See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/231009302

The future of physics of heterostructures: A glance into the crystal (quantum)

ball

Article  in  Physica Scripta · December 2006

DOI: 10.1088/0031-8949/1996/T68/014

CITATIONS READS

3 126

1 author:

C. Weisbuch
University of California, Santa Barbara
354 PUBLICATIONS   18,192 CITATIONS   

SEE PROFILE

All content following this page was uploaded by C. Weisbuch on 17 July 2014.

The user has requested enhancement of the downloaded file.

Physica Scripta. Vol. T68,102-112, 1996

The Future of Physics of Heterostructures: A Glance Into the

Crystal (Quantum) Ball
Claude Weisbuch
Departement de Physique et Laboratoire de Physique de la Mati6re CondensQ, Ecole Polytecbnique, 91120 Palaiseay France

Received June 27,1996; accepted June 28,1996

SOCRATES. I know what you mean. Do you realize that what you structure concepts (i.e. defining the physical limits of electrical to optical
are bringing up is the trick argument that a man cannot try to dis- energy conversion in (thresholdless?)lasers or in LEDs).
cover either what he knows or what he does not know? He would not We then describe some of the future physics on five examples: (i) the
seek what he knows, for since he knows it there is no need of the phonon relaxation issues in quantum dots; (ii) the search for non-epitaxial
inquiry, nor what he does not know, for in that case he does not even deposition techniques (an heterostructure “paint”), with its obvious solu-
know what he is to look for. tion through polymer/molecular electronics, considering organic molecules
MENO. Well, do think it a good argument? as ultimate heterostructure systems; (iii) the basic physics items and the
SOCRATES. No. wide applications stemming from the design of wavelength-scale structures
such as microcavities and photonic bandgap materials; (iv) the develop-
Meno, Plato (c. 429-347 B.C.)
ment of materials and structures for a full quantum control of electron or
photon dynamics; (v) the search for ultimate solutions based on novel
physics for computation or telecommunications.
Abstract This paper does not intend to be a full-fledged document on the future of
The advent of semiconductor heterostructures has opened many heterostructure physics. It should rather be thought as trying to provoke
opportunities for novel physics fields, or has enabled to study with preci- thinking on the many opportunities offered by heterostructures. An exami-
sion many concepts which were diUicult to apprehend in bulk materials. nation of the numerous past achievements of semiconductor hetero-
How can we try to delhe the future developments in the physics of hetero- structures induces one to be very modest in predicting the future: many of
structures? We identlfy four main (overlapping)directions: (i) the physics of the new physical concepts appeared by surprise; Within the predicted
the heterostructures materials system (fabrication processes, materials physics concepts or applications, most did hardly work, and some of the
issues, bandgap discontinuities, approximations and their breakdown at most prominent ones today work for other reasons than predicted!
small dimensions ...); (i) the follow-on of existing physics topics now
carried out in heterostructures; (iii) the novel physical concepts and experi-
ments which might be carried out in the future in heterostructures (a
bottom-up approach to new physics driven by the capacity of hetero- 1. Introduction
structures to act as microlaboratories for demonstrating new physical con- The field of semiconductor heterostructures is one of many
cepts, even originating from other fields of physics); (iv) the novel physics to
be done in heterostructures in order to answer new demands, be it on achievements, in many areas of physics, be it fundamental or
general motivations (i.e. advances towards the limits of computing or applied. Table 1 shows a number of discoveries in low-
communications) or on evaluating the physical limits of existing hetero- dimensional structures, as most of the progress in the recent

Table I. History of Quantum Structures: Physics and Applications

Physics Devices
1970 Proposal for Superlattices 1979 Injection Quantum Well Laser
1974 Resonant Tunneling 1980 TEGFET/HEMT/SDHT/MODFET
1974 QW Size Quantization 1983 Microwave Resonant Tunnel Diode
1978 Modulation Doping 1984 Self-Electro-opticDevice (SEED)
1980 Quantum Hall Effect 1984 Resonant Hot Electron Transistor
1981 Quantum Confined Stark Effect (RHET)
1982 2D Disorder and Localization 1986 Strained QW Lasers
1983 Room Temperature Excitons 1987 Asymetric Fabry-Perot Modulator
1984 QW Non-Linear Etalons 1987 Intraband IR Detector
1985 Giant Intraband Optical Dipoles 1989 Integral MBE Vertical-Cavity
1986 Quantum Wires and Waveguides Surface Emitting Laser (VCSEL)
1987 Quantum Point Contacts 1990 Quantum Wire Laser
1987 Superlattice to Stark Ladder Transition 199 1 11-VI blue laser
1987 Non-Local Propagation 1992 Microwave PBG crystal
1988 2D Wigner Crystal 1994 Quantum cascade laser
1989 Interband IR Giant Non-Linearities 1995 Quantum box laser?
1989 Composite Fermions 1995 GaN blue laser
1990 Electron Refraction and Interferometers
1990 Single Electron Tunneling
1990 Porous Si emission
1991 Quantum Microcavities
1995 Shot noise suppression

Physica Scripta T68

 The Future of Physics of Heterostructures 103

Table 11. Advances in Growth, Characterization and Tech- fields of physics encompassed by heterostructures, and
nology therefore shows how awkward it would be to pretend to
cover the future of such a variety of fields in a limited space
60'ies MBE
(for reviews of heterostructures and quantum semiconductor
70'ies in-situ RHEED structures, see Refs [l-lo]).
Quantum Wells and Tunnel barriers
Distributed Bragg Reflectors
It should also be remarked that heterostructure progress
New Materials (Sb) always originated from the splendid achievements of the
Lithographic Microstructuring materials scientists who were able to master the fabrication
Qualitative X-Ray and Electron Microscopies techniques to levels unsurpassed in any other field of
Interface disorder characterized materials science and who therefore could provide the
80'ies Phase-Locked and Atomic-Layer Epi various structures imagined by physicists. It might be
New Materials (P, U-VI) correct to emphasize that the intrinsic solid state chemistry
CBE, MOMBE properties of the major semiconductor systems studied so
&doping
Strained-Layers (controlled) far helped a lot [ll], as is evident from the difficulties
Nanostructuring: Lithography and Etching associated with some of the more recent materials such as
Nanostructuring: Field Effect the 11-VIS and nitrides. Table I1 provides a very concise
Nanostructuring: Direct Growth view of the various ingredients which allowed the develop-
Quantitative Electron Microscopies ment and assessment of heterostructures. There, even more
STM and AFM and NSOM
Picosecond and Femtosecond Spectroscopy than in the preceding table, progress has continued on every
item all through the period, leading to today's spectacular
achievements. It is clear that future progress of hetero-
structures will continue to rely on advances in materials and
years (i.e. since 1980 approximately) has linked discoveries structures fabrication, and this will be touched upon in this
in heterostructures to some low-dimensional phenomenon. essay.
While such a table is highly subjective and skips many Tables I and I1 give testimony of many achievements and
topics of importance, it bears testimony at once of the huge breakthroughs, but they do not present the way by which

Table 111. Heterostructure physics, past predictions record

~~ ~~~

Physics Devices Fabrication

Predicted and worked out

-Heterostructure concepts (energy barriers,
QW Quantization, variable gap -Double het. laser -Ultra-thin layers
.
quasi E-field . .) -HBT -Superlattice
-Selective/modulation doping -VCSELs pseudo-alloy
-Radiative 2D excitons -QwIPS -Quantitative e-Beam
-Resonant Tunneling -High-efficienc y and X-ray microscopies
-Impurity and Tamm SL states microcav. LED
-Stark ladder localisation -Quantum cascade laser
-Sharp single QD luminescence
-Ballistic electron motion

Unexpected
-RT exciton effects -Electron Bragg reflector -STM/AFM
- Q H Effects (. . .comp. Fermions and -Positive action of strain -Self-organized
Bosons, Skyrmions ..). -High mobility Si/SiGe growth
-Coulomb blockade -Lum. with dislocations (GaN) --Orient. dep. doping
-Strong light coupling -Porous Si -Selective oxydation
-Quantized 1D transport -Impurity-assisted
--Classical and quantum billiards interdiffusion

Undecided, elusive, very hard

Bloch oscillation -Ballistic device -Perfectly flat
Wiper crystal -Interference device interface
Solid state PM -Valence band engineered -Homogeneous
Improved optical features in QWWS, laser (for IVBA, Auger) multiple QDs
QDs @TI -PBG for inhib. spont. em.
Energy relaxation bottleneck in QDs - Q D laser
High mobility in 1D -Hot electron real-space
Direct-gap Si transfer devices
Velocity overshoot -Lateral SL devices
-1D microwave transistors

Worked for other reasons than expected

QW laser
HEMT/SDHT/TEGFET/MODFET

Physica Scripta T68

104 C.Weisbuch

those results were obtained. If instead we try to identify Having shown in retrospect how little has been predicted
which results were searched for, which came unexpectedly in heterostructures, what could be the use of an essay preci-
(although certainly not by chance) and those which hardly sely aiming at somewhat defining the future? It can serve to
came, or became true for other reasons than expected, we show how little has so far been exploited of the many
obtain Table 111. This is certainly a very subjective recollec- opportunities offered by heterostructures, and to draw the
tion of the history of heterostructure development, with attention to the many new avenues which are now opening
some items arguably belonging to several classes. However, due to simultaneous recent progress in materials science (a
the poor past record of predictions should alleviate any major provider of heterostructures) and in other fields of
temptation to wish to develop a heterostructure science base physics. We will do that in a two-fold way. Section 2 will
through predictive arguments. Although one can assess the describe how a systematic approach could be followed to
potential of a field through the leverage one expects on assess future potential of physics successively “of’, “in”,
various phenomena, the detailed outcome will often “with”, “through” heterostructures. Such a division is of
(fortunately) be at odds with expectations. A personal course quite arbitrary as will be evident as many items can
impression is that Nature has in its bag of tricks many more be switched from one category to the other, and as their
fascinating phenomena that could be expected even in the contents shift with time. We only use this classification for
wildest dreams of the “founding fathers” of the field and that the sake of simplicity of presentation, although some might
the unexpected phenomena more than offset those predic- find it unusefully cumbersome. Section 3 will deal with five
tions which did not prove true or which so far have eluded topics, selected in a highly subjective manner, which can
conclusive evidence. A better example of unexpected dis- illustrate some of the issues in the future physics of hetero-
covery is of course that of the quantum Hall effect, in par- structures.
ticular its fractional version: it can safely be said that a
whole “new continent” of condensed matter physics has
been discovered, which is now a major playground for new 2. A systematic approach to future heterostructure physics
theoretical concepts in collective phenomena, phase tran-
2.1. Physics of heterostructures: physics, fabrication and
sitions, elementary excitations, etc.. . . Among those pheno-
characterization of heterostructure building blocks
mena which did not obey predictions, we remarkably find
the two main heterostructure devices, the commercial We are mainly concerned in this part with the physics of the
success of which has helped support heterostructure various building blocks of heterostructures which allow one
research at an enviable level compared to other solid state to design the structures which will serve our purpose, such
physics subjects: the quantum well (QW) laser operates way as doing novel physics and/or devices as described in the
better than the double-heterostructure laser because one latter parts. We outline the various issues in Table IV. There
degree of freedom is frozen for electron motion and there- are clearly too many items to discuss but a few examples
fore the density of quantum states in a QW is just adjusted can show what level of physical understanding is still
to what is needed to achieve the gain required for over- required.
coming losses [l2]. The original idea of a more efficient gain Materials fabrication has still many unknowns. The
per carrier in a 2D system as compared to the 3D double- driving mechanisms in 3D self-organized growth start to be
heterostructure (DH), due to the square density of states identified [13] but more work is needed to master growth
(DOS), only accounts for a factor of two in the total speed directionality if one wishes to use it to obtain low
improvement factor of ten. In a similar vein, the excellent size-dispersed quantum dots [14]. Figure 1 shows a typical
performance of the transistors based on modulation doping polar plot of growth rates for VPE and MOVPE of GaAs
(equivalently labeled as TEGFETs, HEMTs, MODFETs or [lS]. Such diagrams are far from being understood at a
SDHTs) is only marginally due to the high electron mobility level which would allow to explore by simulation the many
but is rather due to a combination of the high intrinsic possibilities of polar and strain-oriented growth, and that
transconductance due to the short gate-to-channel distance field should stimulate more theoretical efforts. In several
and to low access resistances. subjects of this symposium it appeared that perfectly flat

Table IV. Future heterostructure phyics: Physics of building blocks

Physics of materials fabrication Physics of materials processing Physics of interfaces
-Epitaxy capability and compatibility -Etching mechanisms -Band discontinuities
-Lattice mismatch and defect foxmation-Ion-assisted -Defect “control” Adjustable barrier heights
interdiffusion -Passivation processes -Schottky contacts
-Stability issues -Ion-assisted interdiffusion prediction, tuning
-Interface morphology -A1 oxydation -O hm ic : buried, low
-3D self-organized growth mechanisms -3D interconnexion tem@rature fab.
-Fluctuations-free QDs -Functional hetero-bonding -Interfaces characterization
-Materials quality, doping, degradation-Novel materials: 11-VI, theoretical us. real world
GaN ...
-Regrowth capacity
-Perfectly flat interfaces: which way?
-Materials combinations: Semiconductors/Superconductors/
Metals/Insulators/Magnets/Organics/Amorphous.. .
-Scalable fabrication technique

Physica Scripta T68

 The Future of Physics of Heterostructures 105

due to Frenkel excitons with the highly non-linear proper-

ties of semiconductors due to their extended Wannier exci-
tons [21].
As was mentioned in Kroemer’s contribution, the main
phenomena of heterointerfaces, in particular band ofsets,
have been demonstrated and they have proven to be the
basis for the development of heterostructures. There are
however still many uncertainties in the theoretical descrip-
tions of the more complicated interfaces (such as hetero-
valent or incorporating large strains). Also needed are
theoretical descriptions dealing with real interfaces, i.e.
including chemical fluctuations and topological disorder
[22]. The area of Schottky and Ohmic contacts is still one
where experiments dominate the scene. As Kroemer did
bluntly put it: “nobody would consider a prediction if he
can have an experiment !”.
A pragmatic approach to the issue of band offsets can be
considered as follows: whereas it is often said that band
offset values are essential to many heterostructure concepts
Fig. 1. Growth rate polar diagram for VPE (solid curve) and MOVPE and devices, they often elude measurement in a given struc-
(dashed curve) (from Ref. [lS]). ture: this either proves that they are not essential in deter-
mining the properties of the structure, or that they
determine these properties in conjunction with other pheno-
interfaces would be very useful to improve the character- mena of similar importance. Therefore, we should only con-
istics of effects. However growth fluctuations have so far sider those effects and structures where band offsets indeed
prevented such interfaces to be obtained. New growth determine the properties to obtain real-world band-offset
mechanisms incorporating some flattening driving forces values. A good example is provided by the relationship
(surfactants?) are clearly needed. between peak-to-valley ratio and current density in double-
New semiconductor materials such as those extending barrier resonant tunneling diodes (see e.g. Fig. 8.14 of Ref.
bandgaps to higher energies (11-VIS, Nitrides) have yielded [23] or the various papers dealing with resonant tunneling
new growth problems for defect control, n and p doping, in the present volume). Of course, great attention should be
and at the same time some new physical mechanisms: GaN focused on the device design where band offsets play an
displays a remarkably high quantum efficiency in spite of its important role in the performance. It should however be
extremely high dislocation density [161. Radiative and/or reminded that devices are usually complex to analyze and
non-radiative recombination mechanisms are clearly quite require efforts in both measurements and modelisation : the
different from those of usual 111-Vs. mere analysis of leakage currents from active layers into
Stability and degradation issues are also of main impor- confining layers in laser diodes is still a major challenge in
tance for applications. While many of the degradation spite of its importance.
mechanisms of 111-V lasers have been identified, a lot
remains to be done: the stability properties of strained 2.2. Physics in heterostructures
materials [17] have to be better understood (it should be This topic is meant as covering the physical phenomena
reminded that for many years strain was considered a killer “naturally” occurring in heterostructures (see Table V). As
for any long term stability). Some major, very useful pro- such, the future physics there has been covered a lot in the
cessing mechanisms are poorly or not understood such as other lectures and discussion sessions of this symposium.
impurity-induced interdiffusion [18] or AlAs selective oxy- In the past a new type of physics has started to be
dation [19]. In general the most fruitful fabrication pro- addressed using heterostructures as a physics micro-
cesses are based on highly selective processes (chemical, laboratory : a long-standing winner structure in this category
directional kinetic) the basic mechanisms of which are not has been the selectively-doped interface which creates a
identified. quasi-2D metal with an extremely high carrier mobility,
Materials combinations should open many new paths for hence long free paths. Having at hand such a unique elec-
physics and applications. Only a small range of the possible tron system, many important phenomena unraveled, as is
semiconductor combinations have so far been studied. Com- well attested by the dominance of heterostructure studies at
binations with other functional materials could prove useful. International Conferences on the “Electronic Properties of
Already semiconductor-superconductors have been fabri- Two-Dimensional Systems’’ (EP2DS). It is sure that the
cated for basic studies (such as Andreev reflection) or for studies of collective modes, interactions, phases will be con-
applications (superconductor transistor). Semiconductor- tinued in the future, the more so in lower dimensional
magnetic materials combinations have been shown to pre- systems. A “new continent” of physics has been revealed by
serve both semiconducting optical properties and the Quantum Hall effects and will continue to reveal its
ferromagnetic behaviour [20]. Organic insulator- secrets as was beautifully shown by Stormer [24].
semiconductor combinations have been predicted to display Our field started for a large part based on the search for
superior non-linear optical properties due to the com- the Bloch oscillator. While this has proven so far quite
bination of the strong optical coupling of organic materials elusive, at least in its full-electronic version (for a discussion
Physica Scripta T68
106 C. Weisbuch

Table V. Future heterostructure physics: a follow-on approach of physics in heterostructures

“ P m ” physical phenomena Physics in devices Physics of devices

0 Quantum Hall effects 0 injection and 6nite size effects

0 Elementary excitations in 0 Hot electron effects 0 Microcavity effects
low-D structures 0 Recombination processes 0 Velocity overshoot effects
0 Transport in low D: correlation effects, fluctuations, radiative and non rad. 0 Tunability of light
chaos 0 Lateral superlattices effects emission
0 Optics in low D 0 1D high-mobility 0 Hot-electron and resonant
0 Bloch oscillator and Stark ladder emission 0 Electron coherence effects tunneling based devices
0 Tunneling phenomena 0 Novel quantum structures
0 Phases, interactions, collective
modes, Bose-Einstein cond.

of optically-triggered phenomena see Leo’s contributions in and von Klitzing showed that the electron wavefunction can
this volume [25]), many interesting phenomena were play a major role in a variety of experiments such as tunnel-
uncovered in accompanying or related perpendicular trans- ing, interferometric measurements and Coulomb blockade.
port studies such as resonant tunneling, ballistic hot elec-The next step would be the capacity to fully control the
tron transport [26],.. . These concepts were later transferred
dynamics of an electron wavepacket, from the preparation
of a well-defined quantum state to the measurement of the
to in-plane transport in lateral superlattices, resonant tun-
neling lateral structures etc.. . Clearly these studies will con-
final quantum state after the electron has undergone some
tinue as they provide a main tool for the physics of carrierinteractions in a microstructure. This achievement could
transport in various conditions. open the way to quantum state manipulation similar to that
Two fields which are somewhat underinvestigated are being actively pursued today in atomic physics to achieve
those of the physics in and of heterostructure devices. For quantum non demolition (QND) measurements or quantum
the former we think of such phenomena as hot electron gates.
effects (heating and cooling mechanisms), heterostructure Another needed knowledge is that of the interrelation
transport, radiative and non-radiative recombination between the microscopic, mesoscopic and macroscopic
mechanisms .. . (see Table V). For the latter we think of the
effects. So far, we have had some specially-designed struc-
detailed physics analysis of operating devices in order to tures in order to study the various effects, with coupling
provide experimental parameters to perform better model- interfaces which have a very poor efficiency. One would
lizations. The lack of efforts in these two fields might be due
wish to have much better ones in order to be able to use
to the fact that they fall between physics and engineering, microscopic and mesoscopic devices at the usual macro-
and therefore are “orphan” fields. On the other hand it is all
scopic level of electronic devices. This will require research
too natural that they have problems : physics departments efforts in circuit architecture, impedance matching concepts,
noise immunity etc.. .
are often ill-equipped to produce the structures and also are
not always aware enough of the real parameters in a real- As shown in Table VI inputs from other fields of physics
world structure. Conversely electrical engineering depart- might inspire some new physics in heterostructures : atomic,
molecular and cluster physics could provide insight in elec-
ments with their expensively built and run facilities are not
always willing to make the fully-customized structures for tronic structure, in particular help resolve some stability
physical studies while they have many “engineering” struc- issues (cf. magic atom numbers in clusters). Quantum optics
tures on the waiting list. This is all too bad as such studies
should prove a major source of inspiration for hetero-
would provide the ground for mastering future applications structure physics. Many of the newer concepts developed
and as the list of interesting phenomena is long and deals there could be put to use in solid state systems. There is
with many major effects as shown in Table V. For instance, already a surprising convergence of ideas and experiments
after some interesting early results, 1D microwave effects in both fields : sub-shot noise generation of photons through
have seen little activity in the recent years [27], although squeezing is already a major field of optics [29, 301. Besides
one might expect improved performance based on increased its basic interest for understanding and controlling Boson
fields, it will help to detect or amplify very small signals, and
1D mobility [28]. In a similar vein, hot electron devices are
not actively researched, at least at the level which could be
thus achieve new sensors (interferometers, gyros, etc.. .) with
expected in view of their potential [2]. ultimate properties. Sub-shot-noise generation is also now
widely pursued for electrons in heterostructures through
2.3. Novel physics with heterostructures either Coulomb blockade or correlation effects [31-341.
Many more possibilities to do new physics are opened by Antibunching effects in optics have their counterpart in
using heterostructures as c‘microlaboratories”. They stem Table VI. Novel physics in heterostructures
from the use of novel materials investigations and/or using
0 quantum dynamics (“true” electronic quantum effects)
inputs from other fields of physics (Table VI).
0 QBs, QWWs .. . artificial molecules ...
The more novel physics is to originate in grasping new 0 Meeting mesoscopic physics, quantum effects and heterostructures
physical phenomena to be studied in heterostructures. A 0 Interdisciplinary research
very challenging objective is to progress towards a full atomic, molecular and cluster physics : description concepts, stability
control of electron quantum dynamics in solids. In many quantum optics: entangled states, squeezed photons .. .
situations, quantum aspects only deal with electron level statistical physics: fluctuations, coopkxative effects, cell. automata,
chaos
quantization. The various talks by Eaves, Heiblum, Sakaki
Physica Scripta T68
 The Future of Physics of Heterostructures 107

Table VII. Future heterostructures physics based on rele- energy and momentum conservation for QDs sizes below
vance and outside challenges 100 nm. Their model was extended by Benisty et al. [37]
(Fig. 2) to the full relaxation cascade in QDs and extrapo-
0 Provide solutions for microelectronics beyond scaled devices
(allow for fluctuations, soft and hard errors, compatibility, lated to possible positive uses of such relaxation bottleneck
interconneXions, 3D if possible, improved pedormance . . .) effects in intersubband-transition based devices [38].
0 Explore and supply solutions for limits of computing Experimentally the situation is not quite clear: time-
(quantum computing?) resolved experiments show very fast (sub-ns) risetimes of
Supply devices for single photon communication
0
photoluminescence (PL), but usually at rather strong excita-
(sub-Poissonian emitters, single photon detectors . ..)
0 Provide physical limits of devices (i.e. conversion limit for lasers) tion densities, while steady-state PL measurements often
0 Look at life devices as ultimate heterostructures: show strong excited state emission, resulting from non-
novel devices and architectures (3D connectivity, thermalized excitations [39].
plasticity and reconfigurability .. .) The above discussion deals with the “usual” 111-V
quantum dots as obtained from selected-area interdiffusion,
direct-growth or lithographically determined QDs embed-
Coulomb blockade [31-34). While a major emphasis in ded in a semiconductor matrix. When dealing with 11-VI
atomic physics relies on the generation of entangled states QDs in solution, or in a glass or polymer matrix PL relax-
for quantum computing some schemes implementing the ation is always extremely fast wether observed C.W. or in
effect in solid state heterostructures have been proposed pulsed regime [40].
[35]. This convergence is all the more remarkable as one Models have gone beyond the simple “first-order”
deals in the electronic case with Fermions while in optics description. Several second-order mechanisms can enhance
one deals with Bosons. It must however be kept in mind relaxation such as multi-phonon relaxation and Auger pro-
that the Boson interactions leading to the above effects are cesses [39]. A more radical approach is that taken to
mediated by Fermions. Therefore, it might be argued that describe 11-VI relaxation rates where confined phonon
all effects in the end are due to Fermion or to correlated models lead to relaxation rates which increase with decreas-
Fermion properties. ing QD size as R-’ [41], at complete opposition with the
Finally statistical physics effects can be studied or used in decreasing rate as RE for continuum models [38]. A recent
heterotructures (Table VI). Among those most interesting evaluation of ripple modes in InGaAs QDs imbedded in
are the collective properties of cellular automata where one GaAs indeed points out to the importance of interface con-
uses the cooperation between cells of regular lattices to gen- fined phonon modes [42]. Clearly, one should develop the
erate combinatorial properties which could be used in novel science of electron-phonon interaction in confined systems,
computer architectures. Such systems might be at the root in addition to the study of low-D phonons.
of future computers using mesoscopic or microscopic elec-
tron devices. 3.2. Heterostructures on “anything” (towards an
heterostructure “paint”)
2.4. Future physics through demands on heterostructures: a
There is a need to be able to produce heterostructure effects
“top-down” approach ?
on materials systems that do not require epitaxy. As Herb
Although most of the new physics comes unpredicted (see
Plato’s remark at the head of this paper) demand-oriented
research can lead to superb results, even though they were
not precisely forecasted, After all, the prediction of the
electrically-driven Bloch oscillator, although not clearly
(a)
-
capture
-t
non
raciative

demonstrated, has led to the opening of a huge field of

physics in heterostructure perpendicular transport. Simi-
larly, the search for better interface transport in devices led
to 2D transport studies in Si/SiO, systems and to modula-
tion doping, and both resulted in the Quantum Hall effects.
Table VI1 displays a number of demands which could
induce or stimulate major actions in the physics of hetero-
structures. We will discuss some of the possible actions in
Sections 3.4 and 3.5.

3. Selected topics
3.1. Phonons and quantum dot relaxation
The full quantization of electron states in quantum dots
raises a major question about the relaxation of carriers
size (nm)
towards the ground state : while in higher dimensionality La:ira, bcx size (nm)
systems a given phonon will most certainly accommodate
Fig. 2. (a) Schematics of carrier in a fully quantized quantum dot; (b)
energy and momentum conservation in a transition between impact on the radiative recombination efficiency in a continuum phonon
two states of a continuum it is not so in OD systems. Bock- model assuming a non-radiative rate of 1OSs-’; (c) impact of QD lateral
elmann and Bastard [36] were the first to point out that in size quantization on the BLIP temperature of IR intersubband detectors
a continuum model for phonons one cannot satisfy both (from Ref. [37,38]).
Physica Scripta T68
108 C. Weisbuch

Kroemer reminded us heterostructures can provide almost Another fruitful approach is that of organic systems.
any desired function, in particular for devices Indeed organics, either at the individual molecule level or
(“heterostructures for everything”, see Ref. [43]). This collectively, can be considered as heterostructures, some of
approach was beautifully generalized by Capasso through them with semiconducting properties (Fig. 3) [49, SO]. They
his concept of bandgap engineering [2], and more recently of course allow large-scale deposition techniques, and at the
by Yablonovitch through the idea of photonic bandgap same time solve the various issues raised in Table VIII
materials [44-461. A major limitation to many uses of het- through their efficient, defectless fabrication techniques
erostructures is due to the contraints of epitaxial techniques based on the selectivity of the chemical organic reaction. A
(Table VIII). Besides obvious limitations, well-known since number of new basic issues however appear, such as the effi-
the start of the field (need for good-quality substrates with ciency of carrier injection at the metal-polymer interface
lattice matching), new demands are now appearing such as (what is an ohmic contact between a molecule and a
large-surface electronic systems (for displays) for which no metal?), the required carrier transport preferentially to a
epitaxy can be forecasted. At the same time the appearance light emission center (and not to the other electrode), the
of QDs points out to the need for new interconnection solu- efficiency of the formation of a localized Frenkel exciton
tions as they are an even more severe bottleneck than in before recombination etc.. . Also difficult is the stability of
present-day microelectronics. such molecular systems under strong electrical injection.
It should also be reminded that a number of properties Polymers (i.e. molecular electronics) might lie at the root of
that we desire are contradictory to the usual delocalized future electronics, as they can potentially solve a number of
properties of electrons in semiconductors: whereas good issues, but many fundamental concepts will have to be
transport properties rely on the delocalization of the wave- mastered.
function, optical properties, in particular luminescence, are
best when dealing with localized states. Indeed, for free car- 3.3. Microcavities, photonic bandgaps and engineering QED
riers, the recombination time is determined by the occu- c3,5,511
pancy factor of conduction and valence states, and is hence The control of optical modes in microcavities opens the way
quite diminished compared to occupied localized states to ultimate optical systems and devices. This is easily seen
unless dealing with very high excitation leading to degener- considering an optical beam in a limit microcavity with a
ate conditions. In addition, delocalized carriers will often ( A / T I ) ~volume. In such a cavity filled with semiconductor
meet non-radiative (NR) centers before recombining. A lot material one finds x a few lo5 electron quantum states.
of efforts have been devoted to solve this contradiction. These states can generate x loi4 photons per second by
Amorphous Silicon and porous Silicon rely on part on spontaneous emission, typically 10 to 100 times more by
carrier localization to have a reasonable PL efficiency but stimulated emission, 1000 times more if a situation where
their poor transport properties deteriorate electrolumines- strong light-matter coupling develops (see below). These
cence (EL) efficiency. Various attempts have been made to numbers show that a single mode cavity can handle useful
incorporate efficient impurity centers in semiconductors, optical beams, either through electro-optical, non-linear
such as rare-earths in Si or 111-Vs but the excitation transfer optical control devices or through emission devices. This is
efficiency between localized and delocalized excitations has at variance with single QD systems interacting with an
remained poor. The best system is certainly that of hetero- optical beam: as the minimum optical beam waist is of the
structure dots (eventually QDs, but the quantization effect is order of A/n, the optical overlap of the beam with the active
not a requirement here) which associates good transport in QD (so called optical confinement factor) is quite small. In
confining layers and good optical properties of localized addition a single QD can only generate lo5 less transitions
carriers in the dot. A very spectacular experiment has been than the optical microcavity. This asymetry is based on the
recently made by Gerard [47] where he shows that InAs
QDs grown on Si have a PL efficiency similar to that of
Ca
QDs grown on GaAs whereas a test QW has almost com-
pletely lost its carriers to dislocation-related NR recombi- z7zz
nation. Such an experiment shows that one can somewhat
relax the requirements on the transport properties of confin-
ing materials as long as carriers are more quickly trapped in AI
efficient recombination “boxes” than captured by NR ?7zz?z
centers. As natural extension is to use simpler, non-epitaxial
carrier-transporting materials to the QDs, such as conduct- Au
ing polymers. This has indeed been demonstrated by 12.1 eV
Bawendi et al. [48] where they observe EL from CdSe ???779
nanocrystallites grown in solution and later incorporated
into polymer films.
IT0
Table VIII. Heterostructures on anything-Issues m
0 epitaxy mismatch 0 intercomexions - QDs - 3D Polymer 3
0 substrate compatibility 0 chemical, functional bonding
0 large-scale “giant” electronics 0 localized us. delocalized properties
Fig.3. Schematics of “band diagram” for a polymer heterostructure LED
0 defects
(from Ref. [49]).

Physica Scripta T68

 The Future of Physics of Heterostructures 109

Table IX. Diflerent ways to control spontaneous emission

Desired property Implementation concept StNctlllT

Spectral and spatial control Redistribution of spont. emission Planar cavity

Improved efficiency Control and suppress spont. emission “Clever”planar cavity

3D cavity
photonic bandgap material
Change spontaneous emission rate Strong coupling Microcavity resonant with excitons
(requires narrow lines and microcav.)

Fermion nature of electrons (only one electron per state) control of optical modes in microcavities is on the verge of
whereas the Bosons in a single optical mode can be in any having a major impact in the LED markets, which represent
number. These two factors make the microcavity a useful more than half of the total 111-V devices markets.
system, while the single electron Q D is mainly a physics If one aims at still higher values of extraction efficiencies
tool. one needs to further suppress competing optical modes. This
Progress has been quite rapid in microcavities and falls can be achieved by using 3 D microcavities or photonic
into two categories: first, those based on a modification of bandgaps, as described by Yablonovich [U]. Eventually,
the photonic density of states in the perturbative approach combinations of microcavity effects in the vertical direction
to the light-matter interaction. These correspond to the two and in-plane 2D photonic bandgap effects might provide a
first lines of Table IX, whereby one controls directionality, simple, useable solution to the 3D control of optical modes
spectral linewidth, light extraction efficiency. The second [54]. At very high values of mode control, one should reach
category relies on the modification of the light-matter coup- thresholdless laser operation [3, 5, 511.
ling due to the selective interaction of a material exciton The strong coupling regime was recently demonstrated
with a single photon mode, leading to a strong-coupling based on the interaction of 2D QW excitons and 2D micro-
situation whereby the coupled light-matter system oscillates cavity photon modes [55]. In such a system (Fig. 5) only
between its matter and photon excited and ground states. exciton and photon states with the same energy and in-
The first approach is mainly used to control the light plane wavevector are coupled because of the translational
extraction efficiency from semiconductors : whereas only 2% invariances of the system and its Hamiltonian. Therefore the
of internally-emitted light escapes a high-index material strong coupling situation can be realized as both the exciton
(n = 3.6 for GaAs) because of total internal reflection at the damping time and the photon lifetime in the cavity can be
semiconductor-air interface, well-designed planar micro- slower than the Rabi oscillation of the exciton under the
cavity structures raise that amount to 22% [52] (Fig. 4). vacuum electric field. This gives rise to a coupled-mode
What is being achieved there is the relative increase of a behaviour, equivalently called the vacuum-field Rabi split-
vertical cavity mode at the expense of all other optical ting (VRS). Various effects on the strong-coupling regime
modes, such as guided modes and leaky modes in the DBR have been studied, in particular its time response and its
mirror. Scaling modelizations point to the possibility of impact on luminescence properties [55]. Its main interest is
reaching efficiencies up to 40-45% in planar microscavities that it provides a very unique room temperature solid-state
by using dielectric DBR mirrors with index differences of system in which many of the quantum optics concepts of
the order of unity [53]. Such values of DBR index differ- strongly-coupled systems developed in the much bulkier
ences can now indeed be reached by using the newly devel- atomic beam-cavity systems can be applied.
oped A10, DBR mirrors [19]. With such values of The study of 2D excitons in microcavities adds an impor-
extraction efficiencies at hand, it can safely be said that the tant “missing” link into the properties of light-matter inter-
action in systems of varying dimensionality of electronic and
w ! p
A,
+ GeAs
Fadry Perot
enhrrsati matie
photonic modes (Table X). In the bulk the interaction of 3D
(a) 2 O
~~I ; -? s
*A As (b)
des
excitons with 3D photons leads to the well-known exciton-
polariton coupled-mode behaviour due to 3 D translational
GaAdAIAs DBQ
invariance. In such a situation the exciton-photon inter-
action does not lead to dissipation and luminescence
becomes an extrinsic process. A similar situation exists for

Table X. Photon interactions in systems with varying photon

and electron dimensionality
BUk 2D Exciton 0 Strong coupling
3D photon 0 Exciton-polariton
k,, selection rule 0 Extrinsic radiative process
Quantum well 2D Exciton 0 Weak coupling
3D photon
k,,, selection rule Intrinsic radiative process
~

0 io !S ic 25 0
<;,I,-“ ‘r”W

Planar cavity 2D Exciton 0 Strong coupling

Fig. 4. (a) Schematics of a high-efficiency microcavity LED; (b) Schematics 2D photon 0 Cavity-polariton
of the various optical modes in such a cavity; (c) Power output and extrac- k , selection rule 0 Extrinsic radiative process
tion efficiency of the LED (from Ref. [52]).
Physica Scripta T68
110 C. Weisbuch

gates, quantum computation systems [57, 581, quantum

optical taps (photon “cloning” systems) [59].
These experiments are both essential to unravel the very
(a) DER Reflector quantum properties of small systems (through quantum
non-demolition QND measurements for instance) and to
obtain ultimate performance systems. A prerequisite for the
L latter is therefore to generate fluctuations-free electron or
?SA QW
Air photon systems [60]. This possibility has been recently
demonstrated on either photon states (squeezing [29]) or in

LA
Oplicai Field
reflectors electron states through Coulomb blockade or correlation
effects. The next step relies on the quantum manipulation of
quiet electrons and photons. This has already been achieved
(c) COUPLED MODES in a number of situations for optical and atomic systems
such as quantum-state entanglement or quantum cryptog-
raphy.
Heterostructures might play a major role. In the optics
field high-efficiency microcavity emitters might lead to
photon-number squeezed-state generation [44,601. Another
scheme to provide such states would be to use strongly-
coupled systems such as the exciton-based microcavity [60].
2 Conversely, microcavity detectors might improve the noise
properties up to single photon detection levels. In the elec-
CAVITY OSCILLATOR EXCITON OSCILLATOR tronic properties field, one can foresee heterostructure
systems where one would control the electron quantum
Id) External emission angle (deg) dynamics and interactions in order to obtain electronic
entangled states [35]. If this would prove feasible it would
, I

-10 0 10 20 30 40

open the way to a truly single electron electronics. Of course

one would like to obtain “robust” structures and effects at
“reasonable” operating temperatures.

3.5. Heterostructures and the future of microelectronics

It has often been said that heterostructure physics, in partic-
ular nanoelectronics, was preparing the way to future
microelectronics. Should we consider such a statement as
0 2.0 108 4.0 lo6 valid? While it is true that the post-SIA (semiconductor
Wave vector (m- ’1 industry association)-roadmap era is not yet clear, it is not
Fig. 5. (a), (b) Schematics of a strong-coupling microcavity;(c) Schematics obvious that today’s nanoelectronics concepts will provide
of the coupled modes in that cavity; (d) Dispersion curve of the coupled solutions beyond scaled Si CMOS. It is now well-
exciton and photon modes as deduced from the angular dependent photol- established that transistors will operate as we know them
uminescence. today with gate lengths down to 0.07pm. It does not pre-
clude transistors to operate well below that value with an
microcavity excitons and 2D photons where the lifetime acceptable degraded mode of operation. We should also
becomes finite due to the photon leakage out of the cavity. keep in mind that elementary logic gates are already very
The 2D QW situation in a 3D photon field is peculiar: there efficient, operating at the femtojoule speed-power product
is no k-conservation in the optical transition and therefore level. A single electron transition (SET) operating at room-
no strong-coupling can develop. The recombination process temperature under one Volt would “only” be ten thousand
is intrinsic, with a very short lifetime. This was remarked a times better.
long time ago by Agranovitch, and only recently theoreti- It therefore seems that the main gains of future microelec-
cally rediscovered (see Ref. [56]), although it explains why tronics are to be found elsewhere than just through further
QW intrinsic exciton radiative recombination dominates diminished dimensions. Let us first make our point by dis-
impurity-related recombination, at strong variance with cussing computation energy requirements : as is well-known
bulk materials. reversible computation requires zero energy, irreversible
single-bit computation kT log 2 [61]. A good device today
3.4. “Fully quantum” electronics and optics uses 1fJ, i.e. 3.105kT. However a good microprocessor with
The terms quantum electronics and optics have been used 200 MOPS and a 10W power consumption uses 1013kTper
for long to describe phenomena where “some” quantum operation. Of course such an operation contains many
effects take place, such as in light-matter interaction (as in single bit operations but there still is a huge ineffectiveness
laser physics), or in single-photon mode phenomena (as in factor.
QED). We introduce the adjective “fully” in order to focus As is well-known “biological computation machines” are
on those phenomena which rely on the “true” quantum much more efficient. DNA replication uses 20kT per base
dynamics of electrons or photons states, as has become cus- molecule copy [61], including irreversibility (only DNA
tomary in recent experiments such as quantum optical must be copied, RNA must not leave its imprint on the
Physica Scripta T68
 The Future of Physics of Heterostructures 111

DNA!) and proofreading [61, 621. The bee’s brain, contain- structures, new uses of heterostructures as microlaboratories
ing lo6 neurons with 1000 synapses each and operating at for the study of novel physical phenomena. Whereas many
1000Hz has a processing power in the TOPS range, trans- of the discoveries will come from undirected research,
lating in a 1000kT energy per operation assuming a lOpW demands from various applications fields such as the search
power consumption [63]. This clearly an efficient comput- for ultimate computational or telecommunications systems
ing machine ! should also trigger major advances. There is certainly a lot
As is obvious from the preceding numbers a lot of of physics to discover on the way to single electron compu-
progress is to be obtained from better architectures and tation or single photon telecommunications !
better algorithms. It could be added that the neural net-
works of the brain operate in an analog mode [64], with 5. Acknowledgements
very high-efficiency 3D interconnected architectures and a
It is a pleasure to thank all those with whom I collaborated
plasticity capacity which allows learning.
along the years in the various aspects of heterostructures. In
In view of these remarks what could heterostructures
1979-1981, I had the good fortune to be at Bell Labor-
provide to have an impact in future microelectronics? A first
atories, working with Ray Dingle primarily, and interacting
action is on the evaluation of ultimate material properties.
at various degrees with A. Cho, W. Tsang, A. Gossard, P.
One way is to explore the ultimate possibilities of electron
Petroff, H. Stormer, C. Shank, J. Hegarty, M. Sturge, R.
functionality in heterostructures as described above in para-
Fork and many others. In Thomson during 1983-1992, the
graph 3.4 on electron quantum dynamics, with the objective
work on quantum well lasers was mainly carried out by
of achieving “robust” single electron electronic systems.
(then) graduate student J. Nagle, with collaborations with J.
Communications also play a role in many applications and
P. Hirtz, M. Razeghi, P. Him, B. De Cremoux, while the
ultimate single photon communications demonstrations
work on quantum dots was carried out with J. Nagle, H.
using some of the new concepts described in 3.3 and 3.4
Benisty and E. Bockenhoff, with collaboration with C. M.
could have an impact on future systems. Another field of
Sotomayor-Torres from the University of Glasgow. A very
investigation would be to associate SET-like devices in
fruitful collaboration developed with Michael Kelly, then at
mesoscopic systems in order to achieve some deterministic
GEC Hirst Research Center, now at University of Surrey,
combinatorial functionality (such as a simple gate) allowing
during an ESPRIT programme on quantum devices (1984-
for fluctuations, both in fabrication and single device oper-
1987). Another ESPRIT basic research programme
ation. One would certainly need for that feedback actions
(QUANTECS) followed (1988-1991), led by S. Beaumont
and/or some degree of device cooperation. A reverse-
(Glasgow University) and involving J. Kotthaus (Munich),
engineering approach would try to define workable archi-
S. Williamson (Philips), K. Harmans (Delft), F. Briones
tectures and minimal requirements at the single device
(Madrid). The work on the strong coupling in microcavities
performance to achieve deterministic systems output. The
was started as an NTT chair visiting professor at the Uni-
next step would then be to associate these mesoscopic
versity of Tokyo, in Professor Y. Arakawa’s laboratory, and
systems into large-scale microelectronic systems interfaced
also benefited from the collaboration of A. Ishikawa and M.
to the usual microelectronics standards.
Nishioka. The recent work on microcavities is carried out in
A concurrent approach should focus on the weaknesses of
the Ecole Polytechnique Federale de Lausanne with R.
existing microelectronics solutions and evaluate whether
Houdre, R. Stanley, U. Oesterle, P. Pellandini, M. Ilegems.
nanoelectronics could remove some of the bottlenecks of
The present work on photonic bandgap materials and
present-day electronics, such as the 3D interconnection
microcavities at Ecole Polytechnique, Palaiseau, is done
problem which limits the exploration of novel architectures
with H. Benisty, D. Labilloy and P. Mardon. The hospital-
or algorithms (such as neural ones). Of course one of the
ity of the “departement de physique” and “Laboratoire de
solutions could be polymer electronics. On the long term it
Physique de la Matiere Condensee” are gladfully acknow-
might well be the better solution provided the difficulties
ledged. The work is partially supported by the ESPRIT
mentioned in 3.2 are solved, and in addition that determin-
Basic Research project “SMILES”.
istic interconnections are feasible at will with polymers with
the isolation properties that are essential in today’s elec-
6. References
tronic circuits. We are very far today of any fabrication
roadmap or even elementary modelizations to achieve 1. For a general reference in the field, see e.g. Weisbuch, C. and Vinter,
molecular electronics, as only very elementary and poor B., “Quantum SemiconductorStructures”(Academic, Boston 1991).
2. Capasso, F., “Physics of Quantum Electron Devices” (Springer, Berlin
functionalities have been demonstrated at the single mol- 1990).
ecule level, and as we do not know yet how to interconnect 3. Burstein, E. and Weisbuch, C., “Conhed Electrons and Photons:
molecules. New Physics and Applications”(Plenum, New York 1995).
4. See, e.g., van Houten, H. and Beenakker, C. W. J., “Confined Elec-
trons and Photons: New Physics and Applications” (Edited by E.
4. Conclusion Burstein and C. Weisbuch) (Plenum, New York 1995), p. 296; Beenak-
ker, C. W. J. and van Houten, H., Solid State Physics 44, 1 (Academic,
In spite of Plato’s argument (often referred to as the argu- Boston 1991). An outstanding short introduction can be found in Van
ment for “lazyness”) recalled at the beginning of this paper Houten H. and Beenakker C., Physics Today, July 1996, p. 22.
we tried to discuss some of the future physics in hetero- 5. Rarity, J. and Weisbuch, C., “Microcavity and Photonic Bandgaps:
Physics and Applications” (Kluwer,Dordrecht 1996).
structures by showing that the field is still full of potential. It 6. Kelly, M. J. and Weisbuch, C. (Editors), “The Physics and Fabrication
seems that we have only explored so far the tip of the of Microstructures and Microdevices” (Springer, Berlin 1986).
iceberg with much more to come from novel materials I . Kelly, M. J., ”Low-Dimensional Semiconductors: Materials, Physics,
associations, studies of molecules as ultimate hetero- Technology, Devices” (Oxford University Press, Oxford 1995).
Physica Scripta T68
112 C.Weisbuch
8. Davies, J. H. and Long, A. R., “Physics of Nanostructures” (IOP 40. See Bawendi, M., in Ref. [3].
Publishing, London 1992). 41. Nomura, S. and Kobayashi, T., Solid State Commun. 82,335(1992).
9. Reed, M. (Editor), “Semicoductors and Semimetals” (Academic, 42. Knipp, P. A. and Reineke, T. L., Phys. Rev. B52,5923(1995).
Boston 1990),Nanostructured Systems, vol. 35. 43. Kroemer, H., Japan, J. Appl. Phys. 20,suppl. 20-1,39 (1981).
10. Reed, M. and Kirk, W. P. (Editors), “Nanostructures Physics and 44. See e.g., Yablonovitch, E., J. Opt. Soc. Am. B10, 283 (1993),and the
Fabrication” (Academic, Boston 1989). other papers in this special issue of the J. Opt. Soc. Am. on Photonic
11. See e.g. Yablonovitch, E., Science 246, 347 (1989). Bandgap Materials.
12. See e.g. the discussion in Ref. [l]. A more detailed discussion can be 45. Joannopodos, J. D., Meade, R. D.and Winn, J. N., “Photonic Crys-
found in Weisbuch, C., Brazilian J. Phys. 26,21(1996). tals’’ (Princeton University Press, Princeton 1995).
13. Priester, C. and Lannoo, M., Phys. Rev. Lett. 75,93 (1995). 46. See also the special issue of the Journal of Modem Optics 41, Feb-
14. Eberl, K. et al. (Editors), “Low-Dimensional Structures Prepared by ruary 1994 and Soukoulis, C. M. (Editor), “Photonic Bandgap
Epitaxial Growth or Regrowth on Pattemed Substrates” (Kluwer, Materials” (Kluwer, Dordrecht 1996).
Dordrecht 1995). 47. Gerard, J. M., Cabrol, 0. and Sermage, B., Appl. Phys. Lett. 68,3123
15. Horikoshi, H., Ando, S.,Ando, H. and Kobayashi, N., in Ref. [14], p. (1996).
325. 48. Dabbowsi, B. 0. et al., Appl. Phys. Lett. 66, 1316 (1995).
16. Lester, S. D., Ponce, S. A., Craford, M. G. and Steigerwald, P. A., 49. Greenham, N. C., Moratti, S.C., Bradley, D. D.C., Friend, R. H. and
Appl. Phys. Lett. 66, 1246 (1995). Holmes, A. B., Nature 365, 628 (1993);see also Greenham, N. C. and
17. See e.g. Morkoc, H., Sverdlov,B. and Guang-Bo, Gao, Proc. IEEE 81, Friend, R. H., Solid State Physics, 2 (1995).
493 (1993). 50. Petty, M. C., Bryce, M. R. and Bloor, D. (Editors), “An Introduction
18. See e.g. the special issue of Opt. Quantum Electron. 23,S779 (1991). to Molecular Electronics” (Edward Amold, London 1994).
19. Dallesasse, J. M. et al., Appl. Phys. Lett. 58, 394 (1991). 51. For simple references on microcavities, see e.g., Yamamoto, Y. and
20. De Boeck et al., Appl. Phys. Lett. 68,2744(1996). Slusher, R., Physics Today 46,66 (1993);Slusher, R. and Weisbuch,
21. Agranovitch, V. M., Atanasov, R. and Bassani, F., Solid State C., Solid State Commun. 92,149(1994).
Commun. 92,295(1994). 52. Blondelle, J. et al., Electronics Lett. 31, 1286 (1995);also in Ref. [SI.
22. See e.g. Ourmazd,A., Materials Science Rep. 9,201 (1993). 53. Benisty, H. and Weisbuch, C., unpublished.
23. Yokoyama, N. et al., in Ref. [2], p. 253. 54. See e.g. the contribution of Baba, T., in Ref. [SI.
24. Stormer, H. L., this volume. 55. See e.g. Stanley, R. et al., Phys. Rev. B53, 10995 (1996),and references
25. Leo, K., this volume. therein.
26. See the various contributions in Ref. [2]. 56. Andreani, L. C., in Ref. [3], p. 57.
27. Ouda, K. et al., Proc. IEDM ’89,125 (1989). 57. Deutsch, D. and Jozsa, R., Proc. Roy. Soc. Lond. A439, 553 (1992).
28. See e.g. Sakaki, H., Surf. Science 267,623 (1992). See also the contributions by Brune et al. and Lange et al. in Ref. [SI,
29. See e.g. the special issue of Phys. Reports 219,215(1992). Glanz J. in Science 269, 28 (1995);Turchette, Q.A. et al., Phys. Rev.
30. Rarity, J., Physics World June 1994,46(1994). Lett. 75, 4710 (1995), and references therein; Di Vincenzo, D. P.,
31. Kowenhouven, L., Science 271,1689 (1996). Science 268,255 (1995).
32. Lief&, F. et al., Superl. and Microstruct. 16,253(1994). 58. Lloyd, S.,Sci. American, Oct. 95,44(1995).
33. Birk, H., de Jong, M. J. and Schonenberger, C., Phys. Rev. Lett. 75, 59. Bencheikh, K., Levenson, J. A., Grander, Ph. and Lopez, O., Phys.
1610 (1995). Rev. Lett. 75,3422(1995).
34. Kumar, A. et al., Phys. Rev. Lett. 76,2778(1996). 60. Yamamoto, Y., in: “Quantum Optics of Confined Systems” (Edited by
35. Benjamin, S.C. and Johnson, N. F., Phys. Rev. B51,14733(1995). M. Ducloy and D. Bloch) (Kluwer, Dordrecht 1996),pp. 201-281.
36. Bockelmann, U.and Bastard, G., Phys. Rev. B42,8947(1990). 61. Bennett, C. H., Int. J. Theor. Phys. 21, 905 (1982);Bennett, C. H. and
37. Benisty, H., Sotomayor-Torres, C. M. and Weisbuch, C., Phys. Rev. Landauer, R., Sci. American, July ’95,48(1995).
B44,10945 (1991). 62. Radman, M. and Wagner, R., Sci. American, Aug. ’88,24(1988).
38. Benisty, H., in: “Phonons in Semiconductor Nanostructures” (Edited 63. Sejnowski, T. J. and Churchland, P. S.,in: “A New Era in Computa-
by J.-P. Leburton et al.) (Kluwer, Dordrecht 1993),p. 447;Benisty, H. tion’’ (Edited by N. Metropolis and G.-C. Rote) (MIT Press, Cam-
and Weisbuch, C., unpublished. bridge 1993),p. 167.
39. See the various relevant contributions by Arakawa, Y., Bockelmann, 64. Mead, C., ”Analog VLSI and Neural Systems” (Addison-Wesley,
U. and Gerard, J. M., in Ref. [3]. Reading 1989).

Physica Scripta T68

View publication stats