Acta Materialia 55 (2007) 3749–3756

www.elsevier.com/locate/actamat

What is behind the inverse Hall–Petch effect

in nanocrystalline materials?
C.E. Carlton, P.J. Ferreira *

Materials Science and Engineering Program, University of Texas at Austin, Austin, TX 78712, USA

Received 11 January 2007; accepted 18 February 2007

Available online 8 April 2007

Abstract

An inverse Hall–Petch effect has been observed for nanocrystalline materials by a large number of researchers. This effect implies that
nanocrystalline materials get softer as grain size is reduced below a critical value. Postulated explanations for this behavior include dis-
location-based models, diffusion-based models, grain-boundary-shearing models and two-phase-based models. In this paper, we report
an explanation for the inverse Hall–Petch effect based on the statistical absorption of dislocations by grain boundaries, showing that the
yield strength is dependent on strain rate and temperature and deviates from the Hall–Petch relationship below a critical grain size.
! 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Nanocrystalline materials; Dislocations; Grain boundaries; Theory; Plastic deformation

1. Introduction ry ¼ r0 þ kd #1=2 ð1Þ

where r0 is the friction stress in the absence of grain bound-
Nanocrystalline materials show promise for applications
aries, k is a constant and d is the grain size. In other words,
in a wide range of fields, particularly because of their high
the yield stress increases as grain size decreases because
strength-to-weight ratios when compared with their coarse-
pile-ups in fine-grained materials contain fewer disloca-
grained counterparts. However, the high strength expected
tions, the stress at the tip of the pile-up decreases and, thus,
for nanocrystalline materials has been questioned by sev-
a larger applied stress is required to generate dislocations in
eral investigations, which claimed a decrease in the yield
the adjacent grain. In very small grains, this mechanism
stress ry below a critical grain size [1–11]. A great deal of
will break down because grains are unable to support dis-
effort has been expended studying and attempting to
location pile-ups. Typically, this is expected to occur for
explain this effect, but controversy still persists [12–34].
grain sizes below 10 nm for most metals [12,37]. As a
However, as evidence of this unique behavior emerges, a
consequence, a threshold value is expected at which a max-
crucial question arises: As the structural scale reduces from
imum yield stress can be achieved. However, experimen-
the micro to the nanometer range, how does the yield stress
tally, several systems such as Cu and Ni have revealed
change with grain size and what are the operating
that the Hall–Petch slope (k value in Eq. (1)) is reduced
mechanisms?
or becomes negative below a certain grain size [1,2,4–10].
Classically, one would expect an increase in ry for smal-
This is known as the inverse Hall–Petch effect (IHPE). Fur-
ler grain sizes according to the Hall–Petch equation [35,36]
thermore, the transition in the Hall–Petch slope normally
given by
occurs for grain sizes above 10 nm, where dislocation
pile-ups are still possible.
*
Corresponding author. Tel.: +1 512 471 3244; fax: +1 512 471 7681. So far, most of the models proposed to explain the
E-mail address: [email protected] (P.J. Ferreira). IHPE behavior fall into at least one of four categories,

1359-6454/$30.00 ! 2007 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2007.02.021
3750 C.E. Carlton, P.J. Ferreira / Acta Materialia 55 (2007) 3749–3756

namely dislocation-based models [13–17], diffusion-based dence related with the IHPE has indicated that many of
models [18,19], grain-boundary-shearing models [20–23] the proposed models may be inadequate for explaining
and two-phase-based models [24–29]. the IHPE. Among of the most important experimental
The dislocation-based models assume that dislocation findings in this regard are several in situ transmission elec-
motion is the primary agent for plastic flow in nanocrystal- tron microscopy (TEM) observations of dislocation
line materials and explain the IHPE by taking into account motion during tensile deformation and nanoindentation
the ways that very small crystallite sizes can affect the of nanocrystalline materials [38–45]. This is relevant
behavior of dislocations. For example, one model considers because many of the models listed above are based on
how the energy of a dislocation may be reduced by a very the assumption that dislocation motion does not play a sig-
small crystal and how the change in the dislocation’s nificant role in the deformation of nanocrystalline materi-
energy could affect its motion and lead to the IHPE [13]. als. In addition, many of the same TEM studies have
Another dislocation-based model considers how a nano- also shown that plastically deformed nanocrystalline mate-
crystalline grain could affect the operation of Frank–Reed rials contain a very low accumulated dislocation density.
dislocation sources [14]. Additionally, molecular dynamics These observations are contrary to models that rely on
simulations of some metals indicate that deformation of strain hardening as the means to understand the Hall–
nanocrystalline materials can occur via a dislocation mech- Petch relationship [13,15].
anism [16]. Finally, a recent experiment, where the strain rate sensi-
The diffusion-based models assume various diffusion tivity of nanocrystalline copper was determined by nanoin-
processes to explain the IHPE. In one model, the IHPE dentation, revealed that nanocrystalline materials should
is attributed to the effect of Coble creep in strain rate be more strain-rate sensitive than their microcrystalline
regimes and temperature ranges where it is not normally counterparts, but not as sensitive as Coble creep models
seen [18]. Some models use diffusion processes in combina- would suggest [46].
tion with other mechanisms to explain the IHPE. For In the context of this lively and continuing discussion,
example, one model explains the IHPE using a combina- the purpose of this paper is thus to suggest a model for
tion of dislocation motion and creep [15], while another the IHPE in nanocrystalline materials based on the nature
model explains the IHPE as being due to grain-boundary of dislocation absorption by grain boundaries. The paper
shear by a thermally activated grain-boundary-shearing will show that below a critical grain size, typically at the
mechanism [19]. nanoscale, dislocations are more easily absorbed by grain
A third set of models suggest that grain-boundary shear- boundaries. The model matches in situ TEM observations,
ing is the dominant cause for the IHPE behavior. Grain- which show that dislocations remain relatively active in
boundary shearing has been observed in various molecular nanocrystalline materials [38–45], while the dislocation
dynamics simulations [20,22,23]. In these simulations, density remains considerably low.
grain-boundary shearing begins to dominate dislocation
motion as the primary means of deformation at very low 2. A model for the inverse Hall–Petch behavior
grain sizes. There are also analytical models that use
grain-boundary shearing to explain the IHPE. In one of Let us start by considering that dislocations are line
these models, grain boundaries and crystalline grains are defects with length l. As we will see later, from an atomistic
considered to be separate phases, much like the two-phase point of view, this length becomes increasingly important
models mentioned below. In this model the grain boundary as the material’s grain size reaches the nanoscale. Next,
is assumed to deform by an athermal grain-boundary-shear we shall assume that the nature of grain boundaries is
mechanism while the grain interior behaves normally [21]. not scale dependent. In fact, several investigations have
Finally, another set of models explain the IHPE by con- revealed that the structure and width of grain boundaries
sidering a nanocrystalline material to be a composite mate- is essentially the same for nanocrystalline materials and
rial with at least two phases, namely a grain-boundary coarser-grained materials [47–49]. Finally, we shall con-
phase and a grain-interior phase [23–28]. In one of these sider that any grain boundary exhibits a specific probability
models, the grain boundary is considered a continuous to absorb a dislocation, on an atom by atom basis, as the
material that is strengthened by nanocrystalline grains in dislocation approaches the grain boundary. However,
much the same way that conventional materials can be instead of considering the particular crystallography of a
strengthened by precipitates [27]. In another model, the grain boundary, the model proposed herein acknowledges
grain-boundary material and the lattice material are the existence of distinct grain-boundary structures by
assumed to have different yield strengths [26]. A third assuming that each grain boundary exhibits a specific acti-
model assumes quite special mechanical properties for the vation energy for dislocation absorption. This assumption
grain boundary [28]. will be discussed later in the paper in greater detail.
Considering the aforementioned volume and good qual- Assuming the probability of a dislocation being
ity of the work completed on explaining the IHPE in met- absorbed by the grain boundary to be Pdis and the proba-
als, it is necessary to justify the publication of a new model. bility of an individual atom on the dislocation core making
This motivation arises because recent experimental evi- the jump to be Patom, a relatively simple analysis using a
 C.E. Carlton, P.J. Ferreira / Acta Materialia 55 (2007) 3749–3756 3751

Bernoulli distribution function can made. The Bernoulli boundaries is fast enough to be relevant at reasonable
function is a discrete distribution having two possible out- strain rates. Assuming that the strain in the material is
comes, labeled by n = 0 and n = 1, in which n = 1 (‘‘suc- accommodated by the motion of dislocations, it is possible
cess’’) occurs with probability p and n = 0 (‘‘failure’’) to determine the average time !t during which an individual
occurs with probability (1 # p), where 0 < p < 1. Thus, p dislocation is able to traverse the entire grain. As the strain
is the probability of an atom successfully jumping into rate associated with the motion of one single dislocation is
the grain boundary in a single attempt, whereas (1 # p) is given by
the probability of an atom failing to make the jump in a % &'
!x b
single attempt. With this, we can write for Pdis, the follow- e_ ¼ t ð5Þ
d d
ing equation:
N J where !x is the average distance moved by the dislocation
P dis ¼ P Jatom ¼ ½1 # ð1 # pÞ ' ð2Þ
during the time t, then the average time !t taken by each dis-
where J is the total number of atoms on the dislocation location to cross the entire grain (!x ¼ dÞ is given by
core jumping into the grain boundary. A value for J !t ¼ b=ð_edÞ. Hence, the N factor can be rewritten as
can be determined by multiplying the atomic linear den- N ¼ ðbC / mÞ=_ed ð6Þ
sity f of the dislocation core by the length l of the dislo-
cation line. As the length of the dislocation line is where the symbols have the same meaning as before. As a
proportional to the grain size d, the variable J is thus re- result, Eq. (4) shows that the probability of dislocation
lated to the grain size. N is the number of attempted absorption by grain boundaries is strongly dependent on
jumps by dislocation core atoms to the grain boundary the grain size d and the strain rate e_ . As the absorption
during an average time !t between absorption of disloca- of dislocations by grain boundaries will obviously interfere
tions. N is defined by the product of the Debye frequency with grain-boundary strengthening, Eq. (4) predicts that
m, the time !t and the Schmid factor Cu. Thus, the (1 # p)N the greater the grain-boundary absorption that can occur,
term in Eq. (2) represents the chance for one atom failing the more deviation from the Hall–Petch behavior (Eq.
to make the jump after N attempts and the term (1)) should be expected.
[1 # (1 # p)N] describes the chance of one or more jump We are left with the task of establishing a correlation
attempts succeeding. The probability p is thus a modified between the probability of dislocations being absorbed
Boltzmann factor that represents the chance of an atom’s by grain boundaries and the classical Hall–Petch equa-
single jump attempt succeeding, when assisted by a shear tion. To start, consider that the leading dislocation, in a
stress s. Hence, p is given by pile-up of n dislocations, experiences a force due to (i)
# ! " $ the resolved shear stress, s, (ii) the other (n # 1) disloca-
p ¼ exp # DG þ sb3 =k B T ¼ expðMÞ ð3Þ tions and (iii) a backward force due to the internal stress
where DG is the activation energy for atomic migration, b is s0 produced by the grain boundary. According to the clas-
the Burgers vector, kB is Boltzmann’s constant and T is sical Hall–Petch equation, the internal stress s0 is consid-
temperature. M is thus a mobility factor. As we are describ- ered from the point of view of a rigid grain boundary,
ing the jump of atoms from the dislocation core to the where dislocations pile up. This is the case where Pdis = 0
grain boundary, DG is strongly dependent on the specific in Eq. (4). But, what happens when Pdis 6¼ 0? In this case,
structure of the grain boundary, thereby implicitly taking the grain boundary is no longer a rigid barrier and thus
into account the crystallographic nature of the grain the internal stress s0 should be reduced to a term
boundary. Therefore Eq. (2) can be rewritten as s0(1 # Pdis). Assuming that the leading dislocation moves
N J
forward towards the grain boundary by a distance dx, the
P dis ¼ f1 # ½1 # eM ' g ð4Þ work done by the resolved shear stress can be described
From Eq. (4), immediate physical insight is possible. Be- by W1 = nbsdx. Therefore, the motion of the leading
cause the term in the curly brackets is always less than 1, dislocation results in an increased interaction with the
higher values of J will lower the probability Pdis of a dislo- grain boundary, which can be described by W2 =
cation being absorbed by the grain boundary. Hence, the bs0(1 # Pdis)dx. At equilibrium, W1 = W2 and, hence,
larger the grain size d, the larger the value of J will be, the resolved shear stress acting on the leading dislocation
and the lower will be the probability Pdis. On the other of a pile-up composed of n dislocations equals the back-
hand, as N is increased, the probability of all jump at- ward force due to the internal stress s0 produced by the
tempts failing decreases and thus the probability Pdis is in- grain boundary. This can be expressed by
creased. The reason for this behavior is due to the fact that ns
N is the number of attempted jumps by dislocation core s1 ¼ s0 ¼ ð7Þ
ð1 # P dis Þ
atoms to the grain boundary during an average time !t be-
tween absorption of dislocations. Hence, as the time !t is in- where the symbols have the same meaning as before.
creased, N increases and thus the probability Pdis of Assuming n = ds/lb [35,36], where d is the grain size, s is
dislocation absorption increases. In other words, this is a the resolved shear stress, l is the shear modulus and b is
question of whether the absorption of dislocations by grain the Burgers vector, Eq. (7) can be rewritten as
3752 C.E. Carlton, P.J. Ferreira / Acta Materialia 55 (2007) 3749–3756

s1 lbð1 # P dis Þ In addition, the following values were considered, namely

s2 ¼ ð8Þ
d b = a/2[1 1 0] = 0.258 nm, where a = 0.365 nm is the lattice
Considering the Schmid factor Cu = cos h cos u, then parameter for Ni, m = 1013 Hz and f = 6.54 atoms/nm for
s = r cos h cos u and s1 = r1 cos h cos u, where r is the ap- the linear atomic density of the dislocation line in Ni. Fi-
plied stress, r1 is the critical stress at the leading dislocation nally, DG in Eq. (3) was assumed to be the activation en-
to activate a dislocation source on an adjacent grain, h is ergy for grain-boundary self-diffusion. This is a
the angle between the tensile axis and the normal to the slip reasonable assumption as we are considering the absorp-
plane and u is the angle between the tensile axis and the tion of atoms from the dislocation core into a grain bound-
slip direction, Eq. (8) assumes the form ary. For fcc metals, DG has been reported to vary within
( )12 ( )1 the range 0.6–1.5 eV [51] and thus a variety of activation
r1 lb 1 # P dis 2 energies were considered. One other reason to consider a
r¼ ð9Þ
cos h cos u d variation in the activation energies is that grain boundaries
h i12 with specific crystallographic structures will exhibit distinct
Assuming k ¼ cosrh1 lb cos u
, Eq. (9) becomes abilities to absorb dislocations. Finally, assuming
( )12 k = 0.158 MPa m1/2 for the case of Ni [52], the yield stress
1 # P dis
r¼k ð10Þ ry in Eq. (11) can be plotted as a function of grain size for
d
various activation energies (Fig. 1).
which represents the yield strength of a material as a func- As shown in Fig. 1, for specific activation energies, there
tion of grain size and dislocation absorption by grain is a sharp transition at a critical grain size at which the
boundaries, when the lattice friction stress r0 = 0. In the yield stress is suddenly inverted. The critical yield stress
presence of a lattice friction stress, the modified Hall–Petch (CYS) at which the transition occurs represents the onset
relation can be expressed as of the inverse Hall–Petch behavior. At the highest activa-
( )1 tion energy examined (1 eV), the inverse Hall–Petch behav-
1 # P dis 2
ry ¼ r0 þ k ð11Þ ior would not be expected for Ni, even at unrealistically
d small grain sizes. On the other hand, for sufficiently small
This equation describes the yield stress of a material as a activation energies (0.75 eV), dislocation absorption in Ni
function of grain size, taking into account the probability would be expected to dominate.
of grain-boundary absorption. Eq. (11) reverts to the clas- Eq. (11) was also evaluated for different strain rates
sical Hall–Petch equation when the boundary is rigid and (Fig. 2a). Higher strain rates result in less deviation from
Pdis = 0. The essential effect of grain-boundary dislocation the classical Hall–Petch relation, which results in higher
absorption is to reduce the number of dislocations, n, in CYS. Thus, higher strain rates and higher activation ener-
the pile-up, which decreases the stress at the leading dislo- gies have a similar effect on the Hall–Petch behavior.
cation. If all moving dislocations are absorbed by the Finally, calculations of Eq. (11) for different temperatures
existing grain boundaries (Pdis = 1), then Eq. (11) is re- (Fig. 2b), predict that the CYS will be lower for higher tem-
duced to ry = r0, i.e., grain boundaries do not play any peratures and quite sensitive to temperature changes. Thus,
role in strengthening the material. In addition, as the term
Pdis is negative in Eq. (11), reducing the Hall–Petch coef-
ficient, k, is simply an inversion of the Hall–Petch
relation.

3. Results and discussion

Eq. (11) can be evaluated for any material at a given

temperature and strain rate. For illustration purposes, con-
sider the case of Ni deformed at room temperature under a
strain rate of 10#5 s#1. To determine Pdis, the resolved
shear stress, s, acting on the dislocations was assumed to
be s ( r0Cu and r0 = 21.8 MPa [50]. For the Schmid factor
Cu, we assumed (i) a random orientation of grains with
respect to the applied stress axis and (ii) a maximum stress
direction on the slip plane. For an fcc material, the maxi-
mum angle possible between {1 1 1} slip planes is 109.48",
thus giving the average value
Z h¼109=2 Fig. 1. The modified Hall–Petch relationship (Eq. (11)) for Ni deformed
C/ ( cosðhÞ sinðhÞ ¼ 0:33 under a strain rate of 10#5 s#1 at 293 K, showing the inverse Hall–Petch
h¼0 effect at different critical grain sizes for various levels of activation energy.
 C.E. Carlton, P.J. Ferreira / Acta Materialia 55 (2007) 3749–3756 3753

The third important prediction is that the CYS is very

sensitive to temperature changes. As the temperature is
increased, the probability for dislocation absorption
increases, leading to stronger deviations from the Hall–
Petch relation. Although work on the effect of temperature
on the yield strength of nanocrystalline materials is sparse
[52,58], the model suggests that experiments of this nature
could help us understand the mechanisms of deformation
in nanocrystalline materials.
In order properly to consider and confirm the results of
the model, it is important to evaluate how the model fits the
experimental data available in the literature. In doing so, a
few issues arise. First, as tensile tests in nanocrystalline
materials are difficult to perform, strain rates are frequently
not recorded. This is problematic, as the strain rate is a
very important parameter in the model described herein,
while fitting the experimental data becomes very difficult
if it is unknown. Second, the model assumes discrete values
for the activation energy (grain-boundary self-diffusion)
and for the grain size of the material. Yet, in real polycrys-
talline materials, a spread in activation energies (dictated
by a distribution in grain-boundary structures) and a
spread in available grain sizes is expected. Quantifying
the effects of this spreading effect is not trivial and should
be examined in future work. Third, the published data lack
agreement due to the wide variety of experimental condi-
tions employed, namely the use of different nanocrystalline
materials, distinct methods of preparation (electrodepos-
ited, severed deformed, powder compacted) and the use
of either tensile tests under various strain rates and/or
hardness tests.
An example of the variation in the Hall–Petch results for
Fig. 2. (a) Yield stress vs. strain rate at a constant temperature of 293 K
Ni using both tensile and hardness data is shown in Fig. 3.
and (b) yield strength vs. temperature at a constant strain rate of 10#5, in Although several studies claim evidence that either sup-
Ni for various grain sizes and an activation energy of 0.85 eV. ports the classical [58–60] or the inverse Hall–Petch behav-
ior [1,6,9,10], when the data for nanocrystalline Ni are
viewed as a whole it becomes unclear whether it exhibits
changes in temperature will have the opposite effect on the
CYS to changes in activation energy and strain rate.
These predictions are theoretically and experimentally
important for several reasons. First, the activation energy,
which is typically proportional to the melting temperature,
is expected to be different for distinct materials. In fact, cal-
culations of Eq. (1) for Al, Cu and Ni reveal that a devia-
tion from the Hall–Petch equation occurs earlier (coarser
grain sizes) for Al, followed by Cu and Ni. Moreover, since
the activation energy is correlated with grain-boundary
structure, knowledge of the grain-boundary distribution
becomes important to full understanding of the Hall–Petch
behavior.
The model’s second prediction is that the CYS depends
significantly on strain rate, which confirms previous
research [52–57]. This result is not surprising. As the strain
rate is increased, more dislocations impinge on the grain
boundaries per unit time. Therefore, less time is available
for dislocations to be absorbed, leading to higher CYS at Fig. 3. Hall–Petch relationship for Ni determined by several research
faster strain rates. works. (T) denotes tensile tests and (H) denotes hardness tests.
3754 C.E. Carlton, P.J. Ferreira / Acta Materialia 55 (2007) 3749–3756

an inverse Hall–Petch behavior. In fact, a linear regression There are several possible explanations for this. It is pos-
performed on all the data, except for the data reported by sible that the samples indicating an IHPE were flawed in
Wang et al. [9], gives a correlation factor R2 = 0.93. For some way, for instance incomplete densification or the pres-
some of the reported data [58,59] a linear regression fit ence of an additional amorphous phase. It is also possible
gives correlation factors of R2 = 0.98 and R2 = 0.84, that, due to differences in the microstructure, some of the
respectively. materials exhibited IHPE behavior while others did not.
According to our model, a lack of inverse Hall–Petch The theory presented above can rationalize this anomaly if
behavior in Ni would not be surprising. In fact, due to the samples can be shown as having grain boundaries with
the high melting temperature of Ni and consequent high substantially different activation energies of absorption.
activation energy, we could expect it effectively to halt Overall, the data obtained from individual research
grain-boundary absorption of dislocations at reasonable groups and shown in Fig. 4 seem to support the IHPE.
strain rates. However, as mentioned above, polycrystalline However, when the data from single research groups are
materials exhibit a distribution of grain sizes and grain- combined and viewed as a whole, the result is still
boundary structures, which may lead to some form of devi- ambiguous.
ation from the classical Hall–Petch relationship. In the future, to overcome this ambiguity, it will become
The model also suggests that the variation of results, important to report the strain rate and the temperature
such as those in Fig. 3, may differ due to differences in used in the experiments, as well as to characterize the grain
experimental conditions. If the strain rates used in the size distribution and grain-boundary structure.
experiments differed significantly, or different methods for
producing nanocrystalline materials lead to dissimilar 4. Conclusions
grain-boundary distributions and grain size distributions,
this in turn could result in distinct abilities for absorbing The model presented herein uses a simple argument to
dislocations. produce a simple equation that can be used to make some
In the context of this discussion, it is valuable to con- interesting and useful predictions about the effects of grain-
sider the expected behavior of nanocrystalline copper, boundary strengthening in nanocrystalline materials. Sim-
which has been widely investigated in the open literature ply, the model assumes a probability for atoms on a dislo-
[2,4,7,8,46,61–73], According to the model outlined above, cation core to be absorbed by the grain boundary. The
copper would be more likely to exhibit an inverse Hall– larger the grain size, the larger must be the number of
Petch effect and/or exhibit a lower CYS than nickel. This atoms absorbed by the grain boundary, if the dislocation
is because the activation energy for dislocation absorption is also to be absorbed. Thus, for larger grain sizes, the
by grain boundaries for copper is lower than that of nickel. probability for dislocation absorption by the grain bound-
However, upon examination of the experimental results ary is lowered. The predictions provided by the model can
obtained for nanocrystalline copper (Fig. 4), the results be tested in the laboratory, keeping in mind that the strain
reported are even more difficult to interpret than the results rate, the temperature, grain-boundary distribution and
for nickel. As shown in Fig. 4, the two sets of data reported grain size distribution are all important parameters to be
by Sanders et al. [8] appear to support the IHPE for a crit- considered. The results of future experiments dealing with
ical grain size of )16 nm. On the other hand, the remainder these effects will help confirm this model and lead to an
of the data combined are highly inconclusive. increase in our understanding of the mechanical properties
of nanomaterials.

Acknowledgements

The authors would like to thank Professors Desiderio

Kovar and Lew Rabenberg, from the University of Texas
at Austin, for valuable discussions.

References

[1] Chokski AH, Rosen A, Karch J, Gleiter H. On the validity of the

Hall–Petch relationship in nanocrystalline materials’. Scr Metall
1989;23:1679.
[2] Nieman GW, Weertman JR, Siegel RW. Microhardness of nano-
crystalline palladium and copper produced by inert-gas condensation.
Scr Metall 1989;23:2013.
Fig. 4. Hall–Petch relationship for Cu determined by several research [3] Lu K, Wei WD, Wang JT. Microhardness and fracture properties of
works. (D) denotes disk bending tests, (T) denotes tensile tests, (H) nanocrystalline Ni–P alloy. Scr Metall Mater 1990;24:2319.
denotes hardness tests and (C) denotes compressive tests. See above- [4] Nieman GW, Weertman JR, Siegel RW. Mechanical behavior of
mentioned references for further information. nanocrystalline Cu and Pd. J Mater Res 1991;6:1012.
 C.E. Carlton, P.J. Ferreira / Acta Materialia 55 (2007) 3749–3756 3755

[5] Fougere GE, Weertman JR, Siegel RW, Kim S. Grain-size dependent [32] Zhu YT, Langdon TG. Influence of grain size on deformation
hardening and softening of nanocrystalline Cu and Pd. Scr Metall mechanisms: an extension to nanocrystalline materials. Mater Sci Eng
Mater 1992;26:1879. A 2005;409:234.
[6] El-Sherik AM, Erb U, Palumbo G, Aust KT. Deviations from Hall– [33] Wolf D, Yamakov V, Phillpot SR, Mukherjee AK, Gleiter H.
Petch behavior in as-prepared nanocrystalline nickel. Scr Mater Deformation of nanocrystalline materials by molecular dynamics
1992;27:1185. simulation: relationship to experiments. Acta Mater 2005;53:1.
[7] Gertsman VY, Hoffmann M, Gleiter H, Dirringer R. Grain size [34] Meyers MA, Mishra A, Benson DJ. Mechanical properties of
dependence of yield of copper. Acta Mater 1994;42:3539. nanocrystalline materials. Progr Mater Sci 2006;51:427.
[8] Sanders PG, Eastman JA, Weertman JR. Elastic and tensile behavior [35] Hall EO. The deformation and aging of mild steel. Proc Phys Soc
of nanocrystalline copper and palladium. Acta Mater 1997;10:4019. London B 1951;64:747.
[9] Wang N, Wang Z, Aust KT, Erb U. Room temperature creep [36] Petch NJ. The cleavage strength of polycrystals. J Iron Steel Inst
behavior of nanocrystalline nickel produced by an electrodeposition 1953;25:174.
technique. Mater Sci Eng A 1997;237:150. [37] Eckert J, Holzer JC, Krill CE, Johnson WL. Structural and
[10] Schuh CA, Nieh TG, Yamasaki T. Hall–Petch breakdown manifested thermodynamic properties of nanocrystalline FCC metals prepared
in abrasive wear resistance of nanocrystalline nickel. Scr Mater by mechanical attrition. J Mater Res 1992;7:1751.
2002;46:735. [38] Milligan WW, Hackne SA, Ke M, Aifantis EC. In situ studies of
[11] Giga A, Kimoto Y, Takigawa Y, Higashi K. Demonstration of an deformation and fracture in nanophase materials. Nanostruct Mater
inverse Hall–Petch relationship in electrodeposited nanocrystalline 1993;2:267–76.
Ni–W alloys through tensile testing. Scr Mater 2006;55:143. [39] Youngdahl CJ, Weertman JR, Hugo RC, Kung HH. Deformation
[12] Nieh TG, Wadsworth J. Hall–Petch relation in nanocrystalline Solids. behaviour in nanocrystalline copper. Scr Mater 2001;44:475–1478.
Scr Metall Mater 1991;25:955. [40] Kumar KS, Suresh S, Chisholm MF, Horton JA, Wang P. Defor-
[13] Scattergood RO, Koch CC. A modified-model for Hall–Petch mation of electrodeposited nanocrystalline nickel. Acta Mater
behavior in nanocrystalline materials. Scr Metall Mater 1992;27: 2003;51:397–405.
1195. [41] Hugo RC, Kung HH, Weertman JR, Mitra R, Knapp JA, Follstaedt
[14] Lian B Buadelet, Nazarov A. Model for the prediction of the DM. In-situ TEM tensile testing of DC magnetron sputtered and
mechanical behavior of nanocrystalline materials. Mater Sci Eng A pulsed laser deposited Ni thin films. Acta Mater 2003;51:1937.
1993;172:23. [42] Shan Zhiwei, Stach EA, Wiezorek JMK, Knapp JA, Follstaedt DM,
[15] Malygin GA. Breakdown of the Hall–Petch law in micro- and Mao SX. Grain boundary mediated plasticity in nanocrystalline
nanocrystalline metals. Phys Solid State 1995;37:1248. nickel. Science 2004;305:654–7.
[16] Yamakov V, Wolf D, Phillpot SR, Muckherjee AK. Dislocation [43] Haque MA, Saif MTA. In situ tensile testing of nanoscale freestand-
processes in the deformation of nanocrystalline aluminium by ing thin films inside a transmission electron microscope. J Mater Res
molecular-dynamics simulation. Nat Mater 2002;1:1. 2005;20:1769–77.
[17] Schiotz J, Jacobsen K. A maximum strength of nanocrystalline [44] Hattar K, Han J, Saif MTA, Robertson IM. In situ transmission
copper. Science 2003;301:1357. electron microscopy observations of toughening mechanisms in ultra-
[18] Masamura RA, Hazzledine PM, Liaw PK, Lavernia EJ. Yield stress fine grained columnar aluminum thin films. J Mater Res
of fine grained materials. Acta Metall 1998;13:4527. 2005;20:1869–77.
[19] Conrad H, Narayan J. On the grain size softening in nanocrystalline [45] Ma E. Watching the nanograins roll. Science 2004;305:623.
materials. Scr Mater 2000;42:1025. [46] Chen J, Lu L, Lu K. Hardness and strain rate sensitivity of
[20] Van Swygenhoven H, Spaczer M, Caro A. Microscopic description of nanocrystalline Cu. Scr Mater 2006;54:1913.
plasticity in computer generated metallic nanophase samples: a [47] Siegel RW, Thomas GJ. Grain boundaries in nanophase materials.
comparison between Cu and Ni. Acta Mater 1999;47:3117. Ultramicroscopy 1992;46:376.
[21] Takeuchi S. The mechanism of the inverse Hall–Petch relation in [48] Van Swygenhoven H, Farkas D, Caro A. Grain boundary structures
nanocrystals. Scr Mater 2001;44:1483. in polycrystalline metals at the nanoscale. Phys Rev B 2000;62:831.
[22] Yamakov V, Wolf D, Salazar M, Phillpot SR, Gleiter H. Deforma- [49] Kumar KS, Suresh S, Chisholm MF, Horton JA, Wang P. Defor-
tion mechanism crossover and mechanical behavior in nanocrystalline mation of electrodeposited nanocrystalline nickel. Acta Mater
materials. Philos Mag Lett 2003;83:385. 2003;51:387.
[23] Yamakov V, Wolf D, Phillpot SR, Mukherjee AK, Gleiter H. [50] Thompson AW. Yielding in nickel as a function of grain or cell size.
Deformation-mechanism map for nanocrystalline metals by molecu- Acta Mater 1975;23:1337.
lar dynamics simulation. Nat Mater 2004;3:43. [51] Mehrer H, editor. Diffusion in solid metals and alloys. Series:
[24] Lu K, Sui ML. An explanation of the Hall–Petch law in micro and Landolt-Bornstien: Numerical data and functional relationships in
nanocrystalline materials. Scr Mater 1993;28:1465. science and technology, Group 3, Condensed matter, vol. 26. Ber-
[25] Wang Ning, Wang Zhirui, Aust T, Erb U. Effect of grain size on the lin: Springer Verlag; 1990.
mechanical properties of nanocrystalline materials. Acta Metall [52] Gray GT, Lowe TC, Cady CM, Valiev RZ, Aleksandrov IV.
Mater 1995;43:519–23. Influence of strain rate and temperature on the mechanical response
[26] Konstantinidis DA, Aifantis EC. On the anomalous hardness of of ultra-fined grained Cu, Ni and Al–4Cu–0.5Zr. Nanostruct Mater
nanocrystalline materials. Nanostruct Mater 1998;10:1111. 1997;9:477.
[27] Song HW, Guo SR, Hu ZQ. A coherent polycrystal model for the [53] Lu L, Li S, Lu K. An abnormal strain rate effect on tensile behavior in
inverse Hall–Petch relation in nanocrystalline materials. Nanostruct nanocrystalline copper. Scr Mater 2001;45:1163.
Mater 1999;11:203. [54] Dalla Torre F, Van Swygenhoven H, Victoria M. Nanocrystalline
[28] Fan GJ, Choo H, Liaw PK, Lavernia EJ. A model for the inverse electrodeposited Ni: microstructure and tensile properties. Acta
Hall–Petch relation of nanocrystalline materials. Mater Sci Eng A Mater 2002;50:3957.
2005;409:248. [55] Schwaiger R, Moser B, Dao M, Chollacoop N, Suresh S. Some
[29] Chattopadhyay PP, Pabi SK, Manna I. On the inverse Hall–Petch critical experiments on the strain-rate sensitivity of nanocrystalline
relationship in nanocrystalline materials. Z Metall 2000;91:1049. nickel. Acta Mater 2003;51:5159.
[30] Koch CC, Narayan J. The inverse Hall–Petch effect – fact or artifact. [56] Wei Q, Cheng S, Ramesh KT, Ma E. Effect of nanocrystalline
Mater Res Soc Symp 2001;634:B5.1.1. and ultrafine grain sizes on the strain rate sensitivity and
[31] Kumar KS, Van Swygenhoven H, Suresh S. Mechanical behavior of activation volume: FCC versus BCC metals. Mater Sci Eng A
nanocrystalline metals and alloys. Acta Mater 2003;51:5743. 2004;318:71.
3756 C.E. Carlton, P.J. Ferreira / Acta Materialia 55 (2007) 3749–3756

[57] May J, Hoppel HW, Goken M. Strain rate sensitivity of ultrafine- [66] Suryanarayanan IR, Frey CA, Sastry SML, Waller BE, Buhro WE.
grained aluminum processed by severe plastic deformation. Scr Mater Mater Sci Eng A 1999;264:210.
2005;53:189. [67] Lu L, Sui ML, Lu K. Superplastic extensibility of nanocrystalline
[58] Ebrahimi F, Bourne GR, Kelly MS, Matthews TE. Mechanical copper at room temperature. Science 2000;287:1463.
properties of nanocrystalline nickel produced by electrodeposition. [68] Legros M, Elliot BR, Rittner MN, Weertman JR, Hemker KJ.
Nanostruct Mater 1999;11:343. Microsample tensile testing of nanocrystalline metals. Philos Mag A
[59] Xiao C, Mirshams RA, Whang SH, Yin WM. Tensile behavior and 2000;80:1017.
fracture in nickel and carbon doped nanocrystalline nickel. Mater Sci [69] Huang HB, Spaepen F. Tensile testing of free-standing Cu, Ag and Al
Eng A 2001;301:35. thin films and Ag/Cu multilayers. Acta Mater 2000;48:3261.
[60] Hughes GD, Smith SD, Pande CS, Johnson HR, Armstrong RW. [70] Yousseff KM, Scattergood R, Murty KL, Koch CC. Ultratough
Hall–Petch strengthening for the microhardness of twelve nanometer nanocrystalline copper with a narrow size distribution. Appl Phys
grain diameter electrodeposited nickel. Scr Metall 1986;20:93. Lett 2004;85:929.
[61] Merz MD, Dahlgren SD. Tensile-strength and work-hardening of [71] Hansen N. Hall–Petch relation and boundary strengthening. Scr
ultrafine-grained high-purity copper. J Appl Phys 1975;46:3235. Mater 2004;51:801.
[62] Hansen N, Ralph B. The strain and grain-size dependence of the flow- [72] Haouaoui M, Karaman I, Maier HJ, Hartwig KT. Microstructure
stress of copper. Acta Metall 1982;30:411. evolution and mechanical behavior of bulk copper obtained by
[63] Embury DJ, Lahaie JD. In: Natasi M et al., editors. Mechanical consolidation of micro- and nanopowders using equal-channel
properties and deformation behavior of materials having ultrafine angular extrusion. Metall Mater Trans A 2004;35:2935.
microstructure. Dordrecht: Kluwer; 1993. [73] Hansen N. Boundary strengthening in undeformed and deformed
[64] Valiev Z, Kozolv E, Ivanov YF, Lian J, Nazarov AA, Baudelet R. polycrystals. Mater Sci Eng A 2005;409:39.
Deformation behavior of ultra-fine-grained copper. Acta Mater [74] Lu L, Shen Y, Chen X, Qian L, Lu K. Ultrahigh strength and high
1994;42:2467. electrical conductivity in copper. Science 2004;304:422.
[65] Ebrahimi F, Zhai Q, Kong D. Deformation and fracture of [75] Wang YM, Wang K, Pan D, Lu K, Hemker KJ, Ma E. Microsample
electrodeposited copper. Scr Mater 1998;39:315. tensile testing of nanocrystalline copper. Scr Mater 2003;48:1581.