177
Thermal Properties of High Alumina Refractory Mortars Modified With
Lanthanide Chemical Admixtures
THOMAS NOVINSON Naval Civil
Engineering
LabPort
Hueneme, California
HEINZ F. POPPENDIEK Geoscience Ltd Solana
Beach, California
(Received September 16, 1986) (Revised November 17, 1986)
INTRODUCTION
I T
IS OF interest tostudy
the thermalinsulating properties
of certainchemically
modified high alumina refractory
mortars because of their potential
use as heat
resistant
insulating
liners forportland
cement concrete structuresexposed
to heatin the range from 500 to 1000 ° F
Conventional alumina refractories are heat cured in
place,
inengineering
ap-plications [1].
This is not aproblem
forrefractory
linersprepared by gunning
onto the interior steel surface of a furnace or a kiln. For
lining
pcc,however,
thisis a
problem
since flame heatcuring
of therefractory
liner islikely
todamage
theunderlying
concrete meant to beprotected.
For this reason, a
variety
of chemicalmodifiers,
known as &dquo;admixtures&dquo; have beeninvestigated
to lower the curetemperatures
of conventional refractories.Among
the chemicalsreported
in this paper are various derivatives of the lanthanideelements, lanthanum, cerium, neodymium
andpraeseodymium, plus
the
closely
related but non-lanthanide element zirconium. The lanthanide com-pounds
were chosen forpotential
fluxproperties,
thatis, ability
to lower thefusion
point
of alumina[2].
The unmodified controls and modified
(chemical admixtures)
refractories wereheat cured and tested for cold
crushing strength following
the ASTM C133-82aprocedure [3].
Plots ofstrength
vs.temperature
were made to determinephysical
behavior at elevated
temperatures.
Then thermalproperties, including
thermalconductivity, specific
heat and thermaldiffusivity,
weremeasured,
to ensure thatinsulating properties
had notchanged
for the modified materials.Journal
of
COMPOSITEMATERIALS,
Vol.22 -February
19880021-9983/88/02 0177-15 $4.50/0
178
EXPERIMENTAL
Crushing Strength
MeasurementsMortars were
prepared
from 1000 gm ofhigh
alumina calcium aluminate cementcontaining
bauxiteaggregate, plus
150 ml of water(15
%by weight)
con-taining
0(control,
noadmixture), 1 % , 5 %
and 10 %by weight
of each chemical admixture. Therefractory
mortar was cast in theshape
of 2by
2by
2 inch cubes.The cubes were moist cured 24 hours at
95 % humidity
and 75 ° R Then the moldswere
stripped
and the cubes wereslowly
baked in an oven at 220°F to removewater. The
cubes,
which were allowed to cool to roomtemperature,
were then heated in sets ofthree,
totemperatures
of500,
750 and 1000°F The cubes wereagain
cooled to roomtemperature
and were crushed with a Tinnius Olsen test machine. The coldcrushing strengths
were then recorded. The results can beseen in
Figures
1 and 2.Thermal
Conductivity
DeterminationsThe thermal
conductivity
is definedby
the Fourier’sequation,
where
q = heat flow rate k = thermal
conductivity
t =
temperature
y = distance
along
heat flowpath
A = heat transfer area normal to the heat flow
The
guarded
hotplate apparatus (ASTM C-177) [4]
was used to make the ther- malconductivity
measurements.Figure
3 illustrates themajor
elements of thesystem.
Heat from acentral,
flatplate
heater flowsthrough
two hotplates, through
two test slabs to twocooling plates.
Thetemperature
differences acrossthe two test
samples
are determined fromthermocouples
that are embedded in the hot and coldplates
or in the testsamples.
As shown in thefigure,
the central heater is surroundedby
aguard
heater. Athermopile
ispositioned
between the central hotplates
and theguard
hotplates.
Atsteady
state, when the central heater andguard
heater have beenadjusted
so that thethermopile output
is very small(yielding
unidirectional heatflow),
a data set is obtained. From a heat balance on thesystem involving
heater powerinputs,
heat transfer areas, tem-perature
differences andsample thicknesses,
the thermalconductivity
can bedetermined.
Specific
Heat DeterminationsOne convenient way of
measuring
thespecific
heat of cementitious materials isby utilizing
agradient layer
calorimeter. The elements of such asystem
are shown179 Figure 1. Comparison of crushing strengths of unmodified alumina refractory and
refractory
doped with 1 % lanthanide compounds.180
Figure 2. Comparison of crushing strength of alumina refractories with 1 % lanthanum
phosphate
and cerium oxide.181
Figure
3. Guarded hot plate (twin plate) thermal conductivity system.in
Figure
4. A testsample
ispositioned
inside of the heat fluxmeasuring envelope
that is housed in a fluid cooled heat sink. The heat sink in turn is sur-rounded
by
thermal insulation.This calorimeter is based on the
principle
that all of the heat flow into or outof the calorimeter must pass
through
its walls where thetemperature gradient
sensors are located.
Therefore,
the calorimeterenvelope integrates
the total heat flow in thesystem
on an instantaneous basis. The calorimeter walls consist ofspecial
thermoelectric heat flux transducers* thatyield
a DCvoltage output
*A heat flux transducer is composed of a thermopile system that has &dquo;hot&dquo; junction sets at one depth within the
sensor and &dquo;cold&dquo; junction sets at another depth
182
signal.
The calorimeter walls are thin so that low time constants are involved. As heat flowsthrough
thewalls,
a smalltemperature
differencedirectly proportional
to the heat flow is
developed.
In aproperly designed calorimeter,
theoutput sig-
nal is affected
only by
the rate of the heat flow.The utilization of the calorimeter to measure the
specific
heat of a material is done as follows: An unloaded calorimeter that has come to thermalequilibrium
at
temperature level t1
issuddenly exposed
to a newtemperature level,
t2. The heat flow trace as the transient processproceeds
from the initialsteady
state con-dition to the final
steady
state condition can be obtained with arecording poten-
tiometer. The area under this curve isequal
to the heat added to or extracted from the calorimeter and its liner for thesuperposed temperature perturbation. Next,
the calorimeter is loaded with thespecimen
to beinvestigated
andagain exposed
to the
original temperature tl
and allowed toequilibrate.
Then the loaded calorimeter issuddenly exposed
to the newtemperature
datum t2, allowed to reachequilibrium
and thecorresponding
transient heat flow trace recorded. Thearea under this curve is
equal
to the heat added to orgiven
upby
thecalorimeter,
its liner and thespecimen.
Subtraction of the areas under the two transient tracesyields
the desired heat flow for thespecimen
alone. The trace area is related to an energy flow per unit timeby
means of an accurate resistanceheating
calibra-tion for the calorimeter. The
specific
heat for thespecimen
is obtained from theFigure
4. Elements of a gradient layer calorimeter system.183
Figure 5. Thermal diffusivity measurement apparatus.
classical definition
(heat
transferred dividedby
theproduct
of the mass and tem-perature perturbation).
Thedefining equation
for thespecific heat, c,.,,
iswhere
q(O)
= timedependent
heat flowthrough
calorimeter when loaded with testsample
q(6)e =
timedependent
heat flowthrough
calorimeter without the testsample
0= time
0,=
theequilibrium
timeperiod (no
further heatflow)
ms = mass of the test
sample
tl = initial
temperature
datumt2 = final temperature
datum ThermalDiffusivity
DeterminationsIf it is desired to measure the thermal
diffusivity directly (rather
thanby
mak-ing
the individualproperty
measurements andsubstituting
them into thedefining
184
property ratio),
it is necessary to work with atime-temperature
solution of a clas- sicalboundary
valueproblem
where transient heat transfer is inoperation [5].
A
typical example
of such asystem
is a slab that issuddenly
cooled at both sur-faces
by
forced convectioncooling. Symbolically,
the functionalrelationship
ofthe variable can be
expressed
as,where tc, to
and t,
are thecenterline,
surface and initialtemperatures,
respec-tively,
yc, is theproduct
of thespecific
heat and thedensity
and h is the surface heat transfer conductance.The surfaces and centerline
temperatures
of the slab are measured as a function of timeduring
thestep
functioncooling (see Figure 5).
From such data and the mathematicaltime-temperature
solution for thisboundary
valueproblem,
thethermal
diffusivity
can be deduced.Figure
6. Refractory mortar at 100 x magnification. F = flintclay,
M = matrix area,P = pore.
185
Figure 7. Refractory mortar with 1 % cerium oxide at 100 x magnification. F = flint clay,
M = matrix area, P = casting pore, C = cerium oxide.
RESULTS AND CONCLUSIONS
Representative samples
of the control(refractory
withoutadmixture)
and 1%cerium oxide
doped
mortar cubes were vacuumimpregnated
with an epoxypot- ting compound.
The cured cubes were then machinepolished,
first with diamondgrit
down to 1 ¡.tm, then with aluminagrit
down to 0.05 ¡.tm. Thesamples
werethen viewed at several different
magnifications (40X, 100X,
and200X)
for thepetrographic analysis. Photomicrographs
were taken with a Polaroid camera at- tachment. Photos at 100X are included in this paper(Figures
6 and7)
to illustratemajor findings.
Samples
of 8refractory
mortarcubes, including
one control(no admixture)
and one
1 %
cerium oxidedoped
mortar, were examined underhigh magnification using
an ISI(International
ScientificInstruments) scanning
electronmicroscope (SEM).
Thesamples
werechipped
off the corners of therefractory
mortarcubes,
mounted on carbon blocks with aspecial graphite adhesive,
andplaced
in ananalysis
chamber. Thesamples
weresputtered
withgold
forhigh
resolution. The chamber wasplaced
under vacuum and thesamples
were scanned with x-rays for the SEMpictures (recorded
with a Polaroid cameraattachment)
shown inFigures
8 and 9.
Samples
of the control(without admixture)
and 1 % cerium oxidedoped
refrac-tory
mortar cubes werepulverized
to - 325 mesh andchemically analyzed using
x-ray diffraction
equipment (Kevex).
Chemicalanalysis
of the mineralphases
of186
Figure 8. Refractory mortar before finng (75°F), at 500 x magnification.
Figure 9. Refractory mortar after firing (to 700°F) and cooling, at 500 x magnification.
187 the mortars verified the existance of cristobalite
(Si02),
mullite(AI6Si2013)’
andgehlenite (Ca,A1,SiO,), plus
some calcium aluminumhydroxides
that could notbe identified. The mineral cerianite
(Ce02)
was also identified in the 1 % cerium oxidedoped samples.
No other ceriumcompounds
were identified in thedoped refractory.
The cerium oxide did not
chemically
react with thecomponents
of the cements, since inclusions of cerianite were theonly
identifiable cerium com-pounds
observed in the x-ray diffractionexperiments. However,
thepetrographic analysis
at 100Xmagnification (Figures
6 and7) definitely
show a difference incasting density,
thatis,
the cerium oxidedoped samples
are denser than theundoped
controlrefractory
cubes.Perhaps
cerium oxide acts as asuperior
fineaggregate
toreplace part
of the aluminum oxide(bauxite). Unfortunately,
asimilar
study
on the lanthanum oxide could not beperformed
due to time andfunding
limitations.Lanthanum
phosphate
mayactually undergo
chemical reaction with some of the other mortarcomponents, however,
because(a)
this salt is somewhat watersoluble
compared
to the insoluble lanthanide oxides and(b)
the curve forstrength
vs.
temperature (Figure 1)
for lanthanumphosphate
is muchsteeper (hence greater strength)
than that for the lanthanum oxidedoped refractory
mortar.Another reason for the observed
strength
increases may be due tophysical changes
in therefractory
microstructure causedby
the lanthanidecompounds.
Figure
8 shows the surface of the aluminarefractory
at 500Xmagnification
before
firing.
The surface appears like a series ofpottery
cups or vases. After fir-ing
at 700degrees,
the surfacechanges,
as illustrated inFigure 9,
and thefrag-
ments appears to have melted or
annealed,
thusforming
a denser material. The cerium oxide may have aided in the densification of the material at thistempera-
ture,although
it is unclear as towhy
thestrength
decreases between 700 and 1000degrees.
It should be noted that thepeak
instrength
isreproducible
and wasobserved in
laboratory experiments
both at NCEL and at NationalRefractory
andMinerals
Corp.
Thisstrange
increase instrength
at 700degrees
was also notedwith
refractory
cubescontaining
1 % of certainzinc,
barium and strontium salts(results
to bepublished
at a laterdate).
Refractory
mortarscontaining
1 % cerium oxide exhibited thegreatest
coldcrushing strength throughout
the range of 500 to1000 ° F,
with an unusualpeak
or increase in test
strength occurring
at 750 ° F. Lanthanumphosphate (1 % )
alsoproduced
a substantial increase instrength
for therefractory
mortar in the range of 500 to 1000 ° F The other lanthanideoxides, neodymium
andpraseodymium,
had little or no effect or
actually
decreased thestrength
of the heated mortar. The zirconium oxide showedonly slight
increases instrength
when added in1 % ,
5 %and 10 %
by weight
increments to the calcium aluminate mortars.The thermal
conductivity, specific
heat and thermaldiffusivity
vs. densities arerecorded in Tables
1,
2 and 3. In the case of thermalconductivity,
the datapoints
shown in Table 1 indicate that this
property
increasesslightly
with increases inweight
concentration of the mortaradditive,
as shown forZrSiO3.
The thermal conductivities of solid materials are controlled
by
twoprimary
mechanisms, namely,
lattice vibrations(atomic motions)
of the material and free188
Table 1. Thermal
conductivity
measurements for mortar with various additives.Table 2.
Specific
heat measurements for mortar with various additives.Table 3. Thermal
diffusivity
measurements for mortar with various additives.189 electron flow or the electronic
component.
In the case of cementitioustype materials,
thermalconductivity
isprimarily
controlledby
the lattice vibration mechanism.Generally,
one cannotpredict
thermal conductivities for such materials from firstprinciples.
It ispossible,
however topredict
theequivalent
thermal
conductivity
of a mixture ofcomponents
if their thermal conductivities and volume orweight
fractions are known. Thefollowing equations
illustratehow such
predictions
can be made.If the
components
of a mixture werearranged
in &dquo;n&dquo;adjacent
slabs with heat flowalong
thelengths
of theslabs,
the classicalexpression
for the effective orequivalent
thermalconductivity, keq,
iswhere
~y =
density
of thecomposite
materialkn
= thermalconductivity
of the nthcomponent
wn =
weight
fraction of the nthcomponent
yn =
density
of the nthcomponent
If the
components
of such a mixture werearranged
inadjacent
slabs with heat flowperpendicular
to thelengths
of theslabs,
the classicalexpression
for theequivalent
thermalconductivity
isFor the case of one
component (small particles) uniformily
distributed in asecond
component (the carrier)
theequivalent
thermalconductivity (as proposed by Eucken) is
where
k~
= thermalconductivity
of the carrierkp
= thermalconductivity
of theparticles
b = ratio of
particle
volume to total volumea =
3/(2
+kpkc)
The Eucken
equation
is valid in the range 0 < b < 0.5.190
On the basis of the small
percentages
of admixturespresent
and on the basis that the thermal conductivities of thecomponents
are all of the sameorder,
it canreadily
be shown from the threeequations
shown above thatonly
smallchanges
in thermal
conductivity
occur(as
foundexperimentally)
for the mortars contain-ing
the admixtures used in thisstudy.
The
specific
heat measurements for the mortarscontaining Zr02
andZrSiO3
illustrate the fact that theproduct
of thespecific
heat and thedensity
is approx-imately
a constant.From the classical oscillator
theory,
it can be shown that thespecific
heat of amaterial is relatable to the number of
degrees
of freedom of the molecule in- volved and the universal gas constant. Such anexpression
can be used to estimatethe
specific
heat of some materials.Dulong
and Petit relatedspecific
heats ofmetal to their atomic
weights.
Morecomplicated
materials like the cementitious materialsbeing
considered in thisstudy
are more difficult to evaluate on firstprinciple
bases. It ispossible, however,
to describe theequivalent specific heats,
c, , of such materials in terms of the
specific
heats of the individualcomponents
and
theirweight fractions, namely
where
w&dquo; = the
weight
fraction of the nthcomponent
C’n = thespecific
heat of the nth componentCalculations can be made for the
specific
heat of a mortarcontaining
an admix-ture,
using
the mixtureequation given
above.Again,
because of the small admix- turepercentages involved,
one can show that thepredicted changes
inspecific
heat are small as a result of the admixture
additions,
as foundexperimentally.
The thermal diffusivities were found to be very
nearly
a constantquantity
forall admixtures measured. This result is
apparently
related tocorresponding changes
in thermalconductivity
and theproduct
ofdensity
andspecific
heat asthe
compositions change.
It has been shown that small
percentages
of admixtures whichsignificantly
influence the
strength
ofrefractory
mortars, did notsignificantly
affect the thermalproperties.
Analysis
of these data indicates that the refractoriesdoped
with1 %
lanthanideoxides differ little from the unmodified
refractory
mortar control. This informa- tion ispromising
in that it demonstrates thatinsulating properties
of the modified mortars were not lost in those cases where the lanthanide oxidesimparted
in-creased
strength
to the heated refractories.ACKNOWLEDGEMENT
Funds for this work were
provided by
the Naval FacilitiesEngineering
Command, Alexandria,
and the Office of NavalResearch, Arlington, Virginia.
191 The authors wish to thank the materials science
engineering
technicians at NCEL for assistance inpreparing
the mortarcubes, plates
and otherspecimens,
theceramic
engineering
research staff of National Refractories andMinerals, Pleasanton, California,
for modulus ofrupture (cold crushing) data,
andGeoscience staff members for their assistance in
making
the thermalproperty
measurements.
REFERENCES
1. American Concrete Institute "Refractory Concrete," Special Publication SP-57, Detroit, Mich.
(1978).
2. "The Lanthanides," Chap 31, pp. 870-891 in Advanced Inorganic Chemistry, A Comprehensive Text, F. A. Cotton and G. Wilkinson, eds. New York, NY:Interscience Publishers (1962).
3. American Society for Testing and Materials, Philadelphia, PA (1982).
4. Annual Book of ASTM Standards, Part 18, Philadelphia, PA (1981).
5. "Heat Transfer Notes," L. M. K. Boelter, V. H. Cherry, H. A. Johnson and R. C. Martinelli, eds.
Chapter 1, University of California Press, Berkeley and Los Angeles (1946).