University of Nebraska - Lincoln
DigitalCommons@University of Nebraska - Lincoln
3!+(" 2(.-1 2(.- +$0.- 32("1 -#/ "$#,(-(120 2(.-
SiC and Si
3
N
4
recession due to SiO
2
scale volatility
under combustor conditions
James L. Smialek
NASA Glenn Research Center) ,$1+1,( +$*- 1 &.4
R. Craig Robinson
NASA Glenn Research Center
Elizabeth J. Opila
NASA Glenn Research Center
Dennis S. Fox
NASA Glenn Research Center
Nathan S. Jacobson
NASA Glenn Research Center
.++.52'(1 -# ##(2(.- +5.0*1 2 ':/#(&(2 +".,,.-13-+$#3- 1 /3!
9(102("+$(1!0.3&'22.7.3%.0%0$$ -#./$- ""$11!72'$ 2(.- +$0.- 32("1 -#/ "$#,(-(120 2(.- 2(&(2 +.,,.-1-(4$01(27.%
$!0 1* (-".+-2' 1!$$- ""$/2$#%.0(-"+31(.-(-3!+(" 2(.-1!7 - 32'.0(8$# #,(-(120 2.0.%(&(2 +.,,.-1-(4$01(27.%
$!0 1* (-".+-
,( +$* ,$1.!(-1.-0 (&/(+ +(8 !$2'.6$--(1 -# ".!1.- 2' -( -#(
0$"$11(.-#3$2.
(
1" +$4.+ 2(+(273-#$0".,!312.0".-#(2(.-1 NASA Publications
':/#(&(2 +".,,.-13-+$#3- 1 /3!
SiC and
Si3N4
recession due
to
SiO2
scale
volatility
under
combustor conditions
*
JAMES
L.
SMIALEK†,
R.
CRAIG
ROBINSON,
ELIZABETH J.
OPILA,
DENNIS S. FOX and
NATHAN
S. JACOBSON
NASA Lewis Research
Center,
21000
Brookpark
Road,
Cleveland OH
44135,
USA
Abstract-SiC and
Si3N4
materials
were
tested under various
turbine
engine
combustion
environ-
ments,
chosen to
represent
either conventional fuel-lean or fuel-rich mixtures
proposed
for
high
speed
aircraft.
Representative
CVD,
sintered,
and
composite
materials were evaluated in both furnace and
high pressure
burner
rig exposure.
While
protective SiO2
scales form in
all
cases,
evidence is
pre-
sented
to
support
paralinear
growth
kinetics,
i.e.
parabolic
growth
moderated
simultaneously by
linear
volatilization. The
volatility
rate is
dependent
on
temperature,
moisture
content,
system pressure,
and
gas velocity.
The burner tests were used to
map SiO2 volatility
(and
SiC
recession)
over a
range
of
temperature, pressure,
and
velocity.
The functional
dependency
of material recession
(volatility)
that
emerged
followed
the
form:
exp(-Q/RT)
* Px
*
vy. These
empirical
relations were
compared
to
rates
predicted
from
the
thermodynamics
of
volatile SiO and
SiOxHv
reaction
products
and
a
kinetic
model of diffusion
through
a
moving
boundary
layer.
For
typical
combustion
conditions,
recession of
0.2 to 2
µm/h
is
predicted
at
1200-1400°C,
far in
excess of
acceptable long
term
limits.
Keywords:
SiC;
Si3N4;
oxidation;
scale
volatility;
combustor;
water
vapor.
1. INTRODUCTION
SiC
composites
have been
proposed
as
liner
material
for
advanced combustors
in
turbine
engines. Operational pressures
are
in
the
vicinity
of
about
10 atm. Con-
ventional
(lean)
operation produces
a combustion
product
consisting
of
10%02-
8%H20-7%C02-bal. N2
at an
equivalence
ratio,
0,
of
0.5
[1].
Other
combus-
tor
concepts
use
a
rich-burn
pre-chamber
in which a
hypostoichiometric
mixture of
fuel-to-air
(q5
of about
1.5)
is burned.
A
6%H2-12%H20-12%CO-5%C02-bal.
N2
combustion
chemistry
is
projected
here.
The rich-burn
segment
is
followed
by
This document is a U.S. government work and
is not subject to copyright in the United States.
34
a
quick air-quench
and
lean
aft-bum
segment.
Volatile
reaction
products
between
the
Si02
scales and
the
combustion
gases
have
been
a concern on this
program.
It
was shown that
the
allied furnace
TGA
exposures produce
Si(OH)4
(g)
and SiO
(g),
respectively,
when
SiC
is
exposed
to model
lean
and
rich
gases
[2-4].
Although
a
scale is
first
produced by
oxidation,
it then
reacts
with the
gas
to form a volatile sec-
ondary product.
This
gives
rise to
paralinear
kinetics and accelerated
consumption
of the substrate
(recession)
[2].
This
represents
a
non-protective
oxidation
regime,
and
long
term
exposures produce
linear attack rates. These rates are determined
by
the
thermodynamics
of the
equilibrium vapor species,
the ambient
pressure,
and
the
gas velocity.
While
a
number of burner
rig
studies have
been
performed
on
SiC
[5],
only
one has
recently
combined
the use of
high
pressure
and
high
velocity
and
clearly produced
weight
loss
[6].
No
previous
studies have addressed the
effect
of
rich-bum
combustion.
Thus
the
purpose
of
this
study
was
to
examine the
behavior of
pure
CVD
SiC
under a
variety
of
high temperature, pressure,
and
velocity
conditions
in
both
lean-bum
and rich-burn combustion environments. This
was
accomplished
in a
high
pressure
burner
rig
(HPBR).
Some
experimental
results
are
covered
in
more
detail
in
previous
reports
[7, 8],
and chemical mechanisms are covered in
greater
depth
in
[9].
The
volatility
rates in HPBR tests are
compared
to those measured
in the furnace TGA tests and those calculated from the
thermodynamic
diffusion
model. Recession rates were also obtained on CVD and sintered
Si3N4
material for
comparison.
2. EXPERIMENTAL
High
purity
CVD
SiC
and
Si3N4
materials
(Bomas
Machine
Specialties
and
Advanced
Ceramics
Corp.)
were machined
to
0.3
x
1.3
x
2.5 cm TGA
specimens
or 0.3
x
1.3
x
7.6 cm HPBR
specimens.
TGA tests were
performed
with Cahn
1000 microbalances and
furnaces
using quartz
tubes.
Model
lean
exposures
employed
a
50%H20/02
mixture
to
approach
the
high
water
vapor pressure
of a
combustor environment
[2]. (A
moisture content
of
8%
x
10
atm,
or
80%,
would
be
more
representative
of the actual combustor
conditions.)
A
special
double-
chamber
water
saturator,
followed
by
tape-heated
gas
lines,
was
employed
to
insure
accurate
control of the moisture content. The
gas
flow was 4.4
cm/s.
Model
rich
exposures
were made in a
premixed gas flowing through
a similar saturator to
give
4%H2-12%H20-10%CO-7%C02-N2,
flowing
at 0.44 cm/s
[4].
An
existing high pressure
burner
rig
was
extensively
modified to allow both lean
and rich-burn
testing
of
multiple
ceramic
specimens (Fig.
1).
Details of the con-
struction and
operation
can be found
in
[7, 8].
The
basic
operation
entails
pres-
surized fuel and air
injection
(in
an air blast nozzle and swirl
plate
dome
section),
ignition (by hydrogen gas
at a
spark plug),
and combustion
product
formation
(in
an
air-cooled,
TBC-coated
combustor
can).
The
gaseous
combustion
products proceed
through
a water-cooled transition
section,
losing
heat in the
process. Specimens
35
Figure
1. Schematic
diagram
of NASA
Lewis
High
Pressure
Burner
Rig.
are
arranged
in a
wedge configuration
in a water-cooled
specimen
holder,
pneu-
matically
actuated into or out
of
the
flowing gas.
Gas
temperature
is measured
by
a Pt-Ptl3Rh
thermocouple
at a
position just
behind the
samples. Sample
tem-
perature
was measured
by
two-color
optical
and laser
pyrometry
(lean-burn
only).
Rich-burn
temperatures
were determined from a
thermocouple
measurement of
gas
temperature
and
specimen temperature
calibration
curve,
as determined in the lean
mode.
This was
necessary
because the luminous flame in rich
operation
precluded
measurements
by optical pyrometry.
It
was intended to determine
volatility
rates over a
range
of
temperature, pressure,
and
velocity.
To
this
end,
some
flexibility
in
operational
parameters
exists; however,
an
interdependence
between variables resulted
in
certain restrictions. Nominal fuel-
lean
combustion
at $
=
0.8-0.9 and fuel-rich
at 0
=
1.8-2.0
produced
sample
temperatures
from 1200-1450°C. Standard
operating pressure
and
velocity
was
6 atm and 18-25
m/s,
respectively.
For a select number of
tests,
pressure
was also
varied
from 4-15
atm,
while
velocity
for the most
part
was a
dependent
variable.
3. RESULTS
3.1. Furnace tests
In model lean furnace
tests,
mixtures of
10%H20/02
failed to
produce appreciable
differences
in
kinetics from those
performed
in
dry
02
environments.
However,
at
50%H20/02,
which would be more
representative
of a moist environment at
high
pressures,
a
negative
(paralinear)
deviation from
parabolic
kinetics was observed.
Analysis
of
weight
change
data
yields
both a
parabolic growth
constant,
kp,
and
a
linear
recession
rates
[2].
A
typical
curve and fitted
parameters
for CVD
SiC
36
Figure
2. Paralinear
weight change
curves for CVD SiC in
synthetic
lean
50%H20/02
furnace TGA
environment at 1200°C.
The model curves indicate the amount
of
oxygen gain
in the scale
present,
the amount of silicon
(and
carbon)
lost
due
to scale
volatility,
and
the
net
weight change.
oxidized in
50%H20/02 (lean)
at 1200°C is
given
in
Fig.
2. The
actual
(sawtooth)
data and fitted
'net
weight change'
(smooth)
curves are coincident and
provide
a
high degree
of confidence for
the
mathematical
model. The model
'weight gain'
curve is indicative of the amount of scale
present
on the
sample,
while the model
'weight
loss'
curve indicates the amount of Si
(and C)
lost
during
oxidation
and
Si02
scale volatilization.
There was
only
a
slight
temperature
dependence
of kp
and
k,
in lean furnace tests with water
vapor
[2, 11].
A
strong temperature dependence
of
SiC recession
due
to
Si02
volatility
was
produced
in
synthetic
rich
gas
mixtures,
as shown
in
the
composite
TGA curves
of
Fig.
3.
Here
very
little
volatility
could be detected below 1350°C
using
the same
paralinear
model
and
data
analysis
[4].
However,
at 1400°C
and
above,
the
volatility
rates were
appreciable.
A
comparison
of the SiC
weight
loss rates in lean and rich furnace test environ-
ments
is
shown
in
the
Arrhenius
plot
of
Fig.
4.
A
higher temperature dependency
is observed for rich environments. It should be noted that the fuel-rich tests were
performed
at
1/10
the
gas
flow
of
the
fuel-lean
exposures.
Also,
X-ray
diffraction
and SEM
confirmed that a continuous cristobalite scale
was
indeed
present
under
all furnace
exposures
and that active oxidation was not
responsible
for these
weight
losses.