Acta Materialia 260 (2023) 119342

Contents lists available at ScienceDirect

Acta Materialia
journal homepage: www.elsevier.com/locate/actamat

Unraveling kinking: A plasticity enhancing failure mode in high strength

nano metallic laminates
Yifan Zhang a, Miroslav Zecevic b, Aritra Chakraborty a, Rodney J. McCabe a,
Thomas J. Nizolek c, Ricardo A. Lebensohn b, John S. Carpenter c, Nan Li d, Laurent Capolungo a, *
a
Materials Science and Technology Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA
b
Theoretical Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA
c
Metallurgy and Manufacturing Science Division, Los Alamos National Laboratory, NM 87545, USA
d
Center for Integrated Nanotechnologies, Los Alamos National Laboratory, Los Alamos, NM 87545, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Kinking is an important and plasticity-enhancing deformation/failure mode in numerous mechanically aniso­
Kinking tropic materials including high-strength nano metallic laminates (NMLs). However, our current limited under­
Nano metallic laminates standing of the mechanics of kinking and its dependence on microstructural attributes is insufficient for
In situ compression
thoroughly comprehending and eventually being able to control failure behaviors of materials. In this study, we
Strain-gradient crystal plasticity
Dislocation dynamics
investigate kinking dependencies on microstructural attributes in NMLs via in situ micropillar compression,
multiscale microstructure characterization, dislocation dynamic simulations, and crystal plasticity modeling. By
examining several NML systems (Cu/Fe, Ag/Fe, Al-4Mg/Fe), we demonstrate that the development of internal
stresses during loading activates local layer-parallel glide triggering kinking in NMLs. Furthermore, this work
reveals the effect of key microstructural features including layer thickness, layer waviness, interface barrier
strength, and work hardening capacity on kink band formation in NMLs. More broadly, our efforts represent a
generically applicable approach for probing large-strain deformation behavior of complex materials via syner­
getic modeling and experimental efforts.

1. Introduction Different kinking mechanisms have been proposed for various ma­
terial systems to rationalize similar kinking morphologies [10], yet
Kinking / kink banding is a shear localization phenomenon akin to a many of these models are not applicable to kinking in NMLs. For
geometrically constrained ‘folding’ of the microstructure (see Fig. 1). non-crystalline materials such as fiber-reinforced polymers, some prior
Kinking is a deformation/failure mode that has been observed in a wide studies have shown that kinking is likely due to micro-buckling of layers
breadth of material classes including organic, polymeric, ceramic, and and interlaminar sliding [11]. In comparison, kinking in most crystalline
metallic materials [1–9]. These diverse materials exhibit a common materials is associated with anisotropic dislocation activity. For
feature: intrinsic and pronounced elastic and/or plastic anisotropy. instance, dislocation glide constrained to basal planes can result in KB
Interestingly, kinking is often a non-detrimental and controllable failure formation in hexagonal close-packed (HCP) materials [12–14]. Barsoum
mode that can improve the energy-absorbing capability of a material et al. [15] interpreted this as kinking nonlinear elastic (KNE) behavior
[8]. Despite the ubiquity of kinking and its critical importance for me­ due to the hysteretic generation and annihilation of incipient kink bands
chanical deformation and failure, a comprehensive understanding of (IKBs) upon unloading. According to the IKB theory proposed by Frank
kinking processes and conditions promoting kinking remains unestab­ and Stroh [16], kinking starts with a lenticular or elliptical IKB con­
lished. Nano metallic laminates (NMLs), with tailorable microstructural sisting of two walls of dislocations with opposite sign that can grow
parameters, are good model material systems for investigating the ef­ when the maximum shear stress is higher than a critical shear stress (τc).
fects of microstructural features and constitutive behaviors on kinking in IKB nucleation is more difficult in smaller grains as τc is inversely pro­
complex crystalline materials. portional to the square of the grain size [2]. KNE behavior has been

* Corresponding author.
E-mail address: [email protected] (L. Capolungo).

https://doi.org/10.1016/j.actamat.2023.119342
Received 31 May 2023; Received in revised form 5 September 2023; Accepted 7 September 2023
Available online 8 September 2023
1359-6454/Published by Elsevier Ltd on behalf of Acta Materialia Inc.
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 1. (a) Proposed kink band (KB) formation process. (a1) Activation of layer-inclined slip systems in softer material (Cu) due to its lower critical resolved shear
stress; (a2) development of back-stresses on layer-inclined slip systems due to dislocation pileups results in the activation of local layer-parallel slip that can be further
facilitated by local waviness; (a3) local rotation of softer material (Cu) layer caused by local layer-parallel slip (see Fig. S1); (a4) layer-parallel gliding dislocations are
stopped by grain boundaries or local internal stresses leading to accumulations at these locations. (a5) Rotation of adjacent stronger material (Fe) layer to
accommodate Cu layer rotations leads to the activation of layer-parallel slip in Fe layers; (a6) Accumulation of gliding lattice dislocations into geometrically necessary
dislocations (GNDs) at kink band boundaries (KBBs). (a7) Rearrangement of GNDs into geometrically necessary grain boundaries (GNBs). (a8) More GNBs form at
KBBs by incorporating more gliding dislocations. (a9) KB grows by moving GNBs towards unkinked regions. (b) Relative kink banding propensity (KBP) variation
based on perturbing each individual microstructural parameter. (c) Kernel average misorientation (KAM) map showing local misorientation of a kinked Ag/Fe 50 nm
NML after pillar compression. (d) Contour map showing the density of edge dislocations for layer-parallel slip systems in Cu and Nb layers [nm− 2].

observed in coarse-grained materials with grain sizes ranging from 10 to thickness, interface properties, crystal orientations, etc.) have promi­
600 µm [2,6,17,18] and τc less than 100 MPa. Compared to the nent effects on the deformation mechanisms of NMLs. For example,
single-phase materials for which the IKB theory applies, the micro­ shear banding was observed in Cu/Nb [26] and Cu/Ag [27] NMLs
structures in NMLs are considerably more complicated with closely during layer-parallel micropillar compression tests, while layer bending,
spaced phase boundaries and multiple active slip systems in each phase. layer thinning, and bulging occurred in Al/Ti [21] and Cu/Fe [28] NMLs
Thus, it is not clear whether the IKB theory applies to kinking in NMLs. for layer-normal compression. Nan et al. [29] and Liu et al. [30] re­
For the past 20 years, NMLs have drawn significant interest due to ported that deformation mechanisms for NMLs were interface shearing
their exceptional mechanical properties [19–21], good thermal stability and sliding when the loading direction was inclined to layer interfaces.
[22–24], and excellent radiation tolerance [25]. Many studies have been Under shear loading conditions, layer bending, in-plane lattice rota­
conducted to investigate deformation mechanisms of NMLs. Researchers tions, and forced chemical mixing might take place in Cu/Nb NMLs [30,
have demonstrated that both extrinsic (loading conditions, sample ge­ 31]. Snel et al. [32] revealed that deformation of Cu/Nb at 400 ◦ C was
ometry, strain rate, temperature, etc.) and intrinsic factors (e.g., layer mostly accommodated by a stress-assisted diffusion process. Apart from

2
Y. Zhang et al. Acta Materialia 260 (2023) 119342

the extrinsic factors, intrinsic factors including layer thickness, interface thinned to ~500 nm using beam parameters of 30 kV and 0.5 nA and
properties, constituent phase attributes, etc. significantly affect defor­ were further thinned to ~100–150 nm using 30 kV and 100 pA followed
mation behaviors of NMLs. Chen et al. [33] reported that thicker biphase by polishing with a 2 kV beam. TEM and scanning TEM analyses were
layer interfaces could simultaneously increase the strength and performed at 300 kV using an FEI Titan 80–300 STEM. SEM analyses
deformability of Cu/Nb nano laminates. Cui et al. [34] demonstrated were conducted using an FEI Apreo SEM with beam parameters of 10 kV
that shear banding in Cu/Mo NMLs is suppressed by hierarchical and 0.4 nA and a working distance of ~10 mm. Transmission Kikuchi
bicontinuous intertwined interfaces and layer structures. In addition, diffraction (TKD) analyses were conducted on TEM samples using a
the deformation mechanisms and work hardening capability of NMLs − 20◦ pretilt holder inside an FEI Apreo SEM at 30 kV and 13 nA. TKD
depend on layer thickness. Dayal et al. [35] showed that the deformation data were analyzed using the EDAX/TSL OIM Analysis software. The
mechanism shifted from layer bending and bulging to shear banding grain confidence index (CI) standardization and neighbor CI correlation
when the layer thickness went below 20 nm in Al/Pd NMLs. Liang et al. (NCIC) cleanups were performed on collected data sets using a threshold
[36] showed that strength/ductility and co-deformability of Ni-W lam­ CI value of 0.1.
inates were improved with decreasing layer thickness.
Kinking (or kink banding) has been observed in many NMLs during 2.3. In situ pillar compression tests
layer-parallel compression tests [37–45]. However, kinking in NMLs is
significantly less explored compared with other failure mechanisms. Micro-pillars were cut along the TD or RD on polished samples using
Limited aspects of kinking in NMLs have been investigated including the an FEI Helios 600 FIB/SEM. Pillar dimensions were 5 ± 0.5 × 5 ± 0.5
evolution of strain fields in the vicinity of a propagating kink band [5] (cross-section) × 12 ± 1.5 µm (height). Additionally, to investigate the
and the evolution of geometrically necessary boundary (GNB) structures effect of pillar aspect ratios on the kinking behaviors, 4 special pillars
within Cu/Nb NML KB boundaries (KBBs) [46]. These studies revealed with 3.5 ± 0.5 × 3.5 ± 0.5 (cross-section) and two different pillar
important yet limited aspects of kinking; the very process of KB forma­ heights (8.5 and 16 µm) were fabricated for Cu/Fe 50 nm NMLs.
tion and its sensitivity to microstructure in NMLs and other layered In situ pillar compression tests were conducted using a Brucker PI-89
material systems remains unclear. or PI-85 SEM PicoIndenter equipped with an 11-µm diamond flat punch
In this work, we perform a comprehensive experimental and inside an FEI Apreo SEM. Displacement-control modes were used for all
modeling-based study of microstructural dependencies of kinking in pillar compression tests to ensure a constant strain rate of 2 × 10− 3 s− 1.
NMLs. We examine the effects of microstructural factors on kinking Videos were captured simultaneously using an in-lens T1 detector or an
behavior in Al-4Mg/Fe, Ag/Fe, and Cu/Fe NMLs. We corroborate the KB Everhart-Thornley detector (ETD).
formation processes proposed in our previous study [46,47] and reveal
for the first time the relative and absolute roles of layer waviness, layer 2.4. Crystal plasticity modeling
thickness, phase properties, and interface barrier strength on kinking in
NMLs. Results of this work are likely applicable to other crystalline NML mechanical response was simulated using a recent large-strain
engineered or natural layer-structured material systems with compres­ extension of the elasto-viscoplastic model (LS-EVP-FFT) with strain-
sive strength depending on the kinking behaviors. gradient (SG) effects [47,49]. This formulation is a generalization of
the SG version [50] of the small-strain EVP-FFT model [51]. In
2. Experiments and simulations FFT-based models, the equilibrated stress field constitutively related to a
compatible strain field is obtained for periodic unit cells with heterog­
2.1. Material preparations enous mechanical properties using a Green’s function method, solving a
convolution problem in Fourier space [52]. Each single crystal material
Cu/Fe, Ag/Fe and Al-4Mg/Fe NMLs were fabricated using accumu­ point deforms elastically and plastically, where the plastic deformation
lative roll bonding (ARB). High-purity Fe (99.95%), Ag (99.95%) and Al- is accommodated by dislocation glide on available slip systems. Dislo­
4wt% Mg plates were purchased from American Elements (Los Angeles, cations start gliding once the applied stress overcomes the slip resistance
CA, USA). Oxide free high conductivity (OFHC) copper was purchased caused by statistically stored dislocations (SSD) and geometrically
from Revere Copper Products Inc (Rome, NY, USA). As-received mate­ necessary dislocations (GNDs). SSDs contribute to the critical resolved
rials were cold-rolled to thicknesses between 1 and 2 mm and were shear stress while GNDs are accounted for through a SG plasticity model
annealed for one hour at temperatures high enough to completely [53]. The total content of GNDs at a material point is characterized by
recrystallize the microstructures prior to the ARB process (Fe at 550 ◦ C, the Burgers (Nye) tensor, which is in turn used for calculating attracti­
Cu at 450 ◦ C, Ag at 450 ◦ C, and Al-4Mg at 400 ◦ C). The Cu/Fe and Ag/Fe ve/repelling forces acting on dislocations due to the stress fields of
plates were annealed for 2 hours at 550 ◦ C in an inert atmosphere neighboring GNDs. Once a dislocation reaches an interface it may either
following every other ARB step until the average layer thickness was accumulate at the interface or transmit into the neighboring grain/­
300 nm. This was followed by cold rolling to a layer thickness of 50 nm. phase. Resistance to both of these events is given by an interface strength
Al-4Mg/Fe NMLs were created using a combination of room- parameter, k0. The initial NML microstructures are generated by
temperature and elevated-temperature ARB processing. ARB processes randomly sampling the measured distributions of geometrical parame­
include sequential operations of cutting, cleaning, stacking, and roll ters, while the initial crystallographic orientations and material pa­
bonding. Details about our ARB process can be found elsewhere [48]. rameters are homogeneous within a layer. Model parameters are
To examine the effects of annealing on KBP, samples of Ag/Fe 50 nm reported in Table S2, and a predicted stress strain curve is given in
NML were annealed at 450 ◦ C for 1 hour inside a tube furnace under Fig. S5.
high vacuum (<2 × 10− 7 torr). NML samples for micro-mechanical
testing and characterization were first cut along the RD or TD di­ 3. Results and discussions
rections into smaller pieces, ground to an 800-grit SiC, and polished
using 1 µm alumina, 0.3 µm alumina and 0.05 µm colloidal silica. 3.1. Mechanisms of kink band formation

2.2. Microstructure characterization Naturally, the propensity for kink banding is expected to depend,
likely in a highly coupled way, on microstructure (e.g., layer thickness,
Transmission electron microscopy (TEM) samples were prepared by interface morphology), response of the constituents (yield strength,
focused ion beam (FIB) using a Helios 600 dual focused ion beam (FIB)/ work hardening rate), and interface response (transmissibility of dislo­
SEM. 2µm-thick TEM lamellae lifted out from NML samples were cations). The correlations between the kinking propensity and these

3
Y. Zhang et al. Acta Materialia 260 (2023) 119342

characteristics have not been previously studied in a systematic way Misorientation (KAM) maps (Fig. 1(c)) and by slip activity and edge
such that it is unclear which of the aforementioned factors are most dislocation density maps obtained from LS-EVP-FFT simulations (Fig. 1
influential. The methods section details our systematic parametric study, (d)). As shown in Fig. 1(c), KBBs have significantly higher KAM values
based on in situ micro pillar compression tests and large-strain elasto- and, thus, dislocation densities.
visco-plastic fast Fourier transform (LS-EVP-FFT) simulations, to iden­ The locations where GNDs/GNBs accumulate at KBBs are affected by
tify and deconvolve the dependence of kink banding propensity on: layer the presence of grain boundaries within layers, heterogeneities in in­
waviness, average layer thickness, contrast in yield strength between ternal stresses, and layer interface morphology variations. The rotation
materials, strain hardening rate of the constituents, grain size, and of crystals and phase interfaces during kinking further increases the RSS
interface strength. Moreover, efforts have been made to maintain the on layer-parallel slip systems making layer parallel slip even more
same experimental conditions, including indenter-sample alignment, favorable within the kink band. This leads to the formation of several
sample geometry, sample finish, strain rate, and temperature. The most closely spaced GNBs. At the micron length scale this assembly of GNBs
important factors for kinking in NMLs are highlighted in Fig. 1 along manifests as a kink band boundary (KBB). In terms of macroscopic
with our proposed kinking mechanism. mechanical response, the layer rotation induces geometric softening that
The combined experimental and modeling work sheds light on the can result in a load drop depending on how well neighboring layers or
process of kink band formation. During layer-parallel compression, the regions accommodate the fledgling KB. Once the stronger material
softer phase plastically yields first while the stronger material continues plastically yields, initiation of a kink band involves cooperative rota­
to deform elastically carrying more of the load as deformation proceeds. tions over several neighboring layers (steps 3–5 in Fig. 1(a)).
Kinking in NMLs is initiated by a local transition in dominant slip sys­ The relation between local layer-parallel slip and kinking is also
tems. At the beginning of loading, slip is active on systems inclined to observed in our LS-EVP-FFT simulations as shown in Fig. 2. At the onset
layer interfaces, and the resolved shear stress (RSS) (driving force for of plastic deformation, slip activity is primarily on layer-inclined slip
dislocation glide) is small for slip planes/directions nearly parallel or systems (Fig. 2(d)). With increasing strain, layer-inclined slip is deacti­
perpendicular to the compression direction. As deformation proceeds, vated due to back stresses [47,54,55] from dislocation pile-ups at layer
back-stresses created by dislocation pile-ups at interfaces make dislo­ interfaces, leading to activation of local layer-parallel slip. Fig. 2(e)
cation motion more difficult for layer-inclined slip, thus promoting shows that KBs initiate from locations where local layer-parallel slip
activation of layer-parallel slip (steps 1–2 in Fig. 1(a)). For very small occurs. As shown in Fig. 2(f), once kinking starts, the most active slip
layer thicknesses (h<100 nm), layers are unable to accommodate pile- systems inside KB region have layer-parallel slip directions. In com­
ups of multiple dislocations, limiting layer-inclined slip from the start, parison, active slip systems outside KB remain layer-inclined as shown in
thus triggering the transition to layer parallel slip at earlier stage of Fig. 2(g). The change of active slip systems from layer-inclined to
deformation. The local layer-parallel slip activity causes the crystals and layer-parallel slip during deformation is also supported by dislocation
layers to rotate and geometrically necessary dislocations (GNDs) to dynamic simulations as shown in supplementary Fig. S2. Layer-parallel
accumulate at the peripheries of layer-parallel slip zones (steps 3–4 in gliding dislocations can move upwards or downwards depending on
Figs. 1(a), S1) with GNDs evolving into geometrically necessary their Burgers vectors and accumulate at the KBBs (Fig. 1(d)).
boundaries (GNBs). Evidence for this process is obtained from both
transmission Kikuchi diffraction (TKD)-based Kernel Average

Fig. 2. (a–c) EVP-FFT simulation results showing the edge dislocation density in a Cu/Nb 80 nm pillar at different strain levels. (d–g) Directions for most active slip
systems in different locations of the pillar.

4
Y. Zhang et al. Acta Materialia 260 (2023) 119342

3.2. Quantifying effects of microstructure, constituent response, interface measured on an as-rolled Cu/Nb 80 nm NML from our previous study
characteristics and external factors on kink band formation [46]. The relative kink banding propensity (KBP) is defined as KBP =
εkb(ref) /εkb − 1, where εkb(ref) is the strain at which the reference micro­
3.2.1. Relative kink banding propensity analysis structure kinks and εkb is the strain at which the perturbed microstruc­
With the mechanisms of kink band formation established, micro­ ture kinks. To quantify the relative roles of each of the six parameters
structural features and the intrinsic response of the constituent phases considered, we calculate a relative KBP variation using the finite dif­
and layer interfaces (Fig. 1b) can modulate the transition from layer-
ference method: ΔKBP/(Δpi /pi ), where pi is the parameter value and
(ref)
inclined slip to layer-parallel slip thereby changing the kinking pro­
ref indicates the reference parameter value. As shown in Fig. 1(b),
pensity. To quantify the effects of microstructure, interface behavior,
different factors have distinctive effects on the relative KBP. The KBP
and constituent phase intrinsic response on kinking, we use both LS-
increases when either the waviness amplitude, the interface barrier
EVP-FFT simulations and characterization of the response of Cu/Fe,
strength to dislocation transmission, or the CRSS ratio between the
Ag/Fe, and Al-4Mg/Fe NMLs with varying layer thicknesses. In the
constituent phases increases; and the KBP decreases when either the
simulations, we perform local sensitivity analysis [56] by varying each
wavelength, average layer thickness, or strain hardening rate increases.
individual feature of the microstructure, the constituents, and the
Critically, the microstructure and intrinsic constituent response de­
interface response with respect to a reference microstructure (corre­
pendencies derived from the local sensitivity analysis are consistent with
sponding to Cu/Nb) separately, and assess the predicted response of the
experimental observations during pillar compression tests. It is impor­
composite (see methods section). This allows deconvolution and sepa­
tant to note that the dominant features controlling the onset of kinking
ration of the roles of microstructure, constituents, and interfaces on
are waviness amplitude and layer thickness.
kinking. The reference microstructure is defined by the following pa­
rameters: average layer waviness amplitude of 17.8 nm, average layer
3.2.2. Effect of layer thickness
wavelength of 1355.3 nm, layer thickness of 80 nm, infinite interface
In what follows, we detail the specific role of each feature of the NML
barrier strength, null hardening rate of the constituents, and critical
by characterizing microstructure evolution in a wide range of NMLs.
resolved shear stresses (CRSSs) of 262.5 and 210.0 MPa for Nb and Cu,
These observations are rationalized from our experiments and LS-EVP-
respectively [57]. These reference microstructural parameters were

Fig. 3. Effect of layer thickness (h) on kink banding propensity (KBP). Engineering stress-strain curves and example deformed pillars for (a) Al-4Mg/Fe 50 nm, and
(b) Al-4Mg/Fe 300 nm NMLs. (c, d) deformed pillars for Al-4Mg/Fe 50 nm and 300 nm after pillar compression.

5
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 4. EVP-FFT analysis of layer thickness effect on kink band formation. (a1) 60 nm and (b1) 300 nm layer thickness at ε = 0.18. (a2), (b2) The directions of the most
active slip systems for (a1) and (b1), respectively. (c1, c2) Schematics showing the different dislocation glide modes for h < 100 nm and h > 100 nm. (d) Plots showing
the evolution of kink band initiation strain with the layer thickness.

FFT simulation results. Figs. 3 and 4 summarize the effects of average layer (Fig. 4(c2)). For h < 100 nm, layers are unable to accommodate
layer thickness (h) on KB formation in NMLs. Fig. 3(a, b) show transverse dislocation pile-ups, and the transition from layer-inclined slip to
direction (TD) pillar compression tests performed on Al-4Mg/Fe with h layer-parallel slip occurs earlier during deformation (Fig. 4(c1)). More­
= 300 and 50 nm. The KBP is higher for the 50 nm pillars (Fig. 3(c)) as over, grain aspect ratios (the ratio of layer-parallel and layer-normal
all tested pillars kinked and exhibited notable load drops (Fig. 3(a)). In grain size) increases with decreasing h, and the critical resolved shear
comparison, only two 300 nm pillars (P3 and P5 in Fig. 3(d)) are fully stress (CRSS) for layer-inclined slip systems increases significantly
kinked (labeled as kinked in Fig. 3(b)) while the others lack KBs that relative to the CRSS for layer-parallel slip due to Hall-Petch effects [59].
traverse the entire pillar with well-defined KBBs. Additionally, the Al- Consequently, the transition to layer-parallel slip occurs earlier with
4Mg/Fe 50 nm pillars exhibit a KB initiation strain of ~7% that is decreasing layer thickness thereby promoting kinking in NMLs.
significantly lower than the ~13% KB initiation strain of the Al-4Mg/Fe
300 nm pillars. Our LS-EVP-FFT simulations in Fig. 4 demonstrate that a 3.2.3. Effect of annealing
KB forms at ε = 0.18 when h = 60 nm. Accordingly, layer-parallel slip Concurrently, internal stresses due to dislocation content can also
prevails within the KB region (Fig. 4(a2)). In comparison, a KB is not affect the onset of kink banding. Experimentally, we modify the internal
present at the same strain level in the sample with 300 nm layer thick­ stress state by annealing at temperatures that recover some of the
ness, where layer-inclined slip systems are still prevalent (Fig. 4(b1, b2)). dislocation content and potentially alter the intergranular and interface
Our simulations also indicate that the KB initiation strain increases with stresses while maintaining the layer morphology. Fig. 5 shows the
increasing h (Fig. 4(d)), implying a lower KBP. Cu/Fe, Al-4Mg/Fe and kinking behaviors of Ag/Fe 50 nm annealed for 1 h at 450 ◦ C compared
Ag/Fe NMLs exhibit similar layer thickness dependence and these are all to an as-rolled (AR) counterpart. The AR and annealed Ag/Fe 50 nm
consistent with our previous study [8] on kinking in Cu/Nb NMLs with pillars have distinct deformation behaviors, with the AR material clearly
30 nm and 500 nm layer thicknesses. having a higher KBP. First, all tested AR Ag/Fe pillars kinked with two
The layer thickness dependence of KBPs is rationalized as follows. KBs. At a similar strain level, five out of six annealed pillars kinked but
For thicker layers/grains, more plasticity is accommodated by slip on only two exhibited double-kinking (Fig. 5(f)). Second, kinking in the
inclined slip planes before the back-stresses caused by dislocation pile- annealed samples initiated at 11% engineering strain whereas kinking
ups shut these slip systems down [58]. Moreover, our TKD analyses initiated at 9% engineering strain in AR Ag/Fe 50 nm. Third, only one
show that NMLs have polycrystalline intralayer microstructures with third of the annealed Ag/Fe pillars exhibited a load drop, and the
several grains through the layer thickness when h > 100 nm (Fig. 7(e)). average load-drop magnitude was significantly smaller than in the AR
Therefore, slip systems are locally more randomly oriented, resulting in samples (Table 1).
less favorable conditions for layer-parallel slip throughout an entire

6
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 5. Effect of initial dislocation density and related strain hardening rate on KBP. Kernel average misorientation (KAM) maps for (a) as-rolled and (b) annealed Ag/
Fe 50 nm NMLs. Engineering stress-strain curves with deformed pillars for (c) as-rolled (AR) and (d) annealed Ag/Fe 50 nm NMLs. (e, f) Pillars after compression for
AR and annealed Ag/Fe 50 nm samples.

Table 1
Pillar compression results for as-rolled and annealed Ag/Fe 50 nm NMLs.
Sample Yield strength Average work KB initiation Kinked pillar Load drop KB width Load drop magnitude
(MPa) hardening strain fraction probability (µm) (MPa)
rate at 2%< ε <7%
(GPa)

As rolled 882± 50 4.7 ± 0.7 9 ± 0.5% 1 1 1.7 ± 0.3 360 ± 110

Annealed 610± 90 7.2 ± 1 11±0.2% 0.83 0.4 2.3 ± 0.4 70 ± 10

A primary effect of annealing is a reduction in the intralayer and The effect of reducing dislocation densities is modeled with LS-EVP-
interface dislocation densities, therefore decreasing the type III internal FFT by changing the initial work hardening rates for the materials (Fig. 6
stresses that mostly arise from local dislocation content [60,61]. The (d)). Consistent with the annealing experiments, a reduction in initial
reduction in intralayer GND content can be directly seen in the lower work hardening rate decreases the kink band initiation strain (Fig. 6(d)).
KAM values observed for the annealed Ag/Fe compared to AR Ag/Fe Furthermore, reduced dislocation density and reduced internal stresses
(Fig. 5(a, b)). In parallel, the decreased dislocation density in the at layer interfaces might result in a lower interface barrier strength to
annealed samples increases its strain hardening capability. Indeed, re­ dislocation transmission in turn leading to enhanced dislocation trans­
covery restores the dislocation storage capacity of both layer-inclined mission and lower back-stresses on layer-inclined systems within layer
and layer-parallel slip systems (Fig. 6(c1, c2)). The annealed material interiors (Fig. 6(c)) as will be detailed later. This also results in a delay in
thus exhibits a higher strain hardening capacity (Table 1) and more the activation of layer-parallel slip. Moreover, reduced internal stresses
uniform initial strain distribution (Fig. 6(b)). Consequently, layer in­ effectively lower the nucleation probability for KBs as locations with
clined slip can accommodate more of the initial plasticity, postponing high stress concentrations are preferred nucleation sites for KBs [5].
the activation of layer-parallel slip and resulting in a higher KB initiation
strain. The increased work hardening capability can alleviate the 3.2.4. Effect of layer morphology
geometrical softening induced by kinking. Therefore, the As stated previously, the layer waviness of NMLs, including both
kinking-induced load drop magnitude is significantly lower in annealed waviness amplitude (WA) and wavelength (WL), also affects KBP. KB
Ag/Fe (70 MPa±10) than that in as-rolled counterpart (360±110 MPa) nucleation requires either an external local perturbation of the stress
as shown in Table 1. state or an internal perturbation of the microstructure (e.g. waviness)

7
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 6. KAM maps showing local misorientation distribution in (a) AR and (b) annealed Fe/Ag 50 nm NMLs. (c1, c2) Schematic representation showing that the AR
Ag/Fe has a higher initial dislocation density than the annealed Ag/Fe resulting in less work hardening capacity. (d) Plots showing the kink band initiation strains for
different average initial work hardening rates.

[46]. While prior studies have demonstrated a relation between layer (location 2 in Fig. 9(b2)), the local layer thickness is reduced, leading to
waviness and kinking in fiber reinforced polymer composites [62] and increased back-stresses on inclined slip planes. Additionally, an increase
wood [1], the effect of waviness on kinking has not been previously in interface wavelength results in more gradual transitions between
studied in NMLs. Figs. 7–9 shows effects of layer waviness for Cu/Fe 300 regions of small and large thickness resulting in more uniform defor­
nm NML that has different WA and WL along the TD and RD. Pillar mation and delayed KB formation.
compression tests were conducted on Cu/Fe 300 NML along TD and RD
with different WA (198 nm versus 415 nm) and WL (7960 nm versus 3.2.5. Material-dependent kink band formation
7040 nm) (Fig. 7). Fig. 8 shows that four out of five pillars with the KB formation also depends on the materials system and interface
larger WA and smaller WL (RD compression) kinked compared to two out characteristics. The microstructures and kinking behaviors in Cu/Fe 50
of five pillars with the smaller WA and larger WL (TD compression). nm and Al-4Mg/Fe 50 nm NMLs are compared in Figs. 10–12 and
Waviness measurements on individual pillars tested in Fig. 9(a) indicate Table 2. Overall, Al-4Mg/Fe has a higher KBP than Cu/Fe NML with
a greater WA for kinked pillars (~ 1–1.5 h) than for the unkinked pillars differences in their kinking behaviors summarized as follows: (1) The Al-
(0.5–1.5 h). Our LS-EVP-FFT simulations in Fig. 9(c) also demonstrate 4Mg/Fe 50 nm NML has a lower KB initiation strain; (2) The Cu/Fe 50
that a larger WA induces a lower KB initiation strain and, therefore, a nm NML has fewer KBs per pillar; (3) KBs in the Al-4Mg/Fe NML have
lower KBP. In contrast, the influence of WL on kinking is less prominent. smaller widths (Table 2) and sharper KBBs (Figs. 12(c, d), S4).
Layer waviness has two primary effects on kinking. First, the pres­ It is not always possible to experimentally isolate effects of individual
ence of local waviness results in local perturbations of layer morphology microstructure parameters on KBP. The comparison between Cu/Fe 50
resulting in layer-parallel slip systems having non-zero resolved shear nm and Al-4Mg/Fe 50 nm is a case in point. Differences in the micro­
stresses (location 1 in Fig. 9(b2)). Kinking propensity should thus in­ structure factors discussed so far would indicate that KBP should be
crease with growing waviness amplitude and decrease with increasing higher in Cu/Fe. The layer thickness is the same and initial dislocation
wavelength. In addition, when local layers have out-of-phase waviness density is expected to be similar (2.4–3 × 1015 m− 2 ). Most notably, Al-

8
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 7. Microstructure for Cu/Fe 300 nm NMLs. (a, b) SEM images showing layer morphologies in Cu/Fe 300 nm NMLs along the TD and RD, respectively. (c, d)
Waviness amplitude and wavelength statistics along TD and RD in Cu/Fe 300 nm NMLs. (e) IPF orientation map (ND) for Cu/Fe 300 nm NMLs. (f1, f2) IPF maps for Cu
and Fe layers.

4Mg/Fe 50 nm has a considerably lower WA than Cu/Fe 50 nm (23 nm The higher KBP in Al-4Mg/Fe is likely due to differences in interface
compared to 48 nm) (Figs. 10 and S3) which is expected to increase the barrier strengths and constitutive phase properties. Our LS-EVP-FFT
KBP of Cu/Fe relative to Al-4Mg/Fe. It is also observed that Al-4Mg/Fe results in Fig. 13(a) indicate that layer interfaces with higher barrier
50 nm has a smaller layer-parallel grain size (Figs. 10, S3 and Table 1). strength will generate stronger back-stresses that inhibit layer-inclined
Our LS-EVP-FFT simulations (Fig. 14) show that KBP is mostly inde­ slip systems, activate layer-parallel slip systems, and lead to the for­
pendent of grain size when layer parallel grain sizes are larger than 12 h. mation of KBs (Fig. 13(b1, b2)). We attribute the high interface barrier
Since the average grain sizes in both systems exceed 12 h, grain size strength of Al-4Mg/Fe to its higher Koehler stress (τk), smaller fraction
effects are expected to be minor. Interestingly, the narrower KBs in Al- of well aligned slip systems across interfaces, and intermixed nature of
4Mg/Fe NML are likely a consequence of its smaller in-plane grain layer interface. Dislocation transmission across a layer interface requires
size because grain boundaries can impede layer-parallel gliding dislo­ overcoming the energy barrier arising from the change in dislocation
cations and suppress KB growth. core energy, known as the Koehler stress (τk), which can be estimated as

9
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 8. Effect of layer waviness on kink band formation. Cu/Fe 300 nm compressed in the transverse direction (TD) has a significantly lower waviness amplitude and
lower KBP than the samples compressed in the rolling direction (RD). Engineering stress-strain curves for (a) TD compression and (b) RD compression. (c, d)
Deformed pillars for TD and RD compressions.

[63]: Al-4Mg/Fe interfaces are more chemically intermixed due to their

/ negative heat of mixing (− 11 kJ/mol) [66], which could potentially
μ − μ2
τk = 1 μ1 sinθ 8π, (1) increase its interface barrier strength as well [33]. Therefore, the higher
μ1 + μ2
kinking propensity is most likely due to a higher interface barrier
strength in Al-4Mg/Fe NMLs. In addition, larger strength disparities,
Here, θ is the angle between layer interface and slip plane and μi is the
manifested as larger hardness or CRSS ratios between Al-4Mg and Fe
shear modulus of phase i. The calculated Koehler stress is 0.39 and 0.43
layers, can also contribute to higher kinking propensity because kinking
GPa for the Cu/Fe and Al-4Mg/Fe systems, respectively, indicating Al-
initiates in softer layers first (Fig. 13(c)). It is interesting to note that
4Mg/Fe interfaces have a higher barrier strength contribution from
Al-4Mg/Fe 50 nm has much smaller load drop magnitudes induced by
the Koehler stress. Interface character is another factor influencing the
kinking than Cu/Fe 50 nm (Table 2). We assume the larger load drop
barrier strength difference. Generally speaking, good alignment of slip
magnitude in Cu/Fe 50 nm is due to its higher strength anisotropy levels.
planes/directions can facilitate dislocation transmission across the layer
interfaces [64], and the Kurdjumov-Sachs (K-S) orientation relation [65]
3.2.6. Grain size and twinning effect
in bcc/fcc systems provides exact alignment of slip systems: {110}bcc ‖
Fig. 14 summarizes the effect of layer-parallel grain size on KB for­
{111}fcc, <111> bcc ‖<110>fcc (Fig. 11(e, f)). The Cu/Fe 50 nm has a
mation in NMLs. Fig. 14(a) shows the evolution of KB initiation strain
higher fraction of interfaces close to the K-S relationship (~53%)
with grain size normalized by layer thickness (h = 80 nm). A smaller KB
compared to the Al-4Mg/Fe 50 nm (~40%) (Fig. 11(c, d), Table 2),
initiation strain indicates an increased kink banding propensity (KBP).
suggesting that Al-4Mg/Fe interfaces are less transmissible. In addition,

10
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 9. (a) Statistics showing the wavelength and waviness amplitude normalized by layer thickness for individual Cu/Fe 300 nm pillars. (b1, b2) Schematics showing
the effect of layer waviness on KB nucleation. The kink band initiation strain plotted as a function of (c) waviness amplitude and (d) wavelength normalized by
layer thickness.

Notably, the KB initiation strain is constant when the grain size is larger Ag/Fe 50 nm NMLs (Fig. S6). In the current study, deformation twins
than 12 h. As grain size decreases from 12 h to 8 h, the KB initiation were observed in ~ 4% of Ag grains in the as-rolled Ag/Fe 50 nm NMLs
strain decreases slightly because the introduced grain boundaries lead to and ~12% Cu grains in as-rolled Cu/Fe 50 nm NMLs originating from
more stress concentrations for KB nucleation. However, as layer-parallel the ARB process [67]. Moreover, deformation twins observed in ARB
grain size further decreases below 8 h, KB formation is suppressed with a processed NMLs are inclined to layer interfaces at an angle of ~19.5◦
rapid increase in KB initiation strain. As shown in Fig. 14(b3) and (b4), (Fig. S6). Our post-compression analyses reveal no significant increase
the KB formation is fully suppressed when the layer-parallel grain size is in twin density after pillar compression as indicated in Fig. S6(b, c). This
comparable to or smaller than the layer thickness. It is known that the study showed that the activation of KBs is associated with the transition
formation of KBs is achieved due to a transition to layer-parallel slip and from layer-inclined to layer-parallel slip. Thus, pre-existing deformation
subsequent accumulation of GNDs and GNBs near KBBs. Reducing the twins are expected to impede layer-parallel gliding dislocations, leading
layer-parallel grain size impedes the layer-parallel slip by generating slip to a decreased kinking propensity. If deformation twins had formed
and crystallographic discontinuities. Reduced layer-parallel grain size during the pillar compression tests, they would likely lower the kinking
leads to a smaller dislocation mean free paths and smaller differences in propensity of NMLs because twinning would act as competing defor­
the relative slip resistance between layer-inclined and layer-parallel slip mation mechanisms and postpone local layer-parallel slip.
systems. As a result, the layer-parallel slip systems are less operative
than layer-inclined slip systems, resulting in a lower kinking propensity. 3.2.7. Extrinsic factors
Furthermore, grain boundaries also limit the thickening of KBs by The primary goal of this work is to examine how differences in
impeding motion of GNBs into the unkinked regions. intrinsic microstructural attributes affect kink band propensity. Apart
Additionally, we observed deformation twins in as-rolled Cu/Fe and from the intrinsic factors mentioned above, other aspects will likely also

11
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 10. (a–d) IPF orientation maps (ND) and (e, f) BF-TEM images for Cu/Fe 50 nm and Al-4Mg/Fe 50 nm NMLs.

Fig. 11. (a, b) Phase maps overlapped with image quality (IQ) maps for Cu/Fe 50 nm and Al-4Mg/Fe 50 nm NMLs. (c, d) IQ maps with different layer interfaces
highlighted in red (KS interface) and green (non-KS layer interfaces). (e, f) High-resolution TEM images (HRTEM) showing K-S interfaces in Cu/Fe and Al-4Mg/Fe 50
nm NMLs. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

12
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 12. Effect of interface barrier strength to dislocation transmission and constitutive material properties on KBP. (a, b) Engineering stress-strain curves with
deformed pillars for Cu/Fe and Al-4Mg/Fe 50 nm NMLs. (c) Inverse pole figure (IPF) orientation map showing a KB in a kinked Cu/Fe pillar. (d) High-angle annular
dark field (HAADF) showing two KBs in Al-4Mg/Fe 50 nm NML.

Table 2
Comparison of physical properties and pillar compression results for Cu/Fe and Al-4Mg/Fe NMLs.
Sample Koehler Grain Interface Waviness KS type KB KB KBs per/ Strength Load drop Hardness
stress size structure amplitude/ interface initiation width pillar anisotropy magnitude ratio Cu,
(GPa) (nm) wavelength fraction strain (%) (µm) (ε=0.25) (ND/45◦ ) (%) Al/Fe
Cu, Al/ (nm)
Fe

Cu/Fe 0.39 1060 / Sharp, 48 / 2590 53 ± 5% 8.2 ± 2.7 1.7 ± 1.5 6.8 43.9 ± 7.5% 0.64
50 nm 2820 slightly 0.3
miscible
Al-4Mg/Fe 0.43 600 / Intermixed, 23 / 1300 40 ± 5% 7.2 ± 1.8 1.4 ± 1.75 3.3 19.7 ± 6.5% 0.52
50 nm 1880 reactive 0.4

impact KB formation processes in NMLs including testing conditions, have generic significance that agrees well with our EVP-FFT simulations.
sample dimensions, strain rate, and temperature. KB propagation and KB
morphologies observed during micropillar compression will differ (i) Effect of pillar aspect ratio
somewhat from those in bulk compression tests [5] due to sample size,
friction conditions, and sample fixtures. However, since all our pillar Some prior studies [68] show that nucleation of kink bands in some
compression tests were conducted for similar sample geometries and fiber materials are due to elastic buckling of samples. The critical
testing conditions, the trends in KB propensity based on different buckling load (Pcr) for a pillar can be written as:
microstructural attributes measured using micro pillar compression

13
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 13. (a) Kink band initiation strain plotted as a function of interface barrier strength. (b1, b2) Schematics show the effect of interface barrier strength on
activation of layer-parallel slip. (c) A plot of kink band initiation strain with the critical resolved shear stress (CRSS) ratio between BCC and FCC phases.

Fig. 14. (a) A plot showing the evolution of KB initiation strain with grain size normalized by layer thickness. (b1–b4) Von Misses strain field contour maps for
deformed pillars with different grain sizes. The grain size for b1-b4 is 9.17, 4.63, 2.33 and 1.0 h (layer thickness).

π2 EI versus 960 MPa for Pillar 3&4 (Fig. 15(a)). However, if KBs were initi­
Pcr = (2)
(KL)2 ated by elastic buckling, the KB initiation stress would be expected to be
significantly lower for Pillar 1&2 as predicted by Eq (2). Therefore, our
In this equation, E is the Young’s modulus of the material, I is the experiments disqualify elastic buckling as a possible mechanism for
minimum second moment of area of the cross-section of the sample, L is kinking in nano metallic laminates, and that plastic buckling occurs
the unsupported length of the sample (pillar height), and K is the during the kink band formation process.
effective length factor that is ~0.75 for pillar compression. According to
Eq. (2), at the fixed pillar cross-section area, the critical buckling load (ii) Strain rate and temperature effect
should decrease as pillar height increases.
To test whether kinking in NMLs is induced by elastic buckling, we As we discuss in previous sections, the formation of kink bands in
fabricated Cu/Fe 50 nm pillars with similar cross-section areas NMLs involves the transition from layer-inclined to layer-parallel slip
((3.5–3.8) × (3.5–3.8) µm2) and different pillar heights (Fig. 15). All and accumulation of dislocation structures at kink band boundaries
other pillar compression testing conditions were the same. The strain (KBBs). The KB formation is likely to exhibit dependencies on strain rate
rate was 2 × 10− 3 s− 1. Results are shown in Fig. 15. Pillar 2 (aspect ratio and temperature. Several factors are expected to affect the sensitivity of
of 4.6) and Pillar 4 (aspect ratio of 2.8) exhibit similar kinking behavior. KB to temperature and strain rate. Specifically, the generation and
KBs initiated from the pillar top followed by one or more KBs forming at annihilation of dislocations are temperature and strain rate sensitive,
the bottom and/or middle of the Cu/Fe pillars. Interestingly, the KB and will affect the transition from layer-inclined to layer-parallel slip. It
initiation stress was slightly higher for the pillars with the larger aspect is expected that KB propensity will increase with increasing strain rate
ratio. The average KB initiation stress was 1040 MPa for Pillar 1&2 via facilitating the activation of layer-parallel slip and dislocation

14
Y. Zhang et al. Acta Materialia 260 (2023) 119342

Fig. 15. Pillar compression of Cu/Fe 50 nm pillars with different aspect ratios. (a) Stress-strain curves for Cu/Fe 50 nm pillars. SEM images show pillar morphology
for pillar 2 and pillar 4 before and after the pillar compression test.

accumulation at KBBs. However, higher temperature should decrease KB In general, KBs will nucleate from locations with stress and/or
propensity by increasing the dislocation recovery rate and possibly microstructural perturbations including pillar corners, sample surfaces,
activating competing deformation mechanisms. Furthermore, the fric­ distorted layers, and layers with local waviness [46]. A common
tion stress (Peierls stress) is temperature-dependent, particularly in BCC nucleation site is the sample corner during compression experiments.
metals in which significant drops in friction stresses are observed at high This is the case even for samples with millimeter dimensions and good
temperatures. These would likely lead to high dislocation activity, thus lubrication [5]. Due to the inherent limitations of pillar compression,
potentially counteracting the effects of dislocation recovery. experimental conditions including friction between the indenter and
Currently, the limited available studies provide an incomplete pic­ pillar, alignment between the pillar and indenter, and sample geometry
ture of this problem. For instance, Zhen et al. [69] showed that the can affect KB formation by altering the stress condition at the pillar top.
stress-strain responses associated with KB formation in Ti3SiC2 were For instance, the friction on the pillar top can introduce triaxial stress
strain rate and temperature dependent. Maloy et al. [70] investigated states on the pillar top and pillar corners, leading to early activation of
the kinking behavior in single-crystal NiAl at various temperatures (76 K dislocation activities on these locations [72]. Raabe et al. [73] demon­
to 773 K) and strain rates (0.001–2000 s− 1), and revealed that kinking strated that increasing friction coefficient contributes to increasing
propensity decreased with increasing strain rate or decreasing temper­ geometrical stability of pillars. In addition to friction, misalignment
ature. Opposed to this, Ueji et al. [71] studied the kinking behavior of a between indenter and pillar top will generate asymmetrical stress dis­
pearlitic steel at different temperatures ranging from 25 to 600 ◦ C, and tributions and subsequent geometrical instabilities [74].
showed that kinking occurred more frequently in the temperature range While the perturbations determine where a kink will form, it is the
of 25–400 ◦ C. material property (waviness, layer thickness, dislocation density, con­
In NMLs, we postulate that higher strain rates will lead to higher stituents, etc.) that determines how how prone the sample is to the
kinking propensity while higher temperatures have the opposite effect. instability that becomes the kink. In this study, the same processing and
Our hypothesis is based on the assumption that higher strain rates will testing conditions were maintained to minimize variations among pil­
facilitate dislocation accumulation at KBBs while higher temperatures lars. The pillars were cut with similar rectangular cross-section geome­
will increase the dislocation recovery rate and postpone the transition to tries, samples were mounted in a manner to assure good tilt alignment
layer parallel slip. Due to its complexity, the effects of strain rate and between the indenter and the pillars, and the indenter was aligned
temperatures on KB formation are clearly worthy of their own system­ horizontally (Fig. S8). The roughness of pillar tops is similar with some
atic future study. minor variations (Fig. S7), minimizing differences in friction between
pillar tops. The pillars were all tested with consistent experimental
(iii) Effect of variations in experimental setup conditions that should result in similar stress concentrations among
pillars. However, there are likely small but random variations in pillar

15
Y. Zhang et al. Acta Materialia 260 (2023) 119342

top roughness and alignment conditions that could generate slightly Declaration of Competing Interest
different stress conditions at the pillar top. Because the small variations
in testing conditions are random, there are no systematic variations in The authors declare that they have no known competing financial
testing conditions that would explain the obvious microstructural cor­ interests or personal relationships that could have appeared to influence
relations observed in this study. the work reported in this paper.
The correlation of kink band propensity with layer waviness shown
in Fig. 9(a) is a clear example. Ten pillars were tested from two samples Acknowledgements
(Cu/Fe 300 nm RD and Cu/Fe 300 nm TD) that have different average
layer waviness characteristics, but the waviness characteristics of the We acknowledge the funding support from the Laboratory Directed
individual pillars were also directly measured. Of the six pillars that Research and Development (LDRD) programs (20200182DR and
kinked, four were from the RD sample and two were from the TD sample. 20230391ER) at Los Alamos National Laboratory. Electron microscopy
More importantly, the six kinked pillars had the six highest waviness was conducted at the Electron Microscopy Lab (EML) at Los Alamos
amplitudes among the 10 pillars. Again, since the samples were nomi­ National Laboratory. Part of this work was performed at the Center for
nally identical with only small random experimental variations Integrated Nanotechnologies, an Office of Science User Facility operated
(Figs. S7, S8), the chance of the observed correlation being attributable for the U.S. Department of Energy (DOE) Office of Science. Los Alamos
to systematic experimental variations is low. Similarly with the other National Laboratory, an affirmative action equal opportunity employer,
microstructural parameters, the clear trends in kinking propensity is managed by Triad National Security, LLC for the U.S. Department of
cannot be due to random and minor variations in experimental condi­ Energy’s NNSA, under contract 89233218CNA000001.
tions in this study. Hence, the clearly varying trends in kinking pro­
pensity are attributable to variations in microstructural attributes. Supplementary materials
Moreover, these experimentally observed trends match well with the
predictions of our EVP-FFT simulations. Supplementary material associated with this article can be found, in
Additionally, stress concentrations on pillar corners do not usually the online version, at doi:10.1016/j.actamat.2023.119342.
lead to kink band formation in other materials. Stress concentrations on
pillar corners can advance dislocation activities as depicted in step 1 in References
Fig. 1(a). Nevertheless, how pillars deform and fail is mainly determined
by the dislocation glide modes associated with microstructural factors [1] J.S. Poulsen, P.M. Moran, C.F. Shih, E. Byskov, Kink band initiation and band
broadening in clear wood under compressive loading, Mech. Mater. 25 (1997)
including layer orientations, layer thickness, interface properties etc. 67–77, https://doi.org/10.1016/S0167-6636(96)00043-9.
[27,42,75]. It is also noted that pillar corners are not the only available [2] M.W. Barsoum, A. Murugaiah, S.R. Kalidindi, T. Zhen, Kinking nonlinear elastic
nucleation sites for kink bands. As shown in Fig. S9, some kinked pillars solids, nanoindentations, and geology, Phys. Rev. Lett. 92 (2004) 1–4, https://doi.
org/10.1103/PhysRevLett.92.255508.
only have kink band nucleating from their middle portions. [3] A.S. Wronski, T.V. Parry, Compressive failure and kinking in uniaxially aligned
Overall, our consistent and careful experimental setup for pillar glass-resin composite under superposed hydrostatic pressure, J. Mater. Sci. 17
compression results in similar stress distributions in tested pillars. Slight (1982) 3656–3662, https://doi.org/10.1007/BF00752211.
[4] M.E. Manley, E.M. Schulson, Kinks and cracks in S1 ice under across-column
and random variations of friction and misalignment conditions cannot compression, Philos. Mag. Lett. 75 (1997) 83–90, https://doi.org/10.1080/
cause the clearly observed trends in kinking propensity. Therefore, it is 095008397179787.
reasonable to conclude that the effects of microstructural factors on kink [5] T.J. Nizolek, M.R. Begley, R.J. McCabe, J.T. Avallone, N.A. Mara, I.J. Beyerlein, T.
M. Pollock, Strain fields induced by kink band propagation in Cu-Nb nanolaminate
band formation revealed by this work are reliable.
composites, Acta Mater. 133 (2017) 303–315, https://doi.org/10.1016/j.
actamat.2017.04.050.
4. Conclusions [6] M.W. Barsoum, T. Zhen, S.R. Kalidindi, M. Radovic, A. Murugaiah, Fully reversible,
dislocation-based compressive deformation of Ti3SiC2 to 1 GPa, Nat. Mater. 2
(2003) 107–111, https://doi.org/10.1038/nmat814.
In summary, defining the mechanism of kink band formation and its [7] E. Orowan, A type of plastic deformation new in metals, Nature 149 (1942)
dependence on microstructural attributes is necessary for thoroughly 643–644.
comprehending and controlling failure behaviors of NMLs. Moreover, [8] T. Nizolek, N.A. Mara, I.J. Beyerlein, J.T. Avallone, T.M. Pollock, Enhanced
plasticity via kinking in cubic metallic nanolaminates, Adv. Eng. Mater. 17 (2015)
NMLs are model materials for studying kinking of plastically aniso­ 781–785, https://doi.org/10.1002/adem.201400324.
tropic, polycrystalline materials because their microstructure and ma­ [9] R.E. Robertson, Formation of kink bands in oriented polymers, J. Polym. Sci. Part
terial parameters can be readily varied to probe their influences on KBP. A-2 Polym. Phys. 7 (1969) 1315–1328.
[10] T.J. Nizolek, T.M. Pollock, R.M. McMeeking, Kink band and shear band
However, kinking in NMLs cannot be rationalized based on previous localization in anisotropic perfectly plastic solids, J. Mech. Phys. Solids 146 (2021),
models developed for fiber reinforced polymers [76] or single-phase 104183, https://doi.org/10.1016/j.jmps.2020.104183.
materials [2,77]. [11] A.S. Argon, Fracture of composites, Treatise Mater. Sci. Technol. 1 (2013) 79–114.
[12] T. Matsumoto, M. Yamasaki, K. Hagihara, Y. Kawamura, Configuration of
The results presented here are the first in-depth and systematic study dislocations in low-angle kink boundaries formed in a single crystalline long-period
detailing microstructure effects on kinking in NMLs using combined stacking ordered Mg-Zn-Y alloy, Acta Mater. 151 (2018) 112–124, https://doi.org/
experimental and modeling results. We show that kinking initiates due 10.1016/j.actamat.2018.03.034.
[13] M.W. Barsoum, L. Farber, I. Introduction, dislocations, kink bands, and room-
to a transition from layer-inclined to layer-parallel slip dominated
temperature plasticity, Metall. Mater. Trans. A 30 (1999) 1727–1738.
deformation. The transition is accompanied by layer and crystal rota­ [14] M.W. Barsoum, A. Murugaiah, S.R. Kalidindi, T. Zhen, Y. Gogotsi, Kink bands,
tions that further enhance layer-parallel slip and can result in geometric nonlinear elasticity and nanoindentations in graphite, Carbon 42 (2004)
softening. Changes in relevant microstructure parameters that promote 1435–1445, https://doi.org/10.1016/j.carbon.2003.12.090. N. Y.
[15] M.W. Barsoum, T. Zhen, A. Zhou, S. Basu, S.R. Kalidindi, Microscale modeling of
the transition from layer-inclined slip to layer-parallel slip increase the kinking nonlinear elastic solids, Phys. Rev. B Condens. Matter Mater. Phys. 71
kink banding propensity. The roles that waviness amplitude and wave­ (2005) 3–10, https://doi.org/10.1103/PhysRevB.71.134101.
length, layer thickness, interface barrier strength, dislocation density [16] F.C. Frank, A.N. Stroh, On the theory of kinking, Proc. Phys. Soc. Sect. B 65 (1952)
811–821, https://doi.org/10.1088/0370-1301/65/10/311.
(internal stresses), and constituent critical resolved shear stress ratio [17] A.G. Zhou, S. Basu, M.W. Barsoum, Kinking nonlinear elasticity, damping and
were determined, with layer thickness and waviness amplitude having microyielding of hexagonal close-packed metals, Acta Mater. 56 (2008) 60–67.
the most prominent effects. Overall, this study gives new insights into [18] S. Basu, M.W. Barsoum, S.R. Kalidindi, Sapphire: a kinking nonlinear elastic solid,
J. Appl. Phys. 99 (2006), https://doi.org/10.1063/1.2179974.
the kinking behavior in NMLs, which in turn can help us design more [19] Y.F. Zhang, S. Xue, Q. Li, J. Li, J. Ding, T.J. Niu, R. Su, H. Wang, X. Zhang, Size
failure-resistant metallic materials. dependent strengthening in high strength nanotwinned Al/Ti multilayers, Acta
Mater. 175 (2019) 466–476.

16
Y. Zhang et al. Acta Materialia 260 (2023) 119342

[20] J. Wang, Q. Zhou, S. Shao, A. Misra, Strength and plasticity of nanolaminated [45] M.W. Kapp, A. Hohenwarter, S. Wurster, B. Yang, R. Pippan, Anisotropic
materials, Mater. Res. Lett. 5 (2017) 1–19, https://doi.org/10.1080/ deformation characteristics of an ultrafine- and nanolamellar pearlitic steel, Acta
21663831.2016.1225321. Mater. 106 (2016) 239–248, https://doi.org/10.1016/j.actamat.2015.12.037.
[21] Y.F. Zhang, Q. Li, M. Gong, S. Xue, J. Ding, J. Li, J. Cho, T. Niu, R. Su, N.A. Richter, [46] Y. Zhang, N. Li, M.M. Schneider, T.J. Nizolek, L. Capolungo, R.J. McCabe, Kink
Deformation behavior and phase transformation of nanotwinned Al/Ti multilayers, mechanism in Cu/Nb nanolaminates explored by in situ pillar compression, Acta
Appl. Surf. Sci. 527 (2020) 146776. Mater. 237 (2022), 118150, https://doi.org/10.1016/j.actamat.2022.118150.
[22] Y.F. Zhang, R. Su, T.J. Niu, N.A. Richter, S. Xue, Q. Li, J. Ding, B. Yang, H. Wang, [47] M. Zecevic, R.A. Lebensohn, L. Capolungo, Non-local large-strain FFT-based
X. Zhang, Thermal stability and deformability of annealed nanotwinned Al/Ti formulation and its application to interface-dominated plasticity of nano-metallic
multilayers, Scr. Mater. 186 (2020), https://doi.org/10.1016/j. laminates, J. Mech. Phys. Solids 173 (2023), 105187.
scriptamat.2020.04.015. [48] R.J. McCabe, T.J. Nizolek, N. Li, Y. Zhang, D.R. Coughlin, C. Miller, J.S. Carpenter,
[23] S. Zheng, I.J. Beyerlein, J.S. Carpenter, K. Kang, J. Wang, W. Han, N.A. Mara, High- Evolution of microstructures and properties leading to layer instabilities during
strength and thermally stable bulk nanolayered composites due to twin-induced accumulative roll bonding of FeCu, FeAg, and FeAl, Mater. Des. 212 (2021),
interfaces, Nat. Commun. 4 (2013) 1696–1698, https://doi.org/10.1038/ 110204.
ncomms2651. [49] M. Zecevic, R.A. Lebensohn, L. Capolungo, New large-strain FFT-based formulation
[24] Y. Zhang, J.G. Gigax, T.J. Nizolek, J.S. Carpenter, M.M. Schneider, N. Li, and its application to model strain localization in nano-metallic laminates and
L. Capolungo, R.J. McCabe, Tensile and failure behaviors of Cu/Nb nanolaminates: other strongly anisotropic crystalline materials, Mech. Mater. 166 (2022), 104208.
the effects of loading direction, layer thickness, and annealing, Acta Mater. 240 [50] R.A. Lebensohn, A. Needleman, Numerical implementation of non-local polycrystal
(2022), https://doi.org/10.1016/j.actamat.2022.118346. plasticity using fast Fourier transforms, J. Mech. Phys. Solids 97 (2016) 333–351.
[25] W.Z. Han, N.A. Mara, Y.Q. Wang, A. Misra, M.J. Demkowicz, He implantation of [51] R.A. Lebensohn, A.K. Kanjarla, P. Eisenlohr, An elasto-viscoplastic formulation
bulk Cu-Nb nanocomposites fabricated by accumulated roll bonding, J. Nucl. based on fast Fourier transforms for the prediction of micromechanical fields in
Mater. 452 (2014) 57–60, https://doi.org/10.1016/j.jnucmat.2014.04.034. polycrystalline materials, Int. J. Plast. 32–33 (2012) 59–69, https://doi.org/
[26] N.A. Mara, D. Bhattacharyya, J.P. Hirth, P. Dickerson, A. Misra, Mechanism for 10.1016/j.ijplas.2011.12.005.
shear banding in nanolayered composites, Appl. Phys. Lett. 97 (2010), https://doi. [52] H. Moulinec, P. Suquet, A numerical method for computing the overall response of
org/10.1063/1.3458000. nonlinear composites with complex microstructure, Comput. Methods Appl. Mech.
[27] S.J. Zheng, J. Wang, J.S. Carpenter, W.M. Mook, P.O. Dickerson, N.A. Mara, I. Eng. 157 (1998) 69–94.
J. Beyerlein, Plastic instability mechanisms in bimetallic nanolayered composites, [53] M.E. Gurtin, A gradient theory of single-crystal viscoplasticity that accounts for
Acta Mater. 79 (2014) 282–291, https://doi.org/10.1016/j.actamat.2014.07.017. geometrically necessary dislocations, J. Mech. Phys. Solids 50 (2002) 5–32.
[28] J. Wang, C. Yang, P.D. Hodgson, Strain gradients in Cu–Fe thin films and [54] C.J. Bayley, W.A.M. Brekelmans, M. Geers, A comparison of dislocation induced
multilayers during micropillar compression, Mater. Sci. Eng. A 651 (2016) back stress formulations in strain gradient crystal plasticity, Int. J. Solids Struct. 43
146–154. (2006) 7268–7286.
[29] N. Li, N.A. Mara, J. Wang, P. Dickerson, J.Y. Huang, A. Misra, Ex situ and in situ [55] D.L. McDowell, Viscoplasticity of heterogeneous metallic materials, Mater. Sci.
measurements of the shear strength of interfaces in metallic multilayers, Scr. Eng. R Rep. 62 (2008) 67–123.
Mater. 67 (2012) 479–482, https://doi.org/10.1016/j.scriptamat.2012.06.008. [56] A. Saltelli, S. Tarantola, F. Campolongo, M. Ratto, Sensitivity Analysis in Practice:
[30] X. Liu, R. Hao, S. Mao, S.J. Dillon, Shear strengths of FCC-FCC cube-on-cube A Guide to Assessing Scientific Models, Wiley Online Library, 2004.
interfaces, Scr. Mater. 130 (2017) 178–181. [57] B.L. Hansen, J.S. Carpenter, S.D. Sintay, C.A. Bronkhorst, R.J. McCabe, J.
[31] X. Ma, B. Gwalani, J. Tao, M. Efe, M. Olszta, M. Song, S. Yadav, A. Yu, T.J. Nizolek, R. Mayeur, H.M. Mourad, I.J. Beyerlein, N.A. Mara, S.R. Chen, G.T. Gray, Modeling
J.S. Carpenter, Shear strain gradient in Cu/Nb nanolaminates: strain the texture evolution of Cu/Nb layered composites during rolling, Int. J. Plast. 49
accommodation and chemical mixing, Acta Mater. 234 (2022) 117986. (2013) 71–84, https://doi.org/10.1016/j.ijplas.2013.03.001.
[32] J. Snel, M.A. Monclús, M. Castillo-Rodríguez, N. Mara, I.J. Beyerlein, J. Llorca, J. [58] J. Wang, Q. Zhou, S. Shao, A. Misra, Strength and plasticity of nanolaminated
M. Molina-Aldareguía, Deformation mechanism map of Cu/Nb nanoscale metallic materials, Mater. Res. Lett. 5 (2017) 1–19.
multilayers as a function of temperature and layer thickness, JOM 69 (2017) [59] Z.C. Cordero, B.E. Knight, C.A. Schuh, Six decades of the Hall–Petch effect–a survey
2214–2226, https://doi.org/10.1007/s11837-017-2533-1. of grain-size strengthening studies on pure metals, Int. Mater. Rev. 61 (2016)
[33] J.Y. Cheng, S. Xu, Y. Chen, Z. Li, J.K. Baldwin, I.J. Beyerlein, N.A. Mara, 495–512.
Simultaneous high-strength and deformable nanolaminates with thick biphase [60] P.J. Withers, H.K.D.H. Bhadeshia, Residual stress part 2 - nature and origins, Mater.
interfaces, Nano Lett. 22 (2022) 1897–1904. Sci. Technol. 17 (2001) 366–375, https://doi.org/10.1179/
[34] Y. Cui, B. Derby, N. Li, N.A. Mara, A. Misra, Suppression of shear banding in high- 026708301101510087.
strength Cu/Mo nanocomposites with hierarchical bicontinuous intertwined [61] P.J. Withers, H.K.D.H. Bhadeshia, Residual stress part 1 - measurement techniques,
structures, Mater. Res. Lett. 6 (2018) 184–190, https://doi.org/10.1080/ Mater. Sci. Technol. 17 (2001) 355–365, https://doi.org/10.1179/
21663831.2018.1431315. 026708301101509980.
[35] P. Dayal, N. Savvides, M. Hoffman, Characterisation of nanolayered aluminium/ [62] S. Kyriakides, R. Arseculeratne, E.J. Perry, K.M. Liechti, On the compressive failure
palladium thin films using nanoindentation, Thin Solid Films 517 (2009) of fiber reinforced composites, Int. J. Solids Struct. 32 (1995) 689–738, https://
3698–3703, https://doi.org/10.1016/j.tsf.2009.01.174. doi.org/10.1016/0020-7683(94)00157-R.
[36] F. Liang, Z.X. Wang, Y.W. Luo, B. Zhang, X.M. Luo, G.P. Zhang, Enhancing co- [63] S.I. Rao, P.M. Hazzledine, Atomistic simulations of dislocation-interface
deformation ability of nanograined Ni-W layers in the Ni/Ni-W laminated interactions in the Cu-Ni multilayer system, Philos. Mag. A Phys. Condens. Matter
composites, Acta Mater. 216 (2021), 117138. Struct. Defects Mech. Prop. 80 (2000) 2011–2040, https://doi.org/10.1080/
[37] T. Müller, M.W. Kapp, A. Bachmaier, P. Felfer, R. Pippan, Ultrahigh-strength low 01418610008212148.
carbon steel obtained from the martensitic state via high pressure torsion, Acta [64] J. Wang, A. Misra, An overview of interface-dominated deformation mechanisms in
Mater. 166 (2019) 168–177, https://doi.org/10.1016/j.actamat.2018.12.028. metallic multilayers, Curr. Opin. Solid State Mater. Sci. 15 (2011) 20–28, https://
[38] L. Zhao, Q. Guo, Z. Li, D.B. Xiong, S. Osovski, Y. Su, D. Zhang, Strengthening and doi.org/10.1016/j.cossms.2010.09.002.
deformation mechanisms in nanolaminated graphene-Al composite micro-pillars [65] M.J. Demkowicz, R.G. Hoagland, Structure of Kurdjumov–Sachs interfaces in
affected by graphene in-plane sizes, Int. J. Plast. 116 (2019) 265–279, https://doi. simulations of a copper–niobium bilayer, J. Nucl. Mater. 372 (2008) 45–52.
org/10.1016/j.ijplas.2019.01.006. [66] A. Takeuchi, A. Inoue, Classification of bulk metallic glasses by atomic size
[39] Y. Cui, B. Derby, N. Li, A. Misra, Design of bicontinuous metallic nanocomposites difference, heat of mixing and period of constituent elements and its application to
for high-strength and plasticity, Mater. Des. 166 (2019), 107602, https://doi.org/ characterization of the main alloying element, Mater. Trans. 46 (2005) 2817–2829.
10.1016/j.matdes.2019.107602. [67] R.J. McCabe, I.J. Beyerlein, J.S. Carpenter, N.A. Mara, The critical role of grain
[40] A. Röthlisberger, S. Häberli, F. Krogh, H. Galinski, D.C. Dunand, R. Spolenak, Ice- orientation and applied stress in nanoscale twinning, Nat. Commun. 5 (2014) 1–7.
templated W-Cu composites with high anisotropy, Sci. Rep. 9 (2019) 1–9, https:// [68] W. Sun, A.P. Vassilopoulos, T. Keller, Experimental investigation of kink initiation
doi.org/10.1038/s41598-018-36604-9. and kink band formation in unidirectional glass fiber-reinforced polymer
[41] S. Shen, Y. Chen, L. Yang, Y. Feng, J. Xu, Role of columnar structure on the fracture specimens, Compos. Struct. 130 (2015) 9–17, https://doi.org/10.1016/j.
anisotropy of physical vapor deposited Al/SiC nanolaminates, Mater. Sci. Eng. A compstruct.2015.04.028.
786 (2020), 139387, https://doi.org/10.1016/j.msea.2020.139387. [69] T. Zhen, M.W. Barsoum, S.R. Kalidindi, Effects of temperature, strain rate and grain
[42] C.R. Mayer, L.W. Yang, S.S. Singh, J. Llorca, J.M. Molina-Aldareguia, Y.L. Shen, size on the compressive properties of Ti3SiC2, Acta Mater. 53 (2005) 4163–4171.
N. Chawla, Anisotropy, size, and aspect ratio effects on micropillar compression of [70] S.A. Maloy, G.T. Gray III, R. Darolia, High strain rate deformation of NiAl, Mater.
Al-SiC nanolaminate composites, Acta Mater. 114 (2016) 25–32, https://doi.org/ Sci. Eng. A 192 (1995) 249–254.
10.1016/j.actamat.2016.05.018. [71] R. Ueji, H. Somekawa, T. Inoue, T. Hara, Compression temperature to activate
[43] T. Nizolek, N.A. Mara, I.J. Beyerlein, J.T. Avallone, J.E. Scott, T.M. Pollock, kinking in pearlitic steel, Mater. Sci. Eng. A 857 (2022), 144018.
Processing and deformation behavior of bulk Cu–Nb nanolaminates, Metallogr. [72] D. Kiener, C. Motz, G. Dehm, Micro-compression testing: a critical discussion of
Microstruct. Anal. 3 (2014) 470–476. experimental constraints, Mater. Sci. Eng. A 505 (2009) 79–87, https://doi.org/
[44] M. Ardeljan, M. Knezevic, M. Jain, S. Pathak, A. Kumar, N. Li, N.A. Mara, J. 10.1016/j.msea.2009.01.005.
K. Baldwin, I.J. Beyerlein, Room temperature deformation mechanisms of Mg/Nb [73] D. Raabe, D. Ma, F. Roters, Effects of initial orientation, sample geometry and
nanolayered composites, J. Mater. Res. 33 (2018) 1311–1332, https://doi.org/ friction on anisotropy and crystallographic orientation changes in single crystal
10.1557/jmr.2018.107. microcompression deformation: a crystal plasticity finite element study, Acta
Mater. 55 (2007) 4567–4583, https://doi.org/10.1016/j.actamat.2007.04.023.

17
Y. Zhang et al. Acta Materialia 260 (2023) 119342

[74] A. Cornec, E. Lilleodden, Numerical analysis of micropillar compression behaviour [76] B. Budiansky, N.A. Fleck, J.C. Amazigo, On kink-band propagation in fiber
and stress-strain curve estimation verified on glass fused silica, Mater. Today composites, J. Mech. Phys. Solids 46 (1998) 1637–1653, https://doi.org/10.1016/
Commun. 33 (2022), 104971, https://doi.org/10.1016/j.mtcomm.2022.104971. S0022-5096(97)00042-2.
[75] Y. Chen, N. Li, R.G. Hoagland, X.Y. Liu, J.K. Baldwin, I.J. Beyerlein, J.Y. Cheng, N. [77] M.W. Barsoum, L. Farber, T. El-Raghy, Dislocations, kink bands, and room-
A. Mara, Effects of three-dimensional Cu/Nb interfaces on strengthening and shear temperature plasticity of Ti3SiC2, Metall. Mater. Trans. 30 (1999) 1727.
banding in nanoscale metallic multilayers, Acta Mater. 199 (2020) 593–601.

18