The top-down approach limits the dimensions of devices to what is technically achievable using

19
. (a) First, Au nanoparticles are randomly deposited on the wafer; they are then assembled using an atomic force
microscope as a manipulation tool (b and c); and, finally, nanowires nucleate and grow at the nanoparticle positions (d). Scanning electron micrographs are shown of three Au
nanoparticles before nanowire growth(e), after nucleation of GaAs nanowires at the same positions (f), and at an angle to show the nanowires (g). This also shows how the diameter of the
seeding nanoparticles determines that of the epitaxially nucleated nanowires. (© American Institute of Physics 2001.)
REVIEW
FEATURE
University, mainly for the fabrication of nonepitaxially
nucleated nanowires
14-16
. This work has concentrated on the
growth of nanowires directly from catalytic nanoparticles and
the self-assembly of devices and circuits from flowing liquids.
Lieber 
et al. have proposed and demonstrated the potential
for logic circuits from assembly of 
n- and p-doped Si
nanowires. Lieber has also suggested and demonstrated the
use of Si nanowires as highly sensitive biosensor devices,
allowing electrical detection of the selective adsorption of
biomolecules, such as the specific proteins related to certain
types of cancer
17
. Other efforts include those of Yang 
et al.
at the University of California, Berkeley, who have
demonstrated lasing from ZnO nanowires, along with related
studies of GaN nanowires
18
. They have also suggested the
use of nanowires for thermoelectric applications by advanced
design of one-dimensional structures.
Our own work on III-V nanowires started in 1999 and has
focused on the development of epitaxially nucleated
nanowires, with the intention of developing self-assembling,
down-scaled nanoelectronic/photonic devices. We
demonstrated
19
the potential of this approach for GaAs
nanowhiskers using prefabricated and strictly size-selected Au
nanoparticles. Atomic force microscopy was used to assemble
these nanoparticles into structures and patterns (Fig. 7). In
2001, we demonstrated the manipulation of the composition
of nanowires along their length
20
and reported the formation
of sharp interfaces between InAs and GaAs. Furthermore, we
suggested that multiple heterointerfaces could lead to the
insertion of QDs and tunnel barriers inside nanowires. The
properties of such quantum devices were reported for the
first time towards the end of 2001
21
, where crystalline
perfection and atomic abruptness, as well as ideal electrical
performance of heterostructure barriers inside nanowires,
were described for strongly lattice mismatched systems.
These results
22,23
appeared at the same time as the
realization of periodic SiGe superlattices
24
by the Berkeley
group. To complete the picture, the Harvard group also
reported
25
the formation of transitions between GaAs and
GaP inside a nanowire, and demonstrated light emission from
segments of GaAs surrounded by larger band gap GaP. These
four papers
22-25
, which all appeared in February 2002 and are
summarized in Fig. 8, resulted in high expectations of
nanowire heterostructure devices for electronics and
photonics
26
. However, it took until the end of 2002 for such
progress to be made. In the next part, I will describe a couple
of device families that have now been realized and I will
speculate on where this may lead in the future.
Bottom-up fabrication of one-
dimensional heterostructure devices
The basic building blocks for one-dimensional heterostructure
devices are the controlled growth of low and high band gap
semiconductors and the ability to form high quality
heterostructure interfaces between such different materials.
For simplicity, I will concentrate this discussion on a
summary of what has been achieved with the InAs/InP
October 2003
2 7

---

system, which consists of materials with vastly different band
gaps (0.35 eV and 1.4 eV, respectively) and lattice constants
(InAs has the larger lattice constant by ~3.5%). I will also
include some recent results for InAs/GaAs quantum
structures with highly interesting optical properties. 
It should be noted that if one attempts to grow InAs on
top of InP, growth is disturbed by one (or both) of the
following problems: the formation of three-dimensional
islands of InAs via the SK growth mode or misfit dislocations,
resulting in poor quality materials. For this reason, the
InAs/InP combination has so far been employed mainly for
use in optics, transport, or storage applications. 
Our initial results on the incorporation of segments of InP
of varying thickness in an InAs nanowire are shown in Fig. 9.
This shows, first of all, that the segments quickly adjust to
their ‘intrinsic’ lattice constants. This is clearly seen from the
spatial distribution of the origin of split diffraction spots from
the two sub-lattices in Fig. 8e. This has, after an inverse
Fourier transformation to real space, allowed us to map the
InAs (green) and InP (red) lattice constants. From high-
resolution transmission electron microscopy it can be seen
that the nanowires are perfectly defect free, probably
because of their close proximity to the open side surface. The
interface between the two binary semiconductors is abrupt
on the scale of a few MLs
22,23
. 
To assess the electronic properties of the heterointerfaces
we conducted thermally-activated transport measurements
for 
n-type InAs nanowires containing thick (80-100 nm)
barriers of InP. As seen in Fig. 9, the sample containing the
InP barrier exhibits an almost complete blockage of the
current, while the homogeneous InAs nanowire has a low-
impedance and linear current-voltage (
I-V) dependence. The
activation energy obtained by measurement of the
thermionic emission of electrons over such InP barriers gives
REVIEW
FEATURE
October 2003
2 8
Fig. 8 Main image illustrates that atomically abrupt heterostructure interfaces can be formed controllably during nanowire growth. This image, which appeared on the cover of Nano
Letters, is color-coded to show the lattice constants of InAs (green) and InP (red) as deduced from analysis of Fourier transforms of high resolution electron microscope images