Full text of DOI:10.1557/mrs2002.18

www.mrs.org/publications/bulletin
creep in thin films. We then discuss stress
development during film growth, micro-
structural stability during creep tests, and
the relationships between applied stresses,
strain rates, microstructures, and deforma-
tion mechanisms. We next outline zero-
creep stresses and transient strains that must
be considered when evaluating creep prop-
erties, noting their impact on thin-film creep
tests. Finally, we discuss creep deformation
of films on substrates and freestanding
multilayer films.
Diffusional Creep:
Stresses and Strain
Rates in Thin Films
TestTechniques
As with bulk specimens, simple tension
tests are one technique for studying creep
properties. Films can be released from
their substrates and tested in tension as
completely freestanding samples2,3 or in
bending as cantilever structures.4 Such
tests involve challenging sample prepara-
tion and handling. However, they permit
accurate, independent control of applied
stress and test temperature during creep
studies.
and Multilayers
D. Josell, T.P.Weihs, and H. Gao
Abstract
In this article, we discuss creep deformation as it relates to thin films and multilayer
foils. We begin by reviewing experimental techniques for studying creep deformation
in thin-film geometries, listing the pros and cons of each; then we discuss the use of
deformation-mechanism maps for recording and understanding observed creep
behavior. We include a number of cautionary remarks regarding the impact of
microstructural stability, zero-creep stresses, and transient-creep strains on stress–strain
rate relationships, and we finish by reviewing the current state of knowledge for creep
deformation in thin films. This includes both thin films that are heated on substrates as
well as multilayer films that are tested as freestanding foils.
More commonly, creep deformation in
thin films is studied while they remain on
a substrate that deforms only elastically.5
These studies utilize film stresses that arise
due to mismatched thermal expansions
or contractions of the substrate and film.
Since these samples require little prepara-
tion other than deposition, this geometry
is convenient. However, it does not permit
independent control of temperature, stress,
or strain. As a result, creep studies at fixed
stresses are not possible.
A hybrid test method that falls midway
between these two techniques involves
tensile tests of films on substrates that can
deform plastically. This method reduces
the coupling of film stress, strain, and
temperature while easing the handling of
samples.6 However, temperatures in such
tests are limited by the polymeric nature
of the substrates. In nanoindentation creep
tests, one can control loading.7–10 The com-
plex nature of the stress state under the in-
denter tip requires significant modeling,
however, which introduces significant un-
certainties into the values of materials
properties thus obtained.
Keywords: diffusional creep, mechanical properties, thin films.
Introduction
In studying the mechanical properties
deformation as it relates to homogeneous
films on substrates and stacked thin films,
or so-called multilayers.
of materials, most researchers emphasize
yield or fracture properties and test for these
quantities using simple tension tests and
bend tests. However, while yield strength
and fracture toughness are useful proper-
ties for designing materials and structures,
they are rarely experienced in applications
involving thin films and multilayers.
On the other hand, many materials
do exhibit time-dependent deformation,
or creep, under stresses well below those
needed to cause yielding or fracture. This
time-dependent deformation occurs by
means of a variety of deformation proc-
esses. In thin-film applications, creep de-
formation is often driven by stresses that
arise from mismatched thermal expansions
of a film and the substrate on which it is
deposited. Stress relaxation by means of
creep deformation is ubiquitous in the
processing of multilevel, thin-film devices
that require multiple processing steps at
numerous temperatures. Because rates for
several creep mechanisms increase rapidly
as grains get smaller, they cannot be ig-
nored in thin-film structures, even at room
temperature. We therefore discuss creep
As noted in other articles in this issue,
atomistic simulations have become an
important resource for modeling a large
number of deformation processes. How-
ever, even with recent increases in compu-
tational speed, simulations are limited to
extreme cases of total strain or strain rate
that do not yet yield useful predictions in
the case of creep deformation. To predict
plastic strains of a realistic magnitude, one
is restricted to strain rates of 107 s or
ꢀ1
higher1 that are many orders of magnitude
greater than experimental strain rates. Al-
ternatively, if realistic strain rates are used
Regardless of the experimental technique
used, the goal of thin-film creep studies is
to experimentally measure creep rates and
identify creep mechanisms for a range of
applied stresses and temperatures using
specimens that are well defined chemi-
cally and microstructurally.
ꢀ5
ꢀ1
(10 –10ꢀ9 s ), creep strains are limited to
magnitudes (ꢁ10ꢀ12) that are far too small
to be informative. Therefore, this article
focuses on experimental techniques and
analytical models for characterizing creep
deformation in thin films.
In keeping with the goal of this issue of
MRS Bulletin, we emphasize areas where
unique issues arise due to the thin-film
geometries being addressed. We first note
some of the techniques used to study
Creep ofThin Films
Creep during Film Growth
Creep deformation often first appears
during the growth of thin films due to
the development of large stresses during
MRS BULLETIN/JANUARY 2002
39

Diffusional Creep: Stresses and Strain Rates in Thin Films and Multilayers
deposition. The significant role of diffu-
sional processes in the relaxation of these
stresses is discussed in the article in this
issue by Floro et al.
Stability of theThin-Film Structure
Since most thin films are deposited with
fine microstructures and metastable phases,
both chemical and microstructural stabili-
ties are major concerns when character-
izing creep rates and mechanisms. Creep
conditions are usually chosen to avoid con-
ditions where phase transformations or
substantial microstructural evolution (e.g.,
the breakdown of a multilayer into dis-
continuous layers) occur very rapidly. Other-
wise, interpretation of creep properties is
complicated by the need to correlate the
measured properties with the changing
microstructure of the system.
Assuming the phases in thin films and
multilayers are chemically stable, diffu-
sional processes play a significant role in
determining the stability of their micro-
structures because they control the kinetics
of grain growth, grain-boundary grooving
(see Figure 1), and pit formation at triple
junctions. The energetics and kinetics of
microstructural stability in polycrystalline
thin films on substrates can be found in
several works,11–15 as can studies of stabil-
ity in multilayer specimens.16–20 Generally,
a thin film is predicted to be more stable
energetically if it has grain boundaries
with free energies that are low compared
with the free energy of the film surface.
Similarly, a multilayer is predicted to be
more stable energetically if it has grain
boundaries with free energies that are low
compared with the free energy of the inter-
faces between layers. Energetic stability
should be enhanced if in-plane grain sizes
are not significantly larger than the film or
layer thickness; stability should thus bene-
fit from stagnation of in-plane coarsening.
Kinetically, instabilities such as deep pits
or grooves will form more slowly when
diffusional transport is slow.
Figure 1. Cross section of a Ag/Ni multilayer on a thin sapphire substrate after 24 h of
biaxial zero-creep testing at 500ꢆC.20 The image was obtained using secondary electron
imaging after focused-ion-beam milling. Note the significant grooving into the Ni layers
where the Ni grain boundaries meet the Ag/Ni interfaces. (Imaging and milling courtesy of
A. Wilson and G. Spanos, Naval Research Laboratory, who gratefully acknowledge support
from the Office of Naval Research.)
volume, G is the shear modulus; k is the
Boltzmann constant; T is temperature; b is
the Burgers vector; and d is grain size. Dif-
ferent mechanisms control the creep de-
formation of a given material at different
stresses and temperatures. Most broadly,
at high stresses dislocation motion con-
trols deformation, and power-law (dis-
location) creep dominates; at low stresses
atomic diffusion controls deformation, and
diffusional creep dominates. Dislocation
creep is controlled by dislocation glide at
low homologous temperatures and by
dislocation climb at high homologous
temperatures. Similarly, diffusional-creep
processes are controlled by diffusion
along grain boundaries (Coble creep) at
low homologous temperatures and by
diffusion through the bulk of grains
(Nabarro–Herring creep) at high tempera-
tures. Typical values for A, m, and n are
given in Table I for the different mecha-
nisms. Note that the diffusional-creep
mechanisms exhibit a strong grain-size
dependence, indicated by the nonzero
value of n, while the power-law-creep
mechanisms do not. Diffusional creep is
thus expected to dominate deformation
in thin films with sufficiently small grain
sizes. Considering in-plane grain size and
film thickness as separate length scales22
usually does not change this because in-
plane grain size typically scales with film
thickness.
Creep-Deformation Mechanisms
Creep behavior in bulk or thin-film ma-
terials can be summarized by an empirical
equation of the form21
m
n
dꢂ
dt
D
ꢅ2/3
ꢃ
ꢃ
ꢅ
b
d
ꢂ˙ ꢀ
ꢀ A
,
(1)
ꢁ ꢂꢁ ꢂꢁ ꢂ
G
kT
describing the dependence of the creep
rate ꢂ˙ of a material subject to a fixed ap-
plied stress ꢃ. The power of that depend-
ence is m ꢄ 1; A is a coefficient that
depends on temperature and microstruc-
ture; D is a coefficient for diffusion through
the bulk of a grain, along grain bound-
aries, or along dislocation cores; ꢅ is atomic
In an effort to help identify the domi-
nant creep mechanisms over the full range
of achievable stresses and temperatures,
Frost and Ashby developed the concept of
40
MRS BULLETIN/JANUARY 2002

Diffusional Creep: Stresses and Strain Rates in Thin Films and Multilayers
for t ꢀ 10 nm. In situ transmission electron
Table I: Parameters for the Generalized Creep Equation.
microscope studies of creep of gold thin
films found that inclusion of a zero-creep
stress ꢃ0 of the order of 100 MPa explained
why creep rates were ꢄ10,000ꢉ smaller
than those predicted for the specimen
geometry and test conditions.27 Similarly,
substrates on which multilayer films are
deposited have been shown to relax to a
nonzero equilibrium curvature due to the
interfaces in the multilayer film.28 Account-
ing for zero-creep stresses can be neces-
sary for accurate interpretation of creep
data of thin films and multilayers.
Mechanism
Conditions
A*
m*
n*
Nabarro–Herring creep
Coble creep
High temperature, low stress
Moderate temperatures,
low stress, fine grain sizes
Low to moderate temperatures,
high stress
7
50
0
0
2
3
Power-law creep
**
2–6
0
*A is a coefficient that depends on temperature and microstructure, while m and n are empirically
derived constants.
**The values of A vary dramatically, depending on the microstructure that is controlling the
power-law creep.
Steady-State versusTransient Creep
Creep rates depend on the magnitude
of the cumulative creep and thus are not
always constant for a given applied load
or stress. This effect is well known for bulk
structural materials at high temperatures
(T ꢊ 0.5Tm), where creep has been studied
for many decades. As is shown in Figure 4
for an FeCo intermetallic alloy thin film,29
creep data are usually separated into three
deformation-mechanism maps.23 A defor-
mation map for Ni is shown in Figure 2.
These maps are based on diffusion data
as well as experimental data from creep
studies. Different maps are required for
materials with different grain sizes. This
particular map predicts the controlling
creep mechanisms for Ni with a 1.0-ꢇm
grain size. The data used to predict
many of these maps, particularly for the
diffusional-deformation mechanisms, are
frequently limited in quantity. As an ex-
ample, Frost and Ashby based the Coble
creep fields of their deformation maps for
the fcc metals upon diffusion data rather
than experimental creep data. The cor-
responding Nabarro–Herring creep fields
are based on limited creep data at high
temperatures. Predictions of such defor-
mation maps for Coble and Nabarro–
Herring creep-deformation fields should
thus be considered qualitative rather than
quantitative, particularly for materials
with grain sizes significantly smaller than
those available when the maps were de-
veloped. As Frost and Ashby noted, “The
maps are no better (and no worse) than
the equations and data used to construct
them.”23
ure 3b), and thus the zero-creep stress ꢃ0,
can be determined. The value of ꢃ0 can
also be derived through equilibrium thermo-
dynamics arguments. An approximate value
can be obtained from ꢃ0 ꢄ ꢈ/t, where ꢈ is
the free energy of the surface (thin film) or
interface (multilayer) and t is the thickness
of the film or layer. Using ꢈ ꢄ 1 J/m, ꢃ0 is
only 1 MPa for t ꢀ 1 ꢇm, but it is 100 MPa
Zero-Creep Stress
A creep property that is unique to small
structures such as thin films and multilayers
is the existence of a significant range of
stresses for which a positive stress results
in a negative strain rate. This occurrence,
described empirically by an expression of
the type
ꢂ˙ ꢃ (ꢃ ꢀ ꢃ0),
(2)
is a manifestation of the drive to reduce
the area, and thus the free energy, of the
surfaces of a thin film11 or the interfaces
in a multilayer.24–26 Figure 3 (from Refer-
ence 25) shows experimental creep data
from Ag/Ni multilayers at different ap-
plied stresses (Figure 3a) from which
strain rate versus stress-dependence (Fig-
Figure 2. A deformation-mechanism map for Ni composed of 1.0-ꢇm grains (from
Reference 23). The diffusional mechanisms are predicted to dominate creep deformation
for a large range of test temperatures and applied stresses due to the small grain size. ꢇ is
the shear modulus.
MRS BULLETIN/JANUARY 2002
41

Diffusional Creep: Stresses and Strain Rates in Thin Films and Multilayers
distinct stages for both bulk and thin-film
materials tested in uniaxial tension. Dur-
ing the initial or primary creep stage, strain
rates (the local slope of the strain–time
curve) usually drop sharply as the mate-
rial’s microstructure (dislocation density,
grain shape, etc.) evolves toward a steady
state, consistent with the applied load. In
the second, or steady-state, stage, micro-
structure and strain rate remain relatively
constant. In the final, or tertiary, stage, the
material’s microstructure breaks down
through the formation of internal voids
and cracks, and the global strain rate rises
rapidly. Experimental studies of bulk ma-
terials typically analyze the steady-state
creep stage, and most theories of creep-
deformation predict creep rates for this
stage. To date, most researchers have ap-
plied steady-state creep concepts to the
study of thin films. Such application can
be misleading.
At 150–260ꢆC and 21 MPa of tension,
the primary creep stage for bulk Al en-
compasses a plastic strain of 18%.30 In con-
trast, the maximum total strain, plastic and
elastic, that can be induced in an Al film
on a Si substrate by heating from 25ꢆC to
500ꢆC is approximately 1%. Thus, almost
all of the creep deformation in the Al film
might more appropriately be described
by primary creep behavior rather than
steady-state creep behavior, particularly at
lower temperatures.
occurs in the film by the dominant creep
mechanism while the substrate remains
elastic. Stress–temperature plots generally
predict dislocation-creep mechanisms to
dominate at the higher stresses and
diffusional-creep mechanisms to dominate
at the lower stresses (Figure 5). For 1-ꢇm-
thick Cu films on Si substrates, stresses are
predicted to rapidly relax by diffusional
creep as the temperature increases (Fig-
ure 5). Measured stresses do not exhibit
this behavior.32,33
Recent investigations of diffusional creep
in Cu films produced by sputtering and
annealing under ultrahigh-vacuum condi-
tions found that the stress–temperature
curve of the pure Cu films differs signifi-
cantly from that of Cu-1at.%Al alloy films,
where Al surface segregation and oxida-
tion leads to self-passivation.34,35 Assuming
that the diffusional creep is suppressed by
surface passivation in the Cu-1at.%Al alloy
films, the difference between these alloy
films and the pure Cu films could be
quantitatively explained by a constrained
diffusional-creep model,36 in which atoms
are incorporated into grain boundaries via
surface and grain-boundary diffusion with
no sliding and no diffusion at the film/
substrate interface. Under these assump-
tions, grain-boundary diffusion relaxes the
stresses near grain boundaries, but leaves
a significant level of average stress in
the film.
Figure 3. (a) Creep deformation
for a Ag/Ni multilayer composed of
15 bilayers; total film thickness is 20 ꢇm
and width is 1 cm (from Reference 25).
The applied load was changed between
0.1 N and 0.3 N (corresponding to 10 g
and 30 g) every ꢃ104 min. Negative
strain rates were recorded at 0.1 N.
(b) Strain rate versus applied load for
five Ag/Ni multilayer specimens, each
with 21 bilayers; total film thickness
is 24 ꢇm and width is 1 cm (from
Reference 25). The applied load at
which the strain rate would be zero
is ꢃ0.23 N (23 g) for each of the
five specimens.
In contrast, tensile studies of freestand-
ing films allow observation of all stages
of creep, although relatively few studies
have included analysis of the primary
creep stage data (see, e.g., Reference 31).
Tensile studies of freestanding Ag/Ni
multilayers show that a primary creep
stage of 0.1–0.2% precedes the secondary
creep stage when the applied stress is
changed by ꢋ1 MPa in the temperature
range of 650–800ꢆC (as in Figure 3a).25 The
primary creep stage is evident in the
steeper slope at the beginning of the curve
after each change in applied stress. The
magnitude of this strain transient, which
is the same for both negative strain rates
and positive strain rates, is 1000ꢉ larger
than the instantaneous elastic strains asso-
ciated with the stress changes. One might
expect this to be particularly significant
for film-on-substrate curvature experi-
ments, in which stresses typically vary by
hundreds of MPa throughout the experi-
ments, while strains are limited to 1–2%.
Creep Rates of Multilayers
As with thin films, stress and strain
evolution in multilayer materials during
thermal cycling has been modeled by as-
suming that creep deformation occurs
simultaneously in each layer by the domi-
nant creep mechanisms.37–39 The creep be-
havior is generally complicated, as well as
history-dependent, with different defor-
mation mechanisms dominating in the
different layers at different temperatures.
In constant-temperature tension experi-
ments, on the other hand, the creep behav-
ior is expected to be more straightforward.
Upon loading or heating, the less creep-
resistant layer strains plastically and in the
process sheds most of the stress that acts
upon it to the more creep-resistant layer.
Changing elastic strains arising from the
changing stresses accommodate the differ-
ent plastic-strain rates in the layers. This
load redistribution continues until, in the
steady state, both layers are creeping at
the same rate, the slower creeping mate-
rial now bearing the majority of the ap-
plied load. The transient associated with
this stress redistribution was noted for
Ag/Ni multilayers in the previous section
on “Steady-State versus Transient Creep.”
Predictions concerning steady-state creep
Figure 4. Cumulative strain versus
time exhibited by a 150-ꢇm-thick FeCo
intermetallic alloy thin film at 600ꢆC with
an applied stress of 305 MPa (from
Reference 29). Note: even though there
are three distinct stages, only the
middle, steady-state creep, stage is
typically considered.
Wafer Curvature: Films on
Substrates
When thin films on substrates are heated,
stresses can rise rapidly. Stress and strain
evolution during thermal cycles is mod-
eled by assuming that plastic deformation
42
MRS BULLETIN/JANUARY 2002

Diffusional Creep: Stresses and Strain Rates in Thin Films and Multilayers
Figure 6. Creep deformation for a
Ag8/Ni8 multilayer, with layers 2 nm
thick (from Reference 41). The applied
force was held constant while the
temperature was cycled. An applied
force of 0.01 N (stress ꢃ0.7 MPa)
yielded the solid curve in which the
film contracts with each thermal cycle.
An applied force of 0.3 N yielded the
dotted curve that manifests rapid
creep associated with breakdown of
the layered structure at the higher
temperatures. Structures maintain
their layering in region I, but lose it
in region II.
Figure 5. Predicted stress–temperature curves for 1-ꢇm-thick copper films on Si substrates
with dominant creep mechanisms indicated (from Reference 32). Heating rate affects the
deformation mechanism through its impact on stress relaxation. Note the near-complete
stress relaxation at the elevated temperatures.
rates are now compared to experimental
creep results, also for Ag/Ni multilayers.
Using the properties of the slower
creeping Ni in the expression for Coble
creep,23
ignored (Table II). Including the zero-
creep stresses makes all experimental and
predicted creep rates agree in sign. How-
ever, predicted strain rates are still far
higher than those measured experimen-
tally, regardless of the omission (inclusion)
of the zero-creep stress: 5 (5) orders of
magnitude for the ꢃ1-ꢇm-thick layers
and 7 (10) orders of magnitude for the
ꢃ1-nm-thick layers.
Note that the sign of the creep rate is
determined by energetics, that is, whether
contraction or expansion lowers the free
energy of the system. The magnitude of the
creep rate is a manifestation of the kinetics,
that is, how quickly matter moves in re-
sponse to the forces driving creep. The fact
that experimental and predicted creep rates
agree in sign only when zero-creep stresses
are accounted for confirms the impact of
these stresses on the creep of thin films
and layers. The substantial disagreement in
the magnitudes of experimental and pre-
dicted creep rates, which worsens as grains
decrease in size, demonstrates the poten-
tial for user error if creep-deformation maps
are taken as gospel. Creep studies that cor-
relate stress, strain rate, and temperature
for thin-film samples with small but stable
grain sizes are evidently needed.
42ꢅꢌDb ꢍ
kT d3
ꢂ˙ ꢀ
ꢅꢃ
ꢀ
ꢃ0ꢆ,
(3)
where the zero-creep stress ꢃ0 has been
added,25 one can predict the strain rate ꢂ˙.
Predictions are compared with the experi-
mental data from two groups in Table II.
The first group studied Ag/Ni multilayers
with layers of the order of 1 ꢇm in thick-
ness.25 Note the negative strain rates for the
lower tensile stress. The values in Table II
are for the creep data in Figure 3a. The sec-
ond group studied Ag/Ni multilayers with
layers of the order of 1 nm in thickness.40,41
Thermal-cycling creep results for multi-
layers that had only eight atomic planes
per elemental layer are shown in Figure 6.
Note the shrinkage during thermal cycling
for the tensile load.41 Table II includes an
approximate strain rate from constant-
temperature creep studies with this speci-
men.41 The zero-creep stresses used in
Table II come from the creep data in Fig-
ure 3a for the ꢃ1-ꢇm-thick layers and
from ꢈ/t for the ꢃ1-nm thick layers.
Table II: Comparison of Experimental and Calculated Strain Rates for
Ag/Ni Multilayers.
ꢀ
ꢁ˙calc. (sꢂ1
)
ꢀ – ꢀ0
ꢁ˙calc. (sꢂ1
)
d
T
Experimental Data
ꢁ˙exp. (sꢂ1
)
(MPa) for ꢀ0 ꢃ 0 (MPa) including ꢀ0 (nm)
(K)
Layer thickness of
ꢀ8 ꢉ 10ꢀ9
0.5
1.5
2 ꢉ 10ꢀ4 ꢀ0.5
6 ꢉ 10ꢀ4
0.5
ꢀ2 ꢉ 10ꢀ4 670
ꢄ2 ꢉ 10ꢀ4
925
the order of 1 ꢇm ꢄ8 ꢉ 10ꢀ9
(Reference 25)
Layer thickness of
the order of 1 nm
(Reference 40)
ꢀ1 ꢉ 10ꢀ8 sꢀ1
0.7
2.0 ꢉ 10ꢀ1 ꢀ500
ꢀ1.5 ꢉ 102
2
525
Notes: Comparison of experimental strain rates with predictions obtained using deformation
mechanism map parameters for Ni with grain size d.23 Inclusion of zero-creep stress leads to
agreement in sign of predicted and experimental strain rates. It does not rectify the difference in
rates. ꢂ˙exp. is the experimentally derived strain rate, ꢃ is the applied stress, ꢂ˙calc. is the calculated
strain rate, ꢃo is the zero-creep stress, and T is temperature.
The experimental data for the two
Ag/Ni multilayers are inconsistent in both
sign and magnitude with the values that
are predicted when zero-creep stresses are
MRS BULLETIN/JANUARY 2002
43

Diffusional Creep: Stresses and Strain Rates in Thin Films and Multilayers
Res. 14 (1999) p. 2373.
7. A.G. Atkins, A. Silverio, and D. Tabor, J. Inst.
Met. 94 (1966) p. 369.
8. S.N.G. Chu and J.C.M. Li, J. Mater. Sci. 12
(1977) p. 2200.
9. D. Stone, W.R. Lafontaine, P. Alexopoulos,
and C.Y. Li, J. Mater. Res. 3 (1988) p. 141.
10. V. Raman and R. Berriche, J. Mater. Res. 7
(1992) p. 627.
11. M.D. Thouless, Acta Metall. Mater. 41 (1993)
p. 1057.
12. K.T. Miller, F.F. Lange, and D.B. Marshall,
J. Mater. Res. 5 (1990) p. 151.
13. F.Y. Genin, W.W. Mullins, and P. Wynblatt,
Acta Metall. Mater. 40 (1992) p. 3239.
14. D.J. Srolovitz and S.A. Safran, J. Appl. Phys.
60 (1986) p. 247.
15. D.J. Srolovitz and S.A. Safran, J. Appl. Phys.
60 (1986) p. 255.
16. D. Josell and W.C. Carter, in Creep and Stress
Relaxation in Miniature Structures and Components,
edited by H.D. Merchant (The Minerals, Metals,
and Materials Society, Warrendale, PA, 1996)
p. 271.
17. D. Josell, S.R. Coriell, and G.B. McFadden,
Acta Metall. Mater. 43 (1995) p. 1987.
18. J.P. Fain, R. Banerjee, D. Josell, P.M. Anderson,
H. Fraser, N. Tymiak, and W. Gerberich, in
Nanophase and Nanocomposite Materials III, ed-
ited by S. Komarneni, J.C. Parker, and H. Hahn
(Mater. Res. Soc. Symp. Proc. 581, Warrendale,
PA, 2000) p. 603.
41 (1993) p. 3007.
25. D. Josell and F. Spaepen, Acta Metall. Mater.
41 (1993) p. 3017.
26. D. Josell and Z.L. Wang, in Thin Films: Stresses
and Mechanical Properties V, edited by S.P. Baker,
C.A. Ross, P.H. Townsend, C.A. Volkert, and
P. Børgesen (Mater. Res. Soc. Symp. Proc. 356,
Pittsburgh, 1995) p. 357.
27. K.E. Harris and A.H. King, Acta Mater. 46
(1998) p. 6195.
28. A.C. Lewis, A.B. Mann, D. van Heerden,
D. Josell, and T.P. Weihs, in Influences of Interface
and Dislocation Behavior on Microstructure Evolu-
tion, edited by M. Aindow, M.D. Asta, M.V.
Glazov, D.L. Medlin, A.D. Rollet, and M. Zaiser
(Mater. Res. Soc. Symp. Proc. 652, Warrendale,
PA, 2001) p. Y1.3.1.
29. C.H. Shang, R.C. Cammarata, C.L. Chien, and
T.P. Weihs, Acta. Mater. (2001) to be submitted.
30. O.D. Sherby and P.M. Burke, Prog. Mater.
Sci. 13 (1968) p. 325.
31. F.R. Brotzen, C.T. Rosenmayer, C.G. Cofer,
and R.J. Gale, Vacuum 41 (1990) p. 1287.
32. M.D. Thouless, J. Gupta, and J.M.E. Harper,
J. Mater. Res. 8 (1993) p. 1845.
33. R.-M. Keller, S.P. Baker, and E. Arzt, Acta
Mater. 47 (1999) p. 415.
34. E. Arzt, G. Dehm, P. Gumbsch, O. Kraft,
and D. Weiss, Prog. Mater. Sci. 46 (2001) p. 283.
35. D. Weiss, H. Gao, and E. Arzt, Acta Mater. 49
(2001) p. 2395.
36. H. Gao, L. Zhang, W.D. Nix, C.V. Thompson,
and E. Arzt, Acta Mater. 47 (1999) p. 2865.
37. Y.-L. Shen and S. Suresh, Acta Metall. Mater.
43 (1995) p. 3915.
38. Y.-L. Shen and S. Suresh, J. Mater. Res. 10
(1995) p. 1200.
39. Y.-L. Shen and S. Suresh, Acta Mater. 44 (1996)
p. 1337.
40. V. Pelosin and J. Hillairet, Nanostruct. Mater.
4 (1994) p. 229.
41. V. Pelosin, J. Hillairet, and B. Rodmacq,
J. Phys.: Condens. Matter 6 (1994) p. 1099. Strain
rate obtained from creep data in Figure 8.
Conclusion
We have outlined the role that diffusional
processes play in the creep deformation of
thin films. In doing so, we have tried to
highlight some of the difficulties encoun-
tered in applying concepts originally de-
rived for bulk materials to thin films and
multilayers. We have seen that the inclu-
sion of a zero-creep stress is necessary for
analyzing and predicting the sign of creep
rates. These same results have made clear
the shortcomings of extending deformation-
mechanism maps, which are based on bulk
properties, to multilayer geometries for the
prediction of actual creep rates. We have
also discussed the existence and magnitude
of transients associated with the primary
creep regime as well as the implications
for thermal-cycling studies of elemental
thin films adhering to substrates. In sum-
mary, extreme caution should be used in
applying the general concepts of steady-
state creep behavior of bulk materials to
the analysis of thin-film deformation at
elevated temperatures.
Acknowledgment
This work is funded in part by DOE/
OBES grant No. DE-FG-0297-ER45664 and
AFOSR award No. F49620-00-1-0028.
19. D. Josell, W.C. Carter, and J.E. Bonevich,
Nanostruct. Mater. 12 (1999) p. 387.
20. A.C. Lewis, A.B. Mann, D. Josell, J. Tapson,
and T.P. Weihs, in Interfacial Engineering for Op-
timized Properties II, edited by C.B. Carter, E.L.
Hall, S.R. Nutt, and C.L. Briant (Mater. Res. Soc.
Symp. Proc. 586, Warrendale, PA, 2000) p. 249.
21. T.H. Courtney, Mechanical Behavior of Mate-
rials (McGraw-Hill, New York, 2000).
22. G.B. Gibbs, Philos. Mag. 13 (1966) p. 589.
23. H.J. Frost and M.F. Ashby, Deformation-
Mechanism Maps: The Plasticity and Creep of Metals
and Ceramics (Pergamon Press, New York, 1982).
24. D. Josell and F. Spaepen, Acta Metall. Mater.
References
1. J. Schiøtz, T. Vegge, F.D. Di Tolla, and K.W.
Jacobsen, Phys. Rev. B 60 (1999) p. 11971.
2. W.N. Sharpe, B. Yuan, and R.L. Edwards,
J. Microelectromech. Sys. 6 (1997) p. 193.
3. O. Kraft and C.A. Volkert, Adv. Eng. Mater. 3
(2001) p. 99.
4. T.P. Weihs, S. Hong, J.C. Bravman, and W.D.
Nix, J. Mater. Res. 3 (1988) p. 931.
5. W.D. Nix, Metall. Trans. A 20A (1989) p. 2217.
6. M. Hommel, O. Kraft, and E. Arzt, J. Mater.
ꢀ
The Coming of Materials Science
R.W. Cahn
$55 MRS Member
$64 Non-MRS Member
The Coming of Materials Science both covers the discipline of materials science and draws an
impressionistic map of the present state of the subject.
ISBN: 0-08-042679-4
Code: CMS
The first chapter examines the emergence of the materials science concept, in both academia and indus-
try. The second and third chapters delve back into the prehistory of materials science, examining the
growth of such concepts as atoms, crystals and thermodynamics. These two chapters also examine the
evolution of a number of neighboring disciplines, to see what helpful parallels might emerge.
For further information,
or to place an order, contact :
The book contains numerous literature references. Many refer to the earliest key papers and books,
while others are to sources, often books, offering a view of the present state of a topic. Early references are to the
past but as the book continues, it brings the reader up to date with more recent sources.
The author, Professor Robert Cahn FRS (University of Cambridge, Cambridge, UK), has striven to be critical
about the history of the discipline of materials science and to draw general conclusions about scientific practice
from what he has discovered about the evolution of materials science. Further highlights include: What is a
scientific discipline? How do disciplines merge and differentiate? Can a discipline also be interdisciplinary?
Is materials science a real discipline? In other words, this book is not just for reading and reference, but exists to
stimulate thought and provoke discussion as well. 2001, hardcover, 568 pages.
Materials Research Society
Customer Services Department
506 Keystone Drive
Warrendale PA 15086 USA
phone 724-779-3003
fax
724-779-8313
e-mail info@mrs.org
www.mrs.org/publications/bulletin
44
MRS BULLETIN/JANUARY 2002