High Enrth

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

416 Surface and Coatings Technology, 43/44 (1990) 41&-425

CALCULATION OF RESIDUAL THERMAL STRESS IN


PLASMA-SPRAYED COATINGS*

R. ELSING, 0. KNOTEK and U. BALTING


Institut für Werkstoffkunde, Technical University, Templegraben 55, D-5100 Aachen (F.R.G.)

Abstract

Residual internal stresses are induced in thermally-sprayed coating


composite materials during deposition, owing to the different thermophysical
properties of the substrate and coating materials and to the different spray-
ing parameters. These residual stresses have a substantial effect on working
properties, and therefore need to be taken into account in any optimization
of the coating.
Due to the wide variety of material and process parameters determining
residual thermal stresses, systematic experimental testing is extremely effort,
time and cost intensive. It is therefore necessary to determine residual
stresses by means of mathematical techniques which, following experimental
validation of the model, enable the influence of individual parameters on the
residual thermal stress state of coating composite materials to be studied and
the behaviour of the composite under practical conditions to be predicted.
This paper presents a model of this kind. Calculated results for alu-
minium oxide and zirconium oxide on various substrates (process: plasma
spraying) are compared with measured results, showing good agreement.

1. Introduction

Residual thermal stresses are induced in thermally-sprayed coating


materials during deposition. Hot particles, as far as possible in the molten
state, impinge on relatively cold substrates or on an existing partial coating,
where they solidify, releasing their melting heat, and continue to cool. As a
result, the layer on which the particles impinge is heated. This process is
repeated throughout deposition of the coating, until the last particle has
impinged on the surface and the desired coating thickness has been achieved.
The process of coating deposition is thus associated with widely oscillating
temperature loads both on the growing film and on the substrate, tempera-
ture fluctuations in the lower layers of the coating decreasing as the coating
thickness increases. Temperature levels and the temperature curve are de-

*Winner of a 1990 ICMC/ICTF Bunshah Award for Outstanding Paper.

0257.8972/90/$3.50 ~Elsevier Sequoia/Printed in The Netherlands


417

pendent on the various thermophysical parameters of the coating and sub-


strate materials, the melting heat of the particles and the temperature
loading caused by the hot plasma flowing onto the growing film and thus on
the process parameters. We have previously investigated and represented
these interrelationships in detail, using mathematical models [1, 2].
Constant or oscillating temperature loads lead to the formation of
internal stresses in materials consisting of different phases, i.e. especially in
composites or coating composites. In accordance with the fundamental equa-
tion for the calculation of changes in length as a function of temperature,
these stresses are similarly dependent on different moduli of elasticity,
different thermal expansion coefficients and temperature, and hence on all
the other parameters of the materials constituting the composite and on the
process parameters.
During the use of the coating system, internal stresses in a coating
composite material act as pre-stresses on which the mechanical or thermal
operational loads are superimposed. Depending on their orientation and
strength, they may lengthen or shorten service lives or, in extreme cases,
result in immediate spalling of the coating. Their great importance makes the
precise knowledge of residual thermal stresses a topic of interest for both
coating manufacturers and users.
In the past, residual thermal stresses have repeatedly been the subject
of scientific investigation. Experimental determination of residual stresses in
coating composite systems is based on strain measurements, which are
generally destructive and which produce results applicable only to a single
special case. Interpretation of conflicting results is also frequently difficult,
since measurements provide no information on the way in which residual
stresses originate in a particular case and on the resulting level and be-
haviour over a cross-section of the composite system.
Mathematical calculations which take into account the process by
which the composite material originates are of assistance in this respect. A
model for relatively thick coatings is given in ref. 3; residual stress states in
thinner coatings can be characterized according to a model devised by us [4],
which has been considerably sophisticated in the present investigation to
achieve a closer description of the real process. Phase transformations in
films are of importance chiefly in terms of thermal operational loads, not in
terms of film deposition. The cooling rates of the molten particles are so high
that the high temperature modification generally stabilizes and any further
modifications at lower temperatures are suppressed. These are therefore not
allowed for in the present model.

2. Description of the model

The calculation of thermal stresses is based on temperature curves


during deposition of the coating and on subsequent temperature equalization
processes. These are described in a coating structure model and are shown
418

Temperature T

‘:L~~
~ting

~ra~ (7
Fig. 1. Incremental growth of films: temperature curves of coating composite materials over
cross-section and time.

diagrammatically in Fig. 1 over the period of coating deposition. It will be


evident that the film in the model grows in incremental elements, yielding
different temperature curves in the composite material over time. The method
of calculation has been described in detail elsewhere [1, 2].
The fundamental concept of the temperature curve determination model
is based on the sub-division of the coating deposition process into time
elements (approximately 10_c s), the calculation of the existing coating
thickness for each time element and the determination of the temperature
curve for the entire composite from the amounts of heat introduced (via the
hot gas plasma and the particles) and dissipated for each time element. The
thermal constraints of the model are convective heat loss at the substrate
and coating surfaces, ideal contact between the substrate and the coating
and negligible radiation losses [1, 2j.
On the basis of calculated results, the thermal internal strain state of
the coating composite material is then described in a two-stage calculation
model.
(i) The total lateral expansion of the coated workpiece, which is
assumed to be plate-shaped, together with changes due to thermal influences,
are first calculated by numerical integration of the equation

I~~ =x~~E~A~/l0~ ~ —®~~)

where i is an index (s for substrate, c for coating), x, x, i are the path


coordinates along the substrate surface, velocity and acceleration during
thermal expansion and contraction, E, is the Young’s modulus of the coating
and substrate material, A, is the cross-sectional area of the individual
element, is the temperature-dependent coefficient of thermal expansion, m,
~,

is the mass of the individual element, e, is the instantaneous temperature of


each individual element, l~,is the length of the element in the unstressed
419

state and ®~, is the maximum temperature’of the supporting layer on which
the particle impinges during the cooling process of the uppermost lamella.
This equation essentially represents a special form of the basic new-
tonian mechanics equation, applied to the individual substrate and coating
elements. The forces, and thus the stresses, in each of these elements consist
of the deformation-dependent influences resulting from the incorporation of
the element in the composite minus the temperature-dependent strain of the
element itself. The sum of these forces on the left-hand side of the equation
is not equal to zero, because the relatively large acceleration forces which
may occur due to rapid contraction during solidification of the particles and
the rapid heating of the particles immediately below them, also due to
thermal expansions, cannot be neglected.
Incorporation of an impinging particle in the coating composite occurs
at the moment when the supporting layer has just attained its maximum
temperature. The total lateral expansion of the coating composite can hence
be calculated from the instantaneous thermal expansion and the size of the
substrate prior to deposition. The temperature curve model referred to above
assigns the temperatures to various times during deposition of the coating
and to the local coordinates, enabling these values to be integrated directly
in the equation.
(ii) The above equation is used during the period in which the ther-
mally-sprayed coating is being deposited. Following termination of the spray-
ing process, during the ensuing cooling phase and at later times when
operational loading occurs at raised temperatures, the equation
10i {1 + ;(®~)(O,
— ®Oi)}
ii =
is used for each element of the coating and substrate. 1, is the length which
an element would possess at the temperature 0 if it could adjust freely, i.e.
unconstrained by the surrounding elements. Within the composite, however,
the elements interact with and obstruct one another, and these influences are
calculated, taking into account the bending moments and using formulae
similar to those for bimetallic strips.
Table 1 shows the material data used in the temperature curve and
strain model.

3. Results and discussion

Internal strains in plasma-sprayed coatings deposited on austenitic and


ferritic steel using the parameters given below were measured by the familiar
ring-core method [5], modified to include the use of diamond drill bits [6]. The
resiliences were determined by measuring the resulting surface internal
strains in several directions with the aid of specially developed strain gauges.
The deeper the milling depth, the lower the measurable deformations. This
behaviour, typical of the process, was taken into account by defining a
damping function, which can be determined either mathematically or experi-
420

C.)

~I
—I ~-c~
-— C’:,

.~ :;i -:;~ C:,

0 C-)
C
C. ——
C x
0 ._~
— C0-’.~ — CC
—j C’C.
c’i ~ ©

C/i
a
a
a ©

~ ~ —
© ~
C I ~ CC

I ~ C’.’.

a
a0
0 —

—-~.
a ,~- >~
E~
.‘Z
•~
a~
.~

a
I

~ L I E ~ C

< •~ L OX ~ C .0
>~_~ E—o ~lC/ C-)
421

1.50 -
0.75
t o
/00

I ___

100 0 100 pm 200


0 I I
0.25 100 0 100 pm 200

coating substrate coating substrate


Fig. 2. Measured internal strain curves (Al
203 on ferrite).
Fig. 3. Measured internal strain curves (Al203 on austenite).

mentally on a specimen with known, depth-constant internal strain state.


The method is valid given that the strain at right angles to the surface is
zero. This constraint is satisfied exactly only at the surface, but also applies
with sufficient accuracy over the entire cross-section of thin coatings.
Figures 2 and 3 show the results for Al2 03 on ferrite and austenite.
Calculation on the temperature curves and the inferred strains was
based on the following spraying parameters, which were also used to plasma
spray the specimen coatings: substrate thickness, 7.8 mm; gas temperature at
the substrate, 2700 °C; gas velocity at the substrate, 200m s’; torch feed
rate, 5 cm s’; nozzle diameter, 8mm.
Apart from thermophysical parameters, the substrate thickness, the
temperature of the plasma flowing onto the substrate or the growing film, the
gas velocity, the torch feed rate and the nozzle diameter, which determines
the size of the spot, were included in the temperature curve calculation. The
velocity of the particles at the moment of impact was not taken into account.
The melting points of Al2 03 and Zr02 were taken 2 as K1
the particle tempera-
in the spot and
tures. Heat transmission coefficients of 1000W m
lOW m2 K~’at the reverse side of the substrate were used. At the given
torch rate, films of 100 jim (A1
203) and 40 jim (Zr02) were deposited in a
single pass.
422

~ H ~
+

— ~

~ -~i~-~- .—.

+ 1 Aj te~I
%, +

.1

• I Ii, I I,

I I

• .- • . S S •

Fig. 4. Calculated internal strain curves: Al:,O~,on ferrite and austenite.

The results of the strain calculations from the temperature curves are
presented in Figs. 4 and 5.
A comparison between the measured and calculated results shows that
the measured values for internal strains are lower than the calculated
equivalents. This is partly attributable to the assumptions or simplifications
in the model, but the method of measurement also contains some typical
sources of error, as, incidentally, do all other methods for measuring internal
strains in sprayed coatings. The measured absolute value of the internal
strain is, for example, dependent on the lateral convexity produced by
drilling, the calibration function chosen to determine the depth curve, the
drilling feed rate and the size of the hole drilled and hence the size of the
influencing zone.
Conversely, there is good agreement between the qualitative curves of
the measured and calculated internal strains and between the differences
produced by deposition on ferritic and austenitic substrate materials. The
measured and calculated internal strains at the coating surface are twice as
high on austenite as on ferrite, whereas the calculated and measured differ-
ences in internal strains in the various substrates are much less pronounced.
The results from the internal strain curves may be summarized as
follows.
423

COATING3 SUBSTRATE ZrO,/Ferrlte

I ~

i~
t 5)
ZrOO

3 + yflIjle ofralr-
1: — compressIve 011010

2 room - fenlpnraiul
S~.. ~\ • end of spraying orclcess
•~ ~I o 1 sec. offer end of sprvying process

I I 0 L...~ I I I I I
-50 50 100 50 pm 200
S • S S S S

Fig. 5. Calculated internal strain curves: Zr0


2 on ferrite and austenite.

(i) In the four coating—substrate combinations considered, tensile loads


occurred in the coating and compressive loads in the substrate. The transi-
tion from tension to compression occurs in the vicinity of, but not always
exactly at, the interface.
(ii) The loads are time dependent and therefore dynamic. The strongest
loading of the coatings occurs between the end of the spraying process and
complete cooling. This phenomenon may be attributed to inadequate temper-
ature equalization in the coating composite during this phase, reinforced by
the poor thermal conductivity of the coating and/or substrate. Only during
the course of the spraying process and thereafter can the introduced heat
gradually penetrate the substrate zones further from the interface, causing
thermal expansions which have to be taken up by the entire composite
system and hence also by the coating.
(iii) The loads on the coatings after cooling to room temperature are, in
part, far lower than those encountered during cooling. Currently, however, it
is possible to measure only the loads after complete cooling, which yield
hardly any information on the actual loading of the coatings.
(iv) As may be expected from the poor thermal conductivity of the
austenite and the resulting higher temperature gradients during production,
424

TABLE 2
Moduli of elasticity for ZrO:,

Material Young’s modulus (GPa)

Lackey et al. [7/ Hobbs [8/ Hancock [9]

Zr0
2 48 (B) ~- 170 (5)
Zr02 Y:,0., -- 34.5 (S) 32-40 (S)
Zr02-8Y20,, 205 (B) 3.4—6.9 (5)
ZrO:,—2OMgO 97 207 (B) 46.2 (5)
ZrO:,—24Mg0 --- 4.7 (5) -

B, bulk, sintered; S. as sprayed.

the coatings on austenitic substrate materials exhibit the largest internal


strains.
(v) The differences between the thermal expansion coefficients of Zr02
and the two steels investigated are less than in the case of Al2 03. Nonethe-
less, the strains are larger. This may be attributed partly to the changed
temperature curves and partly to the much lower modulus of elasticity of the
compact material, which also explains the higher internal stress state gener-
ally assigned to the Al2 03 as opposed to the Zr02 coatings.
(vi) The internal stresses can be inferred from the calculated internal
strains via the moduli of elasticity, using the values for the sprayed state of
the relevant material. In the equation given above, this is unnecessary, since
the lateral extent of the individual elements may be assumed to be so small
that the material in these zones can be regarded as ideal and not affected by
pores, loosely adhering particles, etc. In real coatings, however, these two
phenomena occur, substantially reducing the modulus of elasticity as com-
pared with that of the compact material, as exemplified in the case of
zirconium oxide in Table 2 [7—9].
The internal strain state therefore appears to provide a much better
basis for evaluating pre-stressing of a coating. The interrelationships noted
above and illustrated in the figures remain valid irrespective of the absolute
value of the modulus of elasticity (as sprayed), and the curves presented
reflect the internal stress curves if the modulus of elasticity fluctuates by the
same mean value at all points within a coating.

4. Conclusions

The determination of thermally-induced residual internal stresses in


coating composite materials is of great significance for their behaviour under
load. The internal stresses depend on the thermophysical data of the sub-
strate and coating materials and on the spraying parameters. Experimental
and mathematical methods are available for determining internal strains. A
425

mathematical method is described and the internal strain curves obtained by


this method are compared with experimental values. The various curves for
different material combinations can be demonstrated both mathematically
and experimentally.
In principle, it is extremely simple to calculate internal stresses from
internal strains. However, precise determination of the required mechanical
parameters for the sprayed state is problematical, and it is suggested that
loading of the coatings should be characterized via the internal strain state.

Acknowledgment

The authors wish to express their thanks to Dr. P. Pantucek of the


Fraunhofer-Institut für Betriebsfestigkeit (LBF) in Darmstadt for perfor-
mance of the internal strain measurements.

References

1 0. Knotek, R. Elsing and U. Balting, Surf. Coat. Technol., 36(1988) 99.


2 H. Elsing, 0. Knotek and U. Baiting, Proc. mt. Thermal Spraying Con!. 1989, London, June
1989.
3 D. S. Rickerby, G. Eckold, U. T. Scott and I. M. Buckly-Jobbs, Thin Solid Films, 154 (1987) 125.
4 H. Eising, 0. Knotek and U. Baiting, Surf. Coat. Technol., 41 (1990) 147.
5 M. T. Fiaman, Experimental Techniques, January (1982) 26.
6 P. Pantucek, VDI-Berchte, No. 679, Verein Deutscher Ingenieure Verlag, Düsseldorf, 1988.
7 W. J. Lackey, D. P. Stinton, G. A. Cerny, L. L. Fehrenbacher and A. C. Schifauser, Ceramic
Coatings for Heat Engine Materials—Status and Future Needs, Proc. mt. Symp. on Ceramic
Components for Heat Engines, October 1 7—21, 1983, Hakine, Japan.
8 M. Hobbs, Surf. J., 16(4) (1985) 155.
9 P. Hancock, Degradation Processes for Ceramic Coatings, E-MRS Con!. on Advanced Materials
Research and Development for Transport, Strasbourg, November 1985.

You might also like