Materials Letters 133 (2014) 151–154

Contents lists available at ScienceDirect

Materials Letters
journal homepage: www.elsevier.com/locate/matlet

Inverse Hall–Petch effect in quasi- and nanocrystalline materials

K.A. Padmanabhan a,n, Sriharsha Sripathi a, H. Hahn b,c, H. Gleiter b
a
School of Engineering Sciences and Technology & Centre for Nanotechnology, University of Hyderabad, Prof. C R Rao Road, Hyderabad 500 046, India
b
Karlsruhe Institute of Technology (North Campus), Institute of Nanotechnology, Hermann-von-Helmholtz-Platz 1,
76344 Eggenstein-Leopoldshafen, Germany
c
KIT-TUD Joint Research Laboratory Nanomaterials, TU Darmstadt, Jovanka-Bontschits-Strasse 2, 64287 Darmstadt, Germany

art ic l e i nf o a b s t r a c t

Article history: Inverse Hall–Petch effect/grain size softening in quasi- and nanocrystalline materials at ambient/low
Received 13 June 2014 temperatures is attributed to mesoscopic (
a grain diameter or more) grain/interphase boundary sliding
Accepted 25 June 2014 controlled flow. Equations for estimating the free energy of activation for the rate controlling process,
Available online 8 July 2014
the free volume fraction present in a basic sliding unit and the average number of grain boundaries that
Keywords: align to form a planar interface during superplastic deformation are given in some of our earlier papers.
Grain size softening/inverse Hall–Petch These predictions are verified here using experimental data pertaining to one quasi- and two
effect nanocrystalline systems. The agreement between the predictions and the experimental observations is
Grain boundary sliding satisfactory.
Rate controlling process
& 2014 Elsevier B.V. All rights reserved.
Quasi-crystals
Nanocrystalline materials

1. Introduction boundary sliding (mGBS) controlled flow in microcrystalline

materials [9] to include the nanocrystalline range as well. The
In micro-mechanics [1] decreasing grain size has a similar correlation of the flow stress with superplastic strain rates and the
effect as increasing temperature so far as creep effects in a method of knowing the atomistic constants, estimating the free
material are concerned. Therefore, it is understandable that energy of activation for the rate controlling process, the threshold
creep/superplastic effects are seen at room temperature in a stress for the onset of mGBS and the free volume fraction present
high-temperature-melting material like palladium when its micro- in a basic unit of sliding were details presented in later works
structure is nanocrystalline [2]. Koch and Narayan [3] found fault [10–15]. In fact, [7,8] appear to have used an earlier paper of ours
with most of the experimental results in which inverse Hall–Petch [16] extensively, which, like [7,8], deals with grain boundary
(IHP) effect was reported because they noted that the samples sliding (GBS) controlled flow. For example, Eq. (2) of [7] becomes
used were not defect-free. According to them [3] only four sets of identical to Eq. (2) of [16] when it is noted that Nv in [7] is
data demonstrated a genuine grain size softening/IHP effect. represented as N (the number of sites along the grain boundary at
Recently, a grain softening effect in an Al–Cu–Fe based nano-/ which atomic jumps take place aided by stress and temperature)
quasicrystalline material was reported [4] The experimental in [16] and (Ab) [7] as γ [16]. (Nv and N are defined somewhat
results (Fig. 3 of [4]) seem to suggest that the grain size depen- differently in the two papers, but the physical picture is very
dence of hardness in the region in which the grain softening similar.) Post-1990, our model was improved upon significantly by
effect is observed follows IHP relationship, as suggested earlier introducing many microstructural details concerning the general
[5,6]. Notwithstanding this, the authors have interpreted their high-angle boundaries [10–15] because by then the controversies
results in terms of a model [7,8] in which the hardness varies as regarding the structure of general high-angle grain boundaries had
log(L) or L, where L is the average grain size, depending on the got settled and understanding in terms of the structural unit
approximation used. model had emerged [17–21].
The present authors [5,6] presented an explanation for the In brief, in the model the rate controlling processes are
occurrence of IHP effect by refining a model for mesoscopic grain assumed to be confined to the grain/interphase boundary regions.
High-angle boundaries, along which boundary sliding is concen-
trated [22,23], are divided into atomic-scale ensembles that
n
Corresponding author. Tel.: þ 91 40 2313 4455.
surround free volume sites present at discrete locations on the
E-mail addresses: [email protected], boundary characteristic of misorientation and the nature and
[email protected] (K.A. Padmanabhan). magnitude of inter-atomic forces. Due to the presence of free

http://dx.doi.org/10.1016/j.matlet.2014.06.153
0167-577X/& 2014 Elsevier B.V. All rights reserved.
152 K.A. Padmanabhan et al. / Materials Letters 133 (2014) 151–154

volume, these ensembles possess a lower shear modulus com- Table 1

pared with the rest of the boundary and hence constitute the basic Degree of fit for the three relations: Hv α ln ðLÞ, H v α L and Hv α L  0:5 .
units of sliding. Microscopic sliding caused by localized shear
System L (nm) H (GPa) Correlation coefficient of H and
persists till it is rendered ineffective by steric hindrances like at a
triple junction. For GBS to develop to a mesoscopic scale, two or lnðLÞ L L  0.5
more grain boundaries need to align to form a plane interface,
which by further interconnection with other similar plane inter- Al62.5 Cu2.5 Fe12.5 [4] 38.8 11.41 0.9487 0.9168 0.9623
23.5 10.46
faces will lead to large scale sliding till it gets stopped by an
20.6 9.31
insurmountable barrier like an extra-large grain or a coarse 17.5 8.56
precipitate. This plane interface formation process, brought about Zn [25] 11.0 1.06 0.9942 0.9813 0.9979
by dislocation emission from grain boundaries or diffusion, both of 7.9 0.84
which are regarded as faster than GBS in the model, gives rise to a 5.9 0.58
Ni 18Fe [26] 13.9 6.15 0.7867 0.8064 0.7767
long-range threshold shear stress, τo , which has to be overcome for
12.7 5.70
mGBS to set in. Free energy minimization and the possibility of the 11.0 5.71
applied stress doing maximum work in this configuration (princi-
ple of maximum work) are the reasons for plane interface forma-
tion. A mathematical analysis that assumes that the basic sliding
unit is of oblate spheroid shape of base diameter ð5ao Þ and height
ð2:5ao Þ, where ao is the inter-atomic distance in the boundary a convenient notation representing
region (
equal to the atomic diameter) and that the grain shape is γB
rhombic dodecahedron, leads to Eqs. (1a) and (1b). From bubble L1 ¼ 3  0:75 215 N PI
 0:5
ð3Þ
Gαf
raft experiments, the work of Argon [24] and MD simulations
[20,21], the free volume fraction inside the basic sliding oblate with αf a form factor (
1) and γ B specific grain boundary energy
spheroid is taken in a first order approximation as
0.10 for all and NPI the number of grain boundaries that align to form a plane
superplastic systems. interface at a given value of L. (For obtaining numerical values for


 NPI using Eq. (2), within a narrow grain size range NPI is treated as
2 Wγ o ν ðτ  τo Þ γ o V o ΔF o
γ_ ¼ sinh exp  ð1aÞ independent of L. But in general NPI is a strong function of grain
L 2 kT kT
size and temperature.) Using isothermal experimental hardness vs.
grain size data reported and Eqs. (2) and (3), not only one can
1
ΔF o ¼ ðβ1 γ 2o þ β2 ε2o Þ GV o ð1bÞ account for the IHP effect, but also predict the number of grains
2 that align to form a plane interface.
In Eqs. (1a) and (1b), γ_ is the strain rate, W the average grain In this paper the data presented in [4] and in [25,26] are
boundary width ð ¼ 2:5ao Þ, γ o the free volume fraction present in re-analyzed.
the basic unit of sliding present in the grain boundaries, which will The degree of fit, as determined by coefficient of correlation, for
approximately be equal to the shear strain (dependent on the all the three expressions H v vs. ln ðLÞ, H v vs. L and H v vs. L  0:5 is
material; the method of determining it has been given), ν the very similar—Table 1. On this basis alone it is fair to say that both
thermal vibration frequency ( ¼ 1013 s  1 or ðkT=hÞ s  1 , with h our approach [5,6] and that of Conrad and Narayan [7,8] account
the Planck constant), τ the applied shear stress, V o the volume of for the experimental results satisfactorily. This finding clearly
the basic sliding unit ð ¼ ð2=3Þ πW 3 Þ, k the Boltzmann constant, underlines the dangers in preferring one atomistic mechanism
T the temperature of deformation on the absolute scale, ΔF o the over another merely based on gross correlations.
free energy of activation for the basic sliding event, G the shear With reference to the preference of [4] for the model of Conrad
modulus of the basic p sliding
ffiffiffi unit, εo the dilatational strain in a unit and Narayan [7,8] the following observations are in order.
sliding event (ð ¼ γ o = 3Þ, if von-Mises yield criterion is assumed) (a) Figs. 4 and 5 of [4] suggest that the activation energy for the
as the oblate spheroid is embedded in a solid matrix, rate controlling process gets doubled when the same is deter-
β1 ¼ 0:944 ð1:59  p=1  p) and β2 ¼ 4ð1 þ pÞ=9ð1  pÞ. Where p is mined by plotting H v vs. L instead of H v vs. ln ðLÞ. This observation
Poisson's ratio. is in conflict with the finding [7,8] that very similar activation
There are no adjustable constants in the analysis. ΔF o , and σ o , energy for the rate controlling process result from both types of
the only unknowns, are obtained directly from the experimental plots. (b) In the earlier analysis [7,8] the effective stress is taken as
results. (Expressions are available for their theoretical estimation
the applied stress, i.e., the strain-rate sensitivity index, m ¼ 1.0.
also-Eq. (1b) above and Eqs. (13) and (14) of [10]). The numerical But, in nanocrystalline materials at room temperature the value of
procedures used to solve the transcendental Eq. (1a) with a view m is in the range 0.02–0.08 [27,28]. In contrast, in our approach
to comparing the predictions of the model with experimental [5,6,9–15] the value of m can vary from a very low value to 1.0,
findings are also available [10,11,15]. depending on the difference between the applied stress and the
It is p
known that if von Mises yield criterion is assumed threshold stress needed to give rise to mesoscopic boundary
ffiffiffi
τ ¼ H v =3 3, when Hv, the hardness of the material, is reported sliding. Our analysis [5,6,9–15] also suggests (a) a method of
on the Vickers scale. How this relationship gets modified, if other calculating the threshold stress, τo (see Eqs. (13) and (14) of [10])
yield criteria like Mohr–Coulomb are used, has also been and (b) the strain rate of deformation in terms of the material and
explained [12,13]. Thus, one obtains Eq. (2) [5,6]. experimental parameters, including the free energy of activation,
ΔF o —Eqs. (1a) and (1b).
m2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
HV ¼ HV a  ðL  Lo Þ ð2Þ In the papers taken up for analysis, the strain rate of the
L
hardness test is not reported. According to [29] this lies in the
here H V is the measured (steady state) hardness, H V a the hardness range of 5  10  2 s  1 5  10  4 s  1 . (It is not clear from [4]
equivalent of the applied
pffiffiffiffiffi stress at the moment of load application, what value of strain rate was assumed, while determining the
m2 , a constant ð ¼ G L1 =CÞ, with pffiffiffiC a conversion factor (from shear activation energy from their Figs. 4 and 5). By assuming that the
pffiffiffi τ ¼ CH V ; C ¼ 1=3 3 for von Mises yield criterion),
to hardness, strain rate range reported [29] is relevant for the results of
Lo ð ¼ 2W 6Þ the grain size at which τo falls off to zero and L1 is [4,25,26], a free energy of activation for the rate controlling
 K.A. Padmanabhan et al. / Materials Letters 133 (2014) 151–154 153

Table 2
Average free volume fraction in the basic unit of sliding, the number of grain boundaries that align to form the plane interface during mesoscopic boundary sliding and the
free energy of activation for the rate controlling GBS process, according to [9–15].

System: Quasicrystalline Al, L; nm ðrangeÞ :17.6–38.8

G ¼ 74 GPa γo NPI ΔF o ; kJ mol

1
for

γ_ ¼ 5  102 ; s1 γ_ ¼ 5  104 ; s1

1:00G; GPa 74.0 0.0700 1.3689 184.1 198.8

System: Ni–Fe, L; nm ðrangeÞ :11.0–13.9

G ¼ 79:3 GPa γo NPI ΔF o ; kJ mol

1
for

γ_ ¼ 5  102 ; s1 γ_ ¼ 5  104 ; s1

0:70G; GPa 55.5 0.0795 18.0 116.7 129.3

0:75G; GPa 59.5 0.0748 20.7 116.6 129.2
0:80G; GPa 63.4 0.0743 23.6 116.5 129.1
System: Zn, L; nm ðrangeÞ :6.0–11.0

G ¼ 43 GPa γo NPI ΔF o ; kJ mol

1
for

γ_ ¼ 5  102 ; s1 γ_ ¼ 5  104 ; s1

0:70G; GPa 30.1 0.0812 7.8 85.2 97.8

0:75G; GPa 32.3 0.0784 8.9 85.1 97.7
0:80G; GPa 34.4 0.0758 10.2 85.0 97.6

process pertaining to the three sets of results is reported. While [4] Mukhopadhyay NK, Ali F, Scudino S, Khoshkhoo MS, Stoica M, Srivastava VC,
predicting the free volume fraction present in basic sliding unit, et al. Grain size softening effect in Al62.5Cu25Fe12.5 nanoquasicrystals. Appl
Phys Lett 2013;103:201914.
the number of grain boundaries that align to form plane interfaces [5] Padmanabhan KA, Nitsche R., Hahn H. On the deformation of nanocrystalline
and the free energy of activation for the rate controlling process metals and ceramics. In: Fourth European conference on advanced materials,
(Eqs. (1a), (1b)–(3) of this paper) for the two nanocrystalline processes., Milano: Associazone Italiana Di Metallurgia; 1995, p. 289–298.
[6] Hahn H, Padmanabhan KA. A model for the deformation of nanocrystalline
materials, the experimental finding [30,31] that when the grain
materials. Philos Mag B 1997;76:559–71.
size is below 15 nm, the shear modulus falls by up to 30% for a [7] Conrad H, Narayan J. On the grain size softening in nanocrystalline materials.
grain size of
5–6 nm is kept in mind. (For the quasi-/nanocrystal- Scr Mater 2000;42:1025–30.
line material the value of G is given.) The results are summarized [8] Conrad H, Narayan J. Mechanism for grain size softening in nanocrystalline Zn.
Appl Phys Lett 2002;81:2241–3.
in Table 2. [9] Padmanabhan KA, Schlipf J. Model for grain boundary sliding and its relevance
For the two nanocrystalline materials the shear modulus in the to optimal structural superplasticity Part 1—Theory. Mater Sci Technol
bulk, as reported by Frost and Ashby [32], was used. Based on the 1996;12:391–9.
[10] Padmanabhan KA, Gleiter H. Optimal structural superplasticity in metals and
present analysis one is able to suggest that in the Al-based nano-/ ceramics of microcrystalline- and nanocrystalline-grain sizes. Mater Sci Eng, A
quasicrystalline material the plane interface is made up of 1–2 2004;381:28–38.
grain boundaries ( ¼NPI). The same increases to 18–24 and 8–10 [11] Padmanabhan KA, Dinda GP, Hahn H, Gleiter H. Inverse Hall–Petch effect and
grain boundary sliding controlled flow in nanocrystalline materials. Mater Sci
respectively for the Ni–Fe and Zn- nanocrystalline materials. The
Eng, A 2007;452–453:462–8.
likely range of activation energy values ðΔF o Þ and free volume [12] Padmanabhan KA, Basariya MIR. A mesoscopic grain boundary sliding con-
fractions ðγ o Þ in the three cases are also presented in Table 2. The trolled flow model for superplasticity in intermetallics. Int J Mater Res Former
predictions concerning the values of NPI and γ o can be checked by Z Fuer Met 2009;100:1543–51.
[13] Padmanabhan KA, Basariya MR. Mesoscopic grain boundary sliding as the rate
high-resolution TEM and/or MD simulations by any interested controlling process for high strain rate superplastic deformation. Mater Sci
reader. Eng, A 2009;527:225–34.
As capability for prediction of new results is a desirable feature [14] Padmanabhan KA. Grain boundary sliding controlled flow and its relevance to
superplasticity in metals, alloys, ceramics and intermetallics and strain-rate
of a model, there appears to be a case in favor of the present dependent flow in nanostructured materials. J Mater Sci 2009;44:2226–38.
analysis. [15] Sripathi S, Padmanabhan KA. On the experimental validation of a mesoscopic
grain boundary sliding-controlled flow model for structural superplasticity. J
Mater Sci n.d.;49:199–210.
[16] Padmanabhan KA. A theory of structural superplasticity. Mater Sci Eng
1977;29:1–18.
[17] Sutton A, Vitek V. On the structure of tilt grain boundaries in cubic metals I.
References
Symmetrical tilt boundaries. Philos Trans R Soc London, Ser Math Phys Sci
1983;309:1–36.
[1] Padmanabhan KA, Vasin RA, Enikeev FU. Superplastic flow: common basis for [18] Sutton A, Vitek V. On the structure of tilt grain boundaries in cubic metals II.
a ubiquitous phenomenon. Springer; 2001. Asymmetrical tilt boundaries. Philos Trans R Soc London, Ser Math Phys Sci
[2] Chokshi AH, Rosen A, Karch J, Gleiter H. On the validity of the Hall–Petch 1983;309:37–54.
relationship in nanocrystalline materials. Scr Metall 1989;23:1679–83. [19] Sutton AP, Vitek V. On the structure of tilt grain boundaries in cubic metals. III.
[3] Koch C.C., Narayan J. The inverse Hall–Petch effect—fact or artifact? In: Generalizations of the structural study and implications for the properties of
Mukhopadhyay NK, Ali F, Scudino S, Khoshkhoo MS, Stoica M, Srivastava VC, grain boundaries. Philos Trans R Soc London, Ser Math Phys Sci 1983;309:
et al. editors. Farkas D, Kung H, Mayo M, Van Swygenhoven H, Weertman J. 55–68.
Symposium proceedings: structure and mechanical properties of nanophase [20] Wolf D. Structure-energy correlation for grain boundaries in F.C.C. metals—III.
materials – theory and computer simulation vs. experiment, Boston, Massa- Symmetrical tilt boundaries. Acta Metall Mater 1990;38:781–90.
chusetts, USA, 28–30 November 2000: Preface. Mater Res Soc Symp—Proc [21] Wolf D. Structure-energy correlation for grain boundaries in f.c.c. metals—IV.
2001; 634, p B5.1.1–B5.1.11. Asymmetrical twist (general) boundaries. Acta Metall Mater 1990;38:791–8.
154 K.A. Padmanabhan et al. / Materials Letters 133 (2014) 151–154

[22] Padmanabhan KA, Davies GJ. Superplasticity: mechanical and structural [28] Varam S, Rajulapati KV, Bhanu Sankara Rao K. Strain rate sensitivity studies on
aspects, environmental effects, fundamentals and applications. Springer-Verlag; bulk nanocrystalline aluminium by nanoindentation. J Alloys Compd 2014;585:
1980. 795–9.
[23] Nieh TG, Wadsworth J, Sherby OD. Superplasticity in metals and ceramics. [29] Maier V, Durst K, Mueller J, Backes B, Höppel HW, Göken M. Nanoindentation
Cambridge University Press; 1997. strain-rate jump tests for determining the local strain-rate sensitivity in
[24] Argon AS. Plastic deformation in metallic glasses. Acta Metall 1979;27: nanocrystalline Ni and ultrafine-grained Al. J Mater Res 2011;26:1421–30.
47–58. [30] Grewer M, Markmann J, Karos R, Arnold W, Birringer R. Shear softening of
[25] Conrad H, Narayan J. Mechanisms for grain size hardening and softening in Zn. grain boundaries in nanocrystalline Pd. Acta Mater 2011;59:1523–9.
Acta Mater 2002;50:5067–78. [31] Grewer M, Birringer R. Shear shuffling governs plastic flow in nanocrystalline
[26] Cheung C, Djuanda F, Erb U, Palumbo G. Electrodeposition of nanocrystalline metals: an analysis of thermal activation parameters. ArXiv14042457 Cond-
Ni–Fe alloys. Nanostruct Mater 1995;5:513–23. Mat. 2014.
[27] Padmanabhan KA. Mechanical properties of nanostructured materials. Mater [32] Frost HJ, Ashby MF. Deformation-mechanism maps: the plasticity and creep of
Sci Eng, A 2001;304–306:200–5. metals and ceramics. Pergamon Press; 1982.