Intense Intrashell Luminescence of Eu-Doped Single ZnO Nanowires

Jun 27, 2014 - ... of semiconductor nanowires, Ph.D., Friedrich-Schiller-University, Jena, 2013; www.db-thueringen.de/servlets/DocumentServlet?id=2207...
1 downloads 0 Views 2MB Size
Letter pubs.acs.org/NanoLett

Intense Intrashell Luminescence of Eu-Doped Single ZnO Nanowires at Room Temperature by Implantation Created Eu−Oi Complexes Sebastian Geburt,† Michael Lorke,‡ Andreia L. da Rosa,‡ Thomas Frauenheim,‡ Robert Röder,† Tobias Voss,§ Uwe Kaiser,∥ Wolfram Heimbrodt,∥ and Carsten Ronning*,† †

Institute of Solid State Physics, Friedrich-Schiller-University Jena, Max-Wien-Platz 1, 07743 Jena, Germany Bremen Center for Computational Materials Science (BCCMS), University of Bremen, Am Fallturm 1, 28359 Bremen, Germany § Institute of Solid State Physics, University of Bremen, Otto-Hahn-Allee NW1, 28359 Bremen, Germany ∥ Institute of Physics and Material Science Center, Philipps-University Marburg, Renthof 5, 35032 Marburg, Germany ‡

ABSTRACT: Successful doping and excellent optical activation of Eu3+ ions in ZnO nanowires were achieved by ion implantation. We identified and assigned the origin of the intra-4f luminescence of Eu3+ ions in ZnO by first-principles calculations to Eu−Oi complexes, which are formed during the nonequilibrium ion implantation process and subsequent annealing at 700 °C in air. Our targeted defect engineering resulted in intense intrashell luminescence of single ZnO:Eu nanowires dominating the photoluminescence spectrum even at room temperature. The high intensity enabled us to study the luminescence of single ZnO nanowires in detail, their behavior as a function of excitation power, waveguiding properties, and the decay time of the transition. KEYWORDS: nanowires, ion beam doping, zinc oxide, rare earth elements, luminescence, density functional theory

S

Zinc oxide (ZnO) is a radiation hard and wide band gap semiconductor, and several studies on doping of ZnO nanowires with RE elements already exist, in which two different routes are basically performed: the majority attempted doping during growth, and a few used subsequent ion implantation. The first approach suffers on operating at thermal equilibrium resulting into two issues: (i) the low solubility limit of the impurities and the necessary high growth temperatures cause in many cases segregation, phase separation, or nonideal nanowire morphologies;10−15 and (ii) furthermore, if the RE elements are incorporated into the ZnO lattice, one expects an occupation of ideal lattice sites replacing Zn during growth and thus resulting into nonluminescent divalent charge states. However, unexpected optical activation of the impurities was achieved, and their emission properties as well as energytransfer processes were investigated, but their intensities were rather low with respect to the near-band edge (NBE) or defect level (DLE) emissions and their origin was not identified.10−15 As an alternative doping method, ion implantation offers precise control on both lateral and depth concentration profiles, but plenty of defects are incorporated into the material dominating all properties after the implantation process. The majority are created Frenkel pair point defectsvacancies and

emiconductor nanowires, especially ZnO nanowires, have been proposed as the next frontier in the miniaturization of light sources and solid-state lasers.1−4 Such devices can generate highly localized intense (monochromatic) light in a geometry ideally suited for efficient coupling into nanophotonic elements such as quantum dots, metallic nanoparticles, plasmonic waveguides, and even biological specimens. Thus, nanowire-based light sources could become critical components in the study and development of novel nanoscale photonic elements. The research community has consequently expended significant effort toward their realization;5,6 however, full device application in photonics requires effective and controlled doping in order to modify the optical properties of the semiconductor nanowires. Doping of materials with lanthanides enables new optical properties, such as these so-called rare earth (RE) elements exhibit a partially filled 4f shell, which does not contribute to the bonding. If trivalent RE elements are incorporated into suitable host matrices, optical intra-4f transitions become possible due to weakening of the spin- and/or parity-forbidden inner-shell transitions induced by the crystal field of the matrix.7 These transitions consequently show long lifetimes in the range of ms and are thus very sharp. Most prominent examples are Er-doped diodes and fibers emitting in the IRspectral range at exactly 1540 nm for telecommunication8 and Nd-doped yttrium aluminum garnet lasers (Nd:YAG lasers) emitting at a based wavelength of 1064 nm.9 © XXXX American Chemical Society

Received: April 27, 2014 Revised: June 13, 2014

A

dx.doi.org/10.1021/nl5015553 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 1. (a) Overview PL spectra of ZnO nanowires implanted with different Eu concentrations at room temperature and subsequently annealed at 700 °C for 30 min in air plotted in a logarithmic scale. (b) PL spectrum of an ZnO nanowire ensemble with an Eu concentration of 1.8 at. % with a respective assignment of the intra-4f luminescence lines (inset: energy levels of Eu3+). (c) High-resolution PL spectra of the respective intense intra4f shell transitions.

See for details ref 18. The PL of ensemble and single nanowires were investigated using a self-built PL system and epifluorescence microscope (details of the PL setup can be downloaded via ref 21) with excitation by a continuous wave 325 nm HeCd laser. The luminescence measurements were performed for temperatures ranging from 4 K to room temperature with varying laser intensities. Figure 1a shows the normalized overview PL spectra of ZnO nanowire ensemble samples taken at 4 K, which were implanted with different europium concentrations at room temperature and subsequently annealed at 700 °C for 30 min in air. The spectra are composed of the recombination of free and bound excitons in the ZnO NBE emission around 3.4 eV22 and a broad emission band centered around 2.4 eV associated with the recombination of carriers at intrinsic ZnO defects (DLE).23 The sharp and structured emission features between 1.5 and 2.1 eV can be clearly attributed to the Eu3+ intra-4f luminescence,4 which can be identified in Figure 1b as a set of relative sharp transitions from the excited 5D0 level to several 7FJ (J = 0−6) ground terms. Thus, successful doping and optical activation of the Eu3+ ions was achieved for the implanted ZnO nanowires. Whereas the signal of the incorporated Eu3+ ions is small for low concentrations (top spectrum in Figure 1a), it gradually increases with increasing concentrations with respect to the NBE and DLE emissions. Finally, it even dominates the spectrum for the highest concentration of 4.3 at. % (bottom spectrum), which was not the case for studies in the past.7−12,15 However, the overall luminescence intensity is optimal for intermediate concentrations in the range of about one atomic percent (e.g., as for the sample shown in Figure 1b) due to resonant Förster transfer processes to nonradiative recombination centers (so-called “concentration quenching”)24 at higher concentrations. It is evident that the transition down to the 7F2 ground level is the most intense one for all concentrations, as shown in Figure 1c and in agreement with refs 10, 12, 13, and 18. Furthermore, the shape and relative intensities of the intra4f transitions slightly change for the different concentrations, which indicates that the Eu3+ ions sense slightly different local environments, which is due to remaining implantation defects next to the incorporated Eu atoms. One would basically expect that europium occupies substitutional lattice sites in the II−VI semiconductor zinc oxide resulting in a formal charge of 2+ in the ideal hexagonal crystal

interstitials of both zinc and oxygendue to the elastic knockoff processes in the implantation cascade, which usually have to be removed via adequate and subsequent annealing procedures.16 Here, the thermal annealing drives the nonequilibrium as-implanted state back to thermal equilibrium. In the past, ion implantation was also successfully used for the incorporation of, e.g., optically active Er and Eu in ZnO nanowires.17,18 The incorporated RE atoms also replaced Zn in the ZnO lattice, and room-temperature luminescence of the intra-4f-transition was observed after annealing to high temperatures; however, the intensity was also as low as for the case of doping during growth. In this study, we identified and assigned the origin of the intra-4f luminescence of Eu3+ ions in ZnO by first-principles calculations to Eu−Oi complexes and utilized the nonequilibrium ion implantation process together with a specific subsequent thermal budget for targeted defect engineering. The optimized implantation and annealing parameters resulted in intense intrashell luminescence of ZnO:Eu nanowires dominating the photoluminescence (PL) spectrum even at room temperature. The high intensity enabled us to study the luminescence of single ZnO nanowires in detail, their behavior as a function of excitation power, waveguiding properties, and the decay time of the transition. ZnO nanowires with diameters of typically about 100−300 nm and lengths of more than 10 μm were synthesized using the vapor−liquid−solid (VLS) mechanism in a horizontal tube furnace19 and subsequently dispersed on clean Si substrates. The ion implantation was simulated using the Monte Carlo package iradina,20 which takes into account the threedimensional nanowire morphology. A combination of several ion energies ranging from 20 to 380 keV yields into a homogeneous doping profile for a depth of about 100 nm. The ion implantation was performed at room temperatures using 153 Eu ions with total ion fluences ranging from of 7.8 × 1013 to 4.5 × 1016 cm−2, resulting in concentrations of 0.01 to 5 at. %, respectively. Thermal annealing was performed in air for 30 min. After implantation and annealing, no other phases, such as europium oxides, were detected by high-resolution transmission electron microscopy (TEM) and no amorphization occurred in any sample. Instead, we observed by energy dispersive X-ray spectroscopy (EDX) homogeneous doping of the single crystalline ZnO nanowires with the desired concentrations. B

dx.doi.org/10.1021/nl5015553 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

lattice; thus, being nonluminescent, as discussed already above. There is even clear experimental evidence in emission channeling studies that RE elements indeed occupy substitutional Zn sites in ZnO, even after ion implantation.25 On the other hand, in our studies and also for the incorporation of Eu during ZnO nanowire or thin film growth,10−15 the specimens clearly show characteristic intrashell transitions of Eu3+. Thus, the local environment of at least a fraction of incorporated Eu atoms must be significantly disturbed. We employed a combination of density functional theory (DFT)26,27 and many-body methods in order to investigate the incorporation of Eu atoms in ZnO. We used DFT in the framework of the projected augmented wave method (PAW)28 as implemented in the Vienna ab initio VASP package.29 The generalized gradient approximation (GGA) approximation was used with the Perdew−Burke−Ernzerhof (PBE) functional to relax the structure. Our tests show that the relaxation with the hybrid functional PBE030 leads to similar results. A 3 × 3 × 2 super cell containing 72 atoms with a single Eu impurity was employed to model the Eu-doped ZnO system. This results in an Eu concentrationof 3%, which is within the range of the experimental concentration. A (2 × 2 × 1) k-point sampling was used for charge density integration. We set an energy cutoff of 500 eV in order to ensure convergence, and the criteria on force convergence was 0.05 eV/Å. The Eu 5s, 5p, 4f, and 6s electrons were treated as a valence shell as well as the Zn 3d electrons. After relaxation, we determined the electronic band structure by many-body calculations using the GW method31−33 in the GW0 approximation. We employed partially selfconsistent GW0 calculations, where only the eigenvalues for the Green’s function G were iterated, and the wave functions were kept at the GGA level. We set a cutoff of 200 eV for the calculation of the response functions and convergence was typically reached after 4−5 iterations. The GW method has the advantage that screening is calculated microscopically, which can be important for systems where screening is inhomogeneous. We have investigated several configurations including Eu at interstitial and substitutional sites as well as europium in the presence of intrinsic defects, as ion implantation produces such defects, as mentioned above. The most common defects in ZnO, not only after ion implantation, are oxygen and zinc vacancies and oxygen interstitials.34−36 While oxygen vacancies in ZnO lead to an occupied state about 1 eV above the valence band maximum, both oxygen interstitials and zinc vacancies lead to empty states in the band gap and can thus potentially oxidize Eu from a 2+ to a 3+ state. Among all configurations investigated, the following were found to be possible candidates to explain the intrashell luminescence in Eu-doped ZnO: (i) substitutional Eu at a zinc lattice position (EuZn) without the presence of any intrinsic defects, (ii) substitutional Eu near a neutral zinc vacancy (EuZn + VZn), and (iii) substitutional Eu near an oxygen interstitial (EuZn + Oi). Europium occupying substitutional sites does not produce strong distortions in the ZnO lattice, as the calculated Eu−O bond lengths are 2.11 Å. The projected density-of-states (DOS) for EuZn, as displayed in Figure 2a, shows that the Eu fstates (red) lie within the band gap of ZnO. While the conduction band (CB) onset is barely visible in the DOS, the GW0 calculation results in a ZnO band gap of 3.7 eV, which is in reasonable agreement with the experimental value. However, in this configuration europium assumes the divalent oxidation state 2+, as expected. Therefore, it cannot explain the observed luminescence, since the f states are fully occupied. In contrast,

Figure 2. Electronic DOS for ZnO doped with Eu in different configurations, as calculated by DFT. The red line represents the projected DOS for (a) EuZn, (b) EuZn + VZn, and (c) EuZn + Oi; whereas, the black line gives the total DOS of the respective system. The dashed line denotes the Fermi level.

both defect complexes EuZn + Oi and EuZn + VZn lead to a formal charge of Eu3+, so that the Eu f-orbitals contain 6 electrons. However, for the EuZn + VZn defect complex, shown in Figure 2b, the occupied f-states lie inside the valence band (VB). This cannot be reconciled with the experimental 7FJ state, as the occupied states in the valence band would not be optically active. Moreover, the unoccupied f states are located at ∼1 eV, having a large energetic distance to the CB, which would hinder carrier transfer. In contrast, for the EuZn + Oi complex, shown in Figure 2c, both occupied and unoccupied Eu f-states are clearly located within the band gap at around −0.5 and 1.5 eV, respectively. This is consistent with the 7FJ ground-state configuration, as it is observed in the experiment and as electrons and holes can be trapped in the Eu f shells from the VB and CB states, respectively. Therefore, we are convinced that the intra-4f luminescence of the ZnO:Eu nanowire samples is due to EuZn + Oi complexes formed during ion implantation. Figure 1c shows that the fine structure of the intrashell luminescence slightly varies as a function of the implantation concentrations, as it can be seen by following the dashed line. We attribute this to second next-neighbor’s defects; thus, higher order defect complexes with intrinsic or extrinsic defects next to the EuZn + Oi configuration resulting into slight distortions of the emitting Eu atom. This is in agreement with the fact that the PL-spectra lose emission features with increasing Eu implantation concentration, as evident in Figure 1c. One Eu-doped ZnO nanowire sample, which was implanted to the optimum concentration of about 1 at. %, was split into three pieces, and each identical piece was annealed subsequent to the implantation in air at temperatures of 500, 700, and 900 °C. No luminescence at all was detected in the as-implanted situation without any annealing. Figure 3a shows the respective PL spectra taken at 4 K, and it is evident that annealing to 500 °C yields to a spectrum with very sharp features, which are also displayed in the high-resolution PL spectra in Figure 3b−d. The overall PL intensity is still low, and the DLE dominates the spectrum (not shown). Thus, this low-temperature annealing yields obviously to the onset of the diffusion of one specific defect species, which is then trapped at the Eu atoms, resulting into the optical activation of Eu within a specific surrounding. C

dx.doi.org/10.1021/nl5015553 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 3. (a) PL spectra of ZnO nanowires implanted with 1.0 at. % Eu ions at room temperature and subsequently annealed in air as a function of annealing temperature. (b−d) High-resolution PL spectra of the intra-4f luminescence of the Eu-doped ZnO nanowires, which were annealed at 500 °C. The DLE background of ZnO host matrix was subtracted.

Figure 4. (a) μ-PL spectra taken at room temperature of three different ZnO nanowires, which were implanted with an concentration of 1.8 at. % and subsequently annealed at 700 °C in air for 30 min. The inset shows an optical image of the about 10 μm long wire and the laser excitation spot (arrow), which provided the top spectrum. (b) μ-PL spectra of one single nanowire as a function of excitation power at room temperature. The inset displays the dependencies of the intra-4f, DLE, and NBE emissions. (c) μ-PL spectra taken at 4 K and at different spots of one single nanowire, while the nanowire was excited at one facet end only.

state. However, the latter are nonradiative; thus, only a lowintensity intrashell luminescence was visible with respect to DLE and NBE (about 1 order of magnitude lower), which is also evident in the bad signal-to-noise ratio of the top spectrum in Figure 3a. Microphotoluminescence (μ-PL) spectra of three individual ZnO nanowires, which were implanted at room temperature with an Eu concentration of 1.8 at. % and subsequently annealed at 700 °C in air for 30 min, are shown in Figure 4a. In all three cases, the strong intra-4f luminescence is clearly visible even at room temperature, but its intensity varies with respect to the NBE emission, which is due to different diameters of the individual nanowires and the limited ion range of the energetic Eu ions of only 100 nm during ion implantation. The larger diameter nanowires simply obtain still unimplanted volume contributing only to the NBE emission. The emission of one single Eu-doped ZnO nanowire was investigated as a function of excitation power. The μPL spectra are dominated by the

Considering the above DFT calculations, we assume that this is due to the diffusion and trapping of interstitial oxygen, as the substitutional diffusion of Eu is very unlikely. Thus, a slight recovery of the implantation damage occurs, and all optical active Eu3+ ions experience identical surroundings. A rigorous assignment by determining the local symmetry and comparing it to the respective Stark-levels should be possible37 for this situation, but is out of scope of this study. Increasing the annealing temperature to 700 °C broadens the luminescence lines, as more defects become mobile in the ZnO lattice and form EuZn−Oi complexes with different trapped second neighbor defects. The recovery of the lattice continues, and the overall luminescence intensity is optimal for this situation, as already shown above. A further increase of the annealing temperature to 900 °C results into a further recovery of the ZnO lattice and the release of interstitial oxygen out of EuZn− Oi complexes resulting in an increasing fraction of substitutional Eu atoms with a perfect surrounding in a divalent charge D

dx.doi.org/10.1021/nl5015553 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

Figure 5. (a) TRPL spectra of an ensemble of Eu-doped ZnO nanowires with a concentration of 1.0 at. % show a slight spectral change at different times after excitation. The respective gate widths are given in parentheses. (b) The 5D0 → 7F2 transients of several types of different sample geometries of ZnO:Eu nanowires with a concentration of 1.0 at. % show all the same multiple-exponential decay with slow lifetimes in the range of ms.

measurement, which could be attributed to confined Förster transfers in a one-dimensional structure as seen in ref 41. Summarizing, we achieved successful doping and excellent optical activation of Eu3+ ions in ZnO nanowires by ion implantation of Eu in the 1% range and subsequent annealing at 700 °C in air. These optimized implantation and annealing parameters resulted into intense intrashell luminescence of single ZnO:Eu nanowires, which dominate the PL spectrum even at room temperature, decay in the ms range, and which are caused by the formation of Eu−Oi complexes. Our obtained defect engineering process can be adapted for doping of ZnO nanowires with all other RE elements enabling an easy choice of the emission color for light-emitting diodes.

intra-4f emission at low excitation densities, as shown in Figure 4b. At excitation powers of ≈300 W/cm2, the ZnO NBE emission is similar in intensity to the Eu3+ emission and further increases in a clear linear behavior as a function of excitation power, as shown in the inset of Figure 4b. On the other hand, both the intra-4f luminescence and the DLE band show a saturation behavior for high power excitations due to their limited amount in the sample. One can further conclude that no light amplification occurred for this particular nanowire, such as in a lasing system, which would result into a superlinear increase of the emission intensities.38 This might be because the diameter is too small in order to act as an efficient waveguide for the 620 nm intra-4f luminescence,4 where the index of refraction is only 2 compared to 2.4 at the NBE region. The experiment displayed in Figure 4c confirms such a conclusion. The figure shows three spectra taken from the nanowire at different positions, while the nanowire was excited at one facet end only. Whereas at the excitation spot as well as at the opposite end, the short wavelength NBE emission is emitted and thus efficiently waveguided through the whole ZnO:Eu nanowire, the intra-4f-luminescence is isotropically emitted and reaches the microscope also from the center body, as shown in the middle spectrum in Figure 4c. Finally, time-resolved photoluminescence (TRPL) measurements were performed at 5 K using a 355 nm excitation by a frequency-tripled Nd:YAG laser (pulse fwhm 3 ns) with an excitation density of 15 μJ/cm2. The detection was performed by a MS257 grating monochromator with an iStar intensified CCD (iCCD) camera. The gate width of the iCCD was varied between 1 μs for short times after excitation up to 1 ms for later times.39 Figure 5 presents the temporal decay of the 5D0 → 7F2 emission in the ZnO nanowire matrix for samples, which were implanted to a concentration of 1 at. %. The emission line shape was detected between 15 and 10500 μs after the excitation pulse and shows slight changes on the low-energy wing (see Figure 5a), which indicates a slightly faster dynamics for those higher Stark-levels. The integrated PL transients were determined for three types of samples with the same Eu concentration and are displayed in Figure 5b. The decay rates are within the typical range of ms for Eu3+ emissions in ZnO40 and follow a multiexponential decay due to the slightly different defect complexes. Furthermore, and obviously, there are no differences between thick (d ≈ 100−300 nm) and thin (d ≈ 40−60 nm) ZnO:Eu nanowires in a diluted or ensemble



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors gratefully acknowledge funding by the German Research Society (DFG) within the projects FOR1616 and GRK1782 as well as computational resources from HLRN (Hannover/Berlin). A.L.R. thanks H. Chacham for fruitful discussions.



REFERENCES

(1) Gevaux, D. Nat. Photonics 2008, 2, 594. (2) Wang, Z. L. Appl. Phys. A: Mater. Sci. Process. 2007, 88, 7. (3) Rühle, S.; et al. Nano Lett. 2008, 8, 119. (4) Voss, T.; et al. Nano Lett. 2007, 7, 3675. (5) Zhang, C.; et al. Opt. Express 2009, 17, 7893. (6) Qian, F.; et al. Nat. Mater. 2008, 7, 701. (7) Diecke, G. H. Spectra and Energy Levels of Rare Earth Ions in Crystals; Wiley-Interscience: New York, 1968. (8) Polman, A. J. Appl. Phys. 1997, 82, 1. (9) Geusic, J. E.; Marcos, H. M.; Van Uitert, L. G. Appl. Phys. Lett. 1964, 4, 182. (10) Lupan, O.; et al. Appl. Surf. Sci. 2013, 282, 782. (11) Rifai, S. A. Al; et al. Russ. J. Phys. Chem. A 2014, 88, 108. (12) Pal, P. P.; Manam, J. J. Lumin. 2014, 145, 340; Radiat. Phys. Chem. 2013, 88, 7. (13) Wang, D. D.; et al. J. Phys. Chem. C 2011, 115, 22729. (14) Luo, L.; et al. Opt. Mater. 2010, 32, 1066. (15) Yang, Y. H.; et al. J. Appl. Phys. 2010, 107, 053502.

E

dx.doi.org/10.1021/nl5015553 | Nano Lett. XXXX, XXX, XXX−XXX

Nano Letters

Letter

(16) Wendler, E.; et al. Nucl. Instrum. Methods Phys. Res., Sect. B 2009, 267, 2708. (17) Wang, J.; et al. Appl. Phys. Lett. 2006, 89, 221917. (18) Geburt, S.; et al. J. Nanosci. Nanotechnol. 2008, 8, 244. (19) Borchers, C.; et al. J. Phys. Chem. B 2006, 110, 1656. (20) Borschel, C.; Ronning, C. Nucl. Instrum. Methods Phys. Res., Sect. B 2011, 269, 2133. (21) Geburt, S. Lasing and ion beam doping of semiconductor nanowires, Ph.D., Friedrich-Schiller-University, Jena, 2013; www.dbthueringen.de/servlets/DocumentServlet?id=22078. (22) Meyer, B. K.; et al. Phys. Status Solidi B 2004, 241, 231. (23) Müller, S.; et al. J. Appl. Phys. 2008, 104, 123504. (24) Förster, T. Ann. Phys. 1948, 437, 55. (25) Wahl, U.; et al. Appl. Phys. Lett. 2003, 82, 1173. (26) Hohenberg, P.; Kohn, W. Phys. Rev. 1964, 136, B864. (27) Kohn, W.; Sham, L. J. Phys. Rev. 1965, 140, A1133. (28) Blöchl, P. E. Phys. Rev. B 1994, 50, 17953. (29) Kresse, G.; Joubert, D. Phys. Rev. B 1999, 59, 1758. (30) Paier, J.; et al. J. Chem. Phys. 2005, 122, 234102. (31) Onida, G.; et al. Rev. Mod. Phys. 2002, 74, 601. (32) van Schilfgaarde, K.; et al. Phys. Rev. Lett. 2006, 96, 226402. (33) Shishkin, M.; et al. Phys. Rev. Lett. 2007, 99, 246403. (34) Janotti, A.; et al. Phys. Rev. B 2007, 76, 165202. (35) Lany, S.; et al. Phys. Rev. B 2008, 78, 2351004. (36) Janotti, A.; et al. Rep. Prog. Phys. 2009, 72, 126501. (37) Gruber, J. B.; et al. J. Appl. Phys. 2011, 110, 023104. (38) Zimmler, M. A.; et al. Semicond. Sci. Technol. 2010, 25, 024001. (39) Kaiser, U.; et al. J. Appl. Phys. 2013, 113, 073506. (40) Armelao, L.; et al. J. Phys. Chem. C 2008, 112, 4049. (41) Chen, L.; et al. Phys. Rev. B 2007, 76, 115325.

F

dx.doi.org/10.1021/nl5015553 | Nano Lett. XXXX, XXX, XXX−XXX