Journal of the Korean Physical Society, Vol. 45, No. 5, November 2004, pp.

11811184

Magneto-Photoluminescence Study on GaAs/AlGaAs Quantum

Wire Superlattices
I. T. Jeong, S. Ahn and D. H. Kim
Center for Quantum Devices, Northwestern University, Chicago IL 60208, USA

J. C. Woo
School of Physics, Seoul National University, Seoul 151-747

S. J. Rhee
Electronic Physics, Hankuk University of Foreign Studies, Kyonggi-do 449-791
(Received 7 July 2004)
A study on magnetic-field-(B )-dependent blue shifts of photoluminescence has been performed
on a quantum wire (QWR) superlattice at B 30 T at LHe temperature. For GaAs/Al0.5 Ga0.5 As
QWR grown by migration enhanced epitaxy on tilted-cut substrate, the effectiveness of 2-D confinement is analyzed by the dimensionality factor RD through the comparison of diamagnetic and
Landau shifts. It is significant that the electronic wavefunction penetrates more into the quantum
barrier in narrow-wire-width QWR, and the excitonic behavior experiences lateral confinement
confirming the presence of 1-D characteristics.
PACS numbers: 78.20.Ls, 78.66.Fd
Keywords: Diamagnetic shift, Effective mass, Exciton, Landau shift, Low-dimensional confinement, Quantum wire (QWR)

I. INTRODUCTION
In recent years, quantum functional semiconductor
structures, such as quantum wells (QW), quantum wires
(QWR) and quantum dots (QD), have attracted attention due to the low-dimensional (low-D) behaviors in
electronic and excitonic properties [15]. In this paper, we would like to report the results of a photoluminescence (PL) study to detect the low-D confinement
effect of a quasi-1D semiconductor, GaAs/Al0.5 Ga0.5 As
QWR superlattice, in a strong dc magnetic field (B ).
The strong B perpendicular to the plane of the QWR
array provides non-destructive perturbation that would
provide an additional planar confinement to electronic
behavior.
The objective of this study is to investigate the effect
of 1-D perturbation provided by QWR superlattice in the
primarily 2-D confined quantum structure system. For
ground state excitons, the quantum barriers sandwiching
QWR would provide 2-D confinement, which influences
the size and shape of the exciton [6], and the applied DC
magnetic field provides an additional 2-D confinement as
well as 2-D characteristic length, namely cyclotron orbital radius, in electronic behavior. It has been reported
E-mail:

[email protected]

that the electron-hole coupling abruptly changes when

the cyclotron radius becomes comparable with that of
the exciton [7]. In this QWR specimen, an additional
characteristic figure that is asymmetric to the excitonic
orbit and the cyclotron orbit is introduced.
This study provides preliminary results on the electronic characteristics confined in 2-D with 1-D quantum
barriers by applying a magnetic field. From the analysis of the photon-energy-shift dependence on the applied
magnetic field by using a 2-D confinement model, it is
significant that the excitonic behavior is still dominant in
the low magnetic field region, where the cyclotron radius
is larger than the QWR superlattice.

II. EXPERIMENT
The GaAs/AlGaAs QWR samples used in this work
were grown by the Petroff method by using MBE [8,9].
Semi-insulating GaAs wafers having primarily a surface
orientation of (100), but cut slightly tilted by 2 and 1
toward the [011] direction, were used as substrates. During the growth of GaAs and AlGaAs buffer and quantum
barrier layers, the temperature of substrate was 600 C
and 620 C, respectively. The growth rates of GaAs and
AlAs were 0.17 m/hour. During the QWR growth, the

-1181-

-1182-

Journal of the Korean Physical Society, Vol. 45, No. 5, November 2004

Fig. 1. TEM image of (30 ML + 30 ML) 60 ML.

migration-enhanced epitaxy technique was introduced,

so that adatoms (Ga, AlGa) could find their stable locations at the step edges, allowing enough migration time.
We used an interruption time of 30 sec for the migration. The background pressure during the growth was
8 108 Torr in an As-rich environment. The QWR
width was determined by controlling the growth rate in
fractions of monolayer (ML). The details on the growth
are described elsewhere [10]. Five QWR samples of various configurations were used in this experiment. The
configurations of QWR are denoted as [(wire width, n)
+ (barrier width, m)] (wire height) in units of ML
throughout the paper. A typical feature of a TEM image of one of the QWR samples is shown in Figure 1.
Magneto-photoluminescence (PL) was performed at
the National High Magnetic Field Laboratory (NHMFL)
in fields ranging from 0 T to 32 T in steps of 0.5 T or 1
T by using resistive DC magnets in Faraday geometry.
The spectral resolution of the spectrum was 0.04 meV or
0.15
A when using a 1800 groove/mm grating. The +
and polarizations were detected by using a /4 plate
and linear polarizer, and also reversing the field polarity.
An Ar+ laser was used as photo-excitation source. All
the magneto-PL was obtained at LHe temperature.

III. RESULTS AND DISCUSSION

Typical magneto-PL spectra obtained for + and
polarizations emitted by radiative recombination of the

exciton, the coupling of the ground state conduction electron and the ground state heavy hole are sketched in
Figure 2. This shows that the PL peak positions tend
to be blue-shifted as the magnetic field increases. The
field-dependent photon-energy shift of PL spectra emitted by radiative recombination of ground-state excitons
is determined by taking the mean value of + and
spectra at a given B, and is summarized in Figure 3.
When the shift taken without using circular polarizers is
not used and it is determined from the peak of the single bell-shape PL spectrum, it appears to have a slightly
lower value. This is so because the + and spectra have different intensities and the spectrum without
polarization distinction is a combination of the two.
The magnetic field (B ) dependent blue shift
E (B) = E (B) E (0)

(1)

in the high-field region (B 10 T) is best fitted to

linear in B, that is
E (B) = B,

(2)

while at low magnetic field (B 7 T), it is best fitted

to quadratic in B, i.e.,
E (B) = B 2 .

(3)

In the high-field region, the electron undergoes a circular motion with cyclotron frequency
c =

e
B,
mc

(4)

as the cyclotron energy is more dominant than other

terms. The energy shift therefore becomes
1
1 ~e
EL = (n + )~c = (n + )
B,
2
2 mc

(5)

which is called the Landau shift. For undoped samples

such as ours, the electron concentration is low enough
to be assumed to occupy only the ground state of the
Landau level. So, by taking n = 0, the energy shift will
be linear in B, with
EL =

1
~e
~c =
B = B,
2
2mc

(6)

On the other hand, in a low field region, the diamagnetic term acts as a dominant perturbation and the paramagnetic term drops in the ground state. So, the diamagnetic shift in a low-dimensionally confined electronic
state can be expressed by introducing Bugajskis model
[11] as
ED = B 2 = RD

Fig. 2. Polarization dependence of magneto-PL spectra of

QWR.

e2 B 2 2
a ,
4c2 B

(7)

where RD is the dimensionality factor, the effective

reduced mass of an exciton in bulk, and aB the Bohr radius. The dimensionality factor RD is theoretically 1 for

Magneto-Photoluminescence Study I. T. Jeong et al.

-1183-

Table 1. Effective mass and coefficient related to dimensionality factor.

Samples
(20 + 40)
(30 + 30)
(30 + 30)
(40 + 20)
(60 + 60)

30
30
60
30
30

(meV/T)
0.20
0.38
0.43
0.40
0.39

( me )
0.30
0.15
0.13
0.15
0.15

(meV/T2 )
0.025
0.021
0.025
0.027
0.020

RD
0.70
0.36
0.34
0.37
0.20

Fig. 4. Effective reduced masses obtained from the fielddependent blue shift in the high-magnetic-field region.

Fig. 3. Field-dependent PL photon energy shifts of QWR

samples obtained from the mean value of + and spectra.

3-D and 3/16 for 2-D. The coefficients and are obtained from the -square fit of the data, which is plotted
in Figure 3. Then, from the coefficients , the effective
reduced masses of QWR samples are calculated from
Eq. (6). By substituting the obtained value of into
Eq. (7), the dimensionality factor RD is calculated from
. All the values obtained for , , and RD for each
QWR sample are summarized in Table 1.
The effective reduced masses of the samples, whose
QWR heights are 30 ML, are plotted in Figure 4 in order to observe the quantum wire width dependence. The
results show that the effective reduced mass is noticeably
larger for the (20 + 40) 30 sample than the rest. In
the case of a quantum well (QW), the exciton binding
energy changes sharply in narrower QW. The change occurs at the vicinity of the well width Lz = 20 ML [11,
12], and a similar effect has been observed from the diamagnetic shift measurement [12]. This phenomenon in
QW is interpreted by the theoretical underestimation of
the electron and hole wavefunction penetration into the
barrier [12]. On introducing a similar argument with the

2-D exciton model, the large of the narrow QWR could

be understood as the more significant penetration of the
electron and hole wavefunction into the quantum barrier
than for the relatively wider ones.
The obtained dimensionality factor RD = 0.71 for the
(20 + 40) 30 sample is significantly larger when compared with RD 0.35 for the (30 + 30) 30, (30 + 30)
60 and (40 + 20) 30 samples. This indicates that
the excitonic behavior of the narrow wire width sample
has little 2-D confinement effect and approaches that of
3-D. When compared with QW, the weakening of low-D
confinement effect occurs for larger QWR height. In the
case of QW, the 2-D confinement is most effective in the
region of Lz of 12 to 15 ML, and becomes less effective
for smaller and larger Lz [13]. This can be understood in
that, in QWR, the exciton may have an egg-shaped ellipsoid with an additional degree of confinement compared
with a pancake-shape ellipsoidal exciton in QW, and the
electronic wavefunction penetrates into the vertical as
well as lateral quantum barriers.
On the other hand, for the (60 + 60) 30 sample,
RD = 0.20 approaches the theoretical value for a 2-D
exciton. When considering the fact that the diameter
of a 2-D confined exciton is in the range of 100
A, as
in QW [14, 15], and the wire width is larger than the
diameter, the lateral confinement is not effective in this
case, and the excitonic property of this wire width will
be influenced only by 2-D confinement.

-1184-

Journal of the Korean Physical Society, Vol. 45, No. 5, November 2004

The results of the dimensionality factors are consistent

with the penetration argument. Furthermore, they also
provide reliable evidence that the electronic properties
of QWR samples are indeed influenced by 1-D confinement. The field-dependent Zeeman results obtained from
the separation of + and spectra provide supportive
evidence on this 1-D effect, which will be reported elsewhere.

authors wish to express their sincere appreciation for the

facilities and technical assistance provided by NHMFL,
Florida State University, Tallahassee, FL, U.S.A, and its
staff.

REFERENCES
IV. CONCLUSIONS
The low-dimensional confinement of excitonic behavior in QWR has been studied with the magnetic-fielddependent photon-energy shift irradiated by groundstate electron-hole recombination. The magnetic field
applied along the direction of the growth direction provides an additional confinement to the planar electronic
orbital behavior, and also introduces the characteristic
length of the cyclotron orbit in addition to the exciton diameter and the QWR width. The magnitude of
the field applied in this experimental work is sufficiently
large for the diameter of the cyclotron orbit to become
comparable to that of the exciton and the QWR width.
The results show that, in the QWR of narrow wire
width, the electronic behavior is strongly influenced by
the tunneling effect due to the 1-D confinement of QWR,
and the 1-D confinement effect becomes less significant as
the wire width becomes larger. For a wire width larger
than the exciton diameter, there is electronic behavior
of 2-D-like confinement. This work provides preliminary
results on the electronic and excitonic behaviors in a
pseudo-1-D confined system. Further study is needed,
to get a clearer understanding and direct analysis of the
1-D excitonic properties.

ACKNOWLEDGMENTS
This work was supported in part by the Korea Research Foundation, Project No. 2003-015-C00185. The

[1] S. H. Park, J. Korean Phys. Soc. 42, 344 (2003).

[2] T. Okamoto, N. Nunoya, Y. Onodera, S. Tamura and S.
Arai, Jpn. J. Appl. Phys. 41, L249 (2002).
[3] Y. Park, J. S. Choe and H. Jeon, J. Korean Phys. Soc.
40, 245 (2002).
[4] M. Lang, W. Bronner, W. Benz, M. Ludwig, V. Hurm, G.
Kaufel, A. Leuther, J. Rosenzweig and M. Schlechtweg,
Electron. Lett. 37, 1247 (2001).
[5] T. G. Kim, J. Korean Phys. Soc. 42, 791 (2003).
[6] D. W. Kim, Y. A. Leem, S. D. Yoo, D. H. Woo, D. H.
Lee and J. C. Woo, Phys. Rev. B 47, 2042 (1993).
[7] H. S. Ko, S. J. Rhee, Y. M. Kim. W. S. Kim. D. H.
Kim, J. H. Bae, Y. S. Kim, J. C. Woo, D. W. Kim and
T. Schmiedel, Superlattice and Microstructures 24.259
(1998).
[8] P. M. Petroff, A. C. Gossard and W. Wiegmann, Appl.
Phys. Lett. 45, 620 (1984).
[9] F. Michelini, L. Ressier, E. Vanelle, J. P. Peyrade, Guy
Fishman and Franccois Laruelle, J. Korean Phys. Soc.
39, 477 (2001).
[10] Y. M. Kim, W. S. Kim, Y. S. Kim, H. S. Ko, D. H. Kim,
J. H. Bae, J. C. Woo and T. Schmiedel, J. Korean Phys.
Soc. 29, 482 (1996).
[11] M. Bugajski, W. Kuszko and K. Reginski, Solid State
Comm. 60, 669 (1986).
[12] K. Oettinger, Al. L. Efros and B. K. Meyer, Phys. Rev.
B 52, R5531 (1995).
[13] H. S. Ko, D. W. Kim, W. Kim, D. H. Kim, Y. A. Leem,
Y. Kim, J. C. Woo and T. Schmiedel, Jpn. J. Appl. Phys.
37, 51 (1998).
[14] S. J. Rhee, H. S. Ko, Y. M. Kim, W. S. Kim, D. W. Kim
and J. C. Woo, Institute of Physics Conference Series
141, 355 (1994).
[15] Y. Kim, J. Korean Phys. Soc. 42, 562 (2003).