TITLE OF THE TERM PAPER

Document Sample
TITLE OF THE TERM PAPER Powered By Docstoc
					Ultrathin Oxide Growth and Characterization


                    By


               Justin Koepke
                 Peter Ong




   ECE 445, SENIOR DESIGN PROJECT

                FALL 2005




             TA: Chad Carlson



             December 6, 2005


               Project No. 21
                                             ABSTRACT

An Ultra-high Vacuum Scanning Tunneling Microscope (UHV-STM) system is modified to allow
growth of ultra-thin oxides in the load-lock chamber. Modifications to the system include the addition
of a contact assembly, a linear manipulator on which to mount the contact assembly, an electrical
feedthrough, a new custom flange on which to mount the linear manipulator and the electrical
feedthrough, and a newly designed fork assembly to hold the sample holder. Modifications to the
sample holder were necessary in order to allow oxide growth.




                                                   ii
                                                         TABLE OF CONTENTS

1.   INTRODUCTION ....................................................................................................................1
     1.1 Purpose ...............................................................................................................................1
     1.2 Specifications ......................................................................................................................1
     1.3 Subprojects .........................................................................................................................2
          1.3.1 Contact Assembly ....................................................................................................2
          1.3.2 Fork Assembly .........................................................................................................2
          1.3.3 Vacuum Hardware ...................................................................................................2
          1.3.4 Oxide Growth Recipe ..............................................................................................3

2.   DESIGN PROCEDURE ...........................................................................................................4
     2.1 Contact Assembly ...............................................................................................................4
     2.2 Fork Assembly ....................................................................................................................4
     2.3 Sample Holder ....................................................................................................................5
     2.4 Vacuum Hardware ..............................................................................................................5
     2.5 Oxide Growth Recipe .........................................................................................................5

3.   DESIGN DETAILS ..................................................................................................................6
     3.1 Contact Assembly ...............................................................................................................6
     3.2 Fork Assembly ....................................................................................................................6
     3.3 Sample Holder ....................................................................................................................7
     3.4 Vacuum Hardware ..............................................................................................................7
     3.5 Oxide Growth Recipe .........................................................................................................8

4.   DESIGN VERIFICATION .......................................................................................................9
     4.1 Testing ................................................................................................................................9
          4.1.1 Initial Assembly .......................................................................................................9
          4.1.2 Bench Testing ..........................................................................................................9
          4.1.3 Post Cleaning ...........................................................................................................9
          4.1.4 UHV Testing ............................................................................................................9
          4.1.5 Sample Heating ......................................................................................................10
          4.1.6 Oxide Growth.........................................................................................................10
          4.1.7 Recipe Optimization ..............................................................................................13
     4.2 Conclusion ........................................................................................................................13

5.   COST ......................................................................................................................................14
     5.1 Parts ..................................................................................................................................14
     5.2 Labor .................................................................................................................................14

6.   CONCLUSIONS ....................................................................................................................15

     APPENDIX A– Block Diagrams.......................................................................................... A.1
     APPENDIX B– Schematics ..................................................................................................A.7
     APPENDIX C– Testing Data ............................................................................................. A.17

     REFERENCES .......................................................................................................................16


                                                                            iii
                                          1. INTRODUCTION

We designed modifications to an existing scanning tunneling microscope (STM) system to allow the in
situ growth of ultrathin oxides on silicon. In order to grow thermal SiO2, a sample of Si and a source of
O2 are required. In order to grow a controlled amount of SiO2; the temperature of the sample, the
oxygen pressure, and the duration of the exposure must be controlled. One method of heating a sample
of Si is to apply a dc bias to the sample causing resistive heating. This project involved the design of the
mechanical parts necessary for controlled sample heating and controlled gas flow. The resulting oxide
was characterized by STM imaging and scanning tunneling spectroscopy (STS).

1.1 Purpose

The purpose of this project was to modify the loadlock chamber of a Scanning Tunneling Microscope
(STM) system to allow for in situ growth of SiO2. The thin SiO2 layers will allow experiments with
single-walled carbon nanotubes (SWNT) using the STM. These experiments seek to gain an atomistic
understanding of how underlying SiO2 affects the local electronic properties of nanotubes, which is
important for understanding SWNT device characteristics. Additionally, it may be possible to use the
STM tip to selectively remove the thin oxide revealing the underlying silicon. This could lead to
systematic studies of the differences between the effects of Si and SiO2 on the properties of carbon
nanotubes (CNTs).

The necessity of in situ growth of the thin oxide arises from the cleanliness required of samples in order
to avoid compromising the preparation chamber pressure or the hydrogen-passivation setup. The
baseline pressure in the preparation chamber is on the order of 10-10 Torr. The main purpose of the
preparation chamber is to prepare samples for imaging with the STM. Samples are degassed via DC
heating for several hours in order to rid the sample of volatile contaminants, flashed to 1250oC to
remove any native oxide present on the sample, and passivated with molecular hydrogen. The removal
of the native oxide, which is generally greater than 1nm, is necessary so that the tunneling currents will
be large enough to prevent the STM tip from crashing into the sample.

1.2 Specifications

Due to the nature of a STM, the insulating film between the tip and the conducting or semi-conducting
substrate must not be too thick. For SiO2, the film must be less than roughly 1.5 nm thick [1] in order to
image the surface properly. In order to have an oxide coating covering the entire surface, the thickness
of the SiO2 must be greater than one monolayer, or about 0.3 nm. In order to have a more robust scan
and the most isolation from the underlying Si sample, the desired growth thickness was approximately 1
nm. However, upon the realization that the methods of metrology available would not give an accurate
estimate of the actual thickness of the oxide film, the specification for the film thickness was changed to
the range of 0.3 – 1.2 nm. For these thicknesses, scanning is possible and a visible change present on
the surface.

The pressure in the loadlock chamber during oxide growth must be maintained below a certain level in
order to ensure the cleanliness of the chamber and the sample during oxide growth. The upper limit for
the growth recipe pressure was chosen as 10-4 Torr, whereas the base pressure of the chamber is roughly
10-7 Torr. Pressures above 10-5 Torr cause increased stress on the filament of the ion gauge used to
measure the pressure of the loadlock chamber. Maintaining pressures above this level for extended
periods of time will decrease the lifetime of the ion gauge filament, necessitating a time consuming and
                                                     1
expensive filament change. The growth rate of the oxide also increases with the oxygen pressure
causing a decrease in the level of control of the growth process.

In order to maintain a good level of control over the oxide growth process, the desired growth time was
specified to lie within the range of 15 to 90 min. However, upon a closer examination of oxide growth
recipes published in peer-reviewed journal articles, the specification was lowered significantly. The
optical pyrometer used to measure the sample temperature does not work below 600oC. Hence the
growth temperature should be at least 50oC higher to ensure the validity of the temperature reading. In
order to ensure that the pressure of oxygen in the chamber is high enough to approximate the total
pressure reading as the pressure of oxygen in the loadlock chamber, the growth pressure must be at least
one order of magnitude higher than the base pressure of the chamber. Therefore, the range of times was
reduced in order to meet the above requirements. The target time for the oxide growth was 30s.

1.3 Subprojects

In order to work more efficiently, the project was divided into several subprojects.

1.3.1 Contact Assembly

The contact assembly was used to apply the bias to the sample holder and the sample for the dc heating
of the sample. It was designed to be moveable so that the contacts were not permanently attached to the
loadlock linear transfer mechanism (LTM). Thus the sample holder is in physical contact with the
contact assembly only during the oxide growth process. This eased the assembly of the apparatus and
increased the modular nature of the design.

1.3.2 Fork Assembly

An insulated apparatus to hold the sample holder on the loadlock LTM was necessary in order to
properly apply the bias to the sample without shorting the two contacts.

1.3.3 Sample Holder

An existing sample holder was modified with the addition of a track for the loadlock fork. The sample
holder consists of a macor spacer sandwiched between two copper pieces. The copper pieces contain
many tracks needed for transferring the sample holder from LTM to LTM between chambers of the
STM.

1.3.4 Vacuum Hardware

The hardware to allow the controlled flow of the oxygen had to be added to the system. A leakvalve
was necessary to properly control the introduction of oxygen into the loadlock chamber. A custom
flange was necessary in order to mount the components necessary to implement the contact assembly.
Additional parts were necessary in order to mount the newly required components to the chamber.




                                                    2
1.3.5 Growth Recipe

The growth recipe needed to be designed in order to ensure the cleanliness of the sample before and
after the oxide growth. Additionally, the actual growth recipe was designed in order to meet the
specifications.




                                                   3
                                     2. DESIGN PROCEDURE

All custom parts not ordered from a vendor were designed using VectorWorks CAD Software. Craig
Zeilenga, from the ECE Machine Shop, custom machined all parts to the specifications of the designs.
After testing and prior to final assembly of the system, we thoroughly cleaned all parts to meet UHV
cleanliness requirements. The stainless steel and copper pieces were electropolished using a
commercially available solution. The stainless steel screws underwent a brief acid etch in HCl and
H3PO4. The macor pieces were dipped in HCl:HNO3 and subsequently baked in a vacuum oven to
remove any adsorbates that may outgas when under vacuum. Additionally, oxygen-free high-
conductivity copper (OFHC Cu) was used for all custom copper parts.

2.1 Contact Assembly

Several design options were weighed when considering how to apply the bias to the sample holder and
thus the sample. One alternative considered was to permanently attach wires from electrical
feedthroughs to the individual forks of the fork assembly. This alternative could have been very
problematic. If one of the wires would catch on the gate valve between the loadlock and the preparation
chamber, then the preparation chamber would have to be vented in order to recover the wire and the
contact system would have to be reassembled. Additionally, it would have been extremely difficult to
attach the wires to the fork assembly once mounted on the loadlock LTM.

Alternatively, commutator rings were considered as a possible method of contact. In this arrangement,
metallic rings separated from the LTM and each other by macor insulating material would have been
attached to the shaft of the loadlock LTM. A fork with electrically isolated prongs would be attached to
the end of a linear manipulator attached to the modified port. The electrical feedthroughs would then be
attached to the forks, which would contact the commutator rings during the oxide growth step. The
commutator rings would have wires to connect them to the LTM fork assembly used to hold the sample
holder. As evidenced by the description, this design would have been very tedious to implement.
Additionally, the location of the rings would have required that the sample not be centered in the top
viewport causing extreme difficulty when attempting to monitor the temperature of the sample.

Given the problems of the above alternatives, the design implemented was chosen as the most
straightforward and robust of the design options for applying a bias to the sample. The contact assembly
is attached to the linear manipulator (LM) that travels perpendicular to the travel of the loadlock LTM.
This design provides electrical contact to the sample holder using the least amount of wiring possible.
The contact assembly consists of one macor spacer sandwiched between two copper blocks designed to
complement the sample holder (Schematics B.6, B.7, and B.10). The electrical feedthrough is connected
to one of the blocks by a wire, while the other block is connected to the chamber which is used as
ground. The contact assembly is moved into place for the oxide growth step by rotating the LM until the
assembly is near the center of the loadlock. The sample holder is then pushed into physical contact with
the contact assembly using the loadlock LTM (Photograph A.6).

2.2 Fork Assembly

The LTM fork assembly mounts on the end of the loadlock LTM and has a fork at the end to hold the
sample holder. In addition to holding the sample holder, the fork assembly is used to press the sample
holder into physical contact with the contact assembly. The prongs of the fork were designed to be
isolated from one another. This prevents the fork from shorting the two sides of the sample holder. A
                                                   4
loadlock LTM fork is already in use on two of the other STM systems in the lab. The main difference
between the new fork and the old design is electrical isolation: the other forks in use consist of one solid
piece of stainless steel or invar. The new design consists of two pieces of stainless steel sandwiched
between two macor blocks attached to stainless steel plates.

The standard dimensions for the Lyding fork were used (Schematic B.1). The spacing between the
prongs of the fork was determined by the measurement of spacing of the prongs on one of the sample
transfer forks in the lab. The dimensions of the prongs of the fork were measured using digital calipers.

2.3 Sample Holder

Once the fork assembly and contact assembly designs were chosen, the necessary modifications to the
sample holder were considered. In order to accurately measure the temperature of the sample using the
pyrometer through the viewport above the sample, the sample must face the ceiling. The set of rails that
were present on the sample holder would not allow the sample holder to sit on the fork assembly in the
necessary orientation. A set of rails perpendicular to the ones already present in the sample holder were
added to allow the sample holder to rest on the fork assembly in the desired fashion. Unlike the other
set of rails in the sample holder, this set of rails does not go all the way through the copper pieces. Thus,
the fork assembly can push the sample holder into contact with the contact assembly. If the rails did go
all the way through the copper pieces, then the sample holder would slide along the fork resulting in
poor contact if any at all with the contact assembly.

2.4 Vacuum Hardware

Once the components requiring custom machining were designed, the necessary vacuum hardware was
considered. The required control over the flow of oxygen into the loadlock chamber necessitated a
leakvalve. The contact assembly required a linear manipulator in order to be moveable. Electrical
contact with the contact assembly required an electrical feedthrough. These components required an
adapter flange in order to mount properly to the loadlock chamber. There were no stock adapter flanges
available. A custom adapter flange was specified and ordered from A & N Corporation. They
generated a CAD file for review prior to finalizing the order. The linear manipulator was customized
with two 2-56 tapped holes to mate with a copper adapter that attaches to the contact assembly. The
design of the copper adapter increases the robustness of the connection between the contact assembly
and the linear manipulator. Mounting the leak valve to the loadlock chamber required a reducing tee to
facilitate the retention of the viewport, necessary for the measurement of the sample temperature using
the pyrometer.

2.5 Oxide Growth Recipe

Unfortunately, the Grove-Deal Model for oxide growth does not apply to the thicknesses of oxide with
which we were concerned. Peer-reviewed journal articles provided parameters for the growth of a
known oxide thickness. The parameters obtained from research provided a good starting point for the
development of the oxide growth recipe.




                                                     5
                                         3. DESIGN DETAILS

3.1 Contact Assembly

The contact assembly is a robust mechanism by which to apply a bias to the sample holder and the
sample. The contact assembly consists of two copper blocks (Schematics B.4 and B.5), between which
a macor spacing block (Schematic B.6) is placed to insulate to two copper contacts. The copper blocks
have a beveled edge to complement the bevel present on the standard Lyding sample holder (Schematic
B.7). By separating the two copper blocks with a macor spacer block, the two blocks are electrically
isolated so that any bias applied will be applied to the sample holder and sample when in contact with
the contact assembly. A copper braid covered with fiberglass insulation connects the electrical
feedthrough to the outer copper contact (Schematic B.7) and is attached by a 2-56 screw and washers
(Photograph A.3).

The macor block has six 2-56 tapped holes. Anchoring the screws that hold the contact assembly
together in the macor was the most straightforward manner to maintain electrical isolation while
assembled. The copper blocks each have three clear counterbored 2-56 clear holes; the screws are
aligned in a complementary pattern so that no screw shares a threaded hole in the macor block. The
screws do not touch the copper block opposite from the one in which they are inserted. Additionally, the
inner copper contact has two 2-56 tapped holes perpendicular to those used to hold the contact assembly
together (Schematic B.6). Two screws attach the contact assembly to the copper adapter piece which
mounts to the linear manipulator.

3.2 Fork Assembly

The fork assembly is designed to be a robust mechanism for securing the sample holder and pressing it
into contact with the contact assembly. The fork consists of two individual stainless steel prongs
(Schematic B.1) that are sandwiched by two macor blocks. The tips of the prongs are beveled to ease
sample holder mounting. We separate the prongs with macor to electrically isolate them and prevent the
fork from shorting any bias applied to the sample holder for heating. The lower macor block (Schematic
B.2) contains precisely machined indentions that hold the two prongs, while the upper macor block
(Schematic B.3) is completely flat and rests on the level surface formed by the prongs and lower macor
block. We only place indentions in the lower block to improve stability and lessen machining
complexity. The two macor blocks also contain size 8-32 clear holes for a bolt.

The two macor blocks are enclosed by a lower stainless steel plate (Schematic B.4) and an upper
stainless steel plate (Schematic B.5) and are held together by a nut and bolt. The upper stainless steel
plate also includes an extension with a ¼” clear hole that allows it to attach to the loadlock LTM. The
end of the LTM shaft is a threaded hole, thus the fork assembly is secured to the LTM by using a ¼”
bolt.

During testing, we discovered a problem where the sample holder tended to fall off the fork assembly
due to a mismatch between the sample holder width and the fork spacing. As a short-term fix, we adding
molybdenum foil spacers to the sample holder to widen it and also bent the fork slightly to ensure a
secure fit. We hypothesize that the beveled tips of the fork cause the prongs to gradually spread apart
under the physical stress that occurs during contact between the sample holder and the contact assembly.
In the future, a more permanent solution will be implemented such as removal of the tip bevel or
modifications to the way the sample holder sits on the fork.
                                                     6
3.3 Sample Holder

The sample holder provides an easy method by which to transfer samples into the system and
manipulate once in the system. We had one of the older style sample holders modified for use with the
oxide growth apparatus. This older style of sample uses two Cu-Be sidepieces of shorter length with a
correspondingly shorter macor spacer. Unfortunately, there are no VectorWorks CAD files available for
this sample holder. However, the critical dimensions are the same. The bevel has the same dimensions,
which is key for proper mating to the contact assembly, and the rail dimensions in the Cu-Be sidepieces
are the same as those in the new style sample holder (Schematic B.7).

The new rails added to the sample are perpendicular to those initially present (Photograph A.4).
Additionally, the rails do not extend all the way through the sample holder sidepieces. As mentioned in
the design procedure section, the rails do not extend all the way through the sample holder sidepieces so
that there is some material to allow the LTM fork assembly to push the sample holder into physical
contact with the contact assembly. The dimensions of the rails added to the sample holder are: 0.070”
wide and 0.070” deep. The new rails are centered along the length of the sample holder.

3.4 Vacuum Hardware

Several items of vacuum hardware were necessary in order to implement the desired functionality of the
system. The leakvalve added to the system was taken from a system that was no longer in use. Use of
this leakvalve required a 4.5” to 2.75” reducing tee in order to retain the viewport necessary for
measurement of the sample temperature. The reducing tee is a stock part that was ordered from Nor-Cal
Products, Inc.

The linear manipulator (LM) precisely moves the contact in one dimension to allow the proper
positioning of the contact assembly for contact with the sample holder. Rotating the knob of the LM
moves the shaft of the LM in the lateral direction. The LM was ordered from Nor-Cal Products, Inc,
who per request added two custom 2-56 tapped holes to the stock linear manipulator, part RLM-133-2.
The center-to-center spacing between the holes was custom specified to 0.223”. The distance from the
center of the first hole to the end of the LM was specified to 0.1970”.

An adapter piece (Schematic B.9) was designed to mate the contact assembly to the linear manipulator.
The adapter piece is a rectangular prism made of copper. The adapter has a 0.25” clear hole designed to
slide over the end of the shaft of the LM. Two 2-56 clear holes pass through the 0.25” hole
orthogonally. These clear holes have the same center-to-center spacing as those custom holes in the
LM. Additionally, there are two 2-56 clear channels. Thus, the lateral location of the contact assembly
along the adapter can be adjusted to obtain the ideal extension length from the end of the linear
manipulator. This allows the precise positioning of the contact assembly to obtain the ideal range of
travel during rotation of the LM. The channel allows for 0.4” of adjustment to the lateral location of the
contact assembly.

In order to mount the LM and the electrical feedthrough to the chamber, a custom flange was necessary.
The flange was ordered from the A&N Corporation, specifying that three 1.33” Con-Flat (CF) tabulated
flanges should be mounted to a 4.5” CF flange (Schematic B.11). We requested that mini-flange 3 be
rotatable. The requirement that flange 3 be rotatable would allow the proper orientation so that the holes
in the LM would be vertical upon mounting on the custom flange.
                                                    7
Upon receiving the custom flange, we realized that A&N did not make mini-flange 3 rotatable as
specified in the order. When mounting the LM on the custom flange, the holes in the LM were not
vertical. Returning the part to the manufacturer was not an option, because the part arrived
approximately two weeks after the estimated date of arrival. However, rotating the custom flange by 90o
in the counterclockwise direction resulted in the proper vertical alignment of the holes in the LM when
the LM was attached to mini-flange 1 (Schematic B.12).

After solving the above problem, the LM was attached to the custom flange and the contact assembly
and adapter were attached to the LM. When attempting to mount the custom flange with associated
parts to the loadlock chamber, we noticed that there was a clearance issue. Specifically, the LM shaft
was too low; so that when the adapter block was inserted over the shaft of the LTM, the adapter block
would not clear the inside diameter of the chamber. After considering what solutions were possible, we
decided to mount the adapter block on top of the shaft of the LM instead of over the shaft of the LM.
This solved the clearance problem, but it also gave rise to a new problem. With the adapter block
mounted on top of the LM shaft, the contact assembly was elevated by 0.25” above the location where it
would have been in the loadlock chamber. As a result, the sample holder contacted the adapter block
before contacting the contact attached to the adapter. Hence, electrical contact would not have been
possible. The solution was to move the contact assembly forward on the adapter so that it was only
attached to the adapter by one screw (Photograph A.3). Contact between the contact assembly and the
sample holder was now possible.

3.5 Oxide Growth Recipe

We referenced a paper by N. Miyata, et al. [2] when designing our recipe. They grew 0.3 nm of oxide,
or one monolayer, under 2x10-6 Torr O2 pressure at 725oC for 5 min. The dose of O2 for this growth
recipe was 600 L (Langmuir), where 1 L = 10-6 Torr * s. Since oxide growth depends on pressure in a
linear fashion, we doubled the O2 dose to 1200 L targeting 0.6 nm of oxide growth, or two monolayers.

The base pressure of the loadlock chamber determined the growth pressure of the oxide recipe. We
decided to use a growth pressure roughly two times the base pressure. The base pressure of the loadlock
chamber is approximately 2x10-7 Torr; the growth pressure chosen was 5x10-5 Torr. Having chosen the
growth pressure, the growth time necessary to obtain a 1200 L dose was approximately 24 s. We chose
to use the same temperature as that used in the paper so that one variable would be eliminated. Thus, the
oxide growth recipe parameters were 5x10-5 Torr O2 pressure at 725oC for 24 s. Additionally, we chose
to keep the turbo pump open to the loadlock chamber during the oxide growth process to maintain a
steady flow of oxygen and to pump out any impurities that might be present.




                                                   8
                                     4. DESIGN VERIFICATION

We will now outline the incremental tests we performed to verify operation of the system. The tests
were performed in a progressive manner during assembly to ensure that no problems would propagate.
Solutions to any problems discovered are mentioned briefly and are explained earlier in this report in the
design section.

4.1.1 Initial Assembly

Before any thorough cleaning of parts, the oxide growth components were assembled outside of UHV to
reveal any mistakes in mechanical specifications or machining.

The parts machined by Craig Zeilenga in the ECE machine shop were machined to specifications except
for the bore in the contact assembly where the linear manipulator shaft is supposed to fit. The bore is
tapered at the end preventing the LM shaft from fitting to the distance we expected, thus preventing the
two 2-56 taps from aligning properly. This problem was fixed by mounting the contact assembly on top
of the LM shaft and attaching it by screws long enough to pass through the entire contact assembly plus
LM shaft.

Another problem discovered during assembly testing was a result of manufacturing errors on the custom
three ported flange we had commissioned from A&N Corp. The port we had designed to fit the linear
manipulator was not manufactured rotatable and caused the linear manipulator plus contact assembly to
misalign. This was fixed by changing the orientation in which we attach the custom flange to the
loadlock.

4.1.2 Bench Testing (electrical)

After assembling the pieces of the system, resistance tests were performed to ensure that no unexpected
shorts existed and that contacts were isolated properly. The measurements were taken using a Fluke 79
multimeter. The sample holder measured an open load between the two side pieces and less than 2Ω
along a single side piece, which indicates that the macor washers in place to isolate the two sides were
not crushed during assembly. Similarly, the resistances between the two prongs of the fork and between
the two contacts of the contact assembly were also measured as an open load on the multimeter. At this
point, a piece of tantalum foil was attached to the sample holder to short the two sides for further
electrical testing. The measured resistance from the external feedthrough to the contact assembly plus
the resistance across the sample holder was ≈ 5Ω. The actual contact of the sample holder to with the
contact assembly was measured later, after assembly in the loadlock, but this provided an estimate of the
system resistance should be. All the contacts that we expected to be isolated measured an open load and
all other paths in the assembly measured on the order of ohms.

4.1.3 Post Cleaning

After all of the pieces were cleaned using various cleaning procedures, they were assembled once again
and resistance tests were repeated to ensure that nothing was altered during cleaning. Aside from some
human error in assembling the pieces again (breaking macor washers), the tests were passed.

4.1.4 UHV Testing

                                                    9
Once all components of the system were assembled and attached to the loadlock, electrical tests were
performed once again. This time, we were also able to test the full resistance of the path used to apply a
bias to the sample. The sample holder was pressed into the contact assembly using the LTM and the
measured resistance was ≈ 5Ω. This indicates that the sample holder is able to make good
physical/electrical contact with the contact assembly and that there is little resistance in the actual
system.

During UHV testing, we noticed a problem with the mechanical stability of the sample holder when
mounted on the fork. We discovered that the fork is slightly wider than the sample holder width, which
caused it to slide laterally and fall off of the fork when the LTM is rotated. This was fixed temporarily
by bending the fork prongs to lessen their distance apart.

While pumping down the loadlock again, we also monitored the ion gauge to ensure that there were no
leaks introduced into the system and that the base pressure still reached something on the order of 10-7
Torr.

The alignment of our new fork with respect to the main LTM fork was also tested by transferring a
sample holder between the forks after the loadlock reached an appropriate pressure.

The system passed all of the tests in this stage, which was rather important because any leaks or errors
with the fork alignment would have taken much time to fix.

4.1.5 Sample Heating

Sample heating operation was then tested by mounting a Si (100) substrate on the sample holder and
connecting an external power supply to the copper feedthrough and chamber ground. The power supply
was set to current limit mode and the current was dialed in to reach a desired temperature. The
temperature of the substrate was monitored using a pyrometer looking through the viewport on top of
the loadlock. The current vs. temperature data collected is included in Appendix C and indicates that the
system can heat a sample above 800oC with ≈ 2.5A, which is in a reasonable range.

4.1.6 Oxide Growth

The final test of our system involved actual growth and characterization of oxide. Our original
characterization method that involved an HF dip and AFM imaging was discarded for an STM
characterization method. We had realized, from Prof. Lyding’s suggestions, that performing an HF dip
on an oxidized surface would not produce the abrupt step that is needed to measure thickness in the
AFM. We also realized that exposing the oxidized surface to ambient conditions would cause the growth
of native oxide, thus limiting the accuracy of our thickness measurements.

The first Si sample we planned on testing was contaminated during the degas procedure due to the
sample holder heat shield contacting the macor insulator, which caused the macor to evaporate onto the
sample.

A second sample (p-Si (100), ρ = 5-20 Ω-cm, Boron doped) from Montco Silicon Technologies was
prepared for oxidation with the assistance of Peter Albrecht. The sample was first degassed in the prep
chamber overnight at ≈ 800oC. It was then flashed three times (above 1200oC for ≈ 30 seconds) to
further remove any contaminants on the surface. After flashing the sample, we hydrogen-passivated the
                                                    10
surface to ensure that it would not become contaminated before oxidation. H-passivation prevents the
surface from reacting with any contaminants in the system by bonding a single hydrogen atom to every
pair of dangling silicon bonds on the (100) surface. The sample was then transferred to the STM
chamber where it was imaged to ensure tip functionality and cleanliness of the surface. As evidenced in
Figure 4.1 and 4.2, the surface is clean and the tip is able to image the dimer (Si and H) rows of the H-
passivated Si surface.




Fig. 4.1: -2V bias, 50pA setpoint, 294x294 Å         Fig. 4.2: -2V bias, 70pA setpoint, 791x791 Å

After verifying functionality of the STM and condition of the surface, we transferred the sample back to
the loadlock to perform oxide growth. The recipe we designed for our initial growth attempt was
adapted from K. Fujita et. al [2] and specified a target O2 pressure of 5x10-5 Torr and temperature of
725oC for a time of 24 seconds. The noted O2 pressure is with the loadlock/turbo valve open, meaning
the oxygen is continually pumped out as new oxygen is leaked in. The actual growth parameters differed
slightly at 730oC (measured by the pyrometer), 5.7x10-5 Torr, and 30 seconds. A current of 1.51 A with
3.67V of bias was supplied by the power supply to reach this temperature. It is also worth mentioning
that the leak valve was opened a full 4 rotations plus an additional 30o to reach the targeted pressure.
After performing the growth procedure, the sample was then transferred back to the prep chamber for a
mild degas at 300oC to remove any contaminants that may exist on the surface from exposure to the
relatively higher loadlock pressure. We then transferred the sample to the STM chamber and obtained
the images in Figure 4.3 and 4.4.




                                                   11
Fig. 4.3: -3V bias, 7pA setpoint, 598x598 Å          Fig. 4.4: -3V bias, 7pA setpoint, 294x294 Å

The first indicator of successful oxide growth is the fact that we can still image the sample. The ability
to image means that the oxide was not grown too thick (above 1 nm) because the STM is still able to
establish a tunneling current, 7pA in this case, without crashing into the sample. Imaging results also
reveal the absence of dimer rows previously seen before oxide growth, meaning that the hydrogen bonds
that form the dimers were broken and oxide was formed in place of them. The amorphous appearance of
the surface is also a good indicator that we are imaging some form of oxide. It is worth noting that not
all of the features in the images are due to surface topography, but are also a cause of electrical
differences in the oxide.

To further verify the existence of oxide, we performed scanning tunneling spectroscopy (STS)
measurements. STS involves defining specific points on the STM image where the microscope will
pause during scanning and perform a voltage sweep while collecting current measurements. This results
in a collection of I-V graphs for all the user-defined points on the surface. The collected data is
presented in Figure 4.5, 4.6, and 4.7 below.


                                                Figure 4.5 displays STS data taken from an H-
                                                passivated p-Si sample from the same wafer as
                                                that of the oxidized sample. The data indicates a
                                                bandgap of approximately 1.1eV, which is
                                                expected for an H-passivated Si substrate. One
                                                can also notice the p-type shift of the bandgap
                                                toward the valence band edge, characteristic of
                                                the high Boron concentration.




Fig. 4.5: STS of H-passivated p-Si sample


                                                   12
Fig. 4.6: STM image indicating STS points             Fig. 4.7: STS of oxidized p-Si sample

STS was performed on the oxidized Si surface by defining a 36 point square rotated 45o. At each of the
36 points, 4 STS spectra were taken and averaged to reduce the affect of noise. Though we only present
data for 3 points in the above figures, the majority of data taken exhibited close similarity. The collected
data indicates a bandgap of approximately 4.5eV and maintains the p-type bandgap shift. Since we
expect an increased bandgap measurement due to the voltage drop across the oxide layer, this data also
supports the existence of oxide.

4.1.7 Recipe Optimization

Due to the extended time spent correcting mechanical errors and characterizing the first growth attempt
we were not able to perform any recipe optimization as specified in our design review. The first attempt,
however, did produce excellent results and further growth experiments will be conducted in the future to
accurately characterize the affect of growth parameters on grown oxide.

4.2 Conclusion

In conclusion, the system passed all of the tests we designed to adequately verify successful growth of
oxide and met all performance specifications. Many early tests revealed problems in mechanical designs
that were fixed with short-term solutions. These solutions allow the system to function properly, but will
be replaced with long-term solutions in the near future.




                                                     13
                                                  5. COST

Following is a cost estimate of the completed project including parts and labor.

5.1 Parts

Table 5.1 is an itemization of the parts necessary to build the oxide growth system. Costs for parts
ordered from manufacturers are actual prices while the costs for parts found in the lab are estimated.
Parts from the ECE machine shop are estimated based on time spent manufacturing.

       Table 5.1 Itemization of parts cost

        Part Description                         Part No.        Manufacturer                 Cost
        Custom Copper Blocks                                     ECE Machine Shop                120
        Custom Macor Blocks                                      ECE Machine Shop                120
        Custom Stainless Steel Fork                              ECE Machine Shop                 60
        Custom Stainless Steel Plates                            ECE Machine Shop                 60
        Custom Flange 4.5” CF w/ (3) 1.33”       450 x 133 (3)   A&N Corp.                       430
        Electrical Feedthroughs (Avail.)                         MDC                             170
        Leak Valve (Avail.)                      9515106         Varian                         1129
        Linear Manipulator 2” w/ custom holes    RLM-133-2       Nor-Cal Inc.                    635
        Miscellaneous Parts/Screws (Avail.)                                                       50
        Pyrometer (Avail.)                       DHS-52          Wahl                            500
        Reducing Tee 4.5” to 2.75” CF            3TR-250-150     Nor-cal Inc.                    235
        Total Parts Cost                                                                      $ 3509

5.2 Labor

The labor cost associated with the project is estimated by Equation 5.1 and assumes an ideal hourly rate
of $50. The time required for the project is an estimate based on time spent in lab and elsewhere doing
design.

               Labor Cost = Ideal Hourly Salary  (Hours Spent + Shop Hours)  2.5                       (5.1)

               Peter: ($50/hour) × 2.5 × 20 hours/week × 13 weeks = $32,500

               Justin: ($50/hour) × 2.5 × 20 hours/week × 13 weeks = $32,500

               Total Labor Cost:       $65,000

The total cost for the project is $68,509. The parts cost is within the usual cost range for projects in the
STM group, however, the labor cost is slightly above the usual rate for graduate research assistants.
Additionally, many of the parts ordered for the system can be used elsewhere in the lab when oxide
growth functions are not needed.




                                                      14
                                          6. CONCLUSIONS

In summary, we were able to accomplish all the goals of the project. Concerning system design, we
successfully designed the mechanical components of the system with enough tolerances to allow quick
solutions to the many problems we encountered during initial assembly. The system passed all
mechanical and electrical tests after slight modifications to the design and did not disrupt the high
vacuum environment of the loadlock. There are still some long-term fixes that need to be implemented
to ensure reliability of the system, but are not critical at this time.

We are also able to grow ultra-thin oxides using the adapted recipe from K. Fujita et al. [2]. The
presence of oxide is verified by the ability to image the surface using the STM, the actual STM image
with absence of dimer rows, and the increased bandgap indicated by STS data. There is still some
uncertainty concerning the detailed characteristics of the oxide like thickness, density, and interface
roughness but the data collected provide very convincing evidence of an oxide layer that is less than 1
nm in thickness.

The next step at the close of this project is to implement some long-term solutions to the previously
mentioned problems. First, we plan to bevel parts of the contact assembly to allow contact of the sample
holder in a more robust configuration. We also plan to modify the fork to prevent it from widening
during normal operation and insecurely holding the sample holder. Aside from mechanical
modifications, we also want to perform more oxide growth iterations to better characterize the recipe
and collect enough data to correlate the three growth parameters (temperature, pressure, and time) to
oxide features like thickness, uniformity, and quality.

Beyond some improvements to the oxide growth system, there are an endless number of experiments
that can be performed involving an ultra-thin oxide. The two experiments we are most eager to run at
this point include the characterization of carbon nanotubes on an oxidized Si surface and the fabrication
of back-gated carbon nanotubes based FET devices.




                                                   15
                                          REFERENCES

[1]   J. Yu, “Silicon/Silicon Oxide Interface Roughness Studied By Scanning Tunneling Microscopy,”
      Ph.D thesis, University of Illinois at Urbana-Champaign, Urbana, IL, 2003.
[2]   N. Miyata, H. Watanabe, and M. Ichikawa, “Nanometer-scale Si-selective epitaxial growth using
      an ultrathin SiO2 mask,” J. Vac. Sci. Technol. B, vol. 17, no. 3, pp. 978-982, May 1999.




                                                 16

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:2
posted:10/4/2012
language:Unknown
pages:19