You are on page 1of 9

Self-assembled InN quantum dots on side facets of GaN nanowires

Zhaoxia Bi, Martin Ek, Tomas Stankevic, Jovana Colvin, Martin Hjort, David Lindgren, Filip Lenrick, Jonas
Johansson, L. Reine Wallenberg, Rainer Timm, Robert Feidenhans'l, Anders Mikkelsen, Magnus T. Borgström,
Anders Gustafsson, B. Jonas Ohlsson, Bo Monemar, and Lars Samuelson

Citation: Journal of Applied Physics 123, 164302 (2018); doi: 10.1063/1.5022756


View online: https://doi.org/10.1063/1.5022756
View Table of Contents: http://aip.scitation.org/toc/jap/123/16
Published by the American Institute of Physics

Articles you may be interested in


Perspective: Toward efficient GaN-based red light emitting diodes using europium doping
Journal of Applied Physics 123, 160901 (2018); 10.1063/1.5010762

Structure formation in Ag-X (X = Au, Cu) alloys synthesized far-from-equilibrium


Journal of Applied Physics 123, 165301 (2018); 10.1063/1.5018907

Guest Editorial: The dawn of gallium oxide microelectronics


Applied Physics Letters 112, 060401 (2018); 10.1063/1.5017845

Burying non-radiative defects in InGaN underlayer to increase InGaN/GaN quantum well efficiency
Applied Physics Letters 111, 262101 (2017); 10.1063/1.5007616

High In-content InGaN nano-pyramids: Tuning crystal homogeneity by optimized nucleation of GaN seeds
Journal of Applied Physics 123, 025102 (2018); 10.1063/1.5010237

Spatially dependent carrier dynamics in single InGaN/GaN core-shell microrod by time-resolved


cathodoluminescence
Applied Physics Letters 112, 052106 (2018); 10.1063/1.5009728
JOURNAL OF APPLIED PHYSICS 123, 164302 (2018)

Self-assembled InN quantum dots on side facets of GaN nanowires


Zhaoxia Bi,1 Martin Ek,2 Tomas Stankevic,3 Jovana Colvin,4 Martin Hjort,4 David Lindgren,1
Filip Lenrick,2 Jonas Johansson,1 L. Reine Wallenberg,2 Rainer Timm,4
Robert Feidenhans’l,3 Anders Mikkelsen,4 Magnus T. Borgstro € m,1 Anders Gustafsson,1
1,5 1 1,5,a)
B. Jonas Ohlsson, Bo Monemar, and Lars Samuelson
1
Division of Solid State Physics and NanoLund, Department of Physics, Lund University, Box 118,
S-221 00 Lund, Sweden
2
Center for Analysis and Synthesis/nCHREM, Lund University, Box 124, S-221 00 Lund, Sweden
3
Niels Bohr Institute, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen, Denmark
4
Division of Synchrotron Radiation Research and NanoLund, Department of Physics, Lund University,
Box 118, S-221 00 Lund, Sweden
5
QuNano AB, Scheelev€ agen 22, S-223 63 Lund, Sweden
(Received 18 January 2018; accepted 1 April 2018; published online 25 April 2018)
Self-assembled, atomic diffusion controlled growth of InN quantum dots was realized on the side
facets of dislocation-free and c-oriented GaN nanowires having a hexagonal cross-section. The nano-
wires were synthesized by selective area metal organic vapor phase epitaxy. A 3 Å thick InN wetting
layer was observed after growth, on top of which the InN quantum dots formed, indicating self-
assembly in the Stranski-Krastanow growth mode. We found that the InN quantum dots can be tuned
to nucleate either preferentially at the edges between GaN nanowire side facets, or directly on the
side facets by tuning the adatom migration by controlling the precursor supersaturation and growth
temperature. Structural characterization by transmission electron microscopy and reciprocal space
mapping show that the InN quantum dots are close to be fully relaxed (residual strain below 1%) and
that the c-planes of the InN quantum dots are tilted with respect to the GaN core. The strain relaxes
mainly by the formation of misfit dislocations, observed with a periodicity of 3.2 nm at the InN and
GaN hetero-interface. The misfit dislocations introduce I1 type stacking faults (…ABABCBC…) in
the InN quantum dots. Photoluminescence investigations of the InN quantum dots show that the
emissions shift to higher energy with reduced quantum dot size, which we attribute to increased
quantum confinement. Published by AIP Publishing. https://doi.org/10.1063/1.5022756

I. INTRODUCTION confinement of the QDs can prevent injected carriers from


recombining via non-radiative defects, which helps improving
A semiconductor quantum dot (QD), with a feature size
the performance of LEDs.9 Furthermore, InN QDs have been
close to or smaller than the de Broglie wavelength of electrons
found interesting for third generation solar cells based on the
has full confinement of charge carriers in all three spatial
concepts of intermediate band solar cells and multiple exciton
dimensions, leading to a discrete density-of-states.1–3 This
generation.16
makes QDs important building blocks for optoelectronic devi-
Nitride QDs have been studied extensively in recent
ces, such as light emitting diodes (LEDs),4–6 laser diodes
years. InGaN and InN QDs have mostly been grown on pla-
(LDs),7–9 and single photon emitters.10–12 For example, by
nar GaN films which commonly suffer from a high density
replacing the quantum well (QW) used in the active layer today
of dislocations (108 cm2) due to heteroepitaxial growth on
with QDs, lasers can benefit from lower threshold current,
non-native and lattice mismatched substrates.4,11–13,17 The
higher temperature stability and radiation resistance.1,2,4,12
InN QD growth on such GaN films is strongly affected by
Nitride semiconductors (GaN, InN, AlN, and their alloys) offer
the dislocations which tend to initiate the QD nucleation and
direct band gaps covering the optical spectrum from infrared
propagate into the QDs.17,18 Currently, bulk GaN substrates
(InN: 0.7 eV) to ultraviolet (AlN: 6.0 eV), and have a large
are available, but the dislocation density is still around
range of lattice constant difference (e.g., InN to GaN: 10.8%
106 cm2.19,20 Dislocation formation can be totally avoided
and to AlN: 13.5%).13,14 This makes them interesting for study-
in the growth of GaN nanowires (NWs),21–26 which makes
ing strain-induced QD growth for related applications in opto-
the GaN NWs a perfect template to study the growth of
electronics.15 Due to the nanometer dimensions and three
dimensional (3D) characteristics, QDs are able to accommo- self-assembled InN QDs, only induced by strain while not
date strain more efficiently than QWs. This makes InGaN QDs by dislocations. h0001i-oriented GaN NWs are formed with
interesting to use in the active layers, especially for long wave- non-polar f1100g m-planes as the side facets. Such m-plane
length green and red nitride LEDs where plastic strain relaxa- side facets are atomically flat, in contrast to the striation mor-
tion typically occurs for planar InGaN QWs. The carrier phology of planar m-plane GaN films.27 Furthermore, the
edges between NW side facets can be energetically favorable
for QD nucleation due to effective strain release as compared
a)
Author to whom correspondence should be addressed: Lars.Samuelson@ to the flat NW facets, providing a way to control the position
ftf.lth.se of the QDs.28,29

0021-8979/2018/123(16)/164302/8/$30.00 123, 164302-1 Published by AIP Publishing.


164302-2 Bi et al. J. Appl. Phys. 123, 164302 (2018)

In this work, we report on binary InN QDs grown on the measurement, the incident photon energy was 70 eV, giving
side facets of dislocation-free GaN NWs. Synthesis was car- the photoelectrons a kinetic energy of 44.5 eV and 46.0 eV for
ried out by metal-organic vapor phase epitaxy (MOVPE). Ga and In, respectively, and therefore, resulting in a very high
An InN wetting layer was observed, on top of which the surface sensitivity. The energy window used for imaging was
InN QDs formed, indicating self-assembly in the Stranski- 0.5 eV.
Krastanow (SK) growth mode.30 By controlling the adatom Structural properties of the InN QDs were characterized
migration, the InN QDs can be controlled to nucleate either by transmission electron microscopy (TEM) and reciprocal
only at the edges between m-plane side facets or on the side space mapping (RSM). Specimens were prepared for TEM
facets as well. We find that the strain in the InN QDs is by gently rubbing a Cu grid with a lacey carbon film over the
relaxed by the formation of a periodic array of mis-fit dislo- substrate to break off and transfer the NWs. For high resolu-
cations, originating at the hetero interface, and by a tilting tion imaging of the interface between the InN QDs and the
of c-planes of the InN QDs with respect to the GaN NW. GaN NW in a h11–20i orientation (beam parallel to the side
Optical properties of these plastically relaxed QDs studied facet), an additional sample was prepared by focused ion
by photoluminescence (PL) reveal that the emission blue beam (FIB) after first spin-coating the substrate with a pro-
shifts as the QD size is reduced, which we interpret as a tective polymer layer.32 This was performed in order to fab-
result of carrier quantum confinement. ricate NWs that were thin enough for lattice imaging. Both
types of samples were analyzed using a JEOL 3000F micro-
II. EXPERIMENTAL DETAILS scope operated at 300 kV. 3D RSM was recorded at the I811
Samples of InN QDs on GaN NWs were grown by beam line at the MAX-IV laboratory using the 12.4 keV
MOVPE, using a reactor equipped with a 3  2 in. close synchrotron radiation in order to determine the crystal struc-
coupled showerhead. We used selective area growth to grow ture of the QDs.33 Measurements were done by illuminating
the GaN NWs. A layer of SiN with a thickness of a few tens around 105–106 NWs with a beam size of about 0.5  1 mm.
The position and the shape of the (0002) Bragg peaks were
of nanometers, fabricated by low-pressure chemical vapor
deposition, was used as a growth mask on a substrate of analyzed in order to determine strain and lattice tilt in the
GaN/Si. Electron-beam lithography and reactive ion etching InN QDs.
Optical properties were evaluated using PL and cathodo-
were used to pattern the SiN mask with hexagonal arrays of
luminescence (CL) spectroscopy at liquid He temperatures
100 nm large openings. The patterned templates were then
of 4 K and 8 K, respectively. The PL was performed using a
transferred into the MOVPE reactor. We used continuously
532 nm Nd-YAG laser and the emission was detected by an
supplied flows of NH3 and triethylgallium (TEG) to synthe-
InGaAs CCD. In order to avoid the luminescence contribu-
size GaN NWs. A low V/III ratio of 11 was used with a TEG
tion from the substrate, the NWs were scraped off from the
flow of 19 lmol/min and the temperature was 1042  C. The
substrate and deposited on an Au-coated Si (100) substrate.
as-grown GaN NWs have a hexagonal cross-section with a
A 1000 nm long pass filter was used in order to block any
diameter of about 180 nm, as observed by scanning electron
higher order emission from the GaN band-edge and yellow
microscopy (SEM). After GaN NW growth, the V/III ratio
band. CL spectroscopy and imaging were performed in a
was increased to 7000 by increasing the NH3 flow in order
dedicated SEM at 5 kV, with a GaInAs photomultiplier tube
to grow a GaN shell around the GaN core NWs, increasing
for signal detection.
the width of the NW side facets. The final diagonal diameter
of the GaN NWs was about 400 nm, with 200 nm wide
III. RESULTS AND DISCUSSION
m-plane side facets. Then, the growth temperature was low-
ered to 600–730  C in order to grow InN QDs. For the InN Figure 1 is an SEM image illustrating the uniformity of
QD growth, the TMI and NH3 flows were in the ranges of GaN NWs standing vertically on the GaN/Si substrate. The
6–33 lmol/min and 100–300 mmol/min, respectively. NWs have a hexagonal cross-section and a diameter of about
The morphology of the InN QDs on GaN NWs was stud- 180 nm and a length of 2.4 lm. The SiN mask and the small
ied with SEM and atomic force microscopy (AFM). SEM footprint of the GaN NWs effectively filter out the dislocations
images were recorded using a FEI Nanolab 600 FIB/SEM sys- from the underlying GaN buffer layer, rendering the GaN
tem. AFM images were obtained using a Nanowizard AFM NWs essentially free from dislocations. The GaN NWs are
from JPK Instruments in the intermittent contact mode. For grown along the h0001i direction and have six equivalent side
this purpose, GaN NWs with InN QDs were redeposited onto facets of non-polar f1010g m-planes. Around such GaN core
a Si substrate, lying on one m-plane side facet. Only the top- NWs, an epitaxial shell of GaN, grown by raising the V/III
most m-plane facet and parts of the adjacent side facets of the ratio, inherits the uniformity and crystal perfection of the as-
NWs were scanned by AFM. grown GaN NWs. This shell is a way to increase the side facet
Photoemission electron microscopy (PEEM) was used to width (NW diameter) for the subsequent InN QD growth,
confirm the existence of the InN wetting layer by spatially while preserving the configuration of the side facets of the
resolved elemental maps of the NW surface.31 The microscope NW cores. The inset in Fig. 1 shows a side view SEM image
was located at beam line I311 at the MAX-IV Laboratory in of a NW with a GaN shell and illustrates the smooth surface
Lund, Sweden. In PEEM, synchrotron x-rays are incident of the side facets. A pyramid delimited by six f1011g planes
on the sample and photoelectrons originating from elemental is observed at the NW top, which is common for GaN NW
specific core-levels are used to form an image. During the growth due to the slow growth rate of the f1011g planes.34
164302-3 Bi et al. J. Appl. Phys. 123, 164302 (2018)

The SEM images show that no growth features are visible


until 30 s after the initiation of the growth. After 30 s growth,
InN QDs were observed to form at the NW edges between
the side facets after interrupting synthesis and cooling down
(top-view SEM image). With prolonged growth, InN QD for-
mation proceeds at the edges while both the QD size and
density increase. After a growth time of 180 s, we observe
that the InN QDs have started to coalesce as shown by the
side-view SEM image. AFM was used to characterize the
InN QDs grown for 38 s and Fig. 3(a) shows a 3D AFM
image obtained by scanning along the NW. The AFM scan
on the top side facet [Figs. 3(b) and 3(c)] verifies that QDs
formed only at the edges between adjacent side facets, while
otherwise a fully homogeneous and flat facet with a surface
FIG. 1. Arrays of dislocation-free GaN NWs. 30 tilted view SEM image of
corrugation of less than 1 nm is observed. The f1011g planes
as-grown GaN NW arrays with a diameter of 180 nm and a length of 2.4 lm. at the top end of the GaN NWs are flat and without any
A GaN shell was grown around the GaN NWs to enlarge the NW side facets traces of QD formation. The sample grown for 22 s was also
for subsequent InN QD growth. The inset is a side view image of a single measured by AFM and no InN QD was observed (see Fig.
NW after the shell growth, showing the smooth m-plane side facets. This
NW was scraped down and lay on the substrate surface for side view S1 in the supplementary material). The preferential nucle-
imaging. ation of the InN QDs at the edges could be attributed to sur-
face stress on the GaN NW side facets, which leads to a
For the InN QD growth, the GaN NWs used as templates were higher free energy at the edges than on the side facets.35
grown to a length of 2.4 lm and a diagonal diameter of Figure 2(b) shows the dependence of the InN QD den-
400 nm (200 nm wide side facets), as shown in the inset of sity (QD number per NW) on the growth time. We used
Fig. 1. side-view SEM images as shown in Fig. 2(a) to count the
Figure 2 shows the SEM images of InN QD growth QDs at the two outer edges of GaN NWs and then multiply it
for different lengths of time, from 22 s to 180 s. Side-view by 3 to obtain the QD number per NW. The InN QD density
and top view SEM images of individual NWs are shown in is increased rapidly from 0 to about 100 as the growth
Fig. 2(a). The side-view images were recorded after the GaN time increased from 22 s to 60 s [Fig. 2(b)]. Such a tendency
NWs were scraped off, lying down at the substrate surface. agrees with the typical growth evolution of SK QDs.36,37

FIG. 2. Growth time dependence of


InN QDs on GaN NWs. (a), Side view
and top view SEM images of GaN NWs
with InN growth for different times.
Scale bar: 500 nm for side view images
and 250 nm for top view images. (b),
Variation of InN QD density versus the
growth time. (c), Dependence of InN
QD size on the growth time in three
dimensions. The insets show how the
sizes were measured. (d), Variation of
total InN volume per NW versus the
growth time. The InN volume was cal-
culated based on the QD density in (b)
and QD size in (c).
164302-4 Bi et al. J. Appl. Phys. 123, 164302 (2018)

FIG. 3. AFM characterization of InN QDs on GaN NWs. (a), 3D plot of an AFM image showing a GaN NW lying on a Si substrate. (b) and (c) are height and
phase AFM images of the top side facet, illustrating that the QDs nucleate preferentially at the edges between side facets.

As a result of the NW geometry, the InN QD size can be to 680  C, the QD density is reduced while the QD height
measured in all three dimensions based on the top and side- is increased [Fig. 5(a)]. When the temperature is raised to
view SEM images, as shown by the insets in Fig. 2(c). The 730  C, no InN QD could be grown, most probably due
results are shown in Fig. 2(c). The QDs grown for 30 s have to decomposition and desorption of InN. The TMI flow rate
a height of about 5 nm and the height increased monoto- affects the InN QD growth significantly, as shown in Fig.
nously to about 20 nm for the longest growth time of 180 s. 5(b). With increased TMI flow rate, the QD density increases
Both base sizes [? and //: refer to Fig. 2(c) for definitions] of while the QD height decreases. Note that the InN QDs are
the QDs are about 3–4 times larger than the height and found to mainly grow at the edges for all samples in Figs. 5(a)
increases with the growth time. The base size ? seems to be and 5(b). Seifert et al. reported an empirical model for the SK
larger than the base size // as shown in Fig. 2(c). Note that QD growth,38 where the QD density depends on the ratio DR
the base size ? may be overestimated as it is measured via (R: growth rate and D: diffusion constant). The density results
the projection of ensembles of QDs along each edge, rather presented in Fig. 5(b) agree with this model as the TMI flow
than on individual InN QDs. Based on the QD size in Fig. rate determines the  2growth
E
rate. The diffusion constant is
 kTD
2(c), the total InN volume per NW was calculated and the given by D ¼ 2kT h a e , where k is Boltzmann’s constant,
results are shown in Fig. 2(d). The InN volume increases lin- T is the absolute temperature, h is Planck’s constant, a is the
early with growth time, and the straight line in Fig. 2(d) indi- lattice parameter, and ED is the energy barrier for surface dif-
cates a growth rate of 3:4  104 nm3/s at the given growth fusion. This shows that raising the growth temperature leads
parameters. The extrapolation to zero QD volume occurs at a
finite growth time (28 s), which indicates the formation of
an InN wetting layer, followed by breaking up of a critically
strained layer and self-assembly of QDs.
Figures 4(a) and 4(b) show a pair of PEEM images of
the sample grown for 22 s, recorded with photoelectrons
from the core levels of Ga 3d and In 4d, respectively. The
NWs in both images are bright, proving that indium is pre-
sent at the NW surface together with gallium. The dark back-
ground observed between the NWs is the Si substrate onto
which the NWs were transferred. The sample with the InN
QD formation at the edges was analyzed by use of PEEM
[Figs. 4(c) and 4(d)]. In both images, there are signals from
Ga 3d and In 4d on the side facets, confirming that a
wetting layer of InN remains on the NW facets after the QD
formation. Apart from microscopy, these two samples were
measured by synchrotron-based x-ray photoelectron micro-
spectroscopy in the PEEM and the data also confirms the
existence of the InN wetting layer after QD formation (see
Fig. S2-S3 in the supplementary material). Therefore, PEEM
characterizations clearly indicate that the InN QDs form in
the SK growth mode, excluding the nucleation by the dislo-
FIG. 4. PEEM characterizations on GaN NWs after InN growth. PEEM
cations as reported in Refs. 17 and 18. Based on the Y-axis images obtained on samples with InN growth times of 22 s [(a) and (b)
intercept in Fig. 2(d) and GaN NW dimensions, the wetting before QD formation] and 38 s [(c) and (d) after QD formation], using Ga
layer thickness is estimated to be 3 Å. 3d photoelectrons [(a) and (c)] and In 4d photoelectrons [(b) and (d)]. Even
though the brightness between NWs in (c) and (d) is not identical, like the
Figures 5(a) and 5(b) show the QD density and height ones in (a) and (b), the presence of indium on the surface of side facets can
dependences on growth temperature and TMI flow, respec- be concluded for both samples, showing that the InN QDs form with an InN
tively. With increasing growth temperature from 630  C wetting layer.
164302-5 Bi et al. J. Appl. Phys. 123, 164302 (2018)

FIG. 5. Impact of growth temperature and TMI flow on the InN QD growth. Density and QD height dependences on the growth temperature in (a) and the
TMI flow in (b). By lowering the growth temperature or increasing the TMI flow, InN QDs grow on the m-plane side facets as shown by an SEM image in (c).
The inset shows the magnified area marked by the green square.

to an increase in the diffusion constant D, which according to In order to reduce the thickness of the NW core to
the model, should result in a decrease in the QD density. This enable high resolution imaging of the QD-NW interface, an
is consistent with the results in Fig. 5(a). The injected TMI additional TEM specimen with about 26 nm high QDs was pre-
amounts (TMI flow rate multiplied by growth time) are equal pared by FIB milling where the GaN core was reduced to a
for the samples in Figs. 5(a) and 5(b). This explains why the similar thickness as the QD. An illustration of the position and
QD height always shows the opposite tendency to the density orientation of the FIB lamella is shown in Fig. 7(a). Due to the
in both figures. In accordance with the discussion above, the decreased thickness of the NW core, it is possible to determine
InN QD growth is controlled by adatom or precursor diffusion that each stacking fault occurs at an MD where a single c-plane
on the side facets of GaN NWs. The preferential QD growth is missing in the QD compared to the NW. Similar MD-SF
at the edges indicates that the diffusion length is larger than arrangements have been observed to be involved in the strain
the half width of the side facets, about 100 nm. Consequently, relaxation in InGaN QWs grown on m-plane GaN.41,42
QD growth on the side facets should be possible by reducing
the diffusion length to less than 100 nm. Figure 5(c) shows
how InN QD growth on the side facets can be achieved
by lowering the growth temperature to 630  C under a high
TMI flow rate of 600 sccm. The QD growth on the side facets
is elongated along the a-axis h1120i, and the base size is
15 6 3 nm and 56 6 9 nm. Such an elongation was previously
reported for InGaN QD growth on m-plane GaN substrates39
and may be related to the anisotropic lattice mismatch (10.8%
along the a-axis versus 9.8% along the c-axis for InN to GaN).
Samples with about 5 and 26 nm high QDs at the edges
were investigated by TEM (Fig. 6). The QDs in both samples
contained regular arrays of I1 type stacking faults [Fig. 6(c)],
i.e., single bilayers with cubic stacking inserted into the sur-
rounding wurtzite structure (…ABABCBC…).40 These are
clearly visible when viewing the QDs in a h1120i direction.
In this orientation, the NWs are imaged with one side facet
parallel to the beam, which means that the NW-QD interface
could not be imaged due to the thickness of the GaN core. In
the h1 100i direction (across an edge), the interface of the QD
along the edge of the NW has a much smaller projected thick-
ness, which allows high resolution imaging. Moire fringes are
formed where the QD and NW lattices overlap due to the dif-
FIG. 6. TEM characterizations directly on the as-grown samples. TEM
ference in the lattice constant. The Moire pattern in the lower images of QDs with heights of about 26 nm (a) and (b) and 5 nm (c) and (d),
part of the QD is tilted due to the downward tilt of InN c- viewed in h11 20i (a) and (c) and h1 100i (b) and (d) directions. The black
planes, in contrast to the Moire pattern parallel to the NW lat- arrow in (a) indicates the growth direction of the GaN NW cores in all four
tice in the top part of the QD. In this orientation, where there images. In the h11 20i viewing direction, regularly spaced I1 stacking faults
are indicated by white arrows. A single stacking fault is shown in higher
is one c-oriented bilayer missing in the QDs compared to the magnification in the inset in (c). In the h1100i viewing direction, the simi-
NW, misfit dislocations (MDs) can be seen at the interface larly spaced MDs are visible instead, as indicated in (d) by mark “>.” In (b),
with an average spacing of 3.2 nm. The observed plane spac- the strong Moire fringes make it difficult to see the MDs. The black guide-
lines in (b) and (d) indicate the (0001) planes in the NW cores and the QDs,
ings and Moire fringes indicate that the InN QDs are close
showing the downward tilt in the lower part of the QDs. This is also seen
to fully relaxed (residual strain below 1%), with the MDs from the Moire pattern in (b), which is parallel to the NW lattice in the top
accounting for 90% of the strain accommodation. part but clearly titled in the lower part (white guidelines).
164302-6 Bi et al. J. Appl. Phys. 123, 164302 (2018)

FIG. 7. TEM characterizations on the FIB-prepared lamella. Schematic draw- FIG. 8. Reciprocal space maps of the (0002) Bragg peaks from the GaN
ing (a) of the FIB lamella prepared from the sample with about 26 nm high NWs and InN QDs. Panels (a)–(c) show the out-of-plane RSMs, where both
QDs together with a TEM overview image of a single QD (b). High resolution GaN and InN Bragg peaks are visible. The InN peak seems to split into two
image of the QD-NW interface (c) from the area marked with a rectangle in b. lobes due to tilt in the QDs. More details about the tilt are visible in panels
In this image, the MDs (marked with >) can be more specifically located to the (d)–(f), which show an in-plane cut through the InN peaks. Multiple peaks
interface of the I1 stacking fault (marked with white arrows) in the QD and the in the InN signal indicate 12-fold (d) and (e) or 6-fold (f) symmetry in cases
NW. The growth direction of the NW core is shown by a black arrow in (c). of the QDs grown at the edges and the side facets, respectively.

Structural properties of InN QDs were also measured from 0.55% to 0.77% for the top part and from 0.16%
with 3D RSM. Figure 8 shows two different cross-sections to 0.26% for the lower part, respectively. The InN QDs
of the (0002) Bragg peaks measured on three different sam- grown on the m-plane side facets show larger residual strain
ples. The first two samples [Figs. 8(a), 8(d) and 8(b), 8(e)] than the ones at the edges with similar size.
had QDs grown mainly at the edges of the NWs and the The detailed pattern of tilts of the QDs grown at the
height of the QDs was 25 and 11 nm, respectively. The third edges and side facets can be seen from the QxQy RSMs [Figs.
sample [Figs. 8(c) and 8(f)] had 10 nm-high QDs grown 8(d)–8(f)] cutting through the InN peaks at Qz ¼ 2.21 Å1.
mainly on the side facets. Panels (a)–(c) show the QxQz Here, we observe that the InN signal actually consists of
cross-section of the (0002) peak, where Qz is parallel to the twelve distinct InN peaks for the samples with QDs grown at
NW growth direction and Qx is perpendicular to the NW the edges (d) and (e) and six InN peaks for the sample with
side facets. We can see strong GaN peaks at higher Qz values QDs grown on the side facets (f). The appearance of such a
and weaker InN signals below. For the relaxed epitaxial InN diffraction pattern can be explained by the symmetry of the
material, we would expect to see the InN peak aligned verti- NWs. For the QDs grown on side facets, each QD is tilted
cally with the GaN peak at Qz ¼ 2.203 Å1. However, the towards one of the six side facets. Therefore, in total, there
InN peak splits into two lobes along Qx and only weak inten- are six unique crystal tilts [Figs. 9(b) and 9(d)]. However, for
sity remains at Qx ¼ 0. This splitting can be explained by the the QDs grown at the edges, each QD has an interface with
symmetric tilt of the QD lattice with respect to the GaN two side facets. Therefore, the QD splits into two parts, tilted
core. The tilt magnitudes are 2.6 , 2.5 , and 2.2 for the 25, towards the neighboring side facets, as shown in Figs. 9(a)
11, and 10 nm QDs, respectively. The lattice tilt can also be and 9(c). As a result, there are 12 unique tilt directions, giving
seen directly in the TEM images, e.g., as a bending of the c- rise to 12 peaks in the InN signal.
planes downwards in the lower part of the QD in Fig. 6(d). Lattice tilt is a common phenomenon in mismatched het-
It should be noted that the thin, top part of each QD is erostructures, including SK QDs,43 which reduces the strain
much less tilted than the thicker, lower part. These non-tilted energy.44–48 A possible mechanism favoring this tilt is the
tips of the QDs contribute to the intensity in the Bragg peak asymmetric shape of the QDs. Since the QDs have a wedge
which corresponds to the non-tilted crystal lattice and is shape at the top part and a free c-plane at the lower part, tilts
shown in a red box in Fig. 8(a). The high sensitivity of the toward the mode free side result in a more efficient elastic
RSM measurements additionally allowed the small residual strain relaxation. Additionally, the 1/3a or 2/3a steps created
strains in the InN QDs to be analyzed (see Fig. S4 in the at the interface by the periodic stacking faults observed in the
supplementary material). The lower part of the QDs with TEM images (Fig. 6) can contribute to the tilt: a 3.2 nm stack-
tilted Moire pattern is more relaxed than the top part with ing fault period and 1/3a steps will e.g., result in a tilt of
parallel Moire pattern. With the QD height reduced from about 1.9 . We believe that these two effects are the main
25 nm to 11 nm, the compressive strain in the QDs increases reason for the observed lattice tilts in the QDs.
164302-7 Bi et al. J. Appl. Phys. 123, 164302 (2018)

all without any capping layer. The reference spectrum is


from the GaN NW templates and the QD height of 0 means
that the sample only has the InN wetting layer. The sample
with a QD height of 0 shows the same luminescence as the
GaN NW templates, originating from the second order of
the GaN NW yellow luminescence band. Despite the high
density of MDs at the base of the QDs with the height of
5.0–19.9 nm, it is possible to detect InN-related emission
from the structures. The shape, linewidth, and peak positions
of the PL spectra are similar to previous reports on InN
QDs.49,50 The inset shows a clear blue shift of the peak emis-
sions with reduced QD height, in agreement with the quan-
tum confinement. The samples show emissions in the range
from about 0.8 eV to the 1.2 eV of the cut-off long-pass filter.
With reduced QD size, the linewidth of the emission
FIG. 9. Tilt of the QDs in two types of samples. (a) and (c) QDs are split in increases from 100 to 340 meV. We attribute the broadening
half, with each part tilting towards the neighboring NW facet. (b) and (d)
QDs tilt around a horizontal axis towards the NW side facet. of the emission to an increase in surface states, due to the
increase in the surface area to volume ratio of the QDs. This
In order to study the optical properties, we used PL leads to an increased electron accumulation in the conduc-
spectroscopy. Figure 10 shows the PL spectra of ensembles tion band and potentially a Burstein-Moss shift.51,52 This is
of single NWs for eight samples including GaN NW tem- less likely in our case, as the high-energy side of the emis-
plates and InN QDs with different QD heights of 0–19.9 nm, sion does not show the exponential decrease at the Fermi-
edge, normally associated with the Burstein-Moss effect.49
The NWs were also studied by CL spectroscopy. The emis-
sion was too weak to do any monochromatic imaging of the
QD emission. For the CL investigation, some of the struc-
tures were capped by a thin GaN layer in order to protect the
InN surface from the electron beam of the CL setup. This
did not improve the emission from the QDs, but the NWs
showed emission at 3.2 eV. This peak also showed up in PL
investigations. Monochromatic images reveal that this emis-
sion originates very close to the surface of the NW side fac-
ets, consistent with emission from the thin wetting layer.53

IV. CONCLUSION
In this work, we demonstrate the growth of InN QDs on
the side facets of GaN NWs by selective area MOVPE, com-
bining zero and one-dimensional structures for nitride opto-
electronics. Self-assembled InN QDs were grown in the SK
growth mode, as evidenced by the presence of a wetting layer
in combination with the QDs. By controlling the adatom
migration length, the preferential nucleation at the NW edges
between the side facets can be suppressed and InN QD growth
on the side facets of the GaN NWs can take place. InN QDs
are fully relaxed even for the smallest size (5 nm high) via
misfit dislocations at the interface between InN QDs and GaN
NWs. Each misfit dislocation seeded an I1 type stacking fault
in the InN QDs. The QD size needs to be reduced further to
eliminate the dislocation in the QDs. Optical measurements
show a significant blue shift of the PL emission with
decreased QD size, consistent with the quantum size effect.

FIG. 10. lPL measurements at 4 K on GaN NW templates and InN QDs SUPPLEMENTARY MATERIAL
with different QD heights. The reference spectrum is from the GaN NW
templates and the sample with the QD height of 0 only has the InN wetting See supplementary material for the AFM data on the sam-
layer. The emission tails for both samples at the energy lower than 1.2 eV
ple grown for 22 s, the data measured by synchrotron-based
are from the second order of the GaN NW yellow luminescence band. The
inset shows PL peak energies of InN QDs at different QD heights, showing x-ray photoelectron microspectroscopy, and residual strain
increased transition energy with reduced QD size. analysis in the InN QDs based on the RSM measurement.
164302-8 Bi et al. J. Appl. Phys. 123, 164302 (2018)

25
ACKNOWLEDGMENTS S. F. Li, S. Fuendling, X. Wang, S. Merzsch, M. A. M. Al-Suleiman, J. D.
Wei, H. H. Wehmann, and A. Waag, Cryst. Growth Des. 11, 1573–1577
The authors wish to acknowledge the support from the (2011).
26
European Union under the project “NWs4LIGHT” (Grant X. J. Chen, G. Perillat-Merceroz, D. Sam-Giao, C. Durand, and J. Eymery,
Appl. Phys. Lett. 97, 151909 (2010).
No. 280773) and from the Swedish Foundation for Strategic 27
Q. Sun, S. Y. Kwon, Z. Ren, J. Han, T. Onuma, S. F. Chichibu, and S.
Research under the project “Energieffektiv LED-belysning Wang, Appl. Phys. Lett. 92, 051112 (2008).
28
Baserad på Nanotrådar” (Grant No. EM11-0015). This work S. Jeppesen, M. S. Miller, D. Hessman, B. Kowalski, I. Maximov, and L.
was also supported by the Swedish Research Council (VR), Samuelson, Appl. Phys. Lett. 68, 2228–2230 (1996).
29
G. Jin, J. L. Liu, S. G. Thomas, Y. H. Luo, K. L. Wang, and B. Y. Nguyen,
Knut and Alice Wallenberg’s Foundation (KAW), and Appl. Phys. Lett. 75, 2752–2754 (1999).
VINNOVA. 30
R. N€otzel, Semicond. Sci. Technol. 11, 1365–1379 (1996).
31
M. Hjort, J. Wallentin, R. Timm, A. A. Zakharov, U. Håkanson, J. N.
1 Andersen, E. Lundgren, L. Samuelson, M. T. Borgstr€ om, and A.
Y. Arakawa and H. Sakaki, App. Phys. Lett. 40, 939–941 (1982).
2 Mikkelsen, ACS Nano 6, 9679–9689 (2012).
M. Asada, Y. Miyamoto, and Y. Suematsu, IEEE J. Quantum Electron.
32
QE-22, 1915–1921 (1986). F. Lenrick, M. Ek, D. Jacobsson, M. T. Borgstr€ om, and L. R. Wallenberg,
3 Microsc. Microanal. 20, 133 (2014).
P. Michler, Nanoscience and Technology: Single Semiconductor Quantum
33
Dots (Springer-Verlag, Berlin Heidelberg, 2009), Chap. 2. T. Stankevič, S. Mickevičius, M. Schou Nielsen, O. Kryliouk, R.
4 Ciechonski, G. Vescovi, Z. Bi, A. Mikkelsen, L. Samuelson, C. Gundlach,
J. Kalden, C. Tessarek, K. Sebald, S. Figge, C. Kruse, D. Hommel, and J.
Gutowski, Nanotechnology 21, 015204 (2010). and R. Feidenhans’l, J. Appl. Crystallogr. 48, 344 (2015).
5 34
I. K. Park, M. K. Kwon, S. B. Seo, J. Y. Kim, and J. H. Lim, Appl. Phys. D. Du, D. J. Srolovitz, M. E. Coltrin, and C. C. Mitchell, Phys. Rev. Lett.
Lett. 90, 111116 (2007). 95, 155503 (2005).
6 35
T. Xu, A. Y. Nikiforov, R. France, C. Thomidis, A. Williams, and T. D. V. Consonni, M. Hanke, M. Knelangen, L. Geelhaar, A. Trampert, and H.
Moustakas, Phys. Status Solidi A 204, 2098–2102 (2007). Riechert, Phys. Rev. B 83, 035310 (2011).
7 36
N. Kirstaedter, N. N. Ledentsov, M. Grundmann, D. Bimberg, V. M. D. Leonard, K. Pond, and P. M. Petroff, Phys. Rev. B 50, 11687–11692
Ustinov, S. S. Ruvimov, M. V. Maximov, P. S. Kop’ev, Z. I. Alferov, U. (1994).
37
Richter, P. Werner, U. G€ osele, and J. Heydenreich, Electron. Lett. 30, H. T. Dobbs, D. D. Vvedensky, A. Zangwill, J. Johansson, N. Carlsson,
1416–1417 (1994). and W. Seifert, Phys. Rev. Lett. 79, 897–900 (1997).
8 38
S. Tanaka, H. Hirayama, Y. Aoyagi, Y. Narukawa, Y. Kawakami, S. W. Seifert, J. Johansson, and N. Carlsson, Jpn, J. Appl. Phys., Part 1 38,
Fujita, and S. Fujita, Appl. Phys. Lett. 71, 1299–1301 (1997). 7264–7267 (1999).
9 39
K. Tachibana, T. Someya, and Y. Arakawa, IEEE J. Sel. Top. Quantum X. Yang, M. Arita, S. Kako, and Y. Arakawa, Appl. Phys. Lett. 99,
Electron. 6, 475–481 (2000). 061914 (2011).
10 40
Z. Yuan, B. E. Kardynal, R. M. Stevenson, A. J. Shields, C. J. Lobo, K. J. P. Hirth and J. Lothe, Theory of Dislocations, 2nd ed. (Wiley
Cooper, N. S. Beattie, D. A. Ritchie, and M. Pepper, Science 295, InterScience, New York, 1982), p. 354.
41
102–105 (2002). A. M. Fischer, Z. Wu, K. Sun, Q. Wei, Y. Huang, R. Senda, D. Iida, M.
11
S. Kremling, C. Tessarek, H. Dartsch, S. Figge, and S. H€ofling, Appl. Iwaya, H. Amano, and F. A. Ponce, Appl. Phys. Express 2, 041002 (2009).
42
Phys. Lett. 100, 061115 (2012). F. Wu, Y. D. Lin, A. Chakraborty, H. Ohta, S. P. DenBaars, S. Nakamura,
12
C. W. Hsu, A. Lundskog, K. F. Karlsson, U. Forsberg, E. Janzen, and P. and J. S. Speck, Appl. Phys. Lett. 96, 231912 (2010).
43
O. Holtz, Nano Lett. 11, 2415–2418 (2011). A. Carlsson, L. R. Wallenberg, C. Persson, and W. Seifert, Surf. Sci. 406,
13
C. J. Humphreys, MRS Bull. 33, 459–470 (2008). 48–56 (1998).
14 44
E. Silveira, J. A. Freitas, O. J. Glembocki, G. A. Slack, and L. J. S. Yoshida, Y. Yokogawa, Y. Imai, S. Kimura, and O. Sakata, Appl. Phys.
Schowalter, Phys. Rev. B. 71, 041201(R) (2005). Lett. 99, 131909 (2011).
15 45
D. J. Eaglesham and M. Cerullo, Phys. Rev. Lett. 64, 1943–1946 (1990). S. Y. Woo, G. A. Devenyi, S. Ghanad-Tavakoli, R. N. Kleiman, J. S.
16
R. E. Welser, A. K. Sood, Y. R. Puri, O. A. Laboutin, L. J. Guido, N. K. Preston, and G. A. Botton, Appl. Phys. Lett. 102, 132103 (2013).
46
Dhar, and P. S. Wijewarnasuriya, Proc. SPIE 7683, 76830R (2010). F. Riesz, Vacuum 46, 1021–1023 (1995).
17 47
O. A. Laboutin and R. E. Welser, Appl. Phys. Lett. 92, 223103 (2008). S. Yamaguchi, M. Kariya, S. Nitta, T. Takeuchi, C. Wetzel, H. Amano,
18
J. G. Lozano, A. M. Sanchez, R. Garcia, D. Gonzalez, D. Araujo, S. and I. Akasaki, J. Appl. Phys. 85, 7682–7688 (1999).
48
Ruffenach, and O. Briot, App. Phys. Lett. 87, 263104 (2005). R. Du and C. P. Flynn, J. Phys.: Condens. Matter 2, 1335–1341 (1990).
19 49
K. Nomoto, B. Song, Z. Hu, M. Zhu, M. Qi, N. Kaneda, T. Mishima, T. P. E. D. S. Rodriguez, V. J. Gomez, P. Kumar, E. Calleja, and R. Notzel,
Nakamura, D. Jena, and H. G. Xing, IEEE Electron Device Lett. 37, Appl. Phys. Lett. 102, 131909 (2013).
50
161–164 (2016). W. C. Ke, C. Y. Kao, W. C. Houng, and C. A. Wei, J. Cryst. Growth 362,
20
I. C. Kizilyalli, P. Bui-Quang, D. Disney, H. Bhatia, and O. Aktas, 353–356 (2013).
51
Microelectron. Reliab. 55, 1654–1661 (2015). V. Cimalla, V. Lebedec, F. M. Morales, R. Goldhahn, and O. Ambacher,
21
W. Guo, M. Zhang, P. Bhattacharya, and J. Heo, Nano Lett. 11, Appl. Phys. Lett. 89, 172109 (2006).
52
1434–1438 (2011). V. Y. Davydov, A. A. Klochikhin, V. V. Emtsev, D. A. Kurdyukov, S. V.
22
J. Ristic, E. Calleja, A. Trampert, S. Fernandez-Garrido, C. Rivera, U. Ivanov, V. A. Vekshin, F. Bechstedt, J. Furthmuller, J. Aderhold, J. Graul,
Jahn, and K. H. Ploog, Phys. Rev. Lett. 94, 146102 (2005). A. V. Mudryi, H. Harima, A. Hashimoto, A. Yamamoto, and E. E. Haller,
23
Y. L. Chang, J. L. Wang, F. Li, and Z. Mi, Appl. Phys. Lett. 96, 013106 Phys. Status Solidi B 234, 787–795 (2002).
53
(2010). A. Yoshikawa, K. Kusakabe, N. Hashimoto, E. S. Hwang, and T. Itoi,
24
S. D. Hersee, X. Sun, and X. Wang, Nano Lett. 6, 1808–1811 (2006). Appl. Phys. Lett. 108, 022108 (2016).

You might also like