chapter 7 study of the thermal properties of mullite...
TRANSCRIPT
129
CHAPTER 7
STUDY OF THE THERMAL PROPERTIES OF
MULLITE COATINGS
7.1 INTRODUCTION
Ceramics are selected for many design applications requiring high
temperature exposure of parts and assemblies, due to their high heat capacity,
low thermal conductivity and coefficient of thermal expansion and high
thermal shock resistance. These properties are generally better in the case of
ceramics than in metals. In this chapter, thermal properties, namely the
thermal conductivity and thermal shock resistance of mullite coated cast
aluminum A 356.0 are measured and comparison with the literature values of
other commercially available ceramic materials is made. The following
thermal properties of ceramic materials are important in various design
considerations ( Dmitri Kopeliovich 2010):
1. Thermal conductivity
2. Thermal expansion
3. Heat capacity
4. Thermal shock resistance
5. Maximum service temperature
Comparison with the thermal shock resistance property of duplex
coated and functionally graded YSZ coatings and with aluminum oxide
coatings, have been made.
130
Comparison of the thermal conductivity of mullite coatings, YSZ
coatings, alumina PEO coatings and mullite rich PEO coatings from literature
reports has been done.
7.1.1 Thermal Conductivity
Thermal Conductivity (k) is amount of heat passing in unit time
through unit surface in a direction normal to this surface when this transfer is
driven by unit temperature gradient under steady state conditions. Thermal
conductivity may be expressed and calculated from the Fourier’s law.
Q/ t = k*a * T/ x (7.1)
where, Q -heat, passing through the surface A; t - change in time;
k - Thermal conductivity; a - Surface area, normal to the heat transfer direction;
T/ x- temperature gradient along x – direction of the heat transfer.
Fourier’s law is analogous to Fick’s first law, describing diffusion
in steady state. Ceramics have low thermal conductivity compared to metals,
due to Ionic-Covalent bonding in ceramics which does not form free
electrons. For example, k of alumina = 6.3 W/m*K and k of aluminum
=231W/m*K.
7.1.2 Coefficient of Thermal Expansion
Thermal Expansion (Coefficient of Thermal Expansion) is defined
as the relative increase in length per unit temperature rise.
= L/ (Lo* T) (7.2)
where, -coefficient of thermal expansion (CTE); L – length increase;
Lo – initial length; T – temperature rise. Thermal expansion of ceramic
131
materials is generally lower, than that of metals. For example, CTE of
SiC = 4.0ºC ¹, CTE of Al = 23ºC ¹.
7.1.3 Specific Heat Capacity
Heat capacity is amount of heat required to raise material
temperature by one unit. Specific heat capacity is amount of heat required to
raise temperature of unit mass of material by one unit.
S= Q/ (m * T) (7.3)
where, S -specific heat capacity; Q – amount of heat; m – Material mass; T
– temperature rise. Specific heat capacity of ceramic materials is higher, than
that of metals. For example, “S” of alumina = 850 J/ kg*K. “S” of steel =
481 J/ kg*K.
7.1.4 Thermal Shock Resistance
Thermal Shock Resistance is an ability of material to withstand
sharp changes in temperature. If a ceramic material is rapidly cooled, its
surface reaches the temperature of cooling environment and tends to contract
(thermal contraction). Since the interior regions of the material are still hot,
thermal contraction of the skin surface is impossible. This leads to formation
of tensile stress (thermal stress) in the skin. Such thermal stresses may cause
cracks and consequent failure, due to the brittle nature of ceramics. Thermal
shock resistance of a material may be estimated in accordance to the formula:
Rs = (k* F)/ ( *E) (7.4)
where, Rs – Thermal shock resistance; k - Thermal conductivity; F – flexural
strength; -coefficient of thermal expansion (CTE); E – Modulus of
elasticity.
132
Sensitivity of ceramic materials to thermal shock may be also
determined by experimental method (Hasselmann Method) for bulk materials.
In this method a specimen (flexural test specimen) is heated to a specified
temperature and then quenched. The specimen cools rapidly by temperature
T (the difference between the specimen temperature before and after
cooling). After quenching the flexural strength of the quenched material is
measured by standard flexure (bending) test. The test results are plotted on the
graph Strength vs T. When T reaches a certain value the specimen strength
falls sharply. This value of T is a parameter indicating thermal shock
resistance of the material. Test pieces (in the simplest case these are bending
bars) are quenched to drop them from a temperature T1 to a temperature T2.
The strength of the samples is measured after the quenching. The curve of
strength against the temperature difference, T = T1- T2, has the shape shown
on figure 7.1. Up to a temperature difference of Tc the strength does not
alter. The strength then drops sharply within a narrow range, T. Up to Tc’
this reduced length then remains constant, falling away again at higher
temperature differences.
Figure 7.1 The strength of thermally shocked bending samplesaccording to Hasselman
133
Test procedures for determining the resistance to temperature
change are specified in DIN V ENV 820-3 standards. Some ceramic materials
have very low coefficient of thermal expansion therefore their resistance to
thermal shock is very high despite of low ductility (e.g. fused silica): Rs of
fused silica / Rs of soda-lime-silica glass = 45.
Glass and some ceramic objects are particularly vulnerable to this
form of failure, due to their low toughness, low tensile strength, low thermal
conductivity, and high thermal expansion coefficients. However, they are
used in many high temperature applications due to their high melting point.
Thermal shock occurs when a thermal gradient causes different parts of an
object to expand by different amounts. This differential expansion can be
understood in terms of stress or of strain, equivalently. At some point, this
stress overcomes the strength of the material, causing a crack to form. If
nothing stops this crack from propagating through the material, it will cause
the object's structure to fail. The problem then becomes how to prevent
thermal shock while still maintaining the temperature extremes required by
the process.
When materials must be tested for their ability to withstand
temperature extremes, they are tested inside a thermal shock chamber. Within
the chamber they are exposed to rapid cycling of extreme hot and cold
temperatures, to determine the temperatures at which the tensile strength of
the material is overcome. This type of testing is used in a very broad range of
industries, including land, air, and spacecraft development, as well as
industrial manufacturing.
Whereas high local thermal stresses in metals merely lead to a
slight local plastic deformation, they can lead to the propagation of cracks in
ceramic materials. For this reason sudden, large changes of temperature
should be avoided whenever possible.
134
7.1.5 Maximum Service Temperature
Ceramic materials retain their properties at elevated temperatures
due to the strong ionic-covalent bonding. Ceramics working at high
temperature are called refractory ceramic materials. Some Borides, carbides
and nitrides, having melting temperature above 3040 ºC, are used in high
temperature applications up to 1800 ºC to 3000 ºC. For example, scaling
(oxidation) temperature of refractory stainless steel AISI 310 is 1150ºC.
7.2 EXPERIMENTAL WORK ON THERMAL CONDUCTIVITY
7.2.1 Introduction
Plasma sprayed coatings are built up by successive accumulation of
molten or semi-molten splats on the substrate surface, forming thin lamellae
upon solidification. The thermal contact between these lamellae is not perfect
and is limited by the presence of pores and cracks at the interface between the
lamellae. The micrographs show these defects. Hence, the thermal properties
of plasma sprayed coatings are quite different from those of the bulk
materials. Thus, measuring thermal conductivity is important for thermally
sprayed coatings and especially for thermal barrier coatings where thermal
properties are of great importance. The thermal conductivity is given by the
following expression:
k= S (7.5)
where, k is thermal conductivity, is thermal diffusivity, is the density, and
S is the specific heat capacity of a material. In this study, the thermal
conductivity of the coated specimens is determined experimentally. Generally
ceramic materials are poor conductors of heat and are extensively used for
thermal barrier applications. In the present section, an attempt is made to
determine the thermal conductivity of mullite coated specimens in order to
135
investigate the potential application of this material for IC engines. Hamed
Samadi (2009) in his work measured the thermal diffusivity and thermal
conductivity of free standing coatings of three ceramic materials, mullite,
forsterite and spinel during both the heating and cooling cycle and found that
the variation of both thermal diffusivity and conductivity is less and the
thermal diffusivity /temperature and the thermal conductivity/ temperature
curves are flatter in the case of mullite and also the values are less compared
to the other ceramics owing to its good thermal barrier properties. The tests
were conducted upto 1000°C. Based on this research study, since the thermal
conductivity did not vary with temperature for mullite, a lower test
temperature of 100°C was chosen for the experiments to determine the
thermal conductivity of duplex coated mullite specimens.
7.2.2 Experimental Set-up
Thermal conductivity was measured by the Lee’s Disc apparatus
shown in figure7.2. The apparatus consists of a steam boiler sufficiently filled
with water and its mouth is closed by a rubber stopper. There is an outlet
through which steam/water vapor escapes. The steam boiler is placed on a hot
plate. The hot plate distributes heat evenly to the steam boiler. A metallic disc
suspended from a stand, has a hole through which a thermometer T2 is
inserted. The insulator specimen whose thermal conductivity is to be
measured is placed over the metallic disc. The diameter of the bad conductor
is the same as that of the metallic disc. A metallic disc is placed on the bad
conductor. A hole is provided through which a thermometer T1 is inserted.
A hollow steam chamber is placed on the metallic disc. It has an outlet and
inlet. Steam is made to pass from the outlet of the boiler to the inlet of the
steam chamber through a rubber tube.
136
Figure 7.2 Lee’s disc apparatus
The following measurements are made before conducting the
thermal conductivity study.
1. The thickness of the bad conductor ‘d’ and metallic disc ‘h’.
2. The mass of the metallic disc, M.
3. The radius of the metallic disc ‘r’.
Then the hot plate is switched on. The steady state temperatures 1
and 2 are noted from the thermometers T1 and T2 . Now the bad conductor is
removed and 2 increases. It is allowed to increase till 2 + 10°C, after which
the steam chamber is also removed. The temperature drops and for every 2°C
fall in temperature the time is noted till 2 – 10°C. Now a graph is drawn with
time ‘t’ along x- axis and temperature along y- axis. The slope d /dt is found.
The time taken for the temperature to drop each degree from 80°C to 70°C
was measured. Slope is found in-between the values of 2 + 1°C and 2 -
1°C.The specific heat of the metallic disc is taken as S= 385 J/kg/K (ASM
handbook, Vol.II).
The coated specimen is shown in the Figure 7.3.
137
Figure 7.3 Coated specimens for thermal conductivity test
The thermal conductivity of the bad conductor is determined from
the formula
K/m/Wh2r2r
)h2r(dddMS
k21
2t
1
(7.6)
where , M is the mass of the metallic disc :
850x10-3 kgs
S1 is the specific heat capacity of the disc : 385 J/kg/K
d / dt is the rate of cooling at :
0.028 °C /s
d is the thickness of the bad conductor : 6x 10-3 m
r is the radius of the metallic disc : 55 x 10-3 m
h is the thickness of the metallic disc 10 x 10-3 m
1 is the steady temperature of the steam chamber : 98 °C
2 is the steady temperature of the bad conductor : 75 °C
138
The measurements were conducted on the following types of samples
1. T6 treated cast aluminum A 353.0 without bond coat and coated
with 100 µm thick mullite (single layer mullite coating).
2. T6 treated cast aluminum A 353.0 without bond coat and coated
with 150 µm thick mullite (single layer mullite coating).
3. T6 treated cast aluminum A 353.0 with bond coat of nickel
chrome (100 µm thick) and with 100 µm thick mullite (duplex
coating).
4. T6 treated cast aluminum A 353.0 with bond coat of nickel
chrome (150 µm thick) and with 150 µm thick mullite (duplex
coating).
Note: The coating thickness varied from 100 µm to 150 µm
for the single layer coating and 200 to 350 µm for the duplex
coating. Nominal values have been used.
7.2.3 Results and Discussion
Five specimens each were tested and the average results are shown
in Table 7.1.
Table 7.1 Results of the thermal conductivity measurements
S.No. Descriptionof specimen
Thermal conductivity inw/m/k
1 Type 1 0.192
2 Type 2 0.181
3 Type 3 0.170
4 Type 4 0.151
139
An average value of 0.151 Wm-1K-1 @ RT( room temperature) may
be taken as the thermal conductivity of the coated specimens of this study.
A comparison of the thermal conductivity of other materials is given in
Table 7.2. Literature reports of the thermal conductivity measurements are
also shown.
Table 7.2 Comparison of thermal conductivity of materials
S.No. Material Thermal Conductivityin Wm-1K-1 @RT
1 Copper 3852 Aluminum 2013 Brass 1104 Lead 355 Card board 0.046 Cork 0.057 Glass 1.08 Rubber 0.159 Wood 0.15
10 YSZ coatings 1.25 to 1.4511 Mullite coatings (by others) 1.512 Alumina PEO coatings 1.513 Mullite rich PEO coatings 0.514 Mullite, bulk 6.06
Note : Reference Carl L. Yaws
Thermal conductivity(TC) of mullite coatings differed by a factor of 10
from other coatings reported due to the thickness of the coated layer (350
µm), the microstructure and the processing technique used. Plasma sprayed
coatings result in a splat structure with discontinuities and porosity of 20%
maximum and hence are good thermal barriers. One coating has used an
advanced technique like PEO (Plasma electrolytic oxidation), (Curran et al,
2007) with coating thickness of 200 µm and the microstructure seems to
140
contain less pores and defects. The TC reported was 1.5 W/m/K for alumina
rich coating and 0.5 W/m/K for mullite rich PEO coatings. Another coating
with thickness of 1 mm (Hamed, 2009), using the air plasma technique
reported a TC of 1.5 W/m/K with 14 % porosity. Also the measuring
apparatus used was a “Lee’s Disc apparatus” in this study, against other
techniques like laser flash method used by others. Factors affecting the TC are
a. Thickness of the coating.
b. Composition of the coating.
c. Microstructure of the coating.
d. Defects like pores, discontinuities, voids and cracks in the
coating.
The TC of bulk mullite is reported as 6.06 W/m/k at room
temperature and 0% porosity (CRC Material Science & Engg. Handbook,
page 287, 3rd edition).Coated mullite will have a lower TC due to the defects
present. Figures 7.4 and 7.5 show the splat structure of the coating, the
micrographs with defects in the coating and the SEM image of a PEO coating.
(a)Figure 7.4 (Continued)
Splatstructureofcoating
141
(b)
(c)
Figure 7.4 (a) and (b) SEM images showing the splat structure of thecoating, the pores and discontinuities, (c) Optical imageshowing coating discontinuities
Figure 7.5 SEM micrograph (SE mode) of a polished section through a200 m thick PEO coating (Curran et al, 2007)
Mullitetop coat
20 µ m
Coatingdiscontinuities
Pores
Voids &Discont-inuities
142
For ceramics, the effects of crystalline nature, lattice imperfections,
and mixtures, factors such as internal porosity, grain boundaries and
impurities can affect this property. Higher or lower levels of thermal
conductivity can be attained in fine ceramic materials by controlling these
factors. In thermal barrier coatings and other ceramic oxides, heat is
conducted by lattice waves. The conductivity is composed of contributions
from a spectrum of waves, interaction between lattice waves (intrinsic
processes), scattering by atomic scale point defects and scattering by extended
imperfections such as grain boundaries (Klemens and Gell 1998).
Heat is transported in solid materials by both lattice vibration
waves (phonons) and free electrons. A thermal conductivity is associated with
each of these mechanisms, and the total conductivity is the sum of the two
contributions, or
k = kl + ke (7.7)
Where kl and ke represent the lattice vibration and electron thermal
conductivities, respectively. The thermal energy associated with phonons or
lattice waves is transported in the direction of their motion. The kl
contribution results from a net movement of phonons from high to low-
temperature regions of a body across which a temperature gradient exists.
Free or conducting electrons participate in electronic thermal conduction. To
the free electrons in a hot region of the specimen is imparted a gain in kinetic
energy. They then migrate to colder areas, where some of this kinetic energy
is transferred to the atoms themselves (as vibrational energy) as a
consequence of collisions with phonons or other imperfections in the crystal.
The relative contribution of ke to the total thermal conductivity increases with
increasing free electron concentrations, since more electrons are available to
participate in this heat transference process.
143
Metals are extremely good conductors of heat because relatively
large numbers of free electrons exist that participate in thermal conduction.
The thermal conductivities of several of the common metals generally range
between about 20 and 400 W/m-K. Nonmetallic materials like ceramics are
thermal insulators inasmuch as they lack large numbers of free electrons.
Thus the phonons are primarily responsible for thermal conduction: ke is
much smaller than kl. Again, the phonons are not as effective as free electrons
in the transport of heat energy as a result of the very efficient phonon
scattering by lattice imperfections. Room-temperature thermal conductivities
range between approximately 2 and 50 W/m-K. Glass and other amorphous
ceramics have lower conductivities than crystalline ceramics, since the
phonon scattering is much more effective when the atomic structure is highly
disordered and irregular. Porosity in ceramic materials may have a dramatic
influence on thermal conductivity; increasing the pore volume will, under
most circumstances, result in a reduction of the thermal conductivity. In fact,
many ceramics that are used for thermal insulation are porous. Heat transfer
across pores is ordinarily slow and inefficient. Internal pores normally contain
still air, which has an extremely low thermal conductivity—approximately
0.02 W/m-K (Callister, Jr. and Rethwisch, 2010)
7.3 THERMAL SHOCK TEST (TST)
7.3.1 Introduction
Thermal Shock is performed to determine the resistance of the part
to sudden changes in temperature. The parts undergo a specified number of
cycles, which start at ambient temperature. The parts are then exposed to an
extremely low (or high) temperature and, within a short period of time,
exposed to an extremely high (or low) temperature, before going back to
ambient temperature. Burner rig durability testing is performed to study the
effect of cyclic thermal gradients on coatings. In this test, a heating period
144
(e.g., 4 min) followed by a cooling period (e.g., 6 min) is applied to the top
coat. Various heating sources have been employed, including a burner and
laser. After the final cycle, external visual examination of the specimens is
performed at 10 X to 20 X. An illegible mark and/or any evidence of damage
to the specimen after the stress test shall be considered a failure. Failure
acceleration due to thermal shock and temperature cycling depends on the
following factors: 1) the difference between the high and low temperatures
used 2) the transfer time between the two temperatures and 3) the dwell times
at the extreme temperatures. For reliability testing or qualification of new
devices, 1000 temp cycles are usually performed, with interim visual
inspection at 200X and 500X.
7.3.2 Experimental Work
Thermal cycling was carried out for the duplex coated T6 treated
samples by placing the specimens in a muffle furnace maintained at 550°C for
a holding time of 10 minutes and blowing compressed air on the specimens
after removing from the furnace, to cool the samples quickly to room
temperature. Repeated runs were carried out to notice any sign of apparent
failure. The test is repeated to check the number of cycles before spallation.
The specimen geometry is shown in figure in chapter 3.
7.3.3 Results and Discussion
During the thermal shock process, there are two main factors
destroying the coatings. One is that the oxidation at high temperatures affects
the combining state of interfaces. The other is that there is a difference of
thermal expansion coefficient between the coating and the substrate materials.
With the temperature up or down, the volume of the two materials changes
differently and this leads to interior stress. The greater the difference of
thermal expansion coefficient, the higher the stress. During the cycling, the
stress keeps rising and when local stress exceeds the strength limitation of
145
coating material, the cracks form and grow continuously until the coatings
spall. In this study six specimens were tested for 100 cycles, without any
noticeable degradation or spallation of the coating. The test was stopped after
100 cycles. The reason for this good performance is due to the high thermal
shock resistance of mullite and its high recrystallization temperature of
1000°C. Figure 7.4 shows the comparison of mullite coating with aluminum
oxide coating. The performance was comparable to functionally graded
coatings reported in literature and presented in Table 7.3.
Notes: Type 1: Al2O3 oxide coating on carbon steel substrates tested at 800°C.Type 2: Mullite coating
Figure 7.6 Thermal shock resistance of mullite coating
Table 7.3 Comparison of mullite coatings with FGC’s
S.No. Description
Yttria stabilizedzirconia/
NiCoCrAlY Hydroxyapatite/Ti-6Al-4V FGC
Fe3Al/Al2O3FGC
Mullitecoatedcast Al(DuplexCoating)
Duplexcoating FGC
1. Bond strengthin MPa 9.3 17.8 38 51 20
2. No.of thermalcycles 15 90 - 119 100
The test temperature used was 550°C only, for an IC engine
application and the recrystallization temperature of mullite being 1000°C no
Coating type
146
apparent damage was expected in the specimens during thermal cycling. As
anticipated, there were no changes in the coated structure seen visually and
also through SEM images. Figure 7.6 below shows the SEM images.
( a)
(b)Figure 7.7 (Continued)
Splatstructureofcoating
Topsurfaceofcoating
147
(c)
(d)
(e)
Figure 7.7 SEM images showing the specimens before testing : (a), (b),
(c) and after testing : (d) and (e)
Topsurfaceofcoating
Topsurfaceofcoating
Topsurfaceofcoating
Machiningmarks
Machiningmarks
148
7.4 STATISTICAL ANALYSIS OF THERMAL
CONDUCTIVITY TEST
The experimental data for the thermal conductivity test is presented
in the Table 7.4.
Table 7.4 Thermal conductivity from experiments in W/m/K
Type of material( Coating thickness is nominal)
Total ofobservations yi
Averages yi
Cast aluminum A 356.0 (T6 treated) withsingle layer mullite coating(100 µm coating thickness)
1.03 0.206
Cast aluminum A 356.0 (T6 treated) withsingle layer mullite coating(150 µm coating thickness)
0.937 0.187
Cast aluminum A 356.0 (T6 treated) withduplex coating (200 µm coatingthickness)
0.85 0.17
Cast aluminum A 356.0 (T6 treated) withduplex coating (350 µm coating thickness) 0.738 0.148
yn = 3.555 yn = 0.178
Since four different material types and in each case five specimens
were tested, there are four levels (or treatments) and five observations at each
level. The objective is to test the appropriate hypothesis about the treatment
and to estimate them. The results of the statistical analysis for the thermal
conductivity of the specimens are presented in Table 7.5.
Table 7.5 Analysis of variance of the thermal conductivity data
Source ofvariation
Sum ofsquares
DOF Meansquare
Ratio of SampleVariance, F0
P-Value
Material type 0.0093 3 3.1x10-3 99.2 < 0.05Error 0.0005 16 3.125 x 10-5
0.0098 19
149
It is seen from the above tables that between treatment mean
squares are many times larger than within treatment or error mean square. The
F ratios for the thermal conductivity is computed as F0 = 3.1x10-3/ 3.125 x
10-5 = 99.2 and compared with an appropriate upper tail percentage point of
F 3,16 distribution. For 5% level of significance (risk), from statistical table
(Douglas C.Montegomery 1997), F 0.05,3,16 = 3.24. Since F0 for the thermal
conductivity is greater than 3.24, the difference in material type (single layer,
duplex layer coating), significantly affects the mean thermal conductivity and
it can be concluded that an upper bound for the P-value is 0.05; that is
P < 0.05.
7.5 CONCLUSIONS
The thermal shock resistance properties of the T6 treated duplex
coating /substrate system is good and it has withstood 100 thermal cycles,
when exposed to an ambient temperature of 550°C in the furnace without
delamination of the coating taking place, due to its high melting point and
creep resistance. The strain rates are low and crack initiation and propagation
in the coating do not take place for operating temperatures of around 600°C,
normally found in IC engines. The thermal cycling test showed the enhanced
thermal behavior of the coated samples. Hence this coating can be used for IC
engine applications. The thermal conductivity of the T6 treated duplex coated
specimens is found to be low and suitable for thermal barrier applications. An
average value of 0.151 Wm-1K-1 was measured. Statistical analysis and
analysis of variance (ANOVA) confirmed the variation in the thermal
conductivity measurements of the specimens and also showed the lower
values for the duplex coated specimens of 350 µm thickness.