie , Marin Petrović , Sven Runte , Fabian Craes , Jürgen Klinkhammer , Marko Kralj , Predrag Lazić , ...
0 downloads 0 Views 3MB Size
Download PDF
Letter pubs.acs.org/NanoLett

The Backside of Graphene: Manipulating Adsorption by Intercalation Stefan Schumacher,*,† Tim O. Wehling,‡,§ Predrag Lazić,∥ Sven Runte,† Daniel F. Förster,† Carsten Busse,† Marin Petrović,⊥ Marko Kralj,⊥ Stefan Blügel,¶ Nicolae Atodiresei,¶ Vasile Caciuc,¶ and Thomas Michely† †

II. Physikalisches Institut, Universität zu Köln, Zülpicher Straße 77, 50937 Köln, Germany Institut für Theoretische Physik, Universität Bremen, Otto-Hahn-Allee 1, 28359 Bremen, Germany § Bremen Center for Computational Material Science (BCCMS), Universität Bremen, Am Fallturm 1a, 28359 Bremen, Germany ∥ Ruđer Bošković Institute, Bijenička 54, 10000 Zagreb, Croatia ⊥ Institut za fiziku, Bijenička 46, 10000 Zagreb, Croatia ¶ Peter Grünberg Institut (PGI) and Institute for Advanced Simulation (IAS), Forschungszentrum Jülich and JARA, 52425 Jülich, Germany ‡

S Supporting Information *

ABSTRACT: The ease by which graphene is affected through contact with other materials is one of its unique features and defines an integral part of its potential for applications. Here, it will be demonstrated that intercalation, the insertion of atomic layers in between the backside of graphene and the supporting substrate, is an efficient tool to change its interaction with the environment on the frontside. By partial intercalation of graphene on Ir(111) with Eu or Cs we induce strongly ndoped graphene patches through the contact with these intercalants. They coexist with nonintercalated, slightly pdoped graphene patches. We employ these backside doping patterns to directly visualize doping induced binding energy differences of ionic adsorbates to graphene through low-temperature scanning tunneling microscopy. Density functional theory confirms these binding energy differences and shows that they are related to the graphene doping level. KEYWORDS: graphene, intercalation, adsorption, doping, work function, STM

T

sublattice of Gr through a Peierls-instability type mechanism, mediated by RKKY-type interactions. Finally, using density functional theory (DFT) calculations, Huang et al.7 find a doping dependent bond strength of H to Gr, with a substantial binding energy increase through both, either n-doping or pdoping. Intercalation of epitaxial Gr by gas species or metals has been widely studied in the past.7−34 It provides a flexible means to modify the interaction of Gr with the substrate,15−20 to bestow Gr with magnetic properties,29−31 or even to enable new reaction pathways.8,9 In the present context most important, it provides a flexible means to drastically modify the chemical potential of the Dirac electrons. If we express the shift in the chemical potential with respect to the Gr band structure as usual by the shift of the Dirac point energy ED measured with respect to the Fermi level EF, a range from ED = 0.8 eV by pdoping through F intercalation32 to ED ≤ −1.3 eV by n-doping through alkali or earth alkali intercalation27,33,34 is obtained.

he electronic properties of graphene (Gr) are extremely sensitive to adsorption. Schedin et al.1 exploited this sensitivity to detect single NO2 molecules on Gr. NO2 molecule adsorption causes chemical doping of Gr and the resulting change of the charge carrier density gives rise to a measurable change in the resistivity of Gr. Subsequently, it was established in numerous experimental and theoretical investigations that even weakly binding adsorbates, not affecting the integrity of the Dirac cone, typically imply doping and thus a shift of the chemical potential within Gr.1−3 Because of the marginal density of states of undoped Gr, even a minor adsorption related charge transfer causes a substantial shift of the chemical potential. Reciprocity implies that the adsorbate binding itself must depend sensitively on the location of the chemical potential in Gr. Indeed, Brar et al.4 demonstrated this sensitivity by the observation of a gate voltage dependent ionization state of a Co adatom adsorbed to Gr. Using a field effect transistor based on bilayer Gr, Sato et al.5 could change the rate of molecular oxygen adsorption to Gr through a gate controlled variation of the chemical potential. Abanin et al.6 predict a doping dependent ordering of covalently bound adsorbates on one © XXXX American Chemical Society

Received: April 11, 2013 Revised: September 30, 2013

A

dx.doi.org/10.1021/nl402797j | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 1. (a) STM topograph after 10% ML Eu intercalated at 720 K (image size 120 nm × 120 nm, sample bias Us = −1.69 V, tunneling current It = 150 pA). (b) A total of 12% ML Eu adsorbed at 300 K (120 nm × 120 nm, −1.58 V, 30 pA). (c) Intercalation of 10% ML at 720 K and subsequent adsorption of 12% ML Eu at 300 K (120 nm × 120 nm, −1.58 V, 58 pA). The arrow in (c) highlights the location of a Eu-p(2 × 2) intercalation island which was not overgrown by the large Eu-(√3 × √3)R30° adsorption layer island. The imaging temperature in (a−c) was 35 K. (d) Schematic cross section along the line indicated in (c). (e) STM topograph after deposition of ∼10% ML Cs at 300 K (45 nm × 45 nm, +1.50 V, 10 pA, imaging temperature 6.5 K). (f) Schematic cross section along the line indicated in (e).

K for adsorption. The Eu flux was 3.5 × 1016 atoms m−2 s−1. Cs was evaporated from a well-degassed Cs dispenser in a background pressure in the low 10−9 mbar range at 300 K. The Cs flux was 5 × 1015 atoms m−2 s−1. Intercalated or adsorbed amounts θ are specified in monolayers (ML), where θ = 1 ML (or 100% ML) corresponds to a full (1 × 1) layer with respect to the Gr lattice. The energetics of Eu and Cs adsorption on Gr/Ir(111), Gr/ Eu-p(2 × 2)/Ir(111), and Gr/Cs-p(2 × 2)/Ir(111) have been theoretically studied by means of spin-polarized DFT39,40 calculations using the generalized gradient approximation (GGA)41 and the projector augmented waves (PAW)42,43 as implemented in the Vienna Ab Initio Simulation Package (VASP).44 The strong local Coulomb interaction of the Eu 4f electrons is accounted for within the GGA+U approach with the Coulomb parameters of U = 7 eV and J = 1 eV that are known to be well suited to describe rare earth systems.45,46 To model Gr/Ir(111), Gr/Eu-p(2 × 2)/Ir(111), and Gr/Cs-p(2 × 2)/Ir(111) we consider a commensurate surface unit cell, where an Ir(111) slab of four atomic layers was fixed to a perfect face-centered cubic (fcc) lattice with a lattice constant of 3.5 Å. Thus, the Ir(111) surface and Gr are forced to have the same lattice constant of 3.5 Å/√2 = 2.47 Å. In the calculations, the adsorbed (intercalated) Eu or Cs atoms were above (beneath) the centers of the C rings which rest above regular Ir hollow sites. The C as well as the Eu/Cs atoms have been relaxed until forces were below 0.02 eV Å−1. Figure 1a displays an STM topograph taken at 35 K after deposition of 10% ML Eu on Gr/Ir(111) at 720 K. Eu deposition at 720 K on Gr/Ir(111) results in its intercalation, where Eu forms a Eu-p(2 × 2) superstructure with respect to Gr,28 denoted in the following as Gr/Eu/Ir(111). Direct evidence for intercalation, details on the intercalation structure, and an analysis of the formation mechanism for Eu intercalation patterns can be found in ref 28 and its supplement. Upon intercalation of an amount less than 25% ML, the amount

The span of 2.1 eV is about an order of magnitude larger than the span of 0.2 eV typically achieved by gating of Gr.5 As intercalation leaves one side of Gr free, it allows for the exploration of the consequences of the chemical potential shift for the interaction of Gr with the environment. Here, we used patterned intercalation of Eu and Cs under Gr on Ir(111) as a unique tool to provide patches of strongly ndoped Gr located above the intercalant and marginally p-doped Gr in contact with the Ir(111) substrate, side-by-side. Controlled adsorption on such a doping pattern directly visualizes large differences in the binding energy of ionic adsorbates to Gr, be it in the form of adatoms, adatom clusters, or adsorption layer islands. This concept of manipulating the strength of adsorbate binding by local doping may be applicable also for other types of adsorbates as well as for Gr structuring. Scanning tunneling microscopy (STM) and spectroscopy (STS) experiments were performed in two ultrahigh vacuum systems for variable (30−700 K) and low (6.5 K) temperatures in Cologne. The STM images were digitally postprocessed with the WSxM software.35 Angle-resolved photoemission spectroscopy (ARPES) experiments were conducted in Zagreb (Eu intercalation) and at the APE beamline of the ELETTRA synchrotron in Trieste (Cs intercalation). All spectra were postprocessed using the Lucy-Richardson deconvolution procedure.36 In all systems, Ir(111) was prepared by cycles of noble gas ion sputtering, heating in oxygen (1 × 10−7 mbar at 1100 K), and annealing to 1500 K, yielding clean terraces with sizes of about 100 nm. A well-oriented, coherent, and closed Gr monolayer was prepared by room temperature ethylene adsorption till saturation, thermal decomposition at 1450 K, and subsequent exposure to ethylene (pressure in the range of 1−5 × 10−7 mbar) at 1150 K for 500−1200 s (TPG + CVD method).37 High-purity Eu38 was evaporated from a watercooled Knudsen cell in a background pressure below 1 × 10−10 mbar at 720 K sample temperature for intercalation and at 300 B

dx.doi.org/10.1021/nl402797j | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

necessary for the formation of a complete Eu-p(2 × 2) intercalation layer, Eu forms a pattern of stripes and islands underneath the continuous Gr sheet, as visible in Figure 1a. Stripes and islands are quantized in size by the moiré unit cell.28 At room temperature, Eu intercalation is inhibited. Therefore, Eu deposition at room temperature causes adsorption of Eu on Gr/Ir(111). For θ = 12% ML the adsorbed Eu phase separates into compact islands with a Eu-(√3 × √3)R30° structure coexisting with clusters of the same internal structure that consist of ∼15 atoms.47 Figure 1b imaged at 35 K to freeze adsorbate mobility displays such a phase coexistence. It is determined by the delicate balance of lowering the kinetic energy of Eu’s valence electrons by delocalization in Gr, the associated Coulomb penalty, and the Eu edge energy related to the finite size of the Eu clusters, as discussed in detail by Förster et al.47 Figure 1c imaged at 35 K visualizes the key experiment related to Eu adsorption. At room temperature, Eu is deposited on a partly Eu intercalated sample. Phase separation of the adsorbed Eu into islands and clusters is again observed. However, Eu clusters and Eu islands are only found on the Gr/ Ir(111) areas but not on the Gr/Eu/Ir(111) ones. The schematic cross section of Figure 1d along the line indicated in Figure 1c clarifies the morphology. The arrow in Figure 1c highlights that the large adsorbed Eu-(√3 × √3)R30° island displays a hole at the location of an intercalated Eu-p(2 × 2) island. The arrangement of the intercalation stripes on the upper terrace is rather inhomogeneous: a large area free of stripes, as under the large Eu-(√3 × √3)R30° island, surrounded by a dense stripe arrangement. Such an inhomogeneous arrangement has never been observed for pure intercalation (see Figure 1a and ref 28). As at the deposition temperature of 300 K intercalated Eu stripes and islands are able to fluctuate in position and shape (compare supplement of ref 28), we tentatively attribute the observed stripe arrangement to be the result of a dynamic response to the growth of the Eu-(√3 × √3)R30° adsorption island. At 300 K, the deposited Eu forms equilibrium structures, as explicitly demonstrated in ref 28. It is highly mobile and thus scans the surface for locations of strongest binding. Upon cooling, the morphology is frozen at some unknown temperature below 300 K. Imaging at 35 K therefore represents the equilibrium situation at this freezing temperature. The absence of adsorbed Eu clusters and islands on intercalated islands or stripes immediately implies that adsorbed Eu is more strongly bound to nonintercalated areas. Upon analyzing about 30 STM images with a 1 μm2 total scanning area of which 40% is intercalated, we find only one small 100 nm2 adsorbed island on an intercalated region, yielding a percentage of 2.5 × 10−4. As we adsorbed 12% ML Eu corresponding to 36% apparent coverage in a Eu-(√3 × √3)R30° structure, this fraction would be 0.36 for a random distribution. Therefore, the probability for Eu to adsorb on an intercalated area is drastically reduced by a factor of 1500. However, without knowing the precise value of the freezing temperature of the morphology (somewhere between 35 and 300 K) and the details of the binding energies within a cluster or island, it is not straightforward to obtain a quantitative value for the binding energy difference ΔEb for the atoms in the two areas. We note that when we increase the amount of Eu deposited beyond the area of acceptance provided by the nonintercalated Gr parts, also the Gr/Eu/Ir(111) areas start to become overgrown until eventually a complete Eu-(√3 × √3)R30° adsorption layer

forms. This does not contradict our inferences above, but is just a consequence of the fact that Eu also binds to Gr/Eu/Ir(111), though weaker than to Gr/Ir(111). A similar binding energy difference for adsorption to intercalated and nonintercalated areas may be inferred from experiments with Cs deposition on Gr/Ir(111) at room temperature. As displayed in Figure 1e for θ ≈ 10% ML, a phase separation of the deposited material is observed. The morphology is clarified by the schematic cross section in Figure 1f taken along the line indicated in Figure 1e. An adsorbed dilute gas of ionized Cs adatoms (the only phase for θ < 2% ML) coexists with large intercalated islands which have a Csp(2 × 2) structure with respect to Gr.48 The intercalation structure is denoted in the following as Gr/Cs/Ir(111). Direct evidence for Cs intercalation through atomically resolved STM and electron diffraction is provided as Supporting Information. In the present context, the key observation is that Cs ions are adsorbed only on Gr/Ir(111) areas, despite the fact that the adsorbed ions repel each other by Coulomb interaction. This is also reflected by the presence of Cs atoms adsorbed on the nonintercalated trenches of the large Gr/Cs/Ir(111) island. The absence of adsorbed Cs ions on intercalated islands again implies that adsorbed Cs ions are more strongly bound to nonintercalated areas, similar to the Eu case. The key to an understanding of why Eu and Cs adsorbates preferentially bind in the nonintercalated areas must be sought in local changes of the Gr electronic structure. Therefore, we conducted ARPES and STS experiments. For reference, Figure 2a displays the Gr electronic structure on Ir(111) along the ΓKM direction. It displays a largely intact Dirac cone with slight p-doping and an estimated Dirac point position of ED = 0.1 eV, that is, the Dirac point is above EF.49 Not visible in this spectrum is a hybridization of the Gr π-band with Ir (presumably with the surface state S3 as labeled in ref 49) right at the Fermi energy, as shown by the detailed investigation of Starodub et al.50 Though Gr on Ir(111) is physisorbed with a characteristic average height of 3.38 Å, the calculations of Busse et al. indicate this slight hybridization to be one of the C2pz orbitals with the Ir5d3z2−r2 orbitals for C atoms that sit exactly atop of Ir ones.51 For a complete Eu-p(2 × 2) intercalation layer a strong n-doping is found with an estimated ED = −1.38 eV, as shown in Figure 2b. Even more instructive are ARPES data after Cs intercalation. After deposition of about 10% ML, an amount corresponding to the situation visualized by the STM topograph of Figure 1e, two Dirac cones are observed in the spectrum of Figure 2c. The first cone is shifted just slightly below EF with ED = −0.12 eV and the second cone is shifted strongly with ED = −1.13 eV. Upon continued deposition the slightly shifted cone diminishes in intensity until it vanishes for layer completion, while the strongly shifted cone gains intensity. Comparison to Figure 1e enables a straightforward interpretation of the situation: the dilute gas of Cs ions induces just a slight n-doping of Gr in its area of adsorption, while the large intercalated Cs-p(2 × 2) islands induce strong n-doping of Gr covering them. Cs intercalation thus creates a doping pattern in Gr. As a side remark, we note that although the intercalated amounts of Cs and Eu for the experiments represented in Figure 2c,e are within the limits of error identical, the intercalated material is distributed quite differently; it forms thin stripes and small islands for Eu, while large islands, often bounded by substrate steps, are found for Cs intercalation. This is not unexpected in view of the fact that the electronic C

dx.doi.org/10.1021/nl402797j | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

nonintercalated areas. For simplicity, we focus here to single adatoms. For the morphology as sketched in Figure 3a, this link is visualized in the energy diagram of Figure 3b. Irrespective of whether Gr is intercalated or not, the work function Φ and the doping level ED display a constant energetic separation. In a Gedankenexperiment, one may partition the adsorption process of an ionic adsorbate to a substrate into three steps as sketched in Figure 3c. As initial state, we consider an atom near the substrate in a height, where a well-defined work function is established, but sufficiently distant to avoid wave function overlap of the atom and the substrate. As it costs no energy to move a neutral atom in this height parallel to the surface through areas of different work function, this initial state is welldefined. In step (1), an electron is removed from the atom. In good approximation, the ionization energy will not depend on the details of the substrate electronic structure. In step (2), the electron is moved to the substrate where it delocalizes. Thereby it lowers its kinetic energy by an amount identical to Φ, irrespective of how the charge is distributed within the substrate. Comparison with Figure 3a,b tells that in this step substantially more energy is gained in the nonintercalated areas (1.48 eV more for the Eu case and 1.23 eV more for the Cs case). In step (3), the positive ion is moved to the substrate. Assuming for the moment that the energy gain in this step does not depend on the substrate, the adsorption of the ionic adsorbate to nonintercalated areas is preferred by just ΔΦ = ΔED. In this simple model the binding energy difference of the ionic adsorbate is given by the difference in the Gr doping level ΔED. In order to see whether and to what extend this simple picture is adequate, we conducted DFT calculations. We consider Eu and Cs adatoms adsorbed to Gr/Ir(111), compare with Figure 4a, as well as to Gr/Eu/Ir(111) and Gr/Cs/ Ir(111), respectively, compare with Figure 4b. In all cases, the most favorable adsorption position of the adsorbed atoms is at the center of the C rings to which the Cs and Eu atoms relaxed freely during energetic optimization (compare also ref 47).

Figure 2. (a−c) ARPES spectra taken in ΓKM direction with a small azimuthal offset ∼1° in (a,c) and ∼1.5° in (b) implying that the shown cut is taken at a finite momentum ky relative to the Dirac point. (a) Gr/Ir(111). (b) After exposure to 25% ML Eu at 720 K resulting in a complete Gr/Eu/Ir(111) layer. (c) After 10% ML Cs deposition. Thin white dashed lines indicate fitted tight binding bands in the first nearest neighbor approximation taking into account the azimuthal offset. The corresponding Dirac point binding energies are (a) ED = 0.1 eV, (b) ED = −1.38 eV, and (c) ED = −0.12 eV as well as ED = −1.13 eV, respectively. In (a) and (c) 40.5 eV p-polarization and in (b) 21.2 eV mixed polarization photons were used. (d) It(z) measurements for Gr/Ir(111) (green) and Gr/Eu/Ir(111) (red) taken with the same tip at Us = −0.5 V sample bias. Here, z = 0 is the start position of tip retraction. Lines indicate fits to the data.

structure (e.g., one 6s electron for Cs and two 6s electrons for Eu) as well as atomic and ionic radii of the two materials differ. Similar to the Cs intercalation, also the Eu intercalation pattern must be assumed to create a doping pattern in Gr on an even finer scale. To establish its existence, we used STS and measured differences in the local work function on adjacent areas of Gr/Ir(111) and Gr/Eu/Ir(111) through the dependence of the tunneling current It on the distance z between tip and sample. It(z) measured with one and the same STM tip on both areas is displayed in Figure 2d. As discussed in ref 47, if measured with the same tip, the difference of the slopes of It(z) for the two cases allows one to determine the work function difference ΔΦ = ΦGr/Ir(111) − ΦGr/Eu/Ir(111). Here, we obtain ΔΦ = (1.5 ± 0.2) eV. This value is in remarkable agreement with the doping level difference given by ΔED = ED,Gr/Ir(111) − ED,Gr/Eu/Ir(111) = 0.1 eV − (−1.38 eV) = 1.48 eV. In view of the DFT calculations of Khomyakov et al.,52 this agreement is well understandable. The authors found that with an accuracy of better than 0.1 eV the difference between the work function Φads of Gr doped by its substrate and the Dirac point energy ED is a constant, namely the free Gr work function Φfree: Φfree = Φads − ED. This implies that for two areas ΔΦ must be identical to ΔED, as we observe experimentally. Therefore, our local work function measurement also reflects the presence of a doping pattern in Gr, which is related to the Eu intercalation pattern. The link between the doping level and the work function of Gr gives also access to understand the preferred binding to the

Figure 3. (a) Schematic cross section through a sample with Gr on Ir(111) which is partly intercalated. (b) Corresponding electron energy diagram for the case of an n-doping intercalant. (c) Partitioning the adsorption process of an ionic adsorbate in a Gedankenexperiment into three steps: (1) Ionization close to the surface, (2) transfer of the electron (e−) to the substrate, and (3) transfer of the remaining positively charged ion to the substrate (see text). D

dx.doi.org/10.1021/nl402797j | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 4. (a,b) Unit cells for Eu and Cs adatom adsorption on (a) Gr/Ir(111) and (b) Gr/Eu/Ir(111) as well as Gr/Cs/Ir(111). (c) Spin-polarized local density of states (LDOS) of the 6s orbitals for Eu adatoms adsorbed to Gr/Ir(111) (full line) and Gr/Eu/Ir(111) (dashed line) as obtained from DFT. The majority (minority) spin is shown on the positive (negative) ordinate. (d) Calculated LDOS of the 6s orbitals for Cs adatoms adsorbed to Gr/Ir(111) (full line) and Gr/Cs/Ir(111) (dashed line). (e−h) Color-coded charge density difference between the system and its parts, the free atom and the rest of the system, at the final relaxed geometry. The cuts are taken along the white lines indicated in (a,b). (e) Cs adatom on Gr/Ir(111), (f) Cs adatom on Gr/Cs/Ir(111), (g) Eu adatom on Gr/Ir(111), and (h) Eu adatom on Gr/Eu/Ir(111). The charge difference ranges from −0.03 (−0.01) electrons per Å3 blue to 0.03 (0.01) electrons per Å3 red for adsorbed Cs (Eu).

what we expect from the work function model but quantitatively they are smaller by a factor of 2 for the Eu case and by 25% for the Cs case. Apparently, the qualitative model sketched above oversimplifies the physical situation. In step (2), the electron gains the entire work function Φ only in the limit of infinitely dilute adsorption. Otherwise, the electron transfer itself will contribute to doping and thereby reduce the energy gain (quantum capacitance effect). More important, we assumed that the energy related to step (3) is independent of the substrate. However, the final electrostatic energy after the ion has moved to the substrate depends on the ability of the substrate to screen the positive ion. Considering only the Gr layer, to first approximation screening will be the better, the higher the LDOS at the Fermi level. Thus, screening will be better for adsorption to the intercalated, n-doped areas of Gr, which possess a much higher LDOS at the Fermi level. Consequently, our expected binding preference ΔEb to Gr/ Ir(111) related to the larger Φ (stronger lowering of the electron kinetic energy) will be diminished by the poorer screening for this case (larger Coulomb penalty). The charge difference plots of Figure 4e−h visualize the total charge redistribution upon adsorption for the four cases under concern. For adsorption to Gr/Ir(111), there is significantly more charge redistribution indicating a larger Coulomb penalty associated with charge transfer. The poorer screening of the cation through Gr for the case of Gr/Ir(111) is also visible, as the charge transfer extends to the Ir substrate atoms, while not even the intercalation layer is involved for the cases of Gr/Cs/ Ir(111) and Gr/Eu/Ir(111). The relevance of screening for the ionic binding energy is underlined by our DFT calculations for adsorption on freestanding graphene. To model the effect of pure Gr doping on the adsorption energy we considered freestanding Gr with

First, we analyze the 6s local density of states (LDOS) of the Eu adatoms on Gr/Ir(111) and Gr/Eu/Ir(111) in Figure 4c as well as of the Cs adatoms on Gr/Ir(111) and Gr/Cs/Ir(111) in Figure 4d. For the Eu adatom the majority spin Eu 6s orbital is in both cases below the Fermi level and occupied, while the minority spin LDOS shows spectral weight only above the Fermi level. As the 6s-orbital of a free Eu atom is doubly occupied, in both cases there is a charge transfer of about one electron from the Eu adatom to the substrate, that is, to Gr/ Ir(111) or to Gr/Eu/Ir(111). For the Cs adatom adsorbed on Gr/Ir(111) or on Gr/Cs/Ir(111) the 6s spectral weight is almost entirely above the Fermi level. As the 6s orbital of free Cs atoms is singly occupied, also the Cs adatom donates in both cases about one electron to the substrate. The major difference between the 6s LDOS of the adatoms in the adsorbed and the intercalated case is a rigid shift. In conclusion, Eu and Cs bind ionically to the bare and intercalated substrate by donation of one electron, as assumed in the model above. In qualitative agreement with the ARPES measurements, our DFT calculations show that intercalated Gr is strongly n-doped, while nonintercalated Gr is undoped. The Dirac point energies are ED ≈ 0 for Gr/Ir(111), ED = −1.2 eV for Gr/Eu/Ir(111) and −0.94 eV for Gr/Cs/Ir(111) deviating less than 0.2 eV from the experimental numbers given above. On the basis of these numbers and the model above we would expect the Eu adatom to bind more strongly by ΔEb = 1.2 eV and the Cs atom by ΔEb = 0.94 eV to Gr/Ir(111) as compared to the corresponding intercalated substrate. For the relaxed structures, we calculated binding energies of Eb = 1.2 eV and Eb = 0.6 eV for Eu adsorbed to Gr/Ir(111) and Gr/Eu/Ir(111), respectively. For the Cs adatom the corresponding values are Eb = 1.5 eV and Eb = 0.8 eV. The binding energy differences are ΔEb = 0.6 eV for the Eu adatom and ΔEb = 0.7 eV for Cs adatom. In sign and magnitude, these large binding energy differences are E

dx.doi.org/10.1021/nl402797j | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

As a last remark, we note that neither differences in diffusion coefficients of the adatoms on intercalated and nonintercalated areas nor variations in local graphene strain are able to account for our observations. First, the absence of adsorbed atoms and adatom clusters on the intercalated areas and their homogeneous distribution on the nonintercalated areas underline that at the adsorption temperature of 300 K adatom mobility is high enough to establish local equilibrium. Second, though slight inhomogeneities in strain are certainly present in Gr/Ir(111), Gr/Cs/Ir(111) and to a lesser extent in Gr/Eu/Ir(111),28 the strain variations do not exceed a few per mill28,57 and extend over mesoscopic scales,57 which is inconsistent with the sharp changes in adsorption behavior on the nanometer length scale associated with intercalation. In conclusion, the weaker binding of adatoms, adatom clusters, and extended islands of cationic adsorbates to Gr areas with strong n-doping is linked to work function changes resulting from the doping. These imply differences in lowering of the kinetic energy upon delocalization of the electron donated by the ion to the substrate system. The Coulomb penalty associated with charge redistribution upon adsorption, including screening, reduces the binding energy differences substantially below those estimated from work function differences. We envision that the doping level of Gr will not only affect ionic binding, but also van der Waals binding of organic molecules through the differences in the π- and π*orbital polarizabilities as a function of their doping-dependent filling. Likewise, we speculate that also covalent bonds, that is, the interaction of radicals with Gr, will be substantially influenced by the Gr doping level. By using a doping pattern, the latter would allow one to imprint a chemical pattern to Gr. This might have far reaching practical consequences for Gr lithography.

adsorbed Eu and added further electrons (together with a compensating positive jellium background) to the supercell. We find decreasing adsorption energies with increasing n-doping, as expected (see Table 1). But for perfectly freestanding Gr the ndoping induced binding energy decrease is even more than twice smaller as the related Dirac point shift (before Eu adsorption). We attribute this to an even stronger variation of screening with the doping level due to the absence of the metallic substrate. A comparison of Eu adsorption on undoped freestanding Gr and metal supported undoped Gr directly illustrates the better cation screening for supported graphene: The binding energy of Eu on Gr/Ir(111) is by 0.4 eV higher than for Eu on free-standing pristine Gr. Before concluding, we analyze in some detail an alternative hypothetical explanation for the preferred binding of the adsorbates to Gr/Ir(111) compared to Gr/Eu/Ir(111) or Gr/ Cs/Ir(111). It is well-known that transition metals form regular cluster arrays with the moiré pitch on Gr/Ir(111).53 A similar adsorption with the moiré pitch has been found also for atomic hydrogen.54 Feibelman55 pointed out that the adsorbates induce a rehybridization of Gr from sp2 to sp3 diamondlike carbon underneath the adsorbates in precisely those locations (with C-atoms exactly atop of Ir atoms), where even without adsorbates a slight hybridization of Gr with the substrate Ir was identified.51 The rehybridization is accompanied by a drastic decrease in C-metal bond distances from ∼3.4 Å characteristic for physisorption to ∼2.1 Å characteristic for chemisorption.55,56 Therefore, one might consider the preferential binding of the ionic adsorbates to Gr/Ir(111) to be rather due to the larger reactivity of the nonintercalated areas compared to the intercalated ones, that is, their larger ability to rehybridize upon adsorption by taking benefit of the already existing slight hybridization of Gr with Ir at specific locations in the moiré. However, this effect is irrelevant here. (i) Even the slightest binding energy inhomogeneity within the moiré unit cell should be reflected by the lateral distribution of Cs adatoms or Eu clusters on Gr/Ir(111), causing ordering with respect to the moiré. The STM topographs of Figure 1, where such distributions were slowly cooled from room temperature to a frozen state, make plain that such an ordering is absent. This is consistent with a continuous variation of the Eu cluster separation upon deposition at low temperatures.47 (ii) Upon adsorption of a Cs or Eu adatom on Gr/Ir(111) there is no change in height of Gr above Ir(111). Such a change would be expected if rehybridization is of any relevance. (iii) There is virtually no change in the width of the 6s resonances in the LDOS plots of Figure 4c,d. This rules out significant changes in the hybridization between the intercalated and the bare regions of the surface. (iv) The same effect, preferential binding to undoped Gr is also observed for free-standing Gr as summarized in Table 1

■

S Supporting Information *

The Supporting Information contains an atomically resolved STM topograph of Gr on a Cs intercalation island, the Fourier transform of a larger scale STM topograph, and a LEED pattern of Cs intercalated Gr. The data provide sound evidence for the formation of a dense Cs-p(2 × 2) intercalation layer. This material is available free of charge via the Internet at http:// pubs.acs.org.

■

ED (eV)

Eb (eV)

0.00 0.01 0.02 0.06

0.00 0.45 0.65 1.01

0.72 0.55 0.45 0.31

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge financial support from Deutsche Forschungsgemeinschaft through SFB608, project MI581/17-2, and INST 2156/514-1, the Ministry of Science of the Republic of Croatia and Deutscher Akademischer Austauschdienst via the project “Electrons in two dimensions”, and the Ministry of Science of the Republic of Croatia via project No. 0350352828-2840. Critical reading of the manuscript and useful discussions with A. Rosch as well as experimental assistance by I. Šrut are acknowledged.

Table 1. Adsorption Energies Eb of Eu on Doped Freestanding Gr Depending on the Charge Doping Level q Related to the Dirac Point Energy ED before Eu Adsorption q (e−/2C-atoms)

ASSOCIATED CONTENT

F

dx.doi.org/10.1021/nl402797j | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

■

Letter

(32) Walter, A. L.; Jeon, K.-J.; Bostwick, A.; Speck, F.; Ostler, M.; Seyller, T.; Moreschini, L.; Kim, Y. S.; Chang, Y. J.; Horn, K.; Rotenberg, E. Appl. Phys. Lett. 2011, 98, 184102. (33) Bianchi, M.; Rienks, E. D. L.; Lizzit, S.; Baraldi, A.; Balog, R.; Hornekær, L.; Hofmann, P. Phys. Rev. B 2010, 81, 041403(R). (34) McChesney, J. L.; Bostwick, A.; Ohta, T.; Seyller, T.; Horn, K.; González, J.; Rotenberg, E. Phys. Rev. Lett. 2010, 104, 136803. (35) Horcas, I.; Fernández, R.; Gómez-Rodríguez, J. M.; Colchero, J.; Gómez-Herrero, J.; Baro, A. M. Rev. Sci. Instrum. 2007, 78, 013705. (36) Rameau, J. D.; Yang, H.-B.; Johnson, P. D. J. Electron Spectrosc. Relat. Phenom. 2010, 181, 35. (37) van Gastel, R.; N’Diaye, A. T.; Wall, D.; Coraux, J.; Busse, C.; Buckanie, N. M.; Meyer zu Heringdorf, F.-J.; Horn-von Hoegen, M.; Michely, T.; Poelsema, B. Appl. Phys. Lett. 2009, 95, 121901. (38) Ames Laboratory (Materials Preparation Center) of the U.S. DOE, Iowa State University, Ames, IA 50011-3020, U.S.A. (39) Hohenberg, P.; Kohn, W. Phys. Rev. 1964, 136, B864. (40) Kohn, W.; Sham, L. J. Phys. Rev. 1965, 140, A1133. (41) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865. (42) Blöchl, P. E. Phys. Rev. B 1994, 50, 17953. (43) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758. (44) Kresse, G.; Hafner, J. J. Phys.: Condens. Matter 1994, 6, 8245. (45) Anisimov, V. I.; Aryasetiawan, F.; Lichtenstein, A. I. J. Phys.: Condens. Matter 1997, 9, 767. (46) Larson, P.; Lambrecht, W. R. L.; Chantis, A.; van Schilfgaarde, M. Phys. Rev. B 2007, 75, 045114. (47) Förster, D. F.; Wehling, T. O.; Schumacher, S.; Rosch, A.; Michely, T. New J. Phys. 2012, 14, 023022. (48) Petrović, M.; et al. Nat. Comm. 2013. (49) Pletikosić, I.; Kralj, M.; Pervan, P.; Brako, R.; Coraux, J.; N’Diaye, A. T.; Busse, C.; Michely, T. Phys. Rev. Lett. 2009, 102, 056808. (50) Starodub, E.; Bostwick, A.; Morescheni, L.; Nie, S.; El Gabaly, F.; McCarthy, K. F.; Rotenberg, E. Phys. Rev. B 2011, 83, 125428. (51) Busse, C.; Lazić, P.; Djemour, R.; Coraux, J.; Gerber, T.; Atodiresei, N.; Caciuc, V.; Brako, R.; N’Diaye, A. T.; Blügel, S.; Zegenhagen, J.; Michely, T. Phys. Rev. Lett. 2011, 107, 036101. (52) Khomyakov, P. A.; Giovannetti, G.; Rusu, P. C.; Brocks, G.; van den Brink, J.; Kelly, P. J. Phys. Rev. B 2009, 79, 195425. (53) N’Diaye, A. T.; Bleikamp, S.; Feibelman, P. J.; Michely, T. Phys. Rev. Lett. 2006, 97, 215501. (54) Balog, R.; et al. Nat. Mater. 2010, 9, 315. (55) Feibelman, P. J. Phys. Rev. B 2008, 77, 165419. (56) Franz, D.; Runte, S.; Busse, C.; Schumacher, S.; Gerber, T.; Michely, T.; Mantilla, M.; Kilic, V.; Zegenhagen, J.; Stierle, A. Phys. Rev. Lett. 2013, 110, 065503. (57) N’Diaye, A. T.; van Gastel, R.; Martínez-Galera, A. J.; Coraux, J.; Hattab, H.; Wall, D.; Meyer zu Heringdorf, F.-J.; Horn-von Hoegen, M.; Gómez-Rodríguez, J. M.; Poelsema, B.; Busse, C.; Michely, T. New J. Phys. 2009, 11, 113056.

REFERENCES

(1) Schedin, F.; Geim, A. K.; Morozov, S. V.; Hill, E. W.; Blake, P.; Katsnelson, M. I.; Novoselov, K. S. Nat. Mater. 2007, 6, 652. (2) Wehling, T. O.; Katsnelson, M. I.; Lichtenstein, A. I. Chem. Phys. Lett. 2009, 476, 125. (3) Liu, H.; Liu, Y.; Zhu, D. J. Mater. Chem. 2011, 21, 3335. (4) Brar, V.; Decker, R.; Solowan, H.-M.; Wang, Y.; Maserati, L.; Chan, K. T.; Lee, H.; Girit, C. O.; Zettl, A.; Louie, S. G.; Cohen, M. L.; Crommie, M. F. Nat. Phys. 2011, 7, 43. (5) Sato, Y.; Takai, K.; Enoki, T. Nano Lett. 2011, 11, 3468. (6) Abanin, D. A.; Shytov, A. V.; Levitov, L. S. Phys. Rev. Lett. 2010, 105, 086802. (7) Huang, L. F.; Ni, M. Y.; Zhang, G. R.; Zhou, W. H.; Li, Y. G.; Zheng, X. H.; Zeng, Z. J. Chem. Phys. 2011, 135, 064705. (8) Sutter, P.; Sadowski, J. T.; Sutter, E. A. J. Am. Chem. Soc. 2010, 132, 8175. (9) Mu, R.; Fu, Q.; Jin, L.; Yu, L.; Fang, G.; Tan, D.; Bao, X. Angew. Chem., Int. Ed. 2012, 51, 4856. (10) Starodub, E.; Bartelt, N. C.; McCarty, K. F. J. Phys. Chem. C 2010, 114, 5134. (11) Vinogradov, N. A.; Schulte, K.; Ng, M. L.; Mikkelsen, A.; Lundgren, E.; Martensson, N.; Preobrajenski, A. B. J. Phys. Chem. C 2011, 115, 9568. (12) Grånäs, E.; Knudsen, J.; Schröder, U. A.; Gerber, T.; Busse, C.; Arman, M. A.; Schulte, K.; Andersen, J. N.; Michely, T. ACS Nano 2012, 6, 9951. (13) Larciprete, R.; Ulstrup, S.; Lacovig, P.; Dalmiglio, M.; Bianchi, M.; Mazzola, F.; Hornekær, L.; Orlando, F.; Baraldi, A.; Hofmann, P.; Lizzit, S. ACS Nano 2012, 6, 9551. (14) Herbig, C.; Kaiser, M.; Bendiab, N.; Schumacher, S.; Förster, D. F.; Coraux, J.; Meerholz, K.; Michely, T.; Busse, C. J. Phys.: Condens. Matter 2012, 24, 314208. (15) Shikin, A. M.; Prudnikova, G. V.; Adamchuk, V. K.; Moresco, F.; Rieder, K.-H. Phys. Rev. B 2000, 62, 13202. (16) Dedkov, Y. S.; Shikin, A. M.; Adamchuk, V. K.; Molodtsov, S. L.; Laubschat, C.; Bauer, A.; Kaindl, G. Phys. Rev. B 2001, 64, 035405. (17) Varykhalov, A.; Sánchez-Barriga, J.; Shikin, A. M.; Biswas, C.; Vescovo, E.; Rybkin, A.; Marchenko, D.; Rader, O. Phys. Rev. Lett. 2008, 101, 157601. (18) Riedl, C.; Coletti, C.; Iwasaki, T.; Zakharov, A. A.; Starke, U. Phys. Rev. Lett. 2009, 103, 246804. (19) Forti, S.; Emtsev, K. V.; Coletti, C.; Zakharov, A. A.; Riedl, C.; Starke, U. Phys. Rev. B 2011, 84, 125449. (20) Robinson, J. A.; Hollander, M.; LaBella, M., III; Trumbull, K. A.; Cavalero, R.; Snyder, D. W. Nano Lett. 2011, 11, 3875. (21) Rut’kov, E. V.; Gall’, N. R. Phys. Solid State 2009, 51, 1738. (22) Virojanadara, C.; Watcharinyanon, S.; Zakharov, A. A.; Johansson, L. I. Phys. Rev. B 2010, 82, 205402. (23) Jin, L.; Fu, Q.; Mu, R.; Tan, D.; Bao, X. Phys. Chem. Chem. Phys. 2011, 13, 16655. (24) Meng, L.; Wu, R.; Zhou, H.; Li, G.; Zhang, Y.; Li, L.; Wang, Y.; Gao, H.-J. Appl. Phys. Lett. 2012, 100, 083101. (25) Sicot, M.; Leicht, P.; Zusan, A.; Bouvron, S.; Zander, O.; Weser, M.; Dedkov, Y. S.; Horn, K.; Fonin, M. ACS Nano 2012, 6, 151. (26) Gao, T.; Gao, Y.; Chang, C.; Chen, Y.; Liu, M.; Xie, S.; He, K.; Ma, X.; Zhang, Y.; Liu, Z. ACS Nano 2012, 6, 6562. (27) Pletikosić, I.; Kralj, M.; Milun, M.; Pervan, P. Phys. Rev. B 2012, 85, 155447. (28) Schumacher, S.; Förster, D. F.; Rösner, M.; Wehling, T. O.; Michely, T. Phys. Rev. Lett. 2013, 110, 086111. (29) Weser, M.; Voloshina, E. N.; Horn, K.; Dedkov, Y. S. Phys. Chem. Chem. Phys. 2011, 13, 7534. (30) Coraux, J.; N’Diaye, A. T.; Rougemaille, N.; Vo-Van, C.; Kimouche, A.; Yang, H.-X.; Chshiev, M.; Bendiab, N.; Fruchart, O.; Schmid, A. K. J. Phys. Chem. Lett. 2012, 3, 2059. (31) Decker, R.; Brede, J.; Atodiresei, N.; Caciuc, V.; Blügel, S.; Wiesendanger, R. Phys. Rev. B 2013, 87, 041403. G

dx.doi.org/10.1021/nl402797j | Nano Lett. XXXX, XXX, XXX−XXX