Present-day electronic technology, based to a large extent on silicon fabricated devices, is surely approaching size limitations arising from quantum effects. The effort to achieve rapid development of electronic devices requires implementation of entirely new ideas that will allow existing technological constraints to be overcome. Among a wide range of concepts, utilization of single organic molecules, acting as active blocks performing logic operations, appears as one of the most appealing and is based on the state-of-the-art use of modern nanotechnology. In this short review, we offer a selection of recent experiments addressing milestones of single-molecule computing devices technology. Along with discussion of the latest achievements in atomic- and molecular-scale technologies, we discuss their future perspectives, challenges, and still unsolved issues standing in the way of practical implementation of single-molecule devices.
Concept of a single-molecule electronic device. The molecule is equipped with various functional groups responsible for logic operations and bonding to an underlying surface and metallic nanoelectrodes. The monomolecular nanodevice is deposited on a thin buffer layer assuring the decoupling from a (semi-)conducting substrate.
Present-day microelectronics have an impact on a broad range of human activities, starting with rather obvious connections to information and communication technologies, through its impact on transport and mobility, security and safety, energy, and environment sectors, up to a recognizable importance for health and wellness areas. Development in the field of transistor-based electronic systems is strongly correlated with miniaturization of a transistor itself. However, limits of the latter will be met soon, oversimplifying, because there is a certain, minimum number of atoms on the surface of a semiconductor required to establish the structure of the transistor in a currently accepted form. Therefore, undoubtedly radically new ideas and approaches are required in order to go beyond the boundaries of progress in the conventional electronics. Already in the second half of the 20th century such novel concepts have appeared, just to mention the seminal paper by Aviram and Ratner 1, with a proposal of using a single molecule as a rectifier, that later on evolved into two different approaches to electronic devices, namely, hybrid-molecular and monomolecular 2. In particular, the monomolecular approach resulted in a brand new modus operandi for electronics, that is to say, the idea of integrating the elementary functions and interconnections required for desired computation into a single molecule 2. Furthermore, such molecular electronic circuits or logic gates do not have to follow the topology of classical electrical circuits 3. From the broader perspective, atomic- and molecular-scale devices may be designed as mechanical machines and transducers (see, e.g., Ref. 4), so that their use is not only limited to computational tasks in electronics and telecommunication.
In the following, we will focus on the concept of single-molecule electronics, present approaches to realizations of that idea and challenges that have yet to be faced. Surely our short survey cannot cover all possible fascinating issues emerging upon experimental realization of the atomic- or molecular-scale devices. Thus, we are forced to limit our discussion to some aspects only. In order to demonstrate the complexity of the field, on the one hand, and its universality, on the other, we decided to highlight three aspects; namely: single-molecule manipulation and characterization, the device reliability and outer-world connections. Although these studies directly devoted to the monomolecular electronics necessarily have to pick up the technical and technological challenges, for instance how to contact electrically a single molecule, the response to them includes more fundamental understanding. In particular, two of three chosen aspects, i.e., single-molecule manipulation and characterization and device reliability, strongly depend on our understanding of molecule–substrate interactions. The latter has a great impact on development of new technologies for fabrication of advanced materials with desired bulk and surface properties. Therefore, studies reviewed in the present text are of importance not only for monomolecular electronics, but for many areas of technology such as solar cell industry, ceramics, gas and biosensors, transducers, and many others.
Our review on atomic- and molecular-scale devices and systems in the context of single-molecule electronics is structured as follows: Section 2 is devoted to single-molecule manipulation and characterization, Section 3 to the device reliability, and Section 4 to the outer-world connections and an outlook on future trends. Each of the sections contains a summary of very recent findings in the field, discussion of ongoing work and open questions.
2 Single-molecule manipulation and characterization
Positioning of a single molecule, expected to perform a predefined operation, in a proper place and knowledge of its properties in such a new environment are of crucial importance for manufacturing reliable nanoscale molecular devices. Nowadays, both aspects are successfully addressed by means of scanning probe microscopes. There are numerous excellent reviews and introductions to the present-day state-of-the-art in the scanning probe techniques 5–13. Here, we would like to focus on a few experiments that in our opinion nicely illustrate recent achievements and future challenges.
In the realm of the atomic- and molecular-scale devices, information on position of individual atoms is essential, and this requirement applies both to atoms in the substrate surface and in the molecular device itself. There are two techniques, namely scanning tunneling microscopy (STM) and noncontact atomic force microscopy (ncAFM, also called dynamic force microscopy) that allow imaging with true atomic resolution. The potential of STM stems from the exponential dependence of the tunneling current on the tip–sample distance and is reserved to conducting and semiconducting samples only. On the other hand, due to short-range chemical forces that are measured by ncAFM, this technique is applicable in experiments with metals, semiconductors, and insulating surfaces. Recent excellent experiments by Sugimoto et al. 14, Gross et al. 15, 16, and Weiss et al. 17, 18 reported on unprecedented resolution that has been achieved with the use of both STM and ncAFM.
Sugimoto et al. 14 present successful attempt to identify three different atom species being constituents of an alloy system. The system under scrutiny consists of silicon, tin, and lead atoms that exhibit similar chemical properties and are found in identical surface positions. Although atomic resolution is achieved, topographic images do not allow the different species to be distinguished. The difficulty is overcome with the aid of dynamic force spectroscopy. Spectroscopic measurements allow quantification of the short-range chemical forces that originate from the temporary chemical bond between the outermost atom of the tip apex and the imaged atom of the surface. The forces are measured as a function of the tip–sample distance and thanks to the normalization procedure introduced by Sugimoto et al., chemical identification of individual atoms is enabled. In the normalization procedure, one has to obtain the relative interaction ratio of the maximum attractive short-range forces between pairs of atomic species of the analyzed system. In their experiment, Sugimoto and coworkers have obtained relative interaction ratios of Sn–Si and Pb–Si pairs of atoms that allowed them to deduce the local chemical composition of the examined alloy with true atomic resolution (see Fig. 1).
Gross et al. 15, 16 present a different approach in achieving higher resolution, namely, they aim at proper functionalization of the microscope's tip apex with suitable, atomically well-defined termination. In their work, ncAFM and STM tips are modified by deliberate picking up of a CO molecule. Such a tip ending allows observation of either the complete chemical structure of individual pentacene molecules with ncAFM (see Fig. 2) 15, or ultrahigh-resolution images of the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of naphtalocyanine molecules with an STM 16. It is worth noting here that there are also reports in the literature on other than CO modification of the STM tip (not always intentional) that results in increased resolution of the measured image, for example, with tip apexes functionalized by pentacene 19, helicene 20, C6021, or O222.
In the case of ncAFM measurements of pentacene molecules, reported by Gross et al. 15, the exceptional enhancement of the atomic-scale contrast allowed the authors to resolve the atomic positions and bonds in the molecule. Furthermore, the comparison between ab initio theoretical calculations and experimental results shows that the source of the atomic resolution lies in the Pauli repulsion. The role of the van der Waals and electrostatic forces is reduced only to addition of a diffusive attractive background 15.
Weiss et al. 17, 18 resolved the inner structure of a complex molecule, such as 3,4,9,10-perylene-tetracarboxylic dianhydride (PTCDA), and imaged individual intermolecular bonds in the layer of PTCDA on Au(111) by applying a “chemically sensitized” version of STM. The latter mode of STM was introduced by Temirov et al. 23 and called scanning tunneling hydrogen microscopy (STHM). In their work, Temirov et al. show that “chemical” resolution may be induced in STM through the presence of hydrogen, H2, or deuterium, D2, molecules in the tunneling junction. Weiss et al. 17, conducted systematic experimental and theoretical studies of the STHM junction that revealed its imaging mechanism. The “chemical resolving power” of STHM lies in the fact that a single H2 or D2 molecule physisorbed in the tunneling junction, for sufficiently small bias voltage values plays a double role, i.e., it is a sensor to the short-range Pauli repulsion between the molecule and the imaged sample, and it performs a transducer action, transforming the force signal into changes of the junction conductance. The latter action is again a consequence of the Pauli repulsion, this time between the molecule and the scanning tip. It has to be pointed out that experiments with STHM 17, 18, 23, although they seem to be similar to the tip functionalization method applied by Gross et al. 15, 16, should be considered as using a new technique. Gross and coworkers modified the tip apex by intentionally picking up of a given molecule, whereas Temirov et al. and Weiss et al. modified the tunneling junction by flooding the microscope chamber with H2 or D2 and thus the whole sample is covered with those molecules.
Methods of chemical identification of atoms on a surface, or resolving the atomic positions and bonds inside a molecule are essential for characterization of the topography of atomic or molecular devices/circuits. The next prerequisite is the ability to measure the properties of the whole device/circuit or its components. For example, the detailed understanding of charge transport through single-molecule wires is still challenging 24, 25. In this context, special attention is paid to measuring changes in the conductance (the resistance) of a molecular wire with increase in its length, as wires of different lengths are important for building the device and connecting it with the outer world.
There are several scenarios that in principle may allow measurements of the molecular conductance, among which quite often the substrate surface plays a role of one electrode, whereas the scanning tip (either STM or metal-coated AFM) plays a role of the other electrode (see, e.g., Refs. 24, 25). However, in many experimental setups results provide only statistical information about the conductance, since the number of molecules contacted to both surface and tip is not controlled. For instance Choi et al. 24 report measurements of the electrical resistance of long conjugated molecular wires (oligophenyleneimine, OPI) as a function of the length of the wire. However, in their approach the measurements are performed on monolayers of OPI on gold substrate to which a metal-coated AFM tip is approached. Therefore, there is no control of the number of contacting molecules and their conformation in the junction. Furthermore, one may synthesize the OPI molecules having different lengths. On the other hand, in the sample-preparation procedure molecules of a single length are used. Consequently, each sample corresponds to a monolayer of molecules of one length only, and the dependence of electrical resistance on the length of the molecular wire is obtained from a series of separate measurements, each of them is done for a fixed length. With these restrictions Choi et al. have analyzed molecular wires of the length from 1.5 to 7.3 nm, and observed a transition in the conduction mechanism from tunneling (the molecular resistance scales exponentially with length) to hopping (linear scaling) near 4 nm. The former regime, i.e., the tunneling conductance, has been recently addressed in a more controlled manner by the ingenious experiment of Lafferentz et al. 25. First, the authors were able to perform on-surface polymerization of dibromoterfluorene (DBTF) to conjugated molecular chains. Secondly, they mastered the procedure of picking up a single-molecular chain from the surface with STM tip (see Fig. 3). Thirdly, Lafferentz and coworkers successfully determined the conductance as a continuous function of the molecule length. Finally, the experimental results are supported by theoretical simulations. The last two points reveal an exponential character of the conductance curves (see Fig. 4), being fingerprints of the tunneling regime of the conductance that is modulated by oscillations. Such a modulation results from detaching one molecular unit after another from the surface upon chain stretching (see Fig. 4).
It is very important to keep in mind that in each experiment, like those discussed above, the measured conductance is in fact a property of the electrode–molecule–electrode junction. In the reviewed experiments, the surface and the tip play a role of electrodes, however, that may not necessarily be the case in future atomic- and molecular-scale devices. Therefore, it still remains challenging to measure the conductance in more realistic configurations, e.g., in which a single molecular wire is contacted with some metallic or molecular nanoelectrodes manufactured on the substrate that does not alter the junction properties. Furthermore, it would be very advantageous if nanoelectrodes gradually evolve into microelectrodes allowing for communication with the outer world. In achieving those goals, it is necessary to develop techniques that would allow for fabrication of electrodes (supposedly of hybrid molecular and metallic character) spanning connection between the “nano” and “micro” domains, as well as, means of decoupling of active parts of the future electronic monomolecular device from the underlying substrate. The latest approaches to those problems will be briefly reviewed in the next two sections.
3 The device reliability
The problems of positioning of the molecule in the proper place within the electronic circuit or interconnecting it with external metal electrodes are interesting, challenging, and stimulating in their own right, however, do not guarantee successful fabrication of a working single-molecule device prototype. The properties of the single-molecule instrument would strongly depend on its internal structure. The chemical and even topological surrounding of the molecule affects and may change the shape of the molecular orbitals and the electronic structure (see, e.g., Refs. 26, 27) or simply sterically restrict access to the active sites of the molecule. Unwanted and unexpected changes in the structure of the molecule will alter the way the single-molecule device operates. Consequently, in order to increase the device reliability, schemes for avoiding a strong coupling of the molecule with underlying substrate have to be developed.
There are possible different scenarios leading to full or partial decoupling of the molecules from the substrate, and some of them have been already tested experimentally to some extent. Let us first consider an approach in which a conducting or semiconducting surface of the substrate is covered with additional epitaxial isolating layers. Several experimental realizations are worth mentioning in this context: a few monolayers of NaCl on Cu(111) 16, 19, 28, NaCl on Ag(111) 29, RbI on Cu(331) 28, or KBr on InSb(001) 20, 30. We shall concisely describe some of those results beginning with the work of Repp et al. 19 on pentacene molecules deposited on a few monolayer thick NaCl film on the Cu(111) surface. Adsorption of molecules on a metal surface typically results in a strong coupling to the substrate. This coupling, apart from influencing the molecular orbitals, introduces a mutual coupling between molecular states through the surface. As a consequence, STM scans of molecules are far from being images of the native orbitals of the free molecule. They are rather a mixture of few molecular orbitals. Repp et al. 19 showed that already a few monolayers of NaCl are enough to decouple pentacene from the electronic influence of the metal surface and at the same time still allow for measurements with the use of STM (i.e., electrons are able to tunnel through the ultrathin insulating film). The conclusion on decoupling the molecule from the Cu(111) surface by few monolayers of NaCl is based on differential conductance spectroscopy measurements and comparison of STM images with the results of theoretical modeling. Spectroscopic data exhibit two distinct features at −2.4 and 1.7 V that are ascribed to the HOMO and LUMO of pentacene, respectively (see Fig. 5). Consequently, STM images obtained with the bias voltage below −2.4 V are highly similar to the HOMO and those with the bias above 1.7 V to the LUMO. Additionally, computer simulations based on density functional theory are applied to model orbitals of the isolated molecule. Comparison of the HOMO and the LUMO obtained through modeling with experimental STM images ascribed to the appropriate orbitals shows no influence of the surface on the molecular electronic structure, i.e., achieved decoupling is sufficient to preserve the inherent electronic properties of the free molecule 19. It is worth noting that the same method of decoupling is used by Gross et al. 16 in observation with the ultrahigh resolution of the molecular orbitals of naphthalocyanine in their experiment with a CO-modified STM tips, as described in the previous section.
Such et al. 20 studied PTCDA molecules adsorbed on an ultrathin layer of KBr grown on an InSb(001) surface. The presence of insulating film changes the behavior and the electronic structure of the molecules significantly. On bare InSb(001) surface PTCDA molecules spontaneously form molecular wires that are oriented along the  crystallographic direction of the substrate surface 31–33. Contrarily, on an ultrathin insulating KBr layer defects of the layer act as adsorption traps, and the formation of molecular wires is hindered 20. The molecules are found either at steps or on flat terraces on both the 1st and the 2nd monolayer of insulator. Moreover, molecules adsorbed on the 1st KBr layer are mainly found as single entities or pairs of molecules oriented with the longer axis of the molecule along the  crystallographic direction of the InSb(001) substrate, and only occasionally as larger groups, whereas on the 2nd layer they always form clusters of at least three molecules. These observations suggest that already the presence of the 2nd KBr layer is enough to minimize the influence of the InSb substrate on molecules organization. Finally, the authors were able to obtain a high-resolution empty states image (at a positive bias voltage of the tunneling junction in STM) of the molecule on the 2 ML KBr film that is extremely similar to the gas-phase LUMO (see Fig. 6b). However, the resolution is not the key result in this work, the most noteworthy issue is the fact that the LUMO is observed at positive bias voltage. On the InSb surface the LUMO of the PTCDA molecule is imaged at negative bias, i.e., filled states image (see Fig. 6a). Due to strong coupling between the molecule and the substrate energetic spectrum of the molecule is shifted with respect to the Fermi level of InSb; consequently the LUMO becomes partially filled and visible at negative bias. Since in the work of Such et al. the LUMO is visible at positive bias, the electronic decoupling from the underlying semiconducting substrate is proven. Consequently, the discussed scenario of decoupling by the ultrathin insulating film is applicable to both conducting and semiconducting surfaces.
We would like to end the above discussion by invoking experiments on two realizations of working molecular switches on an ultrathin insulating layer deposited on a metallic substrate. In the first experiment, Liljeroth et al. 28 demonstrated the single-molecule switch based on hydrogen tautomerization of naphthalocyanine molecules deposited on NaCl on Cu(111). The molecule possesses a pair of hydrogen atoms in the central cavity that can be found in two positions (see Fig. 7b). As a consequence, the LUMO of a free-base naphtalocyanine can have two orientations, formally equivalent to the rotation of the molecule by 90°. The switching between the two orientations of the pair of hydrogen atoms results in a substantial changes in the tunneling current measured at the STM tip positioned over the molecule deposited on the ultrathin NaCl film on Cu(111) (see the red dot and corresponding current vs. time plot in Fig. 7a). The molecular switch is operated by means of increasing the bias voltage in the tunneling junction, i.e., the electrons tunneling through the molecule induce the tautomerization reaction. The authors exclude the possibility of on-surface rotation of the whole molecule as they observed switching of molecules at step edges and in arrays of molecules 28. Accordingly, switching is highly localized, well defined and does not involve changes in the molecular frame. Moreover, the changes are reversible (see current vs. time plot in Fig. 7a) and the tautomerization reaction is not observed when naphthalocyanine molecules are directly adsorbed on a Cu(100) substrate, i.e., the native LUMO of the free molecule relevant to the reaction cannot be probed. Therefore, Liljeroth et al. 28 demonstrate that these molecules possess all the essential features of a single-molecule switch. Such features are required in the scenario of the decoupling from the (semi)conducting substrate by means of few monolayer thick insulating films. At the next step, they can be used as building blocks in more complex arrangements such as logic gates.
In the second experiment we would like to refer to, Mohn et al. 34 demonstrate reversible formation and breaking of the bond between a gold atom and a PTCDA molecule that could be utilized as a molecular switch. The authors in the first step of the sample-preparation procedure covered a Cu(111) surface with 2 ML thick NaCl film and in the next step deposited Au atoms and PTCDA molecules at low coverage regime. Finally, by applying voltage pulses of appropriate polarity to the tunneling junction they were able to switch in a repeatable manner between the bonded (the bias voltage V = −1.5 V) and the nonbonded (the bias voltage V = +1.5 V) configuration of the Au–PTCDA complex. This alteration is accompanied by a change in the tunneling current of about two orders of magnitude. The complex is characterized electronically with STM, its geometry is deduced from atomically resolved AFM images, and the experimental results are corroborated with density functional theory calculations. Mohn et al. argue that the switching stems from a change of the charge state of the Au–PTCDA complex upon bond formation. Furthermore, the stability of the different charge states is ensured by the presence of the insulating NaCl double layer.
The next option that is worth considering as the decoupling method is an approach in which all dangling bonds of the substrate surface are passivated by adsorption of atomic or molecular species. In this regard, it is very convenient to focus on hydrogen passivation of Si(100) or Ge(100) 35. The issue of imaging adsorbed molecular species on such hydrogen-passivated semiconducting surfaces is quite challenging, as reported by Bellec et al. 36. Nevertheless, in their successful attempt Bellec and coauthors examined the pentacene molecules deposited on a hydrogenated Si(100) surface with the use of STM. By obtaining the image of the HOMO of pentacene, they prove that already a single layer of H atoms is sufficient to electronically decouple the molecule from the underlying substrate (see Fig. 8). The result is impressive especially if one compares it to measurements on few ML thick insulating films on metals. However, at this point one should keep in mind that the density of states of the silicon substrate is reduced in comparison to a typical metal, what in turn has an influence on the decoupling method. The authors support their findings by an accompanying theoretical modeling (see Fig. 8). It is worth noting that the possibility of imaging with STM the pentacene molecule on a hydrogenated silicon surface was first shown theoretically by Ample and Joachim 37. In the proposed approach, the buffer layer of hydrogen atoms not only decouples molecules from the underlying substrate, it also makes possible coupling of a molecular device on demand by desorbing hydrogen atoms intentionally one by one with atomic resolution 36, 38–40. The latter procedure allows creation of conductive paths/wires composed of dangling bonds of Si atoms on the substrate. We shall come back to this concept in the last section.
In another scenario, the issue of decoupling is addressed at the stage of the molecular device design. It may be envisaged that the molecular device is composed of an active part of the molecule that performs desired operation, some interconnects that allow communication with the outer world, and some anchor groups that take part in positioning and bonding of the device. If there is a need in increasing bonding ability, those anchor groups may be equipped with some functional groups, for instance carboxylic or thiol groups, depending on which substrate the device will be deposited. Furthermore, anchor groups may also be accomplished in the form of spacers that would lift up the active part of the device above the substrate surface. Increasing the distance between the active part and substrate should result in attenuated coupling between the device and the surface. Molecules with such spacer groups (often called legs) have already been examined on several substrates, for example, Lander molecules have been deposited on Cu 41, 42, TiO243, 44, and InSb 45. Lander family molecules possess legs that are connected to the active board by σ bonds and are partially free to rotate. Due to this fact, the Lander molecules may adapt several geometries in order to fit to underlying substrate. It has also been observed that these molecules can be manipulated by the STM tip in the “walk-like” manner with spacer groups acting as moving legs (see, e.g., Refs. 41–44) or may create hydrogen bonds between each other leading to formation of molecular wires 46.
The active board of Lander molecules is a π-system that in principle could be utilized as a molecular wire. To some extent the effect of decoupling from the underlying substrate is demonstrated by the fact that in the STM image a Lander molecule appears as four lobes arranged in a rhomboidal or rectangular geometry 41–45. Such an appearance is ascribed to tunneling through the spacer legs 41. There is only a single report on a successful attempt to image the active board, i.e., when the molecule is deposited on a TiO2(110) surface 43. Additionally, Moresco et al. 42 report on formation of the contact between a molecular wire, i.e., the Lander molecule, and a metal electrode. In their setup, it is simply a surface step edge that is connected to the molecule, as the molecule on a terrace is decoupled from the underlying substrate. In the experiment, the molecules are deposited on a Cu(111) surface that exhibits Shockley-type surface states. Scattering of the surface-state electrons from adsorbates or step edges is manifested by standing-wave patterns observable by LT-STM. By monitoring these patterns Moresco et al. are able to demonstrate intentionally created contact between the Lander molecule and the step edge. The contact is created by pushing the molecule to the step with the longer axis of its central active board perpendicular to the edge. The interaction between the wire and the step edge induces local electronic structure perturbations that are reflected in alterations of the standing-wave pattern on the upper terrace. On the contrary, no changes in the pattern are observed when molecules are positioned with the active board parallel to the step. The result is corroborated with theoretical calculations showing that the molecular wire interacts via coupling of the π molecular orbitals of the naphthalene end group with the upper terrace surface states 42. In a slightly broader context, the experiment of Moresco et al. may be viewed as a starting point in manufacturing the electrical contacts for the molecular device. At the first step one molecule is contacted with the metal nanoelectrode, then in order to reach the molecular device, subsequent molecules are attached. Molecular units that form the molecular wire may be connected for instance by means of hydrogen bonds due to the presence of additional functional groups at the ends of the active boards.
Finally, in order to avoid the electrical coupling between the future monomolecular device and its support it may be advantageous to use insulating surfaces as the substrate. Such an approach requires development of methods of manufacturing some electric nanocontacts on such surfaces and positioning of active parts of the molecular device in a proper place, as well as advances in the means of characterization of single molecules on insulating surfaces. Most of these issues remain open therefore, advancing technologies in which the insulating surfaces play a role of the substrate would require a considerable effort.
In general, the problem of electrical addressing of the molecular device or even a single molecule itself is difficult. In previously reviewed experiments in the first approach the substrate and the scanning tip play the role of electrodes. When insulting substrates are considered, metallic electrodes spanning connection on a distance from the micro- to the nanoscale have to be manufactured. We shall return to the problem of formation of nano- and mesapads, metallic and molecular nanowires in the last section. Let us assume for a while that the net of connections is prepared on the substrate, now the molecule in question has to be in contact with it. The issue of positioning of a single molecule emerges. The solution to that problem supposedly will involve additional steps in the synthesis of the functional molecules. For example, such a molecule could be equipped with polar or other functional groups that would play the role of anchors stabilizing it in a predefined position 47–49. Furthermore, one may also develop methods of nanostructuring of the insulator surface in order to create preferred adsorption sites. For instance, by irradiation of a KBr(001) surface with 1-keV electrons one obtains straight-edged pits on the surface 50. Those pits may serve as adsorption traps for organic molecules 47, 48 or even be reshaped to some extent by adsorbed molecules 49. As one considers the question of characterization of molecular species on insulating surfaces, the method of choice is the atomic force microscopy that allows both imaging with ultrahigh resolution and tip-induced manipulation (see discussion in Section 2). This technique appears quite challenging in the context of the single-molecule electronics; nevertheless, high-resolution experiments on single isolated molecules on insulators at room temperature have been recently reported 47, 48, 51. Finally, AFM tips may be prepared as conducting and, therefore, allowing for simultaneous AFM and tunneling current measurements, resulting in an approach that still allows consideration of the use of scanning tips as electrodes. In this regard, tuning-fork-based AFM (QPlus configuration) 12, 13 is a highly promising measurement technique. In the QPlus configuration, it is possible to switch between AFM and STM modes while staying on the very same atomic location of the sample. Furthermore, in this setup the sensitivity towards short-range forces is increased, a property that is exploited in achieving ultrahigh-resolution images (see experiment by Gross et al. 15 discussed in the previous section).
4 Assembling the device: The outer-world connections and a few comments on future prospects
All the discussed issues and reviewed experiments hopefully lead us step by step towards a working prototype of a monomolecular electronic device. In recent years, there has been much effort on purely theoretical research that would advance and enrich our understanding of computational and functional potentials of single-molecule devices 38, 52–60. However, if the problem of addressing of monomolecular devices is not solved in the laboratory, these theoretical achievements will remain solely academic exercises for testing our insight into the quantum mechanics and complex systems. Therefore, in order to utilize single-molecule devices, techniques for fabrication of the electrodes spanning connection between them and “the outer world,” i.e., output interface, must be developed. Several courses of action may be proposed in this context 61. For example, there are methods of manufacturing metal nanostructures on different substrates reported in the literature, such as studies on metallic nanowires on InSb(001) 62, or mesapads on Ge(111) 63. Such wires and/or pads could serve as contacts for STM/conducting AFM tips, i.e., the tip would play the role of a microelectrode. However, proximity of such a metal nanostructure that is large in comparison to the monomolecular device supposedly will strongly influence its performance (see discussion in the previous section). Therefore, it may be helpful first to design and produce molecular wires that would connect the device and metal electrodes. In this way, due to the increased distance between the device and metal electrodes the influence of the latter on the operation of the machine will be suppressed. There are already reports in the literature that address the issue of manufacturing molecular wires directly on the surface by on-surface polymerization 25, or by tip-induced reactions 64, 65, or by means of self-assembly processes 46.
The difficulty that remains to be solved is finding a method for the connection between a molecular wire and an active molecule, i.e., the future molecular device. As has been already demonstrated by Lafferentz et al. 25, the wires manufactured by means of on-surface polymerization may be manipulated with STM tips. However, an exact positioning of a lifted-up polymer is a rather difficult task. Nevertheless, there is a very recent and very promising report in the literature addressing the issue of connecting the molecule with a molecular wire. Okawa et al. 64 describe their experiment on controlled formation of conductive polymer nanowires and subsequent connection of those wires to a functional molecule. The authors deposited monomer molecules of the diacetylene compound on a highly oriented pyrolytic graphite, where the molecules formed a self-assembled monolayer (SAM). Thereafter, on top of such a SAM Okawa and coworkers put down phthalocyanine molecules (either metal-free, or as copper, or zinc phthalocyanines). By stimulating with a STM bias voltage pulse in the diacetylene moiety, on which the phthalocyanine molecules are adsorbed, a chain polymerization reaction was initiated. Okawa et al. 64 named that process “chemical soldering.” Such a polymer chain forms a chemical bond with a phthalocyanine molecule and thus allows for the electrical coupling (see Fig. 9). In the polydiacetylene chain, the charge transfer is realized by means of polarons or bipolarons and may proceed from the polaronic states of the chain to the energy levels of phthalocyanine, if the bias voltage is properly adjusted 64. Thus, the polydiacetylenes–phthalocyanine–polydiacetylene (PDA–Pc–PDA) system may serve as a molecular-resonant-tunneling diode. The authors corroborate their experimental results with large-scale density-functional first-principles calculations that allow them to study in details bond formation between the chain and the active molecule, as well as, obtain the energy-level diagram of the formed PDA–Pc–PDA junction. It is worth noting that Okawa et al. 64 accepted the possibility of performing the chemical soldering on other than graphite substrates or even on insulating molecular films.
Possible solutions of the outer-world connection problem are not limited to molecular wires only. The other option is to manufacture atomic wires from adatoms on the surface either by scanning-probe tip-induced manipulation 66, or by means of self-assembly process 67, or with the aid of specially designed molecules 61. Moreover, one may think of a reversed method, i.e., formation of nanowires from atoms intentionally desorbed with atomic resolution from a passivating layer 36, 39, 40. Atomic resolution in the intended desorption of hydrogen from the passivating layer on Si(100) has been already reported in the last decade of the 20th century by Shen et al. 40. Since then it has drawn much attention, and recently has been considered both as a method of formation of conducting atomic wires only 39 and as a method for manufacturing of complete logic gates from the uncovered dangling bonds on a Si(100)H surface 36, 38. In their work, Soukiassian et al. 39 examined experimental conditions for an optimum writing speed of such dangling-bond lines, either for wiring, or as components of logic gates (see Fig. 10). Furthermore, the authors point out sources of possible distortions and changes of the atomic structure of the wire. On the other hand, just recently Ample et al. 38 theoretically analyzed the use of dangling-bond structures as either wires, or logic gates, in the most realistic theoretical approach up to date. In their model, all parts of the system, i.e., the supporting surface, the contacting pads, the dangling-bond structure and, if needed, the active molecules are taken into account. The authors stress that any molecule and atomic-scale circuit logic gate optimized in the absence of any environment will stop functioning when the surrounding is included in the simulations 38. The authors point out several reasons for possible failure of the device. First, there is the surface tunneling leakage current that can induce a short circuit within the gate at the atomic scale. The remedy for this problem may be increasing the distance between the gate and contacting metallic nanopads or utilization of very large surface gap materials as supports (both issues were already discussed in the course of the manuscript). Apart from the leakage current, another problem is the molecular orbital hybridization of the molecule logic gate with the atomic orbitals of its interconnecting wires. Most importantly, the electronic structures of the molecule and the wire are changed due to such interaction. This problem has to be included at the level of design of the molecular device, i.e., one has to treat the system composed of the molecule and the wires as a whole in designing the electronic device.
We would like to finish the present short review by invoking a recent experiment by Soe et al. 52, 53 demonstrating a single-molecule NOR logic gate that has been realized with the use of a trinaphthylene molecule. It is advisable to recall the way the authors communicated with the molecular logic gate. In their setup, the input data was provided by presence or absence of Au adatoms in contact with one or two naphthyl branches of the trinaphthylene molecule and the read-out was realized by means of the tunneling current intensity measurements on the third naphthyl branch (see Fig. 11). The input data is then classical in nature, however, the way the molecular gate works is purely quantum, i.e., the presence of Au atoms in contact with one or two of the naphthyl branches acts nonlocally and changes the electronic conductance through the remote third branch. This is due to the delocalization of the molecular orbitals over the molecular board 53. The Au atoms at the inputs of the molecular logic gate are placed with the use of the STM tip manipulation. For each of the input configurations, i.e., (0,0), (1,0), and (1,1), the authors recorded corresponding STM images (see Fig. 12), I–V characteristics and differential conductance spectra. From the latter, they deduced the optimal bias voltage at which the current through the third branch should be measured as the output of the logic gate. Finally, it was demonstrated that the current values follow exactly the truth table of NOR gate, proving its realization.
Additionally, the work of Soe et al. 52, 53 validated one way of designing the molecular devices, i.e., the quantum Hamiltonian computing (QHC). The authors first proposed and tested theoretically the model system without reference to its realization 53, then demonstrated both theoretically and experimentally that the trinaphthylene behaved like the proposed system 52, 53. In this way, the invoked experiment may be treated as a proof-of-principle that QHC can be utilized for turning a single molecule into a logic gate. Furthermore, Soe et al. 53 suggest that this design can be generalized to more complex logic functions, and thus it creates a link between molecular electronics and quantum design.
Funding for this research has been provided by the EC under the Large-scale Integrating Project in FET Proactive of the 7th FP entitled “Atomic scale and single-molecule logic gate technologies, AtMol” and the Coordination Action “Nanoscale ICT Devices and Systems, NanoICT,” contract no. 216165. S.G. would like to acknowledge support received from the Foundation for Polish Science within a START program (2010 and 2011).