Title: M' Meyyappan
1SWNTs by PECVD Feasibility and Issues
M. Meyyappan NASA Ames Research Center Moffett
Field, CA 94035 Email mmeyyappan_at_mail.arc.nasa.go
v
Acknowledgements Alan Cassell, Brett Cruden,
Lance Delzeit, David Hash, and Ken Teo
2Outline
Various plasma sources Why use plasma in
nanotube growth? Growth results MWNTs
MWNFs Modeling and diagnostics Growth
Results SWNTs Summary of issues in PECVD
growth
3Various Plasma Sources
Classification is primarily based on how power
is supplied to create the plasma all of these
have been reported in the CNT growth
literature DC discharge (electrodes, gt 500
mTorr, inefficient, no longer in use in IC
industry) Radiofrequency (RF) capacitively
coupled - 13.56 MHz is the primary frequency
allowed by FCC - gt 300 mTorr, very popular in
IC industry Microwave discharge - 2.45
GHz is the allowed frequency - Low to high
pressure - Very popular in diamond
deposition Inductive discharge - Simple,
high plasma density compared to capacitive
discharges - Common in the U.S. in IC industry
4PECVD Set-up
5Conditions for SWNT Growth by Catalytic CVD
Theoretical analysis by Kanzow and Ding
(Phys. Rev. B, Vol. 60, 1999) shows that higher
temperatures (900 C and above) and low
supply of carbon favor SWNT growth the
opposite favors MWNTs - High kinetic energies
should be available in the system
(proportional to temperature) to bend the
graphitic plane
6Why Plasma in Nanotube Growth?
Plasma CVD is a variation of the
well-established thermal CVD, offers additional
control parameters such as power, substrate bias
to exert influence on growth if
possible Typically PECVD enables a lower
temperature operation than thermal CVD since
electrons are energetic and lead to the
production of reactive species from the
precursors. Does conventional wisdom of plasma
processing apply here? - CNT growth is catalyst
promoted (VLS mechanism) - Precursor
dissociation on catalyst surface, carbon
dissolution and diffusion, are more important
than gas phase dissociation. - All of the above
require temps. gt 500º C - So, traditional cold
plasma with very low substrate temperatures
may not be possible.
7What is the Role of the Plasma?
Can the reactive radicals produced in the
plasma lead to CNT growth at a lower
temperature? (Their surface reaction
temperature to produce C may be lower than
that for the parent hydrocarbons) Excessive
production of C-bearing species in the plasma may
lead to amorphous carbon contamination. One
recognized advantage is vertical alignment of
nanotubes due to the electric field. Keeping
the C-supply very low, as suggested by Kanzow and
Ding for SWNT growth, is hard
8MWNTs vs. MWNFs
Filaments or fibers consist of stacked-cone
arrangement of graphite basal plane sheets
grow with particles at the tip hydrogen is
believed to satisfy the valences at cone edges
in filaments The orientation angle ? (between
graphite basal planes and tube axis) increases
with increasing hydrogen concentration When ?
0, MWNTs. These are filaments with no graphite
edges, requiring no valence-satisfying species
such as hydrogen Nolan et al provide evidence
(all thermal CVD) that material produced with
CO disproportionation (without any H2) was
MWNTs addition of H2 produced filaments as
of H2 , ? up to 30 Better to call
these structures as MWNFs instead of graphitic
carbon fibers (GCFs) or vapor grown carbon
fibers (VGCFs) both of which denote solid
cylinders
Nolan et al, JPC, B, 1998
9Multiwalled-Nanotubes using an Inductively
Coupled Plasma
Feedstock Ar/H2/CH4 110/20/50 sccm 100 W
inductive power, 20 W capacitive power with -11 V
bias 3 Torr, 900? C Catalyst 1 nm Fe on 5
nm Al
10MWNFs using an Inductively Coupled Plasma
3 Torr, 800? C Methane/H2 20/80 sccm 50
W inductive power 70 W capacitive power
MWNFs invariably are individual, free-standing,
vertically aligned structures in contrast to the
MWNTs.
11MWNTs vs. MWNFs
- TEM Image of MWNTs
- Plasma grown MWNTs
- Base growth
- TEM Image of MWNFs
- Plasma grown MWNFs
- Tip growth
12Raman Spectra at 633 nm Excitation Plasma CVD
Samples
MWNF
D band centered at 1350 cm-1 Tangential G
band at 1590 cm-1 Shoulder peak around 1616
cm-1 only in MWNFs
13Emission Spectroscopy of the Plasma
Emission spectroscopy of plasma, CH
(0, 0) band near 430 nm atomic H
Balmer series (? peak at 680 nm and ? peak at
486 nm) several H2 peaks (strongest at 464
nm) MWNT growth is accompanied by a low peak
intensity of atomic hydrogen. At high
intensities ? MWNFs As power to substrate ?,
MWNFs result may be due to increased
dissociation of H2 from increased ne, Te at
fixed inductive power and pressure Increased
dilution with argon reduces H concentration, as
evident from the intensity, which coincides
with MWNT production
Delzeit et al, JAP (2002), 91, 6027
14Controlled CNT Growth from Stalk Templates
DC-Plasma
Carbon Nanotube Growth Lengths Achievable (1-15
µm), Diameters 50-100 nm
5 ?m
5 ?m
Cassell et al., Nanotechnology, Vol. 15, pp. 9-15
(2004)
1 ?m
2 ?m
15Screening Catalyst Activity of Metals With Good
Work Function Match Using PECVD Growth
Multiwalled CNTs 4.6-5 eV
W
Ti
Ta
Ir
Cr
Co
Fe
Ni
Fe/Ni
Ni/Co
Underlying Contact Metals
Si wafer
Catalyst Layers
Best Combinations
16Carbon Nanotube Growth Morphology Control
Pure Bamboo
Mixed Bamboo/MWNT
Pure MWNT
Bamboo is well-aligned, but not as crystalline
as MWNTs.
TEM of Bamboo type CNTs TEM of MWNTs
170-d Model of Nanotube Growth Plasma
Volume-averaged model (neutral and ionic
species balance, electron density and energy
balances) for high density plasmas 48 species,
563 reactions including electron impact,
electron-ion, ion-ion, ion-neutral,
neutral-neutral reactions Densities (in cm-3)
3 Torr, 800 C, 100 W inductive power, 20/80
methane/H2 at 100 sccm Total density 2.7 x
1016 cm-3
18Species in the Plasma Model Predictions
-
Density, cm-3
- Densities of feedstock gases increase with
pressure - Most radical densities decrease with an
increase in pressure due to a decrease in both Te
and ne - Density of stable species (CH4, C2H2, C2H4,
C2H6) increases with pressure, which is
responsible to maintain reasonable growth rate - Hash and Meyyappan, Journal of Applied Physics,
93, 750 (2003).
19Quick Summary from (MWNT, MWNF) Growth, Reactor
Diagnostics and Modeling
CNF growth is accompanied by increasing H
intensity in the plasma Low bias voltages or
power to the substrate result in MWNT
growth Increasing the substrate voltage/power
results in CNFs. So, it is not surprising that
all dc PECVD papers report CNF growth only, with
their voltages exceeding 500 V There are
plenty of radicals in the plasma phase, besides
the parent hydrocarbon, all of which can be the
C-source for CNT growth. This is in contrast
to thermal CVD where there is barely any
dissociation of the hydrocarbon in the gas
phase since the temperature is normally
maintained below the pyrolysis temperature of
the hydrocarbon.
20Some Recent SWNT Growth by PECVD
Prof. Hatakeyama Group, Tohoku University
Radiofrequency plasma substrate not biased
self-potential 12V Substrate located below
the bottom electrode Reported T 750 C
(bottom side, not the wafer) CH4/H2 3/7 -
This plasma dissociation efficiency is too low.
Most likely CH4 dissociation on the surface.
21Some Recent SWNT Growth by PECVD
Prof. Tim Fishers Group at Purdue
Microwave plasma, 200 W MW Power, -250 V dc
bias maximum External heating 900C, 5 sccm of
methane in 50 sccm H2 (low flow) Raman
Spectra (533 nm 785 nm) confirm
SWNTs Examination of RBMs correlating with dc
bias, SWNT diameter ? with dc bias at -250 V d
is as big as 2.41 nm
22Challenges and Future Directions
What are the species that are responsible for
nanotube growth? What is the role of atomic
hydrogen? Are the radicals deleterious? Do
they lead to amorphous carbon
contamination? Is amorphous carbon
preferentially etched away compared to
nanotubes? Is there a specific role for
ions related to growth? Do they weaken the
particle adhesion to the surface? If so,
what is the dependence on ion energy? Is
there a preferred hydrogen-carrying diluent (H2
vs. NH3)? What is the role of other diluents
such as argon or nitrogen? Meyyappan et al.,
PSST (2003) 12, 205.
23Challenges and Future Directions (Cont.)
Catalyst effects in terms of the transition
metal choice, method of depositing the
catalyst, layer thickness, pretreatment if any,
particle creation, effect of particle size
on nanotube diameter and growth
rate. Can disorders be annealed
away? PECVD grown MWNTs vs. MWNFs vs. SWNTs
parameters dictating this choice. Rate
determining step(s) in CNT growth. Tip vs.
base growth effect process parameters
Alignment mechanism How does electric
field influence growth orientation and resulting
alignment?
24Challenges and Future Directions (Cont.)
Is there a preferred substrate heating method
(resistive heating, hot filament, IR
lamp)? How low a growth temperature is
possible? This is critical for large area flat
panels on glass substrates. Is a large
dc bias appropriate? Does it damage the CNT
structure? What is the effect of the
substrate bias on growth rate, structure, and
alignment? Why is the sub-100m Torr
operation common in IC manufacturing not that
common in PECVD of CNTs? What is the effect of
pressure? Effect of other process
parameters? Is it possible to obtain growth
uniformities over large areas common in IC
manufacturing? What are the advantages of
remote plasma operation?
25Plasma Heating Effect in CNT Growth
Just because there is no external heating, does
not mean room temperature growth. Plasma does
heat the substrate (especially at 600 V in a dc
plasma!)
26Summary
PECVD approach seems suitable to grow
CNT-structures if individually-separated,
free-standing, vertically-aligned structures are
desirable. Free-standing, vertical structures
happen to be MWNFs MWNTs do not seem to enjoy
the same level of vertical orientation or
isolation from the neighbors A variety of
plasma sources has been used to date to
demonstrate interesting growth
results Current understanding of growth
mechanisms is very poor and the field needs
diagnostics and modeling efforts. Just a
couple of recent demonstrations of SWNT growth.
Lot more work is needed to establish true
feasibility.