Académique Documents
Professionnel Documents
Culture Documents
Enhanced light scattering of the forbidden longitudinal optical phonon mode studied by micro-
This article has been downloaded from IOPscience. Please scroll down to see the full text article.
(http://iopscience.iop.org/0957-4484/21/31/315702)
View the table of contents for this issue, or go to the journal homepage for more
Download details:
IP Address: 194.57.171.11
The article was downloaded on 14/10/2011 at 12:55
E-mail: t.stoica@fz-juelich.de
Abstract
In the literature, there are controversies on the interpretation of the appearance in InN Raman
spectra of a strong scattering peak in the energy region of the unscreened longitudinal optical
(LO) phonons, although a shift caused by the phononplasmon interaction is expected for the
high conductance observed in this material. Most measurements on light scattering are
performed on ensembles of InN nanowires (NWs). However, it is important to investigate the
behavior of individual nanowires and here we report on micro-Raman measurements on single
nanowires. When changing the polarization direction of the incident light from parallel to
perpendicular to the wire, the expected reduction of the Raman scattering was observed for
transversal optical (TO) and E2 phonon scattering modes, while a strong symmetry-forbidden
LO mode was observed independently on the laser polarization direction. Single Mg- and
Si-doped crystalline InN nanowires were also investigated. Magnesium doping results in a
sharpening of the Raman peaks, while silicon doping leads to an asymmetric broadening of the
LO peak. The results can be explained based on the influence of the high electron concentration
with a strong contribution of the surface accumulation layer and the associated internal electric
field.
(Some figures in this article are in colour only in the electronic version)
0957-4484/10/315702+06$30.00 1 2010 IOP Publishing Ltd Printed in the UK & the USA
Nanotechnology 21 (2010) 315702 E O Schafer-Nolte et al
Raman spectroscopy, widely used in the study of not of the symmetry-allowed A1 (LO) in those experiments
semiconductors, yields information about lattice properties was explained by strong scattering of the excitation light
such as crystal structure and quality, as well as electronic inside the NW ensemble of the as-grown samples. In
properties like carrier concentration and mobility. It has the contrast, in their study on InN films Cusco et al attributed the
advantages of being non-destructive and contactless and does unscreened LO peak to surface-field-induced scattering in the
not require any special sample preparation. The first-order accumulation layer [17]. To summarize, different mechanisms
phonon Raman scattering for InN follows the selection rules and models have been proposed for the interpretation of the
corresponding to its wurtzite structure [13]. For back scattering LO and PLP peaks. Raman spectroscopy of single NWs
geometry, if the excitation laser light propagates along the c in a controlled scattering geometry, avoiding multiple light
axis, the allowed scattering modes are the non-polar mode E2 scattering in NW ensembles and fluctuations from wire to wire
and the longitudinal phonon A1 (LO), while for the incident as well as along a wire, can contribute to resolve controversies
light perpendicular to the c axis, only the E2 and the transversal in the interpretation of Raman scattering in InN. Raman
optical phonons A1 (TO) and E1 (TO) are allowed. The LO measurements on individual NWs may also offer information
mode can be in polar semiconductors strongly coupled to the about the light interaction with the electronic and phononic
plasmonic system of the free carriers, resulting in a low-energy systems in single InN nanowire devices.
branch PLP (at energy smaller than the TO) and a high- In this paper we present Raman measurements on undoped
energy branch PLP+ (at energy higher than the LO mode). and Mg-doped, as well as Si-doped single nanowires. Taking
The Raman shift of these coupled modes increases with the advantage of the well-defined scattering geometry and the
plasma frequency and can be used to determine the density high spatial resolution, the presented data provide relevant
and mobility of the free carriers. Raman studies on InN information about both the light scattering mechanism and
nanowires reported so far showed a simultaneous occurrence the physical properties of InN nanowires. To the best of our
of a strong peak in the region of the expected uncoupled knowledge, this is a first report on single InN nanowire Raman
longitudinal optical A1 (LO) and E1 (LO) modes and of the measurements.
broad low-energy branch of the LO phononplasmon coupled
mode (PLP ) close to A1 (TO) [14, 15]. 2. Experimental details
The detailed analysis of the PLP lineshape in InN and
the spatial location of the electrons are still controversial. InN nanowires are grown catalyst-free by plasma-assisted
Some papers attribute the scattering mechanism, from molecular beam epitaxy under N-rich conditions on Si(111)
which originates the Raman peak close to A1 (TO), not substrates, as already reported by various groups [1923].
to the low branch phononplasmon coupled mode PLP , Details of the growth mechanism as well as the structural,
but to an impurity-induced one or due to charge-density morphological, optical and electrical properties of undoped
fluctuations [16]. Cusco et al have clearly found for high nanowires have been discussed elsewhere [19, 2429]. The
mobility InN films with electron concentrations in the range obtained nanowires have a diameter between 80 and 120 nm
of 2 1018 2 1019 cm3 that PLP scattering arises from and a length of up to 2 m. Nominal p- and n-type
coupling to the bulk-electron plasmons and can be explained doping is achieved by supplying Mg and Si during growth,
by the standard dielectric model based on dipole-allowed respectively. The dopant fluxes in this paper are given in beam
deformation potential and electro-optic mechanism [17]. In equivalent pressure (BEP). To investigate single nanowires,
contrast, for InN NWs, the PLP peak was attributed not to the they were removed from the growth substrate and spread on
wire bulk but to the spontaneous accumulation of electrons on a graphite substrate (highly ordered pyrolytic graphite) by a
the lateral surfaces of the NWs [14]. The high-energy branch dry transfer method. Raman scattering measurements have
of the LO phononplasmon coupled PLP+ mode has not been been carried out using a WITec alpha 300 microscopy system.
detected in InN NWs nor in compact layers. The absence of a For measurements on nanowire ensembles a 20 (NA 0.4)
clear PLP+ scattering in Raman spectra in spite of a detected microscope objective was used to focus the laser beam and
PLP was explained by a rapid broadening of PLP+ due to to collect the scattered light. Laser radiation with 532 and
high carrier density and the associated fluctuations [17]. It 633 nm wavelength with a power of about 5 mW was employed
is worth noting that at high electron concentrations the PLP as an excitation source. The scattered light was analyzed in
behaves more phonon-like, while the PLP+ has a dominant a 1800 mm1 grating spectrometer with a spectral resolution
plasmon character and therefore is more sensitive to the carrier of about 1 cm1 . Single scan was used for acquisition of
distribution. Another explanation given in the literature for Raman spectra of NW ensembles, with an integration time of 1
the absence of the PLP+ is based on the unscreening, and and 2 h for 532 and 633 nm laser wavelengths, respectively.
consequently de-coupling, of the polar LO phonons in the For Raman studies on single nanowires a 100 (NA 0.9)
case of highly damped plasmons, while an LO-phonon-like microscope objective was used to focus the 532 nm laser
structure occurs as an interaction of the polar phonons with beam to a spot size of about 300 nm. To avoid strong laser
electronhole pairs [18]. But the origin of the unscreened heating, the laser power was reduced to 0.5 mW. The scattered
LO peak in InN is controversial as well. For InN ensemble light was analyzed using a 600 mm1 spectrometer grating
nanowires, the LO peak was attributed to E1 (LO) scattering with a spectral resolution of about 3 cm1 . The sample was
in the NW core with low carrier concentration [14, 15]. The scanned relative to the laser beam by a piezoelectrical stage
observation of the symmetry-forbidden E1 (LO) mode and to obtain spatially resolved intensity maps. Since the Raman
2
Nanotechnology 21 (2010) 315702 E O Schafer-Nolte et al
signal of a single InN nanowire is very weak, the wire was A1(TO) E1(TO) E2H A1(LO) E1(LO)
imaged by mapping the intensity of the elastically scattered
light (Rayleigh peak at 0 cm1 ). It was observed that this 2.5
*
intensity is strongly enhanced when the wire is in the laser
focus. Therefore a good contrast between the nanowire and 2.0
the graphite substrate could be achieved in a relative short
Intensity (a.u.)
acquisition time (about 15 min for a complete map). After 1.5
localizing the nanowire, the laser spot was positioned on the 200"nm
200 nm
wire and a single spectrum with an integration time of 1 h was
1.0 633 nm
recorded.
0.5
3. Results and discussion 532 nm
0.0
Raman spectra of undoped InN nanowire ensembles are shown
400 450 500 550 600 650
in figure 1. The samples are measured as-grown, i.e. with Raman shift (1/cm)
nanowires still attached to the silicon substrate on which they
were grown. Therefore a strong Raman peak from the silicon Figure 1. Raman spectra of undoped InN nanowire ensembles
substrate can be seen at 520 cm1 . Since the nanowires are for 532 and 633 nm excitation wavelengths. The spectra are
normalized to the EH2 peak intensity and shifted for clarity. Black
oriented nearly parallel to the incident laser beam, the nominal dotted lines show Raman mode frequencies reported in the
scattering configuration is z(, )z . Selection rules predict literature [13]. The asterisk marks the Raman signal of the silicon
only the occurrence of EH 2 and A1 (LO) for this geometry. E2
H
substrate. Inset is an SEM image of the as-grown NW sample.
1
is clearly pronounced at 487 cm with an FWHM of about
8 cm1 , while the assignment of the strong peak close to
the position of the LO modes is questionable, as discussed and light trapping in NW ensembles, and thus more intense
above. For 532 nm excitation wavelength, the peak occurs heating, can be expected, even though the same incident
close to the reported A1 (LO) frequency at 586 cm1 , for power of the laser beam was used. However, changing the
633 nm wavelength it is shifted closer to the frequency of laser power within the measurement range used, no significant
E1 (LO) at 593 cm1 . The nanowire morphology, which allows change of the peak position was observed. A similar redshift
that most of the light enters through and scatters from the with photon energy was reported for InN films in [30] and
lateral side of the nanowire, could explain the observation of explained by the difference of the light penetration depth at
the symmetry-forbidden E1 (LO) [14, 15], as well as the weak different wavelengths in the presence of an accumulation layer
occurrence of E1 (TO) (476 cm1 ). However, the peculiarity with a very non-uniform distribution of the free carriers. To
of the LO mode in InN NWs, i.e. its symmetry-forbidden fully understand these phenomena related to the wavelength
occurrence with high intensity, is clearly shown by single NW approaching a resonance condition in InN nanowires, further
measurements, as we discuss in the following. The intensity theoretical and experimental efforts would be necessary.
of the LO mode is strongly depending on the laser wavelength: The origin of the broad feature at 440 cm1 in the spectra
by changing the excitation wavelength from 532 to 633 nm, of figure 1 might be due to the PLP mode. On the other hand,
the intensity of LO relative to EH 2 is roughly tripled. This an assignment to the symmetry-forbidden A1 (TO) seems to be
result is in agreement with the one reported by Kuball et al reasonable due to the unusual light propagation in the nanowire
for InN thin films [30]. A similar increase of the LO intensity sample. The second-order spectrum of silicon also has a peak
for longer excitation wavelength was also observed in Si-and in this spectral range [32], which complicates the evaluation
Mg-doped nanowire ensembles studied in this work. Such of this mode. We can conclude that the Raman spectra of
a strong increase is hardly explicable by distinct penetration our as-grown samples are consistent with those reported in the
depths. Moreover, the internal scattering of the excitation literature for InN NWs.
light should be stronger for shorter wavelength and increases Typical results for undoped single nanowires are shown in
the contribution of E1 (LO) to the LO peak [15]. This is figure 2. Raman spectra and the related Rayleigh peak intensity
in contradiction to the experimental data in figure 1. The maps are presented for two wires with different orientations
most probable explanation is therefore a resonance effect at relative to the laser polarization direction. The polarization
a critical point of the electronic structure which enhances of the incident light is depicted by the arrow in the bottom
scattering of polar modes by the Frohlich mechanism, as was left corner of the map. The scattering geometry in these
suggested in [30] and observed in other materials such as, experiments can be described by a combination of x(y, )x
for example, in GaAs [31]. By decreasing the excitation and x(z, )x , while the contribution of the particular scattering
wavelength, a redshift of the Raman peaks can also be seen in processes depends on the orientation of the nanowire. The
figure 1. This was constantly observed for undoped and doped spectra in figure 2 are measured in different parts of the same
samples. Moreover, the energetic distance between EH 2 and LO wire. The labels A and B in the maps denote the particular
peaks is increased for longer excitation wavelengths. A local origin of the spectra. The relative intensity of these peaks
heating induced by laser beam could explain a redshift of the varies from spectrum to spectrum and in some cases its position
Raman peaks. For shorter wavelengths, stronger absorption does not correspond very well with the literature values. The
3
Nanotechnology 21 (2010) 315702 E O Schafer-Nolte et al
A1(TO) E1(TO) E2H A1(LO) E1(LO) expressed in the spectra of wire 2. The reduction of the EH 2
scattering in wires with parallel orientation to the polarization
of the excitation light is a proof that the wires vertical axis is
wire 1
1500
parallel to the crystallographic c direction, as also shown by
HRTEM analysis [24]. The A1 (TO) mode exhibits a contrary
A
A trend, i.e. it is strongly pronounced in the spectra of wire 2 and
has a rather low intensity for wire 1.
1000
B
B The NW shape influences the light penetration into
the wire. If the NW diameter is much smaller than the
laser wavelength, the electromagnetic field inside the wire,
500
B and consequently the expected Raman scattering modes, are
Intensity (a.u.)
4
Nanotechnology 21 (2010) 315702 E O Schafer-Nolte et al
A1(TO) E1(TO) E2H A1(LO) E1(LO) A1(TO) E1(TO) E2H A1(LO) E1(LO)
-9 -9
Mg BEP =3 10 mbar Si BEP=5 .8 10 mbar
1500
500
1000
Intensity (a.u.)
500
Intensity (a.u.)
0 0
-9 -9
1500 Mg BEP =1 10 mbar Si BEP=1 10 mbar
1000
500
500
0 0
350 400 450 500 550 600 650 350 400 450 500 550 600 650
Raman shift (1/cm) Raman shift (1/cm)
Figure 3. Raman spectra of Mg-doped InN nanowires with different Figure 4. Raman spectra of Si-doped InN nanowires with different
doping levels, acquired from the center of the wire. Black dotted doping levels, acquired from the center of the wire. Black dotted
lines show Raman mode frequencies reported in the literature [13]. lines show Raman mode frequencies reported in the literature [13].
The insets show the related Rayleigh peak intensity maps. The scale The insets show the related Rayleigh peak intensity maps. The scale
bar is 500 nm, arrows denote the polarization direction of the bar is 500 nm, arrows denote the polarization direction of the
incident laser light. incident laser light.
5
Nanotechnology 21 (2010) 315702 E O Schafer-Nolte et al
of the accumulation layer at the wire surfaces can influence [3] Wilke I, Ascabazubi R and Schaff H L W 2008 Appl. Phys.
the Raman spectra via electronphonon interaction and by Lett. 93 221113
changing the light propagation path inside NWs. In [16] it is [4] Wu J, Walukiewicz W, Yu K, Shan W J A III, Haller E,
Lu H and Schaff W 2003 J. Appl. Phys. 94 6477
shown by modeling that, for high electron concentration and [5] Hamazaouni H, Bouazzi A and Rezig B 2005 Sol. Energy
strong Landau damping, the scattering on density fluctuations Mater. Sol. Cells 87 595
can dominate in the visible spectrum over Frohlich and [6] Hsu L and Walukiewicz W 2008 J. Appl. Phys. 104 024507
other mechanisms. A strong electric field in the surface [7] Xu T, Nikiforov A, France R, Williams C T A and
accumulation layer can also induce an enhancement, especially Moustakas T 2007 Phys. Stat. Sol. a 204 2098
[8] Soh C, Liu W, Teng J, Chow S, Ang S and Chua S 2008 Appl.
of the forbidden modes, as suggested in [17]. Indeed, the Phys. Lett. 92 261909
observation of a strong symmetry-forbidden LO scattering in [9] Bhuiyan A, Hashimoto A and Yamamoto A 2003 J. Appl. Phys.
our single NW Raman experiments might deliver additional 94 5
evidence of the formation of a surface accumulation layer [10] Kong Y, Zheng Y, Zhou C, Deng Y, Shen B, Gu S, Zhang R,
Han P, Jiang R and Shi Y 2005 Solid State Electron.
on non-polar surfaces of InN NWs, already suggested by
49 199
photoluminescence and electric transport experiments [24, 25]. [11] Lu H, Schaff W and Eastman L 2004 J. Appl. Phys. 96 3577
[12] Lu Y-S, Ho C-L, Yeh J, Lin H and Gwo S 2008 Appl. Phys.
4. Conclusions Lett. 92 212102
[13] Harima H 2002 J. Phys.: Condens. Matter 14 R967
[14] Lazc S, Gallardo E, Calleja J, Agullo-Rueda F, Grandal J,
In conclusion, Raman studies on single InN nanowires showed Sanchez-Garcia M, Calleja E, Luna E and Trampert A 2007
a strong polarization dependence of A1 (TO) and EH 2 phonon Phys. Rev. B 76 205319
modes in agreement with selection rules. Moreover, the [15] Segura-Ruiz J, Garro N, Cantarero A, Denker C,
occurrence of an intense symmetry-forbidden LO mode was Malindretos J and Rizzi A 2009 Phys. Rev. B 79 115305
[16] Demangeot F, Pinquier C, Frandon J, Gaio M, Briot O,
observed for single wires. A similar LO mode was also
Maleyre B, Ruffenbach S and Gil B 2005 Phys. Rev. B
found in the Raman spectra of as-grown nanowire ensembles. 71 104305
The observed strong excitation wavelength dependence of this [17] Cusco R, Ibanez J, Alarcon-Llado E, Artus L, Yamaguchi T and
mode is pointing to a resonance with electronic transitions, Nanishi Y 2009 Phys. Rev. B 79 155210
as suggested in the literature. The intensity enhancement at [18] Thakur J, Haddad D, Naik V, Naik R, Auner G, Lu H and
Schaff W 2005 Phys. Rev. B 71 115203
longer excitation wavelength, as well as the deviation from
[19] Stoica T, Meijers R, Calarco R, Richter T and Luth H 2006
the expected symmetry found in single NW Raman spectra, J. Cryst. Growth 290 241
indicates that the observed LO peak needs a more complex [20] Shen C-H, Chen H-Y, Lin H-W, Gwo S, Klochikhin A and
explanation in which the phonon interaction with the electronic Davydov V Y 2006 Appl. Phys. Lett. 88 253104
system including the surface accumulation layer is involved [21] Grandal J, Sanchez-Garca M, Calleja E, Luna E and
Trampert A 2007 Appl. Phys. Lett. 91 021902
and the symmetry selection rules are disregarded. The effect [22] Denker C, Malindretos J, Werner F, Limbach F, Schuhmann H,
of magnesium doping is a sharpening of the Raman peaks, Niermann T, Seibt M and Rizzi A 2008 Phys. Status Solidi c
which indicates a good crystalline quality and reduction of the 5 1706
background doping, while for silicon doping the LO phonon [23] Nishikawa S, Nakao Y, Naoi H, Araki T, Na H and
develops an asymmetric lineshape with a pronounced low- Nanishi Y 2007 J. Cryst. Growth 301 490
[24] Stoica T, Meijers R, Calarco R, Richter T, Sutter E and
energy tail. Luth H 2006 Nano Lett. 6 1541
[25] Richter T, Luth H, Schapers T, Meijers R, Jeganathan K,
Acknowledgments Estevez Hernandez S, Calarco R and Marso M 2009
Nanotechnology 20 405206
[26] Frielinghaus R, Estevez Hernandez S, Calarco R and
The authors gratefully acknowledge fruitful discussions and Schapers T 2009 Appl. Phys. Lett. 94 252107
suggestions by Professor H Luth. The authors thank also [27] Richter T, Blomers C, Luth H, Calarco R, Indlekofer M,
K H Deussen for technical support. This work was financially Marso M and Schapers T 2008 Nano Lett. 8 2834
supported by the German Ministry of Education and Research [28] Blomers C, Schapers T, Richter T, Calarco R, Luth H and
Marso M 2008 Phys. Rev. B 77 201301
project EPHQUAM and performed under the auspices of [29] Calarco R and Marso M 2007 Appl. Phys. A 87 499
the US Department of Energy under contract no. DE-AC02- [30] Kuball M, Pomeroy J, Wintrebert-Fouquet M, Butcher K, Lu H,
98CH1-886. Schaff W, Shubina T, Ivanov S, Vasson A and
Leymarie J 2005 Phys. Status Solidi a 202 763
[31] Trommer R and Cardona M 1978 Phys. Rev. B 17 1865
References [32] Temple P and Hathaway C 1973 Phys. Rev. B 7 8
[33] Kuball M, Pomeroy J, Wintrebert-Fouquet M, Butcher K,
[1] Wu J 2009 J. Appl. Phys. 106 011101 Lu H and Schaff W 2004 J. Cryst. Growth 269 59
[2] Che S B, Mizuno T, Wang X, Ishitani Y and Yoshikawa A 2007 [34] Schafer-Nolte E O, Stoica T, Gotschke T, Limbach E, Setter E,
J. Appl. Phys. 10(2) 083539 Setter P and Calarco R 2010 Appl. Phys. Lett. 96 091907