APPLIED PHYSICS LETTERS 88, 061905 共2006兲
Deformation twinning mechanisms in nanocrystalline Ni
Xiao-Lei Wua兲
State Key Laboratory of Nonlinear Mechanics, Institute of Mechanics, Chinese Academy of Sciences,
Beijing 100080, China
En 共Evan兲 Mab兲
Department of Materials Science and Engineering, Johns Hopkins University, Baltimore, Maryland 21218
共Received 8 June 2005; accepted 16 December 2005; published online 7 February 2006兲
Extensive transmission electron microscopy examinations confirm that twinning does occur upon
large plastic deformation in nanocrystalline Ni, for which no sign of deformation twinning was
found in previous tensile tests. Compelling evidence has been obtained for several twinning
mechanisms that operate in nanocrystalline grains, with the grain boundary emission of partial
dislocations determined as the most proficient. © 2006 American Institute of Physics.
关DOI: 10.1063/1.2172404兴
Twinning has recently been identified as a contributing
plastic deformation mechanism in nanocrystalline 共nc兲
metals.1–7 Deformation twins and stacking faults, indicating
the presence of dislocation activities in nc grains and especially the operation of partial dislocation mediated processes,
were observed in Al,1–3 Cu,4 Pd,5,6 and Ta.7 The nc-Al and
nc-Pd cases are especially interesting, as these fcc metals
have high stacking fault energies and little or no chance of
undergoing deformation twinning in their coarse-grained
form 共conventional fcc Cu, in comparison, is known to twin
when deformed at low temperatures and/or high strain rates兲.
However, there appears to be a glaring contradicting example: recent in situ x-ray diffraction studies of fcc nc-Ni
subjected to tensile deformation showed no irreversible peak
broadening.8 It was concluded that the nc-Ni grains stored no
deformation debris such as twins, faults, and dislocations.8
We demonstrate in the following that deformation twinning does occur in nc-Ni, but provided it is exposed to extensive plastic deformation to strains far beyond what was
achievable in a tensile test 共⬃3% before localization and
fracture兲.8 We show that when nc-Ni having very high
strength is forced to deform to large strains, in the deforming
nanostructure there will inevitably be buildup of local stress
concentrations that are at sufficient levels for deformation
twinning to become a competitive response. Twinning becoming probable at high stresses and relatively large plastic
strains is in fact exactly what is predicted by molecular dynamics 共MD兲 simulations9–11 and analytical models1,12,13 for
nc fcc metals.
A more important aspect of fundamental interest is the
mechanisms of twinning in nc metals, when the pole mechanism often invoked to explain twinning in large grains becomes inactive 共mirroring the inability to activate the FrankRead sources to provide dislocations in the interior of
extremely small grains兲.9,10 Several alternative mechanisms
were in fact predicted by MD simulations to be operative in
nc-Al grains.9,10 Recently, three mechanisms were inferred
from the defect morphologies observed in a series of transmission electron microscopy 共TEM兲 experiments on nc-Al
and nc-Cu.2–4 In this respect, more compelling and direct
a兲
Electronic mail: [email protected]
Electronic mail: [email protected]
b兲
evidence is needed in more materials, as well as an assessment of which mechanism is actually the primary one responsible for twinning in nc grains. In this letter, we will
provide such information by illustrating the deformation
twinning mechanisms based on a large number of TEM observations in nc-Ni.
We first cold rolled an electroplated nc-Ni 共the asreceived 140-␮m-thick foil had a nominal average grain size
of 30 nm, as characterized before兲14–16 at cryogenic temperature to 46% rolling strain. Figure 1共a兲 is a high-resolution
TEM 共HRTEM兲 image showing two intersecting mechanical
twins, labeled as T1 and T2. The twin lamella T1 runs in the
horizontal direction with Shockley partial dislocations at the
FIG. 1. 共a兲 HRTEM image of intersecting twins 共T1 and T2, marked with
white lines兲 observed in cold-rolled nc Ni. A Burgers circuit starting at S and
ending at F shows that T2 contains an edge Shockley partial dislocation.
This enlarged view is for the white-boxed area in 共b兲. The overview of this
area inside the grain is shown in 共c兲.
0003-6951/2006/88共6兲/061905/3/$23.00
88, 061905-1
© 2006 American Institute of Physics
Downloaded 23 Mar 2006 to 159.226.230.75. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
061905-2
X. Wu and E. Ma
front boundaries, which are stopped in the grain interior by
T2. At a lower magnification, Fig. 1共b兲, T1 is observed to
come from the right-hand side of the marked box 关details
shown in Fig. 1共a兲兴, and the entire twin plate obviously originated from the grain boundary 共GB兲 共marked by the bottom
white arrow兲 as observed in Fig. 1共c兲 using a still lower
magnification showing the relation of the marked box with
the entire grain. Clearly, the formation mechanism of T1 is
heterogeneous nucleation at the GB and growth 共lengthening
and thickening兲 into the grain interior via partial dislocation
emission from the GB. We found many such examples in our
nc-Ni deformed by rolling and milling, indicating the dominant role of this mechanism in initiating twinning 共see later兲.
Other twin plates 关marked with arrows in Fig. 1共c兲兴, with the
twin boundary 共TB兲 running across the entire grain, are also
visible for the two grains 共A and B兲 included in Fig. 1共c兲.
The twin lamella T2, on the other hand, has both of its
ends terminated inside the grain. As shown with a Burgers
circuit in Fig. 1共a兲, T2 contains an edge Shockley partial
= 61 关2̄11̄兴 on a 共111̄兲
dislocation with a Burgers vector bshockley
p
5
plane. A mechanism to grow a twin inside the grain interior
is shown in Fig. 2, for an nc-Ni with an average grain size of
30 nm formed via ball milling of the surface of a Ni plate.
The twin relationship between T1 and the matrix lattice, A, is
seen in the fast Fourier transform in Fig. 2共b兲. T1 broadens,
as seen in the HRTEM in Fig. 2共c兲, by absorbing extended
partial dislocations from neighboring planes, i.e., via dynamic overlapping of stacking faults on adjacent planes.
Such a mechanism occurs when the concentration of the extended dislocations inside the grain is rather high. This
mechanism explains the formation of twins in a homogeneous fashion inside a grain, such as the T2 in Fig. 1. The
nucleation stage 共with the first two atomic planes兲 was
shown in Ref. 2.
Both of the heterogeneous and homogeneous twin formation mechanisms discussed above were observed before in
conventional fcc metals,10 predicted by MD simulations for
nc-Al,9,10 and discussed for nc-Al and nc-Cu.2–4 A third
mechanism, occurring heterogeneously at GBs, through the
splitting of a GB and subsequent migration of one of the
segments, was discussed in literature before and predicted to
occur in nc-Al by MD simulations as well.10 This mechanism
was believed to be active, in an interpretation by Liao et al.3
of the waviness of TBs observed in their deformed nc-Al. It
is thus useful to look into the possible origins of wavy TBs in
nc-Ni samples that contained many deformation twins. One
example of curved TB is that of a lenticular twin 共not
shown兲, heterogeneously nucleated from GB for example. In
this case the Shockley partials emitted from the GB terminates at different distances from the GB, leading to 共111兲
twin planes connected by incoherent, noncrystallographic
segments. Also, dislocations deposited and accumulated at a
TB can lead to curvature, as seen for the TB between the
matrix of grain B and its twin lattice 共T2兲, shown in Fig.
2共d兲.
Another type of wavy TB was also observed, as shown
in Fig. 2共a兲 for the TB a-b between T1 in grain A and T2 in
grain B. This TB consists of many straight coherent segments
共examples are marked with white lines兲, arranged in a parallel and orderly way, which are connected by incoherent, noncrystallographic segments. Similar features were observed in
nc-Al by Liao et al.3 T1 starts from around a, and broadens
towards the left into A while it lengthens, whereas T2 grows
Appl. Phys. Lett. 88, 061905 共2006兲
FIG. 2. 共a兲 HRTEM micrograph showing growing twins in nc-Ni, T1 in
grain A and T2 in grain B, that lengthen and broaden and eventually meet to
form a curved TB, a-b. A cluster of stacking faults is also observed nearby.
The twin relationship between grains A and B, and that between A / T1,
B / T2, and T1 / T2, are confirmed in the fast Fourier transformation of the
respective regions in 共b兲. 共c兲 HRTEM image showing the growth of T1 by
dynamic overlapping of extended partial dislocations, i.e., absorbing stacking faults on adjacent slip planes. 共d兲 The T2 / B TB is wavy as it is decorated by dislocations. Inset is the low magnification of grains A and B.
in the opposite directions starting from b, as shown in Fig.
2共a兲. As confirmed in Fig. 2共b兲, the two grains 共A and B兲
happen to be in a twin relationship. The twin formation in
each grain renders T1 and T2 in twin relationship as well.
This leads to a curved TB when the two growing twins 共T1
and T2兲 meet.
As such, there can be multiple possible reasons for the
wavy TBs observed before, not necessarily an indication of a
Downloaded 23 Mar 2006 to 159.226.230.75. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp
061905-3
Appl. Phys. Lett. 88, 061905 共2006兲
X. Wu and E. Ma
FIG. 3. HRTEM image of a group of grains in nc-Ni after surface milling
treatment. The formation of twin lamella and its boundaries can be explained based on the grain boundary splitting and migration mechanism.
particular twinning mechanism. We did, however, obtain for
the first time evidence for the GB splitting mechanism. Figure 3 is a HRTEM image showing several nc Ni grains
formed by milling of the surface of a Ni plate. Note that the
TB, A1-A2, is a part of a high-angle GB 共HiAGB兲 marked
with asterisks. The A2-C-B nearby is a HiAGB, which is a
part of a low-angle GB 共LoAGB兲 that extends beyond B.
Also note the presence of another TB, A1-B. These features
are reminiscent of the GB splitting mechanism shown by the
MD simulation:10 the subsequent migration of a GB segment
leaves behind two coherent TBs. Specifically, twinning takes
place via the dissociation of an initial segment of the HiAGB
into a TB 共A1-A2兲 and a new HiAGB. The TB remains at the
same position as before the dissociation, while the migration
of the HiAGB segment to the new 共A2-B-C兲 location produces the twin lamella 共A1-B-C-A2-A1兲, which is separated
from the nontwinned lattice by a second TB 共A1-B兲. That
A2-C-B coincides with a pre-existing LoAGB suggests that
the migrating HiAGB is eventually blocked by the LoAGB.
We stress again that this is the first direct evidence of this GB
splitting mechanism for deformation twinning.
Finally, a remark can be made with respect to the results
of previous tensile tests, where no sign of twinning was
observed.8 For the small plastic strains 共⬃3 % 兲 sustainable in
tension,8 the initial portion of the stress-strain curve involved
apparent bending, in a process possibly using up sources of
dislocations that are pre-existing or relatively easy to escape,
and GB segments that are relatively easy to shear/slide.17
Twinning was apparently not yet competitive with other possible means that can produce the plastic strain. Unfortunately
further tensile straining was not possible beyond the point
where the 共necking兲 instability set in;8 otherwise twinning
could have come in after a certain level of stress 共concentrations兲 is reached, as is the case in our large-strain rolling/
milling experiments. Note that all the experimental twining
evidence in nc metals so far are for deformation conditions
that result in high deformation stresses, owing to high strain
rates and/or large plastic deformation leading to saturated
defect levels, such as indentation,1 grinding,1 high-pressure
torsion,4 high-rate cold rolling,5,6 and ball milling 共sometimes with powders immersed in liquid nitrogen2,3兲. In fact, it
was also observed that there was a threshold in the strain rate
used in rolling, above which twinning became observable in
nc-Pd.5 We conclude that high stress deformation conditions
are needed to initiate deformation twinning in nc metals. But
such high deformation stresses are more likely for the nanoscale grain sizes 共their coarse-grained counterparts may never
get to experience the high twinning stress兲. The latter also
favors partial dislocation mediated processes in general:1,12,13
as the grain dimension decreases, the area associated with the
stacking fault energy goes down faster than the length associated with the dislocation line tension. Exactly when twinning sets in depends on the local stress concentrations and
crystallographic conditions.11,18 It is not clear at present if the
stress state had significant effects on the activation of twinning.
Among the several possible mechanisms that have been
identified, our survey of about 50 grains containing deformation twins led us to conclude that heterogeneous formation
through GB emission of partial dislocations is the most active mechanism responsible for twinning in nc-Ni grains. Apparently, at highly elevated stresses the high density of GBs
provides plentiful sources for the twinning dislocations.
The authors thank S. Cheng for preparing cold-rolled
nc-Ni samples. X.-L.W. was supported by National Natural
Science Foundation of China 共50471086, 10472117,
50021101兲,
National
973
Program
of
China
共2004CB619305兲, and E.M. by U.S. NSF-DMR-0355395.
1
M. W. Chen, E. Ma, K. J. Hemker, H. W. Sheng, Y. M. Wang, and X.
Cheng, Science 300, 1275 共2003兲.
2
X. Z. Liao, F. Zhou, E. J. Lavernia, D. W. He, and Y. T. Zhu, Appl. Phys.
Lett. 83, 632 共2003兲.
3
X. Z. Liao, F. Zhou, E. J. Lavernia, D. W. He, and Y. T. Zhu, Appl. Phys.
Lett. 83, 5062 共2003兲.
4
X. Z. Liao, Y. H. Zhao, S. G. Srinivasan, Y. T. Zhu, R. Z. Valiev, and D.
V. Gunderov, Appl. Phys. Lett. 84, 592 共2004兲.
5
H. Rosner, J. Markmann, and J. Weissmuller, Philos. Mag. Lett. 84, 321
共2004兲.
6
J. Markmann, P. Bunzel, H. Rösner, K. W. Liu, K. A. Padmanabhan, R.
Birringer, H. Gleiter, and J. Weissmüller, Scr. Mater. 49, 637 共2003兲.
7
Y. M. Wang, A. M. Hodge, J. Biener, A. V. Hamza, D. E. Barnes, K. Liu,
and T. H. Nieh, Appl. Phys. Lett. 86, 101915 共2005兲.
8
Z. Budrovic, H. Van Swygenhoven, P. M. Derlet, S. Van Petegem, and B.
Schmitt, Science 304, 273 共2004兲.
9
V. Yamakov, D. Wolf, S. R. Phillpot, A. K. Mukherjee, and H. Gleiter,
Nat. Mater. 3, 43 共2004兲.
10
V. Yamakov, D. Wolf, S. R. Phillpot, and H. Gleiter, Acta Mater. 50, 5005
共2002兲.
11
H. Van Swygenhoven, P. M. Derlet, and A. G. Prøseth, Nat. Mater. 3, 399
共2004兲.
12
R. J. Asaro, P. Krysl, and B. Kad, Philos. Mag. Lett. 83, 733 共2003兲.
13
R. J. Asaro and S. Suresh, Acta Mater 53, 3369 共2005兲.
14
F. Dalla Forre, H. Van Swygenhoven, and M. Victoria, Acta Mater. 50,
3957 共2002兲.
15
Y. M. Wang and E. Ma, Appl. Phys. Lett. 85, 2750 共2004兲.
16
Y. M. Wang, S. Cheng, Q. M. Wei, E. Ma, T. G. Nieh, and A. Hamza, Scr.
Mater. 51, 1023 共2004兲.
17
Y. M. Wang, A. Hamza, and E. Ma, Appl. Phys. Lett. 86, 241917 共2005兲.
18
Y. M. Wang, M. W. Chen, F. H. Zhou, and E. Ma, Nature 共London兲 419,
912 共2002兲; Y. M. Wang and E. Ma, Acta Mater. 52, 1699 共2004兲.
Downloaded 23 Mar 2006 to 159.226.230.75. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp