Strain-compensated (Ga,In)N/(Al,Ga)N/GaN multiple quantum wells for improved yellow/amber light emission K. Lekhal, B. Damilano, H. T. Ngo, D. Rosales, P. De Mierry, S. Hussain, P. Vennéguès, and B. Gil Citation: Applied Physics Letters 106, 142101 (2015); doi: 10.1063/1.4917222 View online: http://dx.doi.org/10.1063/1.4917222 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/106/14?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Surface potential effect on excitons in AlGaN/GaN quantum well structures Appl. Phys. Lett. 102, 082110 (2013); 10.1063/1.4793568 Investigation of the light emission properties and carrier dynamics in dual-wavelength InGaN/GaN multiplequantum well light emitting diodes J. Appl. Phys. 112, 083101 (2012); 10.1063/1.4759373 Effect of n-GaN thickness on internal quantum efficiency in InxGa1-xN multiple-quantum-well light emitting diodes grown on Si (111) substrate J. Appl. Phys. 109, 113537 (2011); 10.1063/1.3596592 Development of green, yellow, and amber light emitting diodes using InGaN multiple quantum well structures Appl. Phys. Lett. 90, 151109 (2007); 10.1063/1.2721133 Improved characteristics of InGaN multiple-quantum-well light-emitting diode by GaN/AlGaN distributed Bragg reflector grown on sapphire Appl. Phys. Lett. 76, 1804 (2000); 10.1063/1.126171

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 180.211.215.249 On: Wed, 20 May 2015 01:45:06

APPLIED PHYSICS LETTERS 106, 142101 (2015)

Strain-compensated (Ga,In)N/(Al,Ga)N/GaN multiple quantum wells for improved yellow/amber light emission K. Lekhal,1 B. Damilano,1,a) H. T. Ngo,2 D. Rosales,2 P. De Mierry,1 S. Hussain,1,3 gue`s,1 and B. Gil2 P. Venne

1 CRHEA-CNRS, Centre de Recherche sur l’H et ero-Epitaxie et ses Applications, Centre National de la Recherche Scientifique, Valbonne 06560, France 2 Laboratoire Charles Coulomb, CNRS-INP-UMR 5221, Universit e Montpellier 2, F-34095 Montpellier, France 3 Universit e de Nice Sophia Antipolis, Parc Valrose, 28 av. Valrose, 06108 Nice cedex 2, France

(Received 19 January 2015; accepted 28 March 2015; published online 7 April 2015) Yellow/amber (570–600 nm) emitting InxGa1xN/AlyGa1yN/GaN multiple quantum wells (QWs) have been grown by metal organic chemical vapor deposition on GaN-on- sapphire templates. When the (Al,Ga)N thickness of the barrier increases, the room temperature photoluminescence is red-shifted while its yield increases. This is attributed to an increase of the QW internal electric field and an improvement of the material quality due to the compensation of the compressive strain C 2015 AIP Publishing LLC. of the InxGa1xN QWs by the AlyGa1yN layers, respectively. V [http://dx.doi.org/10.1063/1.4917222] Nitride-based light emitting diodes (LEDs) have demonstrated their ability to achieve very high external quantum efficiency in the blue wavelength range of the visible spectrum.1–3 When these blue LEDs are combined with specific luminophors,1 white lamps with un-precedent performances can be obtained. Record luminous efficiencies over 200 lm/W and even 300 lm/W have been recently announced.3,4 One reason for the success of these nitride-based LEDs is the large internal quantum efficiency of the active zone constituted by thin InxGa1xN/GaN quantum wells (QWs). For longer wavelengths, corresponding to green and yellow colors, for example, the quantum efficiency of InxGa1xN/GaN QWs strongly decreases.2 Several factors are at the origin of the degradation of the optical properties. All these factors are related to the increase of the In composition required to reach longer wavelengths. First, the InxGa1xN material quality tends to degrade when the In composition increases.5 This is related to the fact that the growth temperature has generally to be decreased to favor a large In incorporation in GaN and the use of low growth temperatures can lead to an increase of extended defects or point defects density. Second, increasing the In composition induces an increase of the stress in the structures. If this stress is too high, there is a risk of defect formation such as V-defects or misfit dislocations.6 Third, the reduction of the oscillator strength of the QW fundamental transition due to the internal piezo-electric field (quantum confined Stark effect) becomes more pronounced when the In composition increases (for QWs grown along the c-plane). To overcome the latter problem, some groups have developed the growth of GaN along semipolar and non-polar crystallographic orientations. Such an approach was used to demonstrate green lasers and yellow emitting LEDs.7,8 However, the In composition in InxGa1xN QWs grown along a semi-polar orientation has to be larger than that for the polar orientation to get the same emission wavelength.9 a)

Author to whom correspondence should be addressed. Electronic mail: [email protected]

0003-6951/2015/106(14)/142101/5/$30.00

One possible solution to overcome some of the problems listed above and therefore to improve the efficiency of yellow emitting nitride-based emitters would be to use (Ga,In)N/(Al,Ga)N/GaN multiple (M)QWs. As AlyGa1yN layers are tensely strained on GaN, they can be used to compensate the compressive strain of the InxGa1xN QWs. In addition, replacing part of the GaN barrier layer by a ternary alloy can increase the magnitude of the in-plane potential fluctuations and hence strengthen carrier localization.10 Very few works have been dedicated to the study of straincompensated InxGa1xN/AlyGa1yN QWs. Strain compensated active zones have been extensively used for heterostructures formed with other semiconductor families such as (Ga,In)As/(Ga,In)(As,P) QWs,11 (Ga,In)As/AlAs superlattices,12 InAs/GaSb or InAs/Al(As,Sb) superlattices,13,14 and InAs/Ga(As,N) quantum dots.15 Coming back to nitride materials, the benefit of using such straincompensated structures to increase gain in lasers emitting at 420–500 nm has been theoretically pointed out.16,17 We have previously shown that for samples grown by molecular beam epitaxy using thick AlyGa1yN layers (>20 nm) in the barrier can help to increase the wavelength emission of InxGa1xN QWs.18 Nagoya university has reported the use of strain-compensated InxGa1xN/AlyGa1yN MQWs grown by metal-organic chemical vapor phase epitaxy (MOCVD) to get room temperature cathodoluminescence up to a wavelength of 525 nm with an enhancement of the structural quality.19 It was recently shown that using a thin AlyGa1yN layer helps to reduce the photoluminescence (PL) intensity decrease of a InxGa1xN single QW when its emission wavelength increases.20 The most significant result is probably the demonstration of green to red LEDs grown by MOCVD with high external quantum efficiency for this wavelength range.21–23 The MQW structure in these works is constituted by a InxGa1xN layer capped at low temperature by a thin 1–1.5 nm AlyGa1yN layer (the maximum value of y is 0.9), while the GaN barrier layers are grown at higher temperature.

106, 142101-1

C 2015 AIP Publishing LLC V

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 180.211.215.249 On: Wed, 20 May 2015 01:45:06

142101-2

Lekhal et al.

We propose in the present work to study yellow/amber emitting (Ga,In)N/(Al,Ga)N/GaN MQWs with variable (Al,Ga)N thicknesses in order to clarify the impact of (Al,Ga)N layers on the MQW strain state and on their optical properties. Note that such structures can be used in devices as blue-to-yellow/amber light converters for the realization of monolithic phosphor-free white LEDs.24,25 Five different samples (A, A1, A2, A3, and B) were grown in a 2-in. vertical home-designed MOCVD reactor on commercially available n-type GaN-on-sapphire templates. Trimethylgallium or triethylgallium, trimethylindium, silane, and ammonia are used as precursors for gallium, indium, silicon, and nitrogen, respectively. The growth started by 800 nm of n-type GaN grown at 1080  C under H2 carrier gas. The H2 carrier gas was then replaced by N2 to grow the MQW (ten periods), whose structure is schematized in Figure 1. The growth procedure of the MQW is the following. First, the temperature is decreased to 715  C for the growth of a 2.6 nm-thick In0.21Ga0.79 N QW followed by a thin GaN cap layer (3 nm-thick) at the same temperature to avoid In desorption during temperature ramps. Then, the temperature is raised to 940  C for the barrier growth constituted by two layers of Al0.2Ga0.8N and GaN. The thickness of the Al0.2Ga0.8N is increased from 0 in sample A to 1.4 nm, 5.2 nm, and 10.6 nm for samples A1, A2, and A3, respectively. For samples A1, A2, and A3 the GaN thickness is reduced to keep a total barrier thickness of 17 nm. Sample B is identical to sample A (i.e., without (Al,Ga)N barrier layer), except that the (Ga,In)N growth temperature is decreased to 700  C instead of 715  C. As a result, the In composition is larger in this sample. The structural parameters of these different samples are summarized in Table I. High resolution X-ray diffraction (HRXRD) was used for the structural characterization. Thick (Al,Ga)N layers were grown with the same conditions than those used for the MQWs to calibrate the Al composition. HRXRD and transmission electron microscopy experiments were performed on separate (Ga,In)N/GaN MQWs to determine the In composition and the (Ga,In)N growth rate. According to these calibrations, we estimated the In composition at 0.21 (samples A, A1, A2, and A3) and 0.22 (sample B) for a growth temperature of 715  C and 700  C, respectively. By fitting the

FIG. 1. Structure of the multiple quantum wells.

Appl. Phys. Lett. 106, 142101 (2015) TABLE I. Structural data of the (Ga,In)N/(Al,Ga)N/GaN multiple quantum wells. T(Ga,In)N is the (Ga,In)N growth temperature and xIn is the corresponding In composition. L(Al,Ga)N and LGaN are the thicknesses of the (Al,Ga)N barrier layer, and the sum of the low temperature cap and GaN barrier, respectively. Sample A B A1 A2 A3

T(Ga,In)N ( C)

L(Al,Ga)N (nm)

LGaN (nm)

715 700 715 715 715

… … 1.4 5.2 10.6

15.2 15.5 15.6 11.9 7.0

2h/x X-ray diffraction spectra of the samples, we deduced the layer thicknesses reported in Table I. The continuouswave PL properties of the samples were measured at room temperature using the 244 nm line of a Ar frequency-doubled laser with an excitation power of 30 mW. Time resolved PL was performed using 266 nm laser radiation at a repetition rate of 82 MHz. The PL spectra were recorded using a 30 cm focal length spectrometer and a 150 groove/mm grating, detected with a Hamamatsu Streak camera with time resolution in the 10 ps range. Figure 2 shows the 2h/x X-ray diffraction spectra of samples A, A1, A2, and A3. The black line in Figure 2 identifies the position of the (0002) diffraction peak of the GaN buffer layer. The dotted line corresponds to the diffraction peak of the mean c-lattice parameter of the MQW of sample A. As the (Ga,In)N QWs are in compressive stress onto the GaN buffer layer in the growth plane, the out of plane lattice parameter (c) is increased. This explains the position of the MQW (0002) diffraction peak at lower angle than the GaN diffraction peak. When (Al,Ga)N is added in the MQW structure (samples A1, A2, and A3), there is a shift of the MQW diffraction peak and of the satellite peaks towards higher angles, showing that the compressive stress due to the (Ga,In)N QWs is compensated by the tensile stress of the (Al,Ga)N layers. The equilibrium lattice parameter, aeq, of the MQW can be obtained by determining the elastic strain energy minimum

FIG. 2. 2h/w X-ray diffraction spectra of samples A, A1, A2, and A3. The black line indicates the position of the diffraction peak corresponding to the (0002) GaN planes. The dotted line corresponds to the position of the mean c- lattice parameter of the multiple quantum well of sample A.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 180.211.215.249 On: Wed, 20 May 2015 01:45:06

142101-3

Lekhal et al.

Appl. Phys. Lett. 106, 142101 (2015)

of the system constituted by the three layers: (Ga,In)N, (Al,Ga)N, and GaN. To simplify the notations, we are going to use in the following: the correspondence 1 ¼ (Ga,In)N, 2 ¼ (Al,Ga)N, and 3 ¼ GaN. Under biaxial strain conditions, the elastic strain energy per surface unit can be written as Eel ðaeq Þ ¼

3 X

Mi Li D2i ;

(1)

i¼1

where Mi is the biaxial modulus, Li is the layer thickness, Di ¼ (aeq  ai)/ai is the strain in the growth plane, and ai is the relaxed lattice parameter of each i layer. By writing that Eel is minimum for dEel/daeq ¼ 0, we find the expression giving the value of the equilibrium lattice parameter 3  X

aeq ¼

Mi L i

j6¼i

i¼1 3  X i¼1

Y  aj

Mi Li

Y

a2j

3 Y

 i¼1

ai :

(2)

the integrated PL intensity. Again, this is the expected variation since the material quality of (Ga,In)N tends to degrade when the In composition increases or when the growth temperature is low, as explained in the introduction. Adding (Al,Ga)N in the barrier layer leads to less obvious consequences. The PL peak wavelength of samples A1, A2, and A3 are red-shifted to 574 nm, 590 nm, and 598 nm, respectively. The very interesting point is that this wavelength shift is also correlated with an increase of the PL intensity. How these observations can be understood? The bandgap of (Al,Ga)N is larger than that of GaN; therefore, if we consider the effect of quantum confinement alone, adding (Al,Ga)N in the barrier should result in a blueshift of the optical transitions. Actually, an increase of the electric field inside the (Ga,In)N QW could be at the origin of this red-shift. A simple expression of the internal electric field, Fj, of a layer, j, in a superlattice was given by Bernardini and Fiorentini26 X

j6¼i

The lattice parameters, stiffness constants of (Al,Ga)N and (Ga,In)N, were calculated using the Vegard’s law. By using the samples parameters given in Table I, we find that aeq is equal to 3.199, 3.197, 3.194, and 3.189 for samples A, A1, A2, and A3, respectively. These lattice parameters have to ˚. be compared to the relaxed value of GaN which is 3.189 A This explains the variation observed on the 2h/x X-ray diffraction spectra of these samples. Note that the relative thicknesses between (Al,Ga)N and GaN in sample A3 allow getting an equilibrium lattice parameter of the MQW equal to that of relaxed GaN. We now discuss the PL properties of these samples. Figure 3 shows the room temperature PL spectra of the different samples. The corresponding integrated PL intensities as a function of the peak wavelength are reported in Figure 6. The peak PL wavelength of sample A is 546 nm, while it is red-shifted to 573 nm for sample B. This is the expected variation since the In composition in the (Ga,In)N QWs increases from 0.21 to 0.22 for samples A and B, respectively. However, this red-shift is accompanied by a drop of

FIG. 3. Room temperature photoluminescence spectra (CW) of (Ga,In)N/ GaN (a) and (Ga,In)N/(Al,Ga)N/GaN (b) multiple quantum wells.

Fj ¼

Li ðPi  Pj Þ ei ; X Li ej i ei

i

(3)

where Pi is the polarization, Li is the layer thickness, and ei is the dielectric constant of each i layer. In our case, the theoretical values of the electric field calculated using (3) are 2.9 MV/cm, 3.1 MV/cm, 3.5 MV/cm, and 4 MV/cm for samples A, A1, A2, and A3, respectively (piezoelectric coefficient and spontaneous polarization values were taken from Ref. 27). Figure 4 shows the comparison of the band diagram of one period of In0.21Ga0.79N MQWs with or without a Al0.2Ga0.8N layer in the barrier. The internal electric fields in the different layers are calculated according to (3). In order to obtain an estimate of the energy variation due to the increase of the electric field, we can make the rough assumption that the electron and hole confinement energies are not affected by the change of the electric field. Then, the PL energy shift of samples Ai relative to sample A equals to eDF1L1, where e is the elemental charge and DF1 is the variation of the electric field compared to sample A. The corresponding values are 52 meV, 156 meV, and 260 meV

FIG. 4. Band diagram of one period of a In0.21Ga0.79N multiple quantum well with pure GaN or GaN/Al0.2Ga0.8N barrier layers.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 180.211.215.249 On: Wed, 20 May 2015 01:45:06

142101-4

Lekhal et al.

FIG. 5. Room temperature time-resolved photoluminescence intensity for samples A, A1, A2, and A3.

for samples A1, A2, and A3. By adding these values to the PL energy of sample A and converting them in wavelength, we obtain 559 nm, 586 nm, and 617 nm. The estimation of these wavelengths is in relatively good agreement with our experimental data. The observed PL wavelength increase when (Al,Ga)N is added in the barrier can therefore be due to an increase of the internal electric field in the (Ga,In)N QWs. The PL decay times were measured at room temperature (Figure 5). For all the samples, the PL decay is non monoexponential as already reported for (Ga,In)N QWs.28 We characterize hereafter the non-exponential decays by the delay, s10, for which the maximum PL signal is divided by 10. We find that s10 values are 1.9 ns, 5.4 ns, 8.4 ns, and 15.4 ns for samples A, A1, A2, and A3. We note that the decay time increases with the emission wavelength. This can be also interpreted as the consequence of the increase of the electric field in the (Ga,In)N QWs. The overlap of the

FIG. 6. Room temperature integrated photoluminescence intensity as a function of peak wavelength of the spectra from Figure 3. Circles correspond to (Ga,In)N/GaN multiple quantum wells with different In compositions (samples A and B). Squares correspond to a series of (Ga,In)N/(Al,Ga)N/GaN with different (Al,Ga)N thicknesses and a fixed In composition (samples A, A1, A2, and A3).

Appl. Phys. Lett. 106, 142101 (2015)

electron and hole wavefunctions is reduced when the electric field increases and therefore the PL radiative lifetime increases. However, it has to be noted that the PL decay time (sPL) is related to the radiative (sr) and non-radiative lifetimes (snr) by 1/sPL ¼ 1/sr þ 1/snr. Therefore, the increase of the decay time can also be due to an increase of the nonradiative decay time. As we find that the PL intensity also improves with the emission wavelength (Figure 6), this means that the non-radiative lifetime increases. This behavior is opposite to what is generally observed for long wavelength emitting (Ga,In)N MQWs.29 We attribute this to the improved quality of the MQWs due to the strain reduction caused by the introduction of the (Al,Ga)N layers in the barriers. In conclusion, we have shown that the high compressive strain in (Ga,In)N/GaN MQWs can be compensated by using tensily strained (Al,Ga)N layers in the barriers. Furthermore, adding (Al,Ga)N in the barrier layers leads to an increase of the internal electric field. The consequences on the optical properties are a redshift of the e1-hh1 fundamental transition of the (Ga,In)N QW and an improvement of the radiative efficiency at room temperature. Therefore, these properties are all benefic for the improvement of long-wavelength (yellow and red) emitters based on nitride materials. The authors would like to thank M. Leroux and J. Massies for the critical reading of the manuscript. We acknowledge support from GANEX (ANR-11-LABX-0014). GANEX belongs to the publicly funded “Investissements d’Avenir” Program managed by the French ANR agency. The authors would like to thank Erasmus Mundus Mobility with Asia (EMMA) program. This work was also partly funded by the French agency for Research ANR 2011 EMMA 004 01 Project “DELMONO.” Daniel Rosales acknowledges the Ph.D. Grant support of international ANR “GASIOPHE.” 1

S. Nakamura and G. Fasol, The Blue Laser Diode (Springer, Berlin, 1997). M. R. Krames, O. B. Shchekin, R. Mueller-Mach, G. O. Mueller, L. Zhou, G. Harbers, and M. G. Craford, J. Disp. Technol. 3, 160 (2007). 3 Y. Narukawa, M. Ichikawa, D. Sanga, M. Sano, and T. Mukai, J. Phys. D: Appl. Phys. 43, 354002 (2010). 4 See http://www.cree.com for the press release of 27 March 2014 regarding white light emitting diodes with a luminous efficiency of 300 lm/W. 5 T. Langer, A. Kruse, F. A. Ketzer, A. Schwiegel, L. Hoffmann, H. J€ onen, H. Bremers, U. Rossow, and A. Hangleiter, Phys. Status Solidi C 8, 2170 (2011). 6 M. Zhu, S. You, T. Detchprohm, T. Paskova, E. A. Preble, and C. Wetzel, Phys. Status Solidi A 207, 1305 (2010). 7 M. Adachi, Jpn. J. Appl. Phys. 53, 100207 (2014). 8 I. L. Koslow, M. T. Hardy, P. Shan Hsu, P.-Y. Dang, F. Wu, A. Romanov, Y.-R. Wu, E. C. Young, S. Nakamura, J. S. Speck, and S. P. DenBaars, Appl. Phys. Lett. 101, 121106 (2012). 9 U. Strauß, A. Avramescu, T. Lermer, D. Queren, A. Gomez-Iglesias, C. Eichler, J. M€ uller, G. Br€ uderl, and S. Lutgen, Phys. Status Solidi B 248, 652 (2011). 10 P. Lefebvre, T. Taliercio, A. Morel, J. Allegre, M. Gallart, B. Gil, H. Mathieu, B. Damilano, N. Grandjean, and J. Massies, Appl. Phys. Lett. 78, 1538 (2001). 11 B. I. Miller, U. Koren, M. G. Young, and M. D. Chien, Appl. Phys. Lett. 58, 1952 (1991). 12 M. P. Semtsiv, M. Ziegler, S. Dressler, W. T. Masselink, N. Georgiev, T. Dekorsy, and M. Helm, Appl. Phys. Lett. 85, 1478 (2004). 13 B. Satpati, J. B. Rodriguez, A. Trampert, E. Tournie, A. Joullie, and P. Christol, J. Cryst. Growth 301–302, 889 (2007). 2

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 180.211.215.249 On: Wed, 20 May 2015 01:45:06

142101-5 14

Lekhal et al.

X. Marcadet, C. Renard, M. Carras, M. Garcia, and J. Massies, Appl. Phys. Lett. 91, 161104 (2007). 15 R. Oshima, A. Takata, and Y. Okada, Appl. Phys. Lett. 93, 083111 (2008). 16 H. Zhao, R. A. Arif, Y.-K. Ee, and N. Tansu, Opt. Quantum Electron. 40, 301 (2008). 17 H. Zhao, R. A. Arif, Y.-K. Ee, and N. Tansu, IEEE J. Quantum Electron. 45, 66 (2009). 18 B. Damilano, T. Huault, J. Brault, D. Lefebvre, and J. Massies, Phys. Status Solidi C 7, 2130 (2010). 19 T. Doi, Y. Honda, M. Yamaguchi, and H. Amano, Jpn. J. Appl. Phys. 52, 08JB14 (2013). 20 D. D. Koleske, A. J. Fischer, B. N. Bryant, P. G. Kotula, and J. J. Wierer, J. Cryst. Growth 415, 57 (2015). 21 T. Shioda, H. Yoshida, K. Tachibana, N. Sugiyama, and S. Nunoue, Phys. Status Solidi A 209, 473 (2012).

Appl. Phys. Lett. 106, 142101 (2015) 22

S. Saito, R. Hashimoto, J. Hwang, and S. Nunoue, Appl. Phys. Express 6, 111004 (2013). J.-I. Hwang, R. Hashimoto, S. Saito, and S. Nunoue, Appl. Phys. Express 7, 071003 (2014). 24 B. Damilano, P. Demolon, J. Brault, T. Huault, F. Natali, and J. Massies, J. Appl. Phys. 108, 073115 (2010). 25 B. Damilano, H. Kim-Chauveau, E. Frayssinet, J. Brault, S. Hussain, K. Lekhal, P. Vennegue`s, P. De Mierry, and J. Massies, Appl. Phys. Express 6, 092105 (2013). 26 F. Bernardini and V. Fiorentini, Appl. Surf. Sci. 166, 23 (2000). 27 I. Vurgaftman and J. R. Meyer, J. Appl. Phys. 94, 3675 (2003). 28 P. Lefebvre, A. Morel, M. Gallart, T. Taliercio, J. Allegre, B. Gil, H. Mathieu, B. Damilano, N. Grandjean, and J. Massies, Appl. Phys. Lett. 78, 1252 (2001). 29 T. Langer, H. J€ onen, A. Kruse, H. Bremers, U. Rossow, and A. Hangleiter, Appl. Phys. Lett. 103, 022108 (2013). 23

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 180.211.215.249 On: Wed, 20 May 2015 01:45:06