Henini M. Handbook of Self Assembled Semiconductor Nanostructures for Novel devices in Photonics and Electronics

Подождите немного. Документ загружается.

Growth and Characterization of Structural and Optical Properties 219

photoluminescence experiment was performed using a triple-pulsed Ti-sapphire laser emitting at

250 nm.The collected luminescence was analysed by a streak camera with 4 ps time resolution

( Fig. 6.24 ).

3.6

Energy (eV)

Time (ps)

PL intensity (log. units)

PL intensity

(lin. units)

200 400 600 800

1000 1200 1400 1600

3.7 3.8 3.9 4.0 4.1

Figure 6.24 Time-resolved photoluminescence at various emission wavelengths. The decay curves are independent

of emission wavelength. The decay is not mono-exponential.

It has to be noted that the time-resolved measurements were also performed as a function of

sample temperature, showing no change up to 100 K. This indicates that non-radiative recom-

bination can be neglected in this temperature range, and thus that the observed photolumi-

nescence decay curves are characteristic of the radiative decay of the quantum dots. The ﬁ rst

striking feature is the fact that the photoluminescence decay curves are independent of the emis-

sion energy. This means that the oscillator strength does not depend on the QD size. This again

is a difference with the behaviour of polar QDs, for which the oscillator strength decreases expo-

nentially with QD size [54, 55] . The second striking feature in the photoluminescence decay

curves is that the decay is very fast as compared to the decay of polar GaN QDs [54, 55] . The

decay in Fig. 6.24 is not mono-exponential, but an average decay time of around 280 ps can be

extracted. This is to be compared with lifetimes ranging between a few ns and several hundred µs

for polar QDs [54, 55] .

From these ﬁ rst experimental observations in cw and time-resolved PL, it can be concluded

that no quantum conﬁ ned Stark effect is observed in these [11

0] quantum dots [13, 56] .

6.6.2 Single dot PL – spectral diffusion and temperature broadening

In order to gain more insights on the properties of conﬁ ned carriers in non-polar GaN QDs,

spatially resolved photoluminescence experiments were performed in order to analyse optical

transitions stemming from single QDs. As compared to other QD systems for which single dot

spectroscopy has been performed for several years, additional difﬁ culties appear. A ﬁ rst difﬁ culty

is that the density of GaN QDs is difﬁ cult to master, and in this case to reduce (see earlier parts in

this chapter). In the present case, the average density of QDs is 2  1 0

c m

 2

. However, the den-

sity is not homogeneous, so that locally the density can be reduced. Another difﬁ culty is the emis-

sion wavelength in the UV, for which good optical components are scarce. In our case, we use an

N A  0.4 refractive microscope objective which can focus the 244 nm doubled argon laser on

a ⬃ 1 μ m diameter spot. This spot size is still too large to isolate a few quantum dots, so an addi-

tional processing of the sample is required. Two techniques were alternatively used to reduce the

number of QDs probed under the excitation spot: (i) mesas deﬁ ned by e-beam lithography were

CH006-I046325.indd 219CH006-I046325.indd 219 6/27/2008 4:22:24 PM6/27/2008 4:22:24 PM

220 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics

etched by inductively coupled plasma reactive ion etching, with sizes ranging between 200 nm

and 5 μ m and (ii) polystyrene nanobeads (diameter 300 nm) were deposited on the samples, cov-

ered by 100 nm aluminium; the beads were then removed, leaving apertures in the opaque alu-

minium mask. Both techniques allow the study of isolated transitions from single QDs and give

similar results ( Fig. 6.25 )

3.5

No processing

d  0.3 µm

Intensity (a.u.)

Ener

y (eV)

d  0.6 µm

100 nm

3.6 3.7 3.8 3.9 4.0 4.1 4.2

Figure 6.25 Left : SEM image of an etched mesa, the aligned surface QDs are visible on the image. Right :

Microphotoluminescence spectra on an unprocessed region, on a 0.6 μ m mesa and on a 0.3 μ m mesa. Isolated sharp

lines are observed on the high-energy side for some of the smallest mesas.

By studying several mesas and nanoapertures, we could isolate sharp spectrally isolated lines.

One ﬁ rst remark is that these sharp lines were always isolated on the high-energy side of the

quantum dot distribution (for energies above 3.8 eV). On the lower-energy side, structured fea-

tures are observed, but are never as sharp and isolated as on the higher-energy side. The reasons

behind this phenomenon are still unclear. As the sharp lines appear on the high-energy side, we

ﬁ rst checked that they correspond to the recombination of single electron–hole pairs, and not to

excited transitions of the QDs. This was done by performing power dependent measurements, in

which the line intensities do have a linear dependency as a function of excitation power.

In such experiments, the ﬁ rst notable feature is the linewidth of the observed transitions. In

our case, the linewidths range between 0.5 and 2 meV (the resolution of the set-up at these ener-

gies is around 0.4 meV). These linewidths are quite large compared to what is usually observed

for QDs in other systems (usually in the 1 μ eV to 100 μ eV range). This is, however, less than

what was reported for polar GaN/AlN QDs. At ﬁ rst, Kako and co-workers [57] reported linew-

idths ranging between 5 meV and 20 meV, i.e. one order of magnitude higher than our ﬁ ndings.

Later, Bardoux and co-workers [58] reported linewidths as narrow as what we observe for small

polar quantum dots; however, the observed lines always come as multiplets of lines spread over a

few meVs. Both these results for polar QDs are explained by spectral diffusion: this phenomenon

occurs when traps near a quantum dot are ﬁ lled and emptied at a pace that is faster than the

observation time. In that case, the time varying Coulomb force due to the presence or not of a

charge in a trap leads to a time-dependent electrostatic potential inside the QD, which changes

the energy of the discrete electronic transitions. In the work of Kako and co-workers [57] , the

volumic density of traps is such that the various electrostatic conﬁ gurations found over the inte-

gration time lead to a large broadening of the transition. In the work of Bardoux and co-workers

[58] , the density of traps is less: a low number of traps are close enough to a given QD to induce

large spectral shifts of the transition, giving rise to multiplets. Each line of the multiplet is also

broadened through an interaction with traps that are farther away.

Returning to the case of [11

0] quantum dots, the experimental situation is that narrow iso-

lated lines or doublets are observed. Spectral diffusion is indeed observed when acquiring successive

CH006-I046325.indd 220CH006-I046325.indd 220 6/27/2008 4:22:24 PM6/27/2008 4:22:24 PM

Growth and Characterization of Structural and Optical Properties 221

spectra with 1 s acquisition time ( Fig. 6.26 ). It can be argued that the effect produced by the emp-

tying and ﬁ lling of a trap on a QD transition will be larger in the presence of an internal electric

ﬁ eld, so that it is reasonable that non-polar QDs undergo less spectral diffusion than polar QDs.

One, however, has to be cautious with such reasoning as the spectral diffusion depends much on

the sample quality so that it is not simple to compare different samples.

0.8

1.2

1.6

2.0

FWHM (meV)

Spectrum number

1.0

0.5

0.0

0.5

1.0

Energy

(eV)

E  E

(meV)

4.0275

4.0250

4.0225

10 20 30 40 50 60 70 80 90 100

Figure 6.26 Spectral diffusion for an isolated PL line. One spectrum is acquired every second. The ﬂ uctuations of

the emission energy and of the FWHM are represented. The typical energy ﬂ uctuations are comparable with the

typical FWHM.

The next question that arises when studying single quantum dots is whether many particle

states such as the biexciton and charged excitons can be observed. The biexciton was, for instance,

observed by Kako and co-workers in polar GaN QDs; in that case the biexciton has a large nega-

tive binding energy ( ⬃ 30 meV). This unusual situation is due to the electron–hole spatial

separation in polar heterostructures which enhances the ratio between the repulsive electron–

electron and hole–hole interactions and the attractive electron–hole interaction, so that by add-

ing a second exciton in the quantum dot is energy costly [57] .

For polar quantum dots, the biexciton could not be observed. The reasons for this are experi-

mental: when raising the excitation power, usually many lines grow in the vicinity of the main

studied line. These new lines might correspond to excited states of other QDs lying lower in

energy (recall that only the fundamental transitions of small dots with a high emission energy

could be observed). The biexciton line of the studied QD might be among these new lines; how-

ever, it could so far not be clearly identiﬁ ed, so that the binding energy of the biexciton in non-

polar GaN quantum dots is so far not known.

The narrow linewidth of the fundamental transition in non-polar GaN QDs allows the study of

the coupling between the conﬁ ned carriers and the acoustic phonons when raising the tempera-

ture of the sample. Indeed, when the temperature is increased, the acoustic phonon population

builds up, leading to an increase of emission and absorption of acoustic phonons in the photo-

luminescence process. These absorption and emission events are observed as “ phonon wings ”

above and below the main line (zero phonon line). We thus studied the single QD emission line

as a function of sample temperature ( Fig. 6.27 ). Below 60 K, the broadening of the line is domi-

nated by the spectral diffusion so that no acoustic phonon broadening is observable. However, for

higher temperature, high and low energy wings can be observed. The asymmetry between both

wings is characteristic of acoustic phonon wings (the acoustic phonon emission processes are

more likely than the absorption processes). These phonon wings were already observed in other

CH006-I046325.indd 221CH006-I046325.indd 221 6/27/2008 4:22:25 PM6/27/2008 4:22:25 PM

222 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics

QD systems [59, 60, 61] . The important feature – which will be developed in the next section – is

that this acoustic phonon coupling is a way of probing the conﬁ ned carrier wavefunctions [59] .

6.6.3 Phonon coupling and oscillator strength: the localization hypothesis

The basic idea behind the probing of conﬁ ned wavefunctions through acoustic phonon coupling

is that the maximum coupling occurs for acoustic phonons with a wavelength that matches the

exciton spatial extension [59] . The properties of the acoustic phonons being known, the phonon

wings can then be modelled with the exciton spatial extension as a ﬁ tting parameter, so that the

ﬁ tting of experimental curves such as the one presented in Fig. 6.28 leads to the measurement of

the exciton spatial extension.

In order to apply such a model, one has to propose a test wavefunction for the conﬁ ned exci-

ton. In the present study, we ﬁ rst of all propose to neglect any internal electric ﬁ eld. Indeed, we

have shown previously that in macrophotoluminescence measurements, no quantum conﬁ ned

Stark effect is observed, so that the internal electric ﬁ eld in these non-polar quantum dots does

not play an important role for the wavefunction geometry. In order to propose a test wavefunc-

tion, we then consider the typical length scales in our system. The typical QD size that we study is

2 nm height and 20 nm diameter. The Bohr radius being 2.8 nm in GaN, we see that the conﬁ ne-

ment due to the QD potential is strong in the vertical direction (growth axis), while it is weak in

the in-plane dimensions. The wavefunction can then be factorized into a vertical component cor-

responding to the GaN/AlN quantum well of 2 nm thickness, and an in-plane component corre-

sponding to a bound 2D exciton whose centre of mass is conﬁ ned by the QD boundaries:

ψππ

Xeh e Z h Z

r,r

zL z L

e( ) cos( )cos( / )





⎛

⎝

⎜

⎞

⎠

⎟

⎛⎛

⎝

⎜

⎞

⎠

⎟

⎛

⎝

⎜

⎞

⎠

⎟

ξπ

 /2

(6.1)

3.94

T  140 K

T  120 K

T  100 K

T  80 K

T  60 K

T  40 K

T  30 K

T  20 K

T  10 K

Intensity (lin. units)

Energy (eV)

T  5 K

FWHM  1 meV

20

PL intensity (arb. units)

Energy (meV)

Acoustic

phonon

emission

Acoustic

phonon

absorption

15 10 50 5101520

3.95 3.96 3.97 3.98 3.99 4.00

Figure 6.27 Left : Broadening of a single QD emission line as a function of temperature. Right : Zoom on the curve

at 140 K, the blue line is a Voigt ﬁ t to the zero phonon line.

CH006-I046325.indd 222CH006-I046325.indd 222 6/27/2008 4:22:25 PM6/27/2008 4:22:25 PM

Growth and Characterization of Structural and Optical Properties 223

The ﬁ rst factor in the proposed wavefunction corresponds to the vertical conﬁ nement in a

quantum well of thickness L

with inﬁ nite barriers (which is here a very good approximation

given the GaN/AlN huge band offsets). For each studied QD, we calculate the corresponding L

from the QD transition energy from a simple QW model, assuming that the transition energy of

the QD is mainly due to the vertical conﬁ nement. The second factor in Ψ corresponds to the 2D

hydrogenoid exciton, r being the electron–hole relative coordinate r  r

 r

and a

the two-

dimensional Bohr radius corresponding to the QW with inﬁ nite barriers of height L

. The last

factor is the conﬁ nement of the centre of mass of the exciton of coordinate R

 ( m

 m

( m

 m

), ξ being the parameter describing the lateral extension of the exciton.

Within this framework, the coupling to acoustic phonons can be modelled, with ξ as ﬁ tting

parameter. The details of the model can be found in [56] and we shall here only describe its main

features. The model is basically the Huang–Rhys model of interaction between a localized elec-

tron and phonons, applied here to a localized exciton in a quantum dot [62] . In general, one has

to consider both the deformation potential and the piezoelectric potential couplings between the

localized exciton and all acoustic phonon branches. For semiconductors with zinc-blende sym-

metry, it can be shown that only the deformation potential coupling to longitudinal acoustic

phonons has to be taken into account [63] . In the present study, we are dealing with a wurtz-

ite symmetry semiconductor so that the piezoelectric coupling potential might not be negligible.

It can be shown that while the deformation potential coupling is mainly sensitive to the over-

all extent of the excitonic wavefunction, the piezoelectric coupling potential is large in the case

when the electron and hole wavefunctions are spatially offset and thus result in a permanent

dipole. For instance, for large enough polar GaN quantum dots, the piezoelectric coupling to

acoustic phonons is large and dominates over the deformation potential coupling [64] . On the

contrary, if the electron and hole wavefunctions are exactly superimposed so that the system

is everywhere locally neutral, then the piezoelectric coupling is strictly zero [64] . Again in our

case, the macrophotoluminescence experiments do not show any evidence of electron–hole spa-

tial separation so that we choose to neglect the piezoelectric coupling of the conﬁ ned exciton to

acoustic phonons. This is justiﬁ ed a posteriori by the good ﬁ ts obtained on the experimental data.

Another difﬁ culty in our case is to be able to ﬁ t the zero phonon line. As already said, the zero

phonon line is dominated by spectral diffusion, which we describe in a very phenomenological way.

In particular, we do not have a modelling of the evolution of spectral diffusion with temperature.

10

T  4K

Normalized µPL intensity (linear units)

Energy (meV)Energy (meV)

T  140KT  100K

Energy (meV)

Phonon

wings

010 10 0 10

10

10 0 10

Figure 6.28 Experimental PL data (open dots), Voigt ﬁ t of the zero phonon line (dotted blue line) and ﬁ t with our

phonon coupling model for ξ  2.1 nm (red line). The inset for 140 K shows the theoretical model without the ZPL,

i.e. only the phonon wings, showing clearly the asymmetry between emission and absorption of acoustic phonons.

CH006-I046325.indd 223CH006-I046325.indd 223 6/27/2008 4:22:26 PM6/27/2008 4:22:26 PM

224 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics

In general, the line shape due to the spectral diffusion can be well ﬁ tted by a Voigt function. For

each temperature, we thus use a Voigt ﬁ t of the zero phonon line, while the ξ parameter that ﬁ ts

the phonon wings is used as a temperature independent ﬁ tting parameter. As the zero phonon

line is typically below 2 meV FWHM and the phonon wings extend up to 10 meV on each side of

the zero phonon line, the ﬁ t of the ZPL and of the phonon wings can be considered as essentially

independent.

The typical result for the ﬁ t is shown in Fig. 6.28 , which displays the experimental data for a

low, an intermediate and a high temperature. The ﬁ t of the zero phonon line is indicated in blue,

while the model including the phonon coupling is displayed in red. It can be seen that the experi-

mental curves can be very well adjusted over the entire temperature range. The ﬁ tting parameter

ξ that describes the centre of mass localization is equal to 2.1 nm in this case. This corresponds

to a full width at half maximum for the in-plane wavefunction of the centre of mass of 5 nm, to

be compared with the diameter of 20 nm of the quantum dot. It can be shown that even with a

GaN quantum dot with the shape of a cylinder with a diameter equal to the top facet of the dot

and surrounded by inﬁ nite barriers (in other words, only the central cylinder of the actual QD is

considered, while its sides are replaced by an inﬁ nite barrier), the lateral extension of the centre

of mass would still be larger ( ξ  3.4 nm) than what we measure due to the phonon coupling

experiment. This means that the exciton localization in the quantum dot is not solely governed

by the barrier conﬁ nement, but that an additional mechanism leads to a supplementary lateral

conﬁ nement.

This last conclusion can actually be probed in another way, namely by giving a closer look at

the time-resolved data acquired on an ensemble of non-polar GaN QDs. Indeed, as described ear-

lier in this article, the average decay time for an ensemble of QDs is around 280 ps. This seems

to be rather short for quantum dots; however, one has to recall that a better way of describ-

ing the ﬁ eld–emitter interaction is the dimensionless oscillator strength for a given transition,

the oscillator strength scales as the square of the emission wavelength. The oscillator strength

deduced from the experimental data is in our case around 9. This is thus the same as for the typi-

cal InAs/GaAs quantum dot. However, the physics in both systems is quite different. The typical

InAS /GaAs QD size is typically 3 nm height and 20 nm diameter – i.e. very similar to the GaN

QDs considered here – while the Bohr radius in InAs is 28 nm, which is an order of magnitude

larger than in GaN. In an InAs QD, the carriers are thus strongly conﬁ ned in all directions: the

conﬁ nement energy dominates over the carrier–carrier binding energy. As already described by

wavefunction 7.1, for non-polar GaN QDs, the carriers are a priori expected to be strongly con-

ﬁ ned in the vertical direction and weakly conﬁ ned in the in-plane directions. The corresponding

physical situation is that of a bound two-dimensional exciton which is coherent over the in-plane

surface of the QD. In that case, the regime of giant oscillator strength should be reached [65,

66] . This effect has, for instance, been clearly characterized for QDs formed by interface ﬂ uctua-

tions of a GaAs/AlGaAs quantum well [67] . We can thus test various hypotheses for the con-

ﬁ nement regime of the non-polar GaN QDs and see in each case which is the expected oscillator

strength. We can examine a ﬁ rst case for which we consider the wavefunction 6.1, with a lateral

conﬁ nement of the centre of mass limited only by the conﬁ ning potential of the AlN barrier. As

discussed earlier, in that case a lower limit for ξ can be taken as 3.4 nm.For such a wavefunction,

the oscillator strength is given by:

⎛

⎝

⎜

⎞

⎠

⎟



(6.2)

where E is the transition energy and E

is the interband matrix element (Kane energy). For E

, the

data found in the literature is scarce and scattered, we choose to use the value (for the conduction

band to heavy-hole valence band transition) of 15.7 eV calculated by Chang and Chuang [68] ,

which is close to the value of 14 eV recommended by Vurgaftman in a review of III-N parameters

[69] . This then gives a value of the oscillator strength of 85, which corresponds to a radiative

lifetime of 24 ps. This is clearly much lower than what we measure.

CH006-I046325.indd 224CH006-I046325.indd 224 6/27/2008 4:22:26 PM6/27/2008 4:22:26 PM

Growth and Characterization of Structural and Optical Properties 225

The opposite situation would be that of laterally strongly conﬁ ned carriers, in which case the

oscillator strength is given by [66] :

(6.3)

in the case of a perfect electron–hole overlap. This value is lowered if the electron and hole wave-

functions are different. The value for f given by the above formula is 4, corresponding to a radia-

tive lifetime of 520 ps, which is larger than the mean decay time that we measure. The measured

mean decay time is thus intermediate between an upper limit to the giant oscillator strength

regime and a lower limit to the strongly conﬁ ned case.

As already said, the Kane energy is not known precisely for GaN, and a value of 7.7 eV is

reported in [70] , but even with this value the precedent conclusion would still hold.

Let us now consider again the conﬁ ned centre of mass wavefunction 6.1, but with this time

the value for ξ  2.1 nm that was deduced from the acoustic phonon coupling analysis. In that

case, equation 6.2 (with E

 15.7 eV) yields f  33, and correspondingly a decay time of 62 ps.

This value is still much lower than what is achieved experimentally. As said earlier, the value of

the Kane energy might be about half of the value with use. But, most importantly, it has to be

pointed out that the enhancement of the oscillator strength in equation 6.2 is due to an exci-

tonic effect. With ξ  2.1 nm (to be compared with a

 1.8 nm), the physical situation is not

really that of a 2D bound exciton which is coherent over a large (much larger than a

) area. The

enhancement of the oscillator strength thus should be reduced in equation 6.2.

The decay time analysis thus shows that the situation is intermediate between the strong con-

ﬁ nement and the 2D giant oscillator strength regime, which is coherent with the acoustic pho-

non coupling analysis.

Going further along this analysis, one might reasonably consider that the localization length

might vary from dot to dot, independently of QD size. In that case, for each emission energy (cor-

responding to a collection of QDs with a given size), there should be a distribution of decay times.

This is indeed what is experimentally observed when looking in more detail at Fig. 6.24 . Each

decay curve is not mono-exponential, and is better ﬁ tted by a sum of exponentials with decay

times ranging between 70 and 600 ps. In order to conﬁ rm this point, we performed time-resolved

measurements on single QD lines. Figure 6.29 displays, for instance, the time resolved signal for

three different quantum dots. The ﬁ rst striking feature is that in that case the decays are mono-

exponential, which is not the case for an ensemble of QDs. The second important fact is that the

decay times for the three different QDs are different, and vary non-monotonously as a function of

emission energy. This supports our interpretation that the decay time is dominated by the locali-

zation length inside the QD, and not by the actual QD size [56] .

0.01

0.1

QDc: τ  240 ps)

QDb: τ  140 ps

QDa: τ  170 ps

PL intensity (log. units)

Time (ps)

(a)

(b)

(c)

Energy

Time (ps)

200 400 600 800 1000 1200

200

400

600

800

Figure 6.29 Left : Time resolved data for single QD lines; the decay is mono-exponential and the decay time varies

from dot to dot. Right : Same data as they appear on the streak camera CCD.

CH006-I046325.indd 225CH006-I046325.indd 225 6/27/2008 4:22:26 PM6/27/2008 4:22:26 PM

226 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics

6.7 Rare earth doping of GaN QDs

As a peculiar application of GaN QDs, their doping with rare earth ions has been achieved in

view of realizing visible light-emitting diodes (LEDs). The idea of combining the RE luminescence

with a semiconductor host came about by using Er and Yb (infrared) in GaAs, GaP, Si and SiC.

Especially, Er-doped Si structures have been well investigated since the research was driven by the

appeal of integration with Si microelectronics. For intense and thermally stable visible light emis-

sion from electroluminescence devices, wide band gap semiconductors are required since efﬁ cient

energy transfers can only occur if the band gap of the host is wider than the energy difference

between RE

3 

ground states and excited state. Based on this idea, Steckl and his co-workers com-

bined the RE

3 

luminescent with the wide band gap of GaN and realized electroluminescence

from the blue to the red spectral range [71, 72] . Furthermore the combination of different col-

ours from RE

3 

ions – red (Eu, Pr), green (Tb, Er, Ho) and blue (Tm, Dy, Ce) – would yield to white

light emission. The choice of GaN as a semiconductor host can be further explained since it

exhibits ﬁ rst a direct band gap, required for a good semiconductor host–RE

3 

ion energy transfer

probability and since it is a III–V semiconductor the RE ions are supposed to occupy metal sites

with a 3  charge, which was later experimentally established [73, 74] .

However, the RE radiative quantum efﬁ ciency strongly depends on the carrier-mediated energy

transfer process, which has to compete with fast non-radiative recombination channels. Non-

radiative processes can occur, for example, at defects, which are found with a density of about

/cm

for GaN, which is a rather high number compared to other semiconductors. The result

is a thermal quenching of the RE luminescence between one and two orders of magnitude from

liquid helium to room temperature for REs in GaN ﬁ lms [75, 76] .

One way to overcome this problem can be the doping of nitride quantum dots (QDs) with RE ions.

Then the carrier-mediated energy transfer to RE ions should improve signiﬁ cantly since QDs are

defect-free regions and act as carrier conﬁ nement centres so that the thermal stability of the observed

RE luminescence should be enhanced drastically. In other words, the energy transfer to RE ions

will be dominant over other possible non-radiative processes in the semiconductor host material.

Following the strategy described above, [0001] GaN QDs have been doped with Eu [74, 77] ,

with Tm [78] and with Tb [79] . As expected, the quenching of RE photoluminescence is consid-

erably decreased in the case of doped QDs with respect to the case of the bulk layer, as illustrated

in Fig. 6.30 .

100

1000/T(K

1

)

0.01

0.1

400 600

GaN:Eu QDs

GaN QDs

CL intensity (a.u.)

Normalize

d PL intensity

Figure 6.30 Photoluminescence intensity as a function of temperature for bulk GaN doped with Eu (open

triangles), for GaN/AlN QDs (black squares) and for EU-doped GaN/AlN QDs (crossed squares). The luminescence

of the EU-doped QDs is stable up to room temperature. The inset displays the cathodoluminescence of GaN QDs and

Eu-doped GaN QDs. The quenching of the QD luminescence indicates an efﬁ cient energy transfer towards the Eu

transitions.

CH006-I046325.indd 226CH006-I046325.indd 226 6/27/2008 4:22:28 PM6/27/2008 4:22:28 PM

Growth and Characterization of Structural and Optical Properties 227

In addition, conﬁ nement in QDs was found to make possible the excitation of levels in RE ions

which are not accessible when using bulk material. As a matter of fact, transitions originating from

and

levels of Tm

3 

ions have been observed in GaN QDs but not in bulk material [80] .

6.8 Conclusion

Although being a latecomer in the family of self-assembled QD forming systems, GaN presents

speciﬁ cities which make it a fascinating one: its versatility to form polar and non-polar QDs, the

presence of an electric ﬁ eld in polar QDs, the possibility, as concerns [0001] QDs, to fabricate

them on a Ga-terminated or N-terminated face, the lack of interdiffusion between dots and bar-

rier, including during the capping process, make this system rather unique. The lack of interdif-

fusion, in particular, makes it a model system to a certain extent and potentially allows one to

probe the interplaying role of mismatch, surface and interface energy.

Optical properties of polar dots are dominated by the presence of an electric ﬁ eld contrary to

the case of non-polar ones. However, in both cases, in-depth understanding of optical proper-

ties is far from being reached, due to a lack of control of residual doping level in both dots and

barriers. Work is presently in progress to overcome these difﬁ culties. Despite them, it is certain

that optical properties of GaN QDs are as fascinating as structural ones: the large conﬁ nement

resulting from the large band gap energy difference between GaN and AlN or AlGaN makes these

nanostructures of potential interest for room temperature optoelectrical applications.

Acknowledgements

The authors acknowledge the contributions of the former and present students who participated

in the work described in this chapter, namely C. Adelmann, B. Amstatt, T. Andreev, J. Coraux, S.

Founta, N. Gogneau, E. Martinez-Guerrero, F. Rol, E. Sarigiannidou, and J. Simon. They also

acknowledge the many contributions from their colleagues: E. Bellet-Amalric, J. Bleuse, C.

Bougerol, V. Favre-Nicollin, J.-M. Gérard, Y. Hori, D. Jalabert, Le Si Dang, H. Mariette, E. Monroy,

N. Pelekanos, D. Peyrade, M.G. Proietti, H. Renevier, J.L. Rouvière, and P. Vennéguès. It is our

pleasure to acknowledge the support of the NOVASiC company which provided high-quality pol-

ishing of SiC substrates.

References

1. See, for instance, the papers in the Special Issue of Philosophical Magazine , 87, 1967 (2007).

2. F. Bernardini and V. Fiorentini , Phys. Rev. B . 57 , R9427 ( 1998 ) .

3. T. Paskova , R. Kroeger , S. Figge , D. Hommel , V. Darakchieva , B. Monemar , E. Preble , A. Hanser ,

N.M. Williams , and M. Tutor , Appl. Phys. Lett. 89 , 051914 ( 2006 ) .

4. M.D. Craven , S.H. Lim , F. Wu , J.S. Speck , and S.P. DenBaars , Appl. Phys. Lett. 81 , 469 ( 2002 ) .

5. H.M. Ng , Appl. Phys. Lett. 8 0 , 4369 ( 2002 ) .

6. C. Chen , J. Yang , H. Wang , J. Zhang , V. Adivarahan , M. Gaevski , E. Kuokstis , Z. Gong , M. Su , and

M. Asif Khan , Jpn. J. Appl. Phys. 42 , L640 ( 2003 ) .

7. P.P. Paskov , R. Schifano , B. Monemar , T. Paskova , S. Figge , and D. Hommel , J. Appl. Phys. 9 8 ,

093519 ( 2005 ) .

8. K. Kusakabe and K. Ohkawa , Jpn. J. Appl. Phys. 44 , 7931 ( 2005 ) .

9. Y. Tsuchiya , Y. Okadome , A. Honshio , Y. Miyake , T. Kawashima , M. Iwaya , S. Kamiyama ,

H. Amano , and I. Akasaki , Jpn. J. Appl. Phys. 44 , L1516 ( 2005 ) .

10. S. Chevtchenko , X. Ni , Q. Fan , A.A. Baski , and H. Morkoç , Appl. Phys. Lett. 88 , 122104 ( 2006 ) .

11. N. Onojima , J. Suda , T. Kimoto , and H. Matsunami , Appl. Phys. Lett. 83 , 5208 ( 2003 ) .

12. M.D. Craven , F. Wu , A. Chakraborty , B. Imer , U.K. Mishra , S.P. DenBaars , and J.S. Speck , Appl.

Phys. Lett. 84 , 1281 ( 2004 ) .

CH006-I046325.indd 227CH006-I046325.indd 227 6/27/2008 4:22:28 PM6/27/2008 4:22:28 PM

228 Handbook of Self Assembled Semiconductor Nanostructures for Novel Devices in Photonics and Electronics

13. S. Founta , F. Rol , E. Bellet-Amalric , J. Bleuse , B. Daudin , B. Gayral , C. Moisson , and H. Mariette ,

Appl. Phys. Lett. 86 , 171901 ( 2005 ) .

14. D.N. Zakharov , Z. Liliental-Weber , B. Wagner , Z.J. Reitmeier , E.A. Preble , and R.F. Davis , Phys. Rev. B .

71 , 235334 ( 2005 ) .

15. M.D. Craven , S.H. Lim , F. Wu , J.S. Speck , and S.P. DenBaars , Appl. Phys. Lett. 81 , 1201 ( 2002 ) .

16. E.J. Tarsa , B. Heying , X.H. Wu , P. Fini , S.P. DenBaars , and J.S. Speck , J. Appl. Phys. 8 2 , 5472

( 1997 ) .

17. G. Mula , C. Adelmann , S. Moehl , J. Oullier , and B. Daudin , Phys. Rev. B . 64 , 195406 ( 2001 ) .

18. C. Adelmann , J. Brault , D. Jalabert , P. Gentile , H. Mariette , G. Mula , and B. Daudin , J. Appl. Phys.

91 , 9638 ( 2002 ) .

19. J. Neugebauer , T.K. Zywietz , M. Schefﬂ er , J.E. Northrup , H. Chen , and R.M. Feenstra , Phys. Rev.

Lett. 90 , 056101 ( 2003 ) .

20. B. Daudin , G. Feuillet , G. Mula , H. Mariette , J.L. Rouvière , N. Pelekanos , G. Fishman , C. Adelmann ,

and J. Simon , Phys. stat. sol. (a) 176 , 621 ( 1999 ) .

21. J.E. Northrup , J. Neugebauer , R.M. Feenstra , and A.R. Smith , Phys. Rev. B . 61 , 9932 ( 2000 ) .

22. M.D. Craven , P. Waltereit , J.S. Speck , and S.P. DenBaars , Appl. Phys. Lett. 84 , 496 ( 2004 ) .

23. C. Adelmann , N. Gogneau , E. Sarigiannidou , J.L. Rouvière , and B. Daudin , Appl. Phys. Lett. 8 1 ,

3064 ( 2002 ) .

24. N. Gogneau , D. Jalabert , E. Monroy , T. Shibata , M. Tanaka , and B. Daudin , J. Appl. Phys. 9 4 , 2254

( 2003 ) .

25. S. Founta , C . Bougerol , H. Mariette , B. Daudin , and P. Vennéguès , J. Appl. Phys. 102 , 074304

( 2007 ) .

26. B. Daudin , F. Widmann , G. Feuillet , Y. Samson , M. Arlery , and J.L. Rouvière , Phys. Rev. B . 56 ,

R7069 ( 1997 ) .

27. B. Damilano , N. Grandjean , F. Semond , J. Massies , and M. Leroux , Appl. Phys. Lett. 75 , 962 ( 1999 ) .

28. S. Tanaka , S. Iwai , and Y. Aoyagi , Appl. Phys. Lett. 69 , 4096 ( 1996 ) .

29. N. Grandjean and J. Massies ,

Appl. Phys. Lett. 71 , 1816 ( 1997 ) .

30. M. Miyamura , K. Tachibana , and Y. Arakawa , Appl. Phys. Lett. 80 , 3937 ( 2002 ) .

31. Y. Hori, O. Oda, E. Bellet-Amalric, and B. Daudin, J. Appl. Phys . 102, 024311 (2007).

32. J.L. Rouvière , J. Simon , N. Pelekanos , B. Daudin , and G. Feuillet , Appl. Phys. Lett. 7 5 , 2632

( 1999 ) .

33. N. Gogneau , D. Jalabert , E. Monroy , E. Sarigiannidou , J.L. Rouvière , T. Shibata , M. Tanaka ,

J.-M. Gérard , and B. Daudin , J. Appl. Phys. 96 , 1104 ( 2004 ) .

34. J. Coraux , B. Amstatt , J.A. Budagoski , E. Bellet-Amalric , Jean-Luc Rouvière , V. Favre-Nicolin ,

M.G. Proietti , H. Renevier , and B. Daudin , Phys. Rev. B . 74 , 195302 ( 2006 ) .

35. B. Daudin , J.L. Rouvière , and M. Arlery , Appl. Phys. Lett. 69 , 2480 ( 1996 ) .

36. J. Northrup and J. Neugebauer , Phys. Rev. B . 60 , R8473 ( 1999 ) .

37. N. Gogneau , E. Sarigiannidou , E. Monroy , S. Monnoye , H. Mank , and B. Daudin , Appl. Phys. Lett.

85 , 1421 ( 2004 ) .

38. D.W. Moon , H.I. Lee , B. Cho , Y.L. Foo , T. Spila , S. Hong , and J.E. Greene , Appl. Phys. Lett. 8 3 , 5298

( 2003 ) .

39. K. Sumitomo , H. Omi , Z. Zhang , and T. Ogino , Phys. Rev. B . 67 , 035319 ( 2003 ) .

40. D. Jalabert , J. Coraux , H. Renevier , B. Daudin , C. Mann-Ho , C. Kwun-Bum , D.W. Moon ,

J.M. Llorens , N. Garro , A. Cros , and A. García-Cristóbal , Phys. Rev. B . 72 , 115301 ( 2005 ) .

41. M.J. Hÿtch , E. Snoeck , and R. Kilaas , Ultramicroscopy 74 , 131 ( 1998 ) .

42. J.L. Rouvière and E. Sarigiannidou , Ultramicroscopy 106 , 1 ( 2005 ) .

43. S. Founta , J. Coraux , D. Jalabert , C. Bougerol , F. Rol , H. Mariette , H. Renevier , B. Daudin ,

R.A. Oliver , C.J. Humphreys , T.C.Q. Noakes , and P. Bailey , J. Appl. Phys. 101 , 063541 ( 2007 ) .

44. J. Tersoff , C. Teichert , and M.G. Lagally , Phys. Rev. Lett. 76 , 1675 ( 1996 ) .

45. F. Widmann , B. Daudin , G. Feuillet , Y. Samson , J.L. Rouvière , and N. Pelekanos , J. Appl. Phys. 8 3 ,

7618 ( 1998 ) .

46. A. Letoublon , V. Favre-Nicolin , H. Renevier , M.G. Proietti , C. Monat , M. Gendry , O. Marty , and

C. Priester ,

Phys. Rev. Lett. 92 , 186101 ( 2004 ) .

47. M.G. Proietti , H. Renevier , J.-L. Hodeau , J. Garcia , J.F. Bérar , and P. Wo l f e r s , Phys. Rev. B . 59 , 5479

( 1999 ) .

CH006-I046325.indd 228CH006-I046325.indd 228 6/27/2008 4:22:29 PM6/27/2008 4:22:29 PM