Henini M. Handbook of Self Assembled Semiconductor Nanostructures for Novel devices in Photonics and Electronics
Подождите немного. Документ загружается.
InAs Quantum Dot Formation Studied at the Atomic Scale 199
47. P.W. Fry et al., Phys. Rev. Lett. 84 , 733 ( 2000 ) .
48. D.M. Bruls , J.W.A.M. Vugs , P.M. Koenraad , H.W.M. Salemink , J.H. Wolter , M. Hopkinson ,
M.S. Skolnick , F. Long , and S.P.A. Gill , Appl. Phys. Lett. 81 , 1708 ( 2002 ) .
49. N. Liu , J. Tersoff , O. Baklenov , A.L. holmes Jr , and C.K. Shih , Phys. Rev. Lett. 84 , 334 ( 2000 ) .
50. O. Wolst , M. Kahl , M. Schardt , S. Malzer , and G.H. Dohler , Phys. E. 17 , 554 ( 2003 ) .
51. A. Lenz et al., Appl. Phys. Lett. 81 , 5150 ( 2002 ) .
52. P. Ballet , J.B. Smathers , and G.J. Salamo , Appl. Phys. Lett. 75 , 337 ( 1999 ) .
53. P.B. Joyce , J. Krzyzewski , R. Bell , and S. Jones , Appl. Phys. Lett. 79 , 3615 ( 2001 ) .
54. Q. Gong , R. Nötzel , J. Hamhuis , J. Eijkemans , and H. Wolter , Appl. Phys. Lett. 8 1 , 1887 (2002).
55. This method subtracts the mean value of the nearest neighbour pixels within the fi lter window:
O x,y I x,y
MN
iN
N
jM
M
Ix iy j()()
/
/
1
2
2
2
2
∑∑
/
/
(, ).
By choosing a size of about
1 1 n m
2
for the fi lter window, long range fl uctuations in the image are suppressed while atomic
details are preserved.
56. A. Rastelli , E. Müller , and H. von Känel , Appl. Phys. Lett. 80 , 1438 ( 2002 ) .
57. M. Stoffel , U. Denker , S. Kar , H. Sigg , and G. Schmidt , Appl. Phys. Lett. 83 , 2910 ( 2003 ) .
58. P.B. Joyce , J. Krzyzewski , R. Bell , A. Joyce , and S. Jones , Phys. Rev. B. 58 , R15981 ( 1998 ) .
59. V.M. Ustinov and A.E. Zhukov , Semicond. Sci. Technol. 15 , R41 ( 2000 ) .
60. K. Nishi , H. Saito , and S. Sugou , Appl. Phys. Lett. 74 , 1111 ( 1999 ) .
61. E.C. Le Ru , P. Howe , T.S. Jones , and R. Murray , Phys. Status Solidi C. 0 , 1221 ( 2003 ) .
62. H.Y. Liu , M. Hopkinson , C.N. Harrison , M.J. Steer , R. Frith , I.R. Sellers , D.J. Mowbray , and
M.S. Skolnick , J. Appl. Phys. 93 , 2931 ( 2003 ) .
63. H.Y. Liu , M.J. Steer , T.J. Badcock , D.J. Mowbray , M.S. Skolnick , P. Navaretti , K.M. Groom ,
M. Hopkinson , and R.A. Hogg , Appl. Phys. Lett. 86 , 143108 ( 2005 ) .
64. J.M. Ripalda , D. Granados , Y. Gonzalez , A.M. Sanchez , S.I. Molina , and J.M. Garcia , Appl. Phys.
Lett. 87 , 202108 ( 2005 ) .
65. S. Frechengues , N. Bertru , V. Drouot , B. Lambert , S. Robinet , S. Loualiche , D. Lacombe , and
A. Ponchet , Appl. Phys. Lett. 74 , 3356 ( 1999 ) .
66. P. Caroff , N. Bertru , A. Le Corre , O. Dehaese , T. Rohel , I. Alghoraibi , H. Folliot , and S. Loualiche ,
Jpn. J. Appl. Phys. 44 , L1069 ( 2005 ) .
67. P. Caroff , C. Paranthoën , C. Platz , O. Dehaese , H. Folliot , N. Bertru , C. Labbé , R. Piron , E. Homeyer ,
A. Le Corre , and S. Loualiche , Appl. Phys. Lett. 87 , 243107 ( 2005 ) .
68. M.V. Maximov , A.F. Tsatsul’nikov , B.V. Volovik , D.S. Sizov , Yu.M. Shernyakov , I.N. Kaiander ,
A.E. Zhukov , A.R. Kovsh , S.S. Mikhrin , V.M. Ustinov , Zh.I. Alferov , R. Heitz , V.A. Shchukin ,
N.N. Ledentsov , D. Bimberg , Yu.G. Musikhin , and W. Neumann , Phys. Rev. B. 62 , 16671 ( 2000 ) .
69. K. Akahane , N. Yamamoto , and N. Ohtani , Physica E Amsterdam 21 , 295 ( 2004 ) .
70. H.Y. Liu , M.J. Steer , T.J. Badcock , D.J. Mowbray , M.S. Skolnick , F. Suarez , J.S. Ng , M. Hopkinson ,
and J.P.R. David , J. Appl. Phys. 9 9 , 046104 ( 2006 ) .
71. R. Songmuang , S. Kiravittaya , and O.G. Schmidt , J. Cryst. Growth 249 , 416 ( 2003 ) .
72. W.M. McGee , T.J. Krzyzewski , and T.S. Jones , J. Appl. Phys. 99 , 043505 ( 2006 ) .
73. P. Offermans , P.M. Koenraad , J.H. Wolter , K. Pierz , M. Roy , and P.A. Maksym , Phys. Rev. B. 72 ,
165332 ( 2005 ) .
74. G. Costantini , A. Rastelli , C. Manzano , P. Acosta-Diaz , R. Songmuang , G. Katsaros , O.G. Schmidt ,
and K. Kern , Phys. Rev. Lett.
96 , 226106 ( 2006 ) .
75. P. Offermans , P.M. Koenraad , R. Nötzel , J.H. Wolter , and K. Pierz , Appl. Phys. Lett. 8 7 , 111903
( 2005 ) .
76. W. Seifert , D. Hessman , X. Liu , and L. Samuelson , J. Appl. Phys. 75 , 1501 ( 1994 ) .
77. H. Yang , P. Ballet , and G.J. Salamo , J. Appl. Phys. 8 9 , 7871 ( 2001 ) .
78. B. Wang , F. Zhao , Y. Peng , Z. Jin , Y. Li , and S. Liu , Appl. Phys. Lett. 72 , 2433 ( 1998 ) .
79. P. Disseix , C. Payen , J. Leymarie , and A. Vasson , J. Appl. Phys. 88 , 4612 ( 2000 ) .
80. P.J. Poole , R.L. Williams , J. Lefebvre , and S. Moisa , J. Cryst. Growth 257 , 89 ( 2003 ) .
81. J. He , R. Nötzel , P. Offermans , P.M. Koenraad , Q. Gong , G.J. Hamhuis , T.J. Eijkemans , and
J.H. Wolter , Appl. Phys. Lett. 8 5 , 2771 ( 2004 ) .
82. S. Fafard , Z. Wasilewski , J. McCaffrey , S. Raymond , and S. Charbonneau , Appl. Phys. Lett. 6 8 , 991
( 1996 ) .
CH005-I046325.indd 199CH005-I046325.indd 199 6/27/2008 4:19:42 PM6/27/2008 4:19:42 PM
200 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
83. H. Marchand , P. Desjardins , S. Guillon , J.-E. Paultre , Z. Bougrioua , R.Y.-F. Yip , and R.A. Masut ,
Appl. Phys. Lett. 71 , 527 ( 1997 ) .
84. C. Paranthoën , N. Bertru , O. Dehaese , A. Le Corre , S. Loualiche , B. Lambert , and G. Patriarche ,
Appl. Phys. Lett. 78 , 1751 ( 2001 ) .
85. K. Ozasa , Y. Aoyagi , Y.J. Park , and L. Samuelson , Appl. Phys. Lett. 71 , 797 ( 1997 ) .
86. Y. Sakuma , M. Takeguchi , K. Takemoto , S. Hirose , T. Usuki , and N. Yokoyama , J. Vac. Sci. Technol.
B. 23 , 1741 ( 2005 ) .
87. N.N. Ledenstov , V.A. Shchukin , M. Grundmann , N. Kirstaedter , J. Böhrer , O. Schmidt , D. Bimberg ,
V.M. Ustinov , A.Yu. Egorov , A.E. Zhukov , P.S. Kop’ev , S.V. Zaitsev , N.Yu. Gordeev , Zh.I. Alferov ,
A.I. Borovkov , A.O. Kosogov , S.S. Ruvimov , P. Werner , U. Gösele , and J. Heydenreich , Phys. Rev. B.
5 4 , 8743 ( 1996 ) .
88. Z.R. Wasilewski , S. Farad , and J.P. McCaffrey , J. Cryst. Growth 201 , 1131 ( 1999 ) .
89. T. Raz , D. Ritter , and G. Bahir , Appl. Phys. Lett. 82 , 1706 ( 2003 ) .
90. C. Çelebi , J.M. Ulloa , P.M. Koenraad , A. Simon , A. Letoublon , and N. B e r t r u , Appl. Phys. Lett. 8 9 ,
023119 ( 2006 ) .
91. G. Saint-Girons , G. Patriarche , A. Michon , G. Beaudoin , I. Sagnes , K. Smaali , and M. Troyon , Appl.
Phys. Lett. 89 , 031923 ( 2006 ) .
92. H.R. Gutiérrez , M.A. Cotta , J.R.R. Bortoleto , and M.M.G. de Carvalho , J. Appl. Phys. 9 2 , 7523
( 2002 ) .
93. J.-S. Lee , H.-W. Ren , S. Sugou , and Y. Masumoto , J. Appl. Phys. 84 , 6686 ( 1998 ) .
94. J.M. Ulloa, C. Çelebi, P.M. Koenraad, A. Simon, E. Gapihan, A. Letoublon, N. Bertru, J. Drouzas,
D.J. Mowbray, M.J. Steer and M. Hopkinson, J. Appl. Phys, 101, 081707 (2007).
95. J.M. Ulloa, I.W.D. Drouzas, P.M. Koenraad, D.J. Mowbray, M.J. Steer, H.Y. Liu and M. Hopkinson.
Appl. Phys. Lett. 90, 213105 (2007).
96 J.M. Ulloa, P.M. Koenraad, E. Gapihan, A. Létoublon and N. Bertru, Appl. Phys. Lett. 91, 073106
(2007).
CH005-I046325.indd 200CH005-I046325.indd 200 6/27/2008 4:19:43 PM6/27/2008 4:19:43 PM
Growth and Characterization of Structural and
Optical Properties of Polar and Non-polar GaN Quantum Dots
B. Gayral and B. Daudin
CEA-CNRS group “ Nanophysique et Semiconducteurs ”, CEA-Grenoble, INAC/SP2M,
17 rue des Martyrs, 38054 Grenoble Cedex 9, France
6.1 Introduction
GaN has become a recent member of the quantum dot-forming semiconductor materials, as a
colateral consequence of the huge effort devoted during the past ten years to III-N semiconduc-
tor materials, in relation to their potentialities for light emission/detection in a wavelength range
spanning from IR to UV.
More generally, the interest in GaN and GaInN quantum dots (QDs) is motivated by several fea-
tures specifi c to the nitride family. First of all, and although the situation is rapidly changing due
to a continuous fl ux of newcomers in the business of bulk, self-standing, GaN and AlN wafers pro-
duction, the high cost of these products means that most nitride devices are presently grown on
sapphire or SiC. Due to the lattice mismatch, nitride thick layers and heterostructures grown
on these substrates usually contain about 10
7
to 10
10
dislocations/cm
2
, depending on growth
technique. Despite such a large density of defects, it will be shown that GaN QDs, due to their
reduced size, are free of structural defects, contrary to quantum wells (QWs), and consequently
exhibit remarkable optical properties. At this stage, it should be noted that the high effi ciency,
demonstrated since the mid-1990s, of light-emitting diodes (LEDs) and laser diodes (LDs) with
InGaN QWs in the active layer is far from being fully understood. The origin of such an intense
luminescence is currently assigned to the presence of localization centres in InGaN QWs, the
nature of which is still controversial [1] . Whatever the nature of this localization is, the experi-
mental evidence is that quantum dot-like centres are present in the InGaN QWs and that they
are responsible for both a reduced diffusion of carriers to non-radiative centres and an enhanced
radiative recombination, further motivating an interest in intentional growth of InGaN QDs for a
better control of LED optical properties.
One specifi city of nitrides is the large band gap difference between InN (0.7 eV), GaN (3.47 eV)
and AlN (6.2 eV). As a consequence, carrier confi nement in GaN (GaInN) QDs embedded in AlN
or in AlGaN (GaN) is expected to be effi cient at room temperature, with no signifi cant thermal
escape, which potentially opens the way to QD-based practical devices.
As extensively reviewed in Chapter 7, an additional specifi city of nitride semiconductors with
respect to other families is the presence of an internal electric fi eld along the [0001] polar direc-
tion, as a result of the non-centrosymmetric wurtzite structure. This electric fi eld, which is a
combination of spontaneous and piezoelectric polarization, may be as high as 9–10 MV/cm in
the case of GaN/AlN heterostructures [2] . As a consequence, both a red shift of the photolumi-
nescence (PL) and a very long PL decay time resulting from fi eld-induced carrier separation limit
CHAPTER 6
CH006-I046325.indd 201CH006-I046325.indd 201 6/27/2008 4:22:10 PM6/27/2008 4:22:10 PM
202 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
the effi ciency of practical devices. Non-polar nitride heterostructures are a current option to over-
come these diffi culties: for this purpose, m -plane or r -plane sapphire as well as [11
2
0] and [1
1 0 0 ]
SiC are used to grow non-polar [11
2
0] and [1
1 00] GaN and GaN/AlGaN heterostructures.
It is our aim to review here the growth and structural characterization of [0001] and [11
2
0]
GaN QDs. As a general overview of nitride QDs optical properties is given in Chapter 7,
the present one will be rather focused on single dot spectroscopy, with the fi nal prospect of pro-
viding the interested reader with a review of the state of the art on GaN QDs, by combining
Chapters 6 and 7.
6.2 Epitaxial growth of nitrides
6.2.1 Generalities
Up to very recently, the lack of bulk nitride substrates, which has often been invoked as a serious
limitation for the structural quality of the c -plane epilayers, is a particularly relevant issue for
the case of the a -plane ones. To date, very few publications report on successful growth of bulk
a -plane nitride layers exhibiting reasonably good structural quality. In a recent article, however,
Paskova et al. managed to grow low-defect-density homoepitaxial a -plane GaN on bulk a -plane
GaN substrates, exhibiting a dislocation density in the 10
5
cm
2
range. These substrates, originat-
ing from Kyma technologies, were obtained by slicing a c -plane HVPE-grown GaN boule parallel to
the [11
2
0] plane [3] . However, this type of substrate, although already available to some extent,
is not yet widely used by the GaN community, and the vast majority of groups are still using pseu-
dosubstrates, e.g. substrates with a nature different from that of the epitaxial layer, to perform
growth.
Two types of pseudosubstrates are currently used for the heteroepitaxial growth of a -plane
GaN or AlN layers: [1
1
02] r -plane sapphire [4–10] and [11
2
0] a -plane SiC [11–14]. This type
of growth results in a high density of structural defects such as stacking faults ( ⬃ 10
5
c m
1
) and
dislocations ( ⬃ 10
9
c m
2
) [3, 4, 14] , an issue that has been partially solved by lateral epitaxial
growth techniques employing SiO
2
stripe patterns [15] . To date, a -plane GaN QDs have been
achieved only on a -plane SiC; this is why we will focus our attention on the works carried out on
this type of substrate.
Although MOCVD is generally used for the commercial growth of nitrides, most of the results
reviewed in this chapter refer to MBE. The reason for this is familiar to growers: MBE being a low
pressure technique allows one to use refl ection high-energy electron diffraction (RHEED) in order
to in situ control the growth process at a fraction of the monolayer scale. Such a tool is of course
of particular interest for the growth of QDs and has also been extensively used for studying the
specifi c GaN growth kinetics.
As a matter of fact, nitride growth mechanisms are strongly dependent on experimental con-
ditions. Regarding the c -plane, two different GaN growth regimes have been identifi ed as a func-
tion of the Ga fl ux (with the N fl ux kept constant), namely the N-rich regime (with an excess of
N with respect to Ga) and the Ga-rich regime (with an excess of Ga with respect to N) [16] . In the
latter, an auto-surfactant effect of Ga (a segregation of Ga on the growing GaN surface) has been
put in evidence [17] , and a coverage diagram, describing the Ga surface coverage during growth
as a function of Ga fl ux and growth temperature, has been established [18] .
Importantly, it has to be emphasized that the nature of the active N itself, i.e. ammonia or N
plasma, drastically infl uences growth kinetics. Actually, the above considerations on the growth
diagram as well as most experimental results reported in this chapter refer to plasma-assisted
molecular beam epitaxy (MBE), active N being obtained by radio-frequency cracking of N
2
in a
plasma cell.
It should be noted that the situation is different when active nitrogen results from cracking of
ammonia. To our knowledge, no systematic study of Ga coverage has been performed in such a
case but it is expected that adatom kinetics be drastically infl uenced by residual H resulting from
ammonia decomposition.
CH006-I046325.indd 202CH006-I046325.indd 202 6/27/2008 4:22:10 PM6/27/2008 4:22:10 PM
Growth and Characterization of Structural and Optical Properties 203
Regarding the a -plane, it has been established that growth also strongly depends on Ga/N fl ux
ratio value. More precisely, it has been found that GaN growth in Ga-rich conditions results in a
rough surface whereas growth in N-rich conditions leads to a smooth surface, as supported by
the AFM images of [11
2
0] GaN grown on [11
2
0] SiC shown in Fig. 6.1 . It is worth emphasizing
that this situation is opposite to what is observed for polar nitrides. It suggests that surface dif-
fusion barriers are lower in N-rich conditions than in Ga-rich conditions for [11
2
0] GaN, which
again is the opposite of what is theoretically predicted [19] and experimentally observed [20] in
the case of [0001] GaN growth.
Figure 6.1 (a) (11–20) GaN grown in N-rich conditions. Rms roughness: 1.6 nm; (b) in Ga-rich conditions. Rms
roughness: 2.7 nm. The thickness is 100 nm.
Figure 6.2 shows the Ga coverage diagram experimentally established for homoepitaxy of
[0001] and [11–20] GaN. Whereas a self-regulated 2 MLs Ga coverage may be found for a proper
Ga/N ratio value for an [0001] GaN surface [18, 21] , no self-regulated regime is observed on an
[11
2
0] surface [22] . The Ga coverage diagram in Fig. 6.2 allows one to defi ne several regimes, i.e.
different Ga/N ratio values for a fi xed active N fl ux.
Regarding Fig. 6.2 a which corresponds to [0001] GaN, four regimes limited by experimen-
tally determined lines can be identifi ed: regime A corresponds to under-stoichiometry (N-rich)
conditions which result in a rough surface, consistent with the reduced adatom diffusion path
[19] . Regime B corresponds to the dynamic equilibrium on the surface of a Ga excess comprised
between 0 and 1 ML. Regime C corresponds to a self-regulated Ga coverage of exactly 2 MLs
[18, 21] . For higher Ga fl ux, Ga accumulation and Ga droplet formation are observed. Note that
for a given substrate temperature, progressively increasing Ga fl ux allows one to successively
experience the four regimes under consideration. Also note that the extension of each regime
along the Ga fl ux axis is strongly dependent on growth temperature. In particular, for a sub-
strate temperature below 680°C, only two regimes are observed, namely the N-rich one and the
Ga-accumulation one.
The case of [11
2
0] GaN is illustrated in Fig. 6.2 b. Note that this diagram has been established
in static conditions, by measuring the desorption time of Ga from an [11
2
0] surface, following
its exposure to increasing Ga fl ux. The full line in the Figure corresponds to the GaN growth rate
at about 730°C which was simply added to static data in order to determine the Ga coverage dia-
gram during growth [22] . In this diagram, A regime corresponds to nitrogen-rich conditions
which lead to a smooth surface, as shown in Fig. 6.1 a nd contrary to the case of [0001] GaN.
Regime B corresponds to the dynamical equilibrium on the surface of a Ga excess comprised
between 0 and 1 ML, followed by Ga accumulation for higher Ga fl uxes. Note that only three
regimes are observed in the case of an [11
2
0] GaN surface. In particular, no regime correspond-
ing to a self-regulated coverage of Ga has been found for this cristallographic orientation.
Figure 6.2 proved to be very useful to identify the suitable growth conditions to achieve the
Stranski–Krastanov growth mode of GaN on AlN giving rise to the formation of QDs, and to
understand how the growth mode can be controlled by playing with the excess of Ga.
CH006-I046325.indd 203CH006-I046325.indd 203 6/27/2008 4:22:11 PM6/27/2008 4:22:11 PM
204 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
In the case of heteroepitaxial growth of [0001] GaN on [0001] AlN, it has been established
that regimes A and B lead to the formation of GaN QDs, according to the Stranski–Krastanov
growth mode. By contrast, in regime C, the presence of a Ga bilayer results in stabilization of a
2D GaN layer and completely inhibits QD formation. However, it has been shown that by stop-
ping GaN growth after deposition of a limited number of MLs and by letting the surface evolve
under vacuum, desorption of the Ga bilayer leads to an instability of 2D GaN and to its spontane-
ous reorganization into 3D islands on a wetting layer: this is the modifi ed SK growth mode, which
results in the formation of GaN QDs similar to those obtained in regimes A and B [23, 24] .
In the case of [11
2
0] GaN grown on [11
2
0] AlN, regime A leads to the formation of a 2D
layer and can be used to grow quantum wells. By contrast, regime B leads to the formation of
QDs according to the SK growth mode [25] .
It should again be emphasized that QW and QD growth conditions are reversed for [0001]
and [11
2
0] orientations, possibly due to a change in adatom mobility on the [11
2
0] surface as
a function of Ga excess. As an alternative explanation of the observed features, it is also possi-
ble that surface energy of (11
2
0) surfaces in Ga-rich conditions be suffi ciently weak to make dot
670 680 690 700 710 720 730 740 750 760
1
C
B
Ga
accumulation
A
670 680 690 700 710 720 730 740 750 760
1
A
B
Ga
accumulation
Substrate temperature (°C)
Ga flux (ML/s)
(a)
(b)
1
Figure 6.2 Ga fl ux dependence of the growth diagram for homoepitaxy of GaN, as a function of substrate
temperature. (a) for [0001] GaN. The N fl ux is 0.35 ML/s. Four regions are identifi ed: A: nitrogen-rich regime; B:
dynamically stable Ga coverage between 0 and 1 ML; C: self-regulated Ga bilayer regime; Ga accumulation regime
(b) for [11 2
–
0] GaN. The N fl ux is 0.28 ML/s. Three regions are identifi ed: A : nitrogen-rich regime; B : dynamically
stable Ga coverage between 0 and 1 ML; Ga accumulation regime. The frontier line between A and B corresponds to
the GaN growth rate at about 730°C.
CH006-I046325.indd 204CH006-I046325.indd 204 6/27/2008 4:22:11 PM6/27/2008 4:22:11 PM
Growth and Characterization of Structural and Optical Properties 205
formation an energetically favourable process, whereas a supposed high surface energy in N-rich
conditions would favour the stabilization of the 2D GaN layer.
6.2.2 Growth of GaN QDs
Due to the 2.5% lattice mismatch between GaN and AlN, growth of GaN deposited on AlN obeys
the Stranski–Krastanov (SK) mode, with formation of three-dimensional islands for a GaN
amount larger than a critical thickness of about 2.3 monolayers (MLs) [26] . However, occur-
rence of this growth mode strongly depends on the used Ga/N ratio value, i.e. on surface kinetics.
It has been discussed in a previous section that SK growth mode is observed when growing GaN
in N-rich conditions, that is in conditions of low adatom mobility [25] . Above stoichiometry, for
Ga/N 1, QD formation is rapidly followed by coalescence, which is assigned to an increase in
Ga diffusion length on a Ga-rich surface. Next, in the extreme case of Ga/N 1, a Ga bilayer
is rapidly formed on growing GaN. The existence of this self-regulated Ga bilayer has been dem-
onstrated both theoretically [21] and experimentally [18] . It results in the opening of a new dif-
fusion path for N between the two Ga layers and a deep change in surface kinetics. When the Ga
fl ux is high enough so that this Ga bilayer is completed before reaching the critical thickness, the
2D/3D transition is found to be completely inhibited and no dot formation is observed.
Actually, when stopping the GaN growth after a few MLs and letting the sample evolve under
vacuum, it has been demonstrated that Ga bilayer desorption is followed by a reorganization of
the GaN bidimensional layer and the formation of GaN QDs similar in shape and morphology to
those grown in the N-rich regime, which suggests that the inhibition of the transition observed in
the Ga/N 1 regime is not of a kinetic nature but rather results from the competition between
elastic and surface energy [24] . It was proposed that this peculiar growth mode, namely the
growth of a 2D, Ga bilayer-stabilized, GaN followed by dot formation under vacuum be named
the modifi ed SK growth mode to differentiate it from the former one.
As mentionned above, the nature of active N source in MBE is a determinant parameter. Along
this view, it should be noted that the above considerations about SK and modifi ed SK growth mode
of GaN QDs on AlN are only valid in the case of plasma-assisted MBE. When using ammonia as
an N source, it has been demonstrated that QD formation only occurs on growth interruption
[27] , suggesting that GaN 2D growth is stabilized in the presence of ammonia, in a way phenom-
enologically similar to what is found to occur in the presence of a Ga bilayer.
As a remarkable consequence of the growth mode dependence as a function of the Ga/N ratio
value, Fig. 6.3 illustrates the possiblity of growing either GaN QDs or QWs, at a given substrate
temperature, by simply tuning the Ga fl ux.
QD
QW
8 nm
Figure 6.3 Quantum wells and quantum dots of [0001] GaN grown at 750°C and embedded in AlN. The QW was
grown in regime C, the QDs in regime B (see Fig. 6.2 a).
Consistent with the speculations about adatom diffusion as a function of Ga/N in the case
of [11
2
0] GaN (see Fig. 6.2 ), the conditions for growing [11
2
0] GaN QDs and QWs have been
found to be Ga-rich and N-rich, respectively, which is contrary to what is observed for [0001]
GaN. As expected from the Ga coverage diagram shown in Fig. 6.2 , no self-regulated Ga coverage
regime is present for [11
2
0] GaN and no modifi ed SK growth mode was found.
Figure 6.4 shows high-resolution transmission electron microscopy (HRTEM) pictures of [11
2
0]
GaN QDs and QWs.
CH006-I046325.indd 205CH006-I046325.indd 205 6/27/2008 4:22:12 PM6/27/2008 4:22:12 PM
206 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
As a conclusion of this section, it appears that GaN QD formation strongly depends on kinetic
conditions. It should be noted at this stage that the inhibition of the 2D/3D transition due to the
self-surfactant effect of Ga is not limited to Ga but has also been observed in the case of [0001]
GaN surface saturation by Eu. Furthermore, in the case of ammonia MBE, it has been found that
[0001] GaN QD formation is only observed when growth is stopped and the material is allowed
to evolve under vacuum. From all these results it is inferred that the occurrence of 2D/3D transi-
tion in the case of a GaN/AlN system is governed to a large extent by surface energy, in addition
to the effect of lattice mismatch.
6.2.2.1 Growth of [0001] GaN QDs
Growth of GaN QDs on AlGaN by MOCVD was fi rst reported by Tanaka et al . [28] . In that work,
the reduced lattice mismatch due to low Al content of the AlGaN buffer was insuffi cient to pro-
mote the SK growth mode. Then, the surface was seeded with Si in order to achieve a micro-
masking which resulted in localized GaN growth.
The SK growth mode of GaN on AlN by MBE was fi rst observed in 1997 by Daudin et al . [26]
and next by Damilano et al . [29] . More recently, it was demonstrated that the SK growth mode of
GaN on AlN could also be observed in MOCVD [30] and for MBE growth of GaN on AlGaN with
an Al content larger than about 50% [31] .
Combining in situ RHEED observations and HRTEM images such as the one shown in Fig. 6.5 ,
it has been established that [0001] GaN QDs are truncated hexagonal pyramids with {1
1
03 }
facets.
Figure 6.4 (a) [11 2
–
0] GaN QDs embedded in AlN. Growth in regime B . (b) QWs grown in regime A (see Fig. 6.2 b).
Figure 6.5 HRTEM observation along [2, 1
–
, 1
–
,0] of a GaN QD embedded in AlN. The 2 H hexagonal stacking is
clearly visible. The distance between each plane is c /2.
CH006-I046325.indd 206CH006-I046325.indd 206 6/27/2008 4:22:13 PM6/27/2008 4:22:13 PM
Growth and Characterization of Structural and Optical Properties 207
A major issue in GaN QD nucleation, related to its optical properties, is the role of structural
defects. In particular, one may wonder whether threading edge dislocations as well as screw and
mixed ones, the density of which is typically 10
7
to 10
10
c m
2
in AlN or GaN templates, play a
role.
To adress this issue, plan view HRTEM observations have been made. As shown in Fig. 6.6 , it
appears that most QDs nucleate close to a threading edge dislocation, due to the local distortion
of the lattice. However, the dots themselves are always free of dislocations and can be viewed as a
nanoisland of perfect material embedded in a very defective matrix [32] .
0 0.25 0.50 0.75 1.00 0 0.25 0.50 0.75 1.00
0
0.25
0.50
0.75
1.00
(a) (b)
µM µM
0
0.25
0.50
0.75
1.00
Figure 6.7 (a) AFM images of GaN QDs with a characteristic loop-like planar distribution associated with the
distribution of TDs. (b) AFM image of 1 mm thick GaN grown on [0001] SiC showing threading edge dislocations
at the edge of atomic steps.
Figure 6.6 HRTEM plan view of a GaN QD embedded in AlN. Note the threading edge dislocation by the side of the
dot, acting as a nucleation centre (after Fig. 6.7 [32] ).
Further atomic force microscopy experiments shown in Fig. 6.7 reveal an inhomogeneous dot
distribution at the microscopic scale. Consistent with results of Fig. 6.6 , this loop-like arrange-
ment has been assigned to the inhomogeneous distribution of TDs which are shown (see Fig.
6.7 b) to decorate the edges of steps on GaN and AlN surfaces.
CH006-I046325.indd 207CH006-I046325.indd 207 6/27/2008 4:22:14 PM6/27/2008 4:22:14 PM
208 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
6.2.2.2 Capping of [0001] GaN QDs
It has been found that the capping of [0001] GaN QDs by AlN is associated with a homogeneous
decrease in size, of about one monolayer (see Fig. 6.8 ). Further Rutherford backscattering spec-
trometry (RBS) experiments have demonstrated that this size reduction is not due to dot–barrier
interdiffusion but is rather assigned to a vertical exchange between Ga and Al, due to the higher
stability of AlN at temperatures used for QD growth [33] .
2 nm
(a)
(b)
(c)
(d)
(e)
4 ML 14 ML
Figure 6.9 HRTEM image of [0001] GaN QDs (bright) covered with 4 and 14 ML of AlN (dark), respectively.
Note the 2 ML thick GaN marker on AlN capping, allowing one to visualize AlN morphology (after [34] ).
1/3[2-1-10]
~ 30°
AlN
GaN
5nm
Figure 6.8 HRTEM image of [0001] GaN QDs on a surface and embedded in AlN. Note the homogeneous size
reduction of the embedded dot with respect to the one on the surface (after [33] ).
More precisely, a detailed study of GaN QD capping by AlN has shown that, following an initial
wetting of dots with AlN up to 4 ML, further AlN growth proceeds in between dots up to 14 MLs
which corresponds to AlN smoothing. As a whole, the AlN capping mechanism is schematized
in Fig. 6.9 [34] . Starting from uncapped GaN QDs (a), their wetting by AlN is fi rst observed (b).
This stage is associated with the Ga/Al vertical exchange illustrated in Fig. 6.8 . Next, homogene-
ous wetting of dots proceeds up to 4 MLs of AlN deposited (c). Further AlN growth preferentially
occurs in between dots up to AlN smoothing (d). In a next stage, 2D homogeneous AlN growth
is observed (e). No change in dot shape and no noticeable interdiffusion are associated with cap-
ping, making the GaN/AlN system a model one and facilitating the study of GaN QD structural
properties, as extensively discussed in next sections.
CH006-I046325.indd 208CH006-I046325.indd 208 6/27/2008 4:22:15 PM6/27/2008 4:22:15 PM