PHYSICAL REVIEW B 86, 075409 (2012)

Electronic properties probed by scanning tunneling spectroscopy: From isolated gold nanocrystalto well-defined supracrystals

Peng Yang,1 Imad Arfaoui,1 Tristan Cren,2 Nicolas Goubet,1 and Marie-Paule Pileni1,*

1Laboratoire des Materiaux Mesoscopiques et Nanometriques (LM2N), UMR CNRS 7070, Universite Pierre et Marie Curie,bat F, BP 52, 4 Place Jussieu, 75252 Paris Cedex 05, France

2Institut des NanoSciences de Paris (INSP), UMR CNRS 7588, Universite Pierre et Marie Curie,4 Place Jussieu, 75252 Paris Cedex 05, France

(Received 8 June 2012; published 3 August 2012)

Scanning tunneling microscopy and spectroscopy at 5 K have been used to determine the electronic propertiesof 7-nm dodecanethiol-passivated Au nanocrystals in three different configurations: isolated nanocrystal, self-organized thin films (few nanocrystal layers), and large three-dimensional well-defined thick films (over 30nanocrystal layers) called supracrystals. The electronic properties of both thin and thick well-ordered supracrystalsare analyzed in scanning tunneling spectroscopy geometry through dI/dV curves and conductance mapping atdifferent bias voltages. The single particles exhibit a typical dI/dV curve with a Coulomb gap of ∼360 meV anda Coulomb staircase. The dI/dV curve of the thin supracrystals presents a Coulomb blockade feature ∼100 meVnarrower in width than that of the single nanocrystal but without well-defined staircase. On the contrary, thethick supracrystals exhibit a dI/dV curve showing a large Coulomb gap with a Coulomb-staircase-like structure.Generally, the conductance mapping is found to be very homogeneous for both supracrystals. Nevertheless,for some bias voltages, inhomogeneities across individual nanocrystals appear. Additionally, some of theseinhomogeneities seem to be related to the supracrystal surface morphology. Finally, these slight variations in theconductance mapping across individual nanocrystals embedded in the supracrystal are discussed in terms of highdegree of nanocrystal ordering, low nanocrystal size distribution, and nanocrystal crystallinity.

DOI: 10.1103/PhysRevB.86.075409 PACS number(s): 73.21.Cd, 73.23.−b, 73.23.Hk

I. INTRODUCTION

The control of the assembly processes of nano-objectsinto highly ordered structures still remains a challengingproblem for nanoscience and nanotechnology.1,2 Systemsof nanoparticles or biomolecules, assembled into periodicstructures, are found in many natural systems, such as atomicand molecular solids, opals, bacterial colonies, etc. The abilityto assemble “artificial atoms,” such as nanocrystals (NCs),into ordered structures is expected to create a novel class of“artificial solids” such as supracrystals.3,4 Periodic structurescan be obtained in the form of films or perfectly facetedcolloidal crystals.5,6 Close-packed NC solids connected withlinker molecules are also exciting new materials for theirpotentially novel electronic properties.7,8 The ordered NCarrays exhibit a wide range of fascinating architectures withemerging physical properties different from those of classicalcrystalline solids.5,9–13 In order to manipulate them andoptimize their elaboration, one has to understand the propertiesof these materials. In particular, the control of the packinguniformity and interparticle distances has led to importantadvances in our understanding of fundamental charge transportphenomena.2,14–16

Transport measurements and electronic properties ofmore or less ordered NC assemblies have been extensivelystudied.14–22 The fact remains that these studies involvedsingle NC, monolayer, or few NC layers. Moreover, mostof these studies used contact electrode measurements inplanar geometry. Here, we illustrate the evolution of electronicproperties from isolated NCs to quasi-two-dimensional (2D)supracrystal (thin film of a few NC layers, thickness <30 nm),and in turn to a three-dimensional (3D) face centered cubic(fcc) supracrystal (over 30 NC layers, thickness of ∼300 nm).

For this purpose, we use scanning tunneling microscopy andspectroscopy (STM/STS), which allows us to measure the elec-tronic signatures of individual NC embedded at the surface ofsupracrystals while, in addition, to acquire information on theoverall electronic behavior of the thick and thin supracrystals.Note that in our previous paper, the electronic properties ofmicrometer-thick supracrystals are briefly reported.23

II. EXPERIMENTAL SECTION

The Au NCs are synthesized by using an organometallicroute described in the previous paper.24 They are coatedby dodecanethiol molecules (C12H25SH) and dispersed intoluene. The colloidal solution is maintained under saturatedtoluene vapor. After 7 days, a film appears at the air-tolueneinterface.25 This film is withdrawn by using a tungsten ringfrom the solution and transferred ex situ directly on highlyoriented pyrolytic graphite (HOPG) substrate. The samplesare then introduced into the STM chamber without any furthertreatments. STM and STS measurements were performed witha commercially available low-temperature ultrahigh vacuumsetup (Omicron Nanotechnology) at a base pressure in the10−11 Torr range. For maximum stability and reproducibility,all measurements were performed at 5 K. Cooling is achievedby means of a cold finger. Chemically etched tungsten tips wereused. STM topographic imaging (800 × 800 data points) ofsupracrystals on HOPG substrate was performed in constantcurrent mode. Simultaneously with the topographic image,I -V curves were recorded with an open feedback loop in thecurrent imaging tunneling spectroscopy mode26 with a gridsize of 200 × 200 points. Then, maps of conductance areobtained numerically from these grids at selected values of

075409-11098-0121/2012/86(7)/075409(6) ©2012 American Physical Society

YANG, ARFAOUI, CREN, GOUBET, AND PILENI PHYSICAL REVIEW B 86, 075409 (2012)

the tunneling voltage. Image processing was performed withthe WSXM program27 and a MATHEMATICA program adaptedin our laboratory. All experiments were repeated severaltimes with different NC batches, and the results presentedin this work were representative and consistent with the morecomprehensive data set.

III. RESULTS AND DISCUSSION

Au NCs coated with dodecanethiol are characterized byan average diameter of 7 nm and a size distribution of 6%[Fig. 1(a)]. A careful study shows that these NCs are composedof a large amount of single-domain crystals, whereas othersare mainly multiple-twinned particles (MTPs) with very fewpolycrystals [Fig. 1(b)]. Because of the low size distribution,the NCs are able to self-order in well-defined supracrystals[Figs. 1(c) and 1(d)] with a thickness of ∼300 nm, corre-sponding to ∼30 NC layers. The small-angle x-ray scattering(SAXRD) pattern [inset of Fig. 1(c)] shows a very brightreflection normal to the substrate, indicating that the NCarrangements are always oriented with (111) as the basalplane on the substrate sharing a common crystallographicaxis normal to the substrate. A long-range ordering withincrystalline domains with a fcc structure is also found. As seenin Fig. 1(d), by tilting the sample, the supracrystal has sharpedges. However, very thin supracrystals of a few NC layersclose to the large supracrystals can be observed as indicated

FIG. 1. (Color online) (a) TEM image of Au NCs. (b) High-resolution TEM image of Au NCs. (c) SEM image of thin (indicatedby the red arrows) and thick supracrystals. Inset: SAXRD patternof the supracrystals. (d) Tilted field-emission SEM image of thicksupracrystal.

FIG. 2. (Color online) (a) Typical topographic STM image oflarge-area, thick Au supracrystals with tunneling current set pointat 50 pA and bias voltage at 2.5 V. Inset: sixfold symmetry of thefast Fourier transform (FFT) pattern of the STM image. (b) Typicaltopographic STM image of the thin supracrystal surrounding thethick supracrystal with tunneling current set point at 200 pA and biasvoltage at 0.5 V. The strips indicate that Au NCs are adsorbed on theSTM tip. The numbered rectangles correspond to the regions wherethe average conductance curves have been deduced.

by the arrows in Fig. 1(c). This is how we distinguish betweenthick and thin supracrystals. Figure 2(a) shows a typicalSTM image of a thick supracrystal similar to those shown inFigs. 1(c) and 1(d). Such STM images are obtained at relativelow current, between 50 and 150 pA. Surprisingly, these highlystable thick supracrystals can be imaged at bias voltages upto 9 V. Figure 2(a) shows that the NCs are highly ordered ina hexagonal network as confirmed by the Fourier transformpattern having a sixfold symmetry [inset of Fig. 2(a)]. Thetypical STM image of thin supracrystals is shown in Fig. 2(b).It is clear that there are a few NC layers in this thin supracrystalon the HOPG surface. The surface of thin supracrystal alsopresents a hexagonal NC array.

Before considering the thin and thick supracrystals, wewill first exhibit the electronic transport through a single NC.During the STM tip scanning, isolated NC can be stuck tothe tip. Such an event is reflected in the STM image by moreor less narrow strips [Fig. 2(b)]. The corresponding dI/dV

spectra taken within strips in areas 1 and 3 show a curve with agap surrounded by some oscillations. Such characteristics areknown to be the signature of Coulomb blockade and Coulombstaircase.17,28,29 In contrast, outside the strips, the spectrum inregion 2 [curve 2 in Fig. 3(a)] exhibits the dI/dV spectrumof HOPG with the expected “V shape.”30,31 In the following,the width of the Coulomb gap is determined approximately bythe distance between the kinks in the dI/dV curve at positiveand negative biases. From curves 1 and 3, we measure that the

075409-2

ELECTRONIC PROPERTIES PROBED BY SCANNING . . . PHYSICAL REVIEW B 86, 075409 (2012)

FIG. 3. (Color online) (a) The tunneling spectra show theCoulomb blockade and Coulomb staircase: Average dI/dV curvesof isolated Au NCs (curves 1 and 3) taken inside regions 1 and3 of Fig. 2(b), respectively. Curve 2 is the average dI/dV curvetaken inside region 2 of Fig. 2(b) and corresponding to the expected“V shape” of HOPG. Inset: (left) schematic diagram of Coulombblockade system where a NC is adsorbed on the STM tip and(right) the equivalent circuit of the double-barrier tunneling junction.(b) The experimental dI/dV curve is fitted using the orthodox theoryat different temperatures, 5 and 100 K, respectively. The parametersused in the orthodox theory are given in the inset.

width of the Coulomb gap of a single NC junction is around360 meV, demonstrating that the tip-particle HOPG junctionpresents a single-electron transport mechanism. It meansthat the charging energy of one electron exceeds the thermalkinetic energy and that the tunneling resistance of the junction(RT ) is significantly larger than the quantum resistance (RQ =h/4e2 ≈ 6.5 k�). The typical tunneling spectrum of anisolated NC deposited on HOPG substrate, forming a double-barrier tunnel junction as illustrated in the inset of Fig. 3(a),is simulated by using the orthodox theory.32,33 The best fit at5 K (dashed line) of the experimental data (dot) is obtained byassuming that C1 ≈ C2 and R1 < R2 [see the values in the insetof Fig. 3(b)]. Although this fit reflects the main features of theexperimental curve, it is not possible to reproduce perfectlythe experimental curve at 5 K. Note that both experimental andsimulated spectra present shifts and asymmetries indicatingeffects of a fractional residual charge Q0 on the tunnelingjunction.33,34 One can improve the fitting by using an effectivetemperature of around 100 K [see solid line in Fig. 3(b)] inthe orthodox model. This temperature increase could have aphysical origin since it has been reported35,36 that some localheating or an inelastic contribution to the tunneling currentcould occur for such a double-barrier tunnel junction made ofdodecanethiol molecules.

FIG. 4. (Color online) (a) Typical 160 × 160 nm STM image ofthe thin supracrystal (left) on HOPG (right) at 5 K with tunnelingcurrent set point at 50 pA and bias voltage at 1 V. (b) Average dI/dV

of HOPG (taken inside region A of the STM image) and the thinsupracrystal (taken inside region B of the STM image). (c), (d),(e) Conductance maps of the topographic STM image shown in (a)at 0, 300, and 80 mV, respectively. (f) Average dI/dV taken insideregion C (blue curve) and region D (red curve) of (e), respectively.

Now, if these isolated NCs are organized in quasi-2D films(see Experimental section) as thin supracrystals [Fig. 1(c)],they are well observed by STM as shown in Fig. 2(b).Figure 4(a) shows another typical 160 nm × 160 nm STMimage of this thin supracrystal. On the left side of Fig. 4(a),NCs are organized in a hexagonal array forming a thinsupracrystal that is located against a defect of HOPG (darkborder) imaged on the right side. The average dI/dV curveof the substrate [curve A in Fig. 4(b)] taken inside the circle A[Fig. 4(a)] shows the typical spectroscopic signature of HOPG.In the same way, the average dI/dV curve of the NC network[curve B in Fig. 4(b)] taken inside the circle B [Fig. 4(a)]demonstrates a symmetrical Coulomb gap. The width of theCoulomb gap is of the order of 250 meV, a bit smaller thanthat of the single NC observed in Fig. 3(a). Moreover, no sharpCoulomb staircase is observed above the gap. Both features arecertainly due to a collective behavior of the NCs forming thethin supracrystal and their interactions with the substrate. Theconductance map associated to the thin supracrystal, at 0 mV,is finite valued. It appears to be very homogeneous with a loweraverage conductance than that of HOPG [Fig. 4(c)]. However,we could expect that it is an effect of direct tunneling betweenthe tip and the substrate without passing through the NCs.The dielectric layer of the thiols capping the NCs lowers the

075409-3

YANG, ARFAOUI, CREN, GOUBET, AND PILENI PHYSICAL REVIEW B 86, 075409 (2012)

tunneling barrier37 compared to vacuum and then allows someelectrons to pass directly between the tip and the surface.16

At 300 mV [Fig. 4(d)], outside the gap, the contrast is stillvery homogeneous and inverted compared to that in Fig. 4(c).In order to observe some singularities in the conductance ofthin supracrystal, the bias should be close to the gap edge.Indeed, in Figs. 4(d) and 4(e) (80 and 300 mV, respectively),we note that there are some variations at the NC scale atthe border between thin supracrystal edge and HOPG. Forinstance, at 80 mV [Fig. 4(e)], the conductance map revealssome singularities at the NC scale with low conductance [bluespot inside circle C in Fig. 4(e)].

These singularities are concentrated at the border betweenthin supracrystal edge and HOPG, but they are also seen farfrom this border. The averaged conductance curves of thesesingularities (circles C and D) are shown in Fig. 4(f) (curvesC and D). The Coulomb gap width associated to these spotsis ∼100 meV larger than the one averaged on the NC networktaken inside circle B [Fig. 4(b)]. Furthermore, Figs. 4(c)–4(e)show a marked change in the conductance for NCs close tothe edge of the thin supracrystal. This gap enlargement may bedue to the fact that NCs at the edge have fewer neighbors thanthose far from the edge. This is in good agreement with the factthat the average Coulomb gap width of the thin supracrystalis reduced compared to that of a single NC. When the numberof neighbors is increased, the capacitance associated withthe junction is increased, and therefore the Coulomb energy(e2/2C) is reduced. The gap enhancement at the supracrystaledge might also be due to an accumulation of single-crystallineNCs at the edge. This is well observed from the transmissionelectron microscopy (TEM) pattern observed in the supracrys-tal with few NC layers.38 Figure 1S shows the bright (a)and dark (b) field TEM images with appearance of veryhomogeneous NCs (either totally bright or dark) at the edgeof the film and heterogeneous NCs forming monolayers.38

Figure 1S(c) points out the region of such nanocrystallinitysegregation.38 Other experiments recently carried out in ourgroup with Au NCs show nanocrystallinity segregation withformation of supracrystals composed exclusively of eithersingle-domain NCs or MTPs.39 Although the conductance ofthin supracrystals is very uniform throughout, we still can findsome areas where the conductance varies markedly from oneNC to its neighbor [Fig. 5(a)]. Figure 5(a) shows a conductancemap superimposed on the corresponding 3D topographic STMimage (26 nm × 26 nm). As shown in Fig. 5(a), the NC labeledas A has a conductance clearly lower than one of its neighborslabeled as B, which is in turn lower than one of its neighborslabeled as C. The corresponding dI/dV curve, taken overall of each NC, shows that the Coulomb gap width does notchange.40 From the STM image shown in Fig. 5(a), the NC sizecan not be assigned to the change in the conductance betweenA, B, and C NCs. By comparing the STM [Fig. 5(a)] andTEM [Fig. 5(b)] images, these differences in the conductancecould be attributed to a change in the nanocrystallinity duringthe supracrystal formation, i.e., either a single crystalline, ora MTP, or a polycrystalline NC as labeled by 1, 2 and 3,respectively, in Fig. 5(b).

The electronic behavior of the thick supracrystals is verydifferent from that of the thin supracrystals. The typical I -Vcurves of well-ordered thick supracrystals exhibit a nonlinear

FIG. 5. (Color online) (a) 26 nm × 26 nm 3D STM image taken ona thin supracrystal at 5 K with tunneling current set point at 100 pA andbias voltage at 1.5 V, where the color corresponds to the conductancedistribution at − 40 mV bias voltage. (b) High-resolution TEM imageof 7 nm Au NCs shows different nanocrystallinity labeled by 1, 2,and 3, respectively.

behavior at 5 K (inset of Fig. 6) totally different from thereported one associated to the thin supracrystals [Fig. 4(b)].The logarithmic plot (Fig. 6) shows a power-law dependenceof tunneling current on bias voltage I ∝ V ζ with ζ ≈ 3.3 whenthe bias voltage is higher than 200 mV at 5 K. Similarly, for thethin supracrystal, there is also a power law but with a ζ equal to∼1.4. A typical 100 nm × 100 nm topographic STM image of

FIG. 6. The logarithmic plot of a typical I − V curve measuredabove a well-defined thick supracrystal at 5 K shows power-lawdependence when the bias voltage is higher than 200 mV. The dashedline is guide for the eyes. Inset: corresponding linear plot of the I − V

curve.

075409-4

ELECTRONIC PROPERTIES PROBED BY SCANNING . . . PHYSICAL REVIEW B 86, 075409 (2012)

FIG. 7. (Color online) (a) Typical 100 × 100 nm topographicSTM image of the well-defined supracrystals with tunneling currentset point at 130 pA and bias voltage at 2.3 V. (b) Conductance mapat 800 mV bias voltage. (c) Average dI/dV curves taken insidethe regions indicated by the colored circles (b). (d) Average dI/dV

curves taken on the spots indicated by the white circles (b).

the thick, well-ordered supracrystal is presented in Fig. 7(a).The surface shows hexagonal NC packing. The dI/dV mapcorresponding to the STM image at 800 mV [Fig. 7(b)] showsthat the conductance is very homogeneous over all the surface.This homogeneity is related to the well-defined structure of thesupracrystals. Note that, at 800 mV, a marked change in theconductance in large areas (three terraces) of the STM imageis observed with lower conductance or higher conductance.Considering the three regions labeled by black, blue, andred circles in Fig. 7(b), the dI/dV curves show a similarCoulomb gap [Fig. 7(c)] with a width nearly equal to that ofa single NC junction (Fig. 3). However, a deeper analysisshows that the Coulomb gap in the black, blue, and redregions is slightly different. Au NCs are characterized by asize distribution of 6% and differ by their nanocrystallinitywith a mixture of single-crystalline, multiple-twinned, andpolycrystalline NCs. Because of this low size distribution, itdoes not seem reasonable to assume that a size segregationtakes place as already proposed in a previous paper to explaina shift in the quadrupolar mode of Ag supracrystals,41 inwhich the size distribution was quite large (around 12%).Moreover, the size segregation can not be assigned in the3D supracrystals (the histogram corresponding to the NCsinvolved in the supracrystals is in the experimental range ofthose obtained from the as-colloidal solution). Here, becauseof the slight change in the Coulomb gap, it seems reasonableto assume that the change in the conductance could be due tonanocrystallinity segregation of NCs during the supracrystalsformation as mentioned for the thin supracrystal previously.Figure 7(b) also shows singularities, at 800 mV, at the NCscale on the conductance map corresponding to the STMimage in Fig. 7(a). Some NCs with lower conductance andhigher conductance are distinguishable. In order to examinethe conductance of these NC singularities, the averaged dI/dV

curves corresponding to each NC (white solid and white

dashed circles) are shown in Fig. 7(d). In both of the twocurves, a staircaselike structure is clearly distinguishable,whereas the Coulomb gap widths remain nearly unchangedcompared to the averaged dI/dV curves [Fig. 7(c)] overlarge-scale scanned area as indicated by the colored circles inFig. 7(b). The staircaselike structure observed may be relatedto the staircase observed for the isolated NCs (Fig. 3). Hence, at800 mV, one can then observe a change in conductance acrossthe NC without changing the Coulomb gap. Unexpectedly,these singularities at the NC scale show a staircaselike structurein their dI/dV curves, which may suggest that these NCsforming the supracrystal partially keep features of isolatedNC (Fig. 3). The change in the staircase structure could beattributed to the interactions between NCs. Note that we cancount more or less the same number of steps in both cases,in Fig. 7(d) and Fig. 3, respectively. This seems to indicatethat these singularities at the NC scale, forming the thicksupracrystals, and surrounded by a large number of NCs,behave as isolated NCs deposited on HOPG. Nevertheless,locally, because of the NC size distribution, these singularitiescould be NCs smaller or larger than 7 nm. This has been alreadyobserved42 by photoinduced emission with a marked changein the behavior of small NCs embedded in a NC monolayer.

IV. CONCLUSION

Electronic properties at 5 K of a 300-nm-thick supracrystalmade of dodecanethiol-passivated 7-nm Au NCs and orderedin a fcc structure called supracrystals are investigated. Thethick fcc supracrystals are sufficiently conductive to beinvestigated by STM/STS even if the coupling of the NCswithin the supracrystal is weak. The dI/dV curves show agap with a width of about 310 meV and a clear staircaselikestructure. The conductance maps are very homogeneous.However, some areas show a slight change in the conductancemay be attributed to the nanocrystallinity segregation of NCsduring the supracrystal formation. Furthermore, some localsingularities at the NC scale are detected and correspond toNCs embedded in the supracrystal surface. These particularNCs mimic the behavior of isolated NCs deposited on HOPG.As with the thick supracrystal, the thin supracrystal exhibitsa homogeneous conductance over the entire surface tested.Finally, at the edge of the thin supracrystal, the conductancediffers from that observed on the overall supracrystal, indicat-ing changes in the electronic transport of NCs either due to thelack of NC neighbors or the change in the nanocrystallinity.

ACKNOWLEDGMENTS

The authors appreciate the help of small-angle x-rayscattering characterization by P.-A. Albouy (Universite Paris-sud 11). The authors thank F. Charra, L. Douillard, andC. Fiorini (Commissariat a l’Energie Atomique) for the fruitfuldiscussion. P.Y. thanks the China Scholarship Council forfinancial support. The research leading to these results hasreceived funding from the European Community’s SeventhFramework Programme (FP7/2008-2011) under Grant Agree-ment No. 213382. The research leading to this paper has beenpartially supported by an Advanced Grant of the EuropeanResearch Council under Grant Agreement No. 267129.

075409-5

YANG, ARFAOUI, CREN, GOUBET, AND PILENI PHYSICAL REVIEW B 86, 075409 (2012)

*marie-paule.pileni@upmc.fr1A. V. Narlikar and Y. Y. Fu, The Oxford Handbook of Nanoscienceand Technology (Oxford University Press, Oxford, UK, 2010).

2M. Suvakov and B. Tadic, J. Phys.: Condens. Matter 22, 163201(2010).

3C. B. Murray, C. R. Kagan, and M. G. Bawendi, Annu. Rev. Mater.Sci. 30, 545 (2000).

4M. P. Pileni, J. Phys. Chem. B 105, 3358 (2001).5M. P. Pileni, Surf. Sci. 603, 1498 (2009).6P. Podsiadlo, G. Krylova, B. Lee, K. Critchley, D. J. Gosztola,D. V. Talapin, P. D. Ashby, and E. V. Shevchenko, J. Am. Chem.Soc. 132, 8953 (2010).

7Y. Suganuma and A. A. Dhirani, J. Phys. Chem. B 109, 15391(2005).

8C. Burda, X. Chen, R. Narayanan, and M. A. El-Sayed, Chem. Rev.105, 1025 (2005).

9M. P. Pileni, J. Phys. Chem. C 111, 9019 (2007).10M. P. Pileni, Acc. Chem. Res. 41, 1799 (2008).11M. P. Pileni, Phys. Chem. Chem. Phys. 12, 11821 (2010).12M. R. Jones, R. J. Macfarlane, B. Lee, J. Zhang, K. L. Young, A. J.

Senesi, and C. A. Mirkin, Nat. Mater. 9, 913 (2010).13L. Motte, F. Billoudet, and M. P. Pileni, J. Phys. Chem. 99, 16425

(1995).14U. Simon, Adv. Mater. 10, 1487 (1998).15T. B. Tran, I. S. Beloborodov, J. Hu, X. M. Lin, T. F. Rosenbaum,

and H. M. Jaeger, Phys. Rev. B 78, 075437 (2008).16A. Taleb, F. Silly, A. O. Gusev, F. Charra, and M. P. Pileni, Adv.

Mater. 12, 633 (2000).17T. P. Bigioni, L. E. Harrell, W. G. Cullen, D. K. Guthrie, R. L.

Whetten, and P. N. First, Eur. Phys. J. D 6, 355 (1999).18A. Zabet-Khosousi and A.-A. Dhirani, Chem. Rev. 108, 4072

(2008).19F. Silly, A. O. Gusev, A. Taleb, F. Charra, and M.-P. Pileni, Phys.

Rev. Lett. 84, 5840 (2000).20P. J. Durston, J. Schmidt, R. E. Palmer, and J. P. Wilcoxon, App.

Phys. Lett. 71, 2940 (1997).21R. C. Doty, H. B. Yu, C. K. Shih, and B. A. Korgel, J. Phys. Chem.

B 105, 8291 (2001).

22I. S. Beloborodov, A. V. Lopatin, V. M. Vinokur, and K. B. Efetov,Rev. Mod. Phys. 79, 469 (2007).

23P. Yang, I. Arfaoui, T. Cren, N. Goubet, and M.-P. Pileni, Nano Lett.12, 2051 (2012).

24N. Goubet, J. Richardi, P. A. Albouy, and M. P. Pileni, Adv. Funct.Mater. 21, 2693 (2011).

25N. Goubet, H. Portales, C. Yan, I. Arfaoui, P. A. Albouy, A. Mermet,and M. P. Pileni, J. Am. Chem. Soc. 134, 3714 (2012).

26R. J. Hamers, R. M. Tromp, and J. E. Demuth, Phys. Rev. Lett. 56,1972 (1986).

27I. Horcas, R. Fernandez, J. M. Gomez-Rodriguez, J. Colchero,J. Gomez-Herrero, and A. M. Baro, Rev. Sci. Instrum. 78, 013705(2007).

28B. Wang, H. Q. Wang, H. X. Li, C. G. Zeng, J. G. Hou, and X. D.Xiao, Phys. Rev. B 63, 035403 (2000).

29L.-P. Xu and S. Chen, Chem. Phys. Lett. 468, 222 (2009).30J. W. McClure, Phys. Rev. 108, 612 (1957).31M. Yamazaki, J. Chem. Phys. 26, 930 (1957).32K. K. Likharev, Proc. IEEE 87, 606 (1999).33A. E. Hanna and M. Tinkham, Phys. Rev. B 44, 5919 (1991).34M. Amman, R. Wilkins, E. Ben-Jacob, P. D. Maker, and R. C.

Jaklevic, Phys. Rev. B 43, 1146 (1991).35M. Tsutsui, M. Taniguchi, and T. Kawai, Nano Lett. 8, 3293

(2008).36Y.-C. Chen, M. Zwolak, and M. Di Ventra, Nano Lett. 5, 621

(2005).37C. R. van den Brom, I. Arfaoui, T. Cren, B. Hessen, T. T. M. Palstra,

J. T. M. De Hosson, and P. Rudolf, Adv. Funct. Mater. 17, 2045(2007).

38See Supplemental Material at http://link.aps.org/supplemental/10.1103/PhysRevB.86.075409 for Fig. 1S.

39Y. Wan, H. Portales, and M.-P. Pileni (private communication).40See Supplemental Material at http://link.aps.org/supplemental/

10.1103/PhysRevB.86.075409 for Fig. 2S.41A. Courty, A. Mermet, P. A. Albouy, E. Duval, and M. P. Pileni,

Nat. Mater. 4, 395 (2005).42A. O. Gusev, A. Taleb, F. Silly, F. Charra, and M. P. Pileni, Adv.

Mater. 12, 1583 (2000).

075409-6