Henini M. Handbook of Self Assembled Semiconductor Nanostructures for Novel devices in Photonics and Electronics
Подождите немного. Документ загружается.
Radiation Effects in Quantum Dot Structures 409
spectra of the bulk CdS
1
x
Se
x
crystals remain unchanged even after irradiation with an electron
fl uence as high as 10
17
c m
2
, it is hardly possible that the considerable changes of the spectra
observed in the QDs samples could result from the radiation defects created at fl uences of
10
13
–10
14
c m
2
.
13.2.4 Quantum dots embedded in a superlattice
Recently, the radiation hardness of InAs QDs embedded in a short-period AlAs/GaAs SL and of
SiGe/Si QDs embedded in a Si/Ge SL was shown to be even higher than that of QDs in a bulk
GaAs or Si barrier, respectively [81, 82] .
To understand these results, it is important to consider that thin-layer Si/Ge SLs exhibit an
enhanced radiation hardness of the PL as compared to Si/Ge QWs and to bulk Si. In the experi-
ments described in [76] , SLs with periods of 10 or 15 monolayers (Si
6
Ge
4
and Si
9
Ge
6
) and double
QWs Ge
n
Si
20
Ge
n
( n 2 or 4) were used. However, the exciting laser light was absorbed mainly in
the SiGe waveguide layer situated on top of the samples, so that its degradation upon irradiation
might have affected the observed result. Therefore, analogous experiments were repeated using
electrical injection. In Fig. 13.18 , the relative degradation of the EL from fully processed QW
and SL diodes is shown. The diodes were exposed to one and the same electron fl uence and the
current density employed in the measurement was kept the same before and after irradiation.
The Figure corroborates the PL data of [76] : whereas the EL intensity of the QW diode drops
below the noise level after electron irradiation, the SL emission is still quite detectable. This means
120
100
60
0
40
20
80
120
100
60
0
40
20
80
1.8 1.9
2.0
1.8
1.9
2.0
2.1
2.1
2.2 2.5 2.62.3 2.4
2.2 2.5 2.6 2.7 2.8 2.92.3 2.4
hν (eV)
hν (eV)
(a)
(b)
Non-irradiated
Non-irradiated
CdS
0.22
Se
0.78
CdS
0.4
Se
0.6
(cm
1
)
10
12
cm
2
10
12
cm
2
10
14
cm
2
10
14
cm
2
10
13
cm
2
10
13
cm
2
10
15
cm
2
(cm
1
)
Figure 13.17 Optical absorption spectra of CdS
0.4
Se
0.6
(a) and CdS
0.22
Se
0.78
(b) QDs in borosilicate glass matrix,
irradiated at RT with 10 MeV electrons [79] .
CH013-I046325.indd 409CH013-I046325.indd 409 6/24/2008 3:45:34 PM6/24/2008 3:45:34 PM
410 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
that we observe an enhanced radiation hardness of the very SL. The following model has been
proposed to explain the effect [76] . As mentioned above, when Si is irradiated with 3–4 MeV
electrons at room temperature, most of the stable radiation defects are formed after long-range
migration of the primary radiation defects, i.e. vacancies and self-interstitials, by subsequent
interaction and formation of complexes with impurities. At doping levels above ⬃ 10
17
c m
3
the
creation rate of the stable defects becomes independent of the doping level and approaches the
primary displacement rate since each primary defect is then captured by an impurity atom [84] .
An analogous picture may basically be assumed for bulk Ge. Hence, even for an impurity concen-
tration as high as 10
19
c m
3
the average distance between neighbouring impurities is still of the
order of 40 Å, and this determines the mean migration length of a primary defect to form a non-
radiative centre. At the same time, the mean diffusion length to reach an interface in a short-
period SL is a few Ångströms only. The interfaces act as sinks and annihilation centres for the
mobile primary radiation defects, thus leading to a lower concentration of non-radiative centres
than in the bulk material.
It should be noted that the role of an SL as a diffusion stop layer in MBE growth is well known.
An SL buffer can effectively getter defects from the substrate and prevent their propagation into
the overlying epitaxial material [433] . Quite recently, a GaAs/AlGaAs SL was again found to min-
imize the defect diffusion into the QD active region during overgrowth at moderate temperatures
[436] . A damage-reducing effect of an intervening SL placed in the surface barrier region was
observed in QW samples subjected to low-energy ion bombardment [434] .
13.2.5 Amorphizing damage
The infl uence on the radiation hardness of reducing the size to nanodimensions is not so unam-
biguous in the case of heavy damage leading to amorphization (usually this is the case of ion
irradiation). So, bulk zirconia ZrO
2
exhibits no evidence of irradiation-induced amorphization at
140014001200 13001200
Intensity (rel. units)
SL
NP
0
5 10
16
e/cm
2
QW
NP
QW
TO
(a)
(b)
(c)
(d)
gain 10
(nm)
Figure 13.18 EL spectra of an Si
6
Ge
4
SL and an Ge
2
Si
20
Ge
2
QW prior to (a, c) and after (b, d) irradiation with
5 1 0
16
c m
2
of 3–4 MeV electrons. The spectra after irradiation were recorded with a gain 10. T
meas
4.2 K.
(a, b) SL, current density j 6.4 A/mm
2
. (c, d) QW, j 2 . 8 A / m m
2
. SL and QW are characteristic luminescence
bands of the SL and QW, respectively. The NP and TO indices refer to no-phonon transitions and their TO-replica,
respectively [83] .
CH013-I046325.indd 410CH013-I046325.indd 410 6/24/2008 3:45:34 PM6/24/2008 3:45:34 PM
Radiation Effects in Quantum Dot Structures 411
doses as high as 110 displacements per atom (dpa). However, ⬃ 3 nm diameter ZrO
2
nanocrystals
(NCs) can be amorphized at a dose corresponding to only 0.9 dpa [85] . Interestingly, in a con-
trol experiment the authors failed to amorphize 3 nm precipitates of Au embedded in amorphous
SiO
2
. The results have been explained by the delicate balance between the bulk free energy and
surface free energy of various zirconia polymorphs and the defect free energy introduced by ener-
getic ions. The excess surface free energy in nanocrystalline zirconia tips the balance in favour of
radiation-induced amorphization. The thermodynamic properties of Au are very different from
those of ZrO
2
: high-temperature metastable phases do not exist, the surface free energy is rela-
tively low, and besides defects can readily recombine due to lattice site equivalence.
The behaviour of Au NCs is not surprising as elemental metals usually cannot be amorphized
via radiation damage, even at low temperatures. However, very recently the authors of [441]
succeeded in amorphizing, at 77 K, ⬃ 2.5 nm small Cu NCs embedded in SiO
2
, after they failed to
do so with ⬃ 8 nm NCs [442] . The different behaviour of the NCs has been partly attributed to
their different size: in the fi rst case half the Cu atoms and in the second case less than 15% of
them reside on the NC–matrix interface.
Similarly, Ge NCs embedded in SiO
2
turned up to be extremely sensitive to ion irradiation and
are rendered amorphous at an ion fl uence ⬃ 40 times less than that required to amorphize bulk,
crystalline standards. This rapid amorphization was attributed to the higher-energy nanocrys-
talline structural state prior to irradiation, inhibited Frenkel pair (i.e. vacancy plus interstitial)
recombination when Ge interstitials are recoiled into the SiO
2
matrix and preferential nucleation
of the amorphous phase at the nanocrystal/matrix interface [86] .
Si NCs in SiO
2
could be amorphized by irradiation at RT either with 30 or 130 keV He ions
or with 400 keV electrons, contrary to bulk Si, in which electron and very light ion irradiation
never leads to amorphization at RT [87] . The effect was explained by an accumulation of point
defects produced by irradiation at the NC surface, leading to amorphization at low displacement
rates (0.1–0.2 dpa). Subsequent publications confi rmed the particular sensitivity of the Si NCs in
SiO
2
to irradiation [88, 89] .
Upon 60 MeV carbon ion irradiation, the defect accumulation rate in nano-Au at 15 K is larger
than that in poly-Au, but for irradiation at 300 K it becomes much smaller than that in poly-Au
[51] . In a subsequent work the authors came to the conclusion that both the cross-sections for defect
annihilation and production in nano-Au increase monotonically as the grain size decreases [90] .
13.2.6 Low-dose effect
There are several reports on an enhancement of the PL intensity in QD structures or of the QD
laser performance due to irradiation with low fl uences of electrons, protons or ions with energies
being high enough to cause the creation of displacement-type defects [62–64, 91–94] . As an
explanation, an increase in the effi ciency of photocarrier transfer into the QDs via defect-induced
energy levels [63] , [64] , reduction of grown-in defects around the QD regions [92] or even
removal of the cap layer due to sputtering [94] have been proposed. A similar “ low-dose effect ”
has been formerly observed in Si [95] , CuInS
2
[96] and GaAs [97] and interpreted as a result of
the dissociation of grown-in defects assisted by the creation of mobile point defects by the irra-
diation. From the viewpoint of the thermodynamics, the low-dose effect manifests a shift of an
imperfect crystal that is in a metastable state after growth to the thermodynamic equilibrium. On
the contrary, high doses of radiation push the crystal off the equilibrium, whereas post-radiation
annealing restores the equilibrium. As the low-dose effect is usually observed in a rather nar-
row fl uence range, the following method of the crystal quality improvement of CuInS
2
crystals
has been proposed in [96] : “ high dose ” irradiation with a posterior annealing at the optimum
temperature. As we shall see in Section 13.4.1, such a method is successfully used to improve the
characteristics of QD and QW structures.
13.2.7 Irradiation with electrons of subthreshold energies and x-rays
Irradiation that is unable to create damage in elastic collisions causes, nevertheless, persistent
radiation effects in QDs based on wide-gap semiconductors.
CH013-I046325.indd 411CH013-I046325.indd 411 6/24/2008 3:45:34 PM6/24/2008 3:45:34 PM
412 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
The luminescence intensity of CdS NCs embedded in a glass matrix degrades upon X-ray and
even light irradiation. The photoinduced defects were found to be different from those induced by
X-rays and were supposed to be located in glass near the NCs [98] . The PL intensity of CdSe QDs
gradually decreases upon laser illumination up to a certain exposure and then remains constant,
contrary to the PL intensity of ZnCdSe QWs which keeps decreasing [99] . The absorption edge
blue shift was observed in glass-embedded CdS [100] and CdS
1
x
Se
x
[101, 102] QDs under X-ray
irradiation as well as in CdSe QDs under intense light illumination [100] , though the transfor-
mation of the confi nement-related optical absorption maxima was reported only in [101] and
[102] and resembled the behaviour under MeV electron irradiation described above (see Section
13.2.3). A strong PL and CL degradation of semiconductor QD composites, formed by highly
luminescent (CdSe)ZnS core–shell NCs embedded in a ZnS matrix, was observed with time [103] .
PL experiments carried out at high laser fl uences (0.5–10 mJ/cm
2
per pulse) showed that the PL
intensity decay with illumination time depended on the size of the NCs and the nature of the sur-
rounding matrix. For instance, close-packed fi lms showed a much slower decay than composite
fi lms. The CL intensity degradation is enhanced at lower temperatures. Partial recovery of the CL
signal could be achieved after thermal annealing at temperatures around 120°C, which indicates
that activation of trapped carriers can be induced by thermal stimulation. The CL and PL decay
in the composite fi lms was attributed to photo- and electroionization of the NCs and subsequent
trapping of the ejected electrons in the surrounding semiconductor matrix [103] .
Electron irradiation with an energy as low as 10 keV quenches the CL in GaN/AlGaN QD
structures [104–107] . It should be noted that it is not surprising because the CL degradation had
formerly been observed in regions with zero dislocations of epitaxial laterally overgrown GaN
(ELOG) layers also under 10 keV electron excitation [108] . The QD CL intensity degradation was
attributed to an increase of the non-radiative recombination due to a recombination-induced
diffusion of pre-existing defects [105] . A comparison of the radiation hardness of heterostruc-
tures containing GaN/AlGaN QDs, QWs and AlGaN barriers together with that of a thick ELOG
layer revealed a much higher resistance of the QDs, especially at high electron beam current
densities [106] .
13.2.8 Hydrogen passivation
An effect that is in its essence opposite to the defect introduction and is often exploited in defect
studies is the hydrogen passivation of the grown-in non-radiative recombination centres like dis-
locations and point defects [109] . A PL intensity enhancement due to the hydrogen passivation
has been reported for the In(Ga)As/GaAs [110–116] , InAs/AlAs [117] , Ge/Si QDs [78] , as well
as for Si nanocrystals embedded in fused silica [118–121] or silicon nitride [122] . This effect has
to be taken into account when, upon proton irradiation, the implanted hydrogen profi le is placed
over the active device layers.
13.2.9 Infl uence of defects on the thermal stability of the luminescence
It is well known that the thermal stability of the QD luminescence suffers from the presence
of defects that act as non-radiative recombination centres (see, e.g., [110] ). The temperature
dependence of the PL intensity in Ge/Si QDs without any treatment (as grown) as well as sub-
jected to 2.4 MeV proton irradiation or atomic hydrogen passivation is illustrated in Fig. 13.19 .
It is clearly seen that the irradiation deteriorates, and the passivation improves the thermal sta-
bility of the PL. The latter is only possible if the as-grown dots contain defects. Indeed, the Ge/Si
QDs studied in [78] had a basis diameter of ⬃ 200 nm, so that the effect of “ self-purifi cation ” [40]
mentioned in Section 2.1 could not occur. In InAs/GaAs QDs, a growth treatment using tetra-
chloromethane eliminated the quenching of the PL intensity up to RT [123] . A photocarrier sta-
tistical model taking into account the variations of the quasi-Fermi level position of the minority
carriers, which are related to the concentration of trapping centres in the GaAs matrix, was
developed [125]. When defects were introduced through proton irradiation, the PL quenching
at RT appeared again [124, 125] . The calculated results for the PL intensity reproduced well the
experimentally observed trends [125] .
CH013-I046325.indd 412CH013-I046325.indd 412 6/24/2008 3:45:34 PM6/24/2008 3:45:34 PM
Radiation Effects in Quantum Dot Structures 413
13.3 Radiation hardness of QD lasers
There is an extensive literature on radiation effects in heterostructure and QW laser diodes (see,
e.g., [48, 126–136] ), so that the logic of the radiation damage in these devices is well under-
stood. From an application standpoint the most important parameters for a semiconductor laser
diode are threshold current and slope effi ciency. Figure 13.20 shows the effect of 5.5 MeV pro-
ton irradiation on the optical power of the strained InGaAs/GaAs QW laser diode operating near
980 nm as the forward current increases from very low currents to the region where laser opera-
tion begins. The optical power increases abruptly after the forward current reaches the threshold
current. The latter grows with increasing proton fl uence. On the contrary, the slope effi ciency
above threshold is independent of fl uence [127] .
6
4
2
0.100.050.00
0.8
0.15 0.20
0.4
0.2
0.6
1
Passivated
Irradiated
As grown
Intensity (a.u.)
1/T (K
1
)
Figure 13.19 Temperature dependences of the PL intensity in the as-grown, passivated and 2.4 MeV proton
irradiated (2 1 0
12
c m
2
) Ge/Si QD samples. The curves have been vertically shifted to avoid superposition of the
points [78] .
Degradation in lasers is usually much greater at low injection conditions, when the laser
operates in the LED mode. The light output (due to radiative recombination below threshold) is
reduced far more at currents below the threshold current compared to the changes that occur
in threshold current when the device enters the laser mode, and light output is dominated by
stimulated emission [136] .
The radiation hardness of In(Ga)As/GaAs QD lasers was investigated for the irradiation with
phosphorous ions [91] and protons [137] in comparison with QW lasers. It was found that
20
0
010 40 7050 6020 30
5
10
15
Current (mA)
Optical power (mW)
Before irrad.
4.6 10
11
cm
2
1.6 10
13
cm
2
7.8 10
12
cm
2
3.9 10
12
cm
2
2.1 10
12
cm
2
Figure 13.20 Degradation of L–I characteristics of a strained InGaAs/GaAs QW laser diode during 5.5 MeV
proton irradiation [127] .
CH013-I046325.indd 413CH013-I046325.indd 413 6/24/2008 3:45:35 PM6/24/2008 3:45:35 PM
414 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
laser diodes can sustain signifi cantly greater 8.56 MeV P ion fl uences ( 100) than comparable
QW-based structures prior to the onset of failure.
Let us consider the infl uence of proton irradiation on laser-device characteristics, i.e. thresh-
old, slope effi ciency, internal differential effi ciency and optical loss following [137] .
The active region of the structures under investigation consisted either of a single In
x
Ga
1
x
As
QW ( x ⬇ 0.2) or of a triple stack of InGaAs/GaAs QDs with a dot density of (3–5) 1 0
10
c m
2
.
The lasing wavelength of the processed broad-mesa ridge waveguide lasers was 1096 nm for
the QW laser and 1155 nm for the QD laser. The mesas were etched down to 400 nm above the
active layer. The stripe width was 200 μ m. For a waveguide thickness of 300 nm this resulted in
weak index guiding of the optical modes and also in the suppression of current spreading. Cavity
lengths of the investigated devices were 1.0 and 1.6 mm. The samples were irradiated with
2.4 MeV protons to a fl uence of 2 1 0
13
c m
2
.
Before and after irradiation, the QD laser operated on the ground state. The lasing wavelength
of the QW devices also remained unchanged after irradiation. Figure 13.21 shows the L–I curves
for the QW and QD lasers for 1.6 mm cavity length before and after irradiation. The extracted
device parameters are summarized in Table 13.2 . As shown in Fig. 13.21 , for these two lasers the
threshold current density and the external differential quantum effi ciency (QE) before irradiation
were very similar: 83 A/cm
2
, 68% and 89 A/cm
2
, 61%, respectively. After irradiation the differ-
ential external QE was reduced in both cases by ⬃ 43% for the QW and by 50% for the QD laser.
The change in threshold current density after irradiation was, however, quite different: the value
for the QW laser worsened by about twice as much as for the QD laser, 950 A/cm
2
and 550 A/
cm
2
, respectively. It was naturally concluded from these results that in the regime of spontane-
ous emission incorporated defects introduced by the proton irradiation result in non-radiative
relaxation channels. These defects in or in the vicinity of the active region are much more criti-
cal in the QW case owing to the higher in-plane diffusion of the carriers, contrasted to QDs. As
already discussed in Section 13.2.2, the localized carriers in the QDs have a reduced interaction
Table 13.2 Threshold current densities and external differential quantum effi ciencies for three
laser devices (1 QD and 2 QWs) before (as-grown) and after irradiation with high energy protons
[137] .
QW laser (L 1.0 mm) QW laser (L 1.6 mm) QD laser (L 1.6 mm)
j
thr
(as-gr.) 92 A/cm
2
83 A/cm
2
89 A/cm
2
η
ext
(as-gr.) 76% 68% 61%
j
thr
(irrad.) 1.2 kA/cm
2
950 A/cm
2
550 A/cm
2
η
ext
(irrad.) 44% 39% 30%
0.30
0.0 0.5 1.5 2.0 3.53.02.5
0.00
0.05
0.10
0.15
0.20
0.25
Current (A)
as-grown
Irradiated
QD
QW
Power (pulsed) (W)
Figure 13.21 Optical output power against current in pulsed operation for the same QW and QD lasers before and
after irradiation (1.6 mm cavity length, 200 μ m stripe width) [137] .
CH013-I046325.indd 414CH013-I046325.indd 414 6/24/2008 3:45:35 PM6/24/2008 3:45:35 PM
Radiation Effects in Quantum Dot Structures 415
with the defects. Once the devices lase, the differential quantum effi ciency is reduced owing to
leakage currents in the barrier that are similar for QWs and QDs. The fast stimulated emission in
the active media bypasses non-radiative recombination of localized carriers, resulting in similar
differential quantum effi ciencies for QWs and QDs.
From the external differential QE for both cavity lengths of the QW device, the internal quan-
tum effi ciency and the internal optical losses before and after irradiation can be estimated. Before
irradiation, an internal QE η
int
of 95% and an internal optical loss of 2.9 cm
1
were found to be
in agreement with results obtained from a larger set of samples ( η
int
84% with a 10% error,
α
int
5 c m
1
). After irradiation, 56% and 3.3 cm
1
were found, respectively. Since the internal
optical loss had a similar value before and after irradiation, the reduction in external differential
QE after irradiation was attributed to the reduced internal QE [137] .
13.4 Radiation technology
Radiation treatment can be used to improve the performance of QSSS-based devices or to modify
their characteristics in a desired manner. New structures can be fabricated due to self-organization
upon irradiation.
13.4.1 Intermixing
Spatially selective intermixing is particularly important in the QW laser fabrication (e.g. in
GRINSCH: graded-index separate confi nement heterostructure) [138] . Intermixing the wells and
barriers of QW heterostructures generally results in an increase in the band gap energy (blue
shift of the luminescence band) and is accompanied by changes in the refractive index.
A very effi cient QW intermixing is achieved using particle irradiation with subsequent anneal-
ing [139, 140] . Independently of the amount of intermixing in the collision cascades (which is
virtually absent when only isolated point defects are created, e.g., by proton or electron irradia-
tion), the mixing can be strongly enhanced by post-irradiation heat treatment and occurs due to
defect-enhanced diffusion [141] . The ion irradiation-induced intermixing has several advantages
over the other intermixing techniques: the amount of defects can easily be controlled by the
ion dose and mass, the defects can be placed at a certain depth by the choice of the ion energy,
and, last but not least, the intermixing occurs at lower annealing temperatures than in the
as-grown structures. The latter fact allows a high lateral band gap contrast using focused
ion-beam implantation or selective ion implantation using masks to be achieved.
It is important to note that in order to realize advanced device parameters, the radiation-
induced damage has to be fully eliminated by annealing, which requires a suffi ciently high
annealing temperature. On the other hand, as mentioned above, the temperature should be low
enough to avoid thermally activated band gap shifts in unimplanted areas. As a compromise, an
annealing temperature close to the onset of purely thermally activated band gap shifts is used
[140] . However, this compromise can only be achieved when the implantation damage is not too
heavy, as heavier damage requires, as a rule, higher annealing temperatures. This fact puts an
upper limit to the useful implantation fl uences.
It is well known that compound semiconductors tend to undergo stoichiometric changes and
defect generation upon prolonged heating. To suppress these detrimental processes, the rapid ther-
mal annealing (RTA) in an inert or forming atmosphere, with the sample surface protected against
effusion by an oxide or nitride layer or proximity caps, is usually applied. It is noteworthy that the
protection by oxide or nitride layers can enhance or retard the intermixing by itself [142] . Typical
annealing conditions for III–V material systems are rapid heating to 600–900°C for times ranging
from several seconds to a few minutes [139] . In some papers, even an improvement of the lumines-
cent parameters of III–V QW samples subjected to As ion [143] , alpha particle [144] , and electron
[145] irradiation or Ar plasma exposure [93] with subsequent annealing has been reported.
The ion mass determines the energy transfer in the collisions (see Section 13.1) and, there-
fore, the defect density. Whereas light ions tend to create isolated point defects or point defect
CH013-I046325.indd 415CH013-I046325.indd 415 6/24/2008 3:45:36 PM6/24/2008 3:45:36 PM
416 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
clusters, heavy ones may produce amorphous regions in a single ion track [146] . As a result, the
intermixing depends on the ion mass [147] .
Besides the implantation dose and ion mass, the ion energy is an important technological
parameter, because it governs the profi les of the damage and of the implanted impurity concen-
tration. Depending on the target material, especially on the defect diffusivities in it, either the
total number of created vacancies or the vacancy concentration in the very QW region may be
important [139] .
The target temperature during the implantation may have a profound effect on the intermix-
ing because of the dynamic defect annealing that is mediated by defect diffusion and annihilation
[148] . So, e.g., ion-beam mixing in Al
x
Ga
1
x
As and InP matrices was measured as a function of
target temperature upon 1 MeV Kr ion bombardment [149] . The mixing increased with tempera-
ture up to a critical temperature T
c
at which point it precipitously dropped. T
c
was identifi ed as the
highest temperature at which the matrix could be amorphized by 1 MeV Kr irradiation. Earlier
the ion implantation into heated InGaAs/InP and AlGaAs/GaAs multilayers was found to induce
compositional disordering at signifi cantly lower temperatures than implantation at RT with sub-
sequent annealing [150] .
The dose rate (ion fl ux) and the channelling effect may also affect the intermixing; the fi rst one
as a result of both temporal and spatial overlap of defects within collision cascades [151] and the
second one due to a modifi cation of the density and depth distribution of the radiation damage
[152] . Besides, channelling allows reducing the lateral straggling of implanted ions thus improv-
ing the steepness of the resulting lateral potentials [153, 154] .
And last but not least, it is evident that the intermixing is strongly material dependent [140,
141] . In addition, in a quaternary system like InGaAs(P)/InP the intermixing is more compli-
cated than that in a ternary one, e.g. AlGaAs/GaAs, as interdiffusion can occur on both group
III and V sublattices, which are characterized by their diffusion lengths. Not only a blue shift but
also a red shift in the band gap was observed in InGaAs(P)/InP QWs, depending on the diffusion
length ratio governed by the RTA temperature after plasma exposure [155] .
Whereas the radiation defect-induced QW intermixing is a mature technology, its application
to QD structures is actively studied at the time. The QD intermixing should differ from that in
QWs due to the large surface area-to-volume ratio, composition gradients inside the dots, and
a peculiar confi guration of strain fi elds around the dots. There are two different aspects of the
chemical disorder via interdiffusion across the QD interfaces, namely: (i) the effect of the strain
relief inside the QDs and (ii) the purely chemical effect due to the group III and group V atomic
species interdiffusion. According to [156] , these effects may be quantitatively comparable, sig-
nifi cantly affecting the electronic and optical properties of the dots.
In the fi rst work on ion implantation-induced intermixing of InAs/GaAs QDs [59] , a blue shift
of the ground state transition of up to 150 meV was achieved after implantation of 1 1 0
13
cm
2
of 50 keV Mn ions and annealing ( Fig. 13.22 ). It is worth noting that a blue shift as large as
100 meV was obtained after implantation of 1 1 0
15
c m
2
prior to any heat treatment, however,
the pristine PL intensity could not be restored after annealing at 700°C, and at 800°C the inter-
mixing began also in the unimplanted sample, thus preventing the implementation of a high lat-
eral band gap contrast required in device applications. Furthermore, the additional blue shift due
to annealing had a larger effect on samples with low implantation dose.
Beside the blue shift, a partial suppression of the dot emission was observed in InAs/(AlGa)As
QD structures implanted with 100 keV Cr ions and annealed at 700°C for 1 min [157] . This sup-
pression was accompanied by an enhancement of the wetting layer (WL) emission and its red
shift ( Fig. 13.23 ). It was concluded that the used treatment drives the system towards a predomi-
nantly two-dimensional character. The unusual behaviour of the WL luminescence was inter-
preted as being due to a compensation of the Ga and/or Al diffusion from the GaAs or AlGaAs
barriers to the WL by the In diffusion from the QDs to the WL (see the inset of Fig. 13.23 ).
Incorporation of more In into the WL would lead to an In-rich and/or a thicker two-dimensional
layer, thus causing a red shift of the WL emission. The same process can lead to the partial sup-
pression and a blue shift of the dot PL in the annealed and/or implanted samples [157] .
A comparison of the InGaAs/GaAs QD intermixing accomplished by means of the 360 keV
P and As ion implantation at 200°C and of the impurity-free vacancy disordering (IFVD) using
CH013-I046325.indd 416CH013-I046325.indd 416 6/24/2008 3:45:36 PM6/24/2008 3:45:36 PM
Radiation Effects in Quantum Dot Structures 417
capping with dielectric SiO
2
or Si
x
N
y
fi lms, always followed by RTA, has shown that in the fi rst
case high spatially selective intermixing can be achieved at a signifi cantly lower annealing tem-
perature than for typical dielectric cap techniques [158, 159] .
In Fig. 13.24 , the implantation-induced energy shift and the full width at half maximum
(FWHM) of the PL spectrum of InAs/InP QDs are shown as a function of the 450 keV P ion fl u-
ence for two RTA temperatures. For low doses the FWHM for the implanted and annealed sam-
ple decreases noticeably compared to that of the annealed but unimplanted QD sample. It was
inferred that ion implantation-induced intermixing reduces the overall variation in QD size,
shape, strain, and composition profi le [160] .
As pointed out above, whatever the technique (IFVD using dielectric caps or implantation-
induced), the intermixing mechanism is the defect-enhanced diffusion requiring the presence of
1 1.1
1.4
1.51.2
1.3
Energy (eV)
PL@12 K,P
laser
50 mW
no Mn implanted
reference
[Mn] 10
13
cm
2
as implanted
[Mn] 10
13
cm
2
600°C (60 s)
[Mn] 10
13
cm
2
700°C (60 s)
[Mn] 10
13
cm
2
800°C (60 s)
53 meV
PL- intensity. (a.u. units)
142 meV
8 meV
34 meV
Figure 13.22 Evolution of the PL spectra as a function of annealing temperature for an Mn implantation dose of
1 1 0
13
c m
2
at T 12 K [59] .
1.2
1.0
1.4
1.6
1.8
QD
WL
Cr12-A1
(AIGa)As
T 10 K
V
1
0.5
Energy (eV)
PL intensity (arb. units)
In
Figure 13.23 Photoluminescence spectra at 10 K for InAs/Al0.3Ga0.7As quantum dots. The spectra refer to the
as-grown (V) sample and that subjected to the implantation of 4 1 0
12
c m
2
Cr ions and 60 s annealing at 700ºC.
Inset: Sketch showing the In interdiffusion in the dot [157] .
CH013-I046325.indd 417CH013-I046325.indd 417 6/24/2008 3:45:36 PM6/24/2008 3:45:36 PM
418 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics
point defects. In this sense, the existence of an optimum ion fl uence for the achievement of the
maximum intermixing (see, e.g., Fig. 13.24 ) is usually interpreted as a result of the formation of
more complex defect clusters at higher fl uences, which reduces the availability of point defects
that cause intermixing. On the other hand, the virgin PL intensity usually cannot be restored
after heavy ion implantation and subsequent annealing at temperatures below the onset of inter-
mixing in unimplanted samples, obviously due to incomplete damage annealing. It follows from
these considerations that any other technique producing a high concentration of point defects
without heavily disordering the crystal lattice should lead to an effi cient intermixing without
degradation of the PL intensity. Two such radiation-based techniques are the high-energy proton
irradiation and the low-energy ion implantation or plasma treatment of the surface.
Due to the large penetration depth as well as low lateral straggling and uniform damage
along the major part of the trajectory (see Fig. 13.5 ), the protons of sub-MeV and MeV ener-
gies are well suited for the treatment of multilayer QD structures. That is why proton beam writ-
ing is a promising technique also in nanolithography: e.g. the energy distribution profi le from a
2 MeV proton beam trajectory over the fi r s t 2 μ m of penetration into a photoresist is essentially
contained within a 10 nm diameter [161] . Using the proton irradiation, it is important to take
into consideration the above-mentioned possibility of the hydrogen passivation of non-radiative
recombination centres like dislocations and point defects [109, 162] , see Section 13.2.8.
In the fi rst investigation on intermixing a ten-layer InAs/GaAs QD structure by means of proton
implantation with subsequent RTA [163] , a blue shift of the PL peak energy up to 94.3 meV was
achieved; however, the virgin PL intensity could not be restored, presumably due to the choice of
too high 100 keV proton fl uences. In a subsequent publication of the same group [115] , the use
of smaller fl uences and energies (50 and 80 keV) led, beside a blue shift of up to 100 meV, to a PL
intensity enhancement by a factor of six. However, as was elucidated by control experiments, the
improvement occurred due to hydrogen passivation of non-radiative centres. In a parallel work
on InGaAs/GaAs QDs [164] , a linewidth reduction was found to be proportional to the amount
of blue shift, and the latter decreased with the implantation temperature increasing from 20 to
200°C due to the dynamic annealing effect during implantation. A non-monotonous depend-
ence of the FWHM on the hydrogen fl uence was observed in [165] . A pronounced enhance-
ment of room temperature PL up to 80-fold induced by 50 and 70 keV proton implantation
and RTA in a multilayer InAs/GaAs QD structure was reported in [114] . As in [115] , secondary
ion mass spectroscopy (SIMS) showed a homogeneous hydrogen distribution over the QD
850°C
850°C
850°C
750°C
750°C
750°C
0.30
0.25
50 100 500
0.08
0.04
0.06
0.02
0.20
510
Dose (x10
11
ions/cm
2
)
Energy shift (eV)
FWHM (eV)
Figure 13.24 The implantation-induced energy shift (defi ned as the difference between the PL energy peaks of
the implanted and unimplanted samples which are annealed at the same temperature) and the full width at half
maximum (FWHM) of the PL spectrum of the InAs/InP QDs vs P ion fl uence. The FWHMs of the PL spectra of the
unimplanted and annealed QDs are also shown (by solid arrows) for comparison. All samples are annealed at either
750 or 850°C for 30 s [160] .
CH013-I046325.indd 418CH013-I046325.indd 418 6/24/2008 3:45:36 PM6/24/2008 3:45:36 PM