18622
J. Phys. Chem. C 2010, 114, 18622–18633
Covalent Anchoring of Re6Sei8 Cluster Cores Monolayers on Modified n- and p-Type Si(111) Surfaces: Effect of Coverage on Electronic Properties Ste´phane Cordier,*,† Bruno Fabre,*,† Yann Molard,† Alain-Bruno Fadjie-Djomkam,‡ Nicolas Tournerie,† Alexandra Ledneva,§ Nikolaı¨ G. Naumov,§ Alain Moreac,‡ Pascal Turban,‡ Sylvain Tricot,‡ Soraya Ababou-Girard,*,‡ and Christian Godet‡ Sciences Chimiques de Rennes, UMR 6226 CNRS-UniVersite´ de Rennes 1, Campus de Beaulieu, 35042 Rennes Cedex, France, Institut de Physique de Rennes, CNRS UMR 6251-UniVersite´ de Rennes 1, Campus de Beaulieu, 35042 Rennes Cedex, France, and Chemical Department NikolaeV Institute of Inorganic Chemistry, Siberian Branch of Russian Academy of Sciences, 3 Acad. LaVrentieV pr., 630090 NoVosibirsk, Russia ReceiVed: July 29, 2010; ReVised Manuscript ReceiVed: September 15, 2010
The electronic properties of redox-active transition metal clusters (Re6Se8) covalently immobilized on modified Si(111) surfaces through linear alkyl spacers have been studied as a function of the cluster coverage (1 × 1013-6 × 1013 cm-2). The latter is controlled by using Si(111)/H surfaces modified by dense mixed alkyl/ acid-terminated monolayers with variable fraction of the acid grafting sites from 5 to 100% in solution. Quantitative X-ray photoemission analysis, spectroscopic ellipsometry, and scanning tunnelling microscopy indicate a covalent attachment of a submonolayer to densely packed monolayer of Re6Se8 clusters, while the vibrational Raman signature confirms the cluster integrity within the monolayer. Electrical band gaps as deduced from scanning tunnelling spectroscopy have been obtained for low Re6Se8 cluster coverage. Using ultraviolet photoemission spectroscopy, electronic properties such as ionization potential changes and energy level alignments at organic/inorganic interfaces are studied. We show that the lowest unoccupied molecular orbital of the Re6Se8 cluster is close to the bottom of the Si conduction band. At high cluster coverage, this affects the current-voltage characteristics measured using a weakly interacting top mercury contact onto the organic monolayer/silicon junctions. Indeed, on n-type silicon, the high level current at low bias and the shape of the conductance G(V) curve indicate a Schottky barrier height lowering. On the other hand, the current-voltage characteristics are the same for both acid-terminated and low coverage Re6Se8 cluster junctions at low bias; the high Schottky barrier height limits the current at low bias. When the forward bias increases, the current is tunnelling limited. As expected from the band alignment deduced from photoemission data, the opposite behavior is obtained on p-type silicon. 1. Introduction The chemistry of M6Li8La6 units containing M6 octahedral metal atom clusters (L ) halogen and/or chalcogen, i ) inner, a ) apical; Figure 1) is particularly well developed with rhenium, molybdenum, and tungsten.1-4 Their properties depend on the number of electrons involved in the metal-metal bonds that generates specific physicochemical properties.5,6 The latter are completely different from those of macroscopic metal particles. Indeed, M6Li8La6 units constitute useful structuring building blocks to carry mechanical, electronic, and luminescent properties in the elaboration of supramolecular assemblies and nanomaterials.7-12 For instance, the octahedral architecture of metal atoms with the presence of a 3-fold axis has an important role in the structuration of metal-organic frameworks including charge transfer salts.13,14 Electronic π-d interactions were evidenced recently between tetrathiafulvalene (TTF) derivatives * To whom correspondence should be addressed. E-mail: (S.C.)
[email protected]; (B.F.)
[email protected]; (S.A.G.)
[email protected]. Tel: +33 (0)223 236 607. Fax: +33 (0)223 236 799. † Sciences Chimiques de Rennes, UMR 6226 CNRS-Universite´ de Rennes 1. ‡ Institut de Physique de Rennes, CNRS UMR 6251-Universite´ de Rennes 1. § Siberian Branch of Russian Academy of Sciences.
grafted on Mo6Xi8 cluster cores (X ) Cl, Br, or I).15 Last but not least, owing to their luminescent properties, characterized by a wide photon emission window from red to near-infrared upon excitation from UV to visible, the cluster units can be used as red dye in the elaboration of nanoparticles,16-18 nanocomposites,19 or liquid crystals20 that offer a broad range of potential applications in bio and imaging technologies. The integration of functional cluster units onto conducting surfaces is a necessary step toward the development of novel electrically addressable and switchable functional devices. On the other hand, hybrid semiconductor-molecule devices have been used to check that some molecular species retain their physicochemical activity once covalently immobilized on semiconducting surface.21-24 This hybrid approach is promising for combining the advantages of semiconductor technology (doping, processing) and the large flexibility in designing molecular structures.25-30 In this frame, we performed recently a full study of molecular junctions based on Mo6Ii8 clusters cores immobilized on Si(111) surfaces via an organic monomolecular layer (OML) containing terminal pyridine groups.31,32 The presence of grafted Mo6Ii8 clusters influences the electronic properties of the junctions and provides new features evidenced in the current density versus voltage J(V) characteristics.
10.1021/jp1071007 2010 American Chemical Society Published on Web 10/08/2010
Re6Sei8 Cluster Cores Monolayers
J. Phys. Chem. C, Vol. 114, No. 43, 2010 18623
Figure 1. (a) Representation of the trans-Re6Sei8(TBP)a4(OH)a2 cluster unit. Selenium, rhenium from the Re6 clusters, and nitrogen atoms are represented as yellow, gray, and blue balls, respectively. (b) Representation of the same units according to a different orientation. (c) Schematic representation of the Re6Se8 cluster immobilization on Si(111) through an acido-basic reaction with carboxylic acid groups ending an alkyl chain.
In this work, in order to evaluate the effect of the type of clusters on the electronic properties, a full study is presented on molecular junctions-based Re6Sei8 cluster cores immobilized on n- and p-type Si(111) surfaces. A simpler immobilization process has been established using Re6Sei8 as compared with the pyridine grafting mechanism used with Mo6Ii8 cluster cores. It consists in the reaction in solution of trans-Re6Sei8(TBP)a4(OH)a2 cluster units precursors with carboxylic acid groups end-capping an alkyl monomolecular layer covalently bound to a hydrogen terminated silicon surface. As shown in previous work, such a reaction between trans-Re6Sei8(TBP)a4(OH)a2 and carboxylic acid leads to the formation of conjugated carboxylates that are in situ grafted on the cluster in place of OH groups.33 This study shows that the cluster coverage can be controlled using mixed alkyl/ acid-terminated monomolecular layers with variable dilution on Si(111)/H surfaces obtained by using mixtures of 1-dodecene and undecylenic acid in the first step of the grafting process. Various experimental techniques including spectroscopic ellipsometry, X-ray and ultraviolet photoelectron spectroscopies (XPS, UPS), Raman spectroscopy, electrochemical impedance spectroscopy (EIS), and scanning tunnelling microscopy (STM) combined to current tunnelling spectroscopy (STS) have been used. The electrical properties of metal-insulator-semiconductor (MIS) diodes prepared from these functional films were also studied using a mercury drop as soft electrical top contact. Typical parameters of Hg/OML-Si junctions (before and after cluster grafting) such as rectification factors and barrier heights are estimated from J(V) characteristics. Data are interpreted on the basis of the band alignment scheme deduced from photoemission measurements without the top contact. The effect of the Re6Se8 cluster coverage on physical properties is discussed and compared with those reported for Mo6I8 cluster monolayers immobilized on Si surfaces. 2. Experimental Section 2.1. Preparation of Rhenium Cluster-Terminated Si(111) Surfaces. Trans-Re6Sei8(TBP)a4(OH)a2 was prepared and recrystallized following the process reported in the literature33
using K4Re6Sei8(OH)a6 · 8H2O as starting precursor.34 The transRe6Sei8(TBP)a4(OH)a2 unit is neutral and chemically very stable, for example, it can be stored in air without any risk of oxidation or exchange of apical ligand by water molecules from the ambient atmosphere. It is based on an octahedral Re6 cluster surrounded by eight inner Se ligands in face-capping positions along with two OH and four tert-butylpyridine (TBP) groups in apical positions (Figure 1). The OH groups are linked to the cluster in trans positions and the four tert-butylpyridine (TBP) groups are linked via the nitrogen atom of pyridine groups. The reaction of trans-Re6Sei8(TBP)a4(OH)a2 with carboxylic acid in refluxing chlorobenzene leads easily to the formation of a carboxylate derivative that in situ anchors the cluster via a Re-O bond. This simple water elimination reaction enables the formation of a stable organic chain/cluster interface. This grafting technique greatly simplifies the cluster immobilization procedure as compared with that used previously for Mo6I8.31,32 Indeed, the ((n-CH4)4N)2Mo6Ii8(CF3SO3)a6 precursor was air sensitive and the exchange of apical ligands by pyridine needed to be balanced by triflate charged moieties with some evidence of uncomplete reaction of apical Ia ligands.31,32 For the following discussion, we must point out that in structures incorporating trans-Re6Sei8(TBP)a4(OH)a2, the tert-butylpyridine groups favor the aggregation of building blocks via π-π stacking interactions.33 The chemicals used for cleaning and etching silicon wafer pieces (30% H2O2, 96-97% H2SO4, and 40% NH4F solutions) were of VLSI semiconductor grade (Riedel-de-Hae¨n). All Teflon vials used for cleaning of silicon were previously decontaminated in 3:1 v/v concentrated H2SO4/30% H2O2 at 100 °C for 30 min, followed by copious rinsing with ultrapure water. Caution: The concentrated H2SO4:H2O2 (aq) piranha solution is Very dangerous, particularly in contact with organic materials, and should be handled extremely carefully. All single side polished silicon(111) samples (n-type, phosphorus-doped or p-type, boron-doped, 5-10 Ω cm, thickness ) 525 ( 25 µm from Siltronix) were cut into 1.5 × 1.5 cm2 pieces from the same silicon wafer to ensure the maximum reproducibility of hydrogen-terminated and further molecular monolayer-modified surfaces. The sample was sonicated for 10
18624
J. Phys. Chem. C, Vol. 114, No. 43, 2010
min successively in acetone (MOS semiconductor grade, Carlo Erba), ethanol (99.8%, VLSI semiconductor grade), and ultrapure 18.2 MΩ cm water (Elga Purelab Classic UV, Veolia Water STI). It was then cleaned in 3:1 v/v concentrated H2SO4/ 30% H2O2 at 100 °C for 30 min, followed by copious rinsing with ultrapure water. The surface was etched with argon-sparged ppb grade 40% aqueous NH4F for 15-20 min at room temperature.35 The NH4F solution was thoroughly deaerated with argon for at least 30 min prior to the immersion of the piranhatreated surface. After etching, the hydrogen-terminated sample (n- or p-type Si-H) was rinsed with argon-saturated water, blown dry with argon, and used immediately for the covalent attachment of the single-component acid-terminated monolayer or mixed n-dodecyl/acid-terminated monolayer. Such monolayers were prepared from the photochemical reaction at 300 nm for 3 h of Si-H with neat undecylenic acid (Acros, 99%, previously passed through a neutral, activated alumina column to remove residual water and peroxides or mixtures of undecylenic acid/1-dodecene (puriss, >99% from Fluka) with molar ratios of 5/95, 10/90, and 33/67.36-42 It has been demonstrated that this direct hydrosilylation route does not lead to appreciable reaction between the carboxyl groups and the surface provided that short UV irradiation times are used (typically, less than 4 h).43 The carboxylic acid-modified surface was rinsed copiously with acetone, then dipped in hot acetic acid at 65 °C for 2 × 20 min and dried under an argon stream.44 The carboxylic acid-modified surfaces were denoted as SiAC100 (for neat undecylenic acid), SiAC5, SiAC10, and SiAC33, respectively. Then the carboxylic acid-modified surface was soaked overnight under argon in a refluxing and stirred chlorobenzene solution (15 mL) containing ca. 5 × 10-4 M Re6Sei8(TBP)a4(OH)a2. The rhenium cluster-modified surface was thoroughly rinsed with freshly distilled CH2Cl2 to obtain a contaminant free perfectly clean surface for subsequent studies. The rhenium clustermodified surfaces prepared from SiAC100, SiAC5, SiAC10, and SiAC33 were denoted as SiRe100, SiRe5, SiRe10, and SiRe33, respectively. 2.2. Spectroscopic Ellipsometry. Spectroscopic ellipsometry (SE) experiments were performed in the range 1.5-4.7 eV at an incidence angle of 70°, using a Horiba (UVISEL) ellipsometer, and analyzed with either two-layer (Si + alkyl) or threelayer (Si + alkyl + cluster layer) models. For the transparent (k ) 0) organic molecular layers, an energy independent refractive index, n, was chosen because no improvement in the fitting result was found with a Schott dispersion formula. To obtain the optical thickness dopt of mixed alkyl-acid layers, their refractive index was set at n ) 1.48, while SE spectra of Re6Se8alkyl-Si surfaces were fitted to the three-layer model using fixed parameters (n ) 1.48, dopt) obtained for the corresponding alkylSi surface. A large index contrast is found between the alkyl layer index (n ) 1.48) and Re6Se8 cluster layer index (n > 1.7). 2.3. Raman Measurements. Raman measurements have been performed on a LabRam HR800 spectrometer (Horiba Scientific/Jobin Yvon). As a continuous background dominates the signal for incident radiations of 633 nm (He-Ne laser), Raman spectra were recorded using a 785 nm exciting laser diode line. Laser line was focused on samples through a 100× objective that gave a diameter of the focused beam spot at around 1 µm. To prevent thermal damage of both porous silicon substrate and of the grafted Re6Se8 cluster, optical density filter has been used on the laser path. Scattered light was collected in back scattering configuration. Spectral resolution was around 0.7 cm-1 per pixel. A double side polished silicon (100) shard (1.5 × 1.5 cm2, 5 × 10-3 Ω cm, p-type, boron-doped, thickness
Cordier et al. ) 375 ( 25 µm, Siltronix) was sonicated for 10 min successively in acetone, ethanol, and ultrapure 18.2 MΩ cm water. It was then cleaned in 3:1 v/v concentrated H2SO4/30% H2O2 at 100 °C for 30 min, followed by copious rinsing with ultrapure water. The sample was then dipped in 50% HF for 1 min and dried under an argon stream. It was pressed against an opening in the cell bottom using a FETFE (Aldrich) O-ring seal and the ohmic contact was made on the polished rear side of the sample with the steel bottom cap (any conducting material was not dropped to avoid surface contamination for subsequent Raman investigation). A platinum counter electrode was used. The hydrogen-terminated porous Si(100) surface was produced by applying a current density of 50 mA cm-2 for 5 min in 50%HF/ ethanol (MOS grade)/ultrapure 18.2 MΩ cm water (2:2:1 vol). The surface was then rinsed with ethanol and dried under an argon stream. The sample was immediately transferred under argon into a pyrex Schlenk tube containing neat undecylenic acid (Acros, 99%, previously passed through a neutral, activated alumina column to remove residual water and peroxides, and heated at 100 °C under argon for 1 h). Temperature was increased to 120 °C and kept at this value for ca. 6 h. As for flat Si(111) crystals, the sample was rinsed several times with hot acetic acid and acetone then dried under an argon stream. The immobilization of the metal cluster on acid-modified porous Si(100) was similar to that described for flat Si(111). 2.4. Scanning Tunneling Microscopy. STM experiments were performed in an ultrahigh vacuum chamber with a base pressure below 10-10 mbar using an OMICRON VT STM operating at room temperature in the constant current mode. Electrochemically etched W STM tips were cleaned in situ by thermal heating before the STM experiments. An ohmic contact was taken at the backside of the silicon substrate to ensure samples grounding (Vsample ) 0). The applied tunnel voltage Ugap is defined by Ugap ) Vsample - Vtip ) -Vtip, and the tunnel current is noted IT. Freshly grafted samples were promptly transferred from the air to ultrahigh vacuum after molecular grafting and degassed for 2 h at 400 K to remove traces of water contamination and improve the tip-sample tunnel contact. All STM images were processed by a simple background plane correction and were not corrected from samples thermal drift that can be neglected on the large scale images presented in this study. 2.5. X-ray and Ultra-Violet Photoemission Spectroscopy Analysis. XPS and UPS measurements were carried out with a base pressure in the 10-10 mbar range. The analysis chamber was equipped with an Omicron UVS 300 high-current UV lamp, a VSW twin anode X-ray source and an Omicron HA100 electron energy analyzer. For the valence band measurements, He I (21.4 eV) and He II (40.8 eV) radiations of the UV light were used. For the near valence band edge measurements, the electron pass energy of the analyzer was set at 10 eV in the fixed analyzer transmission mode, providing an overall energy resolution of 0.25 eV, as determined from the width of the Fermi level step of a clean Au sample. For the onset signal measurements, a fixed retardation ratio of 5 was chosen to reduce the resolution at very low kinetic energies and to avoid the saturation of the analyzer counters. XPS spectra were measured using the Mg KR (1253.6 eV) anode source operating at 120 W, and electron pass energy set either at 40 eV for survey spectra, or at 22 eV for resolved spectra. In this latter case, the overall resolution is 1.0 eV. The zero of binding energy of the UPS is referenced to the Fermi edge measured on a clean Au sample in electrical contact with the spectrometer. The binding energy scale of the XPS is such that the Au4f7/2 peak is set at 84.0 (
Re6Sei8 Cluster Cores Monolayers
J. Phys. Chem. C, Vol. 114, No. 43, 2010 18625
0.1 eV, all the given binding energies being then referred to the Fermi level at the sample surface. 2.6. Electrochemical Characterizations. Cyclic voltammetry and impedance spectroscopy measurements were performed with an Autolab electrochemical analyzer (PGSTAT 30 potentiostat/galvanostat from Eco Chemie B.V.) equipped with the GPES and FRA softwares in a homemade three-electrode Teflon cell. The working electrode, modified Si(111), was pressed against an opening in the cell bottom using a FETFE (Aldrich) O-ring seal. An ohmic contact was made on the previously polished rear side of the sample by applying a drop of an In-Ga eutectic (Alfa-Aesar, 99.99%). The electrochemically active area of the Si(111) surface (namely 0.25 cm2) was estimated by measuring the charge under the voltammetric peak corresponding to the ferrocene oxidation on Si(111)/H and compared to that obtained with a 1 cm2 Pt electrode under the same conditions. The counter electrode was a platinum grid and the system 10-2 M Ag+ | Ag in acetonitrile was used as the reference electrode (+0.29 V versus aqueous SCE). All reported potentials are referred to SCE (uncertainty (0.01 V). Tetra-n-butylammonium perchlorate Bu4NClO4 (Fluka, puriss, electrochemical grade) was used at 0.2 mol L-1 as supporting electrolyte in dichloromethane, previously distilled over P2O5. The (CH2Cl2 + 0.2 M Bu4NClO4) electrolytic medium was dried over activated, neutral alumina (Merck) for 30 min, under stirring, and under argon. About 20 mL of this solution was transferred with a syringe into the electrochemical cell prior to experiments. All electrochemical measurements were carried out inside a homemade Faraday cage in the dark, at room temperature (20 ( 2 °C) and under a constant flow of argon. For impedance spectroscopy measurements, the amplitude of the ac voltage was 10 mV. The differential capacitance C was determined from the imaginary part of the complex impedance Z′′ (C ) -1/ 2πfZ′′ in the frequency range f (typically 50 kHz to 500 Hz) in which the phase angle of the complex impedance was greater than 80°, that is, the range for which the system behaved primarily as a combination of capacitive circuit elements. 2.7. Electrical Transport Measurements. Current densityvoltage J(V) characteristics of Hg/OML-Si junctions were measured with a homemade setup using a Kemula electrode for hanging Hg drop production. The top contact area of the Hg drop (99.999% from Fluka) was adjusted at 3 × 10-3 cm2 (600 µm diameter), as observed with a magnifying lens (10×) and a CCD video camera. An ohmic back contact was made by applying a silver paste electrode on the scratched Si backside. Measurements were performed in the dark, in a glovebox environment under moderate humidity (RH ) 22-25%). Numerical derivation of J(V) provides the differential conductance G(V). The current can be modeled by a thermionic emission mechanism over the barrier at the semiconductor (SC)-metal interface;45 the effective barrier height, Φeff (related to the metal work function and the SC electron affinity), and the diode ideality factor, n (related to the interface states and barrier height inhomogeneities), are fitted to J(V) measured in the forward bias regime (typically 0 < V < 0.3 V) using
( ) (
J(V) ) AT2 exp -
)[
( )]
Φeff qV qV 1 - exp exp kBT nkBT kBT
(1)
Figure 2. Raman spectra showing characteristic peaks ascribed to Se6Re8 cluster obtained on (a) Re6Sei8(TBP)a4(OH)a2 reference powder, (b) Se6Re8/undecanoic acid/porous Si(100), and (c) undecanoic acid/ porous Si(100).
where A is the Richardson constant (A ) 112 A cm-2 K-2 for n-type silicon, A ) 32 A cm-2 K-2 for p-type silicon), q the electronic charge, kB the Boltzmann constant, and T the absolute temperature. The other mechanism that controls the current flow through the Hg/OML-Si diode is tunnelling. The Simmons model46,47 has been widely used in the context of OML/SC junctions.48-51 The tunneling regime can be parametrized to discriminate barrier height (φT), barrier thickness (dT), and free carriers effective mass (m*) effects in the bias regime where the tunnelling mechanism is limiting the current density. For medium-doped Si substrates, the series resistance (Rs) that limits the current at high forward bias is also taken into account to model the current. 3. Results and Discussion 3.1. Raman Analysis. The characterization of M6Li8La6 units by Raman spectroscopy is not usual but the few published results evidence typical vibrations in the 50-400 cm-1 region.52,53 Their assignment is particularly tricky when the symmetry of the M6Li8La6 units deviates from the Oh symmetry. On the basis of the experimental and theoretical Raman investigations performed on Re6Qi8La6 cluster units based compounds by Gray et al.,53 it turns out that the absorption bands correspond to different kind of La6, Qi8, and Re6 breathing modes. The Raman spectroscopy is a very useful technique to detect very small quantities of matter. Indeed, Raman spectroscopy was performed with modified porous Si(100) surfaces to probe the immobilization of the metal cluster on silicon. Such surfaces enable one to graft a larger amount of cluster than on flat silicon using similar chemistries; consequently results are comparable. This technique was preferred to FTIR spectroscopy to observe the vibration bands characteristic from the metal cluster below 400 cm-1. The silicon substrates were etched in HF to produce a highsurface-area porous layer.54 The spectrum recorded for the trans-Re6Sei8(TBP)a4(OH)a2 precursor (Figure 2a) is typical of data reported in the literature for Re6Si8La6 derivatives (Q ) S or Se). It is worth noting that the vibrations of the Re6Sei8La6 cluster unit are found in the Raman spectra of undecanoic acid-modified porous Si(100) after treatment with Re6Sei8(TBP)a4(OH)a2 as shown in Figure 2b. The two spectra represented in Figure 2a,b are almost superposed. The very slight evolution in the position of the bands may be attributed to lower symmetry for the grafted unit than in the starting precursor. From this Raman analysis, it is clear that the Re6Se8 cluster core keeps its integrity during the grafting procedure. 3.2. Ellipsometry Measurements. Since the few grafting experiments performed on p-Si(111) give essentially the same results as found with n-Si(111), only the latter series will be
18626
J. Phys. Chem. C, Vol. 114, No. 43, 2010
Cordier et al.
Figure 3. Spectroscopic ellipsometry data for mixed alkyl/acidterminated monolayers grafted on Si(111), (top) before and (bottom) after immobilization of Re6Se8 clusters. Note that different Re6Se8 cluster coverage was obtained for SiRe100 junctions: ΣSeRe ) 0.65 × 1014 cm-2 (sample 5), ΣSeRe ) 0.40 × 1014 cm-2 (sample 9).
discussed. As shown in Figure 3a, optical properties of all mixed alkyl/acid-terminated monolayers (SiAC100, SiAC33, SiAC10, and SiAC5) grafted to n-Si(111) are very similar. The fitted optical thickness dopt in the range 1.2 to 1.5 nm is consistent with the geometrical length of C12 linear molecules, including a small tilt angle, and no systematic dependence as a function of the acid dilution is found. Figure 3b shows that SE properties of Re6Se8-modified Si(111) surfaces depend on the initial acid dilution with a clear difference observed on the SiRe5 and SiRe100 spectra. This dependence reflects variations in the cluster coverage value. A large refractive index value (n ) 1.71 to 1.94) is found for the denser Re6Se8(TBP)4(OH)2 cluster layers and their fitted optical thickness is dopt ) 1.6 ( 0.1 nm. This optical thickness is consistent with XPS data but appears to be slightly larger than the typical size derived from crystallographic data reported on the Re6Sei8(TBP)a4(OH)a2 building block. As shown in Figure 1, the Re6Sei8(TBP)a4(OH)a2 moiety is included in an oblate spheroid with diameters roughly equal to 0.8 nm (O-O distance) and 1.8 nm (CH3-CH3 distance).33 3.3. STM and STS Data. Figure 4a presents a 500 × 500 nm2 large scale STM image of a mixed alkyl/acid-terminated monolayer-modified Si(111) (SiAC33), before grafting of the Re6Se8 cluster. Large flat terraces (typical width 200 nm) are observed and separated by well-defined parallel steps. The heights-histogram calculated from this image displays 4 distinct peaks, separated by a constant value of 3.0 ( 0.1 Å that matches the Si(111) inter-reticular distance. The observed steps are thus monatomic steps of the underlying Si(111) substrate, and the alkyl/acid-terminated monolayer grafting has preserved the substrate surface morphology. The corresponding root-mean square roughness on this image is 0.27 nm. Small scale STM images measured on a Si(111) monatomic terrace present a flat molecular layer with a tiny and randomly distributed height
Figure 4. (a) A 500 × 500 nm2 STM image (Ugap ) +3.207 V, IT ) 0.202 nA) of SiAC33. Atomic steps of the underlying Si(111) substrate are clearly visible; (b) 200 × 200 nm2 STM image (Ugap ) -3.27 V, IT ) 0.040 nA) of the previous sample after the Re6Se8 clusters grafting (SiRe33). The clusters film has a surface coverage close to unity. The corner of a Si(111) terrace is still visible in the lower right corner of the image; (c) 200 × 200 nm2 STM image (Ugap ) 3.93 V, IT ) 0.021 nA) obtained on SiRe5. The clusters density has been decreased by reducing the anchoring sites density. Isolated clusters are now visible on the surface.
modulation of only 0.25 nm in amplitude. This subnanometric topographic modulation, much smaller than the grafted molecular chains length, attests of the absence of pinholes in the molecular layer and is most likely related to the difference between the acid and alkyl chains length. Similar surface morphology was obtained for a 5% diluted acid layer. From these STM observations, we deduce that the acid-alkyl grafting leads to the formation of a dense and continuous molecular monolayer. The acid and alkyl chains are randomly distributed in the molecular monolayer as expected for this covalent grafting. After grafting of the Re6Se8 cluster, the surface morphology observed by STM on the SiRe33 sample presents major modifications (Figure 4b). A compact granular assembly
Re6Sei8 Cluster Cores Monolayers
Figure 5. Survey XPS spectra before (top, SiAc33) and after (bottom, SiRe33) grafting of Re6Se8 cluster on the mixed al kyl/acid-terminated monolayer.
is now observed, leading to an almost complete coverage of the sample surface. Two Si(111) underlying atomic terraces can still be distinguished at the top and bottom part of Figure 4b. The observed granular morphology is unambiguously attributed to the grafting of the Re6Se8 clusters on the acid chains. Clusters cannot be moved by the STM tip scan with the tunnelling parameters used in this study, confirming the covalent nature of the surface anchorage. The smallest granular medium has an apparent radial extension ranging from 2 to 4 nm but this value cannot be considered as representative of the cluster size due to (i) the observed clear tendency for the clusters to aggregate laterally via π-π stacking interactions between tert-butyl pyridine groups and (ii) experimental convolution of the clusters morphology with the STM tip shape, which can lead to significant size overestimation.55 It should be also noted that some of the largest clusters aggregates observed in Figure 4b present a higher height compared to the rest of the clusters assembly. This could be due to the limited presence of clusters vertical stacking (local formation of cluster bilayers by π-π stacking interactions) or elastic deformation of the molecular chains. As expected, the clusters surface density can be efficiently tuned by changing the acid concentration in the mixed acid-alkyl monolayer. Figure 4c shows a 200 × 200 nm2 image obtained on a cluster assembly grafted on a mixed monolayer prepared using a molar fraction of 5% undecylenic acid in the alkene mixture. The surface clusters concentration has strongly decreased compared to the previous almost continuous layer. From Figure 4b,c, the proportion of the sample’s surface covered by clusters has decreased by a factor 2, due to the reduction of the acid anchoring sites. Finally, discrete clusters can be observed in Figure 4c. The Re6Se8 cluster size can be precisely measured on these single objects by simply measuring the height between cluster and the underlying flat surface. The average Re6Se8 cluster height measured on seven of these isolated clusters is 1.0 ( 0.1 nm and coherent with the expected cluster dimensions (Figure 1).33 The cluster bandgap measured by scanning tunneling spectroscopy on isolated clusters is of 4.0 ( 0.1 eV. 3.4. X-ray Photoelectron Spectroscopy. XPS measurements have been performed on the mixed monolayers at various dilutions of the acid layer, before and after grafting the Re6Se8 cluster. Typical survey scans are shown in Figure 5 after each grafting step. Besides the Si substrate signal, only C1s and O1s signals appear after the first step, while after the second step, characteristic signals originating from the Se, Re, and N atoms indicate the successful grafting of Re6Se8(TBP)4 moieties. On
J. Phys. Chem. C, Vol. 114, No. 43, 2010 18627
Figure 6. XPS spectra showing the evolution of oxygen (left panel), Se and Re intensities (right panel) after grafting of Re6Se8 clusters on different mixed monolayers with variable acid dilution. The residual oxygen has been measured on a pure hexadecyl monolayer grafted on Si(111).
n-type Si(111), the Si2p3/2 maximum appears at 99.6 eV and there is no evidence of oxidation of silicon surface at the first grafting step. Resolved C1s spectra clearly show the main peak at 285.3 eV due to C-C bonds, and a COOH component at 290.1 eV. The intensity of this latter component decreases as the molar fraction of undecylenic acid. All the energies given above are shifted to lower binding values by 0.2 eV on p-type as compared with n-type Si(111). On the cluster grafted surfaces, the N/Re and Re/Se XPS-signal ratios correspond well to the expected values. This constitutes a new experimental evidence that the Re6Se8 cluster is not altered after grafting. As the cluster is grafted to the acid function, we observe the energy shift of the component at 290.1 eV to lower binding energy, corresponding to some contribution of the OdCORe component. A good correlation is found between the Re6Se8 coverage and the COOH headgroup concentration initially present on the surface (Figure 6 and Table 1). An increase of the oxygen on the surface proportionally to the Re6Se8 cluster coverage is observed (Figure 6). This indicates that some H2O molecule produced by the elimination reaction could remain attached to the surface through hydrogen-bond interactions. The total organic layer (acid + alkyl) coverage has been estimated using the total area under the C1s signal.56 The acidterminated chains coverage has been calculated using either the C1s (COOH) or the O1s peaks after the removal of the residual oxygen signal. The coverage of Re6Se8 has been derived from the Re 4f signal. The coverage ratios at each step are given in Table 1. The thickness of the organic layer (dOL) grafted at the two steps has been estimated from the attenuation of the angular averaged silicon signal after each process according to the law OL OL ) where λSi is the mean free path of the Si2p exp(-dOL/λSi OL ) 3.6 nm for electrons traveling through the organic layer (λSi electrons with a kinetic energy of 1158 eV, 100 eV binding energy).57 The value of the organic layer thickness (dOL) found by this method is in agreement with that deduced from spectroscopic ellipsometry data (dopt) within error bars. The electronic structure at the silicon/molecule interface is first studied by XPS and then by UPS (vide infra). The XPS binding energies are sensitive to any band bending induced by electron transfer. The observed binding energy of the Si2p signal is at EF - 99.68 ( 0.08 eV (EF - 99.51 ( 0.08 eV) in n-type silicon (p-type silicon), after grafting of the mixed alkyl/acidterminated monolayers. There is no change in this binding energy after grafting of Re6Se8 cluster cores. This energy is used
18628
J. Phys. Chem. C, Vol. 114, No. 43, 2010
Cordier et al.
TABLE 1: Calculated Acid Termination Coverage and Total Coverage at the Mixed (Acid + Dodecene) Grafting, and Typical Re6Se8 Cluster Coverage on n-Si(111) Surface As Function of the Acid Dilution in Solution acid coverage ratio θ OH acid coverage ΣOH (1014cm-2) acid + alkyl coverage ratio θ1 Σ1 (1014cm-2) at step 1 θ SeRe cluster coverage ratio ΣSeRe (1014 cm-2) at step 2
acid 5%a
acid 10%
acid 33%
acid 100%a
0.05 ( 0.02 0.39 ( 0.05 0.34 ( 0.02 2.6 ( 0.2 0.013 ( 0.002 0.10 ( 0.02
0.06 0.46 ( 0.05 0.41 ( 0.02 3.2 ( 0.2 0.029 ( 0.002 0.22 ( 0.02
0.10 ( 0.01 0.82 ( 0.05 0.39 ( 0.02 3.0 ( 0.2 0.035 ( 0.002 0.27 ( 0.02
0.38 ( 0.02 2.8 ( 0.2 0.38 ( 0.02 2.8 ( 0.2 0.053 ( 0.002 0.42 ( 0.02b
a Similar results have been obtained on p-type Si(111). b This coverage corresponds to the expected coverage for a compact cluster layer. Cluster coverages up to 0.66 ( 0.02 × 1014 cm-2 have been attained on SiRe100 samples.
Figure 7. (a) UPS spectra of SiAC5 (line), SiRe5 (dots) and n-type Si(111)-H (open circles). The right panel shows the features corresponding to the Si valence band and those of the organic layers below the Fermi level. The arrow indicates the position of the HOMO of the organic layers, which is about 4.0 eV under the Fermi level. The left panel shows the onset of the photoemission spectra. The shift on the photoemission onset indicates a +0.40 eV dipole after the grafting of the mixed alkyl/acid-terminated monolayer, then an additional +0.17 eV dipole after the cluster immobilization. (b) Proposed band diagram after the organic layer grafting as deduced from UPS and XPS measurements on n-type Si (left panel) and on p-type Si (right panel), the gray zones indicates the uncertainty in the determination of the LUMO position.
to determine the band bending at the surface after grafting. Assuming EVBM - ESi2p ) 98.74 eV,58 we obtain EF - EVBM ) 0.91 eV for n-type (EF - EVBM ) 0.77 eV for p-type) (Figure 7b), that is, a nearly flat-band condition is found for n-type Si, while a downward band bending occurs for p-type Si. At this stage, the fact that the Fermi level does not move after the Re6Se8 grafting suggests that it can be pinned by defects or interface states. However, we will see below that the modification of the band bending of grafted silicon with an electrode top contact indicates an unpinned position of the Fermi level. 3.5. UV Photoelectron Spectroscopy. UPS characterizes the highest occupied states of the OML-Si system (without top
contact). As already reported,59 and as it will be shown in next sections, it is clear from J-V and electrochemistry measurements that a metallic top contact changes the electronic structure of the system, particularly the band bending in the semiconductor. However, since the molecules are attached chemically to the silicon surface and only physically to the electrodes used here, most of the electronic nature of the junction, that is, the position of the molecular levels relative to the semiconductor band edges, should be captured by UPS. Figure 7a shows UP spectra (He II) with binding energy scale for the clean hydrogen-terminated n-Si(111)-H, SiAC5 and SiRe5 surfaces. The left- and right-hand side of the energy
Re6Sei8 Cluster Cores Monolayers distributions shows the onset of the low kinetic energy secondary electron peak and the valence band structures, respectively. The valence band spectra for the mixed alkyl/acid-terminated chains are similar to those already reported in the literature for linear alkyl chains on silicon.60 The slight differences in the relative heights between our measurements and the literature could be due to the COOH and/or OH functional headgroups. After grafting Re6Se8 cluster, the spectra are not modified in a significant manner, even for larger cluster coverage (not shown). The signal within 4 eV interval just below the Fermi level can be attributed to either states arising from silicon and penetrating the organic layer,61 or to a slight degradation of the organic layer due to irradiation.62 The arrow at the knee indicates the position of the HOMO.61 The ionization potential (IP) is obtained by subtracting the total width of the valence band spectra from the photon energy, IP ) hν - (Ecutoff - EVBM). Ecutoff and EVBM represent, respectively, the energy position of the secondary electron onset and the valence band maximum (VBM) of the substrate surface relative to EF. IPOL for the organic layer is obtained by replacing EVBM by the low binding energy edge of HOMO (EHOMO) in the previous equation, and a value of 8.0 ( 0.5 eV is obtained, close to the values reported in the literature.63,64 The shift of the photoemission onset indicates a +0.40 eV dipole after the acid-alkyl grafting, then an additional +0.17 eV dipole after the cluster grafting. These shifts can be interpreted as arising from interface dipoles (here positive dipoles pointing outward). Using these values, a diagram of the bands alignments is proposed in Figure 7b for n- and p- type silicon. The energy positions ECBM is determined using the electrical gap of Si (1.12 eV). On the other hand, as the ELUMO of the organic monolayers covalently bonded to Si is not well-known, the transport gap (ELUMO - EHOMO) has been taken from published data on similar systems59 and derived from the scanning spectroscopy data after Re6Se8 grafting. With the absence of a top contact, for n-type Si the barrier height for electrons is due to the organic layer, as silicon has flat bands. On the contrary, there is an energy barrier height for the holes on p-type silicon due to the curvature of the bands in the semiconductor. After grafting, the Fermi level lies nearly at the same position at the semiconductor-OL interface. We can see that the LUMO states of the SeRe cluster deduced from this diagram are very close in energy to the minimum of the conduction band ECB. 3.6. Electrochemical Characterizations of Rhenium ClusterModified Si(111) Surfaces. Typical cyclic voltammograms of the rhenium cluster-terminated monolayers in CH2Cl2 + 0.2 M Bu4NClO4 are shown in Figure 8a. In the dark, a single reversible redox system at 0.83 V versus SCE was observed for the modified p-type surfaces whereas no anodic current was measured for the modified n-type surfaces, as expected for a semiconductor under depletion conditions, that is, when few majority charge carriers are available for charge transfer.65 Since the acid-terminated monolayers do not show any oxidation peak within the same potential range, this system can be undoubtedly ascribed to the one-electron oxidation of the bound rhenium clusters and the measured formal potential E°′ (average of anodic and cathodic peak potentials) is approximately similar to that observed for the metallic cluster in solution, namely 0.81 V versus SCE. However, the peak-to-peak separation ∆Ep measured at low scan rates for pSiRe100 is ca. 50 mV, which is larger than the zero value expected for surface-confined reversible redox species.66 Such features can be ascribed to a decrease in the apparent rate of electron transfer of bound metal cluster after immobilization, due to the insulating character of the
J. Phys. Chem. C, Vol. 114, No. 43, 2010 18629
Figure 8. (A) Cyclic voltammograms at 0.1 V s-1 in the dark of pSiRe100 (solid line) and pSiAC100 (dotted line) in CH2Cl2 + 0.2 M Bu4NClO4. (Dashed line) Cyclic voltammogram at 0.1 V s-1 of Re6Se8(TBP)4(COOC10H19)2 at 1.25 mM on Pt (1 mm diameter) disk electrode. (B) Cyclic voltammograms at 0.1 V s-1 in the dark (black) and under illumination of nSiRe5 (magenta), nSiRe10 (purple), nSiRe33 (blue), and nSiRe100 (red). The noise observed on the curves is thought to be caused by fluctuations in the intensity of the light source.
molecular chain between the metal cluster and the underlying silicon surface. Moreover, the bound rhenium cluster is found to be oxidized at a 100 mV lower potential as compared with that of a Mo6-based cluster covalently attached to similar Si(111) surfaces.31,32 Upon illumination, a significant oxidation current response is observed for the rhenium cluster-modified n-type surfaces (Figure 8b), the intensity of which increases with the molar fraction of undecylenic acid in the starting alkene mixture. Such a result is consistent with an increase in the surface coverage of the metal cluster with the fraction of reactive acidterminated chains in the mixed monolayers. Nevertheless, a welldefined anodic peak is obtained only for the nSiRe100 surface at ca. 0.60 V while no such peak is observable for the other modified electrodes within the investigated potential range due to reduced electron transfer kinetics. In contrast with pSiRe100, it is also noticeable that the reverse potential scan shows no significant reduction current that can be ascribed to a low concentration of majority carriers (i.e., electrons) near the surface of the n-type semiconductor electrode.67 To obtain further insights on the electrical properties of the rhenium cluster-modified Si(111) surfaces, differential capacitance measurements have been performed in the same electrolytic medium. To determine the energy levels of the semiconductor bands, it is essential to estimate the flatband potential Efb of the silicon surface, that is, the electrode potential for which there is no space-charge region in the semiconductor. This parameter has been estimated from the commonly used Mott-Schottky plot (Csc-2 vs E, eq 2) that gives the spacecharge capacitance Csc as a function of the electrode potential E under depletion conditions (i.e., depletion of conduction band electrons or valence band holes, respectively, in the space charge region of the n- or p-type surface)45
18630
J. Phys. Chem. C, Vol. 114, No. 43, 2010
C-2 sc )
(
k BT 2 E - Efb 2 q qεε0NA
)
Cordier et al.
(2)
where ε is the relative permittivity of silicon (11.7), ε0 is the permittivity of free space, N is the dopant density of the semiconductor (expressed as Nd the donor density of the n-type Si(111) or Na the acceptor density of p-type Si(111)), A is the area of the electrode, kB is the Boltzmann constant, T is temperature, and q is the electronic charge. Linear Csc-2-E plots are obtained for potentials above -0.5 V and below -0.5 V for the modified n- and p-type surfaces respectively, the intercept and the slope of the curve enable the flatband potential and the dopant density to be determined. Representative linear plots are shown in Figure 9 for nSiRe10 and pSiRe10. The N values obtained for the two surfaces are consistent with the dopant densities derived from the four-probe resistivity measurements of these silicon samples (Nd ) (5.8 ( 0.8) × 1014 donor cm-3 and Na ) (1.4 ( 0.2) × 1015 acceptor cm-3 for the n- and p-type surfaces, respectively). The calculated values of Efb are -0.75 ( 0.05 and +0.05 ( 0.05 V versus SCE for the rhenium clustermodified n- and p-type surfaces, respectively, and are quite similar to those measured for the corresponding acid-modified surfaces (Table 2). Using the band bending in grafted silicon deduced from XPS without the electrode top contact (Figure 6b), the above flat band voltage values show that the electrical contact for electrochemical characterization induces an upward bending of 0.75 eV in both n- and p-type silicon. This displacement is due to a charge transfer required to equilibrate the Fermi level through the structure. The n-type silicon presents then a barrier height of 0.75 eV, and p-type silicon tends toward a very low barrier height. The fact that the Fermi level moves substantially in the band gap is indicative of an unpinned position. This, indirectly, underlines the low interface states density and the good quality of the samples. 3.7. Transport Characteristics of Hg-Re6Se-OMLSi(111) junctions. Current density-voltage J(V) characteristics of Hg/OML-n-type Si junctions show a strong rectification for all mixed alkyl/acid-terminated layers (SiAC100, SiAC33, SiAC10, and SiAC5). Similar characteristics in the forward bias are found for low Re6Se8 cluster coverages. The large values of the rectification factor, R ((1 V) ≈ 1 × 107, are due to a very small reverse current density JREV(1 V) < 10-7 A · cm-2 (Figure 10a) while the forward conductance saturation at 0.1 Ω-1 · cm-2 is due to a series resistance effect (Figure 10b). Indeed, we can see from the fitting curve to the forward current voltage characteristics of SiRe33 (Figure 11), that the current is limited by a series resistance at high voltage, while at low voltage, a Schottky behavior dominates (Φeff ) 0.86 eV, n ) 1.6). In the intermediate voltage domain, the current is limited by tunneling through the organic layer (tunneling barrier 0.60 eV, thickness d ) 2.1 nm). In contrast, a strong coverage dependence is found for the J(V) characteristics of clusterimmobilized surfaces; the reverse current density increases monotonously in the series SiRe5, SiRe10, SiRe33, and SiRe100, up to 2 × 10-3 A · cm-2, as the coverage increases in the range 1-6 × 1013 cm-2. Since junctions SiRe5, SiRe10, and SiRe33 with the lower coverage essentially behave as their parent (acid-terminated) junctions, in the following we essentially discuss SiAc100 and SiRe100 devices. On low doped n-type Si, the acid-terminated OML junctions are strongly rectifying devices with a low reverse current density, JREV < 10-7 A · cm-2 corresponding to a high effective Schottky barrier, Φeff ≈ 0.90 ( 0.05 eV; this value is consistent with the average flatband voltage, Efb ) 0.75 ( 0.05 eV, derived from the
Figure 9. Typical Mott-Schottky Csc-2-E plots at 15 kHz of nSiRe10 (A) and pSiRe10 (B) surfaces in CH2Cl2 + 0.2 M Bu4NClO4.
TABLE 2: Flatband potentials Efb (V versus SCE) Determined from Linear Mott-Schottky Plots at 15 kHz of Acid- and Rhenium Cluster-Modified Si(111) Surfaces in CH2Cl2 + 0.2 M Bu4NClO4
a
surface
n-type
p-type
SiAC5 SiAC10 SiAC33 SiAC100 SiRe5 SiRe10 SiRe33 SiRe100
-0.72 -0.85 -0.75 -0.72 -0.73 -0.80 -0.70 -0.72
a +0.05 a +0.10 a 0.05 a 0.00
Not performed.
Mott-Schottky plot (Figure 9) and with the bulk Fermi level position at 0.25 eV below the conduction band. Both the effective barrier height and ideality factor (n ) 1.30 ( 0.10) are in good agreement with previous studies of C12 alkyl monolayers grafted on low doped n-type Si.65,68 They are indicative of a high quality sample.69 Additional information from p-type Si junctions shows that the difference between the Hg metal work function and the Si electron affinity is a meaningful parameter. For low-doped p-type Si junctions, the acid-terminated OML junctions have a small rectification and a large reverse current density, JREV ≈ 10-2 A · cm-2 corresponding to a low effective barrier, Φeff. After the Re6Se8 cluster grafting step, JREV decreases as observed previously with Mo6I8 clusters31,32 and at high forward bias, a drop in JFWD is observed in the tunnel regime (not shown). This behavior obtained on p-type silicon is expected from the band alignment deduced from photoemission data, however, no signature of the cluster grafting is observed in p-type Si junctions. 3.8. Discussion. In n-type SiRe devices, the signature of the metal cluster in G(V) characteristics is observed for the larger molecular coverage. Using the change in the sign of the G(V) slope near V ) 0, as shown in Figure 12, a threshold coverage
Re6Sei8 Cluster Cores Monolayers
J. Phys. Chem. C, Vol. 114, No. 43, 2010 18631
Figure 12. Re6Se8 cluster coverage dependence of the differential conductance of Hg/OML-n-Si(111) junctions: ΣSeRe ) 0.65 × 1014cm-2 (a), ΣSeRe ) 0.40 × 1014cm-2 (b), ΣSeRe ) 0.15 × 1014cm-2 (c).
Figure 10. Current density-voltage (A) and differential conductancevoltage plots (B) of Hg/OML-n-type Si(111) junctions before (red curve, SiAc100) and after (blue curve, SiRe100) immobilization of Re6Se8 cluster. The characteristics for the Re6Se8 cluster grafted on SiAc10 (black curve) and for SiAC100 are very similar.
Figure 11. Curve fitting of the current voltage characteristics on SiRe33. Open circles represents experimental data, green dots correspond to voltage drop due to series resistance for a given current density, the dark blue dashed and dots curve correspond to the tunneling current calculated using the Simmons model for a Vtunnel voltage across the organic layer, and the dashed red line is the Schottky limited current. As the same current flows through the device, the resulting red line corresponds to the sum of all the voltages for a given current.
value can be defined as the “inversion” coverage, corresponding to a zero-slope G(V) at V ) 0. In this work, a value on the order of 4 × 1013 cm-2 is found for Re6Se8-based junctions. The threshold coverage effect is also observed in electrochemical studies that show that a well-defined anodic peak is obtained
only for the nSiRe100 surface at ca. 0.60 V, while no such peak is observable for the other modified electrodes within the investigated potential range. Hence, this threshold coverage effect is not due to the Hg metal contact alone. For the junctions incorporating the larger cluster coverage, transport characteristics show two main features: (i) a small rectification factor R due to a very large JREV value; (ii) a decrease in the conductance G(V) by more than one decade upon weak forward junction biasing (typically in the range from -0.2 V in the reverse bias regime to +0.5 V in the forward bias regime), the magnitude of the latter effect being dependent on the cluster coverage (Figure 12). To explain these unusual features, three different reasons are examined in the following: (i) The weaker inversion (larger JREV) observed for high cluster coverage could possibly be related to some surface state pinning of the Fermi level, for example, arising from a stronger surface oxidation after the cluster immobilization step. On one hand, this hypothesis can be discarded regarding the fact that this processing step is completely similar for different cluster coverage values (e.g., SiRe10 and SiRe100), which depend only on the acid coverage in alkyl/acid-terminated monolayers. Indeed, comparison of XPS and transport measurements indicates a low interface states density with no efficient Fermi level pinning. Moreover, when we have an evidence of silicon oxidation under a pure alkyl layer, no such behavior on the current voltage characteristics is observed. On the other hand, some observations can imply a role of silicon oxidation in the large JREV values. We have noticed an increase of the reverse current with aging of the SiRe devices with high ReSe coverages, after few days. Moreover, this peculiar behavior of the current has been observed with the Mo6I8 cluster on n-type silicon. In this case, there was clear evidence of silicon oxidation after cluster grafting.32 (ii) The weaker inversion (larger JREV) at high cluster coverage may arise from a change in the work function or interface dipole with redox group coverage. Although only a very small change (0.17 eV decrease in the IP) is observed by UPS after the cluster grafting step, some charge transfer may nevertheless occur upon contact with the Hg electrode. Since the OML is covalently bound to the Si substrate, in contrast with a van der Waals interaction at the Hg contact, application of negative potentials to the n-type silicon electrode is expected to raise the distribution of HOMO levels above the Fermi level at the mercury/OML interface and to provide negative charge transfer from redox groups (QRO > 0) to the mercury surface (QM < 0). This oxidation reaction (or p-type doping of monolayer) leads to a dipole layer (δM/RO) at the mercury/molecule interface, pointing away from the mercury surface. This dipole decreases the barrier height in the reverse bias regime. In this picture, the decrease of the device
18632
J. Phys. Chem. C, Vol. 114, No. 43, 2010
conductance (increase in effective barrier height) as the bias is swept toward the forward regime, is explained by a reduction of the oxidized cluster when the applied bias is in the range from -0.2 to +0.5 V. (iii) While the previous explanation considers homogeneous changes in dipole moments due to the redox properties of the cluster, an alternative explanation can be found in an heterogeneous behavior where parallel transport paths with low and high Schottky barriers coexist on the device due to local variations of the cluster-induced dipoles. In this picture, low Schottky barrier regions dominate the current density in the reverse bias regime, while in the forward regime the tunnel conductance is the rate limiting mechanism. However, surface heterogeneity of the immobilized clusters is not observed in STM images corresponding to high (Figure 4b) and low (Figure 4c) coverage values. A specific effect observed in high cluster coverage junctions is the occurrence of π-stacked cluster layers that could affect the potential distribution in the molecular junction. However, this stacking is unlikely to occur in the Mo6I8 based junctions reported previously, where the minimum in the conductance G(V) has a similar signature. 4. Conclusions An efficient method has been demonstrated to covalently immobilize transition metal clusters (Re6Se8) on modified Si(111) surfaces through linear alkyl spacers; the cluster coverage ranging from 1 × 1013 to 6 × 1013 cm-2 is controlled by using mixed dodecyl/undecanoic acid monolayers. Quantitative XPS analysis, spectroscopic ellipsometry, and STM mapping indicate a covalent attachment of a submonolayer to a dense monolayer of Re6Se8 clusters, while the vibrational Raman signature confirms the cluster integrity within the monolayer. The optical thickness of the Re6Se8 layer is consistent with XPS data, but appears to be slightly larger than the typical size derived from crystallographic data and STM measurements. In Hg/Re6Se8-OML-Si(111) junctions with the denser Re6Se8 molecular coverage, transport characteristics show the following three main features: (i) A small rectification factor R (101-102) due to a very large JREV value, as compared with much larger rectification obtained with acid-functionalized (R > 106) devices. R values decrease with increasing the surface coverage of the metal cluster; (ii) a decrease in the conductance G(V) by more than one decade upon weak forward junction biasing (weak electron depletion, between V ) 0 and the flat band condition V ) VFB). (iii) Observation of the signature of the metal cluster immobilization in G(V) characteristics requires a molecular coverage larger than a threshold coverage value; in this work, a threshold coverage value on the order of 4 × 1013 cm-2 is found for Re6Se8-based junctions. Acknowledgment. This work was financially supported by CNRS, UMR 6226 CNRS/Universite´ de Rennes 1 and Agence Nationale de la Recherche (ANR-07-BLAN-0170-02, Project “CLUSTSURF”). R. Gautier (University of Rennes, France) is acknowledged for frontier orbital calculations. A.L. and N.G.N. are grateful to PECO-NEI ECONET programs for attributions of grants for research at the University of Rennes 1 and A.B.F. acknowledges “Re´gion Bretagne” for a Ph.D. grant. Thanks to the ONIS platform (University Rennes 1) for the use of the HR800 Raman spectrometer, to Dr Bruno Lepine for the development of the I-V curve fitting program, and Arnaud le Pottier for his technical support. References and Notes (1) Selby, H. D.; Roland, B. K.; Zheng, Z. Acc. Chem. Res. 2003, 36, 933–944.
Cordier et al. (2) Gabriel, J.-C. P.; Boubekeur, K.; Uriel, S.; Batail, P. Chem. ReV. 2001, 101, 2037–2066. (3) Prokopuk, N.; Shriver, D. F. AdV. Inorg. Chem. 1999, 46, 1–49. (4) Cordier, S.; Kirakci, K.; Me´ry, D.; Perrin, C.; Astruc, D. Inorg. Chim. Acta 2006, 359, 1705–1709. (5) Hughbanks, T.; Hoffmann, R. J. Am. Chem. Soc. 1983, 105, 1150– 1162. (6) Ramirez-Tagle, R.; Arratia-Perez, R. Chem. Phys. Lett. 2008, 460, 438–441. (7) Golden, J. H.; Deng, H.; DiSalvo, F. J.; Fre´chet, J. M. Science 1995, 268, 1463–1466. (8) Roland, B. K.; Flora, W. H.; Carducci, M. D.; Armstrong, N. R.; Zheng, Z. J. Cluster Sci. 2003, 14, 449–458. (9) Prokopuk, N.; Weinert, S.; Siska, D. P.; Stern, C. L.; Shriver, D. F. Angew. Chem., Int. Ed. 2000, 39, 3312. (10) Me´ry, D.; Plault, L.; Ornelas, C.; Ruiz, J.; Nlate, S.; Astruc, D.; Blais, J.-C.; Rodrigues, J.; Cordier, S.; Kirakci, K.; Perrin, C. Inorg. Chem. 2006, 45, 1156–1167. (11) Szczepura, L.; Ketcham, K. A.; Ooro, B. A.; Edwards, J. A.; Templeton, J. N.; Cedeno, D. L.; Jircitano, A. J. Inorg. Chem. 2008, 47, 7271–7278. (12) Chen, Z. N.; Yoshimura, T.; Abe, M.; Sasaki, Y.; Ishizaka, S.; Kim, H. B.; Kitamura, N. Angew. Chem., Int. Ed. 2001, 40, 239–242. (13) Baudron, S. A.; Batail, P.; Coulon, C.; Cle´rac, R.; Canadell, E.; Laukhin, V.; Melzi, R.; Wzietek, V.; Je´rome, D.; Auban-Senzier, P.; Ravy, S. J. Am. Chem. Soc. 2005, 127, 11785–11797. (14) Shestopalov, M. A.; Cordier, S.; Hernandez, O.; Molard, Y.; Perrin, C.; Perrin, A.; Fedorov, V. E.; Mironov, Y. V. Inorg. Chem. 2009, 48, 1482–1489. (15) Prabusankar, G.; Molard, Y.; Cordier, S.; Golhen, S.; Le Gal, Y.; Perrin, C.; Kalal, S.; Halet, J. F.; Ouahab, L. Eur. J. Inorg. Chem. 2009, 14, 2153–2161. (16) Grasset, F.; Dorson, F.; Cordier, S.; Molard, Y.; Perrin, C.; Marie, A.-M.; Sasaki, T.; Haneda, H.; Mortier, M. AdV. Mater. 2008, 20, 143– 148. (17) Grasset, F.; Molard, Y.; Cordier, S.; Dorson, F.; Mortier, M.; Perrin, C.; Guilloux-Viry, M.; Sasaki, T.; Haneda, H. AdV. Mater. 2008, 20, 1710– 1715. (18) Grasset, F.; Dorson, F.; Molard, Y.; Cordier, S.; Demange, V.; Perrin, C.; Marchi-Artzner, V.; Haneda, H. Chem. Commun. 2008, 4729– 4731. (19) Molard, Y.; Dorson, F.; Brylev, K. A.; Shestopalov, M. A.; Le Gal, Y.; Cordier, S.; Mironov, Y. V.; Kitamura, N.; Perrin, C. Chem.sEur. J. 2010, 16, 5613–5619. (20) Molard, Y.; Dorson, F.; Circu, V.; Roisnel, T.; Artzner, F.; Cordier, S. Angew. Chem., Int. Ed. 2010, 49, 3351–3355. (21) Strother, T.; Cai, W.; Zhao, X.; Hamers, R. J.; Smith, L. M. J. Am. Chem. Soc. 2000, 122, 1205–1209. (22) Ha¨rtl, A.; Schmich, E.; Garrido, J. A.; Hernando, J.; Catharino, S. C. R.; Walter, S.; Feulner, P.; Kromka, A.; Steinmuller, D.; Stutzmann, M. Nat. Mater. 2004, 3, 736–742. (23) Boecking, T.; Kilian, K. A.; Hanley, T.; Ilyas, S.; Gaus, K.; Gal, M.; Gooding, J. J. Langmuir 2005, 21, 10522–10529. (24) de Smet, L. C. P. M.; Pukin, A. V.; Sun, Q. Y.; Eves, B. J.; Lopinski, G. P.; Visser, G. M.; Zuilhof, H.; Sudho¨lter, E. J. R. Appl. Surf. Sci. 2005, 252, 24–30. (25) Joachim, C.; Gimzewski, J. K.; Aviram, A. Nature 2000, 408, 541– 548. (26) Molecular Electronics; Jortner, J., Ratner, M. A., Eds.; Blackwell: Oxford, 1997. (27) Cahen, D.; Hodes, G. AdV. Mater. 2002, 4, 789–798. (28) Cahen, D.; Kahn, A.; Umbach, E. Mater. Today 2005, 32–41. (29) Aswal, D. K.; Lenfant, S.; Guerin, D.; Yakhmi, J. V.; Vuillaume, D. Anal. Chim. Acta 2006, 568, 84–108. (30) Yaffe, O.; Scheres, L.; Puniredd, S. R.; Stein, N.; Biller, A.; Lavan, R. H.; Shpaisman, H.; Zuilhof, H.; Haick, H.; Cahen, D.; Vilan, A. Nano Lett. 2009, 9, 2390–2394. (31) Ababou-Girard, S.; Cordier, S.; Fabre, B.; Molard, Y.; Perrin, C. ChemPhysChem 2007, 8, 2086–2090. (32) Fabre, B.; Cordier, S.; Molard, Y.; Perrin, C.; Ababou-Girard, S.; Godet, C. J. Phys. Chem. C 2009, 113, 17437–17446. (33) Dorson, F.; Molard, Y.; Cordier, S.; Fabre, B.; Efremova, O.; Rondeau, D.; Mironov, Y. V.; Circu, V.; Naumov, N. G.; Perrin, C. Dalton Trans. 2009, 1297–1299. (34) Yarovoi, S. S.; Mironov, Y. V.; Naumov, D. Y.; Gatilov, Y. V.; Kozlova, S. G.; Kim, S. J.; Fedorov, V. E. Eur. J. Inorg. Chem. 2005, 3945–3949. (35) Wade, C. P.; Chidsey, C. E. D. Appl. Phys. Lett. 1997, 71, 1679– 1682. (36) Faucheux, A.; Gouget-Laemmel, A. C.; Henry de Villeneuve, C.; Boukherroub, R.; Ozanam, F.; Allongue, P.; Chazalviel, J. N. Langmuir 2006, 22, 153–162.
Re6Sei8 Cluster Cores Monolayers (37) Mitchell, S. A.; Ward, T. R.; Wayner, D. D. M.; Lopinski, G. P. J. Phys. Chem. B 2002, 106, 9873–9882. (38) Boukherroub, R.; Petit, A.; Loupy, A.; Chazalviel, J. N.; Ozanam, F. J. Phys. Chem. B 2003, 107, 13459–13462. (39) Boukherroub, R.; Wojtyk, J. T. C.; Wayner, D. D. M.; Lockwood, D. J. J. Electrochem. Soc. 2002, 149, H59–H63. (40) Voicu, R.; Boukherroub, R.; Bartzoka, V.; Ward, T.; Wojtyk, J. T. C.; Wayner, D. D. M. Langmuir 2004, 20, 11713–11720. (41) Perring, M.; Dutta, S.; Arafat, S.; Mitchell, M.; Kenis, P. J. A.; Bowden, N. B. Langmuir 2005, 21, 10537–10544. (42) Fabre, B.; Ababou-Girard, S.; Solal, F. J. Mater. Chem. 2005, 15, 2575–2582. (43) Asanuma, H.; Lopinski, G. P.; Yu, H. Z. Langmuir 2005, 21, 5013– 5018. (44) Faucheux, A.; Gouget-Laemmel, A. C.; Henry de Villeneuve, C.; Boukherroub, R.; Ozanam, F.; Allongue, P.; Chazalviel, J. N. Langmuir 2006, 22, 153–162. (45) Sze, S. M. The Physics of Semiconductor DeVices, 2nd ed.; Wiley: New York, 1981. (46) Simmons, J. G. J. Appl. Phys. 1963, 34, 2581–2590. (47) Simmons, J. G. J. Appl. Phys. 1964, 34, 2655–2658. (48) Selzer, Y.; Salomon, A.; Cahen, D. J. Am. Chem. Soc. 2002, 124, 2886–2887. (49) Salomon, A.; Boecking, T.; Seitz, O.; Markus, T.; Amy, F.; Chan, C. K.; Zhao, W.; Cahen, D.; Kahn, A. AdV. Mater. 2007, 19, 445–450. (50) Engelkes, V. B.; Beebe, J. M.; Frisbie, C. D. J. Am. Chem. Soc. 2004, 126, 14287–14298. (51) Wang, G.; Kim, T. W.; Jang, Y. H.; Lee, T. J. Phys. Chem. C 2008, 112, 13010–13016. (52) Schoonover, J. R.; Zietlow, T. C.; Clark, D. L.; Heppert, J. A.; Chisholm, M. H.; Gray, H. B.; Sattelberger, A. P.; Woodruff, W. H. Inorg. Chem. 1996, 35, 6606–6613. (53) Gray, T. G.; Rudzinski, C. M.; Meyer, E. E.; Holm, R. H.; Nocera, D. G. J. Am. Chem. Soc. 2003, 125, 4755–4770.
J. Phys. Chem. C, Vol. 114, No. 43, 2010 18633 (54) Canham, L. T. Appl. Phys. Lett. 1990, 57, 1046–1048. (55) Biro, L. P.; Lazarescu, S.; Lambin, Ph.; Thiry, P. A.; Fonseca, A.; Nagy, J. B.; Lucas, A. A. Phys. ReV. 1997, B 56, 12490–12498. (56) Cicero, R. L.; Linford, M. R.; Chidsey, C. E. D. Langmuir 2000, 16, 5688–5695. (57) Powell, C. J.; Jablonski, A. NIST Electron EffectiVe Attenuation Length Database; National Institute of Standards and Technology: Gaithersburg, MD, 2001. (58) Himpsel, F. J.; Hollinger, G.; Pollak, R. A. Phys. ReV. 1983, B 28, 7014–7018. (59) Shpaisman, H.; Salomon, E.; Nesher, G.; Vilan, A.; Cohen, H.; Kahn, A.; Cahen, D. J. Phys.Chem. C 2009, 113, 3313–3321. (60) Ha¨ming, M.; Zieoff, J.; Salomon, E.; Seitz, O.; Cahen, D.; Kahn, A.; Scho¨ll, A.; Reinert, F.; Umbach, E. Phys. ReV. 2009, B79, 155418. (61) Segev, L.; Salomon, A.; Natan, A.; Cahen, D.; Kronik, L.; Fabrice, A.; Chan, C. K.; Kahn, A. Phys. ReV. 2006, B 74, 165323. (62) Amy, F.; Chan, C. K.; Zhai, W.; Hyung, J.; Kahn, A. J. Phys. Chem. B 2006, 110, 21826–21832. (63) Cahen, D.; Kahn, A. AdV. Mater. 2003, 15, 271–277. (64) He, T.; Ding, H.; Peor, N.; Lu, M.; Corley, D. A.; Chen, B.; Ofir, Y.; Gao, Y.; Yitzchaik, S.; Tour, J. M. J. Am. Chem. Soc. 2008, 130, 1699– 1710. (65) Zhang, X. G. Electrochemistry of silicon and its oxide; Kluwer Academic: New York, 2001. (66) Bard, A. J.; Faulkner, L. R. Electrochemical Methods. Fundamentals and Applications; Wiley: New York, 1980; p 522. (67) Koval, C. A.; Howard, J. N. Chem. ReV. 1992, 92, 411–433. (68) Faber, E. J.; de Smet, L. C. P. M.; Olthuis, W.; Zuilhof, H.; Sudholter, E. J. R.; Bergveld, P.; van den Berg, A. ChemPhysChem 2005, 6, 2153–2166. (69) Seitz, O.; Boecking, T.; Salomon, A.; Gooding, J. J.; Cahen, D. Langmuir 2006, 22, 6915–6922.
JP1071007