Aluminum-Induced Photoluminescence Red Shifts in Core–Shell

Jul 30, 2013 - Department of Applied Physics and Nanomicroscopy Center, Aalto University, P.O. Box 15100, FI-00076 Aalto, Finland. Nano Lett. , 2013, ...
0 downloads 0 Views 4MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Aluminum-Induced Photoluminescence Red Shifts in Core−Shell GaAs/AlxGa1−xAs Nanowires Veer Dhaka,*,† Jani Oksanen,‡ Hua Jiang,∥ Tuomas Haggren,† Antti Nykan̈ en,∥ Reza Sanatinia,§ Joona-Pekko Kakko,† Teppo Huhtio,† Marco Mattila,† Janne Ruokolainen,∥ Srinivasan Anand,§ Esko Kauppinen,∥ and Harri Lipsanen† †

Department of Micro- and Nanosciences, Micronova, Aalto University, P.O. Box 13500, FI-00076 Aalto, Finland Department of Biomedical Engineering and Computational Science, Aalto University, P.O. Box 12200, FI-00076 Aalto, Finland § School of Information and Communication Technology, KTH Royal Institute of Technology, Electrum 229, S-164 40 Kista, Sweden ∥ Department of Applied Physics and Nanomicroscopy Center, Aalto University, P.O. Box 15100, FI-00076 Aalto, Finland ‡

ABSTRACT: We report a new phenomenon related to Alinduced carrier confinement at the interface in core−shell GaAs/AlxGa1−xAs nanowires grown using metal−organic vapor phase epitaxy with Au as catalyst. All AlxGa1−xAs shells strongly passivated the GaAs nanowires, but surprisingly the peak photoluminescence (PL) position and the intensity from the core were found to be a strong function of Al composition in the shell at low temperatures. Large and systematic red shifts of up to ∼66 nm and broadening in the PL emission from the GaAs core were observed when the Al composition in the shell exceeded 3%. On the contrary, the phenomenon was observed to be considerably weaker at the room temperature. Cross-sectional transmission electron microscopy reveals Al segregation in the shell along six Al-rich radial bands displaying a 3-fold symmetry. Time-resolved PL measurements suggest the presence of indirect electron−hole transitions at the interface at higher Al composition. We discuss all possibilities including a simple shell−core−shell model using simulations where the density of interface traps increases with the Al content, thus creating a strong local electron confinement. The carrier confinement at the interface is most likely related to Al inhomogeneity and/or Al-induced traps. Our results suggest that a low Al composition in the shell is desirable in order to achieve ideal passivation in GaAs nanowires. KEYWORDS: GaAs/AlGaAs, core−shell nanowires, MOVPE, MOCVD, Al segregation, TRPL

U

Core−shell GaAs/AlGaAs NWs is the most widely used method for passivation of GaAs NWs. A higher band gap AlGaAs shell grown on GaAs core NWs moves the surface states away and confines the carriers to the core, thereby preventing the carriers from recombining nonradiatively. Further, since Al and Ga atoms have nearly the same lattice constant, AlxGa1−xAs shells can be grown nearly strain free on GaAs NWs. However, the band bending at the GaAs/AlGaAs interface due to traps or doping may lead to the formation of a 2D electron enrichment layer.11 Although there are numerous well-documented research papers on the core−shell GaAs/ AlGaAs passivation, hardly any systematic report exists on the passivation of GaAs NWs as a function of Al composition in the shell and the systematic effects of this 2D accumulation layer on the optical properties of GaAs core NWs. In this study, we report a new carrier confinement related phenomenon observed in Au-catalyzed core−shell GaAs/ AlxGa1−xAs NWs grown on Si substrate using atmospheric

nique properties arising due to their nanoscale radial footprints, III−V semiconductor nanowires (NWs) continue to generate intense interest. NW-based electronic and photonic devices such as solar cells,1 lasers,2 light-emitting diodes,3 photodetectors,4 and transistors5 have already been demonstrated. Among binary III−V semiconductors, GaAs with a direct bandgap of 1.42 eV and with very high electron mobility (∼8500 cm2 V−1 s−1) is particularly promising for photovoltaic and high-frequency applications. However, GaAs has also the highest known surface recombination rate (∼106 cm/s) of charge carriers among III−V materials.6 Further, GaAs NWs, in particular, are associated with a very high density of surface states that promotes undesirable nonradiative recombination channels at the NW surface.7−10 At the surface, the Fermi level is strongly pinned by these surface states to the middle of the bandgap causing a rapid depletion of carriers leading to the formation of a depletion shell.6−9 For lightly doped or undoped GaAs NWs, the surface is totally depleted, and the depletion shell could even extend to the full length of the NW making them highly resistive or even semiinsulating.7,10 Therefore, surface passivation is essential for application of GaAs NWs in functional devices. © XXXX American Chemical Society

Received: April 9, 2013 Revised: July 15, 2013

A

dx.doi.org/10.1021/nl4012613 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

∼40 nm diameters from the top, but a small distribution of thicker wires with ∼80 nm diameter was also observed, most likely due to merging of two or more Au catalyst particles during the pregrowth annealing step. On the contrary, the core−shell NWs (Figure 1) are thicker from the top segment (not visible here) and thinner from rest of the NW body depending on the Al composition in the shell. SEM images of NWs viewed from the top and from the cross-section show that approximately 70% of the NWs have grown in the vertical direction from the substrate. In all core−shell samples, a very high density of about 1010 NWs/cm2 was observed. High-resolution transmission electron microscopy (TEM) measurements were carried out with a JEOL 2200FS double aberration corrected FEG microscope operated at 200 kV. Al composition in the shell was determined using a TEM integrated energy-dispersive X-ray (EDX) spectroscopy tool with a detection limit of 0.2 at. %. Figure 2a−c shows high-

pressure metal−organic vapor phase epitaxy (MOVPE). In our experiments, large red shifts in energy and broadening of the PL spectra from GaAs core NWs were observed in samples with Al compositions exceeding 3% at low temperatures. Furthermore, we show that the shell growth rate increases significantly with increasing Al source flows. Optical properties of GaAs core NWs are discussed in view of Al inhomogeneity in the shells, and the PL red shifts are explained based on a shell−core−shell model suggesting a very strong carrier confinement at the GaAs/AlxGa1−xAs interface leading to indirect electron−hole recombinations. We show that in highly depleted MOVPE grown GaAs NWs the Al composition in the shell should be optimized to achieve an ideal passivation effect. GaAs/AlxGa1−xAs core−shell NWs were grown on Si (111) substrates in a horizontal flow atmospheric pressure MOVPE system using trimethylgallium (TMG), tertiarybutylarsine (TBA), and trimethylaluminum (TBA) as precursors for gallium (Ga), arsenic (As), and aluminum (Al) sources, respectively.12 Hydrogen was used as a carrier gas, and the total reactor gas flow rate was ∼5 slm. A 40 nm diameter colloidal gold (Au) nanoparticles were used as catalyst. After the substrate cleaning followed by subsequent etching of the native oxide from the Si substrate, the samples were loaded in the MOVPE reactor. Prior to the growth, the Si substrate was annealed in situ at 650 °C for 10 min under hydrogen ambient to desorb surface contaminants. The MOVPE growth of GaAs core NWs was started by switching on the TMG and TBA sources simultaneously for 5 min at a fixed growth temperature of 470 °C. The nominal V/III ratio during the growth was ∼25, and the TMG and TBA flows were 7 and 75 sccm, respectively. After the growth of GaAs core, the temperature was ramped to 650 °C for the growth of AlxGa1−xAs shell. All AlxGa1−xAs shells were grown for a similar duration of 20 s, and the molar ratio XAs/XGa+ XAl in the shells was kept constant at 10. Structural properties of the core−shell GaAs/AlGaAs NWs were determined by scanning electron microscopy (SEM) (Zeiss Supra system operating at 3 kV) measurements. Figure 1 shows an SEM image of core−shell GaAs/Al3.1%Ga96.9%As NWs grown on Si substrate. The average length of GaAs NWs without the shells was ∼20 μm and the growth rate ∼4 μm/ min. As-grown NWs without the shell were tapered from the base at this growth temperature with majority of NWs showing

Figure 2. TEM colored profile images showing distinctly the GaAs core and AlxGa1−xAs shells from the top NW segment: (a) 1.2% Al, 6 nm shell; (b) 3.1% Al, 30 nm shell; (c) 15.6% Al, 52 nm shell; (d) EDX line profile showing the variation of the Al (green), Ga (blue), and As (yellow) signal intensity along the red line on a core−shell NW structure. The scale bar corresponds to 20 nm in (a) and 40 nm for (b) and (c). All shells were grown for 20 s. (e) AlxGa1−xAs shell thicknesses as a function of the Al composition for all samples.

resolution colored profile TEM images of GaAs NWs passivated with AlxGa1−xAs shells with varying Al compositions. Though the core and the shell structures can be seen distinctively in Figure 2a−c, an EDX line scan performed on a core−shell NW (Figure 2d) showing the variation of Al intensity along the red line also confirms the shell presence. Figure 2e summarizes AlGaAs shell thicknesses as a function of the Al composition for the samples used in this work. Interestingly, even though the shells growth duration and the

Figure 1. SEM image of core−shell GaAs/Al3.1%Ga96.9% NWs on Si. The core and the shell were grown at 470 and 650 °C, respectively. The scale bar corresponds to 2 μm. Inset shows a high-resolution TEM image of a single core−shell NW. B

dx.doi.org/10.1021/nl4012613 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

total group III flow rate was constant, the shell thickness at the top segment (∼5−7 μm) of the NW increases considerably with increasing Al composition. In particular, a ∼80 nm shell (25% Al) compared to only 6 nm shell (1.2% Al). The tapering of the core-shell NWs from the top can be attributed to the shadowing effect. However, considerably thicker shells at higher Al composition could be attributed to the fact that Al is a very reactive species, and the diffusion length of Al adatoms is shorter compared to the Ga adatoms.13,14 Thus, at higher Al source flows, Al adatoms with a shorter diffusion length traps more ambient species resulting in an enhanced radial growth rate. Combined with the shadowing effect in a dense ensemble of NWs, this result in very thicker uneven AlGaAs shells at higher Al compositions. A similar effect has been observed earlier with MBE grown core−shell nanowires,13 but the effect is more pronounced in MOVPE grown NWs (Figure 2e). The nonuniform AlGaAs passivation layer on GaAs core is not limited to Au catalyzed core−shell NWs, but it can also be seen in wires grown via selective-area growth (SAG), as visible in ref 6 (Figure 2a). This suggests that growing AlGaAs shells requires careful consideration because the shell thickness is dependent not only on the growth time but also on the Al composition, and a high density of NWs should probably be avoided to minimize the effect. To study radial alloy inhomogeneity in the shell, we performed high-angle annular dark field scanning TEM (HAADF-STEM) on the core−shell cross sections. STEM sample preparation was done similar to ref 15, except that a wide angle cut was performed resulting in 40 nm slices of epoxy resin containing the NWs. Figure 3a shows cross-section STEM view of a core−shell GaAs/AlGaAs NW with a 115 nm core and a 30 nm shell. Unusually thicker core indicates that the examined wire cross section is obtained from the tapered base closer to the substrate. As Al is a lighter element than Ga, it shows in a dark contrast in STEM (z-contrast), but since there is a low Al composition (4%) in the shell, the contrast between the shell and the core is not drastic. However, remarkably, six Al-rich dark bands can be clearly seen along the radial ⟨112⟩ direction where two {110} AlGaAs facets meet, indicating Al segregation. A careful observation reveals a 3-fold rotational symmetry of radial Al-rich stripes which alternate in thickness (three thick and three thin bands). Very recently, Zheng et al.16 also observed similar 3-fold symmetry in MBE grown GaAs/ AlGaAs NWs. Notably, in our case, all six Al-rich stripes extend all the way to the core. Al segregation was also observed very recently in few reports on MBE grown core−shell NWs. In particular, Heiss et al.17 reported quantum dot formation in an Al varying region of the AlGaAs shell. Rudolph et al.18 also observed Al segregation in the shell similar to ours. The Al segregation is understood to occur due to the difference in the chemical potential and different Ga and Al adatom mobility on (110)- and (112)-type facets. However, there are scarce reports on Al segregation in MOVPE grown core−shell GaAs/AlGaAs NWs. Sköld et al.15 observed Al segregation in MOVPE grown NWs but on AlInP/GaAs material. Nevertheless, Al segregation seems to be common in both MBE and MOVPE grown core− shell NWs. Figure 3b shows EDX line profile performed along the dark Al-rich line just missing the opposite stripe. EDX analysis reveal that the Al signal was roughly 2 times higher (9% Al) along the Al-rich stripes than elsewhere on the body of the shell (4% Al), in agreement with Rudolph et al.18 Further, Al signal drops completely to zero across the GaAs core, indicating pure GaAs stoichiometric composition (Figure 3b). In GaAs

Figure 3. (a) STEM-HAADF micrograph showing the cross-section view of a core−shell GaAs/Al4%Ga96%As NW with six Al rich stripes along ⟨112⟩ direction within the shell. Alternative stripes vary in thickness giving a 3-fold symmetry along the NW axis. The scale bar corresponds to 30 nm (b) EDX line scan profile performed parallel to one side of the Al-rich dark stripe in (a) while the line scan is slightly missing the opposite stripe (red dotted scan line intentionally slightly shifted away from the stripe for clarity). Al composition is 4% within the AlGaAs shell body and 9% along the Al-rich bands.

core, no Al diffusion or intermixing of any element was observed in EDX analysis. In view of the Al inhomogeneity across the shell, we will next investigate the optical properties of the core−shell structures. For instance, the Al-rich regions at the core−shell interface could act as trap centers which might alter the optical properties of the core. PL measurements were performed in order to study the optical properties of core−shell GaAs/AlGaAs NWs. The core−shell samples were excited by a 532 nm laser with a spot size of ∼100 μm. A closed cycle cryostat was used for lowtemperature measurements at 15 K. A liquid nitrogen cooled germanium detector and standard lock-in techniques were used for signal detection and data capture. Instead of relying on a single NW for PL measurements, PL originating from a statistical average from a large ensemble of NWs (100 μm laser spot size) provides more accurate information on the optical properties. Figure 4a shows low-temperature (15 K) PL spectra from GaAs NWs passivated with AlxGa1−xAs shells. For the purpose of simplicity, the Al compositions (x) in the shells Alx%Ga(1−x)% As are indicated in percent (%) throughout the text. We could not extract any emission from bare as-grown GaAs NWs with our setup, suggesting the presence of high nonradiative channels at the surface. For similar extra-long GaAs NWs as ours, Chang6 and Demichel et al.7 predicted that for critical diameters below 100 nm, the depletion shell due to Fermi level pinning can extend throughout the whole NW diameter, leading to the suppression of radiative channels. Thus, our MOVPE grown GaAs NWs (∼40 nm diameter) surface is likely to be totally depleted of free carriers due to C

dx.doi.org/10.1021/nl4012613 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

possibilities: a high laser excitation could heat the NWs locally thereby changing the band gap, which could result in red shifts.19 However, power dependence PL measurements show a clear blue shift in the PL spectra (discussed in the next section); therefore, we rule out the laser heating effect. Second, we do not have any indication of strain or either diffusion occurring at the GaAs/AlGaAs interface. For strain to occur, the shell has to considerably thicker than reported in this work. To confirm the same, we performed PL measurements on samples with increasing shell thickness at a uniform Al composition of 3%. When the shell thickness exceeds more than 100 nm, blue shifts rather than red shifts were observed (not shown here). Thus, we rule out the possibility of appreciable strain in our core− shell NWs. In contrast, Hocevar et al.,20 very recently, reported the origin of red shifts in their GaAs/AlGaAs NWs to be associated with tensile strain and piezoelectric fields. However, the magnitude of the shifts they reported is rather small (14 meV) compared to our results, and their core−shell structures were grown via MBE using self-catalyzed approach (average length 3 μm). Moreover, their studies are based at a fixed Al composition (35%) in the shell. Next, for diffusion of Al to the GaAs core, again a blue shift is expected. Finally, quantum confinement is also ruled out as the NW diameter (∼40 nm) is well above the GaAs Bohr radius (14 nm).21 We will discuss later that the origin of these red shifts is in fact related to the carrier accumulation at the GaAs/AlGaAs interface. On the contrary, only slight red shifts in GaAs PL emission were observed at RT (Figure 4b). Samples up to 9% Al in the shell show a PL peak close to 871 nm at similar excitation power, as expected from the bulk GaAs material at RT.12 Overall, the shifts are rather negligible compared to low-temperature measurements. We could not detect any presentable PL spectra at RT from the sample with 25% Al in the shell. To get more insight into the PL dynamics, we will discuss next the power excitation dependence of the PL spectra for samples with different Al compositions. Figure 5a shows 15K PL spectra from GaAs core NWs passivated with a Al1.2%Ga98.8%As shell as a function of laser power excitation. At 1.5 mW excitation, two distinct PL peaks are visible: the peak at 820 nm is associated with exciton bound-to-acceptors22,23 (D0, X), and the peak at 831 nm is attributed to neutral donor-to-acceptor transitions pair (DAP)21,22 associated with carbon acceptor impurities. Joyce

Figure 4. Normalized PL spectra from GaAs core NWs passivated with AlxGa1−xAs shells at (a) 15 K and at (b) RT. In (a), a clear red shift and broadening (from 820 nm, 1.2% Al in the shell, fwhm 30 meV to 886 nm, 20% Al, fwhm 91 meV) can be seen in PL spectra from GaAs core NWs with increasing Al incorporation. Laser excitation power was 25 mW in (a) and 75 mW in (b). The dashed lines are guides to the eye.

strong Fermi level pinning by the surface states.7,10 It is interesting to mention here that the TEM measurements performed on bare GaAs NWs grown on Si reveal relatively clean NWs with predominantly a zinc blende structure and with only a few stacking faults12 (not shown here). As seen in Figure 4, compared to unpassivated GaAs NWs, passivation with AlGaAs shells resulted in a very strong PL signal (albeit normalized here) from the GaAs core NWs both at low T (Figure 4a) and RT (Figure 4b) and for Al compositions in the shells. For low Al compositions in the shell up to 3.1%, the main PL peak (∼820 nm) shifts only slightly. However, the PL emission from GaAs core NWs red shifts dramatically and systematically with Al composition exceeding 3.1% in the shells (Figure 4a). Also, a decrease in the PL intensity and broadening in the line widths were observed. The sample with 3.1% Al in the shell shows more than an order of magnitude higher PL intensity (not shown) compared to the one with 25% Al in the shell. Notably, PL from the sample with 1.2% Al in the shell red shifts and broadens from 820 nm (fwhm ∼ 30 meV) to as large as 886 nm (fwhm ∼ 91 meV) when the Al composition in the shell reached 25% (a 66 nm shift). We discuss here few

Figure 5. Normalized PL spectra from GaAs core NWs as a function of excitation power. (a) GaAs NWs passivated with Al1.2%Ga98.8%As shell and (b) GaAs NWs passivated with Al15.6%Ga84.4%As shell. (c) Dependence of the PL peak position on excitation power for different AlxGa1−xAs shells. D

dx.doi.org/10.1021/nl4012613 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

et al.22 have also reported a similar DAP peak in core−shell GaAs/AlGaAs NWs. At lower excitation power, the DAP peak dominates, but with increasing laser power excitation the excitonic peak becomes dominant and the impurity band related DAP peak disappears due to the rapid saturation of carbon related impurity defect states. Clearly, no shifts in peak PL position as a function of excitation power were observed in either excitonic or the DAP peak. This suggests that no appreciable band banding is present in GaAs/AlGaAs NWs at this Al composition. A similar PL dynamics was also observed for sample with 2% Al in the shell. Interestingly, at increasing Al composition, the DAP peak disappears altogether with only a single peak visible which shows rapid red shifts and broadening of the PL spectra. Figure 5b shows PL spectra from GaAs core NWs passivated with Al15.6%Ga84.4%As shell as a function of laser excitation power. At the lowest excitation of 1.5 mW, only a single PL peak is visible at 875 nm. Compared to the sample with 1.2% Al, the PL emission red shifts to as large as 50 nm in this case. As the excitation power was increased, a clear blue shift in the PL spectra can be seen. Figure 5c depicts PL peak positions extracted from GaAs core NWs as a function of excitation power for different AlxGa1−xAs passivating shells. Appreciable blue shifts at lower excitation and saturation thereafter can be seen clearly when the Al composition in the shell exceeds 3.1%. A change in peak PL positions with excitation power indicates substantial band bending present in GaAs/AlGaAs NWs similar to as observed in InP NWs.24 This is due to the fact that when the excitation intensity is increased, the enhanced steady-state concentration of excess electrons and holes reduces the bending of energy band edges at the heterointerface due to an increased screening effect and a consequent reduction in the space charge potential.19 However, a large shift in the PL peak position and the blue shift in the PL spectra with excitation also indicate that the origin of this PL emission is not arising from the electron−hole recombinations in the GaAs flat band region. To gain further insight on the carrier dynamics, time-resolved photoluminescence (TRPL) measurements were performed on an ensemble of core−shell NWs with different Al composition in the shell. TRPL data were extracted at the peak of the steadystate PL spectra (Figure 4a). Best fit to the experimental data was obtained using the biexponential decay model25 with two decay times (τ1 and τ2). Initially, a 400 nm excitation by the mode-locked pulses creates large number of photoexcited carriers both in the shell and the core; after a short thermalization time the carriers rapidly diffuse from the shell to the core and recombine normally first via the defects at the interface (shorter decay time) and later away from the interface (longer decay time), i.e., in the GaAs flat band region. As seen in Figure 6a,b, an initial very fast decay component (τ1) followed by a slower decay time (τ2) was observed for all the samples. Interestingly, the longer decay time of GaAs core increases steadily from 0.6 ns (1.2% shell Al) to 2.6 ns (25% shell Al). This is unusual as the PL emission intensity decreases with increasing Al in the shells, and one would normally expect a shorter decay time at higher Al shell composition. On the other hand, the average faster decay component is of the level of our system response time (80 ps) for all samples except for the sample with 25% Al in the shell (τ1 = 0.5 ns). However, since the steady-state PL data (Figure 4a) shows large red shifts and the emission efficiency decrease at higher Al compositions in the shell, a longer decay time at higher Al composition suggests strongly that the recombinations are indirect in nature

Figure 6. (a) TRPL GaAs spectra from ensemble of core−shell GaAs/ AlxGa1−x As NWs with different Al compositions in the shell. Red lines are biexponential fits to the experimental data shown in different colors. (b) Decay times (τ1 and τ2) as a function of shell Al composition for all the samples.

and are likely to be associated with local carrier confinement at the interface. In confined and indirect transitions, the lifetime increases as electron−hole (e−h) overlap is reduced considerably and carriers decay much more slowly. An increase in the faster decay component to 0.5 ns can be attributed to the fact that a high Al composition in the shell reduces the leakage to the interface, which in-turn can enhance the shorter lifetime component. Thus, the faster decay time at higher Al composition in the shell is attributed to recombination at the interface and/or to minor GaAs flat band recombination, and the longer decay component is attributed to the indirect carrier recombinations from the interface to the GaAs flat band region. In summary, PL data correlated with the TRPL data suggest strongly the presence of indirect electron−hole recombinations at higher Al composition in the shell. We propose a model to understand the shifts in the PL spectra from GaAs core NWs passivated with AlxGa1−xAs shells. The model is based on the electron confinement at the GaAs/ AlGaAs interface induced by the interface defects acting as traps. The interface defects are most likely related to Al inhomogeneity/Al induced traps at the GaAs/AlGaAs interface (Figure 3a). We expect Al segregation to increase furthur with higher Al composition in the shells. Next, a significant bend bending that can confine the carriers locally at the interface could be achieved either with a very high doping (≥1018 cm−2) or due the presence of traps. Carbon impurity doping could also increase the interface trap density. However, compared with the planar 2D growth at temperatures similar to the shell growth temperature, the background carbon impurity is unlikely to exceed ∼1016 cm−2, and this will produce essentially negligible carrier confinement at the interface. Further, we assume that the MOVPE grown GaAs core and the AlGaAs shell are undoped or lightly doped. Usually, undoped or lightly doped core−shell GaAs/AlGaAs structures are slightly n-type.11 E

dx.doi.org/10.1021/nl4012613 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

At room temperature, however, no appreciable red shifts (Figure 4b) were observed, indicating that E transitions mostly disappear with recombination occurring mainly due to band to band recombination (∼871 nm). This suggests that at RT the majority of the carriers will the thermalized to the GaAs flat band region. In order to better understand the effects of band bending on the shifts in the PL spectra, we modeled the band bending and the resulting carrier confinements and PL shifts (1) due to different background impurity levels in the core or the shell, (2) due to interface and surface charge densities, (3) for different temperature, (4) for different excitation levels, and (5) the combinations of the previous items. The band bending and band diagram of the wires in radial direction are obtained by solving the standard continuum Poisson’s equation26,27 adapted to spherical coordinates and with extra terms accounting for the additional trapped surface charge at the interface. In solving the Poisson’s equation, we used separate quasi-Fermi levels for electrons and holes to describe the excitation, accounted for full Fermi statistics of the carriers to enable accurate description of degenerate carrier statistics and accounted for the thermal ionization of the impurities. After solving for the band bending due to doping or interface charges, we use the information from the band diagrams to solve the Schrödinger equation to find the wave functions and ground state energies of the confined carriers (electrons or heavy holes). The lowest possible emission energy is then estimated from the difference between the ground state energies of the electrons and holes. The material parameters needed in the calculations are taken from ref 28, and the temperature dependencies of the parameters that do not directly have tabulated values are interpolated from the known data. To study the effect of trapped interface charges at the GaAs/ AlGaAs interface on the red shifts, we also simulated the PL shifts and band diagrams of the NWs for different interface charge densities. Figure 8 shows the band diagrams and the lowest electron state eigenfunctions calculated for a shell Al composition of 15% at 15 K, with no interface charge in Figure 8a and an interface charge of Qs = 3 × 1012 cm−2 in Figure 8b while assuming an n-type background impurity level of 1016

A schematic illustration of a lightly doped n-type GaAs− AlGaAs core−shell NWs heterojunction with a large interface charge density is depicted in Figure 7.

Figure 7. Model showing schematic for possible transitions labeled A, B, C, and E in shell−core−shell AlGaAs/GaAs/AlGaAs NWs. GaAs core and the AlGaAs shell are assumed to be of n-type.

As shown in Figure 7, a triangular potential notch appears due to the presence of traps at the interface. The depth of this notch depends on the Al composition in the shell. When the core−shell structure is excited with a 532 nm laser, electron− hole pairs are created in the GaAs core as well in the AlGaAs shell. Carriers lying near the GaAs−AlGaAs interface in the shell will be sweeped by this notch, leading to the depletion in AlGaAs shell. This depletion region depends on the shell thickness and the diffusion length of the electrons. Similarly, some carriers from the flat band core will spill over to the notch forming an 2D electron-rich enhanced layer at the interface. Considering a situation where the carriers are present both in the GaAs flat band region and in the notch, electrons recombine with holes in the GaAs core in the form of PL radiative transitions such as A, B (∼820 nm at 15 K, excitonbound-to-acceptors Figure 5a), C (∼831 nm at 15 K, DAP, Figure 5a), or E (Figure 5b). The recombination between the electrons in the notch and the holes in the valence band gives rise to E transitions. Further, electrons collected at the notch tunnel to the n-GaAs side, away from the heterojunction and toward the flat portion of the band, in order to find an appreciable concentration of holes for radiative transition. These transitions are indirect in real space and result in emission which is at energy lower than the bound excitonic recombination. A somewhat similar E indirect transition has been reported previously in GaAs/AlGaAs based thin film structures11 in the 1980s, but it coexisted along with A, B, and C recombinations. At low Al composition (up to 3.1%) in the shell, the effect of the triangular notch formation is negligible with most of the electrons present in flat band region of the GaAs band and no E transitions occurring. At higher Al composition in the notch becomes deeper, prompting more electrons to spill over to the notch from the GaAs flat band and from the shell. At very high Al compositions, all the carriers from the flat GaAs band are sweeped to the notch with essentially very few carriers left in the flat GaAs band for recombination, and the PL is mostly originating from the E indirect transitions, resulting in large PL shifts. Also, the large blue shifts (Figure 5c) and saturation as a function of excitation power at higher Al content in the shells suggest the notch is rather deep, and even at very high laser excitation, majority of the carriers are still in the notch. In addition, at higher Al composition in the shell, indirect transitions and the built-in electric field due to accumulation of carriers in the notch can all contribute towards broadening of the GaAs PL spectra.

Figure 8. Band diagrams and lowest energy electron state probability density |Ψ|2 corresponding to (a) zero interface charge density and (b) 3 × 1012 cm−2 interface charge density as a function of the radial position. The band diagrams have been calculated at T = 15 K and under excitation that has shifted the quasi-Fermi levels of the conduction and valence band into/close to the bands. F

dx.doi.org/10.1021/nl4012613 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

cm−3. The small band bending induced by the n-type background impurity in Figure 8 does not introduce any significant electron confinement or a wavelength shift, but the band bending observed in Figure 8b already changes the wavelength from 816 to 860 nm due to indirect recombination of electrons confined in the notch and the holes in the flat valence band region. The calculated red shift is in good agreement with the observed red shift for sample with 15% Al in the shell. Figure 9 shows how the wavelength corresponding to the lowest energy indirect band-to-band transition depends on the

inhomogeneity/Al induced traps at the GaAs/AlGaAs interface. We expect Al segregation to increase further with higher Al composition in shells. TRPL measurements also support the presence of indirect transitions at the interface at 15 K. Simulations suggest that an interface charge trap density of 3 × 1012 cm−2 is sufficient to induce significant band bending and a strong electron confinement at the interface. We believe that our results should be useful in designing functional NW-based devices requiring optimum passivation of GaAs NWs.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: veer.dhaka@aalto.fi Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Part of this work was supported by the NANORDSUN project (no. 10048), Nordic Innovation Centre (NICe), Norway, and the MOPPI project under Aalto energy efficiency research programme. R.S. and S.A. acknowledge support from the Linné center for advanced optics and photonics (ADOPT) funded by the Swedish Research Council (VR) and from the European Union FP7 Network of Excellence Nanophotonics4Energy (N4E).

Figure 9. Wavelength corresponding to the lowest energy indirect band-to-band transitions as a function of the interface charge density Qs between the core and the shell at T = 15 K. Substantial change in the wavelength takes place at interface charge densities exceeding approximately 1012 cm−2. The simulated red shift is relatively insensitive to any other realistic simulation parameters than the interface charge density.

■

REFERENCES

(1) Wallentin, J.; Antuu, N.; Asoli, D.; Huffman, M.; Åberg, I.; Magnusson, M. H.; Siefer, G.; Fuss-Kailuweit, P.; Dimorth, F.; Witzigmann, B.; Xu, H. Q.; Samuelson, L.; Deppert, K.; Borgström, M. T. Science 2013, 339, 1057−1060. (2) Huang, M. H.; Mao, S.; Feick, H.; Yan, H.; Wu, Y.; Kind, H.; Weber, E.; Russo, R.; Yang, P. Science 2001, 292, 1897−1899. (3) Duan, X. F.; Huang, Y.; Agarwal, R.; Lieber, C. M. Nature 2003, 421, 241−245. (4) Wang, J.; Gudiksen, M. S.; Duan, X.; Cui, Y.; Lieber, C. M. Science 2001, 293, 1455−1457. (5) Thelander, C.; Fröberg, L. E.; Rehnstedt, C.; Samuelson, L.; Wernersson, L. E. IEEE Electron Device Lett. 2008, 29, 206−208. (6) Chang, C. C.; Chi, C.-Y.; Yao, M.; Huang, N.; Chen, C.-C.; Theiss, J.; Bushmaker, A. W.; LaLumondiere, S.; Yeh, T.-W.; Povinelli, M. L.; Zhou, C.; Dapkus, P. D.; Cronin, S. B. Nano Lett. 2012, 12, 4484−4489. (7) Demichel, O.; Heiss, M.; Bleuse, J.; Mariette, H.; i Morral, A. F. Appl. Phys. Lett. 2010, 97, 201907. (8) Parkinson, P.; Lloyd-Hughes, J.; Gao, Q.; Tan, H. H.; Jagadish, C.; Johnston, M. B.; Herz, L. M. Nano Lett. 2007, 7, 2162. (9) Schricker, A. D.; Davidson, F. M.; Wiacek, R. J.; Korgel, B. A. Nanotechnology 2006, 17 (10), 2681−2688. (10) Katzenmeyer, A. M.; Léonard, F.; Talin, A. A.; Wong, P.-S.; Huffaker, D. L. Nano Lett. 2010, 10, 4935. (11) Yuan, Y. R.; Pudensi, M. A. A.; Vawter, G. A.; Merz, J. L. J. Appl. Phys. 1985, 58, 397. (12) Dhaka, V.; Haggren, T.; Jussila, H.; Jiang, H.; Kauppinen, E.; Huhtio, T.; Sopanen, M.; Lipsanen, H. Nano Lett. 2012, 12, 1912− 1918. (13) Zhou, H. L.; Hoang, T. B.; Dheeraj, D. L.; Helvoort, A. T. J.; Liu, L.; Harmand, J. C.; Fimland, B. O.; Weman, H. Nanotechnology 2009, 20, 415701. (14) Chen, C.; Shehata, S.; Fradin, C.; laPierre, R.; Couteau, C.; Weihs, G. Nano Lett. 2007, 7, 2584. (15) Sköld, N.; Wagner, J. B.; Karlsson, G.; Hernàn, T.; Seifert, W.; Pistol, M.-E.; Samuelson, L. Nano Lett. 2006, 6, 2743−7. (16) Zheng, C.; Wong-Leung, J.; Gao, Q.; Tan, H. H.; Jagadish, C.; Etheridge, J. Nano Lett. 2013, DOI: 10.1021/nl401680k.

interface charge density at 15 K. A pronounced red shift can be observed as the interface charge density increases. The red shift can therefore be readily explained by a large trap density and the trapping of either electrons or holes at the interface. However, although plausible, even this explanation is not conclusive because the surface charge density and it is dependence on the trap density and temperature are not currently known in detail. The model predicts red shifts for large interface charge density even at RT. However, we do not know yet how the charge density and trap density depends on one another at RT. At RT, the surface charge density could be almost independent of the trap density. Current model only studies the transition energy of the lowest indirect transition. The PL measurement results suggest that the charge density associated with the interface traps is much larger at LT than at RT. To experimentally verify this, however, further studies are needed. In summary, we report a new phenomenon related to the carrier confinement at the interface in Au-catalyzed core−shell GaAs/AlxGa1−xAs nanowires grown on Si using MOVPE. PL from GaAs core NWs was found to be a strong function of Al in the shell at low temperatures. Large systematic red shifts and broadening of the PL spectra from GaAs core NWs were observed when the Al composition exceeded 3%. STEM crosssectional view of core−shell NWs reveals Al segregation in the shell displaying a 3-fold symmetry. To understand the observed PL red shifts, various possibilities were discussed including a shell−core−shell model based on the carrier confinement at the GaAs/AlxGa1−xAs interface due to the presence of interface defect states acting as traps for the charge carriers leading to indirect electron−hole transitions which are lower in energy. The carrier confinement is most likely related to Al G

dx.doi.org/10.1021/nl4012613 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

(17) Heiss, M.; Fontana, Y.; Gustafsson, A.; Wust, G.; Magen, C.; O’Regan, D. D.; Luo, J. W.; Ketterer, B.; Conesa-Boj, S.; Kuhlmann, A. V.; Houel, J.; Russo-Averchi, E.; Morante, J. R.; Cantoni, M.; Marzari, N.; Arbiol, J.; Zunger, A.; Warburton, R. J.; Fontcuberta I Morral, A. Nat. Mater. 2013, 12 (5), 439−444. (18) Rudolph, D.; Funk, S.; Doblinger, M.; Morkotter, S.; Hertenberger, S.; Schweickert, L.; Becker, J.; Matich, S.; Bichler, M.; Spirkoska, D.; Zardo, I.; Finley, J. J.; Abstreiter, G.; Koblmuller, G. Nano Lett. 2013, 13 (4), 1522−1527. (19) Heiss, M. Phys. Rev. B 2011, 83, 045303. (20) Hocevar, M.; Giang, L. T. T.; Songmuang, R.; Hertog, M. D.; Besombes, L.; Bleuse, J.; Niquet, Y.-M.; Pelekanos, N. T. Appl. Phys. Lett. 2013, 102, 191103. (21) Zhang, G. Q.; Tateno, K.; Sanad, H.; Tawara, T.; Gotoh, H.; Nakano, H. Appl. Phys. Lett. 2009, 95, 123104. (22) Joyce, H. J.; Gao, Q.; Tan, H. H.; Jagadish, C.; Kim, Y.; Fickenscher, M. A.; Perera, S.; Hoang, T. B.; Smith, L. M.; Jackson, H. E.; Yarrison-Rice, J. M.; Zhang, X.; Zou, J. Nano Lett. 2009, 9, 695. (23) Kang, J.-H.; Gao, Q.; Jouce, H. J.; Tan, H. H.; Jagadish, C.; Guo, Y. K. Y.; Xu, H.; Zou, J.; Fickenscher, M. A.; Smith, L. M.; Jackson, H. E.; Yarrison-rice, J. M. Cryst. Growth Des. 2011, 11, 3109. (24) van Weert, M. H. M.; Wunnicke, O.; Roest, A. L.; Eijkemans, T. J.; Silov, A. Y.; Haverkort, J. E. M.; Hooft, G. W. ’t.; Bakkers, E. P. A. M. Appl. Phys. Lett. 2006, 88, 043109. (25) Mattila, M.; Hakkrainen, T.; Lipsanen, H.; Jiang, H.; Kauppinen, E. I. Appl. Phys. Lett. 2007, 90, 033101. (26) Pierret, R. F. Semiconductor Device Fundamentals, 2nd ed.; Addison-Wesley: Reading, MA, 1996. (27) Kivisaari, P.; Oksanen, J.; Tulkki, J. J. Appl. Phys. 2012, 111 (10), 103120. (28) Levinshtein, M.; Rumyantsev, S.; Shur, M. Handbook Series on Semiconductor Parameters; World Scientific Publishing Company: Singapore, 1999; Vol. 2.

H

dx.doi.org/10.1021/nl4012613 | Nano Lett. XXXX, XXX, XXX−XXX