Molecular architectures for hybrid nano-devices

Introduction

A scientific revolution started in the 80’s with the development of Scanning Tunneling Microscopy (Nobel Prize in Physics to G. Binnig and H. Rohrer, in 1986) that disclosed a world of individual atoms and molecules at the nano-scale. We can recognize that this revolution was due not to a single instrument, rather to a myriad of tools with incredible resolution and to the development of methods and protocols that came along with appropriate modelling of nanoobjects. We now refer to the whole set of these innovations and discoveries as “Nanoscience”. Interestingly, we can produce nano-objects, with size ranging from few nm to few tens of nm, following two radically different approaches: top-down methods, which essentially sculpture matter down to the nanoscale and bottom-up approaches which essentially make use of chemical methods to fabricate billions of identical atomic aggregates. In this way, we have learnt to produce, observe and understand the properties of individual nano-objects. Among these, molecules with size of few nm are right at the centre of interest in Nanoscience and Nanotechnologies. We have also started to learn how to assemble different molecular units in an ordered fashion by observing how Nature builds complex biological cells. Achievements obtained in supramolecular chemistry (Nobel Prize in Chemistry to J. M. Lehn, J. Pedersen, D. J. Cramm, 1987) show how it is possible to assemble molecules in complex 3D structures whilst progresses in Surface Science have allowed to produce and observe ordered arrays of molecules/atoms on surfaces (2D) with atomic precision and over a scale of microns. Interestingly, these length scales nicely extend across the ultimate limit of device miniaturization: the size of an electronic processor/memory is currently ~10 nm with our best technologies. The threshold of 10 nm is actually crucial for devices since energy levels become discrete and genuine quantum effects dominate at this length scale. There is therefore an incredible merging of interests and disciplines around the study of nano-scale architectures and devices. If R. Feynman drew our curiosity to the infinitely small claiming that “there is plenty of room at the bottom”, J. M. Lehn replied that “there’s even more room at the top”. We knew that both were right, we just had to tune our goggles at the right length scale.

Next challenge in this field is to combine molecular building blocks (tectons) with nano-structured layers (platforms). The control we have now achieved is already fantastic, yet, the possibility to combine different nanoobjects appears endless, due the myriad of molecules and platforms available and to the large number of possibilities to combine them. Thus, a second revolution in Nanoscience, whose focus is the assembling of different nano-objects in hybrid architectures, is now taking place. Molecular architectonics aims at assembling different molecular objects in hybrid nano-architectures under specific design that − at the end − may allow to exploit specific functionalities of these assemblies. If properly done, this process leads to novel nanodevices.

The easiest example is the assembly of molecules in structures mimicking macroscopic devices, thus we may have molecular transistors, rectifiers or photodetectors etc. Most of the effort in molecular electronics is actually devoted to this type of research. Yet, this picture is very reductive since we have the chance to combine different degrees of freedom (charges, spin, photons, phonons etc.) with unprecedented control. Moreover, since we are imposing unconventional boundary conditions in our hybrid systems, we expect to observe novel phenomena at the nano-scale. Thus we are looking at these nano-architectures with much curiosity and interest for the potential they hold for radically new applications.

The ways to assemble molecular architectures and to connect them to the macroscopic world are disparate and one needs different tools, instruments and approaches. Thus, at the end, molecular architectonics can be considered the art of orchestrating molecular objects with a design of possible device in mind.

Let us briefly consider different approaches to deposit nano-objects on surfaces. They have different levels of complexity, cleanness and selectivity as extensively described in excellent reviews, books and journals entirely dedicated to these topics. Working in UHV is required to guarantee the possibility to prepare the substrate in the cleanest way, by controlling their morphology. Pre-assembled nanoobjects can then be evaporated or synthesised directly on surfaces. This requires building blocks to be robust enough to be heated − for sublimation − and to resist to deposition on surface. The more complex are the tectons, the more delicate they are, thus sublimation is certainly the best approach to deposit atoms or simple aggregates, but we need different methods for fragile objects. Still working in UHV, electrospray deposition is the method used to gently deposit delicate molecules. Using electromagnetic lenses, the electro-spray deposition (ESD) also allows fine mass selection of the aggregates and to control the energy of landing on surface for ionized objects. Yet, being the ionization level limited, this method may be not effective for heavy aggregates and in practice each set up works well with a limited class of molecules. If the constrains of UHV can be relaxed, deposition from liquid phase is the easiest and more versatile method: pre-assembled nanoobjects can be disperse in solution and substrate can be dip and rinsed several times in order to form (multiple) layers.

The way to graft tectons on surface and consequently to assemble them in an ordered way or to position them in specific surface portions essentially resides in the exploitation of different energies for bounding. An excellent overview of the energy scales playing a role is discussed in ref. . Covalent bonds are used to fix tectons on specific positions: for this we exploit the affinity of specific chemical elements and there are well-known templates to be used such as, for instance, thiol group ending with sulphur that easily bonds to gold. When this happens, we classify it as chemical adsorption.

Benzene-like ending groups (like pyridine) form (weak) π-π bonds with allotropes of carbon, like nanotubes or graphene. Non-covalent bonds are instead used for self-assembling processes: extraordinary 2D motifs are formed by the organization of aggregates on surfaces, that results from the minimization of free energy of the system which includes weak lateral interactions between nearest tectons and the coupling to the substrate, comprising omnipresent van der Waals energy. Here we deal with physical adsorption. Deposition of tectons at finite temperature or soft heating provides enough thermal energy to the systems to allow re-organization of the nano-objects which eventually form large ordered patterns. Needless to say that one may also combine different approaches to drive the system to specific surface decoration.

For instance we may organize simple molecules with functional terminations: in this way we obtain a suitable “carpet” on top of which it is possible to anchor functionalized objects to the corresponding terminations.

Even more sophisticated methods can combine top-down approaches, e.g. pre-patterning of surface by lithographic methods followed by decoration exploiting functional groups. Nano-printing techniques are also extensively used to form ordered arrays with pre-defined patterns.

In the following we present few examples of our recent work that illustrate some basic principle of molecular assembling and few examples of molecular devices we have recently developed. Generally speaking, since we are mostly interested in spintronic applications, our favourite molecules − but not all of them − contain spin centers. We first show how to deposit these molecules on surfaces in an ordered way. To use charge current as detection parameter, we use graphene as suitable platform.

To this end, we utilize electron beam lithography to pattern electrodes and we developed electro-burning technique to fabricate molecular junctions. The use of external magnetic field, visible light source or voltage bias through an additional gate electrode helps us to search and exploit intrinsic molecular features.

2 Deposition by liquid phase

We first consider the deposition of molecules from solution, more specifically the case of one molecule, named Cr₇Ni in short, with eight metal ions in a ring shape bridged by pivalates ligands which form an organic shell. This molecule attracted our interest due to the fact that its ground magnetic state at low temperature is a pure two-level state (spin 1/2) that we proposed as suitable candidate for qubit encoding. Cr₇Ni is easily soluble and can be dispersed in different solvents at different concentrations by preserving its integrity and magnetic features. Deposition can thus be done by simply dipping a substrate in solution, with a subsequent rinse to remove possible 3D aggregates. The van der Waals interaction with substrate makes these molecules laying flat in such a way that contact with the surface results is maximized. Conversely, we can make the bonding with the substrate stronger by choosing peripheral threads with different terminations. To test this, we have studied different functionalizations of the Cr₇Ni ring and we deposited them by liquid phase on the Au surface. As an example, if we want to stick molecules on gold we can use thiol groups with sulphur terminations which have high affinity with Au. If we wish to deposit it on graphitic surface, instead, we can use alkyl chains or organic ligands with cyclic carbon terminations that tend to form π- π bonds with the surface. This approach is preferable, for instance, to disperse and fix few molecules on surface and observe them by scanning probe microscopies. On the other hand, while functional terminations make the bond with substrate stronger, they also favour intermolecular interaction thus favouring formation of molecular 3D aggregates that are in general undesired. We have developed a two-step deposition protocol to circumvent this problem: in the first instance we deposit a molecular layer of alkyl chains whose affinity with the surface and large lateral exposure facilitate the formation of ordered arrays. We then use this molecular carpet to anchor Cr₇Ni molecular rings by adding a SH termination to the alkyl chains and a corresponding counter ion to the molecular rings. In the second deposition step the sulphur is de-protonated and forms a covalent bond with the counter ion, thus firmly blocking the Cr₇Ni molecule on the surface as illustrated in fig. 1.

3 Sublimation and self-assembly

In a successive work, we have explored the possibility to deposit our Cr₇Ni molecular qubit in clean conditions (UHV) on metal (gold) surface. Gold is used to make electrodes and any conductive parts in molecular circuits. We found out that Cr₇Ni is a quite robust molecular building block that can be also sublimed at relatively low temperature (200 °C) preserving its integrity. Again we can play with peripheral groups to tune the interactions with the external word. By using S-terminations (thiol groups) the sublimed molecules firmly stick on gold surface (fig. 2 f, g). When, instead we evaporate bare Cr₇Ni molecules it turns out that these are relatively free to float on gold substrate at room temperature (fig. 2 b, c). A 2D layer can be formed and we could observe formation of regular patterns with hexagonal motif (fig. 2d). Interestingly we have also performed DFT ab initio calculations to account for the different molecular interactions, including the van der Walls and lateral interactions between neighbouring molecules, and we found that the energy of the systems is actually minimized for the hexagonal pattern of the 2D molecular assembly, in perfect agreement with our STM images (see fig. 3).

4 Deposition by electro-spray

In many cases, organo-metallic molecules are not robust enough to be heated up to their sublimation temperature or they simply break when their kinetic energy is too high when they impact the surface. A way to gently deposit molecules on surface is to use electrospray deposition (ESD) that is an alternative technique that allows to softly land large and heavy molecules from liquid solution/suspension on to any type of substrate. For instance, ESD has been successfully used for the deposition of giant biomolecules (proteins) and nanoparticles, thus enabling integration of large molecular building blocks in electronic circuits. The basic working principle is depicted in the scheme of fig. 4.

Briefly, after the capillary injecting the desired sprayed solution/suspension, we have one or more stages under differential pressure, leading to an UHV environment where the support (either conducting or insulator) is hold. Sets of electromagnetic lenses act as mass selectors and/or to (de-)accelerate molecules.

We have used a first stage of mass spectrometer to test ESD with simple metallorganic groups in clean conditions. As usual, we checked the integrity of the molecules with topological inspection by STM and by chemical analysis using X-ray photoemission spectroscopy XPS spectroscopy to assess that the element stoichiometric ratio is preserved after deposition. Our tests with Crbased trimers is an exemplary case showing the power and simplicity of this deposition method. The structures of the Cr₃ compound and corresponding positive-ion ESD mass spectra recorded in the 960–1050 m/z mass regions are shown in fig. 5. On the left panel a high-resolution STM image of a single molecule of Cr₃ ESI-deposited on the Au surface is reported. More recently, we adapted electrospry to deposit very big molecules such as graphene nano-ribbons.

These are long ribbons of graphene with atomically controlled edges that can be chemically synthesized by polymerization processes in solution.

Their finite lateral size induces a gap in the electronic band with peculiar optical features, both otherwise absent in 2D extended graphene sheets. The problem we needed to solve was to find an efficient method to transfer them on insulating substrates for making electronic circuits. By using graphene electrodes (see next section), we were able to fabricate a three-terminal device whose central conductive channel was made by graphene nano-ribbons deposited by electro-spray.

5 Patterning surfaces with Focused Ion Beam

Complementary to chemical approaches, we normally use advanced top-down techniques that may sculpture matter with nanometer resolution. Lithographic methods, which make use of electromagnetic radiation (visible, UV or even X-ray) or electron beams, are generally used to create repetitive pattern on surfaces. Another powerful tool is the Focused ion Beam technique that makes use of sharp beams of energetic ions (e.g. Ar⁺ or Ga⁺) to ablate the surface in desired motifs. We were interested in combining this technique with chemical deposition methods to decorate silicon surface with magnetic nano-particles of a Prussian blue cyanided Cs-Ni-Cr derivative. To this end, we used a three-step procedure: i) we first functionalize an H-terminated Si substrate by undecanoic acid; ii) secondly we modify the acidterminated substrate to obtain a dangling tridentate ligand that is able to chelate a Ni (II) metal ion of the nano-particle, and finally iii) we dipped the substrate in an aqueous solution containing the nano-particles to obtain the grafted monolayer.

Between steps ii) and iii), we used FIB patterning to remove the organic ligand from selected regions of the substrate (fig. 6). This approach is straightforward because FIB allows for high-resolution direct lithography without using any chemical etching. Figure 6 shows the STM of the FIB patterning, which provided irradiated stripes 2 μm wide (dark region) alternated with nonirradiated stripes 5 μm wide (light grey).

To evidence the effective magnetic decoration of our stripes, we made use of a Hall probe scanning microscope which is sensitive to the local magnetic stray field of the nanoparticles. Figure 6 (right panel) shows, in false colours, the magnetic fringes of the stripes decorated with magnetic nanoparticles alternating with those of bare silicon.

6 Fabrication of electronic circuits for molecular electronics

Lithographic techniques are currently used in the industry to fabricate devices through multiple steps of patterning, etching and deposition of different types of materials. The current CMOS technology makes use of complex lithographic processes with device definition down to 10 nm over large silicon wafers. In research laboratory we widely use Electron Beam Lithography (EBL) to pattern different substrates and create templates that allow to contact nano-devices with the external world. Our interest is focused on graphene platforms. Needless to remind that graphene itself is an ideal 2D, zero-gap conductive layer on which many studies have been carried out showing fascinating phenomena that have recently led to the Nobel prize (2010, A. Geim and K. Novoselov). Probably worth noticing is that graphene is also an ideal platform to contact molecules, offering the possibility to make carbon-based conductive electrodes on which organic groups, such as alkyl chains or polycyclic aromatic carbon (benzene, pyrene etc.) may well stick.

The procedure to fabricate graphene devices is nowadays consolidated rather well. Typically we start from a “large” graphene sheet on insulating substrate (SiO₂, SiC or BN) and we pattern it in suitable device shape.

A typical template is given by threeterminal devices, similar to conventional transistors, with two electrodes that act as source and drain and a gate that can be either lateral or on top (or at the bottom) with respect to the central channel. Electrical contacts are made by evaporating metal (Au) pads, typically micron in size. Examples of this lithographic process are shown in fig. 7 a-d.

The central conductive channel of graphene can be laterally shrunk even further (down to 100 nm or below) thus forming nano-constrictions. Due to the peculiar electronic properties of graphene, these constrictions effectively work as quantum dots at low temperature, that is their energy levels result discrete and finite conductivity can be obtained only by aligning these levels to the Fermi energy of the electrodes. This electronic band alignment is possible by tuning the gate voltage.

Figure 7e shows an example of graphene nano-constriction we have recently realized by EBL and characterized (see fig. 8).

This type of quantum device has very interesting performances. For instance, since the energy levels of the dot can be filled only by one electron at a time, it effectively works as singleelectron transistor. Moreover, unlike other 2D electron gases in artificial semiconducting heterostructures, graphene has the peculiarity to be directly exposed to the external world thus offering the possibility to “sense” small variations occurring on its surface. Indeed, when tuned at its conductive state by the gate voltage, any perturbation affecting the energy levels of a graphene nano-constriction may drastically change its conductivity. In principle, any local field may perturb its conductance. In practice, local electric field variations affect the conductivity of graphene quantum dot very effectively. For instance, if a molecule is attached to graphene, its chemical potential induces, if not a possible charge transfer, electric signals in the conducting channel, making the device an ideal sensor with sensitivity down to a single molecule. This huge sensitivity obviously exposes the device to any sort of noise, thus we have to find a way to recognize our signals.

Since we are interested in magnetic nano-systems, we have recently by decorating a graphene nanoconstriction with a magnetic molecule. In that case we used a molecule (TbPc₂) with peculiar magnetic features, namely high spin and high uniaxial anisotropy, in order to easily recognize its fingerprints. Low-temperature characterization in applied magnetic field showed that such a device effectively behaves as spin-valve for which magnetic molecules, instead of conventional magnetic electrodes, act as spin polarisers for the current passing through the graphene channel.

7 Molecular junctions and molecular spin transistors

The device scheme previously described essentially comprises a nano-sensor (graphene quantum dot) driven by charge current and functional molecules in direct contact with the developed a spintronic device sensor. This scheme is not invasive (the current does not pass through the molecule itself ) but the sensitivity depends on the coupling between the dot and the molecule. For example, stray magnetic fields, even at nmscale, are in general weak, but if direct (exchange) coupling is established between the molecule and sensor, then the device becomes very sensitive.

A different device prototype consists in injecting a charge current directly through the molecule and using it as the central part of the device. This approach opens the door to exploiting multiple molecular functions such as their switchability, bi-stability, photo-charge conversion and, more in general, their sensitivity to external parameters such as magnetic field, temperature and chemical environment. Prototypes of this scheme are molecular transistors that consist of two electrodes (source-drain) injecting charge current through the molecule and one gate to tune the molecular energy levels. Some smart ways to produce junctions, with gap size of 1 or few nm between two electrodes, have been developed to fabricate molecular electrodes. Among these, fractures produced mechanically (break junctions) or by “electro-migration” are among the most popular in molecular electronics. The latter consists in passing high charge current through a constriction until it burns and breaks in some points. If this process is done in a controlled way (i.e. the current is rapidly reported to zero as soon as the constriction starts to break and controlling all the parameters such as the atmosphere, current ramps and cycles etc.), a fracture with small gap can be produced and this can be suitable to host one or very few molecules. Typically this method is applied to metal (gold) electrodes but we have recently adapted it to graphene for which we found that it works surprisingly well.

To complete the fabrication of such molecular transistors we have then to place molecules within the junction. This can be done by using one of the deposition methods previously described. This process is –of course–random and, since we try to disperse few units all over the whole device, the probability to get one molecule right in between the junction is very low. That’s why we typically prepare tens or hundreds of identical junctions on the same chip on which we then disperse molecules. In principle one may use a scanning probe to position one molecule on the right position in the junction but in practice this makes this long procedure unfeasible. We rather tend to functionalize molecules with anchoring threads specifically designed to graft on the metal/graphene electrodes as previously described. When all these steps are optimized, we can get working devices with satisfactory yield (e.g. 1 or few %).

Combing these methods, we are now able to fabricate molecular transistors (fig. 9). We use graphene electrodes with junctions electroburned and magnetic molecules, such as the TbPc₂ double decker, which is robust enough and possesses specific magnetic features. Experiments are carried out at very low temperature where quantum features of the device can be evidenced and studied in details. Figure 10 shows some preliminary results we have obtained by studying charge transport at 30 mK.

The conductivity of the molecule gets some finite values only for specific range of the gate voltage, thus evidencing the discrete nature of the molecular energy levels.

We are now looking at more specific features of these conductive maps also by applying a magnetic field: this will allow us to recognize some fingerprints of the molecule. We just mention, for instance, that it has been demonstrated that it is possible to read out the nuclear spin state of a single magnetic Tb centre and measure its coherence life-time that turned out to be sufficiently long to allow manipulation by micro-wave pulses. These recent experiments demonstrate that manipulation of quantum states are possible on molecular devices and their performances are comparable to other solid state quantum devices, such as semiconducting quantum dots or superconducting quantum devices. The future of molecular devices looks therefore very exciting and next challenges can be as ambitious as the encoding of quantum algorithms with molecular architectures.

Acknowledgements

We wish to thank our collaborators that contributed to the development of the Molecular Architectonics in Modena. In particular: U. Del Pennino, V. De Renzi and R. Biagi (UniMoRe) for surface studies; P. Fantuzzi, S. Lumetti and L. Martini (PhD at UniMoRe) for fabrication of graphene-based devices; V. Bellini and F. Troiani (CNR-Nano) for DFT calculations and simulations. This work has been partially supported by the EU FP7 FET-Proactive project MoQuaS N. 610449 (www.moquas.eu).