Enhanced tunneling across nanometer-scale metal–semiconductor interfaces

Abstract

We have measured electrical transport across epitaxial, nanometer-sized metal-semiconductor interfaces by contacting CoSi 2 islands grown on Si͑111͒ with the tip of a scanning tunneling microscope. The conductance per unit area was found to increase with decreasing diode area. Indeed, the zero-bias conductance was found to be ϳ10 4 times larger than expected from downscaling a conventional diode. These observations are explained by a model, which predicts a narrower barrier for small diodes and, therefore, a greatly increased contribution of tunneling to the electrical transport. © 2002 American Institute of Physics. ͓DOI: 10.1063/1.1467980͔Electrical transport through metal-semiconductor interfaces has received tremendous interest in the past decades, both experimentally and theoretically. Nevertheless, an important shortcoming of existing models is the restriction to infinitely extending interfaces, so that all parameters vary only in the direction perpendicular to the surface. When the interface size enters the nanoscale regime, many of these models cease to apply. Only a few experiments addressing this topic have been reported. In none of them epitaxial interfaces were used. Scanning tunneling spectroscopy ͑STS͒ of metallic clusters on a semiconductor surface has been used to study small metal-semiconductor contacts. 1 In addition, experiments have been carried out in which the tip of a scanning tunneling microscope ͑STM͒ was used to contact a semiconductor surface 2,3 or a metallic cluster on a semiconductor surface 4 to form a small Schottky contact. Various deviations from the large-diode models were revealed, e.g., enhanced conductance, which was interpreted as a lower effective barrier. Besides the work that addresses a single small diode directly, measurements have been carried out on many small diodes in parallel. 5,6 In this letter, we present measurements of electrical transport through an epitaxial, nanometer-sized metalsemiconductor interface. We argue that the observations can be explained by a simple model for the Schottky barrier thickness in metal-semiconductor interfaces smaller than the free-carrier screening length ͑Debye length, L D ͒. Our model predicts an interface-size-dependent barrier thickness, leading to greatly enhanced tunneling in small Schottky diodes. The CoSi 2 /Si(111) interface used in our experiments is among the few metal-semiconductor interfaces of which reliable Schottky barrier height ͑SBH͒ values exist, mainly because it can be grown as a virtually perfect, abrupt, epitaxial interface. 7 The SBH in this system is 0.67 eV ͑for n-type Si͒ and has been measured with various techniques. 7-9 It is, therefore, a nearly ideal system to study electrical properties of metal-semiconductor interfaces and has been intensely used for that purpose. Both in our model and in the analysis of the measurements, the SBH will be considered as a given quantity, because of the well-determined character of the CoSi 2 /Si (111) interface. We do not expect that ultra-small-size effects as repor...

Direct Measurement of Quantum Confinement Effects at Metal to Quantum-Well Nanocontacts

“…The SBH shows a strong 140 meV systematic increase as the QW width d is reduced to 1 nm. Our measurements agree very well with the predicted increase from simple one-dimensional quantum confinement, adjusted by a smaller calculated decrease (70 meV for a 1 nm QW) due to environmental pinning effects [6,7] and increased image force lowering [10].The samples consisted of a sequence of GaAs QWs (5 10 16 =cm 3 n-type) with width varying between 1 and 15 nm, separated by 200 nm thick Al 0:3 Ga 0:7 As barrier layers (1 10 17 =cm 3 n-type), grown by molecular beam epitaxy on a 200 nm thick GaAs buffer layer (2-3 10 15 =cm 3 n-type) on a GaAs (001) substrate (3 10 17 =cm 3 n-type). A 1:5 m GaAs capping layer with the same doping as the QWs was used as a wide GaAs reference layer.…”

“…The SBH shows a strong 140 meV systematic increase as the QW width d is reduced to 1 nm. Our measurements agree very well with the predicted increase from simple one-dimensional quantum confinement, adjusted by a smaller calculated decrease (70 meV for a 1 nm QW) due to environmental pinning effects [6,7] and increased image force lowering [10].…”

“…There is a critical need to predict how these effects scale with contact size and geometry. Several previous studies have measured current-voltage curves through nm-sized metal contacts on homogenous semiconductor (SC) substrates, to infer effects from nonlocal ''environmental pinning'' and geometry-induced electric fields [6,7]. However, those contacts were not to semiconducting nanostructures, so effects such as quantum confinement were negligible.…”

Scaling of nano-Schottky-diodes

“…The main result is that if the size of the metal-semiconductor interface is smaller than a characteristic length l c , the thickness of the barrier is no longer determined by the doping level or the free carrier concentration, but instead by the size and shape of the diode. The resulting thin barrier in small diodes will give rise to enhanced tunneling, qualitatively explaining measurements of enhanced conduction, 3,4,10 without the necessity of assuming a reduced SBH. Moreover, experimentally observed scaling behavior and deviating IV curve shapes 10 can be explained.…”

“…The resulting thin barrier in small diodes will give rise to enhanced tunneling, qualitatively explaining measurements of enhanced conduction, 3,4,10 without the necessity of assuming a reduced SBH. Moreover, experimentally observed scaling behavior and deviating IV curve shapes 10 can be explained.…”