Structural and optical properties of high quality zinc-blende/wurtzite GaAs hetero-nanowires

Abstract

The structural and optical properties of 3 different kinds of GaAs nanowires with 100% zinc-blende structure and with an average of 30% and 70% wurtzite are presented. A variety of shorter and longer segments of zinc-blende or wurtzite crystal phases are observed by transmission electron microscopy in the nanowires. Sharp photoluminescence lines are observed with emission energies tuned from 1.515 eV down to 1.43 eV when the percentage of wurtzite is increased. The downward shift of the emission peaks can be understood by carrier confinement at the interfaces, in quantum wells and in random short period superlattices existent in these nanowires, assuming a staggered band-offset between wurtzite and zinc-blende GaAs. The latter is confirmed also by time resolved measurements. The extremely local nature of these optical transitions is evidenced also by cathodoluminescence measurements. Raman spectroscopy on single wires shows different strain conditions, depending on the wurtzite content which affects also the band alignments. Finally, the occurrence of the two crystallographic phases is discussed in thermodynamic terms. Comment: 24 pages

Full text

Page 1

Structural and optical properties of high quality zinc-blende/wurtzite GaAs hetero-
nanowires

D. Spirkoska1, J. Arbiol2, A. Gustafsson3, S. Conesa-Boj2, F. Glas4, I. Zardo1, M. Heigoldt1, M.
H. Gass5, A. L. Bleloch5, S. Estrade2, M. Kaniber1, J. Rossler 1, F. Peiro2, J.R. Morante2,6, L.
Samuelson3, G. Abstreiter1 and A. Fontcuberta i Morral1,7

1 Walter Schottky Institut and Physik Department, Technische Universitaet Muenchen
Am Coulombwall 3, 85748 Garching (Germany)
2 Departament d’Electrònica, Universitat de Barcelona, Marti i Franques, 08028 Barcelona,
Spain
3 Solid State Physics, The Nanometer Consortium, Lund University, Box 118, Lund 22100,
Sweden
4 CNRS-LPN, Route de Nozay, 91460 Marcoussis, France
5 SuperSTEM Laboratory, STFC Daresbury, Daresbury WA4 4AD, United Kingdom
6 IREC, Catalonia Institute for Energy Research, Barcelona 08019, CAT, Spain
7 Laboratoire des Matériaux Semiconducteurs, Ecole Polytechnique Fédérale de Lausanne, 1015
Lausanne, Switzerland


Abstract

The structural and optical properties of 3 different kinds of GaAs nanowires with 100% zinc-
blende structure and with an average of 30% and 70% wurtzite are presented. A variety of
shorter and longer segments of zinc-blende or wurtzite crystal phases are observed by
transmission electron microscopy in the nanowires. Sharp photoluminescence lines are observed
with emission energies tuned from 1.515 eV down to 1.43 eV when the percentage of wurtzite is
increased. The downward shift of the emission peaks can be understood by carrier confinement
at the interfaces, in quantum wells and in random short period superlattices existent in these
nanowires, assuming a staggered band-offset between wurtzite and zinc-blende GaAs. The latter
is confirmed also by time resolved measurements. The extremely local nature of these optical
transitions is evidenced also by cathodoluminescence measurements. Raman spectroscopy on
single wires shows different strain conditions, depending on the wurtzite content which affects
also the band alignments. Finally, the occurrence of the two crystallographic phases is discussed
in thermodynamic terms.

Keywords: wurtzite, zinc-blende, bandgap engineering, nanowires, quantum structures


1. Introduction

Heterostructures consist of the combination of two materials, with different band gaps and
electron affinities. Heterostructures of type I are formed when a small bandgap semiconductor is
surrounded by a larger bandgap material having its conduction-band edge at a higher energy and
its valence-band edge at a lower energy than those in the small bandgap material, while
heterostructures of type II are characterized by a staggered band-edge lineup. In heterostructures

Page 2

of type I, electrons and holes will tend to localize in the lower bandgap material, while in type II
these two types of carriers will be spatially separated. Heterostructures and multilayer-structures
have gained huge technological relevance over the past 40 years because the electronic and
optical properties of semiconductors can be tailored and modified with respect to their bulk
counterpart [1,2]. Heterostructures are formed typically with chemically different
semiconductors, such as GaAs and Al1-xGaxAs. In this work, we present a detailed structural
analysis as well as the optical properties of GaAs based ‘heterostructures’, whose junction is
formed by the same material in the two crystalline phases zincblende and wurtzite. Such a
combination presents the typical characteristics of heterostructures, because the bandgap and
electron affinity of semiconductors depend on the crystalline phase.
In the bulk state, GaAs is stable in the zinc-blende structure. When reduced to a nanoscale
volume, such as in the form of nanowire, wurtzite structure becomes stable. In this way, zinc-
blende/wurtzite heterostructures in arsenides and phosphides have been realized [3,4,5,6,7,8]. An
example of such a wurtzite/zinc-blende/ wurtzite quantum well structure is given in Fig. 1. An
aberration corrected high angle annular dark Field (HAADF) scanning transmission electron
micrograph (STEM) of a 3 monolayer thick segment of zinc-blende GaAs embedded in a
wurtzite GaAs matrix is shown in Fig 1a. A model of the atomic positions has been
superimposed. For clarity, this is plotted enlarged in Figure 1b, where the theoretically predicted
band alignment [9] of this structure is also superimposed. In such a potential profile, electrons
are confined in the zinc-blende segment, while holes occupy the higher valence band states in the
surrounding wurtzite structures.

The most successful synthesis method of semiconductor nanowires has made use of the Vapor-
Liquid-Solid mechanism, in which the reactants are supplied in the vapor phase and decomposed
and/or directly incorporated in a metal seed [10]. The thermodynamics of two-dimensional
nucleation of the seeds at the interface between the catalyst droplet and the solid nanowire
determines the crystalline quality of the nanowires [11]. In general, the existence of crystal
imperfections in semiconductors limits the expected performance of optoelectronic devices.
Crystalline imperfections introduced during growth of III-V semiconductor nanowires like InP
and GaAs are very common and include rotational twins and polytypism between wurtzite and
zinc blende structures [12,13,14,15,16,17,18]. From the crystallographic point of view, zinc-
blende and wurtzite structures differ only in the stacking periodicity of the (111)ZB/(0010)WZ
oriented planes [19,20,21]. From the fundamental point of view, the occurrence of polytypism in
III-V nanowires is particularly interesting. As the wurtzite phase is not stable in the bulk form,
the electrical and optical properties are not well known. Twinning or wurtzite/zinc-blende
nanowire heterostructures or superlattices should result in a modification of the band structure,
generating new forms of band offsets and electronic minibands in a chemically homogeneous
material like GaAs [22,23]. From the technological point of view, achievement of control of the
wurtzite/zinc-blende phases in nanowires may enable the fabrication of new kinds of devices
such as quantum wire cascade lasers [24].

The occurrence of wurtzite structure as one of the main crystalline phases of III-V nanowires has
been discussed by several authors. From the thermodynamic point of view, the surface energy of
{1100} wurtzite planes might be low in comparison with {110} and {111}A/B of zinc-blende,
although detailed atomistic calculations are still lacking. As a consequence, for small radii
nanowires, the wurtzite phase would be more stable than the zinc-blende. In the case of GaAs,

Page 3

the critical radius under which wurtzite is expected to be the most stable phase lies between 5
and 25.5 nm, depending on the theory [25,26,27]. Other thermodynamic considerations relate the
formation of wurtzite nanowires with a high supersaturation in the catalyst. As nanowires do not
grow under thermodynamic equilibrium conditions, kinetic theories have also been used to
understand the occurrence of wurtzite nanowires. These theories indicate that the nucleation of
new crystalline layers in the nanowire occurs at the triple phase line [20,28]. This type of
nucleation kinetically favors crystalline structures exhibiting minimal surface energy facets,
again in favor of the wurtzite structure. It is important to note that none of these works refer to
nanowire grown without using gold as a catalyst. The optical properties of wurtzite/zinc-blende
structures have been investigated in the past by photoluminescence spectroscopy. To date,
relatively broad spectra have been obtained, which do not elucidate the quantum nature of these
structures [29,30].
In this work, we present structural and optical properties of very clean GaAs zinc-
blende/wurtzite nanowire heterostructures with different average contents of wurtzite phases
exhibiting localized and sharp emission characteristics. Thanks to the use of local spectroscopy
and time-resolved techniques such as confocal photoluminescence and cathodoluminescence, the
quantum confined nature of these heterostructures is elucidated. The optical characterization is
accompanied with Raman spectroscopy, in order to reveal the presence of strain. Some
theoretical considerations on the growth mechanisms of each of the crystalline phases are also
included.

2. Experimental details

The synthesis of the wires was carried out in a Gen-II MBE system. Two-inch (001) and (111)B
GaAs wafers coated with a sputtered 10 nm thick silicon dioxide were used. The nanowire
growth was performed at a nominal GaAs growth rate of 0.25 Ả/s, a substrate temperature of 630
°C and with 7 rpm rotation. More details on the synthesis process and the growth mechanisms
can be found elsewhere [31,32]. In this synthesis method, the arsenic beam flux has been varied
from 3.5x10-7 mbar to 3.5x10-6 mbar. Here, we present the results on three different arsenic beam
flux conditions, which will be referred in the following as samples α, β and γ γ –see Table 1-.
Nanowires of sample α were deposited at 3.5x10-6 mbar, sample β at 8.8x10-7 mbar and sample γ γ
at 3.5x10-7 mbar. By keeping the gallium rate and substrate temperature constant, the arsenic
beam flux directly influences the supersaturation of the gallium droplet. In this way, the arsenic
beam flux directly determines the growth rate of the nanowires [32]. Nanowires of type α, β and
γ γ correspond to a nanowire growth rate of 3, 1 and 0.3 μm/h, respectively
The samples were prepared for the transmission electron microscopy (TEM) analysis by
mechanically removing the GaAs nanowires from the substrate with a razor blade and diluting
them in a hexane suspension. A drop was then deposited on a holey carbon copper grid. Before
introducing the sample into the microscope, it was introduced into the Plasma Cleaner for 15
seconds. The morphology and structure of the nanowires was characterized by Raman
spectroscopy, high resolution transmission electron microscopy (HRTEM) and scanning
transmission electron microscopy (STEM) in bright field (BF) and high angle annular dark field
(HAADF) modes in a Jeol 2010F field emission gun microscope with a 0.19 nm point to point
resolution. Further Cs-corrected HRHAADF analyses were performed on a dedicated VG HB
501 STEM retrofitted with a Nion quadrupole-octupole corrector.

Page 4

The optical properties were investigated by photoluminescence (PL) and cathodoluminescence
(CL) spectroscopy at the single nanowire level. PL spectroscopy experiments on single
nanowires were carried out by the use of a confocal microscope embedded in a He4 cryostat [33].
The samples could be scanned below the confocal objective (NA=0.65) with piezopositioners
with a spatial accuracy in the nanometer range. The measurements were realized at a temperature
of 4,2 K, using the 632.8 nm line of a He-Ne laser as an excitation source. The spot size of the
laser in the confocal microscope is about 0.8 μm. The luminescence was detected and analyzed
by the combination of a grating spectrometer and Si charge coupled device. The spectrometer
has a resolution of 500 μeV. In order to avoid interference of the PL signal with the underlying
GaAs substrate, the nanowires were first mechanically removed from the initial substrate and
transferred in low densities to an oxidized piece of silicon. Single nanowires could be localized
on the surface by scanning reflectivity measurements over areas of up to 30 μm2. Time resolved
measurements were realized by exciting with a pulsed Ti:sapphire laser and detecting the
photoluminescence intensity as a function of time with a photomultiplier. The laser pulse width
and repetition rate were 100ps and 12 ns, respectively.

For CL measurements, the samples were prepared in a similar way on a gold coated silicon
substrate – gold ensures a good extraction of the impinging electrons such that no charging
occurs-. CL was realized in an adapted scanning electron microscope. The luminescence was
detected and analyzed by the combination of a grating spectrometer and a photomultiplier. The
temperature of the sample was 9 K. The beam was focused to a small spot of 50 nm in diameter
and was scanned over the sample in order to obtain a map of the spectral variations along the
sample. More details on the technique can be found in reference [34].

Spatially resolved Raman spectroscopy was realized with a µ-Raman setup in the backscattering
configuration on single GaAs nanowires transferred on to a silicon substrate. Prior to the
measurements, the nanowires were identified on the substrate by imaging the surface with a
camera. The excitation wavelength was the 514.5 nm line or the Ar+ laser. The used laser power
of the excitation was about 200 μW (equivalent to 70 KW/cm-2), in order to avoid heating of the
nanowire [35]. The scattered light was collected by an XY Raman Dilor triple spectrometer with
a multichannel charge couple device detector. The sample was positioned on a XY piezostage,
which allowed the scanning of the surface (and therefore the nanowire) with a precision of 10
nm.

3. Experimental Results

3.1 Crystalline structure

The crystalline structure of the samples type α, β and γ γ were analyzed by HRTEM and Cs-
corrected HRSTEM in HAADF mode. This method has the advantage that the intensity maxima
in the micrographs correspond to the atomic positions. Additionally, the observed intensity is
nearly proportional to the square of the average atomic number of the elements constituting the
atomic column. We observe that nanowires grown under conditions α –highest growth rate- are
composed of a single zinc-blende structure, while nanowires synthesized under conditions β and
γ γ have an increasing percentage of the wurtzite phase. Remarkably, the typical zig-zag shape of
twinned zinc-blende nanowires was observed in none of the conditions reported here [36]. A

Page 5

typical low resolution TEM measurement of a pure zinc-blende nanowire is shown in Fig.2a. The
lateral facets of these nanowires are of the {110} crystallographic family [37]. Some stripes of
different contrast with thickness of the order of 100 nm are observed. The periodicity of these
zones can vary along the nanowire, and is typically higher for the initial and final stages of
growth, as shown in Fig. 2b. These stripes are caused by a rotational twinning, a 180o rotation
around the growth axis. A high resolution TEM analysis of one such twin interface is shown in
Fig. 2c.

A quite different structure is observed for nanowires obtained under growth conditions β and γ γ.
For these wires, we observe the existence of different sections with wurtzite and zinc-blende
structures. Before we proceed with the statistical analysis of the structure, we present a precise
structural analysis of the zinc-blende and wurtzite sections. A high resolution TEM image of the
nanowire obtained under conditions β is shown in Fig. 2d. A HAADF image of the same region
is superimposed onto the micrograph. Fig 2e shows a magnified image of the boxed region in 2d.
A double twinned interface between two wurtzite domains results in a region composed of two
zinc-blende unit cells. Twinned planes have been marked with dashed lines, while Ga and As
atoms have been marked in orange and green, respectively, for the wurtzite domains, and in red
and blue, respectively for the double unit zinc-blende-cell quantum well. As can be seen clearly
in the sketch, a 180o rotational twin in the zinc-blende structure creates the equivalent stacking of
a single wurtzite unit in the interface. Two consecutive twins, create an atomic stacking
corresponding to 2 wurtzite unit cells, and so on. Following the same model, a 180o rotational
twin in the wurtzite structure would lead to the formation of one GaAs unit cell of zinc-blende -a
rotational twin in wurtzite corresponds to a stacking of ABABCBCBCB-. The atomic
arrangement in the wurtzite/zinc-blende heterostructure shown in Fig. 2d and e is superimposed
in Fig. 2e, so that the different stacking of the planes becomes obvious.

In order to quantify the occurrence of wurtzite structure as a function of the growth conditions, a
detailed statistical analysis of many nanowires was realized. In general, the structural properties
of the nanowires obtained under the same conditions are very similar. The exact stacking
between wurtzite and zinc-blende layers was found to differ, but the average occurrence
coincides. Nanowires of type β typically presented four parts, with different ratios of wurtzite-
zinc-blende phases. HRTEM micrographs of these four parts are shown in Fig.3. In order to
identify the two crystalline phases, we have used a color code. Red denotes wurtzite, while blue
and green refer to the two twinned orientations of zinc-blende. In Fig. 3a the structure of top part
of the nanowire is presented. Typically, the nanowires grown under these conditions finish with
the zinc-blende structure. The density of twins is relatively high, the typical inter-twin distance
about 10 nm. Below this zinc-blende structure that has an extension of about one micron, a zone
composed of an alternation of zinc-blende and wurtzite domains is observed (Fig. 3b). The
domains/regions have a thickness between 5 and 1 nm. This section extends about 0.8 micron on
the nanowire. Below this region, the wurtzite sections increase in thickness up to 10 nm and the
presence of zinc-blende is gradually reduced as it is shown in Fig. 3 c and d. Overall, the part or
the nanowire exhibiting wurtzite structure represents about 30±10% of wurtzite phases.

Nanowires grown under γ γ conditions exhibit a much higher percentage of wurtzite structure.
Typical HRTEM micrographs of the different sections of the nanowire are shown in Fig. 4.
There, the growth of the nanowire also ends with a zinc-blende structure. This time the zone with

Page 6

pure zinc-blende structure is only about 50 nm long. Details on the zinc-blende section are
shown in Fig. 4b. As shown in the detailed micrographs of Fig. 4 c-e, below this initial section
the percentage of wurtzite is very high. There are regions with a high frequency alternation
between wurtzite and zinc-blende phases forming a kind of random superlattice and others with
extremely thin inclusions of zinc-blende layers in wurtzite (narrow quantum wells). We estimate
that the proportion of wurtzite phase under these growth conditions is about 70±10%.
 

3.2 Optical properties

The optical properties of the three types of nanowires were investigated by photoluminescence
and cathodoluminescence spectroscopy respectively at 4.2 and 10K. Typical PL measurements
realized on single nanowires are presented in Fig. 5. For illustration purposes, the spectra have
been shifted vertically by adding an offset. PL spectrum of nanowires type α corresponds to the
top spectrum of the figure. As expected, nanowires from sample α exhibit a single PL peak
which corresponds to the free exciton luminescence of GaAs at 1.515 eV. The PL characteristics
of the other samples are quite different from sample type α –pure zinc-blende. A large downward
shift of the photoluminescence and an increased number of sharp peaks are observed for
nanowires of samples β and γ γ. For sample β, several well defined peaks are present at positions
between 1.51 and 1.46eV, with a FWHM between 2 and 6 meV with some peaks even much
sharper. In the case of sample γ γ, a similar number of peaks is observed, but downshifted to the
energy range between 1.48eV and 1.43eV. No free exciton line of the zinc-blende phase is
observed in these wires.
PL measurements have been performed on various other nanowires corresponding to the growth
conditions β and γ γ. The spectra are in all cases very similar. Several luminescence peaks in the
ranges 1.51-1.46 for sample β and 1.48-1.43 for sample γ γ are observed. The exact position and
intensity of each peak varies from sample to sample and on the position on the nanowire. These
results are in agreement with the fact that the structure varies along the axis of the nanowires.
Additionally, the exact sequence of wurtzite and zinc-blende sections varies from wire to wire. In
order to illustrate this, scanning confocal photoluminescence measurements of one sample of
each of the α, β and γ γ types are presented in Fig. 6a, b and c, respectively. The PL emission of
sample α is rather homogeneous along the wire, with some variations in intensity and linewidth,
but altogether consistent with pure zinc-blende GaAs. A single peak at 1.515 eV is observed,
which corresponds to the free exciton line of zinc-blende GaAs [38]. Sample β exhibits many
more spectral features in the energy range of 1.515 eV and 1.46 eV. The spectral features are not
homogeneous along the length. At the top part of the scan, the luminescence is brighter than at
the bottom part. The free exciton peak at 1.515 eV is present in only some regions of the
nanowire, while lower energy peaks are observed over the whole length. The presence of the
various peaks depends on the position on the nanowire. This may be understood by the variation
in the structure along the wire as presented in Section 3.1. In the wire of type γ γ we find that only
one part of the nanowire shows luminescence peaks, with spectral features between 1.48 and
1.43 eV. Each of the peaks is located in different regions of the nanowire. It is interesting to note
that no PL signal is observed in the top part of the scan up to 1.55 eV – the region between 1.53
and 1.55 eV not shown in the figure-.

Page 7

Before going into detail of the analysis of the PL measurements, we present the results on
cathodoluminescence spectroscopy mapping (Fig. 7). A series of monochromatic
cathodoluminescence images of several nanowires of a sample of type β was recorded. These
nanowires show an emission pattern at energies between 1.51 and 1.46 eV. Monochromatic
images in this energy range exhibit a series of spatially localized bright emission spots, as
illustrated in Fig. 7b-4e. These spots appear at different positions along the wire for different
energies. By comparing the patterns in the monochromatic images, like in Fig. 7e, we conclude
that the different emission energies originate from different locations in the nanowire. By
analyzing the intensity profiles, we can conclude that the carrier diffusion length is very short in
this region of the wire, less than 100 nm. These characteristics advocate highly for the existence
of an array of quantum heterostructures with different quantization energies along the nanowire.
The cathodoluminescence of sample α is also provided for reference in Fig. 7g. These nanowires
show emission related to the free exciton of zinc-blende GaAs from most of its length. This is in
good agreement with the HRTEM measurements and the spatially resolved photoluminescence
studies presented above.

4. Discussion

4.1 Evidence for staggered band-offsets

The photoluminescence peaks and cathodoluminescence maps presented in Fig. 5, 6 and 7 can be
interpreted based on the theoretical band gaps and the band alignment in wurtzite/zinc-blende
heterostructures [9]. The energy gap of unstrained wurtzite GaAs was predicted to be about 33
meV higher than that of zinc-blende [9]. This has been verified recently with PL measurements
of pure wurtzite nanowires [3]. Wurtzite/zinc-blende GaAs heterostructures are believed to
exhibit a type II band alignment as shown schematically in Figs. 1b and 8a. Theory predicts
conduction and valence band discontinuities of 117 and 84 meV, respectively, where the valence
band being higher in wurtzite [9]. Effects of strain and spontaneous polarization at the
wurtzite/zincblende interface are neglected here. With this simple model it is expected that an
electron hole pair confined to the wurtzite/zinc-blende interface will give rise to a spatially
indirect recombination at 1.431 eV -neglecting also the exciton binding energy, which is
expected to be small due to the spatially indirect nature of the exciton. This spatially indirect
recombination should be the lowest energy observable in pure GaAs wurtzite/zinc-blende
multilayer structures. Indeed we observe in wires of type γ γ the lowest PL line typically at 1.43
eV, in surprisingly good agreement with our simple model. When the thickness of wurtzite and
zinc-blende regions are reduced to several nanometers, quantum wells are formed and quantized
levels of electrons and holes appear. Some examples of such type of multilayer structures are
sketched in Fig. 8a. As a consequence of the type II band alignment, electrons and holes are
spatially separated and stored in the zinc-blende and wurtzite regions, respectively. The thinner
the quantum well, the higher is the quantization energy and PL lines are expected between 1. 515
and 1.43 eV. In the case where the wurtzite sections are long enough, one would also expect to
observe luminescence from this phase at about 1.55 eV [39].That we do not observe this pure
wurtzite related luminescence in our samples indicates that the diffusion length of the electrons
in the wurtzite segments is long enough that they are trapped in neighboring zincblende regions
before they recombine.

Page 8

We have calculated the first electron and hole confined states of a wurtzite or zinc-blende
quantum well embedded in a zinc-blende or wurtzite matrix, respectively. These calculations are
indicative and have been realized by assuming the same effective masses for the two GaAs
polytypes of electrons and holes, respectively. The energy values are plotted in Fig. 8b. Due to
the smaller mass of the electron compared to the hole and the higher conduction band
discontinuity, the confinement energy is larger for electrons than for holes. Following these
calculations, as the thickness of the quantum wells is reduced to zero, the first level of the
electron and hole states tends to the value of the conduction and valence band offsets.
Experimentally, we observe that wires of type β β show PL lines at higher energies closer to the
zinc-blende free exciton line, while PL lines in the lower energy range are observed with
increasing wurtzite content (samples γ γ). This cannot be explained with the simple confinement
model presented above. However, one should note that on the nanowire sections where the
wurtzite and zinc-blende regions are below ~3 nm, the quantum wells are not isolated but highly
frequent. We believe that in these regions, the electron and hole wavefunctions overlap and form
minibands that reduce the value of the confinement energy [25,40]. Such kinds of random
superlattices are indeed observed in the TEM analysis of the wires of type γ γ, as shown in Fig. 3b.

Finally, we would like to present further evidence that the observed PL lines are indeed due to
spatially indirect transition from confined carriers in type II quantum heterostructures. The
lifetime of an exciton is a measure of the recombination mechanism. For a direct exciton
recombination in GaAs lifetimes of about a ns or below are expected. On the contrary, spatially
indirect recombinations exhibit longer lifetimes due to the reduced overlap of the wavefunctions.
In order to further assess the nature of the observed peaks, we have performed time resolved
photoluminescence. Results for the 1.515 eV free exciton recombination in a sample of type α α
and a line at 1.46 eV of sample of type β β are shown Fig. 9. The measurements are offset
vertically for clarity. The lifetime of the free exciton line at 1.515 eV is below 300 ps –limited by
the resolution of the experimental set up-, while the lifetime of the peak at 1.46 eV is on the
other hand 8 ns [41]. The lifetime associated with these lower energy peaks is found generally
between about 3 and 8 ns. This is in agreement with recent measurements in similar structures
[29], and it further supports our interpretation that the lower energy peaks are due to spatially
indirect recombinations in type II heterostructures.


4.2. Influence of strain on the optical properties

So far we have neglected the influence of strain on the energy gaps and the optical properties of
wurtzite/zinc-blende nanowires. It is well known that strain in semiconductors induces changes
in the band structure [42]. This manifests itself in a change of the bandgaps, as well as on the
band discontinuities in heterostructures. Spatially resolved Raman spectroscopy is a simple
technique to obtain information about local strain in semiconductors, as built-in strain leads to a
characteristic shift of the phonon frequency [43,44]. We have performed spatially resolved
Raman spectroscopy measurements using a µ-Raman set-up on single GaAs nanowires lying on
a silicon substrate. We know from atomic force microscopy measurements that the nanowires are
lying with one of the facets parallel to the surface. This means that the incident surfaces for the
Raman spectroscopy experiments are of the family {110}ZB or {1010}WZ. In backscattering from
such surfaces the TO phonon is allowed in first order, the LO phonon forbidden. In zinc-blende

Page 9

GaAs the TO phonon at the Brillouin zone center has an energy of about 267 cm-1, the LO
phonon of 291 cm-1 at room temperature, respectively [45]. As it has been shown for GaN, The
phonon dispersion of the wurtzite phase can be estimated from the zinc-blende by a simple
Brillouin zone-folding procedure, as the unit cell of the wurtzite structure is doubled along the
(111) direction. Such a model has been used in the case of GaN [46,47]. With this simple
procedure we expect a back-folded TO mode close to 250 cm-1 in wurtzite GaAs. Raman spectra
of nanowires of the type α, β β and γ γ are shown in Fig. 10. In the spectrum of sample type α, only
the transversal optical (TO) phonon at 266 cm-1 is observed, slightly below the expected bulk
value of zinc-blende GaAs. The LO-Mode at 291 cm-1 is not observed for these wires, as
expected from the selection rules. The TO and weaker LO modes of zinc-blende GaAs are also
observed for sample β. In this case the selection rule for the LO mode may be relaxed due to the
presence of numerous rotational twins. An additional peak is observed at 255 cm-1 that we
attribute to the folded TO phonon being at the zone center in the wurtzite structure. This peak
position is shifted slightly to higher wavenumbers than what is expected from the back-folding
model. For sample γ γ, we also observe all three peaks. The ”zone-folded” TO mode is stronger
and it appears at the expected position (~250 cm-1), while the other two modes are shifted about
4 cm-1 to lower wavenumbers, indicating the existence of tensile strain in the zinc-blende phase.
One should note that both the LO and TO are shifted by the same amount, which is expected for
small values of strain. The existence of strain in wurtzite/zinc-blende structures can be explained
by the difference in lattice constants of the two structures [48]. This means that for zinc-blende
rich nanowires the wurtzite phases may exhibit compressive strain (blue shift of the Raman
mode) while for wurtzite rich nanowires the zinc blende phases are under tensile strain (red shift
of the Raman modes). Details of the Raman studies and the spatial mapping of the wires will be
published elsewhere. Here we just want to emphasis that there is evidence for different strain
conditions in the wires of type β β and γ γ. It is well known that the bandgaps decrease/increase with
tensile/compressive strain, respectively. In addition the band discontinuities vary with strain.
Such effects may have a non-negligible influence on the luminescence energies observed in
different energy regions for samples of type β β and γ .


4.2 Considerations on the occurrence of wurtzite and zinc-blende phases

Comparison of nanowires α α, β β and γ γ shows that, as the As flux increases, the growth rate
increases and the fraction of wurtzite material decreases -similarly, the decrease of the fraction of
wurtzite toward the tops of nanowires β β and γ γ is correlated with some increase of the As flux
during the growth of each nanowire-. These findings are somewhat surprising since the growth
rate is expected to increase with the supersaturation -measured by the difference of chemical
potential μΔ between liquid droplet and solid, including the Gibbs-Thomson effect - and since,
in Au-catalyzed GaAs nanowires, wurtzite phase has been demonstrated to form at high
supersaturation and zinc-blende at lower supersaturation [8].
These seemingly contradictory observations might be reconciled as follows. According to the
analysis of Ref. [29], the nucleation rate for each phase φ is governed not simply by
supersaturation but by the nucleation barrier
Δ
G
edge energy of the 2D nucleus of phase φ that forms at the nanowire/droplet interface and
accounts for the different stacking energies of zb and wz nuclei (
γ .
()
φ
φ
φ
μγΓ−Δ∝
2
, where
φ
γ is the effective
φ
Γ
0 , 0
=>ΓΓ
wzzb
). This edge

Page 10

energy is expected to depend on the vapor pressure, in particular when nucleation happens at the
triple phase line [28], where part of the edge is directly in contact with the vapor. It might thus
happen that, as the As pressure is increased, the supersaturation increases and the nucleation
barriers
G
Δ
decrease. As a consequence, the growth rate of the nanowire increases, whereas
the edge energies of the two phases change. This induces a change of
negative -favoring wurtzite in nanowires γ γ- to positive -favoring zinc-blende in nanowires α α-.
This explanation remains speculative, but the high density of phase changes in nanowires β β and
γ γ indicate that the present conditions are indeed at the border of the zinc-blende/wurtzite
transition. An alternative explanation would be related to the position in which the nucleation
events occur. Nucleation at the triple phase line favors wurtzite, while nucleation away from the
triple phase line favors zinc-blende. Indeed, as the supersaturation increases and the size of the
critical nucleus decreases, the fraction of nucleation events occurring at the triple phase line
decreases at the expense of those away from the triple phase line [28,20]. Given the complexity
of the system, it is clear that more extended studies on the growth mechanisms need to be done
in the future, in order to obtain a more general picture accounting for the existence of wurtzite
and zinc-blende III-V semiconductors in the form of nanowires.


5. Conclusions


In conclusion, we have synthesized GaAs nanowires by molecular beam epitaxy under three
different supersaturation conditions. We have obtained several GaAs polytypes ranging from
pure zinc-blende to wurtzite-rich zinc-blende/wurtzite heterostructures, as demonstrated by
HRTEM. Photoluminescence spectroscopy studies of these nanowires show sharp lines whose
emission energy shifts from 1.515 eV down to 1.43 eV when the percentage of wurtzite material
in the nanowire is increased. This is interpreted as evidence for type II band alignment at the
wurtzite/zinc-blende interface. The results may open the possibility for structural bandgap
engineering of other III-V nanowires and increase the engineering possibilities with nanowires.


Acknowledgements
The authors would like to kindly thank T Garma and M Bichler for their experimental support.
This research was supported by Marie Curie Excellence Grant ‘SENFED’, and the DFG
excellence initiative Nanosystems Initiative Munich and SFB 631. This work was also supported
in part by the Swedish Foundation for Strategic Research (SSF), the Swedish Research Council
(VR) and the Knut and Alice Wallenberg Foundation (KAW). The authors would like to thank
the TEM facilities in the Serveis Cientificotècnics in Universitat de Barcelona.


φ
zb wz
GG
Δ−Δ
from

Page 11

Table 1. Growth conditions, growth rate and crystalline structure of samples α, β and γ .

Type
Arsenic beam flux
(mbar)
3.5x10-6
8.8x10-7
Growth rate (μ μm/h)
Crystalline structure
α 3 Zinc-blende
β 1
Zinc-blende with 30 ± 10%
wurtzite
γ
3.5x10-7 0.3
Wurtzite with 30± 10% zinc-
blende
  

Page 12

Figure 1.Zinc-blende/wurtzite quantum structures. (a) Aberration corrected HAADF High
Resolution Scanning Transmission Electron Micrograph (HRSTEM) of a zinc-blende quantum
well in a wurtzite segment. (b) An atomistic model of a wurtzite/zinc-blende/wurtzite
heterostructures along with a schematics of the band diagram. The Ga and As atoms have been
marked in orange and green, respectively, for the WZ domains, and in red and blue, respectively
for the double unit zinc-blende quantum well.
 
 
 
 
 

Page 13

Figure 2. Transmission Electron Microscopy of GaAs nanowires presenting different crystalline
structures. For illustration purposes, we have indicated each orientation of the zinc-blende phase
domains with blue and green circles, and the wurtzite phase regions with red circles. (a) General
view of the high beam flux nanowires (conditions α α) presenting twins in the zinc-blende
structure. (b) Detail of the twin zinc-blende domains. (c) HRTEM analysis of one twin interface.
(d) HRTEM detail of a nanowire synthesized under conditions β. Inset in (d) corresponds to an
aberration corrected HAADF HRSTEM image of the same region. (e) Magnified Cs-corrected
HAADF HRSTEM detail of the squared region in (d). A zinc-blende quantum well embedded
between two wurtzite regions. Twinned planes have been marked with dashed lines. Simulation
of the atomic positions has been superimposed on the measurement.

Page 14

 
 
 
 
 
Figure
type β.
bottom,
spots co
regions
 
 
 
3. Transmi
From a) to
, correspon
orrespond to
with zinc-b
ssion electr
o d), the se
ding respec
o regions w
blend structu
ron microgr
ctions belo
ctively to th
with wurtzite
ure –with th
aph from di
ng to differ
he latest to
e structure,
he two twin
ifferent segm
rent regions
o initial part
while the b
n orientation
ments of a s
s of the nan
t of the gro
blue and gre
ns-.
 
single GaAs
nowire from
owth proce
een spots co
s nanowires
m the tip to
ss. The red
orrespond to
s
o
d
o

Page 15

 
 
 
 
 
 
Figure 4. Transmission electron micrograph from GaAs nanowire grown under As BF of 3.5x10-
7 mbar –type γ γ. The small insets show closer view of the structure from the selected regions.