28.12.2014 Views

Multiple scattering effects in strain and composition analysis ... - IEMN

Multiple scattering effects in strain and composition analysis ... - IEMN

Multiple scattering effects in strain and composition analysis ... - IEMN

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

APPLIED PHYSICS LETTERS 94, 013112 2009<br />

<strong>Multiple</strong> <strong>scatter<strong>in</strong>g</strong> <strong>effects</strong> <strong>in</strong> stra<strong>in</strong> <strong>and</strong> <strong>composition</strong> <strong>analysis</strong><br />

of nanoisl<strong>and</strong>s by graz<strong>in</strong>g <strong>in</strong>cidence x rays<br />

M.-I. Richard, 1,2 V. Favre-Nicol<strong>in</strong>, 1,3 G. Renaud, 1,a T. U. Schülli, 1 C. Priester, 4 Z. Zhong, 5<br />

<strong>and</strong> T.-H. Metzger 2<br />

1 Institut Nanoscience et Cryogénie/SP2M/NRS, CEA Grenoble, 17 Avenue des Martyrs,<br />

F-38054 Grenoble Cedex, France<br />

2 ID01/ESRF, 6 rue Jules Horowitz, BP220, F-38043 Grenoble Cedex, France<br />

3 Université Joseph Fourier, BP 53, 38041 Grenoble Cedex 9, France<br />

4 Département ISEN, Institut d’Electronique de Microélectronique et de Nanotechnologie, UMR CNRS 8520,<br />

Avenue Po<strong>in</strong>caré, 59652 Villeneuve d’Ascq Cedex, France<br />

5 Institut für Halbleiter-und Festkrperphysik, Johannes Kepler Universität L<strong>in</strong>z, L<strong>in</strong>z 4040, Austria<br />

Received 8 October 2008; accepted 12 December 2008; published onl<strong>in</strong>e 9 January 2009<br />

Experiments <strong>and</strong> numerical simulations based on f<strong>in</strong>ite element model<strong>in</strong>g show that the x-ray<br />

<strong>in</strong>tensity scattered by comparatively large nanostructures on a substrate is not simply related to their<br />

stra<strong>in</strong> <strong>in</strong> experiments us<strong>in</strong>g either graz<strong>in</strong>g <strong>in</strong>cidence or exit because of multiple <strong>scatter<strong>in</strong>g</strong> <strong>effects</strong>.<br />

However, whatever the nanostructure size, the <strong>composition</strong> profiles are correctly extracted from<br />

graz<strong>in</strong>g <strong>in</strong>cidence multiwavelength anomalous <strong>scatter<strong>in</strong>g</strong>. These <strong>effects</strong> are illustrated for the<br />

structural <strong>analysis</strong> of Ge dome-shaped isl<strong>and</strong>s grown on Si001. ©2009 American Institute of<br />

Physics. DOI: 10.1063/1.3064157<br />

a Electronic mail: gilles.renaud@cea.fr.<br />

The adjustment of the electronic <strong>and</strong> optical properties<br />

of nanoisl<strong>and</strong>s or nanowires requires a quantitative <strong>analysis</strong><br />

of their dimension, stra<strong>in</strong>, <strong>and</strong> <strong>composition</strong>, which is a subject<br />

of constant <strong>in</strong>terest. X-ray <strong>scatter<strong>in</strong>g</strong> is widely used for<br />

that sake. 1 Composition <strong>analysis</strong> can also be performed us<strong>in</strong>g<br />

multiwavelength anomalous diffraction MAD. 2 For large<br />

nanoisl<strong>and</strong>s with sufficiently large out-of-plane stra<strong>in</strong> profiles,<br />

the isostra<strong>in</strong> x-ray <strong>scatter<strong>in</strong>g</strong> method 3 can be comb<strong>in</strong>ed<br />

with anomalous diffraction to determ<strong>in</strong>e directly the out-ofplane<br />

stra<strong>in</strong> <strong>and</strong> <strong>composition</strong> profiles. However, often the<br />

scattered <strong>in</strong>tensity is assumed to directly reflect the stra<strong>in</strong><br />

state s<strong>in</strong>ce it is assumed to be its squared Fourier transform:<br />

the k<strong>in</strong>ematical or Born approximation is applied to extract<br />

the stra<strong>in</strong> state <strong>and</strong> <strong>composition</strong> of nano-objects. 4,5 The question<br />

arises to what extent this hypothesis is correct s<strong>in</strong>ce the<br />

x-ray beam has to be kept graz<strong>in</strong>g with respect to the surface<br />

to avoid background <strong>scatter<strong>in</strong>g</strong> from the substrate, result<strong>in</strong>g<br />

<strong>in</strong> multiple <strong>scatter<strong>in</strong>g</strong> <strong>effects</strong>. In this letter the multiple <strong>scatter<strong>in</strong>g</strong><br />

<strong>effects</strong> result<strong>in</strong>g from the graz<strong>in</strong>g <strong>in</strong>cidence geometry 6<br />

around <strong>in</strong>-plane Bragg reflections are <strong>in</strong>vestigated both experimentally<br />

<strong>and</strong> numerically, focus<strong>in</strong>g on isl<strong>and</strong>s grown us<strong>in</strong>g<br />

the Stranski–Krastanow growth mode. 7 They are illustrated<br />

for dome- <strong>and</strong> pyramid-shaped isl<strong>and</strong>s <strong>in</strong> the widely<br />

studied Ge/Si001 system.<br />

Deposition of 6 ML of Ge at 650 °C on Si001 by<br />

molecular beam epitaxy resulted <strong>in</strong> fairly monodispersed<br />

124 nm wide <strong>and</strong> 20 nm high domes, as found by atomic<br />

force microscopy. Graz<strong>in</strong>g <strong>in</strong>cidence x-ray diffraction measurements<br />

wavelength =1.0972 Å as illustrated <strong>in</strong> the <strong>in</strong>set<br />

of Fig. 1 were performed at the BM32 <strong>and</strong> BM02 beaml<strong>in</strong>es<br />

at the European Synchrotron Radiation Facility,<br />

Grenoble. Stra<strong>in</strong> sensitive radial scans <strong>in</strong> the vic<strong>in</strong>ity of the<br />

Si400 <strong>in</strong>-plane Bragg reflection were performed as a function<br />

of the Miller <strong>in</strong>dex h,0,0 h=2a/s<strong>in</strong>2/2, a be<strong>in</strong>g<br />

the Si lattice parameter <strong>and</strong> 2 the <strong>scatter<strong>in</strong>g</strong> angle, for different<br />

graz<strong>in</strong>g <strong>in</strong>cident angles i close to the critical angle<br />

for total external reflection, c =0.156°. The scattered <strong>in</strong>tensity<br />

along the exit angle f was <strong>in</strong>tegrated from 0° to 1.5°<br />

us<strong>in</strong>g a position sensitive detector.<br />

The <strong>in</strong>tensity profile as a function of h Fig. 1 is found<br />

to vary drastically with i . With <strong>in</strong>creas<strong>in</strong>g i , the position of<br />

the <strong>scatter<strong>in</strong>g</strong> peak shifts toward the Si Bragg peak. Above<br />

the critical angle, for i =0.2° <strong>and</strong> i =0.25°, a second peak<br />

hardly visible because of broaden<strong>in</strong>g due to the <strong>in</strong>homogeneous<br />

isl<strong>and</strong> size <strong>and</strong> the related <strong>in</strong>homogeneous stra<strong>in</strong><br />

distribution develops at lower h values h3.925 for i<br />

=0.2°, the orig<strong>in</strong> of which is given below. As all the curves<br />

cannot be superimposed by rescal<strong>in</strong>g, the <strong>in</strong>tensity profile<br />

cannot be directly related to the stra<strong>in</strong> state distribution <strong>in</strong> the<br />

nanoisl<strong>and</strong>s by a simple Fourier transform.<br />

FIG. 1. Color onl<strong>in</strong>e Experimental radial scans around the Si400 Bragg<br />

peak for Ge dome-shaped isl<strong>and</strong>s on Si001 for various values of i . Note<br />

the variation <strong>in</strong> the Ge peak position from h=3.95 to 3.965 as the <strong>in</strong>cident<br />

angle <strong>in</strong>creases. The <strong>in</strong>set displays the experimental setup. The <strong>in</strong>cident <strong>and</strong><br />

exit wave vectors k i <strong>and</strong> k f are kept graz<strong>in</strong>g angles i <strong>and</strong> f with respect<br />

to the surface. The <strong>scatter<strong>in</strong>g</strong> angle 2 is the <strong>in</strong>-plane angle between them.<br />

The one dimensional detector slit setup allows summ<strong>in</strong>g the <strong>scatter<strong>in</strong>g</strong> contributions<br />

along f .<br />

0003-6951/2009/941/013112/3/$23.00<br />

94, 013112-1<br />

© 2009 American Institute of Physics


013112-2 Richard et al. Appl. Phys. Lett. 94, 013112 2009<br />

FIG. 2. Color onl<strong>in</strong>e a Simulated radial scan <strong>in</strong>tensities around the<br />

Si400 reflection for Ge isl<strong>and</strong>s on Si001 whose stra<strong>in</strong> has been calculated<br />

by FDM: a 124 nm wide <strong>and</strong> 20 nm high dome. The <strong>in</strong>tensity is <strong>in</strong>tegrated<br />

between f 0° <strong>and</strong> f 1.5°. The total DWBA <strong>in</strong>tensity is reported for<br />

different <strong>in</strong>cident angles <strong>and</strong> compared to the BA first path <strong>in</strong>tensity. The<br />

substrate <strong>scatter<strong>in</strong>g</strong> has been omitted. b Evolution of t i ,z 2 <strong>in</strong>aGe<br />

dome on Si001 as a function of height z <strong>and</strong> i / c for =1.097 Å. As the<br />

<strong>in</strong>tensity scattered by each z-layer above the surface is proportional to<br />

t i ,z 2 , an x-ray <strong>scatter<strong>in</strong>g</strong> experiment will predom<strong>in</strong>antly reflect the<br />

stra<strong>in</strong> state of a given height depend<strong>in</strong>g on i rather than the full isl<strong>and</strong><br />

stra<strong>in</strong> state distribution. c Same as a but for a 70 nm wide <strong>and</strong> 7 nm high<br />

pyramid.<br />

To underst<strong>and</strong> this effect, four paths accord<strong>in</strong>g to the<br />

distorted wave Born approximation DWBA have to be<br />

taken <strong>in</strong>to account, 8,9<br />

F DWBA =<br />

f j e ik f −k i ·r j + r i f j e ik f +k i ·r j<br />

jQDs<br />

jQDs<br />

+ r f f j e i−k f −k i ·r j<br />

jQDs<br />

+ r i r f f j e i−k f +k i ·r j,<br />

jQDs<br />

where f j is the atomic <strong>scatter<strong>in</strong>g</strong> factor <strong>and</strong> r j is the position<br />

of an atom j <strong>in</strong> the isl<strong>and</strong>; k i <strong>and</strong> k f are the <strong>in</strong>com<strong>in</strong>g <strong>and</strong> exit<br />

wave vectors <strong>and</strong> r i,f are the reflectivities of the <strong>in</strong>cident<br />

<strong>and</strong> scattered waves. The first term <strong>in</strong> Eq. 1, which we call<br />

first path, corresponds to the k<strong>in</strong>ematical <strong>scatter<strong>in</strong>g</strong> of the<br />

<strong>in</strong>cident wave by the isl<strong>and</strong>s or Born approximation BA.<br />

The three other terms imply s<strong>in</strong>gle or double reflections by<br />

the substrate. 8<br />

To <strong>in</strong>terpret these observations, the stra<strong>in</strong> field <strong>in</strong> a coherent<br />

Ge/Si001 dome-shaped isl<strong>and</strong> with the experimental<br />

width <strong>and</strong> height was calculated with the f<strong>in</strong>ite difference<br />

method FDM, 10 assum<strong>in</strong>g cont<strong>in</strong>uum theory of elasticity.<br />

For simplicity, a pure Ge dome was modeled. The scattered<br />

<strong>in</strong>tensity see Fig. 2a was then calculated us<strong>in</strong>g Eq. 1<br />

around the <strong>in</strong>-plane Si400 reflection for different values<br />

of i around c <strong>and</strong> summed up between f 0° <strong>and</strong> f<br />

1.5° to account for the experimental conditions. The isl<strong>and</strong><br />

form factor given by the BA shows two broad peaks that<br />

arise from two prevalent stra<strong>in</strong> states <strong>in</strong> the dome-shaped<br />

isl<strong>and</strong>: the first peak close to the Si Bragg peak corresponds<br />

to the basis <strong>and</strong> also to the ma<strong>in</strong> part of the isl<strong>and</strong>, which is<br />

surrounded by the 113 <strong>and</strong> 15323 facets, while the second<br />

peak results from the stra<strong>in</strong> state of the top part of the<br />

isl<strong>and</strong>, term<strong>in</strong>ated by 105 facets. For i =0.62 c , only one<br />

1<br />

maximum is present <strong>in</strong> the simulated radial scan. It is positioned<br />

at h=3.944, i.e., displaced compared to the real<br />

maxima h=3.925 <strong>and</strong> 3.954 of the stra<strong>in</strong> state given by the<br />

first path. Above the critical angle, the positions of the <strong>in</strong>tensity<br />

maxima are close to the real positions of the maxima of<br />

the stra<strong>in</strong> state given by the first path.<br />

To discuss these <strong>effects</strong>, the isl<strong>and</strong>s can be decomposed<br />

<strong>in</strong>to layers of different heights z, each hav<strong>in</strong>g its own lattice<br />

parameter az. With this hypothesis, a radial position h corresponds<br />

to the <strong>scatter<strong>in</strong>g</strong> by a given layer at z. We also<br />

suppose case of Ge isl<strong>and</strong>s on Si that the higher z, the<br />

larger the lattice parameter <strong>and</strong> hence the smaller h. With<strong>in</strong><br />

the DWBA <strong>and</strong> as k z i,f = 2/ s<strong>in</strong> i,f , for small <strong>in</strong>cident<br />

<strong>and</strong> exit angles, the amplitude scattered at a given h by the<br />

portion of the isl<strong>and</strong> at height z can be written as 11<br />

F z DWBA i , f = F z BA h, i , f t i ,zt f ,z,<br />

where F z BA is the isl<strong>and</strong> form factor at height z <strong>and</strong> t,z<br />

=1+re −4iz/ is a generalized optical function. t i ,z<br />

varies as a function of i <strong>and</strong> z, as illustrated <strong>in</strong> Fig. 2b.<br />

Hence, the <strong>in</strong>tensity distribution <strong>in</strong> Fig. 2a results from the<br />

product of the isl<strong>and</strong> form factor by the generalized optical<br />

function. Because z is directly correlated with h, the optical<br />

function varies along the radial direction. If the value of i is<br />

below or at the critical angle of the substrate, the optical<br />

function can be very strongly enhanced for certa<strong>in</strong> heights z<br />

<strong>and</strong> thus for the correspond<strong>in</strong>g radial positions h. Low <strong>in</strong>cident<br />

angles maximize the generalized optical function for<br />

high z values <strong>and</strong> thus maximize the scattered <strong>in</strong>tensity at<br />

lower h values. For i =0.62 c , this maximum corresponds<br />

to z20 nm see Fig. 2b, i.e., the <strong>in</strong>tensity scattered by<br />

the isl<strong>and</strong>’s top, where it is the most relaxed, is strongly<br />

enhanced. Above the critical angle, the optical function does<br />

not display strong m<strong>in</strong>ima <strong>and</strong> maxima; the multiple <strong>scatter<strong>in</strong>g</strong><br />

<strong>effects</strong> can be neglected, <strong>and</strong> the k<strong>in</strong>ematical approximation<br />

is recovered. The condition for these DWBA <strong>effects</strong> to<br />

be significant is 11<br />

= rzH2/ahd /dz 1,<br />

where H is the total height of the nanocrystal, z is the<br />

parallel stra<strong>in</strong> of an isostra<strong>in</strong> area at height z, <strong>and</strong> rz is its<br />

radius. Hence, this phenomenon is observed only for large<br />

enough isl<strong>and</strong>s, with large enough vertical stra<strong>in</strong> gradient.<br />

For the simulated Ge dome, 10. For smaller isl<strong>and</strong>s with<br />

smaller stra<strong>in</strong> gradient, these multiple <strong>scatter<strong>in</strong>g</strong> <strong>effects</strong> are<br />

not observed. This was checked <strong>in</strong> the case of a small Ge<br />

pyramid on Si001 see Fig. 2c for which the <strong>in</strong>tensities<br />

simulated for different i values can be superimposed. Indeed,<br />

<strong>in</strong> that case, the DWBA <strong>analysis</strong> for isl<strong>and</strong>s on a substrate<br />

is identical to that of isl<strong>and</strong>s buried below the surface:<br />

the total scattered amplitude see Eq. 2 is proportional to<br />

the form factor amplitude, F BA k f −k i , accord<strong>in</strong>g to<br />

F DWBA i , f = F BA k f − k i t i t f ,<br />

where t i,f is the usual amplitude transmission function as<br />

a function of <strong>in</strong>cident exit angle.<br />

Hav<strong>in</strong>g determ<strong>in</strong>ed for which conditions the DWBA has<br />

to be used <strong>in</strong>stead of the BA, the question arises whether the<br />

st<strong>and</strong>ard MAD <strong>analysis</strong> us<strong>in</strong>g the BA is still valid for large<br />

isl<strong>and</strong>s under graz<strong>in</strong>g <strong>in</strong>cidence. Equation 2 shows that the<br />

<strong>in</strong>tensity scattered by a stripe at height z is simply the BA<br />

<strong>in</strong>tensity multiplied by the <strong>in</strong>cident <strong>and</strong> exit generalized op-<br />

2<br />

3<br />

4


013112-3 Richard et al. Appl. Phys. Lett. 94, 013112 2009<br />

Ge content<br />

1<br />

α<br />

0.8<br />

i<br />

=0.12°<br />

<br />

α i<br />

=0.2°<br />

<br />

<br />

<br />

0.6<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

<br />

0.4<br />

<br />

<br />

<br />

<br />

0.2<br />

<br />

<br />

<br />

<br />

<br />

<br />

0<br />

3.9 3.92 3.94 3.96 3.98<br />

h (r.l.u.)<br />

FIG. 3. Color onl<strong>in</strong>e Ge content as a function of h as extracted us<strong>in</strong>g<br />

graz<strong>in</strong>g <strong>in</strong>cidence MAD with 12 different energies around the Ge K-edge.<br />

The Ge content extracted for two different <strong>in</strong>cident angles rema<strong>in</strong>s consistent<br />

<strong>in</strong> spite of the strong Ih dependence with i .<br />

tical functions, which are energy <strong>in</strong>dependent far from c <strong>in</strong><br />

the case of a buffer conta<strong>in</strong><strong>in</strong>g nonanomalous atoms. Therefore<br />

dur<strong>in</strong>g a MAD experiment the extracted structure factors<br />

are F Ge <strong>and</strong> F Si multiplied by these functions. S<strong>in</strong>ce the Ge<br />

<strong>composition</strong> is extracted by a ratio of these factors accord<strong>in</strong>g<br />

to x Ge = f Si F Ge /f 0 0<br />

Ge F Si + f Si F Ge , where f Ge <strong>and</strong> f Si are the<br />

Ge Thomson atomic form factor <strong>and</strong> the Si atomic form factor,<br />

respectively, the generalized optical functions disappear<br />

from the result, <strong>and</strong> the deduced <strong>composition</strong> is not affected<br />

by multiple <strong>scatter<strong>in</strong>g</strong> <strong>effects</strong>. This result was experimentally<br />

checked by compar<strong>in</strong>g the results of graz<strong>in</strong>g-<strong>in</strong>cidence MAD<br />

<strong>analysis</strong> of the dome sample for different <strong>in</strong>cident angles.<br />

Figure 3 shows the Ge <strong>composition</strong> calculated from the Ge<br />

<strong>and</strong> Si structure factors F Si <strong>and</strong> F Ge , which were extracted<br />

for two <strong>in</strong>cident angles without any structural hypothesis,<br />

<strong>and</strong> ref<strong>in</strong>ed through a least-squares m<strong>in</strong>imization us<strong>in</strong>g<br />

a dedicated program. 12 This program evaluates the uncerta<strong>in</strong>ties<br />

of F Ge <strong>and</strong> F Si <strong>and</strong> hence the error bars on the Ge<br />

<strong>composition</strong>. The partial structure factors are affected by the<br />

generalized optical function as is the total <strong>scatter<strong>in</strong>g</strong> amplitude.<br />

However, only slight variations <strong>in</strong> the Ge <strong>composition</strong><br />

are observed, with overlapp<strong>in</strong>g error bars. Hence, the correct<br />

Ge content is extracted, whatever the conditions of <strong>in</strong>cidence.<br />

A BA treatment of anomalous diffraction can thus be<br />

performed to extract the <strong>composition</strong> of isostra<strong>in</strong> regions of<br />

the isl<strong>and</strong>s. The subcritical regime i c yields smaller<br />

<strong>scatter<strong>in</strong>g</strong> from the substrate <strong>and</strong> is thus preferred.<br />

To conclude, for large isl<strong>and</strong>s, multiple <strong>scatter<strong>in</strong>g</strong> <strong>effects</strong><br />

strongly distort the <strong>in</strong>tensity distribution <strong>in</strong> reciprocal space<br />

<strong>and</strong> need a complete DWBA <strong>analysis</strong> to extract the lattice<br />

parameter distribution. By contrast, the <strong>in</strong>-plane stra<strong>in</strong> <strong>in</strong><br />

small isl<strong>and</strong>s can be <strong>in</strong>vestigated us<strong>in</strong>g a BA treatment.<br />

Hence, <strong>scatter<strong>in</strong>g</strong> from small <strong>and</strong> large isl<strong>and</strong>s can be dist<strong>in</strong>guished<br />

simply by vary<strong>in</strong>g the <strong>in</strong>cident angle around the<br />

critical angle c , check<strong>in</strong>g if the <strong>in</strong>tensities can be superimposed<br />

to each other by rescal<strong>in</strong>g. This may be most useful for<br />

bimodal isl<strong>and</strong> growth e.g., pyramids <strong>and</strong> domes <strong>in</strong> the<br />

Ge/Si case to discrim<strong>in</strong>ate the orig<strong>in</strong> of the measured <strong>in</strong>tensity.<br />

Most importantly, the st<strong>and</strong>ard MAD <strong>analysis</strong> can be<br />

applied to all isl<strong>and</strong>s as long as the x-ray energy is far from<br />

an absorption edge of the substrate <strong>and</strong> the substrate does not<br />

conta<strong>in</strong> any elements violat<strong>in</strong>g this condition.<br />

Z.Z. acknowledges support from the FWF, Vienna. We<br />

acknowledge Hubert Renevier for fruitful discussions.<br />

1 J. Stangl, V. Holý, <strong>and</strong> G. Bauer, Rev. Mod. Phys. 76, 7252004.<br />

2 A. Létoublon, V. Favre-Nicol<strong>in</strong>, H. Renevier, M. G. Proietti, C. Monat, M.<br />

Gendry, O. Marty, <strong>and</strong> C. Priester, Phys. Rev. Lett. 92, 186101 2004.<br />

3 I. Kegel, T.-H. Metzger, A. Lorke, J. Peisl, J. Stangl, G. Bauer, J. M.<br />

Garcia, <strong>and</strong> P. Petroff, Phys. Rev. Lett. 85, 1694 2000.<br />

4 T. U. Schülli, J. Stangl, Z. Zhong, R. T. Lechner, M. Sztucki, T. H.<br />

Metzger, <strong>and</strong> G. Bauer, Phys. Rev. Lett. 90, 066105 2003; T. U. Schülli,<br />

M. Stoffel, J. Stangl, R. T. Lechner, E. W<strong>in</strong>tersberger, M. Sztucki, T. H.<br />

Metzger, O. G. Schmidt, <strong>and</strong> G. Bauer, Phys. Rev. B 71, 035326 2005.<br />

5 A. Malachias, S. Kycia, G. Medeiros-Ribeiro, R. Magalhães-Paniago, T. I.<br />

Kam<strong>in</strong>s, <strong>and</strong> R. Stanley Williams, Phys. Rev. Lett. 91, 176101 2003.<br />

6 M. Schmidbauer, D. Grigoriev, M. Hanke, P. Schafer, T. Wiebach, <strong>and</strong> R.<br />

Kohler, Phys. Rev. B 71, 115324 2005.<br />

7 I. N. Stranski <strong>and</strong> L. Krastanow Sitzungsber. Akad. Wiss. Wien, Math.-<br />

Naturwiss. Kl., Abt. 2B 146, 797 1938.<br />

8 M. Rauscher, R. Paniago, H. Metzger, Z. Kovats, J. Domke, H. D.<br />

Pfannes, J. Schulze, <strong>and</strong> I. Eisele, J. Appl. Phys. 86, 6763 1999.<br />

9 J. Coraux, V. Favre-Nicol<strong>in</strong>, M. G. Proietti, B. Daud<strong>in</strong>, <strong>and</strong> H. Renevier,<br />

Phys. Rev. B 75, 235312 2007.<br />

10 V. Ranjan, G. Allan, C. Priester, <strong>and</strong> C. Delerue, Phys. Rev. B 68, 115305<br />

2003.<br />

11 I. Kegel, T.-H. Metzger, A. Lorke, J. Peisl, J. Stangl, G. Bauer, K. Nordlund,<br />

W. V. Schoenfeld, <strong>and</strong> P. Petroff, Phys. Rev. B 63, 035318 2001.<br />

12 V. Favre-Nicol<strong>in</strong>, <strong>in</strong> preparation.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!