See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/228925110

4.1. 3b Residual Stresses in Thick and Thin

Surface Coatings

Article

CITATIONS READS

5 137

1 author:

Bill Clyne
University of Cambridge
290 PUBLICATIONS 10,534 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

both the projects View project

All content following this page was uploaded by Bill Clyne on 16 September 2014.

The user has requested enhancement of the downloaded file.

To be published in “Encyclopaedia of Materials: Science and
Technology”, §4.1 – “Elasticity and Residual Stresses”, PJ Withers (ed.),
Elsevier 2001

4.1.3b Residual Stresses in Thick and Thin Surface

Coatings
T.W.Clyne
Department of Materials Science and Metallurgy
University of Cambridge
Pembroke Street
Cambridge CB2 3QZ, U.K.
tel: 01223 334332
fax: 01223 334567
e-mail [email protected]
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

4.1.3b Residual Stresses in Thick and Thin Surface

Coatings
Introduction

Residual stresses are present in most surface coatings. These can be of considerable
significance, since they may influence characteristics such as the resistance of the coating to
wear and fatigue crack propagation. Furthermore, there is often a danger that the presence of
residual stresses may promote debonding and spallation of the coating. This becomes
increasingly likely as the thickness of the coating is increased, since the release of stored elastic
strain energy as the stresses become relaxed can drive this debonding and the quantity of
energy released, per unit area of interface, normally rises more or less linearly with coating
thickness. This article is not primarily concerned with such effects, but rather is aimed at
providing an understanding of how residual stresses arise in surface coatings. It covers both the
underlying mechanics involved and examples of the physical phenomena which can lead to
stress generation. The treatment encompasses both thin coatings (usually produced by
deposition from the vapour phase), in which the stress level is essentially uniform, and thicker
ones (commonly formed by much faster processes such as droplet spraying), in which through-
thickness variations of stress level can be significant. In both cases, however, the coatings can
normally be taken as sufficiently thin to ensure that significant stresses do not build up in the
direction normal to the plane of the coating. Therefore, at any depth in the system, neglecting
edge effects and assuming an approximately planar specimen geometry, an equal biaxial stress
state is established, characterised by a single stress value. Of course, more complex three-
dimensional stress states can be set up with very thick coatings and shaped components, but an
understanding of such cases, and numerical modelling to predict the stress state, would follow
logically from the treatments described here.

1 Misfit Strains, Resultant Stresses and Associated Curvatures

While the main interest often lies in the stress levels which are generated within the surface
coating (and the substrate), it is instructive to consider the situation in terms of misfit strains -
that is, relative differences between the stress-free dimensions of various layers. The simplest
system is composed of just two layers, the coating and the substrate, but it may be appropriate
(particularly for thick coatings) to consider them being deposited as a series of layers. A
further general point is that it can be very instructive to note how the system will adopt a
curvature as a result of the imposition of a misfit strain. This can be useful, not only in helping
to understand the mechanics of stress generation, but also for using curvature monitoring to
measure residual stress levels.

1.1 Force and Moment Balances

Consider a pair of plates bonded together with a misfit strain ∆ ε in the x-direction - see
Fig.1. The resultant stress distribution, σx(y), and curvature, κ , can be obtained from simple
beam bending theory. Firstly, the misfit strain is removed by the application of two equal and
opposite forces (-P and P). When the two plates are joined, this results in an unbalanced
moment, M. Balancing this moment generates curvature of the composite plate. The moment
is given by

page -2-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

h + H
M = P  2  (1)

where h and H are the thicknesses of deposit and substrate respectively.

Now, the curvature of a beam, κ , (equal to the through-thickness gradient of strain) can be
expressed as the bending moment divided by the beam stiffness, Σ
M
κ = Σ (2)

By calculating the beam stiffness of the composite beam (Clyne, 1996), a general expression is
obtained for the curvature arising from the imposition of a uniform misfit strain, ∆ε, such as
would arise during a change in temperature (∆ε = ∆α ∆T)
6 Ed Es (h + H) h H ∆ε
κ = (3)
Ed2 h4 + 4 Ed Es h3 H + 6 Ed Es h2 H2 + 4 Ed Es h H3 + Es2 H4
where E is the Young’s modulus and the subscripts d and s refer to deposit (coating) and
substrate respectively. It may be noted that, for a given deposit/substrate thickness ratio (h/H),
the curvature is inversely proportional to the substrate thickness, H. This scale effect is very
important in practical terms, since relatively thin substrates are essential if curvatures
sufficiently large for accurate measurement are to be generated.

1.2 Biaxial Stresses and Bifurcation

Eqn.(3) is valid provided the original imposed misfit strain is a uniform one (no through-
thickness gradient) and the system remains elastic while the curvature is adopted. A
modification arises, however, on consideration of in-plane stresses other than those in the x-
direction. For an isotropic in-plane stress state, there is effectively another stress equal to σx in
a direction at right angles to it (z-direction); this induces a Poisson strain in the x-direction.
Assuming isotropic stiffness and negligible through-thickness stress (σy = 0), the net strain in
the x-direction can be written
εx E = σx - ν (σy + σz) = σx (1 - ν) (4)
so that the relation between stress and strain in the x-direction can be expressed
σx E
εx = (1-ν) = E’ (5)

This modified form of the Young’s modulus, E’, is usually applicable in expressions referring
to substrate/coating systems having an equal biaxial stress state.

A further point related to biaxial stresses concerns the possibility of multi-directional

curvatures leading to mechanical instability. Under an equal biaxial stress state, an initially
plane surface will tend to become convex or concave, with a curvature which should be equal
in all directions (ie a spherical surface). However, high curvatures cannot be simultaneously
accommodated in all in-plane directions. On increasing the curvature sufficiently, a bifurcation
point will be reached, at which the curvature increases sharply in one plane and decreases
sharply in the plane normal to this (ie the shape becomes ellipsoidal and approaches that of a
cylinder). This should be avoided, since it is difficult to relate a measured curvature to an
internal stress distribution once a specimen has bifurcated.

page -3-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

In general, it is a complex problem to predict the curvature at the bifurcation point for a
substrate/deposit system under residual stress. However, the case of a thin film on a thick,
rectangular substrate (with a uniform stress in the film) has been analysed (Salamon and
Masters, 1995). The critical curvature depends on the elastic constants, deposit/substrate
thickness ratio (h/H), substrate thickness/length ratio (H/L), substrate width/length ratio (b/L)
and substrate length (L). Predicted behaviour is shown in Fig.2. The critical curvature reduces
as the width/length ratio goes up. It is relatively insensitive to the elastic properties and to the
deposit/substrate thickness ratio. It can be seen from the plot that there will be little or no
danger of a bifurcation instability (ie κ B >~10 m -1 ) if a long (L ~100 mm), fairly narrow
(b/L<~0.2) strip specimen is used. This is convenient, since such specimens are well suited to
accurate measurement of curvature along the length of the strip.

1.3 The Effect of Coating Thickness and Use of the Stoney Equation
A simplified form of Eqn.(3) results when the coating is much thinner than the substrate
(h << H). Since the stress in the substrate then tends to become negligible, and that in the
deposit will vary little as a result of curvature adoption, the misfit strain can be converted
directly to a deposit stress, σd (= E’d ∆ε), so that the equation reduces to the form
6 σd (1 - νs) h
κ = (6)
Es H 2
which is usually known as the Stoney equation (Stoney, 1909) and is commonly used to relate
stress to curvature for thin coatings.

When the condition h << H does not apply, then stresses and stress gradients are often
significant in both constituents. Stress distributions are readily found for the simple misfit
strain case outlined above using the expressions
 E’d H E’s 
σd| = -∆ε h E’d + H E’s + E’d κ (h - δ) (7a)
y=h
 E’d H E’s 
σd = -∆ε h E’d + H E’s - Ed κ δ (7b)
|y=0
 E’d h E’s 
σs = ∆ε h E’d + H E’s - E’s κ (H + δ) (8a)
|y=-H
 E’d h E’s 
σs = ∆ε h E’d + H E’s - E’s κ δ (8b)
|y=0
The stress distributions shown in Fig.1 were obtained using these equations. It can be seen
that the adoption of curvature can effect substantial changes in stress levels and high through-
thickness gradients can result. It may be noted from Eqns.(7) and (8) that (for a given value of
h/H), the stresses at y=-H, 0 and h do not depend on H, ie the stress distribution is independent
of scale.

The Stoney equation is accurate only in the limit where the film thickness, h, tends to zero.
Unfortunately, in this limit the curvature, κ, must also tend to zero. It was suggested some time
ago (Brenner and Senderoff, 1949) that the experimental error arising from the curvature being
too small to measure accurately typically exceeds the error introduced via the approximation

page -4-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

incorporated into the Stoney equation when the ratio of the thickness of the coating to that of
the substrate, h / H, is less than about 5%. This is approximately correct, although the details
are dependent on the Young’s moduli and on the absolute thickness of the substrate. This is
illustrated by the plots in Fig.3, which were produced using Eqns.(3), (7) and (8).

These equations refer to an elastic system exhibiting an equal biaxial stress state, resulting
from the introduction of a uniform in-plane misfit strain, ∆ε, between the two constituents.
Note that, for relatively thick coatings, the adoption of curvature not only changes the coating
stress, but also introduces differences between the level at the free surface and that at the
interface. This is one reason why it is more rigorous to define a misfit strain than a coating
stress, although through-thickness variations within the coating would not be significant for
most thin coatings (produced by vapour deposition).

The plots in Fig.3 confirm that the Stoney equation is expected to be quite accurate for
thickness ratios below a few percent or so, depending on the stiffnesses. Experimental studies
with thin films often satisfy this condition, since a typical value for h is 1 µm and H ranges
from about 100 µm to over 1 mm. However, the plots in Fig.3(a) highlight the problem
identified by Brenner and Senderoff for cases where the h / H ratio is very low. For a misfit
strain of 10 -3, many experimentally-used combinations of H and h correspond to curvatures
below 0.1 m -1 , or even below 0.01 m-1 (100 m radius of curvature). Such curvatures are
extremely difficult to measure with any real precision. Of course, the presence of much higher
misfit strains will push the curvature up to more readily measurable levels, but such strains are
not very common. In any event, it should be recognised that, in the regimes of small h / H and
relatively large H, minor errors in measured curvature lead to major changes in deduced stress
level.

1.4 Optimisation of Curvature Techniques for Stress Measurement

A more rational approach than that of conventional application of the Stoney equation,
although surprisingly little used to date, is to generate relatively large curvatures by using high
h / H ratios and to use Eqs.(3), (7) and (8) to deduce the misfit strains and hence the stress
levels. Curvatures in the range 1 - 10 m-1 are readily measured with high precision. Such
curvatures can usually be generated by depositing relatively thick coatings. Unfortunately, this
can lead to adhesion failure, since thicker coatings lead to higher strain energy release rates for
interfacial debonding (see Drory et al, 1988). One solution is to use thinner substrates,
although there is clearly a limit to this, since very thin substrates become difficult to handle and
prone to fracture or undergo plastic deformation. In practice, deposition stress can usually be
measured accurately via curvature monitoring provided suitable control is exercised over
substrate material, substrate thickness and interfacial toughness. For example, the latter can
often be raised by cleaning / roughening of the substrate surface, use of thin adherent
interlayers etc. It should be noted that the thermal expansivity of the substrate is also relevant,
particularly for films deposited at high temperature, since large differential thermal contraction
stresses will introduce errors in measurement of the deposition stress and may promote
debonding during cooling. Measurement of curvature in situ during deposition, or at least
during subsequent cooling, can alleviate these problems. Care should also be taken to avoid
bifurcation (see §1.2 above), but this is usually straightforward provided suitable strip
dimensions are chosen.

page -5-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

2 Stresses in Thin Coatings formed by Vapour Deposition

For a thin coating (h <<H), the misfit strain is commonly assumed to be taken up entirely
within it, so that the stress level is simply obtained on multiplying by the (biaxial) modulus.
While the system may adopt a measurable curvature (see §1.3), the associated variations in
stress level within the coating can usually be neglected. For all types of coating, the main
sources of residual stress are (a) differential thermal contraction and (b) phenomena occurring
during deposition. It is common for these to be referred to respectively as extrinsic and
intrinsic stresses. Other processes, such as phase transformations, plastic flow, creep etc, can
also effectively generate a misfit strain. These can in principle be handled in much the same
way as thermal stresses.

2.1 Differential Thermal Contraction Stresses

Thermal stresses can readily be calculated from a knowledge of the thermal expansivities of
the constituent materials. The associated misfit strain can, provided these expansivities are
temperature-independent, be written as
∆ε th = ∆T (α s - α d ) (9)
where ∆T is the temperature change. This strain, and the associated stress level in the coating
at room temperature, will clearly be greater when deposition takes place at high temperature
and when the expansivity mismatch is large. While this equation is commonly employed, it
should be noted that neglect of the temperature dependence of the expansivities may be
inaccurate. The misfit strain at ambient temperature, T0 , after cooling from the deposition
temperature T2 , should be obtained from
T0
∆εth = ⌠
⌡(α s - α f) dT (10)
T2

It can be seen from the data presented in Fig.4, which relates to deposition of diamond films on
a tungsten substrate, that use of invariant expansivity values (which would give a linear plot of
misfit strain against temperature) can introduce large errors.

2.2 Deposition (Intrinsic) Stresses

Stresses can arise in several ways during molecular deposition from the vapour phase. For
example, molecular species arriving with high energies can become implanted within the
deposit, where they may occupy interstitial sites (or vacant lattice sites which are
thermodynamically stable) and hence generate compressive stress. Bombardment with
energetic non-depositing species can also promote such site occupancy and hence have a
similar effect, which is often termed “atomic peening”. On the other hand, processes can also
take place during deposition which generate excess vacancies and hence tensile stresses. For
example, preferential removal of a depositing molecular species via etching, immediately after
deposition, can lead to this effect. A further point to note is that there will be a tendency for
both excess vacancies and interstitials / vacancy depletion to be removed by short-range
(surface) diffusion, provided there is sufficient thermal energy available. Both substrate
temperature and the short duration thermal energy injected by bombarding or implanting
species are important in determining the degree to which such annealing processes occur.

page -6-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

Deposition of diamond films requires relatively high substrate temperatures and also leads to
generation of excess vacancies, and associated tensile stresses, as etching occurs during
deposition. The deposition (intrinsic) stress in diamond films deposited onto tungsten is plotted
against methane content and deposition temperature in Fig.5. Preferential etching by atomic
hydrogen of sp2 carbon during diamond deposition is a primary source of excess vacancies.
(The etching rate of sp2 carbon is more than 20 times that of sp3 carbon.) Incorporation of sp2
carbon into the film, on the other hand, tends to produce compressive stress, since its specific
volume is 1.5 times that of sp3 carbon. At low methane levels, there is little initial formation of
sp2 carbon and hence the excess vacancy concentration is low. As the methane level rises, a
higher sp 2 carbon is initially formed, but most of this is etched away by the atomic hydrogen.
This gives rise to a high (tensile) intrinsic stress. At still higher methane contents, however, a
significant amount of sp2 carbon survives the etching by atomic hydrogen and becomes
incorporated into the film. This makes the stress more compressive (and also leads to a
reduction in Young’s modulus and changes in film morphology).

A higher substrate temperature raises the recombination length for atomic hydrogen,
promoting the etching of sp2 carbon and making the stress more tensile (at least for methane
levels where there is substantial deposition of sp2 carbon). On the other hand, higher
temperatures also enhance diffusion and thus tend to reduce the excess vacancy concentration.
This effect, which decreases the tensile stress, tends to become significant only at relatively
high temperatures (Fig.5(b)).

During deposition of diamond-like carbon (DLC), on the other hand, intrinsic stresses are
normally compressive and arise by quite different mechanisms from those operating with
diamond. Deposition temperatures are lower than during diamond formation and there is much
more bombardment by energetic species. Measured intrinsic stress values are plotted in Fig.6
as a function of negative bias voltage, V b . Initially, the stress rapidly becomes strongly
compressive as Vb rises. This is the result of pronounced implantation of bombarding carbon
ions, once they have enough energy to penetrate the structure. Further increases in
bombardment energies lead to intensive local heating (thermal spike) and consequent reduction
in the compressive stress as the structure undergoes thermal relaxation. Also shown in Fig.6 is
a predicted curve based on a simple model (Davis, 1993) describing the variation of
compressive stress in bombarded thin films, based on the competing effects of implantation and
relaxation. Details are given elsewhere (Peng and Clyne, 1998). The level of agreement,
obtained using physically reasonable values for the parameters in the model, suggests that it
does encapsulate the most important features of the process.

3 Stresses in Thick Coatings produced by Droplet Spraying

3.1 Effects with Coatings of Significant Thickness

For a relatively thick coating, it may become inaccurate to assume that a misfit strain is
accommodated entirely within it. This applies to misfit strains arising from differential thermal
contraction, the deposition process or inelastic deformation such as might arise from creep,
plastic flow, micro-cracking etc. Furthermore, while the misfit strain from post-deposition
differential thermal contraction will be uniform within the coating, changes in curvature during
deposition may mean that it should be treated as if it took place by formation of a series of
discrete layers. In fact, an analytical model (Tsui and Clyne, 1997) is available to treat

page -7-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

deposition processes in this way, although it does involve repeated calculations, which are most
conveniently carried out using a computer program. A multi-layer approach within a full
numerical process model may also be advisable in some cases, particularly if the heat flow
taking place during spraying, and the effects of the associated through-thickness thermal
gradients, are to be incorporated. This also allows prediction of the curvature changes taking
place during deposition and subsequent cooling. Comparison between measured and predicted
curvature histories provides a powerful method of validating the input data and boundary
conditions employed (Gill and Clyne, 1994).

3.2 The Quenching Stress

When a droplet impinges on a substrate (or pre-deposited coating layer) during spray
deposition, it will tend to spread into a pancake-like splat, which will then quickly solidify and
cool. During this cooling, the thermal contraction of the splat will be inhibited by the
underlying material, so that a tensile stress will be set up in the splat. Since splats are typically
a few µm in thickness, it can always be assumed that they are on a massive substrate and the
misfit strain associated with the contraction is entirely accommodated within the splat. The so-
called Quenching Stress (Kuroda and Clyne, 1991) is therefore given by the product of the
misfit strain and the (biaxial) modulus of the deposit. However, while it is always tensile, the
value of the quenching stress can vary over a wide range. The nominal value of the misfit
strain is simply the product of the coating expansivity and the temperature drop on cooling
from the melting point to the substrate temperature. Corresponding stress levels are commonly
in the GPa range. However, quenching stress values are normally well below these levels: in
fact, they usually range from a few MPa to a few hundred MPa. There are two reasons for this.
Firstly, the stiffness of sprayed material is often appreciably lower than its handbook value, as a
consequence of the presence of porosity, micro-cracks etc. Secondly, it is common for stress
relaxation processes to come into operation during splat quenching. These may include micro-
cracking, plastic flow, creep, interfacial sliding etc. Some such effects are apparent in the data
presented in Fig.7, which shows measured quenching stress values, as a function of the
homologous temperature. At lower temperatures, all materials tend to show gradual increases
with temperature, which is attributed to improvements in inter-splat bonding. Micro-cracking
is common with many ceramics, leading to low quenching stress values. For metals, depending
on their strength, a reduction in quenching stress tends to occur on moving to relatively high
temperatures, as creep and plastic flow during quenching become pronounced. (Also, of
course, the nominal misfit strain tends to zero as the homologous temperature approaches
unity.)

3.3 Stress Distributions in Thermally Sprayed Systems

As an example of a layer-by-layer simulation of a thermal spray coating process, data are
presented in Fig.8 which refer to a thermal barrier system. These are commonly composed of
an oxidation-resistant metallic bond coat (an MCrAlY of some sort) and a ceramic top coat
(usually zirconia) offering a large thermal resistance. The substrates most commonly protected
in this way are nickel-based superalloys, in aero-engines, or various steels, in land-based
turbines or internal combustion engines. In Fig.8, a comparison is shown between measured
and simulated specimen curvature changes during spraying, with both bond coat and top coat
being divided into about 100 volume elements for the simulation. The periodicity in these plots
is associated with the movement of the spray gun, which involves a cycle of passes over the

page -8-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

specimen, with some cooling occurring between cycles. (Good agreement is also obtained
between measured and modelled thermal histories.)

For this specimen, through-thickness stress distributions were measured by neutron

diffraction, in order to provide an independent validation of the process model - see Fig.9.
The agreement is clearly quite good (although the resolution of the neutron diffraction
measurements was inadequate to confirm the presence of the high stress gradient through the
bond coat). Incidentally, this gradient arose mainly as a consequence of the relatively large
curvature changes occurring during the process, which were deliberately stimulated by having a
thin substrate. In practice, with a thicker (and perhaps non-planar) substrate, curvature changes
would much smaller, but of course the (validated) model could equally well be used to predict
stress distributions in such cases. Finally, it may be noted that the low stress levels in the top
coat are largely attributable to its low stiffness (due to micro-cracks etc), which confers a high
strain tolerance. Under service conditions, however, which involve the free surface of the top
coat reaching high temperatures, this stiffness may rise as a consequence of sintering
phenomena, leading to higher stresses and thus to a greater danger of debonding.

Concluding Points

In all cases, stresses arise primarily as a result of (a) the material being deposited initially in a
non-equilibrium state and (b) differential thermal contraction occurring between the coating
and the substrate, during post-deposition temperature changes. Both mechanisms can generate
either tensile or compressive stress in the coating. While differential thermal expansion
stresses are quite easy to predict, the mechanisms determining the deposition stresses can be
more complex. These can, however, be measured and in many cases they can be rationalised
and predicted, at least in a semi-quantitative manner. Further work is required in order to
understand fully how deposition stresses arise in different systems. The effect of residual
stresses on the thermo-mechanical stability of coatings is also an area requiring concerted
research effort over the coming years.

page -9-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

Bibliography

Brenner, A. and Senderoff, S., Calculation of Stress in Electrodeposits from the Curvature of a
Plated Strip, J. Res. Of Nat. Bur. of Stand., vol.42, (1949) p.105-123.

Clyne, T.W. and Gill, S.C., Residual Stresses in Thermally Sprayed Coatings and their Effect
on Interfacial Adhesion - A Review of Recent Work, J. Thermal Spray Technol., vol.5,
(1996) p.1-18.

Clyne, T.W., Residual Stresses in Surface Coatings and their Effects on Interfacial Debonding
Key Eng. Mat., vol.116/7 (1996) p.307-330.

Davis, C.A., A Simple Model for the Formation of Compressive Stress in Thin Film by Ion
Bombardment, Thin Solid Films, vol.226, (1993) p.30-34.

Drory, M.D., Thouless, M.D. and Evans, A.G., On the Decohesion of Residually Stressed Thin
Films, Acta Metall., vol.36, (1988) p.2019-2028.

Gill, S.C. and Clyne, T.W., Investigation of Residual Stress Generation during Thermal
Spraying by Continuous Curvature Measurement, Thin Solid Films vol.250, (1994)
p.172-180.

Kuroda, S. and Clyne, T.W., The Quenching Stress in Thermally Sprayed Coatings, Thin Solid
Films, vol.200, (1991) p.49-66.

Matejicek, J., Sampath, S., Brand, P.C. and Prask, H.J., Quenching, Thermal and Residual
Stress in Plasma Sprayed Deposits: NiCrAlY and YSZ Coatings, Acta Mater. vol.49,
(1999) p.607-617.

Mortensen, A. and Suresh, S., Functionally Graded Metals and Metal-Ceramic Composites:
Part I. Processing, Int. Mater. Rev. vol.40, (1995) p.239-265.

Peng, X.L. and Clyne, T.W., Formation and Adhesion of Hot Filamant CVD Diamond Films on
Titanium Substrates, Thin Solid Films vol.293, (1997) p.261-269.

Peng, X.L. and Clyne, T.W., Mechanical Stability of DLC Films on Metallic Substrates, Thin
Solid Films vol.312, (1998) p.207-218.

Salamon, N.J. and Masters, C.B., Bifurcation in Isotropic Thin Film/Substrate Plates, Int. J.
Solids & Structures, vol.32, (1998) p.473-481.

Stoney, G.G., The Tension of Metallic Films Deposited by Electrolysis, Proc. Roy. Soc.,
vol.A82, (1909) p.172-175.

Tsui, Y.C. and Clyne, T.W., An Analytical Model for Predicting Residual Stresses in
Progressively Deposited Coatings, Thin Solid Films vol.306, (1997) p.23-61.

page -10-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

Figure Captions

Fig.1 Schematic depiction of the generation of curvature in a flat bi-material plate, as a

result of the imposition of a uniform, linear misfit strain, ∆ε. The distributions of
stress and strain shown were calculated, using Eqns.(7) and (8), for H = h ,
Es = 100 GPa, Ed/Es = 0.1 and ∆ε = 10-3.

Fig.2 Predicted dependence (Clyne and Gill, 1996) on specimen width/length ratio of the
critical curvature at a bifurcation instability, for a residually stressed thin film on a
substrate. The parameter Γ is the ratio of the shear modulus of the deposit to that of
the substrate (Gd /Gs)

Fig.3 Predicted dependence of (a) specimen curvature and (b) stress levels on the ratio of
the thickness of the coating, h, to that of the substrate, H. The plots were obtained
using the exact relationships represented by Eqns.(3), (7) and (8) and using the
Stoney equation (Eqn.(6).) The Poisson ratios of substrate and deposit were both
taken as 0.2.

Fig.4 Thermal expansivity data (Peng and Clyne, 1997) for diamond on tungsten,
expressed as (a) α d (T) and αs(T) and (b) ∆εth(T2), obtained using eqn.(10), with the
polynomial expressions plotted in (a), for an ambient temperature, T0, of 25˚C.

Fig.5 Measured intrinsic (deposition) stresses (Peng and Clyne, 1997) in diamond films
on tungsten substrates, obtained by curvature measurement at room temperature and
subtraction of the thermal stress, as a function of (a) methane concentration, at a
temperature of 825˚C, and (b) temperature, at 1.0% CH4 concentration.

Fig.6 Comparison between the intrinsic stress levels measured experimentally, for DLC
films deposited at 10 Pa pressure, with various negative bias voltages, and the
predicted curve obtained using the Davis model (from Peng and Clyne, 1998)

Fig.7 Experimental quenching stress data (Kuroda and Clyne, 1991) for plasma spraying
in air (APS) or in vacuum (VPS). The values are plotted against the ratio of the
specimen temperature to the melting temperature of the deposit. The data were
obtained from specimen curvature measurements made during spraying.

Fig.8 Comparison between (a) an experimental curvature history obtained during

spraying, and (b) the corresponding prediction from the numerical process model,
for a specimen composed of a CoNiCrAlY bond coat and a ZrO2-8wt%Y2 O3 top
coat, deposited onto a mild steel substrate.

Fig.9 Comparison between stress distributions obtained experimentally (using neutron

diffraction) and by numerical process modelling, for the specimen referred to in
Fig.8.

page -11-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

Figures

y
z
b
x
∆ε (eg ∆α ∆T) h
force
balance
b σ(y) dy = 0
-H
ε (millistrain) σ (MPa)
-1 1 -20 20
0 0
-P -P y=h
y=0

P P y=-H

-1 1 -20 20
0 0
-P -P

M yc =0 y =δ
P M P
h-δ
moment
1/κ balance b σ(y c ) yc dyc = 0

-H-δ

Fig.1 Schematic depiction of the generation of curvature in a flat bi-material plate, as a

result of the imposition of a uniform, linear misfit strain, ∆ε. The distributions of
stress and strain shown were calculated, using Eqns.(7) and (8), for H = h ,
Es = 100 GPa, Ed/Es = 0.1 and ∆ε = 10-3.

page -12-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

100

L =50 mm, H =1 mm, Γ =1

Curvature at bifurcation, κB (m- 1)

L=100 mm, H =1 mm, Γ =1

L =100 mm, H =1 mm, Γ =0.1

10

νs = νd = 0.3
h / H = 0.1

1
0 0.2 0.4 0.6 0.8 1
Width / length ratio of substrate (b /L )

Fig.2 Predicted dependence (Clyne and Gill, 1996) on specimen width/length ratio of the
critical curvature at a bifurcation instability, for a residually stressed thin film on a
substrate. The parameter Γ is the ratio of the shear modulus of the deposit to that of
the substrate (Gd /Gs)

page -13-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

100
-3 (a)
∆ε = 10
E = 100 GPa
s
E = 100 GPa
d
10
Curvature, κ (m - 1)

H = 0.1 mm

Exact
Stoney
0.1
Exact (Ed = 1000 GPa)
Stoney (Ed = 1000 GPa)

0.01 - 3
10 10- 2 10- 1 100
Thickness ratio, h/H

(b)
1
Stress in deposit, σd (GPa)

-3
∆ε = 10
E s = 100 GPa
E d = 100 GPa

0.1
Exact, at interface
Exact, at surface
Stoney
Exact, Ed =1000 GPa, at interface
Exact, Ed=1000 GPa, at surface
Stoney E d=1000 GPa

0.01 - 3 -2 -1 0
10 10 10 10
Thickness ratio, h/H

Fig.3 Predicted dependence of (a) specimen curvature and (b) stress levels on the ratio of
the thickness of the coating, h, to that of the substrate, H. The plots were obtained
using the exact relationships represented by Eqns.(3), (7) and (8) and using the
Stoney equation (Eqn.(6).) The Poisson ratios of substrate and deposit were both
taken as 0.2.

page -14-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

6
Thermal expansivity, α (K- 1 × 10− 6)

1.2

Thermal misfit strain, ∆ ε t h (× 10- 3 )

(a) (b)
5 1

0.8
4
0.6
3
0.4
2 Diamond data points
Tungsten data points 0.2
Diamond fitted curve
1 0
Tungsen fitted curve

0 -0.2
0 200 400 600 800 0 200 400 600 800
Temperature, T (˚C) Deposition temperature, T2 (˚C)

Fig.4 Thermal expansivity data (Peng and Clyne, 1997) for diamond on tungsten,
expressed as (a) α d (T) and αs(T) and (b) ∆εth(T2), obtained using eqn.(10), with the
polynomial expressions plotted in (a), for an ambient temperature, T0, of 25˚C.

page -15-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

1 1
(a) (b)
0.8 0.8
Intrinsic stress (GPa)

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 700 750 800 850 900
Methane concentration (%) Deposition temperature (˚C)

Fig.5 Measured intrinsic (deposition) stresses (Peng and Clyne, 1997) in diamond films
on tungsten substrates, obtained by curvature measurement at room temperature and
subtraction of the thermal stress, as a function of (a) methane concentration, at a
temperature of 825˚C, and (b) temperature, at 1.0% CH4 concentration.

page -16-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

-2.5
Experimental data
Davis model
-2
Intrinsic stress, σi (GPa)

-1.5

-1

-0.5

0
0 50 100 150
Ion energy, E = 0.5 e V b (eV)

Fig.6 Comparison between the intrinsic stress levels measured experimentally, for DLC
films deposited at 10 Pa pressure, with various negative bias voltages, and the
predicted curve obtained using the Davis model (from Peng and Clyne, 1998)

page -17-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

APS VPS
Ni-20Cr
Al
Ni
Quenching Stress σ (MPa)

2 Ni-20Cr Mo Mo
10
Ni-20Cr
Ni
Alumina
q

Mo
1
10
Alumina Al

Ni

1
0 0.2 0.4 0.6 0.8 1
Homologous Temperature

Fig.7 Experimental quenching stress data (Kuroda and Clyne, 1991) for plasma spraying
in air (APS) or in vacuum (VPS). The values are plotted against the ratio of the
specimen temperature to the melting temperature of the deposit. The data were
obtained from specimen curvature measurements made during spraying.

page -18-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

2.5 2.5

2.0 2.0

1.5 1.5
Curvature (m- 1)

top

Curvature (m- 1)
top
coat coat

1.0 1.0

0.5 0.5

0.0 0.0
bond bond
coat (a) coat
(b)
-0.5 -0.5
0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500
Time (s) Time (s)

Fig.8 Comparison between (a) an experimental curvature history obtained during

spraying, and (b) the corresponding prediction from the numerical process model,
for a specimen composed of a CoNiCrAlY bond coat and a ZrO2-8wt%Y2 O3 top
coat, deposited onto a mild steel substrate.

page -19-
 Clyne – 4.1.3b Residual Stresses in Thick and Thin Coatings

300
substrate bond top coat
coat
200
Stress (MPa)

100

-100
Modelled
Experimental

-200
-2.0 -1.0 0.0 1.0 2.0 3.0
Z (mm)

Fig.9 Comparison between stress distributions obtained experimentally (using neutron

diffraction) and by numerical process modelling, for the specimen referred to in
Fig.8.

page -20-

View publication stats