i. COMBINED UHV AND LIQUID PROCESSING (CULP)
The fabrication of thin films and interfaces is an important part of all advanced
microelectronic and optoelectronic devices. The majority of solid interfaces
in use in the microelectronic industry are fabricated by depositions in vacuum
or in reduced atmosphere. Certain interfaces, notably the oxide-Si interfaces,
are formed by solid-state diffusion and reaction in an oxidizing ambient.
By design, the growth morphology of deposited films is usually dominated by
the deposition/reaction kinetics to achieve layer uniformity. As smaller and
more complicated structures are being pursued to increase packing density
and to make use of quantum effects in next-generation devices, there is an
increasing interest in exerting better control on the growth morphology, or
creating novel morphologies. Since surface/interface energies play a very
important role in the growth morphology, it is obviously desirable to be
able to modify these energies to access a wider range of morphologies.
In common deposition techniques, the surface structure of the overgrowing
film is often that of clean crystal surfaces or hydrogen-terminated surfaces.
There is not much room for the manipulation of surface energies under these
With nanoscale science and engineering identified as a focus area for our
national research programs and also for other high-tech nations, the interest
in the self-assembly of mesoscopic structures and quantum devices has surged
in recent years. Self-assembled nanostructures may be used to increase the
functionality of future devices in applications as profound as quantum
computing and single electron transistors and as practical as the control
of nano-pores in low-permittivity (low-K) dielectric layers. To take full
advantage of the increased functionality of nanoscale devices without paying
the high price of advanced lithography, it is desirable to be able to self-assemble
a wide range of materials for different applications. One needs techniques
that can produce nanostructures not only for known device structures, but
also for future devices with yet unforeseen structures. Therefore, it is
important to expand the capabilities and increase the current knowledge of
self-assembly by exploring novel concepts for more flexibility and versatility
Present techniques employed for the self-assembly of quantum dots
essentially use surface/interface energies and/or limited reaction
kinetics to achieve nanometer characteristic size. For one reason
or another, these techniques are limited to specific material systems.
For example, the growth of quantum dots in lattice-mismatched
heteroepitaxial systems depends on thermodynamic driving forces,
i.e. inhomogeneous strain relaxation in the Stranski-Krastanow (SK)
to achieve self-assembly. Thermodynamic forces are
also tapped for the self-assembly of nanoparticles in host materials,
a well-known example of which is the segregation of Si quantum dots
in Si-rich silicon oxide or silicon nitride films.6-8
One notes that the particle-substrate interaction plays an important
role in these examples of nano-fabrication, which limits the applicability
of these techniques to specific material systems. Specifically, the SK
growth mode is limited to hetero-epitaxial systems within some optimum range
of lattice mismatches, e.g. Ge/Si, InAs/GaAs, etc. Also, the phenomenon
of phase separation is system-specific as it depends on the equilibrium phase
diagram. In contrast, the assembly of nanoscale Si particles in the gas
phase,9 in plasma-enhanced chemical vapor deposition, relies
on the control of the duration of the gas supply to adjust the size of
the particles. However, the growth of nano-particles in the gas phase,
or the formation of colloids from solution,10-12 depends on
the chemistry and, therefore, these methods are material-specific
and are not applicable to a wide range of materials. We propose to
investigate an entirely novel approach of nano-scale self-assembly,
which avoids some of the limitations of methods mentioned above and has
good prospect for applications in a wide range of material systems.
This novel approach, tentatively referred to as the "combined UHV and
liquid processing" (CULP) method, makes use of an inert liquid film,
to mediate the growth of physically deposited thin films and to accommodate
the self-assembly of nanoparticles. Investigations along these directions
will also allow us to examine the broader issue of liquid-solid interaction
at the nanometer scale.
The basic idea of CULP is illustrated in Fig. 1,
which shows a typical, single-component, phase diagram with
temperature and pressure as the coordinates.
The liquid phase exists only above the temperature and pressure
of the triple point (t.p.). For common, chemically-inert gases,
the triple point pressure is either above or close to atmospheric
pressure. It thus seems that standard thin-film deposition techniques,
e.g. vacuum deposition, cannot be used to explore the potential
benefits of the liquid phase. (There is a pressure difference
of 13 orders of magnitude between liquid and UHV.)
We suggest that this problem can be circumvented by a two-step
process. In the first step, the intended material is
co-deposited along with chemically inert molecules
(for example, SF6, Xe, etc.) as a mixture at an appropriate
(cryogenic) temperature onto a substrate, or, a substrate
already buffered with a thin layer of adsorbed gas molecules.
Because of the high-vacuum and low-temperature conditions
employed, co-adsorbed films are in the "solid" state, as
represented by point "C" in Fig. 1.
In the second step,
the sample with the solid film is maintained at cryogenic
temperatures while the ambient pressure is increased to above
the triple point pressure (to point P in Fig. 1),
by back filling with cold gas in a special-purpose chamber.
After reaching the desired pressure, the sample is allowed to
warm up, which brings the adsorbed film to the liquid phase
(L in Fig. 1). The high diffusivity in
the liquid film (> 1x10-5 cm2/s) facilitates
the self-assembly of particles with sizes dependent on the
concentration, the film thickness and other parameters.
The thermodynamic driving force for the assembly of particles
is a larger cohesive energy of the solid phase than the atom-liquid
interaction. Partial drying of the liquid can also be explored
at this stage for selective deposition of nanoparticles on
patterned substrate. Alternatively, to reduce the effect
of surface tension, the liquid film can be refrozen into the
solid phase, or taken above the critical point (c.p in
for super-critical drying.
Click here for references.
To the best of our knowledge, the processing of co-adsorbed
deposit/inert-medium films through the solid-liquid boundary
has never been attempted for nano-fabrication. There are
many potential benefits of the proposed CULP technique.
Essentially, this technique combines the cleanliness and fine
control associated with UHV deposition with the favorable
kinetic (physical) processes available in the liquid phase for
self-assembly. This combination can lead to the self-assembly
of nanoparticles of a wide range of materials, including metals,
semiconductors, and insulators. The size of the nanoparticles
in the present approach is controlled not only by the atomic
concentration in the solution but also by the thickness of
the liquid film, which implies that formation of dots smaller
than that achievable in macroscopic liquids may be possible.
Furthermore, because of the additional constraint related to
the film thickness on the self-assembly process, it is
conceivable that CULP can achieve a more uniform distribution
in the size of the self-assembled particles than by some of
the other methods. It is also conceivable that selective
deposition of nano-particles with the existing topographical
and/or material patterns on the substrate can take place
in partially dried liquid due to the line tension of the
liquid droplets. Selective deposition is very advantageous
because of its simplicity and low-cost. In terms of
scientific novelty, the proposed research will probe a
potentially fertile area, which has not been extensively
examined in the past. The co-adsorption of molecules and atoms
in the nanometer thickness regime and the kinetic processes in
ultra-thin liquid films are all interesting subjects of general
scientific interest. They are of particular importance for
the science of nano-fabrication. One expects that the deposition
of nanoparticles or films by the CULP method is essentially
non-disruptive of the substrate surface, i.e. extremely
"soft-landing", which can lead to interface atomic structures
that are significantly different from that found at more reactive
interfaces formed by conventional deposition methods. This
change in the interface structure will likely lead to a different
band-offset/barrier-height for electronic transport across the
interface. The variation of the barrier height revealed in the
proposed experiments is expected to answer some remaining questions
concerning the formation mechanism of the barrier height.
Furthermore, the ability to vary the barrier height is very valuable
in applications such as metal gates and contacts. Last, but
not least, the proposed research uses very simple concepts and is
tied closely to potential applications, making it an ideal vehicle
for the education of graduate and undergraduate students. The
proposed collaboration with an industrial research laboratory
also offers an excellent opportunity to train students for careers
in either industrial or academic environment.
ii.a Surface Adsorption and Wetting
The behavior of thin liquid films at solid surfaces has been actively studied
for decades, and has been vital in the development of processes in the
pharmaceutical, bio-sensor/bio-technology, petroleum, and lubricant industries.
It has long been realized that the contact angle q
of liquid droplets on solid
(in the presence of vapor of the liquid species) is related to the energies
of the liquid-solid interface ssl,
the solid (-vapor) surface ssv, and the
liquid-vapor interface slv by Young's equation,
ssv = ssl +
A measurement of the contact angle, or the related capillary effect,
may be used to deduce the magnitude of the surface and interface energies.
Cahn has pointed out the existence of a surface wetting transition in
the two-phase region of a critical liquid mixture for weak interacting
systems,13 and the scaling behavior has been theoretically
investigated.14 Developments in the theoretical and experimental
aspects of the liquid-solid interaction were recently reviewed.15-17
For the adsorption of simple molecules (inert gases, H2,..)
an approximately "global" wetting behavior has been observed experimentally
and modeled based on the parameters of the van der Waals interactions
between the adsorbate gas molecule and the substrate.18
(Si, GaAs,..) surfaces are expected to have wetting temperatures below
their triple point temperature, because of the strong interaction with
inert gas molecules. When electronegative materials are co-adsorbed along
with an inert gas, the adsorption of the inert gas is expected
to be enhanced over the pure inert gas adsorption. Under such favorable
conditions, the co-adsorption of multiple layers of a mixed inert gas with
a range of materials is likely. The adsorption of larger, yet chemically
inert, molecules such as SF6 and C2F6,
has also been studied extensively.19, 20 However, the prospect
for co-adsorption of mixed films is less certain, because the potential
associated with these molecules are more complex. Based on the higher
triple point temperature of these molecules, the expectation is that
it should be possible to find material-gas combinations which condense
into thick films at reasonable temperatures.
Click here for references.
Some relevant information on selected, chemically inert, gases is given in
The sustained adsorption of gas molecules often occurs at temperatures
significantly below the triple point temperature. On the other hand,
the liquid phase processing requires pressures in excess of the triple
point pressure. Therefore, one looks for gases with high triple point
temperature and low triple point pressure to use as the medium in the
proposed experiments. The triple point temperature should not be too
high, however, so as to avoid "sintering" of the nanostructure at
temperatures required to form the liquid and to desorb the gas completely.
One caveat is that when the liquid contains impurities, such as in the CULP
scheme, all the boundaries between phases in Fig. 1
may be shifted from the
single-component condition. Therefore, in choosing the experimental conditions,
values in Table I are for reference only. Among
the inert gases, xenon is the most convenient for these experiments
because of its triple point temperature of 131.6K and its triple point
pressure of 81.6kPa (about 80% of atmospheric pressure).21
One drawback of Xe is the low temperature required for thick film adsorption.
Previous reports indicated that a temperature of 30-60K was adequate for
thick film adsorption of pure Xe on a variety of substrates.22-24
It is expected that this temperature range will also be sufficient for
the co-adsorption of Xe with noble and late transition metals, because
of the favorable parameters for van der Waals interaction between gas
molecules and these metals. Despite a low triple point temperature,
oxygen adsorption may be possible under the present experimental conditions,
thanks to an unusually low triple point pressure ~ 0.15 kPa. In
choosing gases which can be vacuum-adsorbed in thick films at a more
convenient liquid nitrogen (LN2) temperature (77K), sulfur hexafluoride
(SF6) and hexafluoroethane (C2F6),
with triple point temperatures of 222.3K and 173.1K, respectively,
are found to be attractive among the chemically inert molecules.
Carbon dioxide has been widely used in sol-gel processes with supercritical
drying. Its use is not favored in the present experiments with UHV equipment,
however, because of a high triple point pressure, ~ 518 kPa.
The present technique involves the evaporation of an intended material
along with the flux of chemically inert molecules to form a mixed,
physi-sorbed layer at appropriate (low) temperatures on selected substrate
surfaces. Previously, volatile buffer layers have been used to minimize
the interaction between the overlayer and the substrate. Notably,
thin metal layers were deposited on top of thick adsorbed Xe layers
on compound semiconductor surfaces at cryogenic temperatures.25, 26
The Fermi level position of interfaces formed by such buffered deposition
was reported to be significantly different from that obtained by atomic deposition.
In these early investigations however, the adsorbed Xe layer was thermally
desorbed in high vacuum for subsequent in-situ photoemission measurements.
That procedure led the system through the solid-gas boundary (see dashed line
from C to V in Fig. 1)
without passing through the liquid phase. Large, slender,
irregularly-shaped, islands of metals, with ~ 100nm width and ~ 1um length, were formed as a result of 0.7nm deposition.22, 27
The shape of these islands was likely related to the desorption process
of the buried Xe layer, e.g. pressure rupture from gas bubbles.
The interpretation of the low-temperature photoemission data from isolated
metal islands processed through the Xe layer was called into question later,
when possible effects due to surface photovoltage (SPV) were discovered.28, 29
In contrast to the early studies, the present CULP method uses not only
an inert buffer layer but also co-adsorption of inert molecules along
with the intended material. More importantly, the CULP method attains
a liquid state before the desorption of the volatile gas molecules.
Therefore, the shape and size of particles deposited by the CULP method
are expected to be completely different from that seen in these early experiments.
Surface tension during drying can lead to inhomogeneous surface morphology
of the liquid layer and, subsequently, inhomogeneous distribution of the
self-assembled nanoparticles. A schematic of the heterogeneous morphology
of deposited clusters, as a result of line tension of the liquid meniscus,
is shown in Fig. 2(d). The actual morphology depends on various parameters,
such as the adhesion between the particles and the substrate, the energies of
the liquid-vapor, liquid-substrate, and the liquid-cluster interfaces, etc.
In the limit of strong cluster-substrate adhesion and small liquid-vapor
surface energy, the deposition of the assembled nanostructures is expected
to be homogeneous and independent of the drying process, as shown in a
schematic fashion in Fig. 2(c). In the opposite limit of large surface
energy, the drying of the liquid is expected to leave behind structures
that are heterogeneous and "patchy". However, since the drying process
is sensitive to surface defects and undulation,16, 30 in addition to
surface/interface energies, possibilities exist for self-aligned processing
of nanostructures on substrates pre-patterned with either chemical or
topological disparities. (See Fig. 3.) More frequently, however, it is
desirable to deposit an essentially uniform layer after liquid-phase processing.
There are two common ways to reduce or eliminate surface-tension related layer
inhomogeneity. One is to dry the liquid under super-critical
conditions,31-33 where the surface tension vanishes.
This option is not available in the present experimental set-up
due to the incompatibility of the mechanical design and material
strength used in conventional UHV equipment with high pressure.
(Critical pressures for gases presently proposed are typically ~ 3-6 MPa.)
However, it is recognized that the pressure and the temperature should
be kept as high as possible during liquid processing to minimize the
effect of surface tension. Another method to reduce
"freeze-drying",32, 34, 35 which is feasible in the present experiments.
After the film is processed in the liquid phase, it can be (re-)frozen into the
solid phase. The system is again pumped down, while the film remains solid
(going from point L to point C in Fig. 1).
Warming up the film then crosses
the solid-gas boundary (C to V in Fig. 1)
and allows the gas molecules
to desorb in vacuum, leaving behind the self-assembled film. Surface
tension will be of little significance in the freeze-drying process.
ii.b Self Assembly of Nanoparticles
In the self-assembly of quantum dots, the size of the quantum
dots is often controlled by the supply of material, which
limits the degree of coarsening. For example, the size of the
epitaxial quantum dots grown by strain relaxation depends on the
average thickness of the epitaxial film, and the size of the
nanoparticles assembled in plasma decomposition depends on the pulse
duration of the gas supply.9 The formation of particles from a
dilute solution depends on the concentration of the solution and
some characteristic diffusion length, as shown in Fig. 4(a).
However, it has been possible to achieve uniform distribution in
the size of colloidally formed quantum dots using the self-limiting
nature of the growth.11 In that process, the size of the quantum
dots is not limited by material supply and the solution used is of
macroscopic dimension. In the present CULP approach, there are two
limiting factors on the size of the quantum dots beside the diffusion
length: the amount of the deposited material, as controlled by the
UHV deposition, and the thickness of the liquid film, which may be
in the nanometer range, as illustrated in Fig. 4(b).
The presently proposed procedures are expected to lead to nanoparticles
significantly different from those previously assembled in vacuum or
in macroscopic liquids. The liquids that are chosen for the present
investigation, namely Xe, SF6, C2F6,
etc., are chemically inert, so they are not expected to react with
the deposited materials. Since these molecules are not polar, one
expects few chemical processes in the liquid, i.e. there is no
chemical reaction and no solvation of mobile ions.36
Therefore, the kinetic processes responsible for the self-assembly
are essentially physical in nature. This is the main reason that
this technique may be expected to be applicable to a wide range of
materials. Depending on the interactions and energies at the
liquid/deposit, deposit/deposit, deposit/substrate, liquid/vacuum,
and liquid/substrate interfaces, there are many processes that
could occur in the liquid phase. They could lead to clustering,
porosity, self-organization, uniform deposition, etc., or even
some novel morphologies.
ii.c Schottky Barrier Height Modification
The formation of Schottky barrier height at metal-semiconductor interfaces
and the related problem of heterojunction band-offset formation have been
subjects of intensive research and debate for several decades.37, 38
The lack of a strong dependence of the barrier height on the metal work
function, a phenomenon known as "Fermi level pinning", has prompted the
attribution of the barrier height/band-offset formation to bulk effects
in many theoretical models. However, experimental results obtained on
carefully controlled interfaces, such as epitaxial metal-semiconductor
interfaces39, 40 and precisely fabricated oxide-Si41
or heterojunctions,42 revealed a sharp dependency of the barrier
height/band-offset on the interface structure. Furthermore, it has been
demonstrated that polycrystalline metal-semiconductor interfaces, which
supposedly have "pinned" Schottky barrier height, actually display
significant degrees of barrier height inhomogeneity.43
Some of these apparent inconsistencies were recently removed by a
theoretical work,44, 45 which showed that interface bonds
alone could lead to apparently pinned Fermi level positions. A rough
estimate of the interface bond polarization led to a very satisfactory
explanation of the experimentally observed "strength" of Fermi level
pinning on different semiconductors.45 While a consensus has not yet been
reached as to the complete mechanism of Schottky barrier height and
heterojunction band-offset, the important role played by interface
structure seems firmly established.
There are many applications in which the ability to tune the barrier
height/band-offset is strongly desirable. For example, the contact
resistance to a semiconductor can be dramatically improved with a
reduction in its Schottky barrier height. The ohmic contact issue
is particularly pertinent for wide bandgap semiconductors such as p-type
GaN,46, 47 but it is also important even for common Si ULSI devices.
One notices that the projected specifications for contact resistivity
on SIA's Roadmap cannot be met with any known technology even for the
next technology node (100nm, to be shipped in year 2005).48
Another example where the ability to tune the Schottky barrier height
will be greatly appreciated is in the implementation of high permittivity
(high-K) gate dielectrics and metal gates in next-generation ULSI devices.
Beside good electrical conductivity, metal gates help to keep the crucial
effective oxide thickness (EOT) small by avoiding reaction with the high-k
dielectric and thereby obviating the need for a (lower-k) buffer
layer.48 The latter is a serious problem with the polysilicon
gates presently in use. One approach to the development of metal gates is
to choose a metal with a work function which matches roughly the mid-gap
point of the semiconductor.49 However, to be able to maintain the threshold
gate voltage for the field effect transistor at a convenient voltage,
especially at scaled-back power supply voltages, it is desirable to have
separate Fermi level positions for the gates on n-type and p-type channels.
For this purpose, one needs to control the Schottky barrier height between
the metal gate and the high-K dielectric. Many issues still remain in the
implementation of these metal-gate schemes.
Previously, there have been different approaches to modify the SBH.
The most successful scheme has been to insert a very thin layer of material
between the metal and the semiconductor. For example, layers of
insulators,50, 51 semiconductors,52-54 molecular
dipoles,55-57 and chemical passivation,58, 59
formed on the semiconductor surface, have been shown to modify the barrier
height of the Schottky contact. The manner by which the SBH is affected
by the interlayer is rather unpredictable and system-specific. The creation
of two new interfaces, one between the metal and the interlayer and one
between the interlayer and the semiconductor, and the likelihood of charge
transfer across the interlayer make the experimental "interlayer" situation
rather difficult to analyze. It has not been possible to make meaningful
comparison with theoretical models nor deduce widely applicable information
from these studies. In contrast, the presently proposed approach to modify
the SBH proceeds along a predictable path. Since the Fermi level pinning
effect has been linked to the bond polarization at MS interface, any
reduction in the density or the strength of the interface bonds should steer
the SBH toward the Schottky-Mott condition (away from the "pinned" condition).
It is expected that the bonding at metal-semiconductor interfaces formed by
the CULP method will be weaker as a result of some or all of the following
features: (1) The buffering of the kinetic energy of the incoming metal flux.
This energy has been speculated in the past as leading to defect formation and
metal-semiconductor interdiffusion. (2) The mediation/screening of the
solvate metal atoms by the liquid molecules. (3) The clustering of metal in
the liquid prior to adhering to the substrate. The heat released in metal
clustering has also been suspected in the past as facilitating interfacial
diffusion. (4) The much reduced deposition temperature in the present procedure.
Thus, the expected barrier height changes (n-type) are an increase for metals
with large work functions and a decrease for metals with small work functions,
respectively, from their strongly pinned SBH values. Therefore, a study of
the SBH formation at interfaces with weaker metal-semiconductor bonding, by
employing the proposed fabrication technique in the liquid phase, not only
has a great potential impact on ULSI and Ohmic contact technologies, but
also has an important bearing on our basic understanding of the SBH formation
ii.d Oxide Formation in Liquid Phase
There have been many different approaches to create pores in dielectric
films to lower the permittivity for ULSI interconnect applications.60-64
A reduced dielectric constant of nano-sized silicate particles has been
reported, with an increase in argon pressure during the deposition process.
The experimental arrangement used in the present CULP method may
accommodate two novel ways of growing oxide particles and porous materials.
Firstly, oxygen may be used as the liquid medium, albeit a reactive medium,
to assemble oxide structures. This represents an entirely different method
of oxide reaction: the oxidation of a reducing element by liquid or solid oxygen.
For example, Si may be deposited onto a pre-adsorbed oxygen film. The
triple point pressure of oxygen is small, ~ 0.15 kPa, making processing in
liquid oxygen feasible at low temperatures. Chemical reactions between Si and
O will take place even at cryogenic temperatures. Secondly, oxygen may be
supplied sporadically as an oxidizing agent, in combination with the co-deposition
of a mixed material/medium film, so that small oxide particles can be assembled
in the liquid phase. One example of the latter may be to interrupt the
co-deposition of Si and SF6 every so often to give the surface a dose of oxygen.
The avoidance of the simultaneous supply of oxygen during Si deposition stems
from a practical concern: to avoid the oxidation of e-beam evaporators.
The assembly or the arrangement of the dielectric film in the liquid
phase may present additional opportunity to manipulate the nano-structure
of the dielectric film.
iii. PROPOSED EXPERIMENTS
iii.a Materials Systems
There are many material/substrate/liquid combinations which may be used
to test the feasibility of CULP processing. In order to explore as wide
a range of materials as possible, while taking advantage of evaporation
sources and substrates (including patterned substrates) already available
in this laboratory, we propose to conduct the following experiments: (1)
Semiconductor particles: Si and Ge, assembled in Xe or SF6 on SiO2
substrate. (2) Metal particles: Co and Pt, assembled in Xe or SF6
on H-terminated Si surface. (3) Oxide particles: SiO2 and TiO2,
assembled in O2 on H-terminated Si. (4) Metal or semiconductor
particles on patterned SiO2/Si substrate. TEM, SEM and AFM
will be used to study the dependence of particle size distribution
on processing parameters. In addition to studying the process of
self-assembly in the liquid phase, we also propose to study (5) the
Schottky barrier height of self-assembled metal-semiconductor structures.
iii.b Equipment and Facilities
A multiple e-gun, molecular beam epitaxy (MBE) system with a base
pressure of less than 1x10-10 torr, designed for
silicon-based materials, will be used to perform the proposed
experiments. Until recently, this MBE system, which is equipped
with a double pass cylindrical mirror analyzer (CMA) for Auger electron
spectroscopy, a reverse-view low-energy electron diffraction
(LEED) system, and a sample load-lock chamber, has been used to
grow high-quality epitaxial metal silicide thin films and also
to study thin film reaction under UHV conditions65, 66 at the
Murray Hill location of Lucent Technologies Bell Laboratories.
The sample holder on this system has recently been modified to
have sample cooling capability through contact of the sample with
a liquid nitrogen reservoir on the sample holder, a picture of
which is shown in Fig. 5(a).
Having been donated to Brooklyn College, this chamber has
been moved on campus and is presently being tested. Along
with this chamber, a roomful of laboratory equipment including
TEM sample preparation tools, an optical interference microscope,
wet bench, liquid nitrogen equipment, semiconductor parameter
analyzer, 4-point probe, probe station, variable-temperature probes,
and various electronic equipment were also donated and moved on campus.
There is also a collection of Si wafers, oxide-patterned wafers,
evaporation sources, etc. in our possession. On campus, there are
SEM, STM, and AFM capabilities within the department which are available for our use.
Preliminary experiments with co-adsorbed layers of gas molecules and
evaporant will be carried out inside the Si MBE chamber in its present
geometry. Questions about the cleanliness of the adsorption process
can also be answered. In order to do liquid phase processing, however,
there are some necessary modifications which are being implemented.
The intended study of liquid film behavior by pressurizing the sample
at cryogenic temperature to above the triple point pressure necessitates
the exposure of the cryostat to atmospheric-level pressures. This
step essentially defeats the main advantage of sample-transfer capability
during the experiment. Experiments involving liquid phase processing
are therefore designed to take place in a small "pressure" chamber which
is gated from the main chamber and which can be opened for sample access.
The add-on "pressure" chamber has a total volume of only ~300c.c. and
contains a UHV-compatible cryostat with a control of the temperature of
the sample in the 30-700K range. An additional cryo-shield is also
added whose temperature can be varied from LN2 temperature to 150oC.
During operation, the cryo-shield can be chilled to help maintain
the sample temperature at cryogenic temperature during pressurization.
It also serves as a chiller of the gas that is being introduced.
The small "pressure" chamber can be isolated from the evaporation
chamber and the vacuum pumps via gate valves. A schematic of the
"pressure chamber" is shown in Fig. 5(b).
Since UHV equipment is capable of withstanding a pressure
differential of over 1 atm, one notes that the pressure chamber
can be pressurized to 2 atm without any modifications. Some care
will be exercised to install the gate valves in the right
There are several good physics graduate students who
are close to completing their required courses and are
interested in solid state physics and materials-related experiments.
One student will be chosen to start on these experiments immediately
(winter of 2002-03), with financial support temporarily coming from
the department and the PI's start-up money. Financial support of
this graduate student is requested in the present proposal. Technical
support for the general operation of the MBE system, TEM sample
preparation, and routine laboratory maintenance and repair, is provided
by a skilled technician (Mr. Frank Schrey), who is hired on a part-time
basis with the PI's start-up money. Support for this technician is NOT
requested in the present proposal. TEM analysis will be performed
by Dr. C. H. Chen of Lucent Technologies Bell Laboratories through
collaboration, at no cost to Brooklyn College. Dr. Chen is a world expert on the use of diffraction techniques and energy loss spectroscopy in TEM analyses of microstructures and phase transition. Available to Dr. Chen at Lucent Technologies are two high-resolution TEM/STEM machines with combined capabilities of micro-diffraction, energy dispersive x-ray (EDX), and electron energy loss (EELS). The small probe size of the STEM machine, ~ 0.15nm, is very suitable for detailed analysis of nanostructures.
iii.c Self-Assembly Experiments
A typical experiment in the proposed chamber may proceed as follows.
A sample is introduced and outgassed in the "pressure chamber" after
UHV condition is established. The cryo-shield is chilled and a certain amount
of gas could be allowed to pre-condense on the cryo-shield through dosing with
the leak valve. The sample is chilled to a predetermined temperature and a
buffer layer of molecules could be condensed at this point. The gate valve
to the evaporation chamber is opened and co-adsorption of the material from
the e-gun(s) and the gas molecules takes place. The average thickness of
the deposited film is varied from 5-50nm and the material-to-gas ratio of
the co-deposition can be adjusted in the 1-50% range. Following the
deposition, the valves to the main chamber and to the pumps are closed,
and cold gas(es) is introduced to the chamber. During the introduction
of the gas(es) the temperature of the sample is carefully monitored.
The rate of the introduction of gas(es) can be adjusted to allow the
sample to remain in the solid phase of the phase diagram of
Fig. 1. There are other options
such as cooling the chamber wall (with dry ice or liquid nitrogen)
during pressurization, which may be undertaken, if necessary, to keep
the temperature of the sample low. The gas introduced at this stage
is preferably a gas that does not condense at the prevailing substrate
temperature. Helium is a convenient gas to use as the "backfill gas".
This method also prevents the accidental buildup of excessive pressure
during chamber warm up. (A caveat here is that the single component
phase diagram of Fig. 1 needs to be modified to take into account
that the vapor pressure of that component is only part of the total
pressure.67) The partial pressure of the original gas
(the liquid medium) in the chamber can be increased by raising the
temperature of the cryo-shield to desorb previously adsorbed molecules.
Control of the condition and the duration of the liquid phase can be
exerted by heating only the sample without warming up its surroundings.
This "anneal" process can be likened to rapid thermal anneals (RTA)
commonly used in thin film processing. The control of the duration of
the liquid phase is desirable from the standpoint of avoiding prolonged
partial drying. The sample is "refrozen" after the liquid phase "anneal",
and will be warmed up only after the system is evacuated again.
Alternatively, the entire system may be allowed to warm up to dry
the liquid. The high chemical stability of SF6 and C2F6
makes these molecules difficult to break down. Therefore, they are
sensitive to the environment. The small quantity of these gases used
in the proposed experiments is deemed insignificant. However,
procedures can be implemented to pump the exhaust into sealed canisters.
It is worth mentioning that the semiconductor particles experience only
low or moderate temperatures during their assembly in the present scheme.
It is unknown what degrees of structural order and crystallinity can be
achieved for these particles under these processing conditions.
Post-annealing in vacuum, or in an inert ambient, can be used to
improve the crystallinity of the nanoparticles, if necessary.
iii.d Oxidation Issues for Semiconductor Nanostructures
Contamination is not of primary concern for the proposed
(primarily electronegative) metals and oxides films to be
deposited. However, for semiconductor thin films, one needs to
consider the cleanliness of the process carefully. Even though
the deposition in the proposed experiments is carried out in UHV,
oxidation and contamination during subsequent processing, at near
atmospheric pressures, should be kept to a minimum. As discussed
below, oxidation can be avoided by careful experimental procedures
and by the use of high purity gases. During the extensive bakeout
of the sample holder and the "pressure" chamber, external pumping
is used. Only after the UHV condition has been reached is the
sample holder cooled. This procedure minimizes the amount of
unintentionally condensed gases on the cold tip. The use of
high purity gases (6N) then keeps the adsorbed impurities in
the film on the sample, and elsewhere on the cold tip, to a minimum.
Note that the background pressure of the chamber, ~ 1x10-8
torr without active pumping, is completely negligible, compared with
the rated impurity levels of the gas sources at the triple point pressure.
Without preferential adsorption, the total amount of absorbed impurities
in a 100nm thick film is less than 10-3 monolayer (ML), which is a very
small quantity. However, further condensation of impurity molecules
from the backfill gas may occur, which provides impurities to the liquid.
Since the diffusion coefficient in the liquid phase is usually quite
high, ~ 10-5 cm2 sec-1, it may seem that
contamination is unavoidable in the present scheme. However, we point
out that the high diffusivity of the liquid phase also leads to rapid
assembly or clustering of the evaporant atoms. If the diffusivities
of various atomic species, evaporant and impurities, in the inert liquid
are similar, the initial diffusion can be simply viewed as occurring
within a closed system. Since the composition of the initial solid
film contains typically 1-50% of evaporant and only 10-6
impurities, one expects very efficient assembly of nanostructures in
a clean environment before the arrival of impurities from the gas phase.
Note that the above argument holds independently of the rate of the
temperature rise, as long as the diffusivities of various species are
similar. Therefore, it follows that, even in the case of semiconductor
nanostructures, oxidation will be limited to only the surface of the
semiconductor dots, which is essentially similar to exposing the
semiconductor to air after the nano-assembly. The above argument does not hold, however, if the adsorbed film is highly porous.
iii.e Schottky Barrier Metallization
A shadow mask will be used in the processing of diodes for SBH measurement. Self-assembled metal particles will only appear in diode areas, which range in size from ~ 100- 1000 mm. A thick metal cap layer will be deposited in the diode area for better contact. Standard current-voltage and capacitance-voltage techniques will be used to characterize the Schottky barrier height.
iv. Possible Outcome and Broader Impacts of the Proposed Activities
If successful, the proposed method can lead to the fabrication of a
wide range of materials of nanometer scale and the possibility of
self-organization and selective deposition on patterned surfaces.
Certain quantum dot memory and single-electron transistor devices
can be readily improved by the proposed method. The development of
the CULP method may lead to novel devices, which could have a broad
impact on the current technologies. In the unlikely event that the
present liquid processing technique does not lead to expected
nanostructures, one notices that the high resolution TEM work and
the processing of interfaces under variable cryogenic temperature
still forms a powerful combination to provide new information on
the relationship between the microstructure and the electronic
properties of semiconductor interfaces. Therefore, regardless of
the quality of the nano-structures produced by the present method,
the proposed research will reveal the kinetics involving solute
atoms and nanoparticles in different liquids, as well as creating
novel interfaces whose properties may have a direct bearing on
present theories of barrier height formation.
Because the proposed research uses very simple concepts,
it offers an ideal vehicle for the education of graduate
and undergraduate students. The strong possibility for
applications and the proposed collaboration with an industrial
research laboratory allows the students to gain experience in
industrial research and development. Due to the interdisciplinary
nature of these experiments, the students will be encouraged to
broaden their view by seeking guidance from experts in other areas.
These circumstances of the proposed activities will provide an
excellent training for the graduate students for careers in either
academia or industry. With the Bell Labs collaboration, the PI
expects to bring industrial scientists on campus to share their
experiences, as well as discuss their problems, which should lead
to more opportunities for collaborative investigations. Brooklyn
College traditionally attracts many minority and women students.
The PI is very actively involved in a special MARC program to introduce
and encourage minority students to pursue careers in research.
(See website: http://academic.brooklyn.cuny.edu/physics/tung/MARC.htm.)
He has opened up resources in his laboratory and volunteered his time
in mentoring and tutoring 3 minority undergraduate students (two women).
To broaden the participation of underrepresented groups at an early stage,
funds for a part-time undergraduate student as lab assistant is requested
in this proposal.
All results obtained in this project will be published in journals
and presented at scientific meetings. The students will be requested
to attend and make their presentations at scientific meetings.
These results will also be posted on the Internet for easy access
by the broad scientific community.
(See website http://academic.brooklyn.cuny.edu/physics/tung/CULP.)
The PI makes ample use of the Internet in his research and teaching.
In addition to preparing scientific publications in the
standard format, the PI has developed a tutorial on Schottky
barrier height on his website to facilitate his own teaching
and also to provide service to students and scientists in other
institutions. (See website http://academic.brooklyn.cuny.edu/physics/tung/Schottky.)
Results from this proposal, where appropriate, will also be described
in laymen's language on the web.