Final Report Summary - NANOCONTACTS (Structural and electronic properties of nanoscale metallic contacts fabricated by thermally assisted electromigration)
We study the electric current through wires which are atomically thin. This is important to understand the leads to transistors which become smaller and smaller as the storage densities in computers becomes bigger and bigger. It is simple to understand that for very small metallic wires, the position and chemical nature of every atom in the wire or in the vicinity of the wire counts. The key aim of the project is to correlate the electronic transport properties of nanoscale metallic contacts with their atomic structure. To achieve this, the electronic transport properties of nanoscale metallic conductors have been studied. These contacts were fabricated by Joule heating a nanowire until thermally assisted electromigration sets in and thins the nanowire to form a contact in a computer-controlled process. This computer-controlled process has been improved during the project to allow for unusual behavior in the resistance as a function of voltage. For sufficiently small contacts, i.e. large resistance, a decrease of R(V) while increasing V is observed. We tentatively attribute this effect to the presence of contacts separated by thin vacuum barriers in parallel to ohmic nanocontacts. Simple model calculations indicate that both thermal activation or tunneling can lead to this unusual behavior. We describe our data by a tunneling model whose key parameter, i.e. the tunneling distance, changes because of thermal expansion due to Joule heating and/or electrostatic strain arising from the applied voltage. To study whether the controlled electromigration process could be applied to macroscopic wires, w have studied controlled electromigration (EM) in free-standing copper wires. Oxidation during the EM in air stabilizes the wire against uncontrolled blowing making it possible to thin the conductive part of the wire down to a few conductance quanta G 0 = 2e^2 /h. The decisive influence of oxidation by air on the EM process was confirmed by control experiments performed under ultra-high vacuum conditions. Estimates of the local temperature in the wire are obtained from finite-element-method calculations.
The electromigration thinning process has been applied to magnetic contacts. A ring geometry was used to reproducibly place a magnetic domain wall at the nanocontact. Different resistance levels show the position of the domain-wall can be moved by applying a magnetic field into and out of the nanocontact. The resistance with and without a magnetic domain wall differs by up to 50 percent and exhibits a previously unobserved sign change. Our results can be reproduced by recent atomistic calculations for different atomic configurations of the nanocontact, highlighting the importance of the detailed atomic arrangement for the MR effect. The domain wall pinning strength is increased on reducing the contact size in line with a reduction of the wall energy in narrower constrictions. Furthermore, the angular dependency and symmetry of the depinning field is measured to determine the full pinning potential for a domain wall in a system with a narrow constriction.
We then studied scanning probe imaging of metallic nanocontacts. We used metallic nanocontacts on insulating substrates (oxidized silicon) to avoid short-circuits through the substrate during electrical measurements. We investigated scanning tunneling microscopy, but this technique relies on the tunneling current between the tip and the surface, and we found that for standard electron-beam-lithography made samples, carbon-rich remnants on the surface lead to contamination. To avoid these contaminations we moved to vacuum-fabricated samples and used scanning force microscopy, where the force is measured, and no current is needed. For fabricating samples in vacuum, the design of a modular sample holder was necessary. One of the crucial points is to bridge the gap between the macroscopic leads and the nanostructure itself. This problem is solved by using a set of two different masks. As a result of the fabrication in ultrahigh vacuum, the nanostructures will be accessible to scanning probe techniques without surface contamination. First results show that electrical measurements are indeed possible with this setup.
We have shown scanning probe microscopy measurements of metallic nanocontacts between controlled electromigration cycles. To our knowledge, applying the method of scanning force microscopy to nanoscale metallic contacts has never been done previously with a similar spacial resolution and good image quality. The nanowires used for the thinning process are fabricated by shadow evaporation. The highest resolution obtained using scanning force microscopy is about 3 nm. During the first few electromigration cycles the overall slit structure that starts to form a break of the nanocontact is formed. The slit first passes along grain boundaries and then at a later stage vertically splits grains in the course of consuming them. We find that first the whole wire is heated. Later during the thinning process as the slit forms the current runs over several smaller contacts and only these small spots are heated. We show by electrical measurements that at that time less power is needed than at the start of the thinning process. On wider structures (200 micrometers wide) through electromigration, a 1.5 μm-wide slit is formed, i.e. the slit is much wider than for the 500 nm wide devices where the lit was only 10-20 nm wide. Here, we observe grains inside the slit and on its edge and extensions along the direction of the current flow on the anode side. Such extensions of the slit that had previously not been observed. New grains appear to have formed on the edge of the slit, a process that had not been observed before on narrower structures. These grains are obtained on both, the anode and the cathode side. Kelvin probe force microscopy images show a local work function difference with fluctuations of 70 mV on the metal before electromigration. Between the charged electrodes, disconnected through electromigration, a work function difference of 3.2 V is observed due to local charging.
In metallic nanocontacts, atomic hops are often observed. It is one of the goals of the project to understand the origin and impact of these hops on the electrical properties of nanocontacts by imaging the related processes. However, also in scanning force microscopy such hops are often observed. They lead to jumps in the measured signals and to an increased energy dissipation. We studied such hops by numerical simulations. We attached additional atoms to our cube tip. These additional atoms are placed on purpose at positions were atomic hops are likely. The systematic computations reported can explain the infrequent jumps and very low average energy dissipation observed low temperature in previous scanning force microscopy studies on a KBr(001) sample. Close to the surface we indeed find new states separated by small energy barriers which account for those phenomena. These energy barriers strongly depend on details of the atomic arrangement in the vicinity of the tip apex.
The project aims at a better fundamental understanding of molecular electronics, where a molecule is used to replace a transistor for functional devices. To prepare the study of such nanocontacts with molecules placed between metallic leads, we have studied scanning force microscopy imaging of molecules. The growth of pentacene on KCl(001) at submonolayer coverage was studied by dynamic scanning force microscopy. At coverages below one monolayer, pentacene was found to arrange in islands with an upright configuration. The molecular arrangement was resolved in high-resolution images. In these images two different types of patterns were observed, which switch repeatedly. In addition, defects were found, such as a molecular vacancy and domain boundaries. We have found that, together with molecular islands of upright standing pentacene, a new phase of tilted molecules appears near step edges on KBr. Local contact potential differences (LCPD) have been studied with both Kelvin experiments and
density functional theory calculations. Our images reveal that differently oriented molecules display different LCPD and that their value is independent of the number of molecular layers. These results point to the formation of an interface dipole, which may be explained by a partial charge transfer from the pentacene to the surface. Moreover, the monitoring of the evolution of the pentacene islands shows that they are strongly affected by dewetting: Multilayers build up at the expense of monolayers, and in the Kelvin images, previously unknown line defects appear, which reveal the epitaxial growth of pentacene crystals.
Stepped well-ordered semiconductor surfaces are important as nanotemplates for the fabrication of 1D metallic nanostructures and offer an alternativ route of fabricating atomic-size contacts to molecules. Although measurements of step edges are challenging for scanning force microscopy (SFM), we have investigated the atomic structure using simultaneous atomically resolved SFM and Kelvin probe force microscopy (KPFM) images of a silicon vicinal surface. We find that the local contact potential difference is large at the bottom of the steps and at the restatoms on the terraces, whereas it drops at the upper part of the steps and at the adatoms on the terraces. For the interpretation of the data we performed density functional theory (DFT) calculations of the surface dipole distribution. The DFT images accurately reproduce the experiments even without including the tip in the calculations. This underlines that the high-resolution KPFM images are closely related to intrinsic properties of the surface and not only to tip−surface interactions. The Au-covered vicinal Si(111)-surface, the Si(557)-Au was studied at room-temperature and low-temperature (77 K) with scanning tunneling spectroscopy and voltage-dependent scanning tunneling microscopy. A gapped local electron density of states near the Gamma bar point was observed at different positions of the surface, i.e. at protrusions arising from Si adatoms and step-edge atoms. Within the gap region, two distinct peaks are observed on the chain of localized protrusions attributed to Si adatoms. The energy gap widens on both types of protrusions after cooling from room temperature to T = 77 K. The temperature dependence of the local electronic properties can therefore not be attributed to a Peierls transition occurring for the step edge only. We suggest that more attention should be paid to finite-size effects on the one-dimensional segments.
To make flatter nanocontacts, Pb is a good candidate systems, because it grows flat two-dimensional islands on Si, with a confinement of the electrons in one direction of space. In this system, one can observe a realization of the one-dimensional 'particle in a box' problem known from Modern Physics courses. The electrons confined in a potential well form standing-wave quantum states with quadratically increasing energy as the number of electrons increases. The energy of each state decreases with increasing width of the quantum well. The Fermi energy remains pinned to one state until it the number of electrons exceeds the number of states available. As a result the energy of the highest occupied state oscillates with increasing quantum-well width, the island thickness, as more and more states become populated. STM measurements of the quantum size effect had suffered from a dependence of the phase of the oscillation on the initial electronic state. We have now measured this oscillation directly by Kelvin probe force microscopy (KPFM) on ultra-thin lead islands on Si at liquid nitrogen temperatures. With this method, the work function and consequently the energy of the highest occupied state, the chemical potential, is directly measured. The influence of electrostatic forces on the topography measurement is suppressed. Electron depletion on the Pb islands is observed. The oscillation decays proportional to the thickness to the power of -1. Precise and unambiguous information about the quantum states is obtained without a current applied and at a well-defined voltage at equilibrium. As a consequence the quantum well states could be used for calibration purposes in Kelvin probe force microscopy.
One aim of the project is to exploit superconductivity to study the properties of metallic nanocontacts. As a first step in this direction, we have investigated the electromigration of thin Pb films with a weak link. In the normal state electromigration proceeds in the Pb film in a similar manner as we have observed for Au and Cu. For electromigration started in the superconducting state of Pb, Joule heating is suppressed, because the Pb film is superconducting and electromigration remains inactive in the superconducting film. Upon a sufficiently large current flow, first superconductivity is destroyed by heating from the feeds and by reaching the superconducting transition temperature, and only after that electromigration sets in. Several superconducting transitions are found because the transition temperature is not uniform in the thin film due to a locally varying film thickness.
The electromigration thinning process has been applied to magnetic contacts. A ring geometry was used to reproducibly place a magnetic domain wall at the nanocontact. Different resistance levels show the position of the domain-wall can be moved by applying a magnetic field into and out of the nanocontact. The resistance with and without a magnetic domain wall differs by up to 50 percent and exhibits a previously unobserved sign change. Our results can be reproduced by recent atomistic calculations for different atomic configurations of the nanocontact, highlighting the importance of the detailed atomic arrangement for the MR effect. The domain wall pinning strength is increased on reducing the contact size in line with a reduction of the wall energy in narrower constrictions. Furthermore, the angular dependency and symmetry of the depinning field is measured to determine the full pinning potential for a domain wall in a system with a narrow constriction.
We then studied scanning probe imaging of metallic nanocontacts. We used metallic nanocontacts on insulating substrates (oxidized silicon) to avoid short-circuits through the substrate during electrical measurements. We investigated scanning tunneling microscopy, but this technique relies on the tunneling current between the tip and the surface, and we found that for standard electron-beam-lithography made samples, carbon-rich remnants on the surface lead to contamination. To avoid these contaminations we moved to vacuum-fabricated samples and used scanning force microscopy, where the force is measured, and no current is needed. For fabricating samples in vacuum, the design of a modular sample holder was necessary. One of the crucial points is to bridge the gap between the macroscopic leads and the nanostructure itself. This problem is solved by using a set of two different masks. As a result of the fabrication in ultrahigh vacuum, the nanostructures will be accessible to scanning probe techniques without surface contamination. First results show that electrical measurements are indeed possible with this setup.
We have shown scanning probe microscopy measurements of metallic nanocontacts between controlled electromigration cycles. To our knowledge, applying the method of scanning force microscopy to nanoscale metallic contacts has never been done previously with a similar spacial resolution and good image quality. The nanowires used for the thinning process are fabricated by shadow evaporation. The highest resolution obtained using scanning force microscopy is about 3 nm. During the first few electromigration cycles the overall slit structure that starts to form a break of the nanocontact is formed. The slit first passes along grain boundaries and then at a later stage vertically splits grains in the course of consuming them. We find that first the whole wire is heated. Later during the thinning process as the slit forms the current runs over several smaller contacts and only these small spots are heated. We show by electrical measurements that at that time less power is needed than at the start of the thinning process. On wider structures (200 micrometers wide) through electromigration, a 1.5 μm-wide slit is formed, i.e. the slit is much wider than for the 500 nm wide devices where the lit was only 10-20 nm wide. Here, we observe grains inside the slit and on its edge and extensions along the direction of the current flow on the anode side. Such extensions of the slit that had previously not been observed. New grains appear to have formed on the edge of the slit, a process that had not been observed before on narrower structures. These grains are obtained on both, the anode and the cathode side. Kelvin probe force microscopy images show a local work function difference with fluctuations of 70 mV on the metal before electromigration. Between the charged electrodes, disconnected through electromigration, a work function difference of 3.2 V is observed due to local charging.
In metallic nanocontacts, atomic hops are often observed. It is one of the goals of the project to understand the origin and impact of these hops on the electrical properties of nanocontacts by imaging the related processes. However, also in scanning force microscopy such hops are often observed. They lead to jumps in the measured signals and to an increased energy dissipation. We studied such hops by numerical simulations. We attached additional atoms to our cube tip. These additional atoms are placed on purpose at positions were atomic hops are likely. The systematic computations reported can explain the infrequent jumps and very low average energy dissipation observed low temperature in previous scanning force microscopy studies on a KBr(001) sample. Close to the surface we indeed find new states separated by small energy barriers which account for those phenomena. These energy barriers strongly depend on details of the atomic arrangement in the vicinity of the tip apex.
The project aims at a better fundamental understanding of molecular electronics, where a molecule is used to replace a transistor for functional devices. To prepare the study of such nanocontacts with molecules placed between metallic leads, we have studied scanning force microscopy imaging of molecules. The growth of pentacene on KCl(001) at submonolayer coverage was studied by dynamic scanning force microscopy. At coverages below one monolayer, pentacene was found to arrange in islands with an upright configuration. The molecular arrangement was resolved in high-resolution images. In these images two different types of patterns were observed, which switch repeatedly. In addition, defects were found, such as a molecular vacancy and domain boundaries. We have found that, together with molecular islands of upright standing pentacene, a new phase of tilted molecules appears near step edges on KBr. Local contact potential differences (LCPD) have been studied with both Kelvin experiments and
density functional theory calculations. Our images reveal that differently oriented molecules display different LCPD and that their value is independent of the number of molecular layers. These results point to the formation of an interface dipole, which may be explained by a partial charge transfer from the pentacene to the surface. Moreover, the monitoring of the evolution of the pentacene islands shows that they are strongly affected by dewetting: Multilayers build up at the expense of monolayers, and in the Kelvin images, previously unknown line defects appear, which reveal the epitaxial growth of pentacene crystals.
Stepped well-ordered semiconductor surfaces are important as nanotemplates for the fabrication of 1D metallic nanostructures and offer an alternativ route of fabricating atomic-size contacts to molecules. Although measurements of step edges are challenging for scanning force microscopy (SFM), we have investigated the atomic structure using simultaneous atomically resolved SFM and Kelvin probe force microscopy (KPFM) images of a silicon vicinal surface. We find that the local contact potential difference is large at the bottom of the steps and at the restatoms on the terraces, whereas it drops at the upper part of the steps and at the adatoms on the terraces. For the interpretation of the data we performed density functional theory (DFT) calculations of the surface dipole distribution. The DFT images accurately reproduce the experiments even without including the tip in the calculations. This underlines that the high-resolution KPFM images are closely related to intrinsic properties of the surface and not only to tip−surface interactions. The Au-covered vicinal Si(111)-surface, the Si(557)-Au was studied at room-temperature and low-temperature (77 K) with scanning tunneling spectroscopy and voltage-dependent scanning tunneling microscopy. A gapped local electron density of states near the Gamma bar point was observed at different positions of the surface, i.e. at protrusions arising from Si adatoms and step-edge atoms. Within the gap region, two distinct peaks are observed on the chain of localized protrusions attributed to Si adatoms. The energy gap widens on both types of protrusions after cooling from room temperature to T = 77 K. The temperature dependence of the local electronic properties can therefore not be attributed to a Peierls transition occurring for the step edge only. We suggest that more attention should be paid to finite-size effects on the one-dimensional segments.
To make flatter nanocontacts, Pb is a good candidate systems, because it grows flat two-dimensional islands on Si, with a confinement of the electrons in one direction of space. In this system, one can observe a realization of the one-dimensional 'particle in a box' problem known from Modern Physics courses. The electrons confined in a potential well form standing-wave quantum states with quadratically increasing energy as the number of electrons increases. The energy of each state decreases with increasing width of the quantum well. The Fermi energy remains pinned to one state until it the number of electrons exceeds the number of states available. As a result the energy of the highest occupied state oscillates with increasing quantum-well width, the island thickness, as more and more states become populated. STM measurements of the quantum size effect had suffered from a dependence of the phase of the oscillation on the initial electronic state. We have now measured this oscillation directly by Kelvin probe force microscopy (KPFM) on ultra-thin lead islands on Si at liquid nitrogen temperatures. With this method, the work function and consequently the energy of the highest occupied state, the chemical potential, is directly measured. The influence of electrostatic forces on the topography measurement is suppressed. Electron depletion on the Pb islands is observed. The oscillation decays proportional to the thickness to the power of -1. Precise and unambiguous information about the quantum states is obtained without a current applied and at a well-defined voltage at equilibrium. As a consequence the quantum well states could be used for calibration purposes in Kelvin probe force microscopy.
One aim of the project is to exploit superconductivity to study the properties of metallic nanocontacts. As a first step in this direction, we have investigated the electromigration of thin Pb films with a weak link. In the normal state electromigration proceeds in the Pb film in a similar manner as we have observed for Au and Cu. For electromigration started in the superconducting state of Pb, Joule heating is suppressed, because the Pb film is superconducting and electromigration remains inactive in the superconducting film. Upon a sufficiently large current flow, first superconductivity is destroyed by heating from the feeds and by reaching the superconducting transition temperature, and only after that electromigration sets in. Several superconducting transitions are found because the transition temperature is not uniform in the thin film due to a locally varying film thickness.