Surface Science 605 (2011) 72–80

Contents lists available at ScienceDirect

Surface Science

j ourna l homepage: www.e lsev ie r.com/ locate /susc

Surface interactions of molecular C60 and impact on Ni(100) and Co(0001) filmgrowth: A scanning tunneling microscopy study

Marvin Cummings a,⁎, SebastianGliga c, Boris Lukanov b, Eric I. Altman b,Matthias Bode c, EnriqueV. Barrera a,⁎a Rice University, Department of Mechanical Engineering & Materials Science, 6100 Main Street MS-321, Houston, TX 77005, United Statesb Yale University, Department of Chemical & Environmental Engineering, 55 Prospect Street MEC 111, New Haven, CT 06511, United Statesc Argonne National Laboratory, Center for Nanoscale Materials, 9700 S. Cass Ave. Bldg 440, Argonne, IL 60439, United States

⁎ Corresponding authors. Barrera is to be contacted at713 348 5423. Cummings, Tel.: +1 281 536 0571; fax: +

E-mail addresses: marvinc@rice.edu (M. Cummings)

0039-6028/$ – see front matter © 2010 Elsevier B.V. Adoi:10.1016/j.susc.2010.10.002

a b s t r a c t

a r t i c l e i n f o

Article history:Received 29 June 2010Accepted 1 October 2010Available online 8 October 2010

Keywords:C60

MobilityNi(100)Co(0001)MoiréEpitaxy

Here, the interactions of C60 at the surface of pseudomorphic Ni/Cu(100) and Co/Ru(0001) thin films and itseffect on film growth and morphology were determined using in-situ scanning tunneling microscopy (STM)and Auger electron spectroscopy (AES). The novel development of C60-metallic based nanosystems, such asC60 molecular junction transistors, hinges on our ability to understand the factors governing structuralstability in these nanosystems and the nature of the bond interactions at the C60–metal interface. In this study,C60 deposited onto the Ni(100) film surface is observed to be fairly immobile and uniformly distributed acrossthe Ni surface. On the Co(0001) film surface however, C60 mobility is observed to be severely limited in someregions and highly mobile in others dependent upon Co film surface reconstruction, resulting in a non-uniform distribution of C60 across the Co film surface. Despite the presence of C60 on the Ni surface, there is noobvious influence of the C60 on further Ni film growth. In contrast, during Co film growth, islands onlynucleate and grow from step edges or locally around C60 molecules. The strength of the Co–C60 bondinteraction appears stronger than the Co–Co bond on Co film terrace. Generally, the Ni and Co films bothcontinue epitaxial film growth in the presence of molecular C60. AES results indicate the C60 moleculesmaintain their chemical integrity during growth.

Tel.: +1 713 348 6242; fax: +11 713 348 5423., ebarrera@rice.edu

ll rights reserved.

© 2010 Elsevier B.V. All rights reserved.

1. Introduction

C60 fullerene–metallic nanosystems have been the subject of manyrecent scientific studies, as the unique physical and chemical propertiesof these composite systems (superconductivity and ionocovalentbonding) offer great potential in a wide array of device applications.Some of these applications include the use of C60 fullerenes inmoleculartransistor junctions and composite thin film magnetic media [1–5]. Ineach application, the C60–metal interfaceplays a fundamental role in thefunctionality of the device.

However, the exact properties of C60 critically depend on itsstructural environment. For example in the case of moleculartransistor junctions, Kaun et al. showed theoretical evidence thatchanges in the bond orientation of the C60 molecule at the C60–Auinterface activate different tunneling resonance channels within themetal–fullerene device [6]. The evidence of this has also been foundexperimentally for magnetic (C60–Ni) transistor junctions. Further-

more, for magnetic Co–C60/Fe–C60 composite films, Zheng et al. showthat the bulkmagnetic properties of these thinfilms can bemanipulatedmerely by the grain boundary concentration of molecular C60. Althougheach of the applications above has a promising potential, the fabricationof functional devices has proven difficult and unreliable. Successfulimplementation of these fullerene–metal nanosystems as full-scale,functional devices hinges on both the structural stability of the systemand the nature of the bond interactions at the interface. The growth offullerene–metal superlattices is based entirely on the assumption of astrongbonding interaction betweenC60–fullerene and themetal speciesinvolved; strong bonding at the interface encourages epitaxy, requiredfor the growth of atomically-flat superlattices.

To maximize the true potential and properties of fullerene–metallicnanosystems, the physical phenomena that drive both the growth andinteractions at fullerene–metal interfaces need to be understood. In thispaper we investigate the influence of C60 on the initial growth ofstrained homoepitaxial Ni(100) and Co(0001) films bymeans of in-situscanning tunneling microscopy (STM) combined with Auger electronspectroscopy (AES). We show that both the Ni and Co continuecrystalline growth unperturbed by the presence of C60 adsorbed ontothe strained film surfaces. Eventually, the molecular C60 is encapsulatedbeneath the film surface, incorporating the C60molecule within the filmnetwork.

Fig. 2. Topograph of 0.03 ML C60 adsorbed onto an annealed Ni(100) film surface. Theinsets on the right show from top to bottom: C60 monomer, dimer, trimer, pentamer,and a heptamer cluster.

73M. Cummings et al. / Surface Science 605 (2011) 72–80

2. Experimental procedures

TheC60/Ni/Cu(100) and theC60/Co/Ru(0001) systemswere grown inan ultra-high vacuum (UHV) chamber (base pressure ~5×10−11 mbar)equipped with low-energy electron diffraction (LEED), Auger electronspectroscopy (AES), and in-situ STM. A clean Cu(100) substrate surfacewas prepared by repeated Ar+sputter cycles performed at roomtemperature and 900 K, followed by a 5-minute flash anneal at 950 K inUHV [7]. The Ru(0001) substrate surface was prepared using 3 flashanneals in anO2-ambient pressure of 5×10−9 mbar, followed by2flashanneals in UHV. LEED and AES were used to verify the cleanliness of thesubstrate surfaces. STM images (not shown here) show atomically-flatterraces with widths of ~4000 Å and ≥5000 Å on the Cu (100) and Ru(0001) substrate surfaces, respectively. Ni and Co films were grownfrommetallic rods using e-beam evaporation sources. The Ni filmsweregrown to a total thicknessof 7.6±0.6 MLat a deposition rate of 0.95 ML/min. Afterwards, the Ni films were then annealed at ~455 K for 30 min.Co films were grown to a total thickness of 1.8±0.2 ML at a depositionrate of 0.23 ML/min. After deposition was complete, the Co films wereannealed at 550 K for 8 min.

After the films were grown, the sub-ML amounts of C60 fullerenes(MER-corporation, purity 99.9%) were sublimed from a quartzcrucible held at 740 K onto a room temperature Ni(100) and Co(0001) film surfaces, respectively. The C60 coverage was approxi-mately 0.03 ML for Ni, and the C60 coverage was approximately0.04 ML for Co, as determined from STM. The local effect of C60 on Niand Co film growths was determined on each film (Ni and Co) surface,respectively, by depositing additional monolayers of Ni and Co atroom temperature, post C60 growth. STM imaging of both the Co andNi films was performed at room temperature. During STM imaging,the tip was grounded and sample bias was varied. Typically, the tipcurrent setpoint was set at 20 pA to reduce tip–molecule interaction.

3. Results and discussion

3.1. Sub-ML C60 coverage on annealed Ni(100) films

The sub-ML coverage of C60 on post-annealed Ni(100) films wasanalyzed using STM. In Fig. 1(a), a constant–current STM image of Ni(thickness: 7.6±0.6 ML) grown at room temperature is shown. Thesurface is relatively roughwith multiple islands and 4–5 atomic layersexposed at the film surface, resulting from a multilayer growth mode.Fig. 1(b) shows the same film after annealing at 455 K. The Ni filmmorphology has becomemuch smoother with only two distinct levels

Fig. 1. (a) Topograph of pre-annealed 7.6 ML Ni film grown on Cu(100) substrate at room tematomically-flat upper and lower terraces. Small Ni islands remain atop the upper terrace.

present; the local coverage is now between 7 ML and 8 ML separatedby a monatomic step. The upper Ni film terrace forms a network ofcoalesced ridges which are roughly oriented along [110]-equivalentdirections and up to 400 Å wide. Additionally, small Ni islands with adiameter of about 50 Å or less are observed on the upper Ni terrace. Aswe will see below, these small islands will act as nucleation sites foradsorbed C60. Auger spectra confirmed the absence of any S, O, or C-peak contamination on the Ni surface and LEED patterns of the post-annealed Ni film showed bright well-defined spots with a lowbackground indicating the high quality of the Ni(100) film.

After theNi filmswere annealed, C60was deposited onto the surface.Throughout this study, fullmonolayer (ML) coverage of C60 is defined asthe presence of 1.23 C60 molecules per square nanometer. Approxi-mately 0.03 ML C60 were deposited onto a 7.6 ML Ni film, shown inFig. 2. The C60 molecules preferentially nucleate and cluster at Ni stepedges, on the Ni terraces, and also around small Ni islands. A study byPark et al. investigating the interactions of C60 molecules with small Niclusters showed evidence that C60 adsorption behavior varies with

perature. (b) Ni film after a 20-minute anneal at 455 K. The annealed Ni film surface has

74 M. Cummings et al. / Surface Science 605 (2011) 72–80

cluster-size; and this size-dependence was related to the structure ofthe Ni cluster and the availability of binding sites and electronic chargeat the cluster surface [8]. So, the presence of small Ni islands on the Ni(100) film surfacemay also influence C60molecule nucleation. No directevidence of an adsorption-dependencewith island sizewas observed inthis experimental study. However, controlled Ni film growth studiesinvestigating C60-adsorption behavior with Ni island size may provideadditional insight.Manydifferentmolecular cluster sizeswere found onthe Ni surface. For example, an ordered heptamer that is a cluster ofseven C60molecules can be seen in Fig. 2 (inset). Due to the smaller areaoccupied by the lower Ni terrace (local coverage 7 ML) and the absenceof Ni clusters on this terrace, C60 coverage and clustering are reducedexcept along the Ni step edges. Across the top Ni film terrace (localcoverage 8 ML), C60 coverage is fairly uniform with a high degree ofmolecular clustering. Such clustering across the upper terrace isexpected, due to the much stronger covalent bond interaction betweenthe C60 molecule and the Ni surface compared with that of noble metalsurfaces Au and Ag, limiting the mobility of C60 at the surface [9–11].

Beyond the heptamer, larger clusters of C60, both ordered anddisordered, can be observed on the Ni(100) film surface. Some orderedclusters are shown in the insets of Fig. 2. The observation of short-rangequasi-hexagonal molecular ordering in the heptamer (bottom inset) issurprising at such a relatively low-temperature (300 K), given the highbinding energy between C60 and Ni at the surface. In this sense, the Ni(100) film surface differs from that of the bulk Ni(110) surface whereC60 ordering only occurs at elevated temperatures (≥540 K). This maybe attributed to the more open character of the Ni fcc(100) surface as

Fig. 3. Topograph of (a) 0.03 ML C60 on Ni(100) film surface and after depositing (b) 0.5 ML, (at step edges. C60 molecules also serve as nucleation sites. Box in (a) encloses a section on

compared to the Ni fcc(110) surface, which has a higher corrugation(rougher) making C60 diffusion more difficult (and anisotropic) causedby the increased binding energy or bonding between the electronicorbitals of the C60 and Ni films. The intermolecular interactions of C60lead to close-packed ordering at metallic surfaces. [12]

The average height of C60 molecules measured above the Ni filmterrace is 7.2±0.2 Å. Although the height measured by STM is smallerthan the actual cage diameter of C60 (10.18 Å), it is in agreement withsimilar measurements taken on other metallic surfaces [13–16]. Thereduced apparent height of C60 is a consequence of the relatively lowlocal-DOS at a given energy above themolecule linked to the coupling ofC60 molecular orbitals to substrate states near the EF surface [6,17,18].

At room temperature the C60molecules and clusters showed limitedto no mobility on the Ni(100) film surface. This was confirmed byimaging the same area on the Ni film surface over multiple growthcycles. While some molecular clusters of C60 (dimer and trimer) wereobserved to be unstable during imaging (clusters rotate about a centralaxis normal to the film surface), for a period of over 45 min no long-range mass transport of C60 monomers or clusters was observed acrossthe Ni film surface.

3.2. Ni island growth with 0.03 (low) C60 coverage

To understand how the presence of molecular C60 impacts filmgrowth on the Ni(100) surface, we investigated the growth of Ni onannealed Ni(100) films pre-covered with molecular C60. After eachsubsequent Ni deposition, successive STM images were taken at the

c) 1.5 ML, and (d) 4.5 ML Ni onto the film surface. Ni nucleates on upper/lower terraces,the surface where Ni growth is followed (shown in Fig. 4 below).

75M. Cummings et al. / Surface Science 605 (2011) 72–80

same location. The thickness ofNimonolayer deposits at the surfacewasvaried from 0.5 ML up to 7.5 ML.

Fig. 3 follows the progression of Ni growth at the surface after eachsubsequent deposition. In Fig. 3(a) and (b), topographic STM imagesshow the same location before and after 0.5 ML Ni is deposited onto theNi(100) surface, respectively. Upon the deposition of 0.5 ML Ni, theaverage Ni island size is observed to be≈4.2 nm (≈18 nm2) and the Niislands are beginning to coalesce (Fig. 3b). The Ni islands nucleate onboth the upper and lower Ni(100) terraces. The presence of step edgesor C60 has no obvious effect on the multilayered growth of Ni. Theobservation of Ni islands nucleating away from C60 molecules at thesurface indicates thatNi has a high energy diffusion barrier and suggeststhat the diffusion rate is too low for Ni adatoms to adequately sampleenough sites; and this limits the number of Ni–C60 encounters thatoccur.

In regions on the Ni(100) terrace where no C60 is present, the Niislands nucleate forming a square shape, consistent with the previousgrowth studies of Ni/Cu(100) systems and also consistent with thestudies of homoepitaxial systems of Ni and Fe(100) [7,19,20]. Thesquare-shapedNi islands are an indication of the ability of the impingingadatoms to preferentially diffuse along the [011] step edge direction(closed-packed). In regions on the Ni(100) film surface where C60 ispresent, Ni adatoms treat the C60 molecules as alternative nucleation

Fig. 4. Topograph of (a) C60 on Ni(100) film surface and after depositing (b) 0.5 ML, (c) 1.5 Msame location following Ni film growth with C60 at the surface. C60 imagery in linescan obtaiasp?ID=217&othername=Bucky%20Ball].

sites on the Ni film surface, leading to Ni island growth around themolecular clusters. These Ni islands encapsulate the C60 clusters andtend to take-on the general shape of the cluster being surrounded,suggesting that C60molecules do have a localized influence on Ni islandgrowth at the surface.

At 1.5 ML Ni, an additional monolayer Ni has grown on the Ni filmsurface with another partial monolayer exposed at the surface, asevidenced by the small islands on the larger terraces [Fig. 3(c)]. Atapproximately 2.5 MLNi, only about 10% (or ~0.8 Å) of the C60moleculeremains visible above the Ni film surface; and after the deposition of4.5 MLNi, noC60 remains exposedon the surface [Fig. 3(d)]. Still asmoreNi is deposited at the surface, the growing Ni film maintains crystalline(epitaxial) growth, essentially uninterrupted by the presence of C60.

The region within the white box in Fig. 3(a) was tracked during Nigrowthas shown in Fig. 4 alongwith its corresponding line section. Fig. 4(a) shows a topographic STM image of a region on the Ni film surfacewhere C60molecules have adsorbed at the bottom edge of a Ni step; thedashed line represents the path taken for the corresponding line section,shown in Fig. 4(e). As more Ni is deposited onto the film surface (0.5to 4.5 ML), the Ni islands nucleate and grow on both the upper andlower Ni(100) film terraces. As shown in the topographic images inFig. 4(b)–(d) and the corresponding line sections in Fig. 4(e), the Niislands also grow around the C60 molecules, trapping the molecules

L, and (d) 4.5 ML of Ni deposited onto the film surface. (e) Sequential line section of thened courtesy of [“Bucky Ball@3Dchem.com,” http://www.3dchem.com/moremolecules.

76 M. Cummings et al. / Surface Science 605 (2011) 72–80

along the lower edge of theNi step. After 4.5 MLNi have beendeposited,the C60 molecules are completely encapsulated beneath the surface ofNi.

At even higher C60 coverages (up to 0.2 ML), the Ni film continuesto grow with a remarkably high degree of crystallinity as determinedby the regularly shaped island whose edges are roughly orientedalong low-index crystalline directions. Indeed, we also studied 0.2 MLC60 coverage with the film grown slightly above room temperature(~350 K). Fig. 5 shows the Ni film morphology after 3.8 ML Ni weredeposited onto the annealed Ni film surface. At 3.8 ML Ni, there is noevidence of C60 at the surface similar to what was observed at lowerC60 coverages. The atomically-flat Ni film terraces also appear widerand more coalesced (Fig. 5), as compared with the lower C60

coverages due to the slightly elevated substrate growth temperature.

Fig. 6. Auger spectra shows shift and attenuation in C60 C-peak as 1 to 3 ML Ni aredeposited onto annealed Ni(100) film surface pre-covered with C60. The C-peakposition for each individual spectra is labeled. The Ni(848 eV) and Cu(920 eV) peaks areidentified by black andwhite arrows, respectively. The Ni- and Cu-peaks of each spectraare offset by 25 eV only for clarity.

3.3. Auger spectral analysis of Ni growth

In order to confirm that the C60 molecules are being buried (asopposed to diffusing on top of the film surface), we have performed acoverage-dependent Auger electron spectroscopy (AES) study. TheAES study was performed with 4-grid Auger optics. Successive MLs ofNi were deposited onto annealed Ni(100) films pre-covered with0.2 ML C60, under very similar conditions as in the Experimentalprocedures section, (base pressure 7×10−10 mbar). C60 film coveragewas determined by STM. Auger electron spectra were taken after eachsuccessive ML of Ni was deposited onto the film surface and comparedwith the spectra taken from an annealed Ni(100) film, pre-coveredwith 0.2 ML C60. Each of the AES spectra is plotted in Fig. 6. The peaksof interest identified in this figure are the C (272 eV), Ni (848 eV), andCu (920 eV) peaks. Changes observed in the C60 C-peak height andposition due to Ni deposits are reported and discussed below alongwith the associated changes in the Ni-to-Cu peak ratio. The peakheights in this study were determined after background subtraction.

Fig. 5. Topograph (200 Å×400 Å) of 3.8 ML Ni grown annealed Ni(100) film pre-covered with 0.2 ML C60. Epitaxial growth of Ni film continues at higher C60 coverages.

The AES plot (Fig. 6) shows a clear shift in the C60 C-peak position— or AES chemical shift. Such chemical shifts are attributed to changesin the chemical environment or the interactions of nearest neighboratoms. As Ni is deposited onto the film surface, a reduction in the C-peak height and changes in the lineshape are also observed. Theattenuation of the C-peak is consistent with the C60 molecules slowlybeing buried below the Ni surface during deposition; due to therelatively short mean-free path of Auger electrons, even the first fewatomic layers of Ni lead to a significant reduction of the C signal [21].With increased Ni coverage at the surface, the Cu peak is also observedto reduce relative to the Ni-peak (Ni-to-Cu peak ratio increases), asexpected. In Table 1, shifts (attenuation) observed in the C-peakposition (height) are reported along with associated changes in theNi-to-Cu peak ratios. An explanation of the changes observed in theC60 C-peak lineshape is less clear without more detailed theoretical/experimental studies. For molecules, it is well-understood thatchanges in the AES (and XAS) lineshape can generally be attributedto changes in the molecular bond orientations at the substrate surfaceand many-body processes (unresolved vibrational excitations accom-panying AES transitions within the molecule), which are the result oflong-lived final states [22–24]. The AES results presented hereindicate that the C60 molecules maintain their chemical integrityduring the Ni growth, in good agreement with the literature whichreports that the decomposition of C60 on Ni(110) surface occurs onlyat elevated substrate temperatures (≥750 K), well-above the tem-peratures used in this study [25].

3.4. Co Film growth and morphology on Ru(0001) substrate

Fig. 7(a) and (b) shows Co films grown on the Ru(0001) substratesurface at room temperature before and after annealing, respectively.

Table 1C60 C-peak position and height (normalized) after 0.2 ML C60 deposited onto annealedNi(100) film surface and after each subsequent Ni deposition. Ni-to-Cu peak ratioincreases with each Ni deposition as expected.

Layer Peak position Height Ni/Cu ratio

C60–Ni 259.8 5.77 0.8851 ML Ni 263.5 4.85 1.3372 ML Ni 265.0 4.49 1.4473 ML Ni 266.7 3.57 1.598

77M. Cummings et al. / Surface Science 605 (2011) 72–80

The total Cofilm thickness deposited is 1.84 ML. Pre- and post-annealedSTM imageswere taken at different locations on the film surface. The Cofilms grown at room temperature on the Ru(0001) substrate display avery roughmultilayer surfacemorphology, as shown in Fig. 7(a), similarto that of the pre-annealed Ni films. However after the Co films areannealed, the morphology changes significantly (Fig. 7b). The resultantCo film surface is relatively complex and has atomically-flat terraces aswide as 500 Å.

In thepost-annealedCofilm, local coverages between1 MLand3 MLcan be found on the film surface, accompanied by regions on the surfacewhere the Ru(0001) substrate is exposed. In the first Comonolayer, wefind co-existing smooth and quasi-periodically corrugated regions witha lateral periodicity of about 35±1 Å. These regions have differentapparent heights of 1.82 Å and 2.34 Å above the Ru(0001) substrateterrace, respectively, and are labeled unreconstructed (u) and recon-structed (r), in Fig. 7(b). We speculate that the reconstructed region isequivalent to the strain-driven formation of a dislocation networkreported in a LEED study by Gabaly et al. performed under slightlydifferent conditions, i.e. on films grown at an elevated substratetemperature of 523 K [26]. In this study evidence was found that filmstrain is relieved in the first Comonolayer only after the appearance of atemporary phase, which consists of a network of misfit dislocations.Gabaly et al. also observe that the phase change occurs only afterapproximately 0.9 ML Co have been deposited onto the Ru(0001)substrate surface at 523 K. It is described as a “metastable” phase(dislocation network), which disappears within a minute. In contrast,the quasi-periodic corrugation pattern appearing in our STM imagesobtained at room temperature was stable for hours after the annealingprocess.

The existence of a stable quasi-periodically reconstructed (r)region in the first Co monolayer differs from the findings of Lundgrenet al. for Co films grown on Pt(111) [27]. In their study, dislocationsin the first Co monolayer are lifted by the growth of the second Colayer. However as evidenced in Fig. 7(b), the areas of reconstructionobserved in the first Co monolayer are stable and adjacent to regionsshowing no reconstruction. The reconstructed regions display a moirépattern; and the presence of adjacent (r) and (u) regions within thefirst (and second) Co filmmonolayer suggests (i) that the partial-layerreconstruction (dislocation network) is not fully lifted by the second(or third) Co monolayer, respectively, and (ii) that the partially-reconstructed regions also play a role in relieving film strain energywithin these monolayers. The above differences cited with respect toother experiments are likely related to the preparation method of ourCo films.

Fig. 7. (a) Topograph of pre-annealed Co(0001) film surface (arrows denote terraces of undfilm surface. Numbers (1–3) denote the CoML; zero (0) denotes a region where the Ru substrin (b). The dashed lines denote terraces of the underlying Ru substrate. Unreconstructed andand fcc-stacked regions on the 2nd and 3rd Co ML are labeled (h) and (f), respectively (I=

With the exception of some tiny unreconstructed patches in thesecond monolayer, all higher coverages were found to be fullyreconstructed and display a moiré pattern with a periodicity of 35±1 Å. Although the local coverage across the Co film surface lookslargely homogeneous in the topographic (constant–current) images,significant contrasts could be identified in the differential conduc-tance (dI/dV) maps measured at U=−0.4 V [Fig. 7(c)]. Closeinspection of the dI/dV map reveals that for both local coverages oftwo and three monolayers, two discrete contrasts can be observed.These contrasts are caused by changes in the Co film's electronicstructure. We believe that this variation in the local electronicstructure originates from differently stacked regions. It is well-known that variations of the stacking sequence in quasi-hexagonalsurfaces (hcp vs. fcc) leads to energetic shifts of surface states andsurface state resonanceswith d3z2− r2 symmetry near the Γ-point. [28–30] Closer analysis reveals that out of the five 3rdmonolayer islands inFig. 7(b), four were found to exhibit a low differential conductivity(labeled 3h) and only one exhibits a high differential conductivity(labeled 3f). A similar observation is made for the 2nd monolayer inFig. 7(c), where histogram analysis of the Co film surface (not shownhere) reveals that the hcp (h) region represents 78% and the fcc (f)region represents 22% of the second Co monolayer. Since both the Rusubstrate as well as the Co film are hexagonal in nature, we interpret(h) and (f) regions in the second and thirdmonolayers as hcp- and fcc-stacked Co domains, respectively, i.e. the hexagonal stacking is morelikely. Such stacking faults are quite common in hetero-epitaxialsystems where both constituents exhibit a bulk hcp crystallinestructure. The resultant moiré patterns are the product of atomicsurface distortions induced by the large lattice-mismatch (7%)between the Co film and the Ru (0001) substrate. The atomicdistortions relieve the stress built-up within the Co film and aremediated by the formation of a network of dislocation on the filmsurface [31]. The changes observed in the moiré pattern across the Cofilm surface have a direct impact on the C60 interactions with the filmand will be discussed in more detail below.

3.5. Sub-ML C60 coverage on annealed Co(0001) films

After the Co(0001) films were annealed, sub-monolayers of C60were deposited onto the film surface. Fig. 8 shows a Co film covered by0.04 ML C60: while the evaporated molecules impinge onto the entireCo surface with equal probability, it is evident that the C60 distributionacross the surface is non-uniform. We observe that the non-uniform

erlying Ru substrate). (b) Topograph at different locations on post-annealed Co(0001)ate is exposed. (c) Differential conductance dI/dVmaps of post-annealed Co film shownreconstructed regions on the 1st Co ML are denoted (u) and (r), respectively. The hcp-2 nA, V=−400 mV).

Fig. 8. Topograph of 0.04 ML C60 deposited onto an annealed Co(0001) film surfaceshows C60 distribution is non-uniform across the film surface.

78 M. Cummings et al. / Surface Science 605 (2011) 72–80

coverage is a direct consequence of variations in the local filmproperties of Co [32].

In the upper left corner of Fig. 8, a large quasi-hexagonal cluster ofC60 sits atop a region on the surface where the Ru(0001) substrate isexposed (indicated by a white arrow); small molecular chains of C60have also formed along the bottom step edge of the 1st Co monolayer.The C60 molecules, adsorbed on both the Ru substrate surface and theCo film surface, exhibit a height of 6.8±0.1 Å, i.e. there is no distin-guishable height difference.

A relatively high number of C60 monomers as well as manyrelatively small C60 clusters (3–5 molecules) are present on the firstCo monolayer. In contrast, there is no C60 in the unreconstructed

Fig. 9. Topograph of Co film growth tracked in the same location before and after 0.23 MLindicate the position of corresponding line sections from the topographs (b). C60 imagery inmoremolecules.asp?ID=217&othername=Bucky%20Ball].

regions of the first Co monolayer. In Fig. 8, individual C60 monomersand small C60 clusters are also observed on the reconstructed regionsof the second Co monolayer. However unlike the first monolayer, thereconstructed regions of the second Co monolayer exhibit differentstackings, i.e. hcp and fcc. Very few C60 molecules seem to adsorb onthe hexagonally stacked (2h) surface, likely due to highmobility of C60in these regions. In comparison, more C60 monomers and clusters canbe found on fcc-stacked (2f) regions, where the dislocations impedethe mobility of C60 molecules.

On top of the third Co monolayer (labeled 3h) in Fig. 8, large quasi-hexagonal clusters of C60 (10 molecules or more) are observed onlyalong the edges of the top terrace and adsorbed along the bottom of thethird monolayer Co step (black arrows in figure). In contrast, no C60molecules are observed at the center of these third monolayer Coterraces. A high number of C60 molecules are also adsorbed along thebottom step edge of these terraces, indicating the high C60 mobility inthese regions. The high mobility of C60 in these (3h) regions correlateswith the increased step decoration observed along the 3rd Co mono-layer step.

The effect of Co (stacking/reconstruction) on C60 mobility is alsorevealed by quantifying the percent monolayer coverage of C60 oneach Co layer at the surface. In the STM topograph of Fig. 8, Cooccupies approximately 90% of the surface area, and the area coveredby each of the three Co monolayers is roughly equivalent (1:1:1). Theremaining portion (10%) of the surface is the exposed Ru substrate. Asdetermined by STM, the C60 coverage on the Ru substrate surface is0.047 ML. While the analysis shows that the C60 coverage on Comonolayers 1 and 2 is 0.054 ML, for the third monolayer the C60coverage is determined to be 0.027 ML, nearly half the amount foundon the Ru substrate surface and lower Co monolayers. The disparityobserved in the local coverage of C60 can only be explained bydiffusion of the C60 molecule across the 3rdmonolayer step of Co fromthe upper terrace to a second ML lower terrace. In comparison, themobility of C60 seems to be limited on the first and second Co

Co is deposited onto the film surface (a). The dashed lines in STM topographs (above)linescan obtained courtesy of [“Bucky Ball@ 3Dchem.com,” http://www.3dchem.com/

79M. Cummings et al. / Surface Science 605 (2011) 72–80

monolayers, such that step diffusion across these monolayers isnegligible compared with the third Co monolayer.

3.6. Co island growth with low C60 coverage

To determine how Co growth will proceed with C60 present on thefilm surface, additional Co (0.2 ML–1.6 ML) was deposited at roomtemperature onto the annealed Co film with 0.04 ML C60 present atthe surface. After each subsequent Co deposition, successive STMimages were taken at the same location. The STM image in Fig. 9(a)shows C60 molecules adsorbed onto a room temperature grown andannealed 1.84 ML Co film. As described in the previous section, thehigh mobility C60 on the second and third Co monolayer leads to thestep edge decoration.

Indeed, the 3rd monolayer island in the top-right corner of Fig. 9(a)is surrounded by five C60 molecules. A line section crossing two C60molecules and a third monolayer Co-island is shown in the lower panelof Fig. 9(a). The apparent height of the C60 molecules in the profileindicates that it is adsorbed on the lower second layer terrace.

On top of this surface, an additional 0.23 ML Co was deposited atroom temperature. After the deposition, the same locationwas imaged[Fig. 9(b)]. The comparison of the STM images (upper panels) andcorresponding line section (lower panels) before and after growthindicates two processes: (i) the step edges of existing islands aredecorated with Co. The data indicate that the C60 molecules that werepreviously adsorbed at the step edges, retain their original position.As a consequence, the C60 molecules are further away from the stepedges after each successive deposition (some examples highlightedby black arrows). (ii) Some C60 molecules act as nucleation sites forthe formation of new Co islands in the second and third monolayers(white arrows).

However unlike Ni, Co atoms impinging on the film surface nevernucleate on the bare Co terrace. Co island nucleation is only observedto precede from the step edges, via step flow, or around C60 moleculesadsorbed on top of the film terrace. These C60 molecules behave likeartificial nucleation sites for the Co diffusing across the film surface;and as the islands grow and extend away from the C60 molecule,{100}, {111} microfacets along the edges of the growing island beginto develop. The absence of Co island formations away from molecularC60 or step edges indicates that the strength of the Co–C60 bondinteraction (binding energy) is stronger and more stable than the Co–Co bond atop the film terrace.

In Fig. 10, an additional 0.5 ML Co has been deposited onto the filmsurface. A fourth monolayer of Co has nucleated and continues togrow around C60 clusters. The Co islands extend approximately3.4 nm beyond the edge of the C60 cluster forming smooth atomically-flat islands with {100}, {111} microfacets at the island edges. Thehigher adatom edge diffusion rate (lower diffusion barrier) along the{111} step edge microfacets, as reported by Jacobsen et al., leads to

Fig. 10. Topograph of large Co islands with {111}, {100} step edgemicrofacets nucleatedaround C60 clusters on a three ML-thick Co film after 0.5 ML Co deposited. Arrows pointto 4th ML on Co surface. Circled region on topograph points to a triangular dislocationnetwork identified on the 3rd ML of Co film surface.

the anisotropic hexagonal island shape observed [30,33,34]. Incontrast with our earlier observations, in this case the C60 moleculeshave nucleated forming ordered clusters at the center of a thirdmonolayer (moiré) terrace. In this figure, dislocation networks areobserved on the 3rd monolayer Co terrace. These dislocationnetworks (sinks) are atypical on the third monolayer Co surfaceand may play a role in the formation of the observed C60 clusterformation. More studies are needed to understand the role that thesedislocation networks play in the formation of large C60 clusters at thesurface.

After the deposition of 1.6 ML Co, the Co islands begin to coalesceand form terraces 160 Å in width (shown in Fig. 11). These films weregrown at room temperature, similar to the pre-annealed Co filmsdiscussed earlier (Fig. 8). However unlike the pre-annealed Co films,the Co films grown here have the same morphology as the underlyingannealed Co surface. The film growth is crystalline, similar to thepreviously described Ni growth. The terraces remain atomically-flatwith a majority of the coalesced terraces displaying a moiré-reconstructed surface. The C60 molecules present on the surfacecontinue to act as pseudo-nucleation sites for the diffusing Co atomsat the surface, and the molecules combine rather coherently withincrystalline network of the Co film. Line sections (Fig. 10b) show thatthe height of C60 molecules above the film surface has decreased afterthe deposition of 1.6 ML Co. Many of the C60 molecules are stillsurrounded by 1 ML Co. However, a fewmolecules at the surface havea second monolayer of Co nucleated around them. A third monolayerof Co nucleated around the C60 was not observed for the 1.6 ML Cocoverage.

Fig. 11. (a) Topograph of Co film surface after 1.6 ML Co deposited. Circled regions ontopograph point to C60 molecules around which one ML of Co has nucleated. Trianglespoint to molecules around which two ML of Co have nucleated. Arrows point tomolecules on the surface where line sections were taken as shown in (b). The linesections follow the direction of thewhite arrows and show twoML Co nucleated aroundC60 (black solid line) and one ML Co nucleated around C60 (red dotted line).

80 M. Cummings et al. / Surface Science 605 (2011) 72–80

4. Conclusions

The effect of C60 on film growth andmorphologywas studied usingSTM for both the Ni/Cu(100) and the Co/Ru(0001) systems. The C60

molecules deposited onto the Ni(100) film surface were observed tobe fairly immobile and were uniformly distributed across the Ni filmsurface. On Co(0001) film surfaces, we observe regions on the surfacewhere C60 mobility is highly mobile and regions where mobility isseverely limited, resulting in a non-uniform distribution of C60 mole-cules across the Co film surface. Despite the presence of C60 moleculeson the Ni surface, the C60 shows no obvious effect on Ni film growth, asa high Ni diffusion barrier is shown to limit the occurrence Ni–C60encounters at the Ni surface. AES results indicate that the C60 mole-cules maintain their chemical integrity during Ni growth. For Co films,Co islands only nucleate and grow from the step edges and locallyaround the C60 molecules which serve as pseudo-nucleation sites. Thestrength of the Co–C60 bond interaction appears stronger than the Co–Co bond on the Co film terrace. The Co films continue to grow ascrystalline films with well-defined reconstructed (moiré) terraces.These findings demonstrate that the nature of the C60–metallic bondat the surface influences thin film systems, but does not inhibitcontinued epitaxial growth.

Acknowledgements

We acknowledge the National Science Foundation (NSF Cooper-ative Agreement Number HRD-0450363) for their financial support,the Welch Foundation (grant no. C-1494), and the use of the YaleCenter for Research on Interface Structures and Phenomena facilities(NSF grant number DMR-0520495). Use of the Center for NanoscaleMaterials was supported by the U. S. Department of Energy, Office ofScience, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.

References

[1] Abhay N. Pasupathy, et al., The Kondo effect in the presence of ferromagnetism,Science 306 (5693) (October 1, 2004) 86.

[2] Lam H. Yu, Douglas Natelson, The Kondo effect in C60 single-molecule transistors,Nano Letters 4 (1) (January 1, 2004) 79.

[3] Lingyi A. Zheng, et al., Formation of nanomagnetic thin films by dispersedfullerenes, Applied Physics Letters 77 (20) (2000) 3242.

[4] Lingyi A. Zheng, Enrique V. Barrera, Robert D. Shull, Formation and stabilization ofnanosize grains in ferromagnetic thin films by dispersed C60, Journal of AppliedPhysics 92 (1) (2002) 523.

[5] Lingyi A. Zheng, Enrique V. Barrera, Robert D. Shull, Magnetic properties of theCo–C60 and Fe–C60 nanocrystalline magnetic thin films, Journal of Applied Physics97 (9) (2005) 094309.

[6] Chao-Cheng Kaun, Ryan Jorn, Tamar Seideman, Spontaneous oscillation of currentin fullerene molecular junctions, Physical Review B 74 (4) (July 13, 2006) 045415.

[7] J. Shen, J. Giergiel, J. Kirschner, Growth and morphology of Ni/Cu(100) ultrathinfilms: an in situ study using scanning tunneling microscopy, Physical Review B 52(11) (1995) 8454.

[8] E.K. Parks, et al., Adsorption of C60 on nickel clusters at high temperature, PhysicalReview B 59 (20) (May 15, 1999) 13431.

[9] T. Quast, et al., Excited-state photoemission with combined laser/synchrotronpulse excitation from C60 chemisorbed on Ni(110), Journal of Applied Physics 83(3) (1998) 1642.

[10] Eric I. Altman, Richard J. Colton, The interaction of C60 with noble metal surfaces,Surface Science 295 (1–2) (September 20, 1993) 13.

[11] Eric I. Altman, Richard J. Colton, Determination of the orientation of C60 adsorbedon Au(111) and Ag(111), Physical Review B 48 (24) (December 15, 1993) 18244.

[12] M.R.C. Hunt, et al., Charge transfer and structure in C60 adsorption on metalsurfaces, Physical Review B 51 (15) (April 15, 1995) 10039.

[13] Tomihiro Hashizume, et al., Intramolecular structures of C60 molecules adsorbed onthe Cu(111)–(1×1) surface, Physical Review Letters 71 (18) (November 1, 1993)2959.

[14] M. Abel, et al., Scanning tunneling microscopy and X-ray photoelectron diffractioninvestigation of C60 films on Cu(100), Physical Review B 67 (24) (June 9, 2003)245407.

[15] J. Weckesser, et al., Binding and ordering of C60 on Pd(110): investigations at thelocal andmesoscopic scale, The Journal of Chemical Physics 115 (19) (2001) 9001.

[16] M.Grobis, X. Lu,M.F. Crommie, Local electronic properties of amolecularmonolayer:C60 on Ag(001), Physical Review B 66 (16) (October 30, 2002) 161408.

[17] P.W. Murray, et al., Growth of C60 on Cu(110) and Ni(110) surfaces: C60-inducedinterfacial roughening, Physical Review B 55 (15) (April 15, 1997) 9360.

[18] Philippe Sautet, Images of adsorbates with the scanning tunneling microscope:theoretical approaches to the contrast mechanism, Chemical Reviews 97 (4) (June1, 1997) 1097.

[19] E. Kopatzki, et al., Homoepitaxial growth on Ni(100) and its modification by apreadsorbed oxygen adlayer, Surface Science 284 (1–2) (March 10, 1993) 154.

[20] M.C. Bartelt, J.W. Evans, Transition to multilayer kinetic roughening for metal(100) homoepitaxy, Physical Review Letters 75 (23) (December 4, 1995) 4250.

[21] M.P. Seah, W.A. Dench, Quantitative electron spectroscopy of surfaces: a standarddata base for electron inelastic mean free paths in solids, Surface and InterfaceAnalysis 1 (1) (1979) 2.

[22] A. Goldoni, et al., C70 adsorbed on Cu(111): metallic character and molecularorientation, The Journal of Chemical Physics 116 (17) (2002) 7685.

[23] Joachim Stöhr, NEXAFS Spectroscopy, Springer, 1992.[24] T.R. Ohno, et al., C60 bonding and energy-level alignment on metal and

semiconductor surfaces, Physical Review B 44 (24) (December 15, 1991) 13747.[25] V. Saltas, C.A. Papageorgopoulos, Adsorption and decomposition of C60 on Ni(1 1 0)

surfaces, Surface Science 488 (1–2) (August 1, 2001) 23.[26] Farid El. Gabaly, et al., Structure and morphology of ultrathinCo/Ru(0001) films,

New Journal of Physics 9 (3) (3, 2007) 80.[27] E. Lundgren, et al., Thin films of Co on Pt(111): strain relaxation and growth,

Physical Review B 62 (4) (July 15, 2000) 2843.[28] L. Diekhoner, et al., Surface states of cobalt nanoislands on Cu(111), Physical

Review Letters 90 (23) (June 10, 2003) 236801.[29] O. Pietzsch, et al., Spin-polarized scanning tunneling spectroscopy of nanoscale

cobalt islands on Cu(111), Physical Review Letters 92 (5) (February 6, 2004)057202.

[30] J. Wiebe, et al., Scanning tunneling spectroscopy on Co(0001): spectroscopicsignature of stacking faults and dislocation lines, Physical Review B 70 (3) (July 8,2004) 035404.

[31] Raghani Pushpa, Shobhana Narasimhan, Reconstruction of Pt(111) and domainpatterns on close-packed metal surfaces, Physical Review B 67 (20) (May 23,2003) 205418.

[32] Jakob Bork, et al., Potential energy landscape of metallic Moiré patterns, NewJournal of Physics 11 (11) (11, 2009) 113051.

[33] Joachim Jacobsen, et al., Island shape-induced transition from 2d to 3d growth forPt/Pt(111), Physical Review Letters 74 (12) (March 20, 1995) 2295.

[34] Staffan Ovesson, Alexander Bogicevic, Bengt I. Lundqvist, Origin of compacttriangular islands in metal-on-metal growth, Physical Review Letters 83 (13)(1999) 2608.