com
Abstract
The plastic deformation of micropillars is known to be aected by whether dislocations can escape easily from the material volume,
and the extent to which the dislocations mutually interact during the deformation. In this work, pre-straining and coating are used to
modify the initial dislocation content and the constraints on the escape of dislocations. Aluminum micropillars in the size range from 1
to 6 lm, with or without thin coating by tungsten deposition and pre-straining by 7%, were compressed using a at-punch nanoindenter to study their plasticity behavior. The results reveal very dierent behavior between the size regime of a few microns and that of
1 lm. For pillars a few microns large, coating leads to signicant strengthening, and pre-straining by 7% also produces a mild strengthening eect. The proof strength also exhibits good correlation with the square root of the residual dislocation density measured by transmission electron microscopy after deformation, indicating that strength in this size regime is controlled by dislocation interactions as in
traditional Taylor hardening. Coating evidently helps retain dislocations inside the pillar, and pre-straining increases the initial dislocation content; both eects lead to more severe strain hardening during deformation. For smaller pillars 1 lm in size, however, pre-straining results in softening, although coating still leads to strengthening, and the strength exhibits no correlation with the residual dislocation
density, which remains close to the initial value even with coating. These suggest dislocation starvation in this small size regime and that
strength is controlled by the availability of mobile dislocations. Coating cannot eectively trap dislocations inside the pillar, but can still
strengthen the pillar, presumably because dislocation nucleation is more dicult at the coated surface.
2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
Keywords: Plastic deformation; Compression tests; Dislocations; Nanoindentation
1. Introduction
The last decade or so has seen tremendous interest in the
yielding of micropillar forms of metals. In addition to their
power-law, size-dependent strength [111], these micropillars also deform in a jerky manner with continuous occurrence of strain bursts in a stochastic manner [1,2,1217]. To
explain these unusual behaviors, several theories have been
proposed. One important concept is dislocation starvation [13,15], which refers to the easy zipping of dislocations through the small crystal without accumulation and
multiplication, therefore resulting in the crystal staying in
Corresponding author.
1359-6454/$36.00 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.actamat.2012.07.048
6103
6104
Fig. 1. Normal stressnormal strain curves of 7% pre-strained aluminum pillars of diameter (a) 5.6 lm, (c) 3.3 lm and (e) 1.2 lm. Colored curves (online
version only) are for coated pillars with dierent volume fractions Vw of the tungsten coating; black curves are for uncoated pillars of similar sizes.
Representative SEM images of the deformed coated pillars with diameter of (b) 5.6 lm, (d) 3.3 lm and (f) 1.2 lm are also shown. (For interpretation of the
references to color in this gure legend, the reader is referred to the web version of this article.)
5.6 and 3.3 lm coated pillars, Fig. 1a,c indicate that the
deformation is smooth with continuous strain-hardening
up to 5% strain, but at larger strains the strain-hardening
ceases and a plateau follows. For the smaller 1.2 lm pillars in Fig. 1e, the softening at 5% strain is so rapid that it
corresponds to a huge and uncontrollable strain burst of
10%. The post-deformation SEM images of the coated
pillars shown in Fig. 1b, d and f reveal the occurrence of
only one or two intensive slip bands, around which the
tungsten coating shows signs of cracking and aking o
the pillar surface. As for the 1.2 lm pillars, which deformed
with a large avalanche, the post-deformation SEM morphology in Fig. 1f shows an intensive shear step corresponding to almost complete shear fracture of the pillar
into two segments. The large strain burst is thought to be
related to the cracking of the coating at a large stress
[22,24], and as this happens substantial slip of the pillar
core would proceed at the cracked region.
3.2. Strengthening of tungsten-coated specimens
The stress plotted in Fig. 1a, c and e is the nominal stress
r, which is the applied load divided by the gross cross-sectional area of the coated pillars including the coating thickness. To obtain the stress rAl applied onto the aluminum
core of the coated pillars, a composite model is used as
in Ref. [22]:
r rAl V Al rw V w ;
where VAl and Vw are the aluminum and tungsten deposition volume fractions, respectively, and rw is the stress sustained by the tungsten coating. The intrinsic strength of the
tungsten coating deposition fabricated by the present FIB
system was investigated previously [22], and its 2% proof
strength was found to be 100 MPa. Hence the 2% proof
strength rAl of the aluminum core can be calculated from
the overall nominal stress r using the above rule of mixtures. Such a rule-of-mixtures calculation assumes simple
load sharing between the constituent phases without mutual interactions, and the objective here is indeed to see
whether the pillar cores intrinsic strength can be explained
by such an eect, since if this is the case, the calculated rAl
would agree with the strengths of the uncoated pillars. The
results for dierent cases, including the uncoated pillars
with no pre-straining (i.e. the pristine group), are shown
in Fig. 2. For the coated pillars in both pre-straining conditions, their rAl data calculated from the rule of mixtures
are always higher than the strengths of the uncoated pillars
of the same size in accordance with previous ndings [22],
implying that the strength of the pillar core is aected by
some interaction eects from the coating. However, the
coated and pre-strained pillars (the red dotted line in
Fig. 2) are substantially weaker than the coated pillars with
no pre-straining (the black dotted line) at sizes of 1.2 lm:
the 2% proof strength of the coated/pre-strained group at
1.2 lm size is 172.6 24.1 MPa which is only about
54% of the coated group without pre-straining. At sizes
6105
6106
Fig. 3. (a and b) TEM montages of annealed and 7% cold reduction aluminum samples [36] prepared by the twin-jet method. (c) TEM image of a FIBprepared specimen from the annealed bulk taken near the [1 0 1] pole.
Fig. 4. (a and c) Montages of TEM images of the longitudinal sections of 5.6 lm uncoated pillars without (a) and with (c) 7% pre-strain. The top of the
pillar is towards the right side (the same orientation is used for all subsequent TEM images). (b and d) Higher-magnication TEM images of each
condition.
6107
Fig. 5. (a and c) Montages of TEM images of the longitudinal sections of 1.2 lm uncoated pillars without (a) and with (c) 7% pre-strain. (b and d)
Higher-magnication TEM images of each condition.
Fig. 6. (a and c) Montages of TEM images of the longitudinal sections of 5.6 lm tungsten coated pillars without (a) and with (c) 7% pre-strain. (b and d)
Higher-magnication TEM images of each condition.
6108
Fig. 7. (a and c) Montages of TEM images of the longitudinal sections of 1.2 lm tungsten coated pillars without (a) and with (c) 7% pre-strain. (b and d)
Higher-magnication TEM images of each condition.
Table 1
Details of deformed aluminum micropillars for TEM characterization.
Diameter
(lm)
Pre-strain
Tungsten volume
fraction
Max. stress
(MPa)
Final strain
5.6
5.6
1.2
1.2
5.6
5.7
1.1
1.2
None
7%
None
7%
None
7%
None
7%
None
None
None
None
12%
20%
10%
20%
58
65
240
209
172
227
333
295
0.05
0.11
0.07
0.10
0.06
0.19
0.04
0.04
Corresponding
gures
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
Fig.
4a and b
4c and d
5a and b
5c and d
6a and b
6c and d
7a and b
7c and d
for the case without pre-strain, or 1.0 1015 m2 with prestrain. On the contrary, for the 1.2 lm coated pillars, no
dislocation cell could be observed in the TEM montages
shown in Fig. 7a and c with or without pre-strain. In these
samples, dislocations are scarce and arrange in a random
manner, and their densities are estimated as 5.0 1013 m2
without pre-strain and 7.0 1013 m2 with pre-strain, i.e.
still very close to the background value of 1013 m2.
Fig. 8 summarizes the measured dislocation densities of different types of pillars after compression. It shows that in the
case of the 5.6 lm pillars, coating and pre-straining can
result in a signicant increase in the dislocation density during deformation. On the contrary, in the small 1.2 lm pillars, the dislocation content does not vary much with or
without coating or pre-straining, and stays at the low magnitude of 1013 m2, comparable to the initial state prior to
compression.
4. Discussion
The present results in Fig. 2 indicate that coating the
aluminum micropillars always leads to strengthening irrespective of the sizes of the pillars within the range studied,
and whether or not pre-straining was applied. However,
the eect of the 7% pre-straining is very dierent between
large and small pillars. For larger 5.6 lm pillars, prestraining results in a small strengthening eect in both
the coated or uncoated conditions, but for the small
1 lm pillars, pre-straining leads to softening in both the
coated and uncoated conditions, and the eect on the
coated condition is larger. The results in Fig. 8 also show
that at the small size of 1.2 lm, the dislocation content
remains close to the initial state in all groups, but this is
not so in the large pillar size of 5.6 lm. These indicate
that the dislocation mechanisms are very dierent between
the large and small pillars.
6109
Fig. 9. Strength vs. square root of residual dislocation density for the
5.6 lm pillars.
size, the strength data in Fig. 2 are in the following descending order: coated with pre-strain > coated without pre-strain > uncoated with pre-strain > uncoated without prestrain, and the dislocation density data in Fig. 8 follow
exactly the same order. Hence, for the 5.6 lm pillars,
strength evidently correlates with the residual dislocation
density. Fig. 9 plots the strength data vs. the square root
of the residual dislocation density data for the 5.6 lm pillars, and a proportionality relation between these two quantities is apparent. Hence, the natural conclusion for this size
regime is that strength is controlled by mutual dislocation
interactions as in traditional Taylor hardening, i.e.
p
r / q. This conclusion veries the key assumption of Taylor hardening in a number of theories on size eect of
strength [18,19,36], as well as the prediction of dislocation
dynamics simulations [21]. In this larger size regime, coating
produces strengthening because it traps dislocations inside
the pillar and their frequent interactions lead to multiplication, so that the density of mobile dislocations is maintained
at a high level [22]. Without coating, some dislocations can
annihilate at the pillars free surface, but since the pillar size
is not small, not all dislocations can travel freely to the free
surface without interaction with other dislocations. The
result is that, even without coating, the residual dislocation
density is still higher than the initial state with signicant
production. Pre-straining by 7% in this size regime evidently
results in a mild increase in the initial dislocation content
(cf. Fig. 3a and b), and indeed in Fig. 8, the residual dislocation density is higher with pre-straining in both the coated
and uncoated case. The higher initial dislocation content
from pre-straining leads to a small strengthening eect via
the Taylor mechanism as shown in Fig. 2, both for the
coated and uncoated case. In all, in this larger size regime,
the eects of coating and pre-straining on strength and
residual dislocation density can be understood as arising
from the Taylor hardening mechanism.
6110
density in Fig. 8 has not grown much from the initial low
value, but the strength data in Fig. 2 vary by a large extent
depending on whether the pillar is coated or pre-strained.
There is therefore no correlation between strength and
the residual dislocation density, i.e. the results here suggest
that strength in this regime is not controlled by the Taylor
mechanism. Another important phenomenon from Fig. 2 is
that pre-straining softens the pillars in both the coated and
uncoated groups at this small size, and this is in fact opposite to the Taylor eect where more existing dislocations
should produce strengthening. Therefore, the natural conclusion to draw for this size regime is that strength is controlled by the availability of mobile dislocations, in the
sense of the dislocation starvation concept as described
in Section 1, or simply that pertaining to the Orowan equation e_ qbm if the pillar is not completely depleted of dislocations. Pre-straining evidently results in more initial
dislocations so that strength is decreased at the same strain
rate. However, these mobile dislocations zip through the
small crystal easily, so that the residual dislocation content
stays at the same range. Strength is controlled by the mechanism(s) by which new dislocations are generated, presumably at the pillars surface.
The eects of coating are more dicult to comprehend
in this small size regime. Fig. 2 shows that coating still
strengthens the 1.2 lm pillars as in the larger pillars,
but with pre-straining the strengthening eect is less than
that without pre-straining. The persistently low residual
dislocation density even with coating means that the latter
is not eective in trapping the generated dislocations in this
small size regime. Two factors may explain this. First, the
applied stress in this size regime is now very high, and secondly, with the same volume ratio Vw of the coating it is
now very thin, compared to the larger pillars. Therefore,
the coating may fail easily when interacted upon by an
approaching dislocation, and may not be able to retain
the latter at the Alcoating interface. However, in Fig. 2,
the 1.2 lm coated pillars are always stronger than the
uncoated ones, and this is particularly so without prestraining. This seems to indicate that although the coating
cannot trap the generated dislocations, it can make the dislocation generation process itself more dicult. It is quite
possible that nucleation of dislocations at a coated surface
of the pillar is more dicult than at a free surface. The
present results therefore suggest that the strengthening
mechanism of coating is quite dierent in the small size
regime, as compared with dislocation trapping in the larger
size regime.
5. Conclusions
Eects of coating and pre-straining on deformation of
aluminum micropillars were investigated by compression
experiments using a nanoindenter with a at-ended diamond punch. The results indicate that in the larger size
regime of a few microns, the strength of the pillars correlates well with the square root of the residual density of
[31]
[32]
[33]
[34]
[35]
[36]
[37]
6111