-
This content has been downloaded from IOPscience. Please scroll
down to see the full text.
Download details:
IP Address: 128.113.7.98
This content was downloaded on 15/12/2015 at 14:43
Please note that terms and conditions apply.
Analysis of the stability of InGaN/GaN multiquantum wells
against ion beam intermixing
View the table of contents for this issue, or go to the journal
homepage for more
2015 Nanotechnology 26 425703
(http://iopscience.iop.org/0957-4484/26/42/425703)
Home Search Collections Journals About Contact us My
IOPscience
iopscience.iop.org/page/termshttp://iopscience.iop.org/0957-4484/26/42http://iopscience.iop.org/0957-4484http://iopscience.iop.org/http://iopscience.iop.org/searchhttp://iopscience.iop.org/collectionshttp://iopscience.iop.org/journalshttp://iopscience.iop.org/page/aboutioppublishinghttp://iopscience.iop.org/contacthttp://iopscience.iop.org/myiopscience
-
Analysis of the stability of InGaN/GaNmultiquantum wells against
ion beamintermixing
A Redondo-Cubero1,2, K Lorenz1, E Wendler3, S Magalhães1, E
Alves1,D Carvalho4,5, T Ben4,5, F M Morales4,5, R García4,5, K P
O’Donnell6 andC Wetzel7
1 Instituto Superior Técnico, Universidade de Lisboa, Campus
Tecnológico e Nuclear, Estrada Nacional 10,2695-066 Bobadela LRS,
Portugal2Departamento de Física Aplicada y Centro de Microanálisis
of Materiales, Universidad Autónoma deMadrid, E-28049 Madrid,
Spain3 Friedrich-Schiller-Universität Jena, Institut für
Festkörperphysik, Max-Wien-Platz 1, D-07743 Jena,Germany4Department
of Materials Science and Metallurgic Engineering, and Inorganic
Chemistry, Faculty ofSciences, University of Cadiz, Spain5 IMEYMAT,
Institute of Research on Electron Microscopy and Materials,
University of Cádiz, Spain6 SUPA Department of Physics, University
of Strathclyde, Glasgow, G4 0NG, Scotland, UK7Department of
Physics, Applied Physics, and Astronomy, Rensselaer Polytechnic
Institute, 110 EighthStreet, Troy, NY 12180-3590, USA
E-mail: [email protected]
Received 7 June 2015, revised 27 July 2015Accepted for
publication 5 August 2015Published 1 October 2015
AbstractIon-induced damage and intermixing was evaluated in
InGaN/GaN multi-quantum wells(MQWs) using 35 keV N+ implantation at
room temperature. In situ ion channelingmeasurements show that
damage builds up with a similar trend for In and Ga atoms, with a
highthreshold for amorphization. The extended defects induced
during the implantation, basal andprismatic stacking faults, are
uniformly distributed across the quantum well structure. Despite
theextremely high fluences used (up to 4×1016 cm−2), the InGaN MQWs
exhibit a high stabilityagainst ion beam mixing.
Keywords: ion beam mixing, InGaN, quantum wells,
implantation
(Some figures may appear in colour only in the online
journal)
1. Introduction
InGaN/GaN multiple quantum wells (MQWs) are the basis ofmany
modern optoelectronic devices, including bright lightemitting
diodes, laser diodes, light displays, solar cells, etc.However, the
strain induced by the lattice mismatch, inaddition to the
spontaneous polarization in III-N hetero-structures, affects the
internal quantum efficiency through theso-called quantum-confined
Stark effect (QCSE), whichinduces a spatial separation of electron
and hole wavefunc-tions [1], in particular for high InN-content
devices, with wideMQW emitting in the green spectral region. One of
the
strategies suggested to mitigate QCSE in GaN-based materi-als is
the fabrication of intermixed or graded quantum struc-tures by ion
beam mixing [2, 3]. This approach has beensuccessfully applied in
AlGaAs/GaAs [4, 5], AlGaInP/GaInP [6] and MgZnO/ZnO [7] MQWs, but
it remainsunexplored in InGaN/GaN systems.
Despite the outstanding radiation resistance of GaN
(withamorphization thresholds above 10 displacements per
atom),recent experimental results in GaN/AlN MQWs [8] haveproved
that partial intermixing can be induced at low tem-peratures (15 K)
for sufficiently high Ar fluences. Thisintermixing induced by
collision cascades was shown to be
Nanotechnology
Nanotechnology 26 (2015) 425703 (6pp)
doi:10.1088/0957-4484/26/42/425703
0957-4484/15/425703+06$33.00 © 2015 IOP Publishing Ltd Printed
in the UK1
mailto:[email protected]://dx.doi.org/10.1088/0957-4484/26/42/425703http://crossmark.crossref.org/dialog/?doi=10.1088/0957-4484/26/42/425703&domain=pdf&date_stamp=2015-10-01http://crossmark.crossref.org/dialog/?doi=10.1088/0957-4484/26/42/425703&domain=pdf&date_stamp=2015-10-01
-
more efficient than thermal interdiffusion [8], and motivatesthe
present studies. We consider here conventional InGaN/GaN
nanostructures under more adaptable conditions forindustrial
scalability, i.e., room temperature implantation withnative ions.
Thus, in this work we investigate the damagebuild-up during ion
mixing processes in InGaN/GaN MQWswith 35 keV N+ ions, comparing
the behavior with thickInGaN layers.
2. Experiment
The InxGa1−xN/GaN MQW sample was grown by metal-organic chemical
vapor deposition on a (0001) GaN/sapphiretemplate [9]. The active
layers consist of five InxGa1−xN/GaN periods with thicknesses of 3
nm/20 nm (measured bytransmission electron microscopy (TEM)) and an
average InNmolar fraction in the QWs of x=10% (determined
byRutherford backscattering spectrometry (RBS)). A ∼20 nmGaN
capping layer was deposited to protect the active
layers.Photoluminescence and ionoluminesce measurements reveal
agreen QW emission centered at ∼530 nm (2.34 eV).
Ion implantation and in situ RBS in channeling mode(RBS/C)
studies were carried out in the double-beamchamber at the Institut
für Festkörperphysik (Jena) [10]. Theimplantations were performed
at room temperature with35 keV N+ ions, being the energy selected
to place themaximum of the damage profile in the central region of
theMQWs. According to Monte Carlo simulations with theSRIM code the
mean N range is 55 nm, with a longitudinalstraggling of 25 nm [11].
The ion fluence was increasedsequentially from 5×1012 to 4×1016
cm−2 and the samplewas tilted by 7° off-axis during the
implantation to avoidchanneling effects. After each implantation
step, the damageaccumulation was evaluated in situ by RBS/C spectra
of thesample using a 2.2 MeV He+ collimated beam (∼1 mm2).RBS/C
spectra were taken in the 〈0001〉 aligned direction,using a 3-axis
goniometer for sample orientation.
The final as-implanted sample, corresponding to a totalfluence
of 4×1016 cm−2, was further studied ex situ withx-ray diffraction
(XRD), TEM, and grazing incidence RBS.XRD data were acquired in a
D8Discover high resolutiondiffractometer (Bruker-AXS) using Cu(Kα1)
radiation, anasymmetric two-bounce Ge(220) monochromator, and a
NaIscintillation detector. To reduce the divergence, the
incomingx-ray beam is collimated with a 0.2 mm slit while the
dif-fracted beam is collimated with a 0.5 mm motorized slit and
a0.1 mm fixed slit, being the approximate final 2θ/ω
angularresolution ±0.01°. XRD 2θ/ω curves were simulated usingthe
dynamical theory of XRD [12] The influence of theimplantation is
taken into account by including the staticDebye–Waller parameter
(describing the attenuation causedby thermal motion) on the
structure factor of individual lay-ers, thus attenuating the
derived intensity and considering a c-lattice expansion following
the work of Boulle andDebelle [13].
TEM experiments were carried out using a JEOL 2010Fmicroscope
operated at 200 kV. Structural and compositional
features were studied from micrographs collected by HR-TEM and
high angle annular dark field (HAADF) imaging inscanning-TEM (STEM)
mode, while nanoprobe analyses oflocal proportions of atomic
elements were based upon energydispersive x-ray spectroscopy done
in the same mode (EDX-STEM). Grazing incidence RBS experiments were
carried outwith a 2MeV He+ beam using the Van de Graaff
acceleratorat LATR/IST (Portugal).
3. Results and discussion
Figure 1(a) shows RBS/C spectra of the sample after
theimplantation to different total N fluences. Remarkably, the
Insignal does not show any shift to higher energies even for
thehighest fluence, what confirms that there is neither
appreci-able diffusion of In towards the surface, nor
significantsputtering of the GaN capping layer. Two energy
windowswere defined for In(w1) and Ga(w2) at equivalent
depthscorresponding to the central region of the damage profile
inthe MQWs and avoiding the surface peak zone. The minimumyield
(χmin), determined as the ratio of the aligned and therandom
backscattering yield, was obtained for both windows.The increase of
χmin with fluence determines the damageaccumulation. The damage
level is then described as thedifference in the minimum yield of
the implanted and virginsample, Δχmin=χmin(implanted)−χmin(virgin).
As a refer-ence, the initial χmin values for the as-grown sample
were 4.4(1)% and 2.2(1)% for In and Ga, respectively. For the
max-imum fluence of 4×1016 cm−2 the aligned level almostoverlaps
with the random level (χmin=100%), indicating alevel of damage that
is maximal for measurement by RBS/C.
Figure 1(b) shows the damage level obtained for In(w1)and Ga(w2)
for the entire range of fluences. Both curves havea similar
behavior in two main regions of interest. For lowfluences (
-
Solid lines in figure 1(b) represent the fits using Heck-ing’s
model, which takes into account the production (Ppd),recombination
(Rpd), and clustering (Cpd) of point defects insemiconductors, as
well as the production of amorphousregions (Pa) and their growth
(Ga) [17, 18]. The results
obtained from these fits are summarized in table 1. The
var-iations observed with the values obtained from Wendler et
al[16], in particular the low values of Pa obtained here,
areascribed to the different temperature and not to the thicknessor
the structure of the InGaN layers. This is supported by thefact
that the damage build-up curve does agree well with thedata from
Kucheyev et al [14] at room temperature, pointingout the relevant
role of dynamic annealing processes.
Figure 2(a) shows the XRD 2θ/ω scans for the 0002reflection of
the sample before and after the implantation(corresponding to a
total fluence of 4×1016 cm−2 andnamed as-implanted) as well as the
fits to these curves. Thespectrum of the as-grown sample is well
described con-sidering the nominal structure without any roughness,
defor-mation or disorder (Debye–Waller factor of ∼1) but allowinga
slight variation of individual layer thicknesses below0.5 nm. The
as-grown sample shows up to 13 superlatticepeaks, in good agreement
with the simulated structure,reflecting excellent crystal and
interface quality. After theimplantation, a clear broadening of the
main peaks isobserved and several superlattice peaks are missing as
a resultof the damage build-up. The fit assumed the same SL
struc-ture as the as-grown sample but allowing slightly higher
Figure 1. (a) RBS/C spectra for different N+ fluences (note
thesemi-log scale). The surface energy positions for In and Ga
aremarked for better identification of the elements. Energy
windowsused for the individual analysis of damage are marked as w1
(In) andw2 (Ga). Fluence values are given in cm
−2. (b) Damage level forwindow w1 (In) and w2 (Ga) as a function
of the fluence. Thecorresponding fits using Hecking’s model are
also shown (lines).Results for thin InGaN films implanted with Au
at room and lowtemperature are also shown as a reference (adapted
from [14, 16]).
Table 1. Main parameters obtained from the Hecking model.
Parameter In (w1) Ga (w2)
Ppd (10−16 cm2) 4.5(5) 2.9(5)
Rpd (10−14 cm2) 1.5(2) 1.2(2)
Cpd (10−16 cm2) 1.2(2) 0.4(2)
Pa (10−18 cm2) 1(1) 1(1)
Ga (10−17 cm2) 4(1) 8(1)
Figure 2. (a) HR-XRD (0002) 2θ/ω scans and simulations for
as-grown and as-implanted (4×1016 cm−2) MQWs. (b) and (c)Reciprocal
space mappings for both samples. The color barrepresents intensity
in a log scale.
3
Nanotechnology 26 (2015) 425703 A Redondo-Cubero et al
-
variation of layer thicknesses (up to 0.7 nm deviation from
thenominal structure) and a decreased Debye–Waller factor(∼0.5 for
the first three periods increasing until reachingagain unity for
the deeper layers). Furthermore, a deformationof the implanted
layers was taken into account yielding aperpendicular strain of
maximum ∼1% for the three firstperiods and decreasing to zero for
the deeper unimplantedlayers.
The in-plane (Qx) and out-of-plane (Qz) components ofthe
scattering vector were obtained from the corresponding1015( ¯ )
reciprocal space maps of both samples, shown infigures 2(b) and
(c). In addition to the broadening, the mapsconfirm the strong
suppression of the SL peaks (the signalabove GaN being due to the
remaining Kα2 contribution fromthe monochromator). The elongated
peak below GaN isassociated to the expansion of the c-lattice
parameter in theheterostructure, an effect well-documented for
implantation inthin nitride films [19, 20]. It should be noted that
the dyna-mical theory may not be well suited to describe XRD of
thehighly damaged sample after high fluence
implantation.Nevertheless, the results are in good agreement with
the RBSand TEM results presented in this paper suggesting a
deterioration of the crystal quality, in particular of the
firstthree periods, while intermixing remains below the
detectionlimits of the employed techniques.
Figure 3 shows TEM images of the defects formed in
theas-implanted maximum-fluence sample. Dark field images,recorded
using 0002, 1–120 and 1–100 reflection g vectors,respectively
(figures 3(a)–(c)), reveal a heavily damagedsurface region. Figure
3(a), taken under the g=0002 con-dition, indicates the presence of
a large number of pointdefect clusters, interstitials or vacancies,
also confirmed infigure 3(b). Figure 3(c), exciting the g=1–100
reflection,shows that, in addition to the defect clusters, a
complexnetwork of basal stacking faults (BSFs), both intrinsic (I1
andI2 type) and extrinsic (E), and prismatic stacking faults[21,
22] has been generated during the implantation. A highresolution
example of one I2 BSF inside the implanted regionis exhibited in
figure 3(d). The formation of these extendeddefects agrees well
with previous reports on both GaN[15, 19] and InGaN [23], where
clusters and planar defectshave been pointed out for relatively
high implantation flu-ences. A similar defect structure of the
ternary and binarycompounds can explain the homogenous distribution
of
Figure 3. Dark field TEM images of the as-implanted sample
(4×1016 cm−2) acquired with g=0002 (a), g=1–120 (b), and g=1–100(c)
orientations. (d) HR-TEM image corresponding to the implanted
region, where an I2 intrinsic BSF is visible.
4
Nanotechnology 26 (2015) 425703 A Redondo-Cubero et al
-
defects observed in the present case, where the presence ofthe
QWs does not seem to affect the defect distribution andthe
migration/annihilation properties.
Figure 4 shows HAADF micrographs of (a) the as-grownand (b) the
as-implanted samples. After the implantation, thetop first QWs are
not clearly distinguishable due to the highdamage levels, which are
known to distort the unit cell eitherby accumulation of point
defects or strain fields associatedwith extended defects, both
cases disimprove on the contrastof HAADF-STEM images [24, 25], and
can render the directinterpretation of contrasts in terms of
chemical compositionsineffective. However, local EDX spectra reveal
that most ofthe In remains close to the original QW structure (not
shown).This fact was confirmed by grazing incidence (60°
tiltbetween the surface normal and the incoming beam)
RBSmeasurements, which clearly resolve the first two InGaNQWs which
are completely separated in the spectrum.Figure 4(c) directly
compares the RBS spectra of the as-grown and as-implanted sample;
no major difference can bediscerned. Both spectra are compatible
with the targeted ori-ginal depth profile. Nevertheless, some
degree of intermixingcannot be completely excluded due to the
limited depthresolution of the technique. These results differs
from thoseobtained for AlN/GaN MQWs where a partial intermixingcan
be induced by 100 keV Ar+ implantation [8]. This fact
might be related to the different efficiency of the
ballisticintermixing in both cases (which is expected to be three
timeslower in the current study due to the lower energy and
ionmass) but other factors such as the electronic
energy-losscontribution, the stress state of the material, and the
solubilitylimits of the elements cannot be ruled out and could also
playa role in this regard.
4. Conclusions
We have demonstrated that InGaN/GaN MQWs exhibit ahigh
resistance to ion beam induced mixing at room tem-perature. Damage
builds up with a similar trend for In and Gaatoms, leading to the
formation of defect clusters and exten-ded planar defects. Such
defects are homogeneously dis-tributed throughout the MQWs, but
despite the high damagelevel attained both EDX-STEM and grazing
incidence RBSconfirmed that compositional intermixing in the
QWsremains low.
Acknowledgments
We thank Dr P Ruterana for fruitful discussions and
sug-gestions. We acknowledge support by FCT Portugal
(bilateralproject DAAD/FCT 2011-2012, PTDC/FIS-NAN/0973/2012,
SFRH/BPD/74095/2010, Investigador FCT) andJuan de la Cierva program
(under contract number JCI-2012-14509). TEM experiments were
carried out in the ElectronMicroscopy -DME- and Sample Preparation
-LPM- Divisionsof the Central Services of Science and Technology of
theUniversity of Cádiz (SCCYT-UCA), being financed by theprojects
MAT2010-15206 (CICYT, Spain), EU-COST ActionMP0805, and
P09-TEP-5403 (Junta de Andalucía with EU-FEDER participation).
References
[1] Miller D A B, Chemla D S, Damen T C, Gossard A C,Wiegmann W,
Wood T H and Burrus C A 1984 Phys. Rev.Lett. 53 2173
[2] O’Donnell K P, Auf der Maur M, Di Carlo A and Lorenz Kand
the SORBET consortium, 2012 Phys. Status Solidi RRL2 49
[3] Jain I P and Agarwal G 2011 Surf. Sci. Rep. 66 77[4] Ralston
J, Wicks G W, Eastman L F, De Cooman B C and
Carter C B 1986 J. Appl. Phys. 59 120[5] Tan H H, Jagadish C,
Williams J S, Zou J, Cockayne D J H and
Sikorski A 1995 J. Appl. Phys. 77 87[6] Beernink K J, Sun D,
Treat D W and Bour B P 1995 Appl.
Phys. Lett. 66 3597[7] Davis J A et al 2008 Nanotechnology 19
055205[8] Redondo-Cubero A, Lorenz K, Wendler E, Carvalho D, Ben
T,
Morales F M, García R, Fellmann V and Daudin B
2013Nanotechnology 24 505717
[9] Wetzel C, Salagaj T, Detchprohm T, Li P and Nelson J S
2004Appl. Phys. Lett. 85 866
[10] Breeger B, Wendler E, Trippensee W, Schubert C andWesch W
2001 Nucl. Instrum. Methods B 174 199
Figure 4.HAADF micrographs showing the as-grown (a) and the
as-implanted (b) sample. (c) Grazing incidence (60°) RBS spectra of
as-grown and as-implanted sample. The simulation for the
nominalstructure is also shown for comparison. Markers for the
surfaceenergies of In and Ga are displayed.
5
Nanotechnology 26 (2015) 425703 A Redondo-Cubero et al
http://dx.doi.org/10.1103/PhysRevLett.53.2173http://dx.doi.org/10.1002/pssr.201100206http://dx.doi.org/10.1016/j.surfrep.2010.11.001http://dx.doi.org/10.1063/1.336852http://dx.doi.org/10.1063/1.359358http://dx.doi.org/10.1063/1.113799http://dx.doi.org/10.1088/0957-4484/19/05/055205http://dx.doi.org/10.1088/0957-4484/24/50/505717http://dx.doi.org/10.1063/1.1779960http://dx.doi.org/10.1016/S0168-583X(00)00433-X
-
[11] Ziegler J F, Biersack J P and Littmark U 1985 The
Stoppingand Range of Ions in Solids (New York: Pergamon)
[12] Bartels W J, Hornstra J and Lobeek D J W 1986
ActaCrystallogr. A42 539
[13] Boulle A and Debelle A 2010 J. Appl. Crystallogr. 43
1046[14] Kucheyev S O, Williams J S, Zou J and Jagadish C 2004
J. Appl. Phys. 95 3048[15] Ruterana P, Lacroix B and Lorenz K
2011 J. Appl. Phys. 109
013506[16] Wendler E, Wesch W, Alves E and Kamarou A 2004
Nucl.
Instrum. Methods B 218 36[17] Hecking N, Heidemann K F and
Tekaat E 1986 Nucl. Instrum.
Methods B 15 760[18] Wendler E 2009 Nucl. Instrum. Methods B 267
2680
[19] Lacroix B, Leclerc S, Declémy A, Lorenz K, Alves E
andRuterana P 2011 EPL 96 46002
[20] Leclerc S, Lacroix B, Declémy A, Lorenz K and Ruterana
P2012 J. Appl. Phys. 112 073525
[21] Ruterana P and Nouet G 2001 Phys. Status Solidi B
227177
[22] Gloux F, Wojtowicz T, Ruterana P, Lorenz K and Alves E2006
J. Appl. Phys. 100 073520
[23] Kucheyev S O, Williams J S, Zou J, Pearton S J andNakagawa
Y 2001 Appl. Phys. Lett. 79 602
[24] Grillo V, Carlino E and Glas F 2008 Phys. Rev. B 77
054103[25] Grillo V, Mueller K, Volz K, Glas F, Grieb T and
Rosenauer A
2011 17th Int. Conf. on Microscopy of SemiconductingMaterials
2011 vol 326
6
Nanotechnology 26 (2015) 425703 A Redondo-Cubero et al
http://dx.doi.org/10.1107/S0108767386098768http://dx.doi.org/10.1107/S0021889810030281http://dx.doi.org/10.1063/1.1649459http://dx.doi.org/10.1063/1.3527944http://dx.doi.org/10.1063/1.3527944http://dx.doi.org/10.1016/j.nimb.2003.12.033http://dx.doi.org/10.1016/0168-583X(86)90407-6http://dx.doi.org/10.1016/j.nimb.2009.05.059http://dx.doi.org/10.1209/0295-5075/96/46002http://dx.doi.org/10.1063/1.4758311http://dx.doi.org/10.1002/1521-3951(200109)227:13.0.CO;2-7http://dx.doi.org/10.1002/1521-3951(200109)227:13.0.CO;2-7http://dx.doi.org/10.1063/1.2357845http://dx.doi.org/10.1063/1.1388881http://dx.doi.org/10.1103/physrevb.77.054103
1. Introduction2. Experiment3. Results and discussion4.
ConclusionsAcknowledgmentsReferences