Molecular clusters as building blocks for nanoelectronics: the first demonstration of a cluster single-electron tunnelling transistor at room temperature

This content has been downloaded from IOPscience. Please scroll down to see the full text. Download details: IP Address: 130.60.206.42 This content was downloaded on 07/07/2014 at 09:51 Please note that terms and conditions apply. Molecular clusters as building blocks for nanoelectronics: the first demonstration of a cluster single-electron tunnelling transistor at room temperature View the table of contents for this issue, or go to the journal homepage for more 2002 Nanotechnology 13 185 (http://iopscience.iop.org/0957-4484/13/2/311) Home Search Collections Journals About Contact us My IOPscience iopscience.iop.org/page/terms http://iopscience.iop.org/0957-4484/13/2 http://iopscience.iop.org/0957-4484 http://iopscience.iop.org/ http://iopscience.iop.org/search http://iopscience.iop.org/collections http://iopscience.iop.org/journals http://iopscience.iop.org/page/aboutioppublishing http://iopscience.iop.org/contact http://iopscience.iop.org/myiopscience INSTITUTE OF PHYSICS PUBLISHING NANOTECHNOLOGY Nanotechnology 13 (2002) 185–194 PII: S0957-4484(02)24991-0 Molecular clusters as building blocks for nanoelectronics: the first demonstration of a cluster single-electron tunnelling transistor at room temperature S P Gubin1, Yu V Gulayev2, G B Khomutov3, V V Kislov2, V V Kolesov2, E S Soldatov3, K S Sulaimankulov4 and A S Trifonov3 1 N S Kurnakov Institute of General and Inorganic Chemistry, Russian Academy of Sciences, 119991 Moscow, Russia 2 Institute of Radioengineering and Electronics, Russian Academy of Sciences, 101999 Moscow, Russia 3 Faculty of Physics, Moscow State University, 119899 Moscow, Russia 4 Institute of Chemistry and Chemical Technology, National Academy of Sciences, 720071 Bishkek, Republic of Kirgistan Received 17 May 2001, in final form 18 February 2002 Published 14 March 2002 Online at stacks.iop.org/Nano/13/185 Abstract This work is the result of coherent effort of a multi-disciplinary research team working for a considerable number of years in the former USSR in the area of nanocluster molecular electronics. For the first time the successful demonstration of a single-electron tunnelling transistor working reliably at room temperature and based on a single molecular metallorganic cluster is presented. A broad spectrum of different molecular clusters was investigated. Our group has developed a complete cycle of custom-designed molecular cluster manufacturing, deposition, characterization and modification of nanoelectronic devices based on a single molecular cluster. It was shown that the atomic and electronic structure of nanoclusters containing from 3 up to 23 metal atoms had no crucial importance for the transistor fabrication. At the same time extensive research into characteristics of nanoelectronic devices based on single molecular clusters and their tunnelling properties is summarized. 1. Introduction Conventional technologies of microelectronics are limited by the size of a single element (transistor) and the density of the element arrangement on a crystal [1]. First of all it is connected with the resolution of the modern lithographic equipment (limit 5 nm) and the power consumption of semiconductor transistors. Besides, with the transition to the nanoscale the influence of the chemical heterogeneity of layers in which elements are formed as well as surface defects becomes essential. The density of the arrangement of elements on a crystal is limited not only by its size, but also by the high heat dissipation, which may result in circuit destruction. Thus, the search for new alternative ways of electronics development is focusing on the direction of further miniaturization of individual elements, as well as the development of new principles of information processing. Nowadays, the single-electron tunnelling (SET) transis- tor [2, 3] is considered to be one of the candidates for a fu- ture nanocircuit element. The characteristics of such an el- ement are considerably better (smaller power consumption, smaller size) than that of a traditional one. A single-electron transistor working at low temperatures was first fabricated by Kuzmin and Likharev [4] and independently by Fulton and Dolan [5]. Later the single-electron device development us- ing various methods was the subject of much literature [2, 3]. 0957-4484/02/020185+10$30.00 © 2002 IOP Publishing Ltd Printed in the UK 185 http://stacks.iop.org/na/13/185 S P Gubin et al According to these studies the traditional technologies allow one to produce tunnelling nanostructures with relatively large sizes (about 100 nm, that needed cooling up to very low temper- atures [6–11]), a very low reproducibility of junction param- eters and a wide parameter dispersion (for example, granular structure) [12–15]. Thus, the problem is not only the reduc- tion of separate elements sizes, but also the reduction of the parameter dispersion of single-electron junctions. During recent years interest in single-electron devices at high (up to room) temperature has greatly increased [16–31]. New physical processes have been investigated and the theory verified with experiment. In our opinion the new approach of modern nanoelectronics is connected with choosing optimal objects—future nanoelectronic elements—and providing their reproducible working at room temperature5. The possibility of molecular single electronics and bioelectronics was suggested in 1987 [32], but the wide use of their advantages in electronics was not technologically possible then. It is necessary to note that molecules suitable for nanoelectronics applications should have the following characteristics: (1) small sizes, i.e. the electrical capacitance of tunnel junctions should be small enough, if we want the temperature fluctuations to be smaller than the single- electron recharging energy (C < 10−18 F at room temperature) and (2) stability at a single-electron transfer process. We find that so-called small molecular clusters satisfy all these requirements and are the best prospective candidates for applications in nanoelectronics devices. The metallorganic clusters differ from other organic and inorganic molecules in that cluster molecules consist of a compact heavy core of metal atoms (with the symmetry close to spherical) surrounded by a ligand shell of light atoms or simple molecules [33]. The electronic structure of cluster molecules has some peculiarities. On the one hand, the presence of a great quantity of close-together upper-filled molecular orbitals and corresponding low vacant molecular orbitals (both are weakly binding) causes the plural convertible single-electronic transitions in clusters [34] and, on the other hand, provides sufficient stability of the cluster molecule skeleton when adding or removing an electron [35]. Fairly often such cluster molecules are called ‘electron reservoirs’. In particular, this means that as a rough approximation it is possible to consider a cluster molecule as a nanogranule of metal atoms surrounded by a dielectric ligand shell with highest occupied molecular orbital–lowest unoccupied molecular orbital (HOMO–LUMO) separation as in quantum dots. The redox properties of such molecules are determined quite reliably by electrochemical methods and vary over wide limits, which depend on the nature of the metal, ligand shell and cluster structure. The main properties of cluster molecules can be described as follows. (1) Clusters, like any molecular compound, may be purified by routine chemical methods and after purification all molecules of the sample are strictly identical both in composition and in structure. Hence, the parameter 5 The first successful fabrication of a tunnelling nanostructure realizing controlled correlated electron transport at room temperature on the base of a single cluster molecule using STM was demonstrated in 1996 [20–22]. scattering of cluster nanostructures caused by element dissimilarity will be appreciably less than in the case of traditional technologies. (2) The sizes of clusters are very suitable for the production of tunnelling molecular nanostructures. The tunnel barrier of the basic molecular cluster may be noticeably less than for any element formed by methods of modern nanolithography and elements are absolutely identical! Thus, it will allow us to essentially increase the circuit complexity of elements and, hence, to increase the information volume. (3) The currents passing through tunnelling nanostructures from a cluster element are sufficiently small and the time of switching of such an element (time of tunnelling) is also very small—10−12 s. Therefore, the power consumption of one element will be sufficiently low that even at a very high integration level of elements the electrical nanocircuit will not be destroyed. (4) The chemical and physical properties of clusters are already quite well investigated. Now there are various technologies to deposit cluster molecules on a substrate not only by physical (sedimentation, LB technique) but also by chemical methods. (5) Therefore, such cluster molecules can be supposed to be among the most preferable candidates for the role of simple basic ‘elements’ for fabrication of nanostructures for the purposes of future molecular electronics. In this work systematical research on so-called ‘small’ clusters (with a number of metal atoms from three up to 23) (table 1)6 is carried out. The opportunities for their deposition on atomically flat surfaces of highly oriented pyrolitic graphite (HOPG) by various methods are investigated. Their topography and local electronic characteristics are investigated by methods of scanning tunnelling microscopy (STM) and spectroscopy (STS). Single-electron devices (analogues of the diode and the transistor) on their basis working at room temperature were fabricated. 2. Experimental details The various molecular clusters (table 1) were synthesized and purified by routine chemical techniques. Their chemical properties and spectral characteristics (IR spectra in a range of 4000–700 sm−1) were in accordance with the well known standard data. HOPG was used as the substrate for making the tunnelling nanostructures. The preparation of Langmuir–Blodgett mono- layers on a water surface and their transfer to the solid substrate were carried out on the setup described in [38]. Furthermore, specially created LB equipment with a two-section underwater camera [39] was used for the deposition of heterogeneous bi- and multi-layer tunnel nanostructures. The chemically pure solvents tetrahydrofuran (THF), CH2Cl2, CHCl3, stearic acid C18H36O2 from Serva, and 6 Similar research on molecular clusters has already been carried out. As a result of casual adsorption of rhodium and platinum clusters Rh4(CO)12 and [Pt12(CO)24]2− on an HOPG surface separate ‘knobs’ were formed [36]. Also single huge clusters were investigated by STM and STS methods, for example Pt309Phen36O30 [37]. 186 Molecular clusters as building blocks for nanoelectronics Table 1. Molecular clusters investigated: their shapes, dimensions and HOMO–LUMO gaps. Chemical formula Contour and size (Å) �EHOMO–LUMO (eV) References 1 Pt3(CO)3[P(C2H5)3]4 Torus, 15 × 6 a 2 Pt4(CO)5[P(C2H5)3]4 13 × 11 a 3 Pt5(CO)5[P(C2H5)3]4 13 × 11 a 4 Pt5(CO)6[P(C2H5)3]4 Ellipsoid, 13 × 11 2.2 a 5 Pt5(CO)7(P(C6H5)3)4 13 × 11 a 6 Pt17(CO)12(P(C2H5)3)8 20 × 8 a 7 Pd3(CO)3[P(C6H5)3]4 Torus, 15 × 6 ∼2.7 a 8 Pd10(CO)12[P(C4H7)3]6 Sphere, 18 a 9 Pd23(CO)20[P(C2H5)3]8 25 × 25 a 10 (C5H5)4Fe4S4 Sphere, 7 × 7 0.66 b 11 [Fe6C(CO)16]2− + 2∗[(C2H5)4N]+1 9 × 9 1.6 c 12 Carborane, C2B10H12 7 × 7 ∼4.2 d 13 1, 7-(CH3)2-1,2-C2B10H9Tl(OCOCF3)2 10 × 14 e 14 Fullerene, C60 Sphere, ∼8 f a Eremenko N K, Mednikov E G and Kurasov S S 1996 Carbonylphosphine compounds of Pd and Pt Usp. Chim. 54 (4) 671. b Trinh-Toan, Boon Keng Teo, Ferguson J A, Meyer T J and Dahl L F 1977 Electrochemical synthesis and structure of the tetrameric cyclopentadienyliron sulfide dication, [Fe4(η5-C5H5)4(µ3-S)4]2+: a metal cluster bonding description of the electrochemically reversible [Fe4(η5-C5H5)4(µ3-S)4] n system (n = −1 to +3) J. Am. Chem. Soc. 99 408. c Churchill M R, Wormald J, Knight J and Mays M J 1971 J. Am. Chem. Soc. 93 3073. d Lipscomb W N 1975 Boron Hydride Chemistry ed E L Muetterties (New York: Academic) p 39. e Yanovskii A I, Antipin M Yu, Struchkov Yu T, Bregadze V I, Usyatinskii A Ya and Godovikov N N 1982 Structure of thallium bis (trifluoroaceto)-1,7-dimethyl-m-carbonyl (α, α′, -dipyridyl) Izv. Akad. Nauk, Ser. Khim. 293. f Hirsch A 1994 The Chemistry of Fullerenes (Studgard: Time). deionized water (conductivity of 20 M� sm−1) from a ‘Milli- Q’ water treating system from Millipore were used in this work for fabrication of complicated nanostructures. The Nanoscope STM was used for the investigation of the topography and the measurement of current–voltage characteristics (CVCs) of samples at 300 K. The images were recorded in the ‘constant-current mode’ [41] at a tunnel current of 100–500 pA and a tunnel voltage of 150–500 mV. The estimated tip–sample gap (from the experimental dependence of tunnel current on Z-distance by the usual technique, see for example [55]) was about 1 nm. 2.1. Deposition of cluster molecules on the HOPG surface At the first stage of the research it was necessary to develop reproducible methods of cluster deposition on the surface of a solid-state substrate, which allow us to obtain samples with well fixed nano-objects (clusters). The clusters (see table 1) did not contain specific functional groups for attaching on a graphite surface. It is clear that the robust fixing on the surface of cluster molecules in such conditions is due to the presence of graphite surface structure defects and mainly van der Waals interactions. Based on this assumption two main techniques of cluster deposition on a substrate surface were used in this work for producing nanostructures. 2.1.1. Deposition from solutions in organic solvents. Investigated cluster molecules dissolve well enough in such organic solvents as acetone, tetrahydrofuran (THF), chloroform (CHCl3) and methylene chloride (CH2Cl2). All the prepared solutions contained the cluster molecules isolated one after another. After deposition and drying of a diluted solution drop the dissolved cluster molecules remained on the substrate surface. Samples containing isolated cluster molecules of platinum Pt5(CO)6(PBu3)4, Pt17(CO)12(PEt3)8, palladium Pd10(CO)12(PBu3)6 and also iron [Fe6C(CO)16]2− on the HOPG surface were made (STM image shown in figure 1(a)). The cluster size on graphite was determined by STM methods and was compared with the x-ray data of monocrystals of the same compounds. Satisfactory agreement was obtained. However, the STM topography of isolated clusters was not easily reproducible because of a weak molecular adhesion. The STM tip moved them on the surface while scanning. More reliable and reproducible results were obtained by the following method of sample preparation: the HOPG substrate was cleaved along the layer directly inside the cluster solution in the organic solvent (in situ). It was expected that part of the interlayer interaction energy (16.8 J mol−1) was used for the cluster sorption on the freshly cleaved graphite surfaces. After washing and drying, the samples were ready for STM measurements. A typical topography of the sample prepared by such technology is shown in figure 1(b). However, the methods described above allow us to fix the single clusters on the HOPG surface but they are not practically feasible to create well ordered nanostructures from isolated cluster molecules. 2.1.2. Langmuir–Blodgett technique for cluster monolayer formation and cluster fixation on the HOPG surface. We developed a reproducible method allowing us to rigidly fix the clusters with given density on a atomically smooth surface by the Langmuir–Blodgett technique [38]. The Langmuir– Blodgett technology enables us to work with molecules which float on the water surface. A ‘two-dimensional gas’ of freely floating molecules cooperating poorly with each other 187 S P Gubin et al a b c 3D-view of (c) d e Figure 1. STM topography of clusters deposited on a highly oriented pyrolitic graphite (HOPG) surface by various methods: (a) by drying a solution of clusters [Fe6C(CO)16]2−, in chloroform (CHCl3) on an HOPG surface, (b) by cleaving a substrate along the atomic layers inside the solution of clusters [Fe6C(CO)16]2− in chloroform (CHCl3) and subsequent drying of a fresh-cleaved graphite surface, (c) LB technique, single clusters of Pt5(CO)6[PPh3]4 on an HOPG surface, in the three-dimensional view the white–black range corresponds to 1 nm on the z-axis, (d) LB technique, chains of clusters of Pt5(CO)6[PPh3]4 on an HOPG surface, (e) LB technique, two-dimensional lattice of carborane clusters 1,7-(CH3)2-1,2C2B10H9Tl(OCOCF3)2 on an HOPG surface. was formed on the surface while adding a small quantity of amphiphilic substance on the water surface of a Langmuir bath. In the process of ‘two-dimensional gas’ compression (with the help of movable walls of the Langmuir bath) various states of amphiphilic molecules in the monolayer on the water surface may occur (by analogy to a three-dimensional case): ‘gas’, ‘liquid’, ‘liquid-crystal’ and ‘solid’ states [38,40]. The investigated metallorganic clusters were not amphiphilic molecules. Therefore it was necessary to use stearic acid to make a stable monolayer. The LB monolayer was formed of a mixture of solution clusters and stearic acid in chloroform on a water surface. Later the monomolecular film was transferred onto an HOPG surface using a horizontal lift [38]. The cluster monolayer deposition on a substrate was achieved at its contact with the compressed monolayer as a result of its adsorption on the substrate surface. We used such a technique to exclude bilayer formation by Shaeffer method transposition [38]. The obtained LB films were monolayers. Depending on deposition conditions (the surface pressure at the moment of monolayer transfer, cluster concentration in a solution) it was possible to obtain separate cluster molecules on a substrate surface (figure 1(c)), as well as ordered structures (figures 1(d) and (e)). 3. Results of STM study 3.1. Topography The investigation of the graphite surface with deposited cluster molecules by STM methods showed that the topography differs essentially according to the deposition technology. It is possible to distinguish several typical features of a surface profile according to deposition conditions. 3.1.1. Single clusters on HOPG surface. Single separate peaks were observed on flat atomic smooth surfaces of HOPG. Such a peak has size 12 × 10 × 9 Å3 (see figure 1(c)). It may be single metallorganic clusters on the graphite surface, probably covered by solvent molecules. The sizes of investigated clusters were the following: for Pt5(CO)6(PBu3)4, 7 × 10 × 5 Å3; for Pt17(CO)12(PEt3)8, 20 × 7 × 8 Å3; for [Fe6C(CO)16]2−, 9×9×9 Å3. The important argument that the 188 Molecular clusters as building blocks for nanoelectronics a b Figure 2. (a) The Coulomb blockade in an ‘STM tip–cluster–HOPG’ double-junction tunnel system with Pt5(CO)6[PPh3]4 cluster. (b) The Coulomb staircase in an ‘STM tip–cluster–HOPG’ double-junction tunnel system with the carborane cluster molecule 1,7-(CH3)2-1,2-C2B10H9Tl(OCOCF3)2. a b Figure 3. (a) STM topography of the Tl-substituted carborane cluster molecule (1,7-(CH3)2-1,2-C2B10H9Tl(OCOCF3)2). (b) The projection of this cluster molecule onto a lattice of HOPG. above-mentioned single peaks correspond to cluster molecules is really made by the current–voltage characteristics (CVCs) recorded by the STM tip precisely on top of a peak. A typical picture is shown in figure 2(a) [42]. CVCs at points that are far from the cluster give typical results for a flat graphite surface. At the same time in CVCs recorded by the STM tip directly above the cluster the distinct blockade region is observed in the vicinity of the point of origin, where the conductivity is suppressed strongly (up to tenfold). Such CVCs are typical for the cluster molecules investigated. The similar single clusters Pd10(CO)12(PBu3)6, Agn and Agn(PPh3)m and their CVCs were observed in [42, 43]. 3.1.2. Cluster chains. The regular one-dimensional cluster chains were formed on a graphite surface (HOPG) [53]. A typical example is shown in figure 1(d) [42,44]. The distance between peaks is larger (by 20–30%) than the average cluster diameter determined from x-ray study [45]. Such a cluster arrangement is surely more flexible than its arrangement in a solid crystal lattice, where distances between cluster molecules correspond to van der Waals contacts [45]. Usually such chains were formed along the line of cleaving on the graphite surface. Often the cluster aggregates were formed on a graphite surface after the drying of the cluster solution (the first method was described above). The diameter of such a formation was about 100 Å. Probably the cluster chains on the graphite surface were sorbitized along the defects of the deformed crystal lattice. 3.1.3. Two-dimensional structures of clusters. In the case of cluster deposition by LB or Shaeffer methods [38], the obtained samples of nanostructures represented monomolecular layers of stearic acid with incorporated molecular clusters deposited on freshly cleaved HOPG. Their sizes varied from 10 to 50 Å. STM images of a stearic acid monolayer with incorporated clusters are shown in figure 1(e). Analysis of the image (for the samples with metal-containing carborane clusters) allows us to find the lattice parameters of a periodic two-dimensional structure of the electronic density distribution [46]: a = 28.0 ± 4.0 Å, b = 20.0 ± 4.0 Å, ±α = 70◦, that corresponds to that known from x-ray analysis of the same clusters [45]. Thereby it is possible to expect that the 189 S P Gubin et al a b Figure 4. The scheme of a double-junction tunnel structure without the control electrode (a) and with the control electrode (b); 1—the insulator (Al2O3), 2—the gold film. revealed periodic structure in STM images was formed by cluster molecules. At higher magnification it is possible to observe that the projection of peak tops of cluster molecules onto a plane is not symmetric (differs from an ideal circle). A typical picture is shown in figure 3(a). The form and the area of a light spot in figure 3(a) correspond to a projection of the investigated cluster on the HOPG surface. To check this assumption, a cluster projection on a graphite lattice was constructed in the real scale of lengths and angles by using x-ray data [46]. The results are shown in figure 3(b). The dotted curve shows a contour of a white spot in the central part of the STM image of the cluster molecule (figure 3(a)) in the same scale [47]. In our opinion, this is a good argument that this white spot in STM image is just the cluster molecule. It is necessary to note that the recorded images of cluster molecules are well reproduced at repeated scanning of the same region of a sample surface. The STM images of various samples contained similar two-dimensional structures as described above. Thus, it is shown that the separate cluster molecules reliably fixed in an LB monolayer matrix of stearic acid and were suitable for STM measurements. 3.2. Electronic characteristics of single-cluster tunnel systems The method of scanning tunnelling spectroscopy [41] was used to investigate electronic characteristics of the cluster tunnel systems. Basic setups are shown in figures 4(a) and (b). The series of CVCs was recorded for investigation of the electronic transport in the various regions of a substrate: above flat regions of a surface (without clusters), as well as above cluster molecules. At the STM tip position above a cluster molecule some peculiarities were registered on CVCs. On the flat regions of a graphite surface sufficiently smooth CVCs were registered (without any peculiarities). It is necessary to note that these CVCs (for the single molecules and other substrate regions) were made during a single scan and, consequently, in the same environmental conditions (including humidity). This allows us to assign the recorded difference in CVCs solely to the passing of the tunnel current through the single cluster molecule. A typical CVC recorded in the ‘STM tip–cluster– substrate’ system [20–22] at room temperature is shown in figure 2(a). The region with the suppressed conductivity at small voltage on the CVC specifies the realization of a SET mode (‘Coulomb blockade’) [48] in the investigated double- junction tunnel structure. The main single-electronic features of CVCs (blockade, staircase) were well reproducible in time (on one and the same molecule) and with various STM tips and cluster molecules7. Let us consider in more detail the dependence of CVC parameters (value of ‘Coulomb blockade’ and others) on nuclearity, composition, structure and molecule shape (symmetry) of a cluster molecule in the investigated clusters. 3.2.1. Ligand shell influence. The molecular metallorganic clusters can be presented as a compact central kernel consisting of metal atoms surrounded by a light shell composed of the organic (for example hydrocarbonaceous) molecular chains— the ligands. It is necessary to note that there is a fixed number of molecular ligands per metal atom in the cluster molecule. Therefore the bonding in small molecular clusters is stronger than in the case of bulk metal. In our preliminary report [43] it was shown that in the ‘STM tip–molecular cluster–substrate’ nanosystem in the case of using a ‘naked’ cluster (a metal nanoparticle) only one tunnel junction (‘STM tip–cluster core’) was realized. In this case the CVC has a linear shape. For fabrication of the two tunnel junctions (‘STM tip–cluster–substrate’) which are needed for observation of a SET [48] it is necessary to use a molecular cluster surrounded by a dielectric ligand shell. To obtain the required tunnel system the substrate with already deposited ‘naked’ silver clusters was processed by an organic solution containing potential ligands–molecules of triphenylphosphine (PPh3). At the same time the sample topography was not changed, but a strongly pronounced region with suppressed conductivity (‘blockade’) appeared in the CVC as a result of such treatment. The value of the energy gap (HOMO–LUMO separation [33]) was equal to zero for the molecular clusters without a ligand shell, while clusters surrounded by a ligand shell had a gap in the energy spectrum whose value was in the region from 0.7 to 4.6 eV (table 1). Thus, our experimental results show that metallorganic cluster molecules used for the fabrication of reliable single-electron tunnel structures at room 7 We can neglect the influence of environmental conditions (for example the appearance of a ‘fluid’ contamination layer on the sample surface as a result of humidity) in the first approximation for the following reasons. (1) The measured dependence of current through the tip–sample system was exponential, corresponding to the real tunnelling process, which would be impossible in the case of an environment with high conductivity. (2) We had an atomic resolution in all experiments. This would be impossible for a medium with high conductivity. Thus at low conductance of the environment its influence on the electron transport in a tunnel system can be disregarded. 190 Molecular clusters as building blocks for nanoelectronics temperature should have an insulating ligand shell around a metal core (nanoparticle). 3.2.2. Nuclearity and cluster size. The ligand-free metal clusters (nanoparticles) are in several aspects completely different from their ligated counterparts and consequently, as expected, they exhibit different physical properties. In fact, the results of the ligation process are as follows: the energy gap between the highest filled and the lowest unoccupied molecular orbitals is sufficiently small in a ligand-free metal cluster; next after a ligation process the energy gap of a cluster in the ligated form becomes substantial (more than 1.5 eV). Thus the metallic features of the cluster metal core reduced after a ligation process. It is expected that this energy gap decreases when the size of the molecular cluster increases. Such an energy modification in an extended system would correspond to a metal–insulator transition. The ligand-free silver clusters were very suitable for experiments with the ligation process in situ. Nearly half of the atoms in the ‘naked’ silver cluster with a diameter of 3 nm are external and locate on the surface. The very small number of free electrons in this cluster leads to weakening of internal bonds. More effective Ag–Ag metal bonding is encouraged by coordination of the phosphine ligands (PPh3) with the external atoms of the cluster because the lone-pair orbitals of these ligands are effectively overlapped by the metal 5s, 5p and 4d orbitals. The additional electron pairs introduced by the phosphine ligands occupy the lowest orbitals, which are strongly metal–metal and metal–phosphorus bonding. As a result there is a shift of all energy levels (including the LUMO) to high energies and at the same time the relative position of energy levels remains practically unvaried. The reduction of the cluster nuclearity influences the structure of the electronic spectrum. As a result the relative contributions of metal and ligand orbitals in the HOMO–LUMO of the small clusters become comparable. In other words in such small clusters (N < 11–13) it is impossible to distinguish the individual contributions of the molecular ligand and the metal to the energy levels. Research on various cluster molecules (table 1) with ap- proximately identical sizes and structures, but with increas- ing nuclearity, Pd3(CO)3[P(C6H5)3]4, Pt5(CO)6[P(C2H5)3]4 and Pd10(CO)12[P(C4H7)3]6 [49], was carried out. A clus- ter with the same structure but with enlarged size (2.8 nm)— Pd23(CO)20[P(C2H5)3]8—was also studied. The qualitative tendency of variation of the ‘Coulomb blockade’ value with the geometrical size of a molecular cluster is shown in figures 5(a), (b) and (c). The CVC of the nanostructures with the biggest Pd23(CO)20[P(C2H5)3]8 clusters looks like a cubic parabola. This result is in good agreement with the theory that assumes the cluster energy variation to be less than the energy variation of thermal fluctuations. Therefore the ‘Coulomb blockade’ on the CVC in this case is completely diffused. 3.2.3. Structure of cluster energy levels. All previous estimates were made within the framework of so-called ‘orthodox theory’ [48]. The basic statement of this theory is the presence of a continuous electronic spectrum of nano-objects. However, for the majority of investigated clusters it is not Figure 5. CVCs in double-junction ‘STM tip–cluster–HOPG’ tunnel systems with clusters (a) Pd3(CO)3(PPh3)4, (b) Pt5(CO)6(PEt3)4 and (c) Pd10(CO)12(PBu3)6. correct. The distances between energy levels in the spectrum may be higher than the thermal fluctuations energy (∼25 meV) and, hence, it is necessary to take into account the contribution of discreteness of the electronic spectrum to analyse the electronic characteristics. The presence of discreteness of the electronic spectrum has the result of the appearance of additional ‘steps’ (in addition to those caused by ‘charging effects’ [48]) in the CVC. The position of the ‘steps’ is defined by the energy of the corresponding level in the electronic spectrum relative to the Fermi energy. As shown in [50] the distances between energy levels in a molecular cluster with a small number of metal atoms (N < 11) were basically equal to 10–30 meV. However the some single levels were situated far from each other. Thus, it is possible to observe the discreteness of the electronic spectrum of a separate cluster molecule at room temperature. As a result single steps appeared in the CVC in addition to the usual ‘charging effect’ steps [48]. The discreteness of an electronic spectrum of the tunnel nanostructure will also influence the signal characteristics of 191 S P Gubin et al a b Figure 6. The control characteristics of an ‘STM tip–cluster–HOPG’ tunnel system with the control electrode (a) for the carborane cluster molecule (1, 7-(CH3)2-1,2-C2B10H9Tl(OCOCF3)2), (b) for the cluster molecule Pt5(CO)6[PPh3]4. the electronic device. In general (if we take into account both effects, the single-electron effect and the effect of discreteness of the electronic spectrum) the period of the control characteristic �Vgate is equal to the sum of the contributions of each effect [52]: �Vgate = C/Cgate(�E/e + e/Cgate) where �E is the typical distance between energy levels and Cgate the value of control capacitance. The contribution of the first term caused by the discreteness of the electronic spectrum (∼50 mV) is negligible in comparison with the second term. The main defining parameter is control capacity. In our experiment (see below) sufficiently small control capacitance (∼1 × 10−19 F) was realized. Thus the contribution to the control characteristics caused by the discreteness of the level structure in a cluster should be insignificant in this case. 4. SET transistor On the basis of manufactured tunnel nanostructures, we carried out a series of experiments for registration of the controlling effect (the signal characteristics) at room temperature using an independent external source of controlling voltage. In reality this was the demonstration of functioning of a molecular SET transistor at room temperature. To control the tunnel current through the cluster a control electrode was made by electronic nanolithography in the form of a thin bilayer (a layer 50 nm Au deposited on a layer 50 nm Al2O3) strip on an atomically smooth surface of pyrolitic graphite (HOPG). Then the mixed film of stearic acid with incorporated clusters was transferred onto the substrate using Shaefer’s method [38]. Two completely different molecules were used as the basis of a molecular tunnel nanosystem: the first from the metallorganic cluster family (Tl-substituted carborane [20, 51]), the second a Pt5(CO)6[P(C2H5)3]4 cluster [53, 54]. Their availability for molecular single- electron system fabrication is conditioned by the fact that such cluster molecules represent ready ‘elementary cells’ for tunnel nanostructures (the metal core − electronic reservoir + the tunnel barrier − a ligand shell) with fixed well determined parameters. This method essentially simplifies the decision on the important problem of low dispersion of parameters in nanostructures with a large number of tunnel junctions. To produce a molecular matrix the Langmuir–Blodgett (Shaefer) technique [38] was used. This technique allows us to form controllably regular monomolecular layers from various molecular mixtures providing their strong fastening on a substrate. The structure of samples is shown in figure 4(b). If the STM tip is situated directly above a cluster a three- electrode (analogue of the transistor) system is formed in the area, where the electric field of the controlling electrode may influence the tunnel current through the ‘STM tip–cluster molecule–substrate’ tunnel system. The CVC of such a nanostructure in the experiment using a carborane cluster is shown in figure 2(b) [20]. The steps on the CVC are equidistant in current and in voltage, that coincides with the theory (a so-called ‘Coulomb staircase’) [48] and characterizes the CVC of strongly asymmetric (i.e. with a large difference between the tunnel junction conductivities) single-electron transistor. The CVC of the nanostructure with a platinum cluster differs from these results: the presence on the CVC of a ‘Coulomb blockade’ but no ‘Coulomb staircase’ corresponds to a symmetric transistor. A series of control characteristics (i.e. dependence of the tunnel current through a double-junction tunnel nanostructure on the voltage of the control electrode at fixed tunnel voltage on the double-junction structure) were recorded after the CVC measurements. It has been shown that the current through the investigated molecular transistor structure changes periodically for monotonic change of the control electrode voltage (with the period (700 ± 50) mV for a carborane [20] and (2000 ± 100) mV for a platinum cluster [54]). This is shown in figure 6. According to the theory [48] such a behaviour is typical for the single-electron transistor and the period of control characteristics corresponds to the change of an effective cluster charge by one electron charge value. The control characteristics recorded in the regions without any cluster molecules show the absence of any control voltage influence on the tunnel current. It is possible that the difference of the period of control characteristics for the two types of 192 Molecular clusters as building blocks for nanoelectronics cluster can be explained by the difference of the investigated system geometry (the various distances between the control electrode and cluster molecules) and, hence, by the difference of the control capacitance Cgate. The estimate of electrometric sensitivity of the system from the maximal slope of the control characteristic and experimentally observed noise (amplitude peak to peak of ∼150 pA at a bandwidth of 30 kHz and a CVC measurement time of 3 ms) gives the value of ∼7 × 10−4 e Hz−1/2. This value approaches the typical values for traditional thin-film single-electron systems [2, 8]. Obviously the recorded noise was caused mainly by mechanical vibrations in the STM. Using the measured data of the ‘Coulomb blockade’ value, the ‘step’ size, the maximal slope of the control characteristic and the period of the control characteristic we have estimated the transistor junction capacitances and the gate capacitance Cgate (‘control electrode–cluster molecule’). These values are in good agreement with the real geometry of tunnel nanostructure and with the theory of single electronics [48]. The good agreement of our experimental data for the tunnel system with the central electrode in the form of the single cluster molecule and the ‘orthodox’ theory [48] founded for metal electrodes is the obvious evidence that the electron behaviour on transport through a molecular cluster nanostructure and a metal nanoparticle is similar to a strong degree. This may be caused by the fact that the nanoparticle has an electronic spectrum similar to that of a cluster molecule [33]. At high temperatures thermal smearing of energy levels is sufficiently large and therefore at small currents the molecular cluster behaves similarly to a nanogranule of metal (d ∼2–5 nm). The reason for this is the following: the real metal nanoparticle (with the above- mentioned sizes) is always covered (because there is a high surface energy ∼0.5–2 eV per atom [33]) by an adsorbate consisting of light atoms (H, O, C, N) from the external medium. Thus, to the point this is a pseudocluster molecule consisting of a metal core and ligand shell. It is possible to explain the suitability of the semiclassical theory [48] for the qualitative description of the correlated electron tunnelling through such especially quantum objects as single cluster molecules. The experimental results have shown that CVCs of types such as the ‘Coulomb blockade’ are observed almost for all series of investigated cluster molecules (table 1). These cluster molecules differ from each other in the nature of metal atoms (Fe, B–Tl, Pd, Pt) and their number (3–23), in the presence or absence of an extra charge, in the nature of ligand (CO, PPh3, S), in the size (0.7–2.8 nm) and, most importantly, in the value of HOMO–LUMO separation (0.2–4.2 eV). For all the investigated cluster molecules almost the same CVCs are observed. It is possible that this means that the area of energy levels near HOMO–LUMO separation (this parameter is responsible for the chemical reactions and the basic spectral characteristics) is not a governing factor for the electron transport through the cluster molecule, and it determines only the common shape of the I–V curves. The peculiarities of I–V curves may be determined by other parts of the electronic spectrum of the multielectronic cluster molecule. At the same time the agreement of the obtained results with the theory of single electronics should mean that in the investigated quantum objects there is a set of energy levels similar to that of the pseudometal state of nanogranules, and just above these energy levels the electronic tunnel current penetrates the ‘STM tip–cluster–substrate’ system. The cluster molecules differ from other molecular compounds in the set of filled, closely located, weakly binding d–d levels that have a similar set of corresponding vacant orbitals in the top part of the spectrum. It is natural to assume that this set of molecular orbitals determines the electronic transport through the cluster molecules. The external electric field forces the electron to overcome the first barrier of the two-barrier system and to locate on one of the excited cluster molecule vacant levels. It is very important that the mode of correlated electron tunnelling was independently realized on the base of all molecules given in table 1. Thus, we showed that the chemical composition, the atomic and electronic structure of the molecular cluster do not play decisive roles in the choice of cluster molecules. This is very suitable for production of single-electron devices working at room temperature. 5. Conclusions We propose a method of manufacturing nanoelectronic devices based on single cluster molecules. The following conclusions from this study can be made. (1) For all investigated clusters the CVC had a region with suppressed conductivity at small voltage that can be interpreted within the single-electron model of electron tunnelling. The ‘Coulomb staircase’ effect was observed. The CVCs of a single molecular cluster and a flat graphite substrate were different at all times. (2) Despite the appreciable difference in the electronic structure of investigated clusters (HOMO–LUMO gap, ionization potential, electronic spectrum) there was no difference between the current–voltage characteristics of tunnel nanostructures with various molecular clusters. (3) The following tendency was revealed: the value of ‘Coulomb blockade’ decreased with the growth of nuclearity (number of metal atoms in the cluster molecule) in a set of cluster molecules of the same type. (4) The fundamental role of the ligand shell in transformation of the ‘STM tip–cluster–substrate’ system from single junction into double junction was experimentally shown for the first time for metallorganic clusters. (5) We have developed technology that allowed us to produce nanostructures with preset characteristics: they can be separate clusters fixed at certain distances from each other, as well as systems of molecular clusters—groups of several clusters, one-dimensional chains of clusters or two-dimensional regular arrays of clusters. As shown the clusters with nonspherical shape were more suitable for this technology. (6) Using this technology a molecular SET transistor on the base of a single cluster molecule operating at room temperature was fabricated for the first time. It was shown that the SET transistor had high potential sensitivity to the electric charge. This, in turn, allows for a number of effects in two-dimensional planar structures, when the behaviour of these structures (and collective phenomena) will depend on external electromagnetic field [40]. 193 S P Gubin et al (7) It was experimentally shown that the atomic and electronic structure of molecular clusters containing from three up to 23 atoms of any metal has no crucial importance for the realization of the three-electrode scheme of control by correlated electron tunnelling through the cluster nanostructure (SET transistor). For the first time it was shown that small metallorganic cluster molecules could be a basis for producing single- electron nanostructures operating at room temperature as units for nanoelectronic devices. Probably memory chips and pre- cision standards on the basis of such elements (for example, the standard of current) will be the most widely implemented in the future. Acknowledgments This work has been supported by INTAS 99-864 and the Russian Foundation for Fundamental Research (grant 99-03- 32218). References [1] Keyes R W 1988 IBM J. Res. Dev. 1988 32 24 [2] Likharev K K 1999 Proc. IEEE 87 606 [3] Ahmed H and Nakazato K 1996 Microelectron. Eng. 32 297 [4] Kuzmin L S and Likharev K K 1987 JETP Lett. 45 495 [5] Fulton T A and Dolan G J 1987 Phys. Rev. Lett. 59 109 [6] Delsing P, Claeson T, Likharev K K and Kuzmin L S 1989 Phys. Rev. Lett. 63 1184 [7] Geerligs L J, Anderegg V F, Holweg P, Mooij J E, Pothier H, Esteve D, Urbina C and Devoret H 1990 Phys. Rev. Lett. 64 2691 [8] Zorin A B, Ahlers F-J, Niemeyer J, Weimann T, Wolf H, Krupenin V A and Lotkhov S V 1996 Phys. Rev. B 53 682 [9] Keller M W, Martinis J M, Zimmerman N M and Steinbach A H 1996 Appl. Phys. Lett. 69 1804 [10] Pothier H, Lafarge P, Orfila P E, Esteve D, Urbina C and Devoret H 1991 Physica B 169 573 [11] Ralph D C, Black C T and Tinkham M 1997 Phys. Rev. Lett. 78 4087 [12] Wan J C, McCreer K A, Anand N, Nowak E and Goldman A M Coulomb blockade on imaged mesoscopic lead grains Preprint University of Minnesota [13] Klein D, Roth R., Lim A K L, Alivisatos A P and McEuen P 1997 Nature 389 699 [14] Takahashi Y, Namatsu H, Kurihara K, Iwdate K, Nagase M and Murase K 1996 IEEE Trans. Electron Devices 43 1213 [15] Chen W, Ahmed H and Nakazato K 1995 Appl. Phys. Lett. 66 3383 [16] Pashkin Yu A, Nakamura Y and Tsai J S 2000 Appl. Phys. Lett. 76 2256 [17] Altmeyer S, Hamidi A, Spangenberg B and Kurz H 1997 J. Appl. Phys. 81 8118 [18] Shirakashi J, Matsumoto K, Miura N and Konagai M 1998 Appl. Phys. Lett. 72 1893 [19] Nakamura Y, Chen C D and Tsai J S 1996 Japan. J. Appl. Phys. 35 L1465 [20] Soldatov E S, Khanin V V, Trifonov A S, Presnov D E, Yakovenko S A, Gubin S P, Kolesov V V and Khomutov G B 1996 JETP Lett. 64 556 [21] Gubin S P, Kislov V V, Kolesov V V, Soldatov E S and Trifonov A S 1999 Nanostruct. Mater. 12 1131 [22] Gubin S P, Gulayev Yu V, Kislov V V, Kolesov V V, Soldatov E S, Trifonov A S and Yudin S G 1997 J. Commun. Technol. Electron. 42 1303 [23] Takahashi Y, Nagase M, Namatsu H, Kurihara K, Iwdate K, Nakajima Y, Horiguchi S, Murase K and Tabe M 1995 Electron. Lett. 31 136 [24] Ohata A, Niyama H, Shibata T, Nakajima K and Torumi A 1995 Japan. J. Appl. Phys. 34 4485 [25] Leobandung E, Guo L, Wang Y and Chou S Y 1995 Appl. Phys. Lett. 67 938 [26] Leobandung E, Guo L and Chou S Y 1995 Appl. Phys. Lett. 67 2338 [27] Matsumoto K, Ishii M, Segawa K, Oka Y, Vartanian B J and Harris J S 1996 Appl. Phys. Lett. 68 34 [28] Ishikuro H, Fujii T, Sarya T, Hashiguchi G, Hiramoto T and Ikoma I 1996 Appl. Phys. Lett. 68 3585 [29] Matsumoto K 1997 Proc. IEEE 85 612 [30] Andres R P, Bein T, Dorogi M, Feng S, Henderson J I, Kubiak C P, Mahoney W, Osifchin R G and Reifenberger R 1996 Science 272 1323 [31] Porath D and Millo O 1997 J. Appl. Phys. 81 2241 [32] Likharev K K 1987 Mikroelektronika 16 195 [33] Gubin S P 1987 Chimiya Klasterov (Moscow: Nauka) [34] Manning V C and Trogler W C 1981 Coord. Chem. Rev. 38 89 [35] Trinh-Toan, Fenhammer W P and Dahl L F 1972 J. Am. Chem. Soc. 94 3389 [36] Fujimoto T, Fukuoka A, Nakamura Y and Mychikawa 1989 J. Chem. Soc. (Chem. Commun.) 845 [37] Dubois J G A, Shafranjuk S E, Geritsen J W, Boon E J G, Van Kempen H and Shmid G 1996 Europhys. Lett. 33 279 [38] Gaines G L 1966 Insoluble Monolayers at Liquid–Gas Interface (New York: Wiley) [39] Yudin S G et al 1992 Thin Solid Films 210/211 46 [40] Kislov V V et al 1996 Mathematical Methods in Contemporary Chemistry ed S I Kuchanov (Amsterdam: Gordon and Breach) pp 413–32 [41] Hansma P R and Tersoft J 1987 J. Appl. Phys. 61 R1 [42] Gubin S P, Khanin V V, Soldatov E S and Trifonov A S 1996 Inorg. Mater. 32 1111 [43] Gubin S P, Soldatov E S, Trifonov A S and Khanin V V 1997 Mendeleev Commun. 30 [44] Trifonov A S, Yakovenko S A, Khanin V V, Gubin S P, Khomutov G B, Soldatov E S and Obidenov A Yu 1998 Abstracts Int. Symp. on Nanostructures: Physics and Technology (St Petersburg, 1998) p 245 [45] Brage D and Grepioni F 1994 Accounts Chem. Res. 27 516 [46] Yakovenko S A, Trifonov A S, Soldatov E S, Khanin V V, Gubin S P and Khomutov G B 1996 Thin Solid Films 284/285 873 [47] Yakovenko S A, Gubin S P, Khanin V V, Soldatov E S, Trifonov A S and Khomutov G B 1996 Inorg. Mater. 32 1118 [48] Averin D V and Likharev K K 1991 Mesoscopic Phenomena in Solids ed B L Altshuler, P A Lee and R A Webb (Amsterdam: Elsevier) p 173 [49] Soldatov E S, Trifonov A S, Gubin S P, Kolesov V V and Yudin S G 1997 Inorg. Mater. 33 1216 [50] Pacchioni G and Rosher N 1990 Inorg. Chem. 29 2901 [51] Averin D V and Nazarov Yu V 1992 Single Charge Tunneling ed H Grabert and M H Devoret (New York: Plenum) [52] Zubilov A A et al 1994 Pis. Zh. Eksp. Teor. Fiz. 20 41 [53] Khomutov G B, Yakovenko S A, Trifonov A S, Soldatov E S, Khanin V V and Gubin S P 1998 Thin Solid Films 327–329 550 [54] Soldatov E S et al 1999 Abstracts Int. Symp. on Nanostructures: Physics and Technology (St Petersburg, 1999) p 467 [55] Kuk Y 1992 Scanning Tunneling Microscopy I ed H-J Guntherodt and R Wiesendanger (Berlin: Springer) p 17 194