ArticlePDF Available

Assessing the Fracture and Fatigue Resistance of Nanostructured Thin Films

Authors:

Abstract and Figures

Fatigue failure through sustained loading of ductile materials manifests in irreversible motion of dislocations, followed by crack initiation and growth. This contrasts with the mechanisms associated with brittle ceramics, such as nanostructured physical vapor deposited thin films, where inhibited dislocation mobility typically leads to interface-controlled damage. Hence, understanding the fatigue response of thin films from a fundamental viewpoint – including altered atomic bonds, crystal structures, and deformation mechanisms – holds the key to improved durability of coated engineering components. Here, a novel method utilizing quasi-static and cyclic-bending of pre-notched, unstrained microcantilever beams coupled with in situ synchrotron X-ray diffraction is presented to study the fracture toughness and fatigue properties of thin films under various loading conditions. Investigating a model system of sputter-deposited Cr and Cr-based ceramic compounds (CrN, CrB2, and Cr2O3) demonstrates that the fatigue resistance of such thin films is limited by the inherent fracture toughness. In fact, cantilever cycling close to the critical stress intensity is sustained up to 10⁷ load cycles on all materials, without inducing noticeable material damage, structural or stress-state changes. The observed variation in fracture toughness is put into context with linear-elastic fracture theory and complementary micro-pillar compression, thereby elucidating the wide range of values from as low as 1.6±0.2 MPa√m for Cr1.79O3 up to 4.3±0.3 MPa√m for Cr1.03B2, respectively. Moreover, possible mechanisms governing the elastic-plastic deformation response of all coatings, both in quasi-static and cyclic-loading conditions, are discussed. Our findings contribute key-insights into the underlying mechanisms dictating the damage tolerance of PVD coated components by relating fatigue strength limits to fundamental material properties.
Content may be subject to copyright.
Acta Materialia 239 (2022) 118260
Contents lists available at ScienceDirect
Acta Materialia
journal homepage: www.elsevier.com/locate/actamat
Full length article
Assessing the fracture and fatigue resistance of nanostructured thin
films
L. Zauner
a , , R. Hahn
a
, E. Aschauer
a
, T. Wojcik
a , b
, A. Davydok
c
, O. Hunold
d
, P. Polcik
e
,
H. Riedl
a , b
a
Christian Doppler Laboratory for Surface Engineering of high-performance Components, TU Wien, Austria
b
Institute of Materials Science and Technology, TU Wien, Austria
c
Helmholtz-Zentrum Hereon, Institut für Werkstoff physik , Germany
d
Oerlikon Balzers, Oerlikon Surface Solutions AG, Liechtenstein
e
Plansee Composite Materials GmbH, Germany
a r t i c l e i n f o
Article history:
Received 11 April 2022
Revised 6 August 2022
Accepted 8 August 2022
Available online 10 Aug ust 2022
Keywo rds:
Fatigue
Thin films
Synchrotron diffraction
Physical vapo r deposition
Micromechanics
Ductility
Ceramics
a b s t r a c t
Fatigue failure through sustained loading of ductile materials manifests in irreversible motion of dislo-
cations, followed by crack initiation and growth. This contrasts with the mechanisms associated with
brittle ceramics, such as nanostructured physical vapor deposited thin films, where inhibited disloca-
tion mobility typically leads to interface-controlled damage. Hence, understanding the fatigue response
of thin films from a fundamental viewpoint including altered atomic bonds, crystal structures, and de-
formation mechanisms –holds the key to improved durability of coated engineering components. Here, a
novel method utilizing quasi-static and cyclic-bending of pre-notched, unstrained microcantilever beams
coupled with in situ synchrotron X-ray diffraction is presented to study the fracture toughness and fa-
tigue properties of thin films under various loading conditions. Investigating a model system of sputter-
deposited Cr and Cr-based ceramic compounds (CrN, CrB
2
, and Cr
2
O
3
) demonstrates that the fatigue re-
sistance of such thin films is limited by the inherent fracture toughness. In fact, cantilever cycling close to
the critical stress intensity is sustained up to 10
7
load cycles on all materials, without inducing noticeable
material damage, structural or stress-state changes. The observed variation in fracture toughness is put
into context with linear-elastic fracture theory and complementary micro-pillar compression, thereby elu-
cidating the wide range of values from as low as 1.6 ±0.2 MPa
m for Cr
1.7 9
O
3
up to 4.3 ±0.3 MPa
m for
Cr
1.0 3
B
2
, respectively. Moreover, possible mechanisms governing the elastic-plastic deformation response
of all coatings, both in quasi-static and cyclic-loading conditions, are discussed. Our findings contribute
key-insights into the underlying mechanisms dictating the damage tolerance of PVD coated components
by relating fatigue strength limits to fundamental material properties.
© 2022 The Authors. Published by Elsevier Ltd on behalf of Acta Materialia Inc.
This is an open access article under the CC BY license ( http://creativecommons.org/licenses/by/4.0/ )
1. Introduction
Creating a fundamental understanding of the fracture and fa-
tigue resistance of state-of-the-art materials is invariably associ-
ated with predicting the accessible lifespan for engineering com-
ponents in aerospace, electronics, or energy applications [1] . The
term “fatigue” comprises the cumulative material damage induced
by cyclic loading, frequently leading to fracture and failure of struc-
tural components. In metals, where plasticity takes the dominat-
ing role in controlling cyclic deformation, fatigue damage accumu-
Corresponding author at: Christian Doppler Laboratory for Surface Engineering
of high-performance Components, TU Wien, Getreidemarkt 9, 10 60 Wien, Austri a.
E-mail address: lukas.zauner@tuwien.ac.at (L. Zauner) .
lates upon irreversible plastic flow, initiated by irreversible dislo-
cation motion. The latter mechanism cumulates in the formation
of cracks at preferential sites such as persistent slip bands and
related surface irregularities ( i.e. , intrusions or extrusions), high-
angle grain boundaries, or twin-boundaries [2–5] . Conversely, on
the opposite end of the ductility spectrum, highly brittle ceramic
materials are characterized by strong ionic and/or covalent bonds
with limited dislocation mobility. In turn, the resistance to fracture
and fatigue failure in these brittle solids is inherently determined
by the atomic bond strength and activation of pre-existing flaws
( i.e. , grain boundaries, voids, inclusions, etc.) under loading [ 2 , 6 , 7 ].
Yet , regardless of the apparent bonding character, the material re-
sponse under repeating loads is closely linked to the microstruc-
ture at hand.
https://doi.org/10.1016/j.actamat.2022.118260
1359-6454/© 2022 The Authors. Published by Elsevier Ltd on behalf of Acta Materialia Inc. This is an open access article under the CC BY license
(
http://creativecommons.org/licenses/by/4.0/ )
L. Zauner, R. Hahn, E. Aschauer et al. Acta Materialia 239 (2022) 118260
By decreasing the overall microstructural domain size, as is the
case for physical vapor deposited micro- or nanostructured thin
films, it is obvious that the governing deformation mechanisms
are affected by the increased fraction of interfaces, grain bound-
aries, and free surfaces. These features introduce additional con-
straints on the availability of dislocation sources, their activity, and
mutual interaction so that macroscopic dislocation features are re-
duced and interface-mediated damage, similar to brittle materials,
becomes prevalent [ 3 , 8-11 ].
While general, bulk-scale fracture toughness and fatigue exper-
iments provide means of identifying the crack resistance and lifes-
pan of a material, assessing the failure-responsible atomistic or
structural origin is difficult to accomplish. Thus, shrinking the ex-
amined material volume to characteristic thin film dimensions al-
lows to fundamentally probe the influence of such features and un-
ravel the origins of material failure [12–15] .
Using micro sized cantilever beams and pillars, the intrinsic
fracture toughness and failure strength of thin films especially
that of hard protective coatings –has been investigated extensively
[16–23] . In contrast, reports on fatigue testing of coating materials
are disproportionally less, with a strong focus towards metallic ma-
terials ( e.g. , copper, nickel-base alloys, etc.) [24–28] or macroscale
testing of the entire coating-substrate arrangement [29–35] .
Employing TiN coated Ti-6Al-4V tension-tension fatigue speci-
men, Bai et al. [36] recently identified a stress-sensitive crack ini-
tiation mechanism. While below a critical stress amplitude dislo-
cation pile-up and slip-steps in the substrate induced crack forma-
tion close to the coating interface, increasing the applied stress re-
sulted in accelerated, coating-fracture-induced cleavage cracking of
the substrate (see also Ref. [37] ). Based on these mechanisms, en-
hancing the coating toughness or introducing a ductile interlayer
is suggested to alleviate the adverse effect of the coating on the
substrate fatigue properties. In addition, Sivagnanam Chandra et al.
[ 38 , 39 ] discussed the influence of varying both the thickness and
compressive stress state of Ti-Al-N coated Ti-alloys by correlating
the observed fatigue failure with the imposed stress profile. There,
critical aspects of the residual stress state across the coating and
interface region were highlighted with respect to preferred crack
nucleation sites.
Shifting the viewpoint to microscale fatigue testing, Kiener et al.
[40] first performed fully-reversed in situ bending experiments on
free standing single-crystalline copper micro cantilevers, thereby
excluding any interference of substrates or microstructures with
the resulting material behavior. The combined theoretical and ex-
perimental approach provided insights into the deformation mech-
anisms during fatigue loading at the micro-scale, up to a total
number of n = 100 load cycles. Using an analogous approach for
low cycle bending tests, a study by Kirchlechner et al. [3] cou-
pled fatigue experiments of single-crystal copper beams with in
situ synchrotron micro Laue diffraction to reveal the interplay of
dislocation motion and slip planes under reversible load. Moreover,
in works by Wimmer et al. [41] and Lavenstein et al. [42] , vastly in-
creased numbers of load cycles were accessed for cantilever bend-
ing experiments on poly-crystalline copper ( n 1.5 ×10
5
) and
a single-crystalline Ni-base superalloy ( n 10
7
), respectively, pro-
viding insights into small-scale fatigue failure within the high cy-
cle fatigue regime. Especially the approach of the latter study is
unique in its utilization of a permanent bond between the actuator
tip and cantilever free-end, which allows to maintain zero mean
stress while avoiding load gaps during strain reversal. In addition,
similar to a work by Schwaiger and Kraft [10] , a stiffness based
technique was employed to probe cyclic damage accumulation and
discuss the fatigue crack evolution [43] .
Despite the extent of studies carried out, only little is known
about the micro-scale material response of cyclically loaded ce-
ramic thin film materials. Especially the aspect of varying ma-
terial classes, which includes altered interatomic bond charac-
ters, crystal structures, and hence available deformation mech-
anisms, lacks an in-depth understanding. In this study, various
free-standing micro-geometries fabricated from Cr-based thin film
materials ( i.e. , body-centered cubic Cr, face-centered cubic CrN,
hexagonal CrB
2
, and rhombohedral Cr
2
O
3
) are utilized to unravel
the fracture and fatigue properties of nanostructured thin films.
Both, micro-cantilever bending and -pillar compression tests are
conducted to discuss the apparent fracture resistance and elasto-
plastic deformation behavior in relation to the structural proper-
ties. An approach based on synchrotron X-ray nanodiffraction cou-
pled fatigue experiments reveals the in-plane stress distribution
of statically and dynamically loaded microcantilevers. Moreover,
the methodology provides a direct comparison between the ob-
served high cycle fatigue properties and the individual single cycle
strength, i.e. , the fracture toughness.
2. Experimental
2.1. Coating deposition
Cr-based thin films were deposited onto (100)-oriented silicon
substrates by unbalanced DC magnetron sputtering (DCMS) using a
modified Leybold Z400 deposition system. The metallic Cr as well
as the CrN and Cr
2
O
3
compound thin films were deposited from a
powder-metallurgically prepared Cr target (99.95 % purity, Plansee
Composite Materials GmbH) operated in pure Ar-, mixed Ar/N
2
-,
or mixed Ar/O
2
-atmosphere, respectively. The CrB
2
thin film was
synthesized from a powder-metallurgically prepared CrB
2
target
(99.3 % purity) in pure Ar-atmosphere. For all thin films, a constant
target-to-substrate distance of 40 mm was maintained. Prior to
each deposition, all substrates were ultrasonically pre-cleaned us-
ing acetone and ethanol consecutively. After reaching a base pres-
sure of p 0.2 mPa, the substrate was heated to the desired tem-
perature T
dep
, followed by an Ar-ion etching step with an applied
substrate bias potential of U
S = -150 V (pulsed DC, 150kHz, 2496
ns pulse duration) at a total Ar gas pressure of p = 1.3 Pa. The
subsequent depositions were carried out using parameter sets ( i.e. ,
average cathode power density P
C
, total deposition pressure p
dep
,
process gas composition, substrate bias potential U
S
, etc.) deter-
mined during preliminary studies. Moreover, the deposition tem-
perature ( T
dep
) was adjusted to maintain a uniform value of 0.35
for the homologous temperature ( T
H
) of all Cr-based compounds,
while that for metallic Cr was reduced to T
H 0.22 to prevent ex-
tensive grain coarsening observed for higher T
H
values. The depo-
sition time was individually adjusted to yield a uniform thickness
of h = 3 μm for all coatings. A summary of all deposition parame-
ters and the resulting chemical composition including the content
of undesired residuals Res ( i.e. , argon and oxygen) are included in
Table 1 .
2.2. Coating characterization
Investigations on the coating structure were conducted using
X-ray diffraction (XRD) on a PANalytical XPert Pro MPD system
equipped with a Cu-K αradiation source ( λ= 1.5 418 ˚
A) and op-
erated in θ- θgeometry. The X-ray probe size obtained a constant
width of 14 mm and a decreasing length from 15 to 5 mm with
increasing diffraction angle (the coated sample area measured 18
×7 mm
2
). The chemical composition of all Cr-based thin films
was determined in top-view configuration by energy dispersive X-
ray spectroscopy (EDS) utilizing an FEI Philips XL30 SEM equipped
with an EDAX EDS detector (15 kV acceleration voltage). Further-
more, scanning electron microscopy (SEM) in an FEI Quanta 200
FEGSEM system (operated at 10 kV) was used to characterize the
2
L. Zauner, R. Hahn, E. Aschauer et al. Acta Materialia 239 (2022) 118260
Tabl e 1
Detailed overview of the deposition conditions and chemical composition of the synthesized coatings.
Coating
3"-target P
C T
dep T
H p
dep Ar N
2 O
2 U
S Res
(purity, %) (Wcm
2
) ( °C) (-) (Pa) (sccm) (sccm) (sccm) (V) (at.%)
Cr Cr (99.95) 4.0 200 0.22 [44] 0.37 30 - - -75 0.8
Cr
0.94
N Cr (99.95) 4.6 400 0.34 [45] 0.37 16 16 - -60 0.4
Cr
1.03
B
2 CrB
2
(99.3) 5.5 600 0.35 [46] 0.37 30 - - -50 0.4
Cr
1.79
O
3 Cr (99.95) 4.9 600 0.34 [47] 0.37 26 - 6.5 -50 0.6
Fig. 1. SEM micrograph of a FIB prepared, pre-notched micro-cantilever as used
for fracture toughness and fatigue testing experiments. The insert shows a post-
mortem SEM micrograph of the fractured cantilever cross-section with the pre-
notch depth a .
film growth morphology based on fracture cross-sections of single-
side coated silicon substrates. Complementary information on the
growth characteristics and grain size distribution was obtained via
transmission electron microscopy (TEM) on an FEI TECNAI F20 sys-
tem (200 kV acceleration voltage) utilizing samples prepared in
top-view orientation. Both, the hardness ( H ) and Young’s mod-
ulus ( E ) were determined by an ultra-micro indentation system
(UMIS) equipped with a Berkovich diamond tip. For each thin film,
at least 30 load-displacement curves were analyzed according to
Oliver and Pharr [48] in the load ranges of 5 to 22 mN (steps of
0.5 mN), with additional measurements up 45 mN to probe for
any substrate influence. Moreover, the recorded E values were fit-
ted using a power law as function of the penetration depth and
subsequently extrapolated to zero indentation depth to extract the
film-only modulus [49] .
Following the methodology introduced by Di Maio and Roberts
[50] , micromechanical bending experiments were performed on
cantilever geometries to characterize the critical fracture force ( F
C
)
and fracture toughness ( K
IC
) of all thin films investigated (see
Fig. 1 ). Therefore, the silicon substrate of a mirror polished frac-
ture cross-section was initially dissolved in aqueous KOH (40 wt.%
concentration at a temperature of 70 °C) to yield a section of free-
standing thin film material. Focused ion beam milling (FIB) us-
ing Ga
+
-ions on an FEI Quanta 200 3D DBFIB system was subse-
quently employed to produce microcantilever geometries in accor-
dance with the guidelines given in Ref. [51] . A beam current of 1
nA was used for initial milling sequences, subsequently reduced to
0.5 nA to minimize ion induced material damage. The cantilevers
were produced to a final geometry of length ( l ) ×width ( b ) ×
height ( h ) = 22 ×3 ×3 μm
3
, respectively (see Fig. 1 ). An ini-
tial notch was fabricated at a beam current of 50 pA, resulting
in final depths of 300 - 400 nm for all thin films (depth fluctu-
ations between cantilevers of the same material were below 50
nm). The notch did not extend over the entire cantilever width, in-
stead small material bridges (50 - 100 nm wide) were left to initi-
ate a very sharp crack at the notch base upon fracture (see Fig. 1 -
insert).
During testing, the cantilevers were loaded with an in situ Fem-
toTools FT-NMT04 nanoindentation system equipped with a di-
amond wedge tip (contact length of 10 μm) in an FEI Quanta
200 FEGSEM. The experiments were performed in displacement-
controlled mode with a deflection rate of 5 nms
1
, so that the
maximum force at failure ( i.e., F
C
) could be determined. All load-
displacement curves were further analyzed to assure pure linear-
elastic material behavior. The fracture toughness was then evalu-
ated according to Matoy et al. [16] :
K
IC
=
F
C
l
bh
3
2 ×f
a
h
(1)
with
f
a
h
= 1 . 46 + 24 . 36
a
h
47 . 21
a
h
2
+ 75 . 18
a
h
3
(2)
where a is the depth of the initial notch. A total of five bending
experiments were performed for each material system.
Furthermore, using a micropillar geometry, all coating materi-
als were investigated in their plastic deformation response under
uniaxial compression. The micropillars were analogously prepared
by FIB milling of the as-deposited coatings at a height to diame-
ter aspect ratio of h:d 3:2 ( h equals the coating thickness) and a
taper angle below 2 °. Stepwise decreased ion-currents from 15 nA
to 50 pA were employed for coarse and fine milling, respectively.
The pillars were compressed in displacement-controlled mode at 5
nms
1 using a diamond flat punch tip (contact diameter of 5 μm)
mounted to the above mentioned in situ nanoindentation system.
The recorded load - displacement curves were used to calculate the
engineering stress following an approach by Wheeler and Michler
[52] , where the top diameter of the pillar is taken as the refer-
ence contact area. The engineering strain was obtained from the
displacement data using the coating thickness as the initial pillar
height reference. Moreover, the displacement data was corrected
by accounting for the base compliance following Sneddon’s correc-
tion [53] given by
L
Sned d on
=
1 ν2
Sub
E
Sub
F
d
(3)
where νSub
and E
Sub
are the Poisson’s ratio and Young’s modulus of
the substrate, respectively, and F is the applied force.
Additional investigations on the fatigue resistance of all Cr-
based coatings under oscillating, unidirectional load were per-
formed through cyclic cantilever bending experiments ( i.e. , fatigue
tests), employing the same cantilever specimen and nanoinden-
tation system (see Fig. 1 ). The test setup was targeted towards
cyclically loading the cantilever close to the instantaneous frac-
ture threshold within the boundaries given by geometrical de-
viations between the samples and thus differences in the actual
fracture limit –to promote the accumulation of fatigue damage
in the material and eventually cause failure of the micro speci-
men. Therefore, the oscillating force F ( t ) was based on the previ-
ously recorded, material specific critical stress intensity, consisting
of a mean force F
M = 0.75 ×F
C
overlaid with a sinusoidal am-
plitude force F
A = 0.15 ×F
C
. This resulted in the applied stress
3
L. Zauner, R. Hahn, E. Aschauer et al. Acta Materialia 239 (2022) 118260
intensity oscillating between 60 and 90% of K
IC
, where a margin of
10% was maintained to accommodate for any geometrical inconsis-
tencies between the cantilevers . The fatigue tests were conducted
in force-controlled mode, meaning that F
M
and F
A
are maintained
throughout the test duration, irrespective of any cantilever deteri-
oration ( i.e. , crack growth). Furthermore, all experiments were per-
formed at a loading frequency of f = 500 Hz for a duration of
t = 334 min, which equates to a total number of n = 10
7 load
cycles. In addition to the load - displacement data, also the dy-
namic cantilever stiffness S was recorded, which is defined as the
real part of the ratio between the applied force- and displacement-
amplitude ( w
A
) and takes the form:
S =
F
A
(
t
)
w
A
(
t
)
cos
(
ϕ
) (4)
where ϕdenotes the phase angle between the load and displace-
ment response. Within the framework of this work, changes to
the parameter “stiffness” are used as the main indicator for cyclic
degradation of the cantilever specimen.
2.3. Coupled synchrotron X-ray nanodiffraction and fatigue
experiments
Initially, a double-side polished, cross-sectional substrate-
coating lamella with a thickness of 35 μm in beam direction was
prepared by mechanical polishing of a Cr
0.94
N coated Si substrate.
The lamella was subsequently secured to a sample holder, which
allows for a precise horizontal alignment, guaranteeing a 90 °con-
tact angle of the coating surface to the nanoindenter tip, while
avoiding any interference with diffracted beam paths during in situ
cross-sectional X-ray nanodiffraction experiments (see Fig. 2 a). Us-
ing the above-mentioned FIB workstation, larger microcantilever
specimens with a dimension of l ×b ×h = 25 ×25 ×3 μm
3
were
ion-milled into the sample lamella. Therefore, a gradually reduced
Ga
+
-ion current of 20, 5, and 1 nA was employed to remove the
silicon substrate material underneath the coating on a length of 30
μm, producing a clamped cantilever of the coating material. The
final geometry of the cantilever ( i.e. , releasing one of the clamped
ends, milling the contour to final shape) was fabricated at an ion-
current of 1 nA. Analogous to the smaller cantilever specimens, an
initial pre-crack was introduced in the form of a through-thickness
notch at a beam current of 0.1 nA. Prior to the fatigue experiments,
conventional bending tests were performed on three of these can-
tilevers to determine a new reference value for F
C
and K
IC
. Here it
is important to mention that the fracture toughness evaluation for
K
IC
is conducted outside the geometrical boundaries given in Ref.
[16] , with the sole purpose of obtaining the apparent fracture limit
of the material in the given geometry.
The coupled synchrotron X-ray nanodiffraction and fatigue ex-
periments were performed at the nanofocus endstation of beam-
line P03 at PETRA III located within the synchrotron facility
“Deutsches Elektronen-Synchrotron” DESY (Hamburg, GERMANY).
The monochromatic X-ray source was operated at a wavelength
of λ= 0.82656 ˚
A (beam energy of 15 keV) and focused to a
probe cross-section of 250 nm. The above mentioned nanoin-
dentation system was fixtured inside the measurement endstation
such that the cantilever samples were placed in transmission ge-
ometry with the X-ray beam in line to the pre-notch [54] . Fur-
thermore, the X-ray beam was centered at the pre-notch position,
while the diamond wedge tip of the indenter system was located
23 μm from this position at the opposite end and centered along
the cantilever width (see Figs. 2 a and b). Positioning of the entire
nanoindenter-sample arrangement was conducted using a hexapod
(tilt alignment) and linear nanopositioning stages (x-y- as well as
z-alignment). The 2D diffraction signals were recorded on an Eiger
X M9 photon-counting detector, with an acquisition time of 10
s per frame. The exact detector arrangement with respect to the
sample position was calibrated using a standardized LaB
6
reference
powder, resulting in a fixed sample-detector distance of 319.2 mm.
This allowed for the recording of diffraction patterns with Bragg
angles of up to 27 °.
Area scans covering a 20 ×20 mesh-grid in steps of 300 nm
along the y- and z-axis were performed prior to the fatigue ex-
periments to characterize the cross-sectional area of the cantilever
around the pre-notch position (see Fig. 2 c). Thereby, three dif-
ferent static loading scenarios were investigated, with the can-
tilever incrementally loaded from ( i ) no force applied, to ( ii ) 35
% of K
IC
(not shown here), and to ( iii ) 70 % of K
IC
. Subse-
quently, fatigue tests were conducted in alignment with the above-
mentioned procedure and parameters, with the applied forces ad-
justed to F
M = 0.65 ×F
C
and F
A = 0.15 ×F
C
, respectively, result-
ing in the overall stress intensity oscillating between 50 and 80
% of K
IC
. During these experiments, three distinct positions along
the cantilever height, spaced 1 μm apart (top, middle, bottom; see
Fig. 2 d), were repeatedly scanned for the duration of the loading
procedure. This allowed for a detailed analysis of the phase and
stress evolution within the cantilever material up to a total num-
ber of n = 5 ×10
6 load cycles.
2.4. Synchrotron X-ray nanodiffraction data analysis
The recorded 2D diffraction patterns were analyzed with re-
spect to the phase evolution and the in-plane stress state σ
, both
for static as well as dynamic loading conditions. To evaluate the
stress state of the cantilever cross-section σ
( y, z ) an integration of
the recorded patterns was performed in direction of the azimuthal
angle ψin segments of 10 °from ψ= 0 to 90 °(see Fig. 2 e). Within
the so obtained radial intensity distributions I( θ, ψ ) the positions
of distinct diffraction peaks 2 θ(ψ)
hkl
( e.g. , (200)-peak for CrN),
and thus also the orientation-dependent lattice spacing d (ψ)
hkl
,
were determined by fitting a Pseudo-Voigt function to the spec-
trum (see Figs. 2 f and g). Subsequently, the orientation-dependent
lattice strain ε (ψ)
hkl
was calculated following
ε
hkl
(
ψ
)
=
d
(
ψ
)
hkl
d
0
(
ψ
)
hkl
d
0
(
ψ
)
hkl
(5)
where d
0
( ψ
)
hkl
is the lattice spacing in the stress-free direction
ψ
hkl
, which itself is expressed by a set of X-ray diffraction elastic
constants (XEC) s
1 ,hkl
and
1
2
s
2 ,hkl
in the form of
sin
ψ
hkl
=
2 s
1 , hkl
1
2
s
2 , hkl
(6)
Both the XEC were adopted from literature, e.g. , as s
CrN, ( 200 )
1 ,hkl = -
0.3013 ×10
3 GPa
1 and
1
2
s
CrN, ( 200 )
2 ,hkl = 2.584 ×10
3 GPa
1 for
CrN [ 55 , 56 ]. Following the sin
2
( ψ)-method and the assumption of
an elastically isotropic material under a biaxial stress state the fun-
damental equation for the in-plane stress state follows
ε
hkl
(
ψ
)
=
1
2
s
2 ,hkl
σ
(
y, z
)
×sin
2
(
ψ
)
+ 2 s
1 ,hkl
σ
(
y, z
) (7)
The described 2D pattern and data processing was conducted
using the open-source software package DPDAK [57] as well as
a self-written MATLAB script [58] (The MathWorks Inc., version
R2019b, Natick, Massachusetts).
3. Results & discussion
In the following, the results of the experimental work are
presented and discussed thoroughly. Fig. 3 shows a schematic
overview of the entire coating synthesis and characterization
4
L. Zauner, R. Hahn, E. Aschauer et al. Acta Materialia 239 (2022) 118260
Fig. 2. (a) Schematic view of the in situ cross-sectional X- ray nanodiffraction setup. A Cr
0.94
N microcantilever sample with dimensions 25
×25 ×3 μm
3
(length
×width
×height), cut from a substrate-coating lamella, was scanned with an X-ray beam obtaining a cross-section of 250 nm. The microcantilever was loaded using a diamond
wedge tip with a contact length of 10 μm, positioned at a distance of 23 μm from the pre-notch, and centered along the cantilever width. The 2D Debye-Scherrer diffraction
patterns were recorded using
an Eiger X M9 photon-counting detector. (b) Detail showing the geometry and deflection w ( t ) of the microcantilever under a static and/or
cyclic load F ( t ), applied using the nanoindenter tip. (c) During static loading, the X-ray beam was scanned along the y- and z -axis in steps of 30 0 nm creating a 20 ×20 grid
centered around the pre-notch position, whereas during (d) cyclic loading, three positions along the cantilever height (“top”, “middle”, “bottom”; located at the pre-notch
position in y -direction) were repeatedly scanned for the duration of the loading experiment. (e) Representative intensity plot of the recorded Debye-Scherrer patterns. The
intensities of the (200)-peak were integrated in direction of the azimuthal angle ψwithin 10 °segments (dot-marked area) for ψ= 0 to 90 °and (f) plotted as function of
the diffraction angle. (g) Lattice parameters derived from the peaks plotted in (f) as function of sin
2
(
ψ) including a corresponding linear fit for stress analysis.
workflow ( Figs. 3 a-e), as well as exemplarily the material re-
sponse during fracture and fatigue experiments ( Figs. 3 f-j). In
Section 3.1 , the general framework of this study is established,
outlining the materials selection and thin film properties with
respect to phase formation, crystal structure, chemical composi-
tion, and apparent growth morphology ( Figs. 3 a-b, and f). This
framework subsequently allows for closer insights into the frac-
ture resistance ( Fig. 3 c-left) and elasto-plastic deformation behav-
ior ( Fig. 3 d) of thin film materials under quasi-static loads, as dis-
cussed in Section 3.2 . Therein, focused ion beam machining is used
to produce free-standing microcantilever and micropillar speci-
men to determine the material specific fracture toughness ( Fig. 3 g)
5
L. Zauner, R. Hahn, E. Aschauer et al. Acta Materialia 239 (2022) 118260
Fig. 3. Illustration showing the methodical approach to assess the fracture and fatigue resistance of thin films. (a-b) depict the fundamental thin film constitution in the
form of the apparent crystal structure and growth morphology, respectively. (c) comprises microcantilever bending experiments to determine the coating fracture toughness
(left) or the material response under prolonged cyclic loading (right). (e) shows the configuration for coupling in situ synchrotron X- ray nanodiffraction with cyclic bending
experiments. Exemplary data obtained following the approach in (a-e) is further presented: (f) thin film fracture cross-section revealing the growth morphology of a Cr
thin film; (g) force - displacement
data obtained from fracture toughness experiments; (h) force - displacement data obtained from micropillar compression experiments
indicating a ductile and brittle material response; (i) microcantilever stiffness and loading force data obtained from cyclic cantilever bending experiments; (j) diffraction
pattern obtained from in situ synchrotron X- ray nanodiffraction.
and amenability to plastic deformation (
Fig. 3 h), respectively. Tak-
ing the observed deformation mechanisms into account, the es-
tablished knowledge is further developed towards dynamic load-
ing conditions in Section 3.3 . Performing cyclic cantilever bend-
ing experiments up to the high cycle fatigue regime ( Fig. 3 c-
right) provides details on the fatigue resistance of each coating
material ( Fig. 3 i), and highlights the correlation between the fa-
tigue strength and fundamental material properties. In Section 3.4 .,
the micromechanical fatigue experiment is coupled with a novel
in situ synchrotron X-ray nanodiffraction setup ( Figs. 3 e and j).
Through this approach, the direct effect of imposing cyclic loads
on larger cantilever structures is reviewed, revealing the cross-
sectional stress distribution, phase evolution, and fatigue resistance
of selected thin films.
3.1. Phase formation and growth morphology
The structural analysis of all Cr-based thin films obtained by X-
ray diffraction is summarized in Fig. 4 . The diffractograms are ar-
ranged from bottom to top as metallic Cr ( bcc , squares), Cr
0.94
N
( fcc, circles), Cr
1.0 3
B
2
( hcp , triangles, space group 191) , and Cr
1.7 9
O
3
( rh , diamonds), respectively, and further complemented by the
standardized 2 θreference peak positions [59–62] . The data clearly
shows the phase-pure crystal structure of all coatings, alongside
an almost perfect stoichiometry for all Cr-based compounds with
respect to their nominal composition. Moreover, all diffractograms
indicate a polycrystalline, randomly oriented crystal growth except
for Cr
1.0 3
B
2
, which obtains a clearly preferred orientation in the
(001)-direction.
All coatings were additionally investigated in more detail re-
garding their growth morphology and grain boundary constitu-
tion using cross-sectional SEM and top-view TEM, respectively.
Figs. 5 a,c,e, and g depict the fracture cross-sections of single side
coated silicon substrates, whereas Figs. 5 b,d,f, and h include the
Fig. 4. XRD diffractograms of all Cr-based thin films measured on (100)-oriented
silicon substrates (cubic, open star,
[63] ) including the corresponding 2 θpeak po-
sitions of standardized reference patterns
[59–62] . The diffractograms are arranged
with metallic Cr ( bcc , squares), Cr
0.94
N ( fcc, circles), Cr
1.0 3
B
2 ( hcp , triangles, space
group 19 1) , and Cr
1.7 9
O
3
( rh , diamonds) from bottom to top, respectively.
corresponding top-view TEM micrographs prepared from the iden-
tical specimen. All thin film materials obtain a dense, homogenous
growth morphology over the entire coating thickness, which even-
tually concludes in a smooth top surface. Here, especially the thin
6
L. Zauner, R. Hahn, E. Aschauer et al. Acta Materialia 239 (2022) 118260
Fig. 5. (a), (c), (e), and (g) show SEM fracture cross-sections of all Cr-based thin
films deposited on (100)-oriented silicon substrates. (b), (d), (f), and (h) present cor-
responding top-view bright-field TEM micrographs including the average crystallite
grain size of the film cross-sections depicted in (a), (c), (e), and (g).
films synthesized from metallic Cr, Cr
0.94
N as well as Cr
1.7 9
O
3
ex-
hibit a pronounced formation of crystal columns, growing perpen-
dicular to the coating-substrate interface. Typical for DCMS sput-
tered thin films, the columns appear to be interrupted and regu-
larly forced to re-nucleate due to competitive crystal growth [64] .
Contrary, Cr
1.0 3
B
2
presents a rather featureless cross-section, lack-
ing in clearly separated morphological features. These findings are
largely supported by the corresponding bright-field TEM micro-
graphs, presenting a dense and void-free growth structure for all
coatings. Furthermore, the micrographs of metallic Cr and Cr
0.94
N
especially highlight the individual crystal columns by a distinct
orientation dependent contrast, whereas those for Cr
1.0 3
B
2
and
Cr
1.7 9
O
3
obtain a slightly less pronounced visual separation. Using
a line intercept method, the average crystal size was determined
for all thin film materials. Within the analyzed volumes, the pure
Cr thin film was evaluated with the largest crystal size of 277
nm, followed by Cr
1.0 3
B
2
with 201 nm, and then Cr
0.94
N as well
as Cr
1.7 9
O
3
with 143 and 123 nm, respectively. These results
suggest that the coatings synthesized in pure Argon atmosphere
tend to form larger crystals, whereas both coatings sputtered in
reactive environment are subject to enhanced re-nucleation and
thus obtain slightly smaller grains. The TEM micrographs allowed
for further microstructural analysis of the grain-boundary constitu-
tion. Here, the absence of distinct grain-boundary phases was con-
firmed, which is in line with the exact stoichiometry obtained for
all Cr-based coatings and the results received during XRD analysis.
These results outline the framework for all subsequent fracture
and cyclic loading experiments. Considering that the ensemble of
synthesized coatings shows a vastly comparable thin film structure,
the observed material properties will be particularly connected to
the apparent bonding nature of the individual thin film [65] . This
is considered as an essential aspect for elucidating possible origins
for varying fracture and fatigue characteristics.
3.2. Fracture characteristics
A comprehensive overview of the thin film properties recorded
and discussed within this Section is given in Table 2 . In addition,
literature references to corresponding bulk and thin film materials,
as well as results from density-functional-theory (DFT) calculations
are included to reveal possible scaling effects caused by the PVD
synthesis route.
Micromechanical bending experiments on all four coating ma-
terials reveal a pronounced variation in the fracture resistance as
depicted by the raw indenter force-displacement curves shown in
Fig. 6 a. In addition, Fig. 6 b presents the data normalized in terms
of stress intensity and bending strain. Upon examining the frac-
ture cross-section of the cantilevers, all materials are found to
fracture preferably along the grain boundaries. The results show
that Cr
1.0 3
B
2
obtains the highest critical fracture force during mi-
cro cantilever bending tests at F
C 810 ±65 μN. Consequently, this
also translates into the highest calculated fracture toughness with
K
IC 4.3 ±0.3 MPa
m. Compared to previous studies reporting on
the fracture toughness of protective thin films, this material system
appears to outperform many industrially relevant coatings such
as Ti-Si-N ( 3 MPa
m) or α-Al
2
O
3
( 4 MPa
m) [ 66 , 67 ]. More-
over, the fracture toughness is increased over monolithic bulk CrB
2
( 3 MPa
m, [68] ), likely due to the differences in microstructure
and the intrinsic stresses caused by the high number of point-
defects within the PVD thin film. In contrast, the data recorded
for Cr
0.94
N and Cr
1.7 9
O
3
indicates drastically lower critical fracture
loads of F
C 332 ±14 and 232 ±22 μN, thus resulting in fracture
toughness values of K
IC 2.1 ±0.1 and 1.6 ±0.2 MPa
m, respec-
tively. The values obtained for the metallic Cr coating are located
in between these boundaries, obtaining a critical fracture force of
F
C 462 ±31 μN and a corresponding fracture toughness of K
IC
3.6 ±0.3 MPa
m. Here, it is essential to note that the metallic ma-
terial is particularly prone to dislocation motion and hence plas-
tic deformation, which manifests in a slight curvature of the load
- displacement curve prior to the ultimate fracture- or yield-point
(see Fig. 6 a). This, however, stands in partial violation of the testing
conditions outlined within the frameworks of linear elastic fracture
mechanics (LEFM). Consequently, the fracture toughness values for
the Cr thin film should be treated carefully, especially when com-
pared to results deduced from different procedures. Overall, the re-
sults for the Cr thin film are considerably lower in contrast to the
bulk material fracture toughness [69] –a known effect for sputter
deposited metallic thin films [70] (see Table 2 ).
Upon specifically regarding the elastic deformability of the thin
films, all Cr-based compounds show a comparable maximum bend-
ing strain in the range of ε= 1-1.2 %, whereas the metallic coat-
ing exhibits an extended displacement up to w > 1.5 μm and ε
> 1.5 % prior to failure. Apart from varying maximum fracture
strengths, this is especially interesting when considering the ab-
sence of elastic differences as obtained from instrumented nanoin-
dentation experiments, where an almost identical Young’s modu-
lus of E 310 GPa was determined for all investigated coatings
(see Table 2 ). These observations eventually raise several intrigu-
ing questions: What are possible origins for the marked difference
in fracture toughness of the coating materials? Is there a particu-
7
L. Zauner, R. Hahn, E. Aschauer et al. Acta Materialia 239 (2022) 118260
Fig. 6. (a) Indenter force - displacement curves of all Cr-based thin films recorded during in situ bending experiments on free-standing microcantilever geometries. Critical
fracture loads and deflections are indicated using the material corresponding symbols introduced in
Fig. 4 . The mean value and standard deviation of the critical fracture
load is included for each thin film on the right axis. (b) Normalized stress intensity - bending strain curves calculated from the data in (a) and post-mortem analysis of the
individual geometry of each fractured cantilever cross-section. Critical stress intensities and bending strains, as well as mean values for the apparent fracture toughness of
all thin films, are again included analogous to (a).
lar connection to intrinsic (toughening) mechanism that counteract
crack propagation?
Assuming morphological consistency between the thin films, a
primary connection can be drawn to the apparent bonding struc-
ture and strength of each material ( i.e. , metallic, covalent, ionic,
or a combination thereof). In principle, any fracture process ul-
timately involves the rupture of interatomic bonds occurring
on intra-crystalline lattice planes or generic grain boundary sites.
Thus, the stronger the bonds, the higher the resistance against
fracture. This holds particularly true for brittle fracture processes,
which are entirely determined by atomistic events at the crack
tip location [85] . This becomes particularly evident, when put into
context with an expression for the fracture toughness of thin films
in the form [86] :
K
IC
=
4 γE (8)
where γdenotes the surface energy –so that a principal origin
for the measured variation may be attributed to the altered surface
energies between the coating materials (given the minute variance
in elastic moduli). Note, γcan also be interpreted as/exchanged
by the ideal debonding energy along an interface E
sep
hkl
[ 87 , 88 ]. Ap-
parently, this explanation can only capture the studied thin films
and their prevailing bonding character, thus may not be gener-
alized to the corresponding material families ( i.e. , not all nitride
and oxide thin films show inferior fracture toughness compared
to borides). In consequence, using Equation 8 assumes that any
contribution from dissipative mechanisms ( e.g. , crack tip plasticity,
morphology induced micro-cracking or crack-branching) to coun-
teract crack growth is negligible, especially within mode I based
cantilever bending experiments.
However, calculating a theoretical estimate for the fracture
toughness of CrN based on surface energies derived from DFT cal-
culations [89] , only a maximum value of K
IC 1.2 MPa
m is
obtained (see Equation 8 , Young’s modulus taken from Table 2 ).
Hence, regardless of the variation in γ, this result proposes that
intrinsic mechanisms may still contribute to the apparent frac-
ture toughness. In fact, recent studies proposed that plasticity re-