Plan-view transmission electron microscopy investigation of GaAs/(In,Ga)As core-shellnanowiresJavier Grandal, Mingjian Wu, Xiang Kong, Michael Hanke, Emmanouil Dimakis, Lutz Geelhaar, Henning Riechert

, and Achim Trampert

Citation: Applied Physics Letters 105, 121602 (2014); doi: 10.1063/1.4896505 View online: http://dx.doi.org/10.1063/1.4896505 View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/105/12?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Growth of size and density controlled GaAs / In x Ga 1 x As / GaAs ( x = 0.10 ) nanowires on anodic aluminamembrane-assisted etching of nanopatterned GaAs J. Vac. Sci. Technol. B 28, 1111 (2010); 10.1116/1.3498753 Strain influenced indium composition distribution in GaN/InGaN core-shell nanowires Appl. Phys. Lett. 97, 181107 (2010); 10.1063/1.3513345 Electron microscopy of GaAs/MnAs core/shell nanowires Appl. Phys. Lett. 97, 072505 (2010); 10.1063/1.3481066 Room temperature luminescent InGaAs/GaAs core-shell nanowires Appl. Phys. Lett. 93, 083117 (2008); 10.1063/1.2977475 Measurement of the effective electron mass in GaInNAs by energy-loss spectroscopy Appl. Phys. Lett. 84, 1453 (2004); 10.1063/1.1650906

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: 62.141.165.1

On: Mon, 22 Sep 2014 13:41:05

Plan-view transmission electron microscopy investigationof GaAs/(In,Ga)As core-shell nanowires

Javier Grandal,a) Mingjian Wu, Xiang Kong, Michael Hanke, Emmanouil Dimakis,b)

Lutz Geelhaar, Henning Riechert, and Achim TrampertPaul-Drude-Institut f€ur Festk€orperelektronik, Hausvogteiplatz 5-7, 10117 Berlin, Germany

(Received 23 June 2014; accepted 12 September 2014; published online 22 September 2014)

Plan-view transmission electron microscopy in combination with electron energy-loss spectroscopy

have been used to analyze the strain and the chemical composition of GaAs/(In,Ga)As core-shell

nanowires. The samples consist of an GaAs core and a radially arranged (In,Ga)As layer as quan-

tum well and GaAs outer-shell. The nominal parameters of the quantum well in the two samples

under investigation are: an indium concentration of 25% and a quantum well thickness of 22 nm

and 11 nm, respectively, while the core and the external shell dimensions are fixed. Scanning trans-

mission electron microscopy using high-angle annular dark field detector was performed to verify

the actual dimensions of the layers. Geometric phase analysis was carried out in order to examine

the local strain of the radial (In,Ga)As quantum well, while the local chemical composition was

determined by means of spatially resolved electron energy-loss spectroscopy. Finite elements cal-

culations were carried out in order to simulate the multi-shell structure and extract the actual strain

distribution. The results indicate that there is a uniform strain distribution at the coherent interfaces.

In addition, based on calculations, we show that there is no region in the considered core-shell

structure absolutely free of strain. VC 2014 AIP Publishing LLC.

[http://dx.doi.org/10.1063/1.4896505]

Nanowires (NWs) with core-shell heterostructures are

receiving much attention since their unique optical properties

have some potential applications,1–7 in particular, because

these heterostructures take up a large volume fraction of the

entire NW.8,9 Obviously, the precise control of interfaces

and chemical compositions at nano-scale level is very impor-

tant for such three-dimensional (3D) structures. In addition,

it is expected that beyond the critical dimensions of the NWs

(as in the planar case), the epitaxial strain can be relieved by

generating surface undulations or edge-type misfit disloca-

tions both along and/or around the NWs.10,11 Due to the NW

geometry and the resulting 3D strain distribution, a detailed

structural characterization is required, for instance, by means

of transmission electron microscopy (TEM), to analyze

whether the shells or quantum wells (QWs) are either relaxed

or strained and, if so, which is the mechanism for the men-

tioned strain relief.

One of the key aspects in the characterization of the

NWs by TEM is the combination of several projections in

order to obtain comprehensive data. In the case of core-shell

NWs, projections parallel and perpendicular to the wire axis

are generally used, named as plan-view and cross-section,

respectively. In this letter, we will show only results based

on plan-view configuration although cross-section projec-

tions were also taken in order to determine the crystal struc-

ture of the NWs.

The GaAs/(In,Ga)As core-shell NW samples were

grown on Si(111) substrates by molecular beam epitaxy

(MBE) in the self-assisted approach. The NW density is in

the order of �107 cm�2. The nominal structure consists of a

GaAs core with a diameter of 50 nm followed by a 50 nm-

thick shell grown at lower temperature. Since TEM cannot

distinguish between this shell and the core, in what follows

we will refer to the combination of both, as the inner-shell

plus the core (ISC). The structure continues with a radially

arranged (In,Ga)As QW and finally, an external GaAs outer-

shell layer of 30 nm thickness that was grown after the QW.

The two samples under investigation have the nominal in-

dium concentration of 25%, whereas the nominal QW thick-

ness was 22 nm for sample A and 11 nm for sample B.

Details about the growth parameters of a similar structure

are given elsewhere.12

The plan-view TEM specimen preparation was per-

formed by a two-step procedure adapted for achieving better

reproducibility when working with NWs longer than 1 lm.13

First, the NWs were mechanically stabilized (without detach-

ing the NWs from the substrate) spreading 30 ll of a solution

of EPO-TEK glue from Epoxy Technology on acetone

(3:100) over the sample and hardened at 150 �C for 30 min.

In the second step, the NW samples were thinned, from the

back-side only, using standard methods of grinding, dim-

pling, and Argon ion milling until electron transparency was

reached.

The samples were characterized by conventional and

high-resolution (HR) TEM either on a JEOL JEM-3010 or

JEM-2100F microscope. Electron energy-loss spectroscopy

(EELS) was performed on the JEM-3010 microscope operat-

ing at 300 kV. Scanning (S) TEM images were obtained on

the JEM-2100F system operating at 200 kV and recorded

using high-angle annular dark field (HAADF) detector for

Z-contrast imaging conditions.14 The In M4,5 edges and the

a)Author to whom correspondence should be addressed. Electronic mail:

grandal@pdi-berlin.deb)Present address: Helmholtz-Zentrum Dresden-Rossendorf (HZDR),

Bautzner Landstraße 400, 01328 Dresden, Germany.

0003-6951/2014/105(12)/121602/5/$30.00 VC 2014 AIP Publishing LLC105, 121602-1

APPLIED PHYSICS LETTERS 105, 121602 (2014)

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: 62.141.165.1

On: Mon, 22 Sep 2014 13:41:05

As L2,3 edges were used to quantitatively measure the

indium concentration, an InAs sample was used as a refer-

ence to extract the local In composition.15 Here, the EEL

spectra were recorded inside the (In,Ga)As QW of sample A

(nominal QW width of 22 nm) using a spot size of 6 6 2 nm

(measured at the full width at half maximum of the incident

electron beam). Since the strain calculations of realistic mor-

phologies are hardly possible within an analytical attempt, a

numerical approach based on finite element method (FEM)

may serve as a powerful alternative. FEM calculations were

carried out using the commercial program package

MarcMentat. Thereby, the full elastic anisotropy was consid-

ered by elastic constants, c11, c12, and c44, assuming

Vegard’s law for the ternary compound (In,Ga)As.

Figure 1 depicts plan-view conventional TEM images of

sample A recorded at low (a) and higher magnification (b).

The low density of the NWs allows both their collective and

individual characterization. The NWs clearly present a

hexagonal-like shape with well-defined sharp side facets, as

it is expected for the growth of zinc-blende (ZB) NWs along

the axial h111i direction. This orientation is experimentally

confirmed by selected area electron diffraction (SAD) using

both cross-section (not shown) and plan-view [Fig. 1(c)] pro-

jections. Comparing the results from imaging and diffraction

[i.e., Figs. 1(b) and 1(c)] we found that the side facets consist

of non-polar f1�10g lattice planes, as it has been reported16

for self-assisted grown GaAs-based NWs. Moreover, the

plan-view conventional TEM of the NWs ensemble, points

to a defined NW alignment with respect to each other (and

we assume that also with the substrate) since a small average

twist angle of 63.5� is verified by measuring the relative

rotation angle of the f1�10g planes of all NWs observed in

Fig. 1(a).

Figure 2 shows the HAADF-STEM image of a typical

NW of sample A [Fig. 2(a)] taken along the h111i zone axis.

In this Z-contrast image, the brighter (In,Ga)As QW formed

around the GaAs ISC is clearly distinguishable from the ISC

and the GaAs outer-shell, retaining the hexagonal symmetry.

Therefore, it is possible to measure the thickness of the

outer-shell, the (In,Ga)As QW and the ISC. Based on inten-

sity scan profiles across the NW [Figs. 2(b) and 2(c)], the

QW thickness is determined for all six sidewalls facets, with

values ranging from 20 nm to 23 nm for a selected NW of

sample A (Table I) and 8 nm to 11 nm for a selected NW of

sample B (not shown). These measured thicknesses corre-

spond well to the nominal values; nevertheless, the 5%–10%

variation in thickness on different facets of the same NW is

significant, and will affect the optical properties.17 Similar

variations were observed for the thickness of the GaAs

outer-shell for the same selected NWs as can be seen in

Table I. However, these variations in the dimensions of the

QW and outer-shell are also present for different NWs within

the same sample and so cannot be taken as statistical values.

The fluctuations in the QW and shell thicknesses and in the

facets longitudinal dimensions can be attributed to a shadow-

ing effect (still present even having a low NW density) dur-

ing the growth, since they were grown in a self-assembled

fashion and not in a patterned way, which minimizes the

mentioned shadowing by controlling the spacing between

NWs. All the NWs observed present a distortion from the

ideal hexagonal shape,18 being even more pronounced in

those that grow very close as those observed in [Fig. 1(d)],

where it is obvious that the extreme distortion of the shadow-

ing effect caused in the external shell and QW dimensions.

On the other hand, the contrast along the QW facet

[Fig. 2(d)] appears almost constant, at the examined resolu-

tion, indicating a homogeneous lateral indium distribution

and the absence of indium clusters, respectively.

Given the complex geometry of the NWs (3D objects),

in order to extract the actual strain of the QW, it is required

to know the actual indium composition. The analysis of the

FIG. 1. Conventional bright-field plan-view TEM images taken along h111izone axis at low magnification (a) and at higher magnification (b). It is

clearly visible that the NWs present a well defined hexagonal shape. (c)

Corresponding SAD pattern of NWs reflecting the zinc-blende structure

along the axial h111i direction. In combination with the TEM images in (a)

and (b), it is deduced that the side facets consist of non-polar f1�10g lattice

planes. (d) Bright-field TEM image of two NWs showing strong shadowing

effect. The white dotted hexagons selecting the (In,Ga)As quantum well are

a guide for the eye.

121602-2 Grandal et al. Appl. Phys. Lett. 105, 121602 (2014)

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: 62.141.165.1

On: Mon, 22 Sep 2014 13:41:05

local chemical composition was carried out by means of spa-

tially resolved EELS in the TEM (Fig. 3). The indium con-

centration measured at different facets of a single NW and

also for different NWs, varies between 22% 6 3% and

28% 6 3% (Table II); these values are in the range of the

nominal value of 25%. Consequently, the core-shell NWs

show a similar behavior in terms of variation in QW thick-

ness and composition between the different facets of the

same NW. The correlation between the In composition varia-

tion and the variation of the QW dimensions needs further

investigation.

HR micrographs were taken along the h111i zone axis

in order to examine the interface quality and to extract the

local strain of the radial (In,Ga)As/GaAs heterostructures

using the geometric phase analysis (GPA) method.19 The

GPA strain represents a relative lattice strain that is generally

defined as eGPA ¼ dloc�dref

dref, where dloc is the lattice distance of

the area of interest and dref is the lattice distance of a refer-

ence area taken as zero-strain level in the same image under

analysis. Figures 4(a) and 4(b) show the GPA strain map of

exx and eyy components, respectively, at the edge of two adja-

cent facets. According to the definition of GPA strain indi-

cated above and the relative orientation of this image it can

be interpreted that exx ¼dInGaAsf�2�24g �dGaAs

f�2�24gdGaAsf�2�24g

, where df�2�24g is the

f�2�24g plane distance for the (In,Ga)As QW or the GaAs ISC

taken as reference and that eyy ¼dInGaAsf2�20g �dGaAs

f2�20gdGaAsf2�20g

, where df2�20g is

the f2�20g plane distance for the (In,Ga)As QW or the GaAs

ISC taken as reference. For the experimental GPA calcula-

tions, we assumed that the GaAs ISC is in a bulk state. The

strain maps reveal larger radial lattice distances for the

f2�20g planes, as compared with the f�2�24g planes, in the

(In,Ga)As QW with respect to the GaAs ISC reference lat-

tice. Since no misfit dislocations are observed in the GPA

maps, we can conclude that the interfaces are coherently

strained. The fluctuations in the strain maps observed in

Figs. 4(a) and 4(b) are due to the noise in the corresponding

HRTEM lattice image. The local eyy GPA strain as a function

of distance across the (In,Ga)As/GaAs QW is plotted in Fig.

FIG. 2. (a) Plan-view HAADF-STEM

image taken along the h111i zone axis

of a core-shell NW of sample A. The

labels (a–f) refer to Table I, in which

the QW and external shell thicknesses

of all six facets are indicated. (b)

Intensity profile taken across the NW

as indicated in Figure 2(a) (blue rec-

tangle). (c) Detailed intensity profile of

the QW across one side facet (red

rectangle). (d) Intensity variation

measured along the QW of the NW in

Figure 2(a) (green area).

TABLE I. Thickness variation of the QW and the external shell in one single

nanowire (Fig. 2(a)), the error values are extracted from the measurement.

a b c d e f

QW thickness (nm) 20 6 1 23 6 1 22 6 1 22 6 1 20 6 1 20 6 1

Shell thickness (nm) 26 6 1 29 6 1 31 6 1 29 6 1 27 6 1 26 6 1

FIG. 3. Typical electron energy-loss spectrum of the (In,Ga)As QW. The

inset depicts two different NWs for which the In concentration extracted on

the labeled facets is provided in Table II. The scale bar of the insets repre-

sents 50 nm.

121602-3 Grandal et al. Appl. Phys. Lett. 105, 121602 (2014)

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: 62.141.165.1

On: Mon, 22 Sep 2014 13:41:05

4(c) for a sector at the center of the facet (see inset) giving

an average GPA strain of �2.2%.

In order to strengthen and support our experimental find-

ings, we have applied FEM, which provides total strain,

comparable to the experimental GPA strain. According

to structural dimensions and chemical compositions

provided by TEM and EELS, we have created a GaAs/

(In0.25Ga0.75)As/GaAs core-shell NW with an inner GaAs

core diameter of 150 nm and two subsequent QW and outer-

shell of 20 nm thickness. The outer-shell thickness in FEM

calculations is set to 20 nm in order to directly compare with

the dimensions of the NW used for GPA maps in Figs. 4(a)

and 4(b). The in-plane components exx and eyy obtained by

FEM calculations are depicted in Figs. 4(d) and 4(e), respec-

tively, in cross-sections through the core-shell NW suffi-

ciently far away from the NW top facet and the base. The

length of the NW is considered long enough so the boundary

conditions do not have any impact on the total strain. Similar

to the experimental GPA analysis, a reference must be cho-

sen to which the total FEM strain is given. Therefore, as we

assumed the GaAs ISC to be in a bulk state, we used bulk

GaAs (infinite in size and thus non-strained). Since the NW

growth is supposed to take place in a pseudomorphic way,

the strain ezz yields a constant (not shown), whose value

depends on the volume ratio GaAs vs. (In,Ga)As and reflects

a vertically expanded GaAs lattice within core and outer

shell, and a vertically compressed (In,Ga)As QW. Based on

FEM, we derived numerically maximum values of eyy of

�2.4%, in good agreement with the average experimental

eyy GPA strain, �2.2%. It is worth to mention that the strain

distribution obtained by FEM is qualitatively similar to the

one observed experimentally in the GPA strain maps (Figs.

4(a) and 4(d) for exx and Figs. 4(b) and 4(e) for eyy). Then,

the common general argument of an effective strain relaxa-

tion in NWs (due to the lack of geometrical constraints)

obviously fails in the case of core-shell structures.

Figure 5 shows a HRTEM image of the two (In,Ga)As

QW interfaces. At macroscopic level, the HAADF-STEM

images and the GPA show sharp interfaces, however, the

detailed observation of HR micrographs reveals an atomic

scale roughening at both interfaces of the QW and also at the

surface of the outer-shell (not shown). There are several

mechanisms that may induce this roughening, for instance,

relaxation by changing 2D to 3D growth mode.20 At this

FIG. 4. (a) exx and (b) eyy experimental GPA strain at the edge of two adjacent facets. (c) Line scan performed perpendicular to the (In,Ga)As QW taken from

an area of the experimental eyy GPA map shown in the inset (blue rectangle). The eyy GPA map of the inset is taken from a different facet of the same nano-

wire. (d) and (e) are the numerically calculated strain exx and eyy , respectively. Bulk GaAs serves as reference, thus, the data are directly comparable to the ex-

perimental GPA strain shown above. Cartesian coordinates are used to represent all the figures, so the hexagonal symmetry yields in a 2-fold symmetry. The

white dotted lines are a guide for the eye.

TABLE II. Indium composition variation of the QW in different positions of

two different NWs, see Figure 3, the error values are extracted from the

measurement.

a b c d e

Indium composition (%) 28 6 3 26 6 3 22 6 3 26 6 3 22 6 3

121602-4 Grandal et al. Appl. Phys. Lett. 105, 121602 (2014)

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: 62.141.165.1

On: Mon, 22 Sep 2014 13:41:05

respect, even the GPA strain indicates that the interfaces

between the (In,Ga)As QW and the GaAs ISC and outer-

shell are coherently grown, and provided that these are 3D

objects and that the distortion of the lattice is much more

complicated than in the planar QW case (tetragonal distor-

tion), as shown by FEM, we cannot conclude that there is not

an inherent relaxation of the QW. A different origin of this

atomic scale roughening could be the intrinsic atomic

arrangement of the f2�20g planes, as in the case of the growth

of GaAs (110) planar films, and finally, we can also attribute

it to non-optimal growth conditions of the inner GaAs shell

(lower growth temperature) that can induce the roughening

of the first interface and subsequent interfaces and sidewalls.

It is worth to mention that most of the NWs present this

roughening at the sidewalls surface (GaAs outer-shell); how-

ever, we have also found cases in which the roughening is

not as pronounced. In order to elucidate the actual origin of

this roughening, further investigation is necessary.

In conclusion, we have investigated GaAs/(In,Ga)As

core-shell NWs grown by molecular beam epitaxy by plan-

view transmission electron microscopy. The NWs present a

hexagonal-like shape with a distortion probably due to shad-

owing effects during the growth. In addition, and likely due

to this effect, the dimensions of the QW exhibit a significant

difference in thickness of 5%–10% for different facets of

the same NW and also for different NWs. The indium com-

position, measured by EELS inside the QW, also presents a

local deviation of 612% from the nominal value. We have

observed a homogeneous GPA strain distribution across the

NWs investigated. FEM calculations indicate that there is no

region in the considered core-shell structure completely free

of strain as has been also observed in the external shell of the

experimental GPA strain maps. They further indicate a rather

complex relaxation scenario especially close to the NW

edges, which needs further investigation. The atomic rough-

ening observed at the QW interfaces with the core and the

external shell might be a source of relaxation.

The authors would like to acknowledge Mrs. D. Steffen

for the sample preparation and Mrs. A. Pfeiffer for technical

assistance. Thanks are also due to Dr. E. Luna for critical

reading of this manuscript and valuable comments and

discussion.

1L. J. Lauhon, M. S. Gudiksen, D. Wang, and C. M. Lieber, Nature 420,

57–61 (2002).2F. Qian, S. Gradecak, Y. Li, C. Wen, and C. M. Lieber, Nano Lett. 5,

2287–2291 (2005).3A. Fontcuberta i Morral, D. Spirkoska, J. Arbiol, M. Heigoldt, J. R.

Morante, and G. Abstreiter, Small 4, 899–903 (2008).4J. A. Czaban, D. A. Thompson, and R. R. LaPierre, Nano Lett. 9, 148–154

(2009).5D. Spirkoska, A. Fontcuberta i Morral, J. Dufouleur, Q. Xie, and G.

Abstreiter, Phys. Status Solidi RRL 5, 353–355 (2011).6A. Biermanns, T. Rieger, G. Bussone, U. Pietsch, D. Gr€utzmacher, and M.

I. Lepsa, Appl. Phys. Lett. 102, 043109 (2013).7M. V. Nazarenko, N. V. Sibirev, K. W. Ng, F. Ren, K. S. Ko, V. G.

Druvrovskii, and C. Chang-Hasnain, J. Appl. Phys. 113, 104311

(2013).8B. M. Kayes, H. A. Atwater, and N. S. Lewis, J. Appl. Phys. 97, 114302

(2005).9E. Garnett and P. Yang, Nano Lett. 10, 1082–1087 (2010).

10R. Popovitz-Biro, A. Kretinin, P. Von Huth, and H. Shtrikman, Cryst.

Growth Des. 11(9), 3858–3865 (2011).11G. Perillat-Merceroz, R. Thierry, P.-H. Jouneau, P. Ferret, and G. Feuillet,

Appl. Phys. Lett. 100, 173102 (2012).12E. Dimakis, U. Jahn, M. Ramsteiner, A. Tahraoui, J. Grandal, X. Kong, O.

Marquardt, A. Trampert, H. Riechert, and L. Geelhaar, Nano Lett. 14(5),

2604–2609 (2014).13E. Luna, J. Grandal, E. Gallardo, J. M. Calleja, M. A. S�anchez-Garc�ıa, E.

Calleja, and A. Trampert, “Investigation of III–V nanowires by plan-view

transmission electron microscopy: InN case study,” Microsc. Microanal.

(published online).14S. J. Pennycook and P. D. Nellist, in Impact Electron Scanning Probe

Microscopy on Materials Research, edited by D. Rickerby, G. Valdre, and

U. Valdre (Kluwer Academic Publishers, 1999), p. 161.15D. B. Williams and C. B. Carter, Transmission Electron Microscopy, 2nd

ed. (Springer, 2009).16F. Jabeen, V. Grillo, S. Rubini, and F. Martelli, Nanotechnology 19,

275711 (2008).17A. Trampert, J.-M. Chauveau, K. H. Ploog, E. Tourni�e, and A. Guzm�an,

J. Sci. Vac. Technol. B 22, 2195 (2004).18O. Brandt, S. Fern�andez-Garrido, J. K. Zettler, E. Luna, U. Jahn, C. Cheze,

and V. M. Kaganer, Cryst. Growth Des. 14, 2246 (2014).19M. J. H€ytch and L. Potez, Philos. Mag. A 76, 1119–1138 (1997).20S. Guha, A. Madhukar, and K. C. Rajkumar, Appl. Phys. Lett. 57, 2110

(1990).

FIG. 5. HRTEM micrograph of the QW taken along the h111i zone axis.

Roughening at the interfaces between the QW and the core and shell is

observed.

121602-5 Grandal et al. Appl. Phys. Lett. 105, 121602 (2014)

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: 62.141.165.1

On: Mon, 22 Sep 2014 13:41:05