Letter pubs.acs.org/NanoLett
Quantum Confined Stark Effect in a GaAs/AlGaAs Nanowire Quantum Well Tube Device: Probing Exciton Localization Bekele H. Badada,† Teng Shi,† Howard E. Jackson,† Leigh M. Smith,*,† Changlin Zheng,‡ Joanne Etheridge,‡,§ Qiang Gao,∥ H. Hoe Tan,∥ and Chennupati Jagadish∥ †
Department of Physics, University of Cincinnati, Cincinnati, Ohio 45221-0011, United States Monash Centre for Electron Microscopy and §Department of Materials Engineering, Monash University, Victoria, 3800 Australia ∥ Department of Electronic Materials Engineering, Research School of Physics and Engineering, The Australian National University, Canberra, ACT 2601, Australia ‡
S Supporting Information *
ABSTRACT: In this Letter, we explore the nature of exciton localization in single GaAs/AlGaAs nanowire quantum well tube (QWT) devices using photocurrent (PC) spectroscopy combined with simultaneous photoluminescence (PL) and photoluminescence excitation (PLE) measurements. Excitons confined to GaAs quantum well tubes of 8 and 4 nm widths embedded into an AlGaAs barrier are seen to ionize at high bias levels. Spectroscopic signatures of the ground and excited states confined to the QWT seen in PL, PLE, and PC data are consistent with simple numerical calculations. The demonstration of good electrical contact with the QWTs enables the study of Stark effect shifts in the sharp emission lines of excitons localized to quantum dot-like states within the QWT. Atomic resolution cross-sectional TEM measurements and an analysis of the quantum confined Stark effect of these dots provide insights into the nature of the exciton localization in these nanostructures. KEYWORDS: Nanowires, heterostructures, quantum well, photocurrent, Stark effect
G
is narrow enough to result in exciton localization at low temperatures. In these experiments, a longitudinal electric field is applied along the length of the nanowire using metal contacts at either end. We use simultaneous photocurrent (PC), photoluminescence (PL), and photoluminescence excitation (PLE) spectroscopy in single nanowire QWT devices to probe both the delocalized and localized states within the QWT. Such measurements provide insights into the nature of the localization observed in narrow QWTs and thus how these strongly quantum confined structures might be optimized in the future. Quantum Well Tube Structure: Insights from HighResolution STEM and Band Structure Modeling. A schematic of the cross sectional structure of the quantum well tube (QWT) and expected band structure is shown below in Figure 1a. The QWT nanowires were grown using MOCVD.12,30 The core GaAs was first grown by a standard two-temperature technique using a 50 nm gold catalyst. After growth of the core, four layers of alternating AlGaAs and GaAs shells were grown resulting in a thick AlGaAs (90 nm) shell
roup III−V semiconductor nanowires have been extensively studied for their potential use for future compact and efficient optoelectronic devices ranging from solar cells to light-emitting diodes to field effect transistors.1−9 Strongly quantum confined electronic states in radial nanowire heterostructures are being explored particularly for accessing new one-dimensional geometries that are expected to be distinctively sensitive to external magnetic and electric fields.10,11 Such states with extremely high optical efficiency have been observed in GaAs quantum well tubes embedded inside an AlGaAs shell, along with a transition from delocalized one-dimensional states to zero-dimensional dot-like states with narrower well widths.10−15 Separately, quantum dots localized within the AlGaAs shell have been reported in MBE-grown QWT nanowires,16,17 along with the first attempts of modulation doping these structures.13,18 In both self-assembled quantum dots,19−26 dots that are formed naturally through well-width fluctuations in 2D quantum wells27 and dots in nanowires defined by axial heterostructures,28,29 the application of electric fields laterally and vertically can provide important information on the potential profile and size of the quantum dots. This Letter describes experiments where this quantum confined Stark effect is observed in a semiconductor nanowire QWT where the well © 2015 American Chemical Society
Received: June 22, 2015 Revised: November 6, 2015 Published: November 12, 2015 7847
DOI: 10.1021/acs.nanolett.5b04039 Nano Lett. 2015, 15, 7847−7852
Letter
Nano Letters
Figure 1. (a) Schematic of a QWT nanowire cross section and expected band diagram. (b) HAADF-STEM images for 4 and 8 nm QWTs. (c) Atomic resolution HAADF-STEM image of 4 nm QWT. (d) Cross-sectional Al concentration map with unit cell resolution. (e) Al concentration line profile across the GaAs quantum well, integrated across the width of the region indicated by the white rectangle in (c). The atomic resolution images show the extremely high crystalline quality of the NWs which is consistent with their high efficiency luminesence emission.12,14
image was recorded with an electron beam semiconvergent angle of 16.2 mrad and detector collection angles from 46−200 mrad. The corresponding aluminum concentration map is shown in Figure 1d. A typical composition line profile across the quantum well is shown in Figure 1e. The average aluminum concentration in the neighboring barrier layer measured from the line profile “below” (0−3 nm region) and “above” (8−10 nm region) the QW is 39.9 ± 2.6%. From such extremely high-resolution images, we see that the barrier/QW interfaces are not atomically sharp. The aluminum concentration transitions between AlGaAs to GaAs and back again are at least 1 nm in width, as well as some fluctuation in the QW width in the field of view.30 In addition, fluctuations of the aluminum concentration of up to ±3% are observed. These measurements provide accurate, high spatial resolution measurements of the Al composition across the cross-section of the wire, but only small volumes of the total nanowire can be sampled. Moreover, it is not yet known what the fluctuations in well width or aluminum composition are along the length of the wire. To gain insight into this localization process using the quantum confined Stark effect, two terminal single QWT nanowire devices were fabricated photolithographically as described in the Supporting Information (S−II). These devices were found to be extremely photosensitive with dark currents of a few picoamperes increasing by 4 orders of magnitude under 800 nm illumination. Using spectroscopic techniques described below, we show that these contacts are making excellent electrical contact with the QWT. Photoluminescence and Photocurrent Spectroscopy of QWT Nanowire Devices. Figure 2 shows photoluminescence (PL) spectra at 10 K of both the 8 and 4 nm QWT nanowire devices as a function of bias. The devices are photoexcited well above the AlGaAs barriers by 200 fs pulses at 590 nm with an average power of less than 4 μW and focused onto the NW to a 2 μm spot using a 50X long working length objective. The emission from the QWT dominates over the core emission at 10 K. Both NW devices exhibit weak emission from the core. The 8 nm quantum well tube nanowire shows intense luminescence at 1.56 eV that is 50 meV higher than the GaAs band edge (see Figure 2a), while the 4 nm quantum well tube has intense broader luminescence centered at 1.67 eV, 140
surrounding the core followed by a 4 or 8 nm GaAs quantum well, then an outer 25 nm AlGaAs shell and finally a 10 nm GaAs cap to prevent oxidation. The thin GaAs layer sandwiched between the two AlGaAs shells defines the quantum well tube. Figure 1b shows cross-sectional HAADFSTEM images of the nominally 4 and 8 nm QWT nanowire samples. In the images, the GaAs QWT are well-defined narrow stripes embedded within the AlGaAs barriers which surround the 50 nm GaAs core. Using the QWT widths obtained from the cross sections, we can calculate numerically (see Supporting Information, Table S1) the confined state eigenvalues of electrons and holes in the QWT by solving the Schrodinger equation for an equivalent cylindrical quantum well.12 From this information, we can thus predict the energies of the optical transitions (Supporting Information, Table S2 and schematically in Figure S2). We find that there are two confined electron states for the 8 nm QWT but only one for the 4 nm QWT. These optical transitions predicted from calculations are used to analyze the single nanowire spectroscopic data presented below. The 8 and 4 nm QWT widths were chosen to lie on either side of a transition from delocalized excitonic states (the 8 nm well) to localized states (the 4 nm well) at low temperatures.14 To understand the nature of the disorder responsible for this localization, we utilize an aberration-corrected high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) using a Titan3 80-300 standard Schottky field emission gun TEM/STEM operated at 300 keV to probe the morphology and composition of the cross-sectional quantum well tube. The aluminum concentration in a rectangular section within ∼3 nm of either side of the GaAs quantum well is measured using the quantitative HAADF-STEM method.31 In this approach, the intensity distribution in the HAADF-STEM image is converted to a map of aluminum concentration via comparison with comprehensive frozen-phonon dynamical electron scattering calculations, incorporating all of the relevant electron-optical parameters (measured separately).32−35 This approach requires a much lower electron dose than spectroscopic methods, thus minimizing the risk of electron beam damage. Figure 1c shows an atomic resolution HAADFSTEM image of the QW and the surrounding AlGaAs barrier layer in the nominally 4 nm QW sample. The HAADF-STEM 7848
DOI: 10.1021/acs.nanolett.5b04039 Nano Lett. 2015, 15, 7847−7852
Letter
Nano Letters
The 4 nm QWT emission appears to quench completely at biases above 6 V, which is slightly higher than what was observed in the 8 nm QWT. This may reflect an increased exciton binding energy in the smaller width QWTs. Photocurrent Spectroscopy of QWT Devices. The ionization of excitons at low temperature in the QWT devices strongly suggests that the contacts are making good electrical contact to the quantum wells. To confirm this, we take photocurrent spectra from each device, which should enable the observation of the ground and excited states of the QWT. Figure 3a shows a photocurrent spectrum from the 8 nm QWT
Figure 2. (a,b) Photoluminescence spectra at different applied biases for 8 and 4 nm QWT, respectively. (c) Plot of integrated intensity versus the applied bias for the 8 nm QWT and (d) plot of PL integrated intensities for high energy side (>1.67 eV) and low energy side (
