The reaction kinetics and film morphology of molybenum films deposited by LPCVD on a silicone
surface
by Edward James Flanigan
A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in
Chemical Engineering
Montana State University
© Copyright by Edward James Flanigan (1987)
Abstract:
Molybdenum films were deposited by Low Pressure Chemical Vapor Deposition (LPCVD) on silicon
substrates by the hydrogen reduction of molybdenum hexafluoride. The reaction kinetics were studied
in order to determine a rate equation. Extensive scanning electron microscopy (SEM), electron
spectroscopy for chemical analysis (ESCA), auger electron spectroscopy (AES) and rutherford
back-scattering (RBS) studies were conducted to characterize and analyze the morphology of the
deposited molybdenum films.
The hydrogen reduction of molybdenum hexafluoride was determined to be one half order in hydrogen,
approximately zero order in molybdenum hexafluoride and had an activation energy of 76,000 J/mol at
temperatures from 250 to 3507deg;C, and total pressures from 0.9 to 10.0 torr. The preexponential
factor was determined to be 2.02 x 106 nm s-1.torr -0.5.
The main features of the films were the high oxygen (up to 28%) and impurity content. Although pores
could not be seen by the instrumental techniques provided, film impurities and surface roughness were
attributed to a porous deposit. Rough interface conditions were explained by the competing silicon
reduction reaction, and good quality deposits were seen to be produced at high temperatures and low
pressures.
Results of this experiment were compared with previous studies involving tungsten and molybdenum
LPCVD. One difference observed was the apparent lack of linear film growth with time. The other
difference was that a molybdenum film had a resistivity of three times less than any previously
reported resistivity for a LPCVD molybdenum film. THE REACTION KINETICS AND FILM MORPHOLOGY OF MOLYBDENUM
FILMS DEPOSITED BY LPCVD ON A SILICON SURFACE
by
Edward James Flanigan
A thesis submitted in partial fulfillment
of the requirements for the degree
of
Master of Science
in
Chemical Engineering
MONTANA STATE UNIVERSITY
Bozeman, Montana
July 1987
main lie-
APPROVAL
of a thesis submitted by
Edward James Flanigan
This thesis has been read by each member of the thesis
committee and has been found to be satisfactory regarding
content, English usage, format, citations, bibliographic
style, and consistency, and is ready for submission to the
College of Graduate Studies.
Chairperson,Graduate Committee
Approved for the Major Department
7 mi
,
Approved for the College of Graduate Studies
Graduate yDean
iii
STATEMENT OF PERMISSION TO USE
In presenting this thesis
requirements
for
a
in partial fulfillment of the
Master's
degree
at
Montana
State
University, I agree that the Library shall make it available
to borrowers under rules
from this thesis
are
of
the Library.
Brief quotations
allowable without special permission,
provided that accurate acknowledgment of source is made.
Permission for extensive
of this thesis may be
quotation from or reproduction
granted
his/her absence, by the
by my major professor, or in
Director
opinion of either, the proposed
scholarly purposes.
Any copying
this thesis for financial gain
of Libraries when, in the
use
of the material is for
or
use of the material in
shall not be allowed without
my written permission.
Signature,
Date_____
7
T
VJ>97?C
iv
TABLE OF CONTENTS
Page
TITLE PAGE ........................................
i
A P P R O V AL............ ..............................
ii
..............
iii
TABLE OF C O N T E N T S .............................. ..
iv
LIST OF T A B L E S .............. ...................
vi
STATEMENT OF PERMISSION TO USE . . .
LIST OF F I G U R E S ............ ..
ABSTRACT ..........
•
...................
ix
..............................
INTRODUCTION ........ ............................
BACKGROUND ..................
vii
•
....................
I
5
Integrated Circuit Technology . . ..............
Importance of Molybdenum as a Contact or Inter
5
connect Metal ........................
'
Molybdenum Deposition (Basic Reaction)..........
Hydrogen and Silicon Reduction of Molybdenum
5
9
Pentachloride .............. ..................
Hydrogen and Silicon Reduction of Tungsten
10
Hexafluoride ........................ ........
Hydrogen Reduction of Molybdenum Hexafluoride . .
13
Analyses by A E S ...................... .. . . . .
14
16
Analyses by S E M ................ ................
Analyses by E S C A ........ ......................
18
Analyses by R B S ............ ....................
22
20
V
RESEARCH OBJECTIVES
........................
24
EXPERIMENTAL EQUIPMENT . . . . . . . . . . . . . . .
Gas Flow Control System ........
25
. . . . . . . .
25
Reactor and H e a t e r ........................
Pressure Read-Out and Control System ..........
27
29
Pumping S y s t e m ............
Acid Gas Detection System .......................
After Burner and Alumina Trap S y s t e m ..........
29
30
31
Chemical H a z a r d s ..............
32
EXPERIMENTAL PROCEDURES
............
34
Sample Cleaning ................................
Deposition Procedure ..............
AES Analysis Procedure ........................
SEM Analysis Procedure
ESCA Analysis Procedure . . . . . .
•
Acid Dissolution Procedure ....................
34
34
35
38
38
39
RBS Analysis Procedure
39
. . . . .
..............
RESULTS AND DISCUSSION . . . . . . . . . . . . . . .
40
Thickness Determination of Deposited Samples
40
Reaction Kinetics ..............
. .
. . . . . . . .
40
Order of Reaction ...............................
43
Rate Equa t i o n ..................................
Characterization of Molybdenum Films ..........
46
48
SUMMARY AND CONCLUSIONS
RECOMMENDATIONS
..........................
70
. . . . . . . . . . . . . . . . . . .
72
REFERENCES CITED . . . . . .
APPENDICES . . . . .
......................
73
..............................
78
Appendix - Sample Calculations
................
79
LIST OF TABLES
Free Energy Changes for the HoF6 Reaction . .
Temperature, Time and Partial Pressure
Variations for the Kinetic Study of the
Hydrogen Reduction of Molybdenum Hexafluoride
Molybdenum Film Thickness Measurements by
Acid Dissolution (for the Kinetic Data) . . .
Relative Molybdenum Film Impurities from AES
Depth Profiles at Molybdenum's Highest Peak
Value . ......................... . . . . . .
vii
LIST OF FIGURES
Figure
Page
1.
Schematic View of a HOSFET Cross Section . . . .
6
2.
LPCVD Reactor System . . . . .
26
3.
Substrate Heater/Holder for the LPCVD Reactor
4.
Molybdenum Thickness as a Function of Time
at T = 300 °C and Pfcofc = 0.9 torr . . . . . . .
................
.
28
42
5.
Arrhenius Plot
6.
Plot for Determining Order of Reaction With
Respect to Hydrogen Partial Pressure ..........
45
Plot for Determining Order of Reaction With
Respect to Molybdenum Hexafluoride Partial
Pressure .......... ............................
47
AES Depth Profile of a Mo Film Deposited
at T = 300 °C and Pfcofc = 4 torr ...............
49
ESCA Spectra for a Mo Film Deposited at
T = 350 °C and Pfcofc = 5 torr . .................
52
SEM Surface Micrograph of a Mo Film Deposited
at T = 400 °C and Pfcofc = 0.9 torr . . . . . . .
55
SEM Surface Micrograph of a Mo Film Deposited
at T = 250 °C and Pfcofc = 2 . 5 torr . . . . . . .
56
SEM Surface Micrograph of a Mo Film Deposited
at T = 300 C and Pfcofc = 1.3 torr . . . . . . .
58
SEM Surface Micrograph of a Mo Film Deposited
at T = 250 °C and Pfcofc = 7.3 t o r r ............
59
SEM SurfaceriMicrograph of a Mo Film Deposited
at T = 350 C and Pfcofc = 5 torr . . . . . . . .
61
AES Elemental Point Scan of the Surface in
Figure 14(b); point I ............ ............
62
7.
8.
9.
10.
11.
12.
13.
14.
15.
.......... . .......... . •
44
viii
Figure
16.
Page
AES Point Elemental Scan of the Surface in
Figure 14(b); point 4 .........................
63
Rutherford Backscattering Scan of Mo Films . . .
65
18. x SEM Cross-Section of a Mo Film Deposited at
T = 350 °C and P^ot = 5 torr . . . . . . . . . .
67
17.
19.
SEM Cross-Sections of a Mo Film Deposited at
T = 400 C and P^0^. = 0.9 torr . . . . . . . . .
68
ix
ABSTRACT
Molybdenum films were deposited by Low Pressure Chemical
Vapor Deposition (LPCVD) on silicon substrates by the
hydrogen reduction of molybdenum hexafluoride. The reaction
kinetics were studied in order to determine a rate equation.
Extensive scanning electron
microscopy (SEM)„ electron
spectroscopy for chemical analysis (ESCA), auger electron
spectroscopy (AES) and rutherford back-scattering (RBS)
studies were conducted to characterize
and analyze the
morphology of the deposited molybdenum films.
The hydrogen reduction of molybdenum hexafluoride was
determined to be one half order in hydrogen, approximately
zero order in molybdenum hexafluoride and had an activation
energy of 76,000 J/mol at temperatures from 250 to 350°C,
and total pressures from 0.9 to 10.0 torr. The
preexponential factor was determined to. be 2.02 x 10 nra*
s-1'torr-0°3.
The main features of the films were the high oxygen (up
to 28%) and impurity content.
Although pores could not be
seen by
the
instrumental
techniques
provided, film
impurities and surface roughness were attributed to a porous
deposit. Rough interface conditions were explained by the
competing silicon reduction reaction, and good quality
deposits were seen to be produced at high temperatures and
low pressures.
Results of this experiment were compared with previous
studies involving tungsten and molybdenum LPCVD.
One
difference observed was the apparent lack of linear film
growth with time.
The
other difference was that a
molybdenum film had a resistivity of three times less than
any previously reported resistivity for a LPCVD molybdenum
film.
I
I
INTRODUCTION
In the semiconductor industry
known and important metal.
today, aluminum is a well
This metal is used for contacts
and interconnects in the microelectronics area of very large
scale integration (VLSI).
industry
is
growing
Competition in the semiconductor
at
a
rapid
pace,
competitive condition brings about the
with the capabilities of
though.
This
need for a new metal
making smaller, less expensive and
more productive components for integrated circuits.
Aluminum
is
metallization
the
of
most
silicon
widely
devices
temperature resistivity and
used
due
excellent
other deposited silicate glasses
to
material
its
low
for
room
adhesion to SiOg and
Ell. However, aluminum has
a relatively low melting point (660°C), and upon approaching
this temperature the transfer
increased,
causing
of momentum from electrons is
transport,
conductive material.
This
or
migration,
of
the
e^ectromigration is a potential
source of breakdown in aluminum interconnect lines [2,31.
The failures brought about
by using aluminum has opened
the door to new contact metals such as the refractory metals
of molybdenum, tungsten and
tantalum.
has shown potential applications
interconnect,
but
in
other
silicides have been formed
on
not
areas
Recently, molybdenum
only
as
as a contact or
well.
Molybdenum
refractory metals to protect
Schottky diodes have been shown to approximate the ideal the
2
metal
against
high
characteristic
temperature
properties
oxidation
of
Schottky
E4-6J.
Mo/Si
diodes
C'73.
Molybdenum has been used for diffusion or corrosion barriers
and, because of its
shape E83.
There
high
has
ductility, for parts with complex
also
been considerable interest in
molybdenum as a gate electrode for VLSI fabrication E9,103.
Molybdenum as a
contact
metal
has
shown promise in
VLSI because of its high melting temperature (2610 °C), high
corrosion
resistance,
coefficient that
is
low
close
resistivity
to
that
interconnect resistance and thus
of
and
expansion
silicon.
A low
a high operating speed can
also be obtained by using molybdenum E103.
Today,
modern
surface
provide the capability of
use in microelectronics.
to investigate
such
science
the
condition
deposited on a
affected
by
as
deposition.
In
parameters of the
the
reaction
pressure of the
uniformity and purity
interface
This
deposition
chemical
the
porosity of a deposited film,
the
surface.
the
techniques
These methods allow the researcher
things
of
other
studying metal films designed for
throughout film deposits, the
and
and
after
a
metal
is
film morphology is greatly
parameters
and
the
rate of
vapor
deposition (CVD), the main
are
the substrate temperature,
reactant
gas mixture and the ratio of
3
hydrogen to metal halide.
parameters has great
The separate variation of these
influence
on
the deposition rate, on
the amount of impurities, on the crystal structure and grain
size, and on the
number
of
micro-bubbles that may develop
under certain circumstances along
the
reaction
morphology
C113.
the grain boundary during
Improved
and the kinetics
more efficient metal
knowledge
of
contact
of
the
film
the reaction may provide a
and manufacturing process for
the semiconductor industry.
Research on the
low
pressure chemical vapor deposition
(LPCVD) of
molybdenum
molybdenum
pentachloride
£7,9,12-163.
Some
source of Mo
£17-203.
is
limited.
as
studies
the
have
Other
authors
formed
have
on
deposition
rates,
high
pressures in order to
have
considered
the
molybdenum
or diodes were
molybdenum £7,213 and Mo
n-type polysilicon lines £223.
the
£23,243
deposits for metallurgical
of
the carbonyl as the
barriers
of
studied
molybdenum deposition
source
used
Schottky
formed by selective deposition
layers have been
Several studies used
or
silicides
formed
by
have
studied molybdenum
applications
which require high
temperatures
cover
a
large
hydrogen
and
area.
reduction
atmospheric
A few authors
of
molybdenum
hexafluoride for integrated circuit technology £25,263.
In this study,
surface by
hydrogen
molybdenum
reduction
was
deposited
on a silicon
of molybdenum hexafluoride.
4
This was performed over a temperature range of 250-350°C and
a total pressure range of
this
research
morphology
and
deposition.
was
0.9-10.0
to
study
torr.
investigate
the
The objective of
the
kinetics
molybdenum
of
the
film
molybdenum
The amount of deposition was determined by acid
dissolution of the molybdenum
films.
This acid dissolution
method was employed in determining a reaction rate equation.
The
film
Microscopy
Electron
morphology
(SEM),
was
Auger
Spectroscopy
for
studied
Electron
Chemical
Rutherford Backscattering (RBS).
by
Scanning
Electron
Spectroscopy
Analysis
(AES),
(ESCA)
and
5
BACKGROUND
Integrated Circuit Technology
Electronics basically began
(IC) which was invented
early
primitive
electronic
by
forms,
devices
with the integrated circuit
Kilby
ICs
in
have
containing
1958
C23.
evolved
hundreds
of
From the
into
complex
thousands
of
individual components on a single chip of silicon.
The transistor is the most important component of an IC.
This is what tells current to
open or close.
One
such
flow
type
or not, or a circuit to
of transistor is the metal
oxide field effect transistor (HOSFET) pictured in Figure I.
This has an n-type
substrate.
The
positive charge
source
letters
and
n
carriers,
drain are contacted by
is applied to the gate
source and drain.
This
gate oxide which causes
and
p
refer
respectively.
metal
connected to a power supply.
drain implanted in a p-type
(e.g.
to negative and
The
source and
Al or Mo) contacts and
When a threshold voltage (V^)
electrode, current flows between the
voltage
the
to n-type, thus creating a
creates a field across the
adjacent p substrate to invert
conductive n channel between the
source and drain E23.
Importance of Molybdenum as a Contact or Interconnect Metal
Metallization, as
described
above,
has many important
METAL CONTACTS
GATE
ELECTRODE
DEPOSITED
DIELECTRIC
ON
Figure I
Schematic View of a MOSFET Cross Section
I
factors associated, with the
used.
Metallization
type of contact or interconnect
requires
mechanical properties, adhere
a
contact
firmly
to
have
good
during both formation
and subsequent processing, not cause excessive stress in the
underlying semiconductor and
A contact must also
be
low electric resistance.
compatible
used for the interconnection
to electromigration and
have
with the metal system
technology, not be susceptible
corrosion,
and, finally, be easily
patterned by a straight forward process C33.
Most silicon
MOS
and
bipolar
integrated circuits now
manufactured are metallized with
Al
Aluminum has
high
and
both silicon
and
resistance
aluminum.
to
Despite
problems
shallow.
silicon
contacts
silicon E U .
some
conductivity
in
Good
p—type
these
VLSI
step
excellent adhesion to
and
also
heavily
forms lowdoped n-type
advantages, aluminum also has
coverage
is
where
hard
junctions are
to
achieve with
At the present time, physical vapor deposition is
difficult
to
get
for
difficult to obtain a low
Thinner
depositing
the
prevent junction spiking and
switching.
one of its alloys.
It
applications
the only method available
very
dioxide.
or
proper
and
It is
composition to
electromigration.
It is also
resistance contact for high speed
deposits
densities and higher resistances.
hillock formation
alloy
aluminum.
etching
produce
higher
current
Particulate interference,
difficulty
are
also other
problems which are explained in the literature EI,23.
Molybdenum,
advantages
on
over
deposition.
the
aluminum
CVD.
With
achieved.
hand,
mainly
due
provides
to
the
several
method
of
CVD
low bulk resistivities.
I
surface adhesiveness and smaller
homogeneous films, better
grained (large grains
other
cause
electromigration) films can be
Also, CVD employs simple equipment and offers the
capability of coating a large
number of silicon wafers at a
time relatively inexpensively.
The most important advantage
of using
CVD
molybdenum,
however,
is
its
improved step
coverage and selectivity.
Molybdenum could be
an
excellent interconnecting metal
because it can be selectively deposited on silicon leaving a
silicon dioxide surface uncovered;
therefore, the number of
process steps can be reduced by eliminating lithography E23.
Lithography is the process
of transferring geometric shapes
on a mask to the surface of a silicon wafer.
Selective deposition
the process steps
in
of
molybdenum
VLSI
eliminates some of
fabrication.
The advantages
realized by eliminating these
process steps could result in
a higher
a
product
yield
Molybdenum is not without
different sets
produce
high
of
and
its
conditions,
resistivity
and
expansion different from that
substantial
cost savings.
disadvantages though.
molybdenum
porous
Under
was reported to
films,
a
thermal
of silicon and voids formed
at the Mo-Si interface due to an undesired reaction with the
underlying silicon.
9
Molybdenum Deposition (Basic Reaction)
Low pressure
chemical
vapor
»
of molybdenum
deposition
from MoF^ takes place by the following reaction:
*°(s)
M°F6(g) + 3H2(g )
Another reaction that takes
hydrogen
reduction
of
MoF6
+ SHF,,)
place
is
some time during the
the
silicon
reduction
reaction:
2Mo
2MoF6(g> + 3S1(s)
It is known that this
reaction must take place sometime
during the reaction process
Mo layer some
Si
is
+ 3S1F4(g)
(S)
because during formation of the
consumed
(about
twice the volume of
deposited metal) C253.
Free energy changes
both reactions
are
shown
the
that this reaction is more
intent
of
this
takes
research
reduction reaction.
Table I demonstrate that
thermodynamically
free energy change for
reduction reaction
in
possible
E273.
The
silicon reduction reaction show
favorable.
place
was
Although the silicon
during
to
Mo deposition, the
observe
the
hydrogen
10
Table I.
Free Energy Changes for the MoF6 Reaction
T = 300°C
I
AG0
I
(kcal/mol)
i
i
i
i
i
i
L Reduction
-65
I
Si Reduction
I
-218
I
|
Hydrogen and Silicon Reduction of Molybdenum Pentachloride
Recent studies
involving
molybdenum :stemmed
possible use in microelectronics.
compounds have been used
CVD molybdenum.
in
from its
A variety of molybdenum
studying the potential uses of
These include MoFfi, MoClfi and Mo(CO)fi.
The
literature on molybdenum carbonyl CMo(CO)fi], however, showed
this compound to be a
poor
the incorporation of too much
comtaminated
films
are
source of molybdenum because of
carbon in the films.
unacceptable
for
Carbonuse
in
microelectronic applications.
In
Mo
CVD
investigations,
molybdenum
pentachloride
(MoClg) is probably the most studied compound of molybdenum.
11
Hydrogen reduction of MoCl5 has been accomplished at several
temperatures
and
pressures
molybdenum's possible
uses
electronic components.
of 3-15 torr with
and a
in
range
order
the
Studies
variable
temperature
in
investigate
production
of various
performed at low pressures
MoCl5
of
to
and H2 partial pressures
700-1100
°C
showed
that by
I
adjusting the deposition
parameters,
a thin, dense coating
with good adhesion can be produced [15].
A similar study on Mo by the hydrogen reduction of MoCl5
incorporated
the
same
pressure ranges as
temperatures
performed
the
between
with
low
the
pressure
previous
500-800
deposited molybdenum as a
gate
Here the deposition
was
rate
surface reaction and was
and
partial
discussion but with lower
°C
potential
ranges
C93.
This
study
was
of
using
the
application
metal source in transitors.
thought
proportional
to be controlled by
to
the 3/2 power of
hydrogen partial pressure in the region of the surface.
The
films deposited in this temperature and pressure range had a
thickness uniformity for a batch
of 25 wafers within 5% and
the films were not oxidized.
Molybdenum films
deposited
pressure and a temperature near
investigation ES,14].
grain size.
These
from
600
films
MoCl5
at atmospheric
°C was another area of
were thick and of large
The main interest in the atmospheric deposition
studies was to examine the
film resistivities as a function
12
of the purity of molybdenum chloride, poisoning contaminants
and growth rate.
Quite
pure molybdenum films were found by
H2 reduction of HoCl5 at atmospheric pressures.
that
increased
resistivity
were
microcrystallinity and metalloid
were
MoO2Cl2
and
°C
found
contamination.
incompletely
Temperatures above 500
The factors
reduced
MoO2
to
be
Impurities
and
MoO.
minimized the negative influence
of oxygen.
Still another area of
source of Mo
comes
investigation
from
the
Refractory metal silicides,
using MoCl5 as the
application
such
as
of MoSi2 films.
MoSi2 , are also being
studied as highly conductive interconnect and gate-electrode
materials.
This silicide (MoSi2 ) is also being studied as a
2
potential material in reducing the gate dimensions in I L
(integrated injection logic) circuits
E243. Studies in this
area all produced good
quality, highly oriented thin films.
The films may also
easily
be
not form hillocks.
chemically dry etched and do
Molybdenum
disilicide
films are also
resistant to HCl, HNO3, H2PO4 , H3PO4 , and HF solutions.
Although
hydrogen
reduction . of
purity thin films, there
are
method of
Most
deposition.
MoCl5
produces
several disadvantages to this
of
excessive film resistivity E153.
the
techniques reported
There were also reports of
unreasonable amounts of film contaminants E123.
disadvantage in using
MoCl5
high
is
The biggest
that at temperatures below
13
150 °Cr the chlorides
pentachloride also
condense
can
not
on any surface.
be
obtained
commercially and
special equipment is needed to produce this gas.
two disadvantages make MoF6
as
the
Molybdenum
These last
source of Mo much more
desirable.
Hydrogen and Silicon Reduction of Tungsten Hexafluoride
.
Molybdenum and tungsten
expect similar CVD
hydrogen
chemistry
behavior.
reduction
of
is similar enough to
Tungsten
tungsten
deposition by the
hexafluoride
for
use in
microelectronics is well known E323.
Kinetic work by McConica
and Krishnamani, and Broadbent
and Ramiller, showed deposition of tungsten from MF6 is very
similar to molybdenum
deposition
found hydrogen reduction of MF6
and zero order in
MF6
with
10
MoF6 [28,293.
They
to be 1/2 order in hydrogen
an
J/mol (0.71eV) at temperatures
pressures from 0.1 to
from
activation energy of 69000
from
torr.
250
°C
to 500 °C and
The preexponential factor
was determined by McConica and Krishnamani to be 6.2 X IO4
nm/S'Pa 0 "5 . McConica's and Krishnamani's investigation also
reported the rate limiting step could be either the addition
of
adsorbed
fluorinated
monotomic
tungsten,
However, Broadbent and
mechanism
to
be
the
hydrogen
or
to
hydrogen
Ramiller
adsorbed,
fluoride
partially
desorption.
reported the rate limiting
Krishnamani,
Ramiller, also reported dissociation
of
and
Broadbeht
and
Hg adsorbed on the
14
surface.
McConica
and
tungsten
deposition thickness and
initial native oxide
observed in the
to
absence
depositions
reduction
reaction.
a
limiting
structure that was dependent upon
characteristics.
limiting
have
of
This condition was
hydrogen,
were
only
Other
or
shown
studies
in other words,
in
the
silicon
involving
the low
pressure chemical vapor deposition of tungsten have reported
similar findings Cl,8,30,31,323.
Hydrogen Reduction of Molybdenum Hexafluoride
As stated earlier,. the
information on LPCVD molybdenum
from the hydrogen reduction of
have been only two very
MoF6 is very limited.
There
recent studies that have conditions
somewhat similar to the research conditions utilized in this
research.
In
a
study
by
Woodruff,
deposited by hydrogen
and
silicon
The temperature ranged from
from 2-5 torr.
were carried out in
a
hot
al.,
Molybdenum
was
reduction of MoF6 [263.
200-500 °C, the pressure ranged
The hydrogen
MoF6 flow rate range 5-25
et.
flow rate was 100 seem and the
seem.
Experiments for this study
wall, low pressure CVD reactor.
The MoF6 was introduced by bubbling hydrogen through MoF6 .
It was found that
against deposition on
the reaction was completely selective
silicon
dioxide.
The reaction with
15
silicon took place at
a
very
limiting, in contrast to
and Si.
To
prevent
rate and was not self-
the analogous reaction between WFfe
the
molybdenum and thus the
high
silicon
reduction reaction with
severe etching of silicon, attempts
were made to put a capping layer between the silicon and the
depositing molybdenum.
sputtered TiW and Mo
Films of selective CVD tungsten, arid
were
the MoFg-Si reaction.
ineffective
More
as barrier films to
importantly, it was discovered
that the Mo films deposited over silicon were high in oxygen
content and porous.
A similar study by Lifshitz, et. al., was performed in a
hot walled, tubular reactor
400 °C and a pressure
at
range
a temperature range of 200-
of
0.2
to 0.9 torr C253.
Mo.
films were deposited at these conditions by LPCVD on silicon
substrates by the
reduction
of
hydrogen and argon atmospheres.
molybdenum hexafluoride in
The deposition proved to be
extremely selective, with no Mo observed on silicon dioxide.
Reduction
contribute
by
to
both
the
hydrogen
and
deposition
extremely high deposition rates.
was observed.
Again
the
main
extreme porosity - about 30%.
silicon
with
were
shown
approximately
to
equal,
Mo self limiting thickness
feature of the deposits was
The
films grew in a loose,
open structure which could
be easily penetrated by reactant
gases.
used
This porosity
was
high deposition rates, high
to
explain such things as
resistivity and the continuing
reaction of Si with the molybdenum hexafluoride reactant.
16
The
only
other
hydrogen reduction
Delval C83.
substrates
copper,
This
but
existing
of
MoF6
study
rather
study
was
was
temperature ranged from
performed
various
and
600
LPCVD
Mo
by the
by Schroff and
not performed using a silicon
using
stainless-steel
of
substrates
molybdenum.
to
1100
°C
such as
The deposition
and the pressure
varied from 5 to 760 torr within a HgZMoF6 ratio of I to 60.
This investigation was performed
to study the dependence of
the deposition thickness
on
the
number of defects in the
films were measured along with the
amount of gaseous impurities,
was found that bubble
HgZMoF6 ratio was
kept
pressure kept below 20
700
free
in
particular, fluorine.
The
It
coatings were obtained when the
within
torr
reaction parameters.
the
range
of
3 to 6., the
and the temperature kept above
Low fluorine content deposits were observed to lead
to thermal instability.
Analyses by AES
Auger Electron Spectroscopy
(AES)
is
a technique that
may be employed in measuring the quality and quantity of CVD
deposits.
The technique is based on the Auger process.
Auger process itself is
The
preceeded by an excitation process,
which leaves an electron hole in a core level of an element.
In
this
case,
the
preliminary
excitation
process
is
17
stimulated by absorption of
beam.
After the
energy
excitation
process,
filled by an electron occupying
is, either a shallower core
another
a higher energy level, that
or a valence state. This
the
process and the energy is
electron.
receive enough energy to leave
Auger electron.
Auger
the core hole can be
level
electron must lose energy in
transferred to
from a primary electron
This
last
electron may
the system, thus becoming an
electrons
are
then detected by an
analyzer that measures their energy.
AES can be used
to
conduct
deposited molybdenum.
profiling is
The
equivalent
to
a
depth profiling
equipment
a
used
gas, usually Argon.
the
The
AES depth
standard Auger spectrometer
except for the prescence of an ion gun.
to bombard the surface of
in
of the
The ion gun is used
sample with ions of an inert
bombardment
removes the surface
atoms of the sample at a slow rate [333.
If a quantitative Auger analysis is carried out stage by
stage
during
interruptions
in
the
ion
bombardment, the
results will give the composition of the sample at different
depths with respect to the
molybdenum on
original
surface.
The depth of
silicon can then be obtained from:
z = 3.6 X IO-4EMZplj S.
P
18
where z equals the erosion
rate
in
pm/hr, M is the atomic
weight of the target atom in
amu's, p is the density of the
target in g/cm3, j
primary ion current density in
is
the
yA/cm2 and S is the sputter yield E331.
The
Auger
several
depth-profiling
factors
like
impurities in the ion
content in the
ion
chamber,
on
mixing,
is
affected
surface
by
roughness,
beam, residual gas adsorption, oxygen
These all contribute to
Further details
technique
uneven
current distribution, etc.
poor depth resolution and accuracy.
AES
can
be
found
in
the literature
[33,341.
Analyses by SEM
A very
powerful
tool
in
electron microscopy (SEH).
surface
This
studies is scanning
technique enables one to
observe and analyze phenomena: occuring from a scale of about
50
A
to
several
centimeters.
microscope not only permits
but also
reveals
the
The
analysis
spatial
scanning
electron
of very tiny objects,
or structural relationships
between components analyzed E353•
In the scanning electron
analyzed is irradiated with
microscope,
a
which is rastered across the
the surface to be
finely focused electron beam
surface
of the specimen.
The
types of signals produced when the electron beam impinges on
a
specimen
surface
include
secondary
electrons.
19
backscattered electrons. Auger
rays, and photons of
various
electrons are preferred
to
energies.
the
they provide higher contrast
in the
case
of
electrons, characteristic st­
rough
secondary electrons
is
backscattered elctrons, as
due to their enhanced emission
surfaces
only
emerge from the specimen
Secondary emission
E363.
The detector for
sensitive
with
less
to
than
electrons that
50 keV of energy
E373.
The incident beam of electrons
the specimen is similar to
A directly synchronized
cathode ray tube, and
modulated
by
the
detector.
In other
that
raster
the
used in a television tube.
pattern
intensity
signal
from
words,
the screen will depend
that scan the surface of
the
is
displayed on a
of the moving spot is
the
secondary
electron
brightness at any point on
on
the
strength of the signal from
the corresponding point on
the
specimen.
image of the specimen surface is
In this way, an
built up on the CRT, point
by point E383.
In using SEM
obtained
from
morphological.
several
ways
techniques
the
for
secondary
Morphological
E393.
electron
studies
For
molybdenum) on a surface, a very
recognition of shape or crystal
analyzed is well known, it
film studies, information
images
Can
deposition
is usually
be performed in
of
metal
(i.e.
useful technique is direct
habit.
If the system being
may be possible to recognize the
20
various constituents present and obtain an estimate of their
relative size and concentration.
Scanning
electron
microscopy
quantitative technique
The operation of the
is
not
as
a
without
equipment,
qualitative
and
its flaws however.
the sample preparation and
the interpretation of the instrument parameters are all very
complicated.
A very
required to operate
affect
SEM
skilled
SEH
results
and experienced technician is
equipment.
are
specimen
scattering, channeling patterns,
Other factors that
contamination, multiple
etc.
These are explained
in detail by Cocks [353.
Analyses by ESCA
Another technique used
in
the
electron spectroscopy for
chemical
like AES,
sensitive
is
a
surface
analyzes electrons released from
study
of thin films is
analysis (ESCA).
technique.
ESCA,
ESCA also
a surface by their kinetic
energies.
Impinging x-rays on a
sample surface cause the emission
of electrons by the photoelectric effect.
The electrons are
analyzed by
more importantly,
their
kinetic
their binding energy.
energy,
and
The binding energy is determined by:
KE = hi) - BE - Og
21
where KE is the kinetic
energy
of emitted electrons, hu is
the photon energy, BE is the binding energy of the electrons
and Org is the spectrometer
work function.
The spectrometer
work function and photon energy are known entities.
Like
Auger
electrons,
electrons in ESCA
produces
them.
possible.
The
are
binding
characteristic
This
binding
bonding of an atom.
combined with
the
makes
energy
of
of
the
the element that
elemental
identification
also indicates the chemical
For example,
oxygen,
energy
it can tell if an atom is
flourine
or
other such impurities.
providing
details on the surface
This information aids
in
chemistry of a film.
Further
details on ESCA may be found
in the literature C40-443.
In
analyzing
thin
films
electronic applications,
important.
the
ESCA analysis
of
molybdenum
purity
may
be
of
for possible
the
film is very
used in combination with
ion sputtering to analyze
successive film depths throughout
a layer of
Each
molybdenum.
analyzed to see what
place from the
sample
energy values for
the
literature
chemical
element
shifts or changes are taking
surface
elements
[443.
or impurity can be
and
From
to
the interface.
Binding
compounds are tabulated in
these
values
and
their
corresponding spectra, the chemical make-up of a film can be
determined and
also
speculation
porosity, resistivity, etc.
can
be
made
as to film
22
Analyses by RBS
Rutherford
backscattering
used in CVD film analysis.
previous ones, uses a
(RBS)
beam
This technique
charged particles.
of monoenergetic and collimated
impinge perpendicularly on a
incorporates
The
another technique
This technique, similar to the
alpha particles (He-nuclei) to
target.
is
particles
devices which collimate or focus
a high-energy beam of
pass through a series of
the beam and filter it for
a selected type of particle and energy.
When
the
high
surface of a
energy
sample,
alpha
some
particles
particles
sample, some particles pass through
of
a
tliin
target)
backwards at
direction.
angles
These
and
Other
greater
particles
90°
printed out in
As in ESCA
possible
in
a
are
from
particles
the detector generate an electrical
amplified and processed by
are implanted in the
the sample (in the case
than
backscattered
penetrate the
the incident
that impinge on
signal.
computer.
scattered
This signal is
The data is then
the form of a spectrum.
and
RBS
AES,
since
elemental
identification is also
it . produces
spectra
that
are
characteristic of the parent atom. However, the advantage of
RBS resides in the
perceive depth
speed
of
distributions
surface without
modify the sample
sputtering
and
lead
the technique, its ability to
of
(ion
atomic
species below the
sputtering
can sometimes
to erroneous conclusions), and
the quantitative nature of the results.
Further details on
23
RBS are available in the literature E453.
RBS in thin film
studies
film purity.
Observation of
film is pure
(clean
elements
sharp
(interdiffusion
provided by an RBS spectra
is important in analyzing for
an RBS spectra shows whether a
peaks)
of
can
or
peaks).
combined with other
The
information
be used in conjunction with
ESCA and AES in confirming the chemical state of a deposited
film.
24
RESEARCH OBJECTIVES
The objectives
of
this
research
are essentially two­
fold:
1.
Study
the
kinetics
of
the
hydrogen
reduction
of
molybdenum hexaflouride on a silicon surface.
2.
Characterize
deposited
molybdenum
films by AES, ESCA,
SEM and RBS and compare the findings with those published in
the literature.
25
EXPERIMENTAL EQUIPMENT
The stainless steel reactor system is diagramed in Figure
2.
This
low
pressure
chemical
system consisted of the
vapor deposition (LPCVD)
following
primary components:
the
gas flow control system, the reactor and heater, the chamber
pressure control system, the pumping system, the trap system
and the acid gas
heater used
in
detection
this,
system.
system . was
silicon wafer received a
The contact substrate
designed
such
that the
uniform heating distribution which
is required for obtaining accurate kinetic data.
Gas Flow Control System
Hydrogen (99.9995% pure,
pure,
Matheson
Co.)
pure, SERAC Co.)
controllers.
and
flows
All the
MKS 247B four channel
flow
ratios
gas flow
ranges
were
sccm/min
for
by
MKS type 1259
controllers were connected to a
and
power supply unit.
The
were set independently, but could
of
0-101
and
hexaflouride (99.9%)
controlled
readout
have been set.as
hexaflouride.
molybdenum
were
gas flows in each channel
helium
Matheson Co.), helium (99.9999%
one another.
sccm/min
0-21
The controllable
for hydrogen, 0-145
sccm/min
for
molybdenum
Wood
flee and
Prcaaura
Control
Clattronlta
Tamp. Control
Varlac
Flow Control
throttle Valve
Pneumatic Valve
After Burner
Pneumatic
V e l,v e
Alumina Trap
Figure 2.
LPCVD Reactor System
I
27
Reactor and Heater
The low pressure
2) was wrapped
with
stainless
steel reactor system (Figure
fiberglass
water vapor inside the reactor
reactor was accessed
by
heater was attached.
steel block was
unit.
Centered
a
A
used
as
in
this
was
2.5cm
the
with
Ni-chrome
(Figure
3).
The
wires
for
wire
four equally spaced
running
through them
connected
in
a
series
of the system was 2.1
Ni-chrome wires were attached to
in
were
power supply controller.
which the substrate
were
were
copper feed-throughs welded
copper feed-throughs
The
3.5cm X 5.0cm stainless
resistance
the
to insure that
substrate holder and heating
tubes
The leads
to
X
block
arrangement and the total
tape
kept at a minimum.
flange
ceramic
ohms.
heat
in
the
turn
The
access
flange.
connected
These
to a variac
manually controlled heating
arrangement was capable of temperatures up to about 500°C.
Two 1/4" stainless steel rods which were screwed into the
feed-through flange acted as
the
support for the substrate
holder.
the
substrate
The
underside
machined grooves for
of
positioning
this
holder had two
holder on the rods.
The substrate holder was spot-welded to the rods.
Centered at the surface
of
the substrate holder was the
Alumel-chromel thermocouple.
The
attached
AlumeI-chromeI feed-throughs
to
larger
diameter
thermocouple wires were
Thermocouple
. : :i
N)
OO
To
Variac
SIDE VIEW
TOP
Figure 3 .
Substrate Heater/Holder for the LPCVD Reactor
VIEW
29
which were welded in the access flange.
recorded
by
a
digital
The temperature was
thermometer
and
was
manually
controlled by varying the supply voltage through a variac.
Silicon slices were
cut
to
fit
into, the top grooved
portion of the substrate heater/holder.
Pressure Read-Out and Control System
The
chamber
pressure
was
capacitance manometer gauge, MKS
was mounted on top
of
the
sensed
baratron
by
an
absolute
type 222B.
reaction chamber.
This
The pressure
was displayed on a MKS power supply and digital readout PDRD-I.
A MKS throttle valve type 253-1-40-1 Was quick clamped to
the bottom of the
reaction
and pressure sensor
valve
controller
were
type
chamber.
The throttling valve
interfaced
through a MKS exhaust
253A.
The
chamber
pressure was
controlled by regulating the throttle valve opening and thus
the pumping speed.
Pumping System
The pumping
system
consist
pumps connected in parallel.
of
One pump (secondary pump) is a
Precision vacuum pump, model D25.
below the reactor and
chamber.
This pump has
evacuated
a
two mechanical roughing
The pump sat on the floor
directly from the reaction
maximum
speed of 1500 1/s (0.88
30
CFM).
The pump was used
0.02 torr and was
to achieve the initial vacuum of <
isolated
valve during the reaction.
from
The other pump (primary pump) is
a Leybold-Heraues, model D4A,
1500 1/s.
This pump was
the pump used during
the system by a pneumatic
which
located
the
has a maximum speed of
in a exhaust hood and was
reaction
to exhaust the residual
reaction gases.
As stated earlier, the
throttle valve was connected to
the stainless-steel reaction chamber.
Between the throttle
valve and the mechanical pump
in the hood were two
located
residual gas traps, an oil trap
pneumatic valve was a
safety
oil backstreaming in the
The
other
feature.
mechanical
Both
and a pneumatic valve.
feature to protect the pump's
lines
pump
The
in
case of a power outage.
also
pneumatic
contained
valves
were
this
safety
activated
by
pressurized argon gas and were designed to fail closed.
Acid Gas Detection System
Since HF acid was one of the by-products of the reaction,
a sodium hydroxide neutralizing generator
exhaust hood to
neutralize
exhausted from the
roughing
bubbled through a sodium
base
reaction
precipitate.
took
HF
gas.
pump
in
was set up in the
When the gases were
the
hood, they were
hydroxide solution where the acid-
place
Phenophthalein
leaving
was
added
a
sodium
to
flouride
the NaOH as an
31
indicator.
The
purple
indicator
turned
clear
if
the
solution became acidic.
The purpose of
provide an
this
indication
acid
of
how
system was in adsorbing the
the pump from this
dilute.
gas.
neutralizing
efficient
station was to
the alumina trap
residual. HF gas and to protect
The
NaOH solution was made very
to
change the alumina trap each
The procedure was
time a NaOH solution turned clear.
After-Burner and Alumina Trap System
A stainless steel, in-line
were
positioned
between
the
roughing pump located in the
stainless steel can
The can was
after-burner and alumina trap
filled
throttling , valve
hood.
with
fiberglass
the Ni-chrome wire and
The after-burner was a
stainless steel shavings.
insulation
secured
The Ni-chrome wire was connected
constant temperature of
reactor
the
wrapped with ceramic-insulated Ni-chrome wire.
Another layer of
the
and
was
around
reduced
on
was wrapped around
to the stainless steel can.
to
a variac and kept at a
300°C.
the
Residual MoF^ from
heated
stainless-steel
surface.
Two copper rods mounted on a stainless-steel feed-through
flange were quick-connected to the side of this after-burner
and
had
spot-welded
thermocouples
stainless shavings in the
in
after-burner.
contact
with
the
This flange and
32
the bottom flange (going
60°C by
cold
tubing.
The alumina
alumina.
water
This
to
the
running
trap
high
pump) were both cooled to
continuously
in
1/8" copper
was a stainless-steel can filled
surface
area
alumina
adsorbs
and
neutralizes the HF gas by forming AlFg.
Chemical Hazards
The
safety
hazards
associated
mainly dealt with MoFfif Hg
and
with
LPCVD molybdenum
HF gasesf and also NaOH and
NaF solutions.
Fluorides from the reaction could potentially be emitted
into
the
atmosphere
fluorides such
toxic.
as
or
the
pumping
MoFfi and
HF
Molybdenum hexafluoride,
airf liberates
corrosive
hydrofluoric
substance.
are
highly irritable and
a
very
exposure
inflammation and congestion of the lungs.
HF causes
irritation,
burns
dangerous
effects
skin
of
sclerosis of the bones.
and
severe
contact
is
and
irritating and
will
cause rapid
Skin contact with
pain.
from
The most
HF
causing
Sclerosis of the bones is caused by
the fixation of calcium by fluorine.
also corrode
Inorganic
when exposed to moist room
acid,
Severe
system.
severely
reduce
Hydrogen fluoride will
the
operating
life of
pumps.
As stated
MoFfi on the
earlier,
the
stainless-steel
adsorbs any HF gas on
the
after
burner reduces residual
shavings
high
and the alumina trap
surface area alumina. The
33
sodium hydroxide neutralizing generator
back-up
in
indicating
adsorption system.
The
the
produces
Sodium hydroxide is
a
The NaF and NaOH
Sodium
efficiency
also added as a
of
the
alumina
acid-base reaction that takes place
between NaOH and HF
severe burns.
is
a
toxant
and
fluoride
solutions
NaF precipitate and water.
may
is
irritant, and can cause
an inorganic fluoride.
be disposed of by flushing
them down the drain with copious amounts of water.
Hydrogen gas is
extremely
allowed around any source of
explosive
ignition.
and should not be
In large quantities
in an enclosed area, hydrogen may also cause asphyxiation.
34
EXPERIMENTAL PROCEDURES
Sample Cleanincr
The silicon slices used
in dilute HF acid
to
acetone and methanol
for deposition were first soaked
remove
to
any
oxides and cleaned with
remove
organics.
All stainless
steel flanges, traps and various accessories associated with
the reaction chamber were all
boiled in distilled water and
rinsed with acetone and methanol.
boiled, ultrasonically cleaned
The substrate holder was
and
rinsed with acetone and
positioned
on the substrate heater
methanol.
Deposition Procedure
A silicon slice
and placed in the
was
chamber.
lines were then evacuated
gas
controller
valves
to
300°C and
the
cooling
after-burner flange.
gas
lines
were
closed
prevent
after
wrapped
temperature above 7O0C to
and
the
lines
The
were
any moisture leak through.
burner
water
The
reaction chamber and gas
approximately 0.02 torr.
were
pressurized with gas to
During this time the
The
was stabilized at about
was
running
to protect the
reaction
chamber
as well as the
with
aid
heat
tape
and
kept
at a
in evaporating any remaining
water vapor and to keep the MoFfe from condensing.
35
The substrate heater
was
turned
on slowly (to protect
from thermal shock and breaking of the wires) and stabilized
at the deposition temperature
(see
slowly introduced turning
the
simultaneously.
on
Table 2).
Hydrogen was
exhaust valve controller
When the reaction chamber was stabilized at
the test pressure and the substrate heater was stabilized at
the deposition temperature, the MoFfe was introduced into the
chamber.
The
residence
time
approximately five seconds.
on and the substrate
within + 2°C.
of
the
MoFfe gas
was
Immediately a timer was turned
temperature was manually controlled to
Careful
attention
was given to the color of
the NaOH-phenophthalein solution.
At the end of deposition the hydrogen and MoFfe were shut
off. The throttle valve was opened completely and helium was
introduced into the chamber to
helium
purged
the
system
temperature was reduced.
was closed and
cool down the
temperature
helium
about
(for
was
purging the throttle valve
introduced
When
65°C,
the
the
atmospheric pressure with helium
While the
5 minutes), the substrate
After
substrate.
of
purge the system.
into the chamber to
substrate
chamber
was
reached a
brought to
and the sample was removed
for analysis.
AES Analysis Procedure
When the sample was
removed
from the reaction chamber,
it was labeled and transported to the Montana State
36
Table 2.
Temperature, Time and Partial Pressure Variations
for the Kinetic Study of the Hydrogen Reduction of
Molybdenum Hexafluoride
I time
I
I (min)
I
I
pH2
(torr)
I
P
I
rtot
(torr) I
I
0.056 I
0.844
I
0.900 I
I
0.30
I
1.00
I
1.30
I
I
1.50
I
i.o.o
I
2.50
I
3.00
I
1.00
I
4.00
I.
I
0.30
I
4.70
I
5.00
I
I
0.30
I
7.00
I
7.30
I
I
7.00
I
1.00
I
8.00
I
1_
0.30
I
9.70
I
10.00
0.056 I
0.844
I
0.900 I
I
0.30
I
1.00
I
1.30
I
I
1.50
I
1.00
I
2.50
I
3.00
I
1.00
4.00
I
0.30
I
4.70
I
5.00
I
0.30
I
7.00
I
7.30
I
I
7.00
I
1.00
I
8.00
I
I
0.30
I
9.70
I
10.00
I
0.844
I
10
I
I
T ,.
(0C)
250
A
5
300 & 350
I
I
A
5
400
Ten, 16 and 20
pMoFe
(torr)
I
0.056 I
minute
I
0.900 I
experiments were also run
partial pressure ratio, total pressure and T = 300 °
37
University
analyses.
Physics
Each
Department
sample
for
piece
was
sections so that each section
of
Auger
cut
depth profiling
into
I
cm x I cm
the same sample
could be
analyzed by different methods.
The
AES
depth
profiling
Physical Electronics
(PHI
scans
595)
were
performed
by a
scanning Auger microprobe.
Each sample was sputtered by an Argon (Ar+ ) ion beam for 1/2
to 5 minute intervals.
area sputtered by the
At
ion
the
beam
primary electron beam had a
current of approximately
composition
0.20
data
was
voltage
analyzed by AES.
of
pA.
surface composition after each
surface
end of each interval, the
3.0 keV and a beam
The analyses gave the
successive ion sputter.
was
The
recorded
digitially
The
on
a
magnetic disc with the use of a DEC PDP 11/04 computer.
In using AES
as
an
instrumental analysis technique, a
few problems were encountered.
roughness
and
Due to the amount of surface
impurities
associated with each sample,
not be obtained C463.
(oxygen,
flourine,
carbon)
a constant sputter yield could
Without a constant sputter yield, the
AES depth equation was invalid.
Realizing these problems, AES
quantitative and qualitative
was
basis.
then used on a semiIn other words, even
though definite amounts of molybdenum deposited could not be
38
determined for each
sample,
relative amounts of molybdenum
and impurities present could be determined.
SEM Analysis Procedure
Samples for
SEM
Science Department
analysis
at
were
Montana
taken
State
were cut to be 5 mm x
5
surface
SEM.
The
the
backside
analyses
performed
by
by
scoring
fracturing along the
prepared,
University.
Samples
mm in size for a cross-section and
scored
processed,
to the Vetinary
and
cutting
line.
of
procedure
the
sample
was
and
The samples were then
photographed
by
the
operating
technician in the department.
ESCA Analysis Procedure
The ESCA scans of
performed
by
a
spectrometer (L-H
the molybdenum deposited samples were
Leybold-Hereaeus
EAlD.
The
x-ray
radiation
photoelectron
source was non-
monochromatized Mg Hot, (HE = 1253.6 eV) x-rays.
The ESCA
data
digital recorder.
was
recorded
A special
modified by the Montana
with
an analog and
program developed by XEROX and
State University Physics Department
aided in analyzing the ESCA data.
capability to smooth the data
locations of the peaks.
both
This program provided the
and obtain the kinetic energy
39
Acid Dissolution Procedure
Acid etching was performed
consisting of;
ml of CHgCOOH
5 ml of HgPO4
(99%)
and
by
starting with a solution
(85%), 30 ml of HNOg (70%), 4
150
ml
HgO.
The solution was
contained in a pyrex beaker.
The I x I cm
sample
was
carefully placed in the acid
solution, agitated
gently
measured on a CAHN
29 automatic electrobalance model number
C-29.
Each
sample
was
and
weighed.
dipped
and
The
weighed
weight was
at 5 minute
intervals until there was ho significant weight change.
solution
was
tested
for
selectivity
of
The
molybdenum over
silicon and was found to be completely selective.
RBS Analyses Procedure
RBS
Physics
was
performed
Department.
in
the
The
Montana
State
non-commercial
incorporated a beam of 1.4 MeV (He nuclei).
University
instrument
40
RESULTS AND DISCUSSION
Thickness Determination of Deposited Samples
Acid etching produced the values
given in Table 3.
The
etching proved to be a good method in determining the amount
of molybdenum deposited.
checked by measuring
The validity of this technique was
the
film
micrograph cross-sections.
were
compared
to
thicknesses pictured on SEM
The
acid
SEM thickness measurements
dissolution
film
thickness
measurements and were shown to be in good agreement.
As discussed in the experimental section of this report,
AES
was
used
to
determine
impurities present in
each
sputter yield
for
varied
the
differing
sample.
samples
amounts
of
This was because the
of
the same thickness;
therefore, the AES depth equation for determining the amount
of deposition was not used.
Reaction Kinetics
Figure 4 shows
constant
total
temperature of
relationship.
molybdenum
reaction
300°C.
thickness
pressure
The
of
plot
The reaction shows
versus
0.9
indicates
time at a
torr
and
a
a non-linear
a high rate of deposition
as time increases (past 15 minutes) and there is no observed
self-limiting behavior.
has been observed by
The
other
lack of self-limiting behavior
researchers, but there has been
no report of
molybdenum
deposition
on
silicon being non­
linear [263.
By using Figure 4 and the Auger analysis of
41
Table 3 . Molybdenum Film Thickness Measurements by Acid
Dissolution (for the Kinetic Data).
P /p
Hz MoFe
tot
Film
Thickness*
(0C)
(nm)
(torr)
0.9
I
I
1.3
I
I
2.5
I
I
I
22
3.3
I
22
I
0.67
I
12
4.0
I
0.33
I
69
5.0
I
15
I
57
I
7.3
I
23
I
60
I
8.0
I
I
22
32
I
90
15
I
53
.3.3
I
105
I
118
250
I (10 min)
1
I
O
I
I
I
0.9
. I
I
I
1.3
I
2.5
I
I
O
I
iH
I
0.14
0.67
OJ
OJ
I
15
300
[
I
(5 min)
|
in
I
I
7.3
I
I
CO
I
I
10.0
I
32
|
287
I (10 min)
|
0.9
I
15
I
93
I (16 min)
|
0.9
I
15
I
170
I (20 min)
I
0.9
I
15
I
320
■ O
I
264.
I
4.0
-
O
I
15
I
139
I
23
I
185
i— f
o .
O
138
* Thickness measurement results were repeatable within ± 4 nm
Thickness
(nm)
42
T ime
Figure 4.
(min)
Molybdenum Thickness as a Function of Time at
T = 300 °C and Pfcot = 0.9 torr
43
samples deposited at
determined
that
250
ten
°C,
300
minutes
°C
of
sufficient thicknesses of 10
nm
and
350 °C, it was
reaction
or
time
provided
greater for the 250 °C
samples and that five minutes of reaction time also provided
sufficient thicknesses of 10 nm or greater for the 300°C and
above samples. These
times
were
also
short enough for a
linear initial rate to be used in analysis.
Figure
5
temperature
shows
versus
an
r
Arrhenius
(rate
of
plot
for
reciprocal
deposition)
for
total
pressures of 0.9 and 5.0 torr and for total deposition times
of five and ten
minutes.
kept constant at 15:1.
total
pressure
The
A
curve.
hydrogen to MoF6 ratio was
slope
From
was calculated from each
these
values
an
average
activation energy of 76,000 ± 1500 J/mol was determined.
Order of Reaction
The
reaction
order
with
determined as a function
of
The partial pressure
MoF6
of
respect
the
to
hydrogen
was
partial pressure of MoF6 -
was
kept constant while the
hydrogen partial pressure varied from 1.0 to 9.7 torr within
a temperature range of 250-350°C.
of In
r
(rate
of
(partial pressure of
deposition)
as
hydrogen).
respect to hydrogen was
plots.
Figure 6 shows the plots
a
The
determined
function
of In P^2
reaction order with
from the slope of these
The average value of the reaction order with respect
to hydrogen was determined to be 0.5.
44
-9300
(nm/min)
slope
5.0
0.9
slope
1.5
-9030
1.6
1.7
T -1
Figure 5.
Arrhenius Plot
torr
torr
-E
/R
1.8
( I O - 3 K " 1)
1.9
2.0
0.3
to r r
slope
0.51
slope
0.54
3.0 (]
C)
(n m / m i n )
(300
(250
u C)
In P
Figure 6.
(t o r r )
Plot for Determining Order of Reaction
Respect to Hydrogen Partial Pressure
With
46
Similarly, the reaction order
determined
hydrogen.
as
a
function
The hydrogen
at 1.0 torr while the
of
with
the
partial
respect to MoF6 was
partial
pressure
of
pressure was held constant
MoF6 partial pressure varied from 1.5
to 7.0 torr within a temperature range of 250-350°C.
Figure
7 shows the results of the In r (rate of deposition) plotted
as a function of In
PMoF6
order of reaction with
the slope values.
and
were
(partial pressure of MoF6 ).
respect
The
to MoF6 was determined from
The values obtained had a slight variance
statistically
average, a reaction order
averaged.
of
0.0
From
the statistical
with respect to MoFfc was
inferred.
Rate Equation
For all of
factor ko
was
the
experimental
calculated
order of reaction and the
knowing
runs, the pre-exponential
the
activation energy,
reaction rate at each temperature
from:
r/(e-E/RT>(P0 -Sh 2P0mof6,
where
r = Reaction rate, (nm/s)
kQ = Pre-exponential factor, (nm/s/torr0,5)
PHz = Partial pressure of hydrogen, (torr)
PMoFe = Partial pressure of MoFfc, (torr)
4 .O
P
=
1.0
to r r
H2
= 0 .07
3
0
(n m / min)
(300
°C)
2.0
Vj
C
I .0
— □ -------
D
Q
slope
(250
= 0.0
°C)
0 ---------------------------------------- —
-3.0
-2.0
- 1.0
0
1.0
2.0
3.0
ln pMoF6 (t°rr)
Figure 7.
Plot for Determining Order of Reaction With
Respect
to
Molybdenum
Hexafluoride Partial
Pressure
48
E = Activation energy of the reaction, (J/g-mol)
R = Universal gas constant, (J/g-mol-K)
T = Temperature, (K)
A statistical average of all the kQ values was performed
and k was determined
O
s
•torr
.
the growth
Using
rate
to
be
2.02
x
10
±1.2x10
nm•
kO and the previous rate expression.
molybdenum in units of nm/s due to
of
hydrogen reduction of MoF^ can be expressed as:
2.02 X IO6 e ( 9140/T) p0
MoFe
,0.5
Characterization of Molybdenum Films
Figure 8 shows an AES
at 4 torr of total
300 °C.
This
molybdenum
highest peak
pressure
sample
content
comprising the
depth
had
with
other
10%
value).
profile of a sample formed
and a substrate temperature of
a
5%
oxygen content and an 85%
silicon,
of
This
carbon,
impurities
profile
and
fluorine
(at molybdenum's
displays
one of the
better depth resolutions.
Table 4 shows
the
relative
impurities present in samples
conducted.
Here,
oxygen
amounts
of molybdenum and
for which depth profiles were
is
shown
percentage impurity in the samples.
to
be
the
highest
The rows which do not
show impurity percentages were poor quality deposits and had
Atomic
Concentration
(%)
100
Sputter
Figure 8.
T ime
(min.)
AES Depth Profile of a Mo Film Deposited
T = 300 °C and Ptot = 4 torr
50
Table 4.
T
(0 C)
250
I
^tot
I
Re!.
Rel.
Rel.
Rel.
Rel.
I
%
%
%
%
%
I(t o r r ) I
Mo
0
Si
F
C
I
0.9
I
55
18
17
I
1.3
I
2.5
I
4.0
I
5.0
I
I
7.3
I
I
1
I
1
3
Est.
Error
±
7
5
—
70
13
10
4
3
4
I
61
28
6
3
2
5
0.9
I
71
14
8
3
4
4
1.3
I
61
22
6
7
4
5
2.5
I
58
18
16
3
5
6
85
5
7
I
2
3
——
——
18
8
2
5
I
O
30 0
Relative Molybdenum Film Impurities from AES Depth
Profiles at Molybdenum's Highest Peak Value
I
5.0
I
I
7.3
I
I
0.9
I
I
1.3
|
2.5
|
—
—
—
4.0
I
40
27
20
7
6
7
58
20
5
13
4
6
77
11
6
4
2
5
350 1
in
I
7.3
I
400 I
0.9
I
O
I
66
—
—
6
—
—
—
51
to be repeated.
The
poor
quality deposits had a very high
oxygen content, a poor surface
away.
Due
to
the
color and a film that flaked
insufficient
time
available,
the
molybdenum films that were deposited again were not analyzed
by AES.
In a
study
performed
by
Lifshitz,
et.
al., excess
oxygen was determined to be dependent upon the temperature
at which the reactor was opened to, air; however, the reactor
in this study was never opened
E253.
Oxygen content
was
upon reactor cleanliness.
to air at a high temperature
also
speculated to be dependent
When the reactor was baked at 400
°C before the reaction, the films were of better purity.
An ESCA analysis of a molybdenum film deposited at 350°C
and 5 torr yields the
figure
shows
two
oxygen
Is
overlapping
presence of oxygen throughout
energy (BE) band indicates
and the low BE
porosity
techniques
could
band
be
available,
presence of adsorbed
The porosity of the
bands
the
the
oxygen
film
metal Oxide.
with
the
to
Although film
the
investigators
also
indicates
film. , The high binding
observed
other
and
The
presence of adsorbed oxygen
indicates
not
band in Figure 9(a).
instrumental
attributed the
the film porosity [25,263.
explained such phenomena as
extremely high deposition rates, uninterrupted Si reduction,
uniform oxidation
and
similar studies with
high
resistivity
tungsten,
films
C253.
In
silicon reduction was shown
to reach a limiting film thickness [283.
52
I
I
I
I
I
I
I
I
I
110.7
Binding Energy (eV)
(b)
531.2 523.6
Binding Energy (eV)
(a)
695.7
686.2
67(
Binding Energy (eV))
(c)
I I
I
I
I
I
I
I Ti
223.6
Binding Energy (eV)i
(d )
Figure 9.
ESCA Spectra for a Mo Film Deposited at
T = 350 °C and P. . = 5 torr. a)0 Is band, b) Si
2p band, c) F ls° band, d) Mo 3d band. I) as
received, 2)after 120 seconds sputter etching, 3)
after 1.2 hours sputter etching, 4) after I.§3
hours sputter etching and annealing at 350 C
under vacuum for 600 seconds, 5) after additional
annealing at 800 C for 600 seconds.
53
Adsorbed oxygen
silicon as seen
The
highest
by
BE
in
the
the
films
silicon
shoulder
to
energy close to that
of
energy shoulder on the
After sputtering for
silicon
1.2
additional
intensity
0.63
of
these
left
has
energy which
is that of elemental silicon.
peaks
the
are
of
peaks
bands in Figure 9(b).
oxy-fluoride. The lowest
hours,
hours
also combined with
The peak in the middle has
silicon
right
oxy-fluoride
2p
the
corresponds to oxidized silicon.
was
oxidized silicon and
greatly
diminished.
sputter-etching
even
more.
An
reduced
the
Considering
the
decreases in impurities as the interface is approached, this
figure suggests that the film
is more porous at the surface
than at the interface.
The fluorine Is spectra
in
Figure
trend as the silicon spectra.
show oxy-fluoride (high
BE)
is approached.
[30],
believe
Schroff ,
that
to
The two bands in the figure
and
Again, the oxy-fluoride peak
9(c) shows the same
metal
fluoride (low BE).
is diminished as the interface
et.
al.
obtain
a
CU]
and Chin, et. al.
higher quality film and
reduce fluorine content. it seems reasonable to work at high
temperatures,
at
high
ratios
of
H2
to
halide,
at low
pressures and with no moisture or carbon.
The carbon can come from
or from the atmosphere
as
the oil of the mechanical pump
volatile hydrocarbons or COg and
CO adsorbed on the pore surface.
54
It is
difficult
molybdenum
spectra
to
in
speculate
Figure
informatively about the
9(d).
The
only
trend
noticeable is that the Mo spectra is asymetric toward higher
binding energies.
shift
toward higher binding energy
suggest that some metal-oxides
are formed, probably as thin
oxide
layers
The
on
the
pore
essentially no source of
surfaces.
oxygen
Since
there
was
in the reactor (except for
maybe a minute amount adsorbed on the reactor walls), the
large amount of oxygen
in
the
Mo
layer
would have to be
incorporated after removal from the reactor.
Impurities may also
be
seen
scans of a deposited film.
by
Figures 10 and 11 show surface
SEM micrographs of some of the samples.
film
formed
temperature
at
of
a
total
400°C
was deposited
pressure.
at
the
Figure
11
pressure
and
smoothness and the fewest
analyzing the surface
of
displays
amount
highest
was
Figure 10 depicts a
of
0.9
the
best
"bubbles".
temperature
deposited
torr
at
and
surface
Figure 10
and the lowest
a low temperature
(250°C) and a relatively low total pressure (2.5 torr).
deposit depicted by
Figure
11
a
The
showed more surface bubbles
than the deposited pictured in Figure 10.
The surface micrographs
increasing temperature.
of processes.
show
Bubble
bubbles
to decrease with
formation involves a number
Some authors attribute the presence of
55
, * I :;
Figure 10.
SEM Surface Micrograph of a Mo Film Deposited at
T = 400 C and P.tot. = 0.9 torr
56
Figure 11.
SEM Surface Micrograph of a Mo Film Deposited at
T = 250 0C and Pfcot = 2.5 torr
57
bubbles to flourine
most
widely
impurities
accepted
in
hypothesis
trapped in the crystal
lattice
during deposition [47].
the deposits, while the
is
that
gas
atoms are
in the supersaturated state
Shroff
and Delval explain bubble
growth as followss
gas atoms move through the lattice and form tiny
clusters which
precipitate
as bubble nuclei;
further expulsion of gas. from the lattice cause the
individual bubbles to
grow
and creates more
bubbles, the coalescence
rate increasing with
increasing temperature [83.
Figure 12 shows an SEH micrograph of a surface deposited
at 300 0C and 1.3
and Figure 12(b)
hour.
torr.
is
Figure 12(a) is before sputtering
after
sputtering
the surface for 1/2
The film shown in Figure 12 has a smooth surface, and
after sputtering not much change in uniformity is observed.
Figure 13 shows a SEM
O
250 C and 7.3 torr. This
hour.
The
high
film was also sputtered for a 1/2
pressure
conditions initially
sputtering (Fig.
micrograph of a film deposited at
produced
13(b),
the
low
a
temperature
rough
surface
surface
lower total
micrograph
however,
was
pressure
formed
than
film
thickness
and
The film
at a higher temperature and
Figure
analyses, Figure
and after
showed a non-uniform
condition or the presence of white and dark spots.
shown in Figure 12
formation
12
13.
shows
From the surface
a
uniformity
better deposit;
are
two
other
'
■
v
S
«
K
*
*
^
B
V5IiES
m
i i s s ,
. .m
. '■ .i.■
•-..
#
#
:
Figure 12.
#
#
•*>.
■' * ■ ' %
-
<
%
#
r - .. -
. '' ■
•** »
, M
"■• ~.'.f
; # •'
*
-
SEM Surface Micrograph of a Mo Film Deposited at
T = 300 °C and P.tot. = 1.3 torr
(a) as received;
(b) after sputter ethcing for 0.5 hours
'
•• - ■■••*■•-
•
*■'•'.'
V-
Figure 13.
SEM Surface Micrograph of a Mo film Deposited at
T = 250 °C and P.
= 7.3 torr (Numbers 1-5 are
for AES Point StuSies)
(a) as received
(b) after sputter etching for 0.5 hours
60
important parameters that dictate the quality of a deposited
film.
The
white
and
dark
spots
seen
analyzed for their composition by AES.
shown to be concentrated
in
in Figure 13 were
The white spots were
molybdenum
and the dark spots
concentrated in silicon.
Figure 14 shows a SEM
micrograph
of a film formed at a
total pressure of 5 torr and a temperature of 350°C.
14(b) was obtained after sputtering
14(c) after sputtering for
Figure
for one hour and Figure
2
hours.
If Figure 14(b) and
14(c) are closely observed, a
change
in the shading of the
surface after continued sputtering is noticed.
Figure 14(c)
also
shows that the
molybdenum-silicon
white and dark spots
extends
mixture
depicted by the
throughout the deposited film
and close to the interface.
Points I and 4
elemental analyses.
in
Figure
14
were
analyzed by an AES
Figures 15 and 16 show the result.
The
spectra for the dark spot (I)
gives
for Si.
gives strong peaks at 186 and
The white
222 eV, which are
spot
the
(4)
doublet
peaks
spectra also shows moderately strong
eV depicting the
presence
the Auger scans detect
of
the
are
combined
silicides, etc.).
of molybdenum.
This
peaks at 509 eV and 92
oxygen and silicon. Although
presence of oxygen, silicon and
molybdenum, it does not provide
elements
a strong peak at 92 eV
(i.e.
information as to how these
metal
oxides,
molybdenum
Ficrure 14.
SEM Surface Micrograph of a Mo Film Deposited at
T = 350 C and P. . = 5 torr (Numbers 1-5 are
for AES Point Studies)
(a) as received;
(b) after one hour of sputter etching;
(c) after two hours of sputter etching
7
2
I
92 Si
O
100
200
300
4 00
500
Kinetic
Figure 15.
AES Elemental Point Scan
Figure 14 (b); point I
of
600
700
Energy
the
800
(e V)
Surface in
900
1000
7
w
-K
W
B
3
100
200
300
400
500
Kinetic
Ficrure 16.
AES Elemental Point Scan
Figure 14 (b); point 4
600
700
Energy
of
the
ROO
900
(e V )
Surface in
nnro
64
The presence.of silicon
film
cap
also
be
on
seen
backscattering spectra.
the surface of a molybdenum
by
analyzing
the
Rutherford
Figure 17(a) shows the spectra of a
molybdenum film deposited at
350°C
and 5 torr superimposed
on the spectra of a molybdenum film deposit at 300°C and 0.9
torr (b).
The high
intensity peak at the right corresponds
to that of molybdenum and the low intensity peak on the left
is that of silicon.
Figure 17(b) displays
a
sharp
molybdenum peak with no
overlap into the energy range of the silicon peak.
of figure 17(a)
however
shows
silicon into the molybdenum
even though figure 17(a)
the high pressure
conditions may
be
(5
or
The peak
some
interdiffusion of the
vice
versa.
In this case,
is the higher temperature deposit,
torr)
enough
associated with the deposition
to
decrease
the
purity of the
sample.
Schroff reports that
of microdefects C113.
the direction
is
This
perpendicular
direction parallel.
molecules
higher
small,
At
low
is
to
pressure deposits are full
because growth is faster in
the
surface
pressures
consequently
so
the
is
than
in the
number of gas
the
number
of
defects.
A SEH
cross-section
of
the
film
above
(Fig. 17(a))
deposited at 350°C and a total pressure of 5 torr is shown
INTENSITY (a RB. UNITS)
IOOO
1200
ENERGY M
Ficrure 17.
Rutherford Backscattering Scan of Mo Films
(a) Deposited at T = 350 °C and P. . = 5 torr;
(b) Deposited at T = 300 °C and PfcoJ; = 0.9 torr
66
in Figure
again,
18.
may
The
explain
condition of
this
reduction of
the
high
the
pressure deposition conditions,
rough
film.
It
MoF6
could
interface
also
and non-uniform
appears
that silicon
attribute
to
the interface
cross-sections
of
two different
,roughness.
Figure 19
shows
SEM
sections of a sample deposited at a high temperature (400°C)
and a low pressure (0.9
torr).
micron sample shows good
interface.
interface.
some
The
The
silicon
adhesion
film
in
reduction
"wormholeing".
with a relatively smooth
figure
roughness
section of this figure
In figure 19(a), the 2.25
of
may
have
shows
from
shows
a
rough
this interface again suggest
occurred.
The middle
a film characteristic called
The wormholeing may
deposition or possibly
19(b)
have been caused by the
fracturing
the sample for SEM
analysis.
For this research study,
films produced the best
on tungsten
and
finding CS,113..
however
is
quality films.
molybdenum
The major
their
uninterrupted Si
high temperature, low pressure
deposition
porosity.
reduction
and
films were reportedly due to
continually diffusing through
in
hydrogen
and
silicon
also supports this
drawback of the molybdenum films
extreme
the deposited films E251.
Previous literature
Phenomena
uniform
such as
oxidation
of the
the reactant and product gases
the
open porous structure of
Film porosity was observed
reduction
of
MoF6
and
both
most
67
Background
Mo
Interface
Si
Figure 18.
SEM Cross-Section of a Mo
T = 350 °C and Pfcot = 5 torr
Film
Deposited at
Fiaure 19.
SEM Crosg-Section of a Mo Film
T = 400 °C and Pfcot = 0.9 torr
Deposited at
(a) smooth interface
(b) rough interface with wormholeing
69
recently
in
tungsten
films
by
silicon
reduction
C483.
Tungsten and molybdenum films showed similar characteristics
when
analyzed
by
interfaces to be
SEM.
SEM
rough,
was
deposited by silicon reduction
Although the results of the most recent experimenters
seem similar to
major
showed W-Si
with characteristic protusions and
wormholes when tungsten
C253.
cross-sections
the
difference
results
is
found
that
in
Lifshitz,
Woodruff, et. al. C263 used
this experiment, a
et.
al.
C253
and
a hot—wall reactor instead of a
substrate heater.
LPCVD
of
tungsten
and
produced high resistivity
explained by the
high
films
porosity
molybdenum film resistivity
which is an order
C253.
of
molybdenum
was
[25,26].
of
the
reported
also
reportedly
This again was
films.
to
The best
be 60 pft-cm,
magnitude higher than bulk molybdenum
In this study, a molybdenum film deposited at a total
pressure of 0.9 torr and
resistivity of 20 pSl-cm.
of most of the
other
other films were not
to be within
the
a
temperature of 400°C produced a
A comparison of the resistivities
samples
of
accuracy
range
determining
ranged
pA'cm
2
The
sufficient thickness or uniformity
which was used for
between
films was unattainable.
to
of
the four-point probe
resistivities.
several
hundred
Other films
ptt•cm.
general, low pressures favored lower film resistivity.
In
70
SUMMARY AND CONCLUSIONS
The objective of this research was to examine molybdenum
CVD films by AES, RBS, ESCA,
kinetics
for
the
and SEM, and to study reaction
hydrogen
reduction
of
MoF^.
The
significant findings of this investigation are as follows:
1.
Examination of the
CVD
molybdenum
films showed oxygen
content in some films to be as high as 28%.
essentially no oxygen
in
the
oxygen was assumed to
be
incorporated
removal from the
reaction
Since there was
chamber during the reaction,
chamber.
in the sample after
I
The incorporation of
oxygen and impurities in a sample inferred a porous deposit.
This porosity
could
also
explain
the
high resitivity of
molybdenum LPCVD films.
2.
Although
deposition
there
was
parameters
no
and
absolute
the
deposited film, high temperatures
purity
trend
and
between
quality
the
of a
and low pressures seem to
produce the best results.3
3.
A
kinetic
study
showed
the
hyrogen
reduction
of
molybdenum hexafluoride to be 1/2 order in hydrogen and zero
71
order in molybdenum
hexafluoride
of 76000 ± 1500 J/mol.
IO6 ± 1.21 x IO6
nm
A
s-1
with an activation energy
preexponential factor of 2.02 x
torr-0"4
5
was determined.
These
results were similar to those found in tungsten hexafluoride
kinetic studies.
4.
Due to the
observed
porosity
roughness
hexafluoride does
of
not
of
the
appear
the
molybdenum
Mo-Si
to
films and the
interface w
be
a
molybdenum
good compound for
molybdenum LPCVD, under the conditions investigated.
72
RECOMMENDATIONS
Based on the results
of
this experimental work and the
literature review, the following recommendations are made:
1.
Further testing needs to
be
performed to determine the
film purity dependence on temperature and pressure.
2.
A transmission electron microscopy
(TEM) study would be
helpful in analyzing for voids and pores in the films.3
3.
In order to use
AES
depth profiling in determining the
+
amount of molybdenum deposited, an oxygen (0 ) ion beam is
.
needed and also the
with
oxygen.
capability
This
of flooding the AES chamber
modification
would
negative effects of oxygen as an impurity.
eliminate
the
73
REFERENCES CITED
74
REFERENCES CITED
1.
Shaw, J.M., and Amick,
1970.
2.
Sze, S.M.,
1983.
3.
Ghandi, S.K. "VLSI Fabrication
and Sons: New York, 1983.
4.
Beidler, E.A.; Powell, C.F.; Cambell, I.E.; and Yntema,
L.F. J . Electrochem Soc., 98, 21 (1951).
5.
Berezhoni, A.S. "Silicon
and its Binary Systems",
Transl. from Russ., p. 173, Consultants Bureau, New York
(1960).
6.
"Refractory Molybdenum Silicides", Bulletin Cdb-6A, p.
9, Climax Molybdenum Co. (May 1963).
7.
Simeonov, S.S., Kafedjiiska, E.I., and Guerassimov, A.L.
Thin Solid Films. 115, 291-220 (1984).
8.
Shroff,
A.M.,
Delval, G.
Hiorh
Pressures. Vol. 3, p. 695 (1971).
9.
Yasuda, K., and Murota, J. Japanese J . of Appl. Phys.,
Vol. 22, No. 10, p. L615 (1983)
.
"VLSI
J.A.
RCA
Technology",
Review, p. 306, June,
McGraw-Hill:
New York,
Principles", John Wiley
Temperatures-High
10. Fukumoto, M.; Inoue, K.; Ogawa, S.; Okada, S.; and
Kugimiya, K. "I jam Mo Gate MOS Technology", 1981
Symposium on VLSI Technology, Publ. IEEE, New York
11. Schroff, A.M., Thomson-CSF "Influence of the Pressure on
the Deposition Characteristics of CVD Tungsten", 6th
Plansee Seminar 1968, Edited by Benesovsky, Springer
Verlag, 1969.
12. Seto,
D.K.,
Doo,
V.Y.,
Dash
S.,
"Growth and
Characterization of Low
Temperature CVD Molybdenum
Films", Chem. Vapor Deposition, Int. Conf., 2nd, 659-92,
Ed. by: Blocher, J.M., Jr., Electrochem. Soc.. New York,
(1970).
13. Hieber, K.; Stolz, M.;
100(3), 209-18 (1983)
Siemens,
AG
Thin Solid Films,
75
14. Casey, J.J.; Verderber, R.R.; and Garnache, R.R. J.
Electrochem Soc., Vpl. 114, No. 2, 201 (1967).
15. Schroff, A.M. High
415-421 (1974)7
Temperatures-Hiqh Pressures. Vol. 6,
16. Inoue, S.; Toyokura, N.; Nakamura, T.; Maeda, M. and
Takugi, M. J. Electrochem. Soc.. Vol. 130, No. 7, 1603
(1983).
17. Kaplan, L.H., and d 'Heurle,
117, (5), 693 (1970).
F.M. J. Electrochem. Soc..
18. McCreary, W.J.
Fifth
International
Chemical Vapor Deposition, 714, (1975).
19. Carver, G.E., and Seraphin,
(4) (1979).
B.O.
Conference
on
Ap p I . Phvs. Lett. 34.
20. Carver, G.E. Thin Solid Films. (63), 169 (1979).
21. Pinneo, G.G. Proc. of 3rd Int. Cont. on CVD, F.A.
Glaski, Ed., American Nuclear Society, Hinsdale, ILL, p.
462 (1972).
22. Ayugawa, M. Paper # 14p-C-14,
1984.
Fall Meeting of the JAPS,
23. Guivarc'h, A.; Auvray, P.; Berthou, L.; LeCun, M.;
Boulet, J.P.; Henoc, P.; and Pelous, G. J. Ap p I . Phvs..
49, (I) (1978).
24. Sasaki, Y.; Ozawa, O.? and Kameyama, S . IFFK Trans. on
Elect. Dev.. Vol. ED-27, No. 8 (1980).
25. Lifshitz, N.; Williams, D.S.; Capio, C.D.; and Brown,
J.M. "Selective Molybdenum Deposition by LPCVD", AT&T
Bell Labs, Murray Hill, New Jersey. To be published.
26. Woodruff, D.W. and Sanchez-Martinez, R.A., In "Tungsten
and Other Refractory Metals for VLSI Applications", E.K.
Broadbent, Ed., MRS, Pittsburg (1986).
27. JANAF Thermochemical Tables, 2nd Edition, D.R. Stull and
H. Prophet, Editors, National Bureau of Standards,
Washington, D.C., 1971.
28. McConica, C.M., and Krishnamani, K. "The Kinetics of
LPCVD Tungsten Deposition in a Single Wafer Reactor",
AICHE Seattle Meeting (1985).
76
29. Broadbentr E.K . and Ramillerr
Vol. 131r No. 6 , 1427 (1984)
C.L. J. Electrochem. Soc.
30. Chinr J . Nucl. Sci. Abstr. , 23(16), 31326, (1969).
31. Cheung, H. Nucl. Sci. Abstr. , 26(12), 28599, (1972).
32. Blewerr R.S., editor, "Tungsten and Other Refractory
Metals for VLSI Applications," E.K. Broadbent, Ed.,
(Materials Research Society, Pittsburg, PA 1986).
33. Margaritondo, G. and Rowe,
J.E., In "Treatise on
Analytical Chemistry", Elving, P.J., Bursey, M.M.,
Kolthoff, I.M., Eds., John Wiley and Sons, New York,
1981, Part I, Vol. 8, Chap 17.
34. Werner, H.W. and Morgan/ A.E. In "Treatise on Analytical
Chemistry", Elving, P.J., Bursey, M.M., Kolthoff, I.M.,
Eds., John Wiley and Sons, New York, 1981, Part I, Vol.
10, Chap 5.
35. Cocks, G.G. In "Treatise on Analytical Chemistry",
Elving, P.J., Bursey, M.M., Kolthoff, I.M., Eds., John
Wiley and Sons, New York, 1981, Part I , Vol. 8, Chap 16.
36. Thornton,
P.R.
"Scanning
Electron
Microscopy,
Applications to Materials and Device Sciences", Chapman
and Hall Ltd., London, 1968, Chap. I.
37. Lee, H.; Swartz, M.L. and Stoffey,
Electron Microscopy of Polymers and
Princen, Ed., Interscience Publishers,
No. 16, Chap. I.
D.G.
"Scanning
Coatings", L.H.
New York, 1971,
38. Wells, O .C . "Scanning Electron Microscopy", McGraw-Hill,
Inc., New York, 1974, Chap. I .
39. Suite, E.,
(1956).
and
Uyeda,
N.
Proc.
Jpn.
Acad., 32, 117
40. Carlson, T.A. "Photoelectron and Auger Spectroscopy",
Plenum Press, New York, 1975, Chap. I-V.
41. Fadley,
C.S.
In
"Electron
Spectroscopy; Theory,
Techniques and Applications", Brundle, C.R. and Baker,
A.D., Eds.; Academic Press, New York, 1978; Vol. 2,
Chap. I.
77
42. Fadley,
C.S.
and
Hagstrom,
S.B.M.
In
"X-ray
Spectroscopy", Azaroff, LiV., Ed., McGraw-Hill, New
York, 1974, Chap. VIII.
43. Riggs, W.M. and Parker, M.J. In "Methods of Surface
Analysis"; Czanderna, A.W., Ed.; Elsevier Scientific,
New York, 1975, Vol. I, Chap. IV.
44. "Handbook of
Photoelectron
Spectroscopy"; Physical
Electronics Division, Perkin Elmer Corp., 1979.
45. Chu, W.; Mayer, J.W. and Nicolet, M.A. "Backscattering
Spectrometry", Academic Press, Inc., New York, 1978,
Chap. I .
46. Sahin, T., Personal Reference, Montana State University
Bozeman, Montana, (1987).
47. Farrell, K., Federer, J.J., Schaffhauser, A.C., "Gas
bubble formation in metal deposits", in CVD Second
International Conference, Los Angeles (Electrochemical
Soc., New York, 1970) .
48. Green, M.L.; All, Y.S.; Boone; Davidson, B.A.; Feldman,
L.C. and Nakahara, S. To be published.
78
APPENDIX
Sample Calculations
79
This
is
the
calculation
for
determining
preexponential factor, Ic q . .
The kinetic equation for calculating kQ is:
the
0.
= r/(e"E/RTPt -0.5p
rHoFe 1
where;
r = Reaction rate, (nm/s)
_0 5
k = Pre-exponential factor, (nm/s/torr '3)
Pu = Partial pressure of hydrogen, (torr)
PH^p = Partial pressure of MoF6 , (torr)
ET = Activation energy of the reaction, (J/g-mol)
R = Universal gas constant, (J/g-mol*K)
For this example calculation the following will be used:
r
= 0.1808 nm/s
P f y = o.84 torr
pMoFe = °-06 torr
EraoS6VefOOO J/mol
Ra= 8.314 J/g-mol'K
T = 573 K
0.1808/£expE-76000/(8.314)•(573)3•(0.84)0*5 •(0.06)0
and;
1.67 x IO^ nm/s'torr0
MONTANA STATE UNIVERSITY LIBRARIES
3 1762 10024276 5
DATE DUE
HIGH S M ITH REORDER #45-230
I