Letter pubs.acs.org/NanoLett
Electrical and Surface Properties of InAs/InSb Nanowires Cleaned by Atomic Hydrogen James L. Webb,*,† Johan Knutsson,† Martin Hjort,† Sepideh Gorji Ghalamestani,‡ Kimberly A. Dick,‡,§ Rainer Timm,† and Anders Mikkelsen*,† †
Division of Synchrotron Radiation Research, Lund University, Lund 221 00, Sweden Division of Solid State Physics, Lund University, Lund 221 00, Sweden § Center for Analysis and Synthesis, Lund University, Lund 221 00, Sweden ‡
ABSTRACT: We present a study of InAs/InSb heterostructured nanowires by Xray photoemission spectroscopy (XPS), scanning tunneling microscopy (STM), and in-vacuum electrical measurements. Starting with pristine nanowires covered only by the native oxide formed through exposure to ambient air, we investigate the effect of atomic hydrogen cleaning on the surface chemistry and electrical performance. We find that clean and unreconstructed nanowire surfaces can be obtained simultaneously for both InSb and InAs by heating to 380 ± 20 °C under an H2 pressure 2 × 10−6 mbar. Through electrical measurement of individual nanowires, we observe an increase in conductivity of 2 orders of magnitude by atomic hydrogen cleaning, which we relate through theoretical simulation to the contact-nanowire junction and nanowire surface Fermi level pinning. Our study demonstrates the significant potential of atomic hydrogen cleaning regarding device fabrication when high quality contacts or complete control of the surface structure is required. As hydrogen cleaning has recently been shown to work for many different types of III−V nanowires, our findings should be applicable far beyond the present materials system. KEYWORDS: nanowire, STM, InSb, InAs, III−V, heterostructure emiconductor nanowires represent a field of active research with regards to developing new types of semiconductor device, particularly for optoelectronic (solar cells1,2 and lighting3) and computing (nanowire field effect transistors and tunnel diodes4−6) applications and in terms of fundamental physics research. In particular, the antimonide compounds (InSb, GaSb) are of interest for a range of applications due to their narrow bandgap7 and because they can ideally form a broken bandstructure at a heterostructure interface with other III−Vs such as InAs.8−10 Advances in the defect free synthesis of such heterostructured nanowires from the initial work by Park,11 Carloff,12 and Ercolani et al.13 has made it possible to grow such nanowire structures for experimental study where the crystal structure, aspect ratio, and ultimately electrical properties of the nanowire can be highly controlled through careful variation in growth parameters.14,15 Due to the potential for 1D confinement and large surface area, the electrical conductance properties of nanowires are extremely sensitive to their surface properties, in terms of structure, oxidation state, or due to the influence of any surface adsorbents on the nanowire. A precise understanding of the nanowire surface is thus required to fully understand the behavior of the nanowire, for which highly surface-sensitive techniques such as X-ray photoemission spectroscopy (XPS) or scanning tunneling microscopy (STM) in ultrahigh vacuum are ideally suited. For STM, scanning on the nanowire is hindered by the presence of the native surface oxide layer which forms on exposure to air after growth. Several techniques exist to remove
S
© XXXX American Chemical Society
such an oxide layer but these are either ex situ such as chemical etchingmeaning the nanowire will reoxidize before it can reach the STMor will damage the nanowire surface such as dry etching using reactive ion or argon ion (Ar + ) techniques.16−19 A nondestructive alternative is to use atomic hydrogen cleaning (H*) whereby hydrogen atoms, split from their molecular form in a thermal cracker, are directed at a nanowire surface that has been preheated to a suitable temperature in order to react with and remove the native oxide.20 Our previous work has shown this process to work for GaAs,20 InP,21 InAs-only nanowires and for STM and AFM scanning gate microscopy measurements on InAs/GaSb nanowires.22,23 Having a perfectly clean nanowire surface for a nanowire device configuration opens up the possibility for unprecedented atomic scale measurement and control, with the tip acting either as a gate or as a probe for determining local density of states. To date, no STM studies of InSb nanowires has been carried out. In fact, InSb surfaces have generally received little attention compared to other III−V materials until very recently. Previous work on planar substrate by Bell et al. has identified temperatures of the order of 380 °C to clean InSb with atomic hydrogen (along with an Sb4 flux),24 similar in temperature to Received: January 23, 2015 Revised: May 5, 2015
A
DOI: 10.1021/acs.nanolett.5b00282 Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
Figure 1. (a), (b) Schematic and SEM image of 500/600 nm length InAs/InSb nanowires as grown on InAs(111)B substrate taken at an angle of 30° to the substrate plane. Nanowire diameter is 60 nm for InSb and 30 nm for InAs. (c) TEM image of the heterojunction area, showing the primarily wurtzite structure narrower InAs section at the bottom, widening to the thicker zincblende InSb at the top. Native oxide thickness can be estimated at approximately 1 nm from the image.
that required to clean InAs.25 Some previous work has been undertaken in the past on planar InSb, particularly the distinctive c(8 × 2) reconstructed surface of InSb (001)26,27 where recent work has focused on the adsorption of large, organic molecules onto the InSb surface.28,29 The cleaved InSb(110) surface has been studied by Whitman et al.,30 showing an unreconstructed and unpinned surface similar to other cleaved III−V (110) surfaces. The InSb (111)A and B surface has been studied by Edamoto et al.31 and Nishizawa.32 Other work has focused on epitaxial growth of InSb on other substrates, such as Si or GaAs.33 A dramatic recent surge in interest for InSb has been in the contacting of InSb (or the InSb part of) nanowires to a superconductor and studying the conductance behavior of such a system with regards to creating Majorana states at the ends of a nanowire34−37 or in alternative InSb nanostructures such as the nanocrosses fabricated by Plissard et al.38 Such behavior is commonly understood to be extremely sensitive to the nature of the interface between the materials, in terms of the magnitude of the proximity induced superconducting gap energy39 and due to the spin sensitivity of the transport processes involved such as Andreev reflection.40−43 Ideally, a good quality tunnel barrier (free of defects with native oxide removed or passivated) or a clean (ohmic) semiconductormetal contact is desired. This is commonly achieved through either Ar ion milling44 or etching to create a transparent interface (followed by tunnel barrier oxide deposition if desired), which are techniques that can induce unwanted effects or damage to the nanowire surface. Alternatively, with no such processing, the native oxide layer on the nanowire is retained along with undesired interfacial states. Both of these can result in less than ideal junction and nanowire properties.45,46 Alternatively, recent work by Krogstrup et al.47 and
Chang et al.48 have used deposition of Al onto as-grown InAs nanowires in situ without breaking vacuum to achieve the desired superconductor−nanowire interface for devices. Although possible for a range of III−V materials, this has only been demonstrated for nanowires grown by molecular beam epitaxy (MBE) and can only be done at the initial stage of nanowire growth as opposed to on a patterned or fully fabricated device. A further possibility is capping the nanowires after growth in situ with a thin layer of As or Sb that can be later removed through thermal annealing to leave an oxide-free nanowire surface.49,50 This is, however, again limited to MBE grown nanowires and raises difficulties with device fabrication because (by most common nanowire contacting schemes) the decapping would have to take place after lithographic patterning but prior to deposition. This risks hard baking the organic resist used for lithography, prohibiting good quality liftoff of metallic contacts. In situ H* cleaning of the InSb nanowire potentially provides an alternative to these other techniques in terms of creating a clean nanowire surface, onto which a high quality contact can be made, on any range of III−V materials not limited to those grown by MBE. In this work, we show that H* cleaning can be used at a much later stage of fabrication, demonstrated to work effectively for nanowires deposited on a typical device substrate such as Si/SiO2 at sufficiently low temperatures to avoid nanowire device damage. The nanowire growth was performed on InAs(111)B substrate, decorated with size-selected Au aerosol nanoparticles. In this work, we use two different types of nanowire: (1) For XPS measurements, nanowires with average lengths 350 nm and 1.5 μm and approximate diameters 30/60 nm for InAs/ InSb, respectively, and (2) for STM and electrical measureB
DOI: 10.1021/acs.nanolett.5b00282 Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
such as instability and picking up nanowires with the tip, which we attribute to poor adhesion between the nanowire and substrate. We found Vtip ≈ −0.1 V at tunnel current 300 pA with feedback loop gain 1−2% to be most stable for InSb, closer to suitable parameters for scanning on metal due to the low bandgap of InSb. XPS measurements were performed at the I311 beamline at the Max IV Laboratory, Lund, Sweden. Sample cleaning was again performed under UHV in a preparatory chamber (base pressure 5 × 10−11 mbar) using a cracker of the same efficiency and manufacturer, filament temperature (1700 °C) and distance to the sample as compared to the STM. We considered the chamber pressure of H2 in the larger beamline preparation chamber to be equivalent to the pressure used for cleaning in the STM system, something that was verified by test cleaning with III−V substrate (InAs) in both systems. The sample was heated using a rear-mounted electron beam heater rather than a resistive heater as used in the STM. Cleaning was again performed between 20 and 30 min, with approximately 20 min run-up and sample heating time beforehand. A slightly longer cooldown period of 30 min afterward was required in order to obtain sufficiently low pressure to transfer to the beamline analysis chamber, with the transfer performed in situ at pressure
