Philosophical Magazine, 1 Nov1 Dec 2003

Vol. 83, Nos. 3134, 36093622

Dislocation nucleation in the initial stage

during nanoindentation
H. Y. Liang, C. H. Wooy
Department of Mechanical Engineering, Hong Kong Polytechnic University,
Hong Kong SAR, China

Hanchen Huang
Department of Mechanical, Aerospace and Nuclear Engineering,
Rensselaer Polytechnic Institute, Troy, New York 12180, USA

A. H. W. Ngan
Department of Mechanical Engineering, University of Hong Kong,
Hong Kong SAR, China

and T. X. Yu
Department of Physics, Hong Kong University of Science and Technology,
Hong Kong SAR, China
[Received 17 June 2002 and in nal form 27 May 2003]

Abstract
The microstructure origin of the elasticplastic response of a Cu substrate
during nanoindentation is studied using molecular dynamics simulation.
The elastic response is found to deviate from the Hertzian solution observed
experimentally. The departure can be traced to the small tip radius used in
the simulation. Further penetration sees the development of an inhomogeneous
microstructure. Even at the same strain rate, dierent parts of the contact
surface deform via dierent mechanisms: some elastically, some via the
dislocation bow-out and some via the nucleation and growth of Shockley
partials that sometimes interact to form stair-rod locks. The resultant eect
produces the observed quasi-elastic behaviour on the loaddisplacement curve,
characterized by interspersed minor yields. The present computer simulation
shows in some detail the corresponding dislocation structure development. The
stair-rod lock formation is found to provide a more satisfactory explanation
to the experimentally observed time-delayed occurrence of pop-in below the
spontaneous pop-in load.

} 1. Introduction
As the nanoscale counterpart of the traditional microhardness tests, nanoindentation can provide much more fundamental insight into the mechanical properties of

yAuthor for correspondence. Email: chung.woo@polyu.edu.hk.

Philosophical Magazine ISSN 14786435 print/ISSN 14786443 online # 2003 Taylor & Francis Ltd
http://www.tandf.co.uk/journals
DOI: 10.1080/14786430310001605579

3610

H. Y. Liang et al.

materials with atomic resolution. Indeed, during nanoindentation, the movement of

even just a few dislocations can be detected directly, through the response of the
indentation load. However, the interpretation of nanoindentation results is complex,
because of the large number of dependent variables that aect the loading behaviour,
such as the nature of the surface, the size and tip geometry of the indenter, the
penetration depth, the crystallographic and elastic anisotropy, the thickness of the
substrate and the strain gradient eects (Gouldstone et al. 2000, Gerberich et al.
2002, Swadener et al. 2002).
Two interesting observations have been conrmed by many nano-indentation
experiments. Firstly, single crystals can sustain ultrahigh stresses approaching the
theoretical shear strength. This phenomenon has been attributed to plasticity in
a defect-free region, to a triaxial stress state that stabilizes the structure and raises
the ideal strength in shear, or to surface eects (Gane and Bowden 1968, Krenn et al.
2002). It is often presumed that the substrate behaves elastically before the rst
plastic deformation takes place. However, the possibility of a pre-existing dislocation
around the indenter cannot be excluded (Gane and Bowden 1968), and some
dislocation movements may have already taken place well before the detection of
the rst major plastic response. Indeed, minor dislocation activities have been
reported to appear in the Hertzian elastic stage by Kiely et al. (1999).
Secondly, plasticity during nano-indentation usually occurs as pop-in (in
force-controlled experiments) or sudden force drop (in displacement-controlled
experiments) phenomena, resulting probably from the nucleation, multiplication,
motion and pile-up of dislocations, or phase transition (Gane and Bowden 1968,
Corcoran et al. 1997, Kiely and Houston 1998, Kiely et al. 1999, Gouldstone et al.
2000). Interpretation of experimental results is complicated by the dominance
of surface factors such as steps, asperities, oxides, contamination, or mechanical
hardened surface regions (Kiely and Houston 1998, Kiely et al. 1998, Kramer et al.
2001, Liu and Ngan 2001). This is particularly the case during the very early stages of
loading. Nevertheless, it appears that the rupture of the surface oxide layer is not a
primary mechanism responsible for the displacement excursion (Gouldstone et al.
2000, Minor et al. 2001, Chiu and Ngan 2002). Discrete yielding events during
nanoindentation have been observed by several workers on a clean single-crystal
Au surface with a large at terrace (Corcoran et al. 1997, Kiely and Houston
1998, Kiely et al. 1999). Recently, by recording the entire process of nanoindentation
on lm, Minor et al. (2001) found that the onset of plastic deformation is directly
connected to the movement of dislocations initiated from the surface.
Despite a growing body of experimental studies, atomistic simulation studies still
provide a large proportion of the information concerning the evolution of microstructure during nanoindentation, especially in the initial stage. Using molecular
dynamics (MD) simulation, Kelchner et al. (1998) studied displacement-controlled
indentation on an Au(111) surface and found that dislocation loops started to
appear to one side of the indenter axis, when the applied shear stress reached
7 GPa, well above the critical shear stress level of 2 GPa. Zimmerman et al. (2001)
investigated the eect of surface steps on dislocation nucleation under an indenter of
tip radius of 4 nm and found a signicant reduction in the required load, if applied
next to the step. Rodr guez de la Fuente et al. (2002) studied the formation of
hillocks near indentation points and found that they were composed of four
Shockley partials and a stair-rod dislocation. More recently, the MD simulation
results of Li et al. (2002) suggested that the incipient dislocation loops, which

Initial dislocation nucleation during nanoindentation

3611

were nucleated homogeneously at the gigapascal stress level, were only responsible
for the minor relaxation. Acting as dislocation sources operating at the megapascal
stress level, these then became responsible for most of the plastic behaviour.
In this paper, MD simulation of nanoindentation on a Cu(100) surface is performed, aiming to gain insight into the relationship between the displacementload
behaviour and the associated dislocation structure development.
} 2. Simulation procedure and results
2.1. Simulation procedure
In the following, the embedded-atom method potential of Doyama et al. (1999)
for Cu is adopted, with parameters determined from the experimental values of
cohesive energy, Born stability, elastic constants C11, C12 and C44, formation energy
of a vacancy and stacking-fault energy. The Cu substrate is a rectangular crystal
lattice with dimensions 188 A
188 A
108 A, containing 324 480 atoms (gure 1).
The top surface of the substrate is the (001) crystal plane, and the other two side
surfaces are (100) and (010) planes respectively. Periodic boundary conditions are
applied to the two side faces. The top surface is traction free, and the two atom layers
at the bottom are held xed. The temperature of the simulation cell (i.e. all the
moving atoms) is kept constant at 300 K through the NoseHoover thermostat.
Following Kelchner et al. (1998), a repulsive potential is used to model a spherical
indenter of radius R 40 A, penetrating into a frictionless surface.
The indentation velocity is about 3 m s1, similar to that used by other workers,
such as Christopher et al. (2001) who used a velocity of 40 m s1, and Gannepalli and
Mallapragada (2002) who used a value of 50 m s1. This velocity is nine orders of
magnitude higher than the experimental values, a necessary evil due to the usual
restrictions of MD in terms of both temporal and spatial scales. The sensitivity of
the simulation results to the indentation velocity has to be considered before the
simulation and experimental results can be directly compared. We found that, while
a higher indentation velocity in MD simulation leads to a higher strength, the
relationship between the applied load and the indentation depth remains relatively

Indenter

Z [001]
Y [010]
X [100]

Figure 1.

Schematic diagram of the simulation set-up for indentation on the Cu(100) surface.

3612

H. Y. Liang et al.

unaected. At the same time, the ensuing dislocation structures are also not
substantially dierent. More importantly, the values will quickly converge as
the velocity decreases and approach approximately 3 m s1. In fact, no obvious
dierence is apparent when the velocity is below 3 m s1. Nevertheless, we must be
aware that convergence within MD here does not necessarily ensure that the eects
of time-dependent mechanisms, such as diusion-driven mechanisms, have been
properly accounted for.

2.2. Elastic regime

The resulting indentation depth h is plotted as a function of the applied load F in
the displacementload curve in gure 2. As expected, for small indentation depths of
less than 4.0 A, the displacementload curve exhibits a smooth elastic behaviour.
However, in contrast with published experimental results, which normally follow the
Hertzian relation F / h3=2 , our results in gure 2 do not. This can be understood
because the Hertzian elastic solution requires the indentation depth h to be much
smaller than the tip radius R (i.e. h  R), a condition that is not met in the present
case. The Hertz model has to be corrected for the geometric nonlinearity before it
can be considered for the case where h  R. Indeed, indentation experiments with tip
radius (R 50250 A) comparable with the penetration depth also found a deviation
from Hertzian solution (Kracke and Damaschke 2000, Fraxedas et al. 2002).
Replotting these experimental results, a relation of F / hn , where n  1:9, is
obtained for a depth of less than 15 A. This compares well with the relation
250

0.60
200

0.13

Force (nN)

4
150

3
2

0.06
100

1
50

0
0

10

12

Indentation depth (Angstrom)

Figure 2.

Loaddisplacement curve plotting indentation depth h as a function of applied

load F, from the MD simulation.

3613

Initial dislocation nucleation during nanoindentation

F / h1:85 in the present case obtained by tting to the part of the displacementload
curve before point 1 in gure 2.
The contours of the displacement along the Z axis at a penetration depth of
about 4.0 A are shown in gure 3. This deformation behaviour is typical in the elastic
stage prior to point 1. Figure 3 (a) shows the displacement of the atoms on the
top surface XY. Figure 3 (b) shows the displacement contour on the cross-section
(YZ plane) along the indenter axis (X 0). Because of the elastic anisotropy of the
fcc Cu substrate, the elastic displacement eld is orientation dependent. Indeed, it
can be seen in gure 3 (b) that the vertical atomic displacements a few atomic layers
beneath the surface are not symmetrical with respect to the axis of the indenter,

(a)
Displacement ()

()

[100]

()

[010]

(b)
Displacement ()

()

Z ( )
[001]

[010]

Figure 3. Contours of displacement (Z component) during the elastic stage at a penetration

depth of 4.0 A: (a) the displacement of atoms on the top surface; (b) the displacement
contour on the cross section (YZ plane) along the indenter axis.

3614

H. Y. Liang et al.

indicating larger shear stress o the indenter axis. From gure 3, it can also be seen
that the stressed volume is suciently small relative to the simulation cell size.
2.3. Incipient plasticity
In the loaddisplacement curve in gure 2, there are ve inection points
produced by minor yielding due to inelastic deformation under the indenter. Their
occurrence suggests specic events in the development of the dislocation structure
during the indentation. Between the inection points, the curve shows a relatively
stable and smooth deformation behaviour, suggesting the steady development of the
incipient dislocation structure. The magnitude of the inelastic displacements at
points 1, 3, and 5 are 0.06, 0.13 and 0.60 A respectively, suggesting an increasing
magnitude of the underlying dislocation activities. Beyond the load drop at point 5,
large excursions appear frequently and the curve no longer has any smooth part,
suggesting that the evolution of the dislocation structure has become more complex.
We shall further discuss these observations in relation to the microstructure
development during the indentation in the next few paragraphs.
For an increased radius of the tip, the number of atoms directly under the
indenter also increases. As a result, the relative number of atoms involved in the
nucleation and movement of individual dislocations will decrease, when averaged
over all the atoms under the indenter, most of which are displaced only elastically.
This may be the reason why minor dislocation events in the early stages of loading
are not often observable experimentally (Kiely et al. 1999, Kramer et al. 2001).
In our MD simulation, the size of the indenter is one to two orders of magnitude
smaller than that used experimentally. The stressed volume beneath the indenter
is suciently small that the number of atoms involved in the dislocation events
constitutes a substantial percentage of the total. The dislocation events thus show
up as minor yields in gure 2.
At this point, we note that the present simulation exercise has been repeated for
Cu(111) and Cu(110) surfaces and, similar to the results of Kelchner et al. (1998) and
Zimmerman et al. (2001), the loaddisplacement relation that we obtained also
follows a smooth curve until a sudden large load drop, without the prior occurrence
of minor yields such as those in gure 2.
2.4. Microstructure development
To correlate between the dislocation structure development beneath the indenter
and the loaddisplacement behaviour, one needs to identify the details of dislocation
structure. Several techniques are available, such as common neighbour analysis
(Clarke and Jonsson 1993), atomic stress tensor (Kulp et al. 1993), Voronoi polyhedra analysis (Brostow et al. 1998), centrosymmetry parameter (Kelchner et al.
1998), and slip vector (Zimmerman et al. 2001). The slip vector, which is closely
related to the Burgers vector, is adopted in the present paper to analyse the dislocation structures. A slip vector s
can be dened for each atom as
s


n
1X
x
  X
 ,
ns 6

where x
 and X
 are the vectors linking the atom
with all its nearest neighbours 
in the current and reference positions, and ns is the number of slipped neighbours.

Initial dislocation nucleation during nanoindentation

[001]

[1 1 0]

(a) h = 4.0

(b) h = 4.8

3615

(c) h = 5.6

[110]

(d ) h = 6.4

( g) h = 8.3

(e) h = 6.9

( f ) h = 7.8

( h) h = 8.9

Figure 4. Microstructure evolution beneath the contact surface corresponding to the

inection points in gure 2: (a) a piece of surface elastically deformed prior to point
1; (b) point 1, where the formation and growth of Shockley partials and their interaction to form a LomerCottrell lock; (c) appearance of faulted loops, leading to the
deection near point 2; (d ) growth of the faulted loop and the bow-out of the unlocked
segment on (111) from lock 1 between points 2 and 3; (e) formation of lock 2 and lock
3 perpendicular to lock 1 at point 3; ( f ) conguration after point 4, showing the bowout of the unlocked segment in lock 2 and lock 3; (g) conguration at the beginning of
the large deviation of point 5; (h) conguration just after point 5.

Figure 4 presents a sequence of atomic congurations, in which atoms with slip

vectors having a magnitude js
j ranging from 0.1 to 3.0 A are identied. The sequence
describes the development of the dislocation structure during indentation at various
inection points in gure 2.
Thus, gure 4 (a) represents the deformed surface of the specimen under
the indenter at an indentation depth of 4.0 A. The lack of specic features to indicate
any presence of dislocations suggests that the Cu substrate is deforming elastically,
corresponding to the early part of the loading curve, that is prior to point 1 in
gure 2. As the pentration increases beyond a depth of about 4.8 A (point 1 in
gures 2 and 4 (a)), a pair of Shockley partials is nucleated at the point of highest
stress concentration beneath the contact surface, with (111) and 1 1 1 slip planes

3616

H. Y. Liang et al.

(gure 5 (a)). The Shockley pair grows to intersect with the surface and with
each other, forming a stair-rod dislocation and a LomerCottrell lock (lock 1)
(gures 4 (b) and 5 (b)). In the schematic diagram in gure 6 (a), the locked segment
bc is of pure edge character, formed from the dislocation reaction 16a1 1 2
1 
1 
6a112 3a110 (a is the lattice constant of Cu). The unlocked and non-edge segments ab and cd (with Burgers vector 16a1 1 2), be and cf (with Burgers vector
1 
6a112) connect the locked segment bc to the surface. The slip that occurs during
the formation of lock 1 produces the minor yield at point 1 (gure 2). After the lock
is formed, the crystal hardens, as reected by the resumption of load increase. We
note that, as the indentation depth increases, the contact area also increases beyond
the lock where the deformation strain is smaller and remains in the elastic regime.
Further displacement of the indenter is thus accommodated only partly plastically,
by the glide of the non-edge and unlocked components in the form of bow-outs,
causing the faulted region to expand along the h110i direction (gures 4 and 6). The
balance of the deformation occurs elastically. This explains the complex quasi-elastic
behaviour of the displacementload curve.
The foregoing description is similar to simulation results reported in the literature, such as the formation of dislocation loops o the indenter axis (Kelchner et al.
1998), the nucleation of Shockley partials and stair-rod locks (Rodr guez de la
Fuente et al. 2002) in Au, and the development of the pyramid defect structure on
{111} faulted planes with edges of stair-rod nature (Gannepalli and Mallapragada
2002) found using a truncated pyramid indenter.
Although the tip of the indenter is perfectly spherical, Shockley partials form
on only two of the four possible {111} slip planes in our simulation, despite the

(a)

[ 001 ]

[ 001 ] (111)

[ 1 10 ]

( 1 1 1)

[ 110 ]

(b)

[ 001 ]

[ 001 ]
[ 1 10 ]

[ 110 ]

Figure 5. Formation of lock 1: (a) a pair of Shockley partials nucleate at the high stress
region beneath the contact surface; (b) the partials grow and interact to form a stairrod dislocation and the LomerCottrell lock.

Initial dislocation nucleation during nanoindentation

3617

(a)
[001]
d

[1 10]
a

(111)

[110]

( 001)

( 1 1 1)

(b)
d

b
c

(c)

l
a

b
c

bb
l

Figure 6.

Schematic diagram of the bow-out of dislocation loops from lock 1 b 16a1 1 2;
b0 16a1 1 2 .

symmetry of the load. This symmetry-breaking occurrence is typical of bifurcations,

and is due to the stochastic nature of nucleating events at nite temperatures.
The presence of the lock restricts the plastic ow of the material inside the
wedged-shape region bounded by the two {111} slip planes and therefore hardens
the material inside the wedge. The load transmitted by the indenter through the wedge
creates a stress concentration at its tip, which generates, at a suciently large load,
the proper condition for the nucleation of a pair of faulted loops (gure 7), with
Burgers vectors 16a1 1 2 on the (111) plane and 16a1 1 2  on the 1 1 1 plane, which grow
along both sides of the wedge, producing the shear that allows the wedge to slip
vertically downwards (compare the size of the fold between gures 8 (a) and (b))
under the load, thus producing the minor yield at point 2 (gure 2). As the faulted
loops grow (gure 7 and gures 8 (a) and (b)), the thickness of the wall of lock 1
changes from two atomic layers (intrinsic fault) to three (extrinsic fault). When the
faulted loop reaches the surface, the quasi-elastic behaviour resumes.

3618

H. Y. Liang et al.
[ 001 ]

[ 001 ]

(a)
[ 110 ]

[ 1 10 ]

(111)

( 1 1 1)

Fault loop

[ 001 ]

[ 001 ]

(b)

[ 110 ]
[ 1 10 ]

Fault loop
Figure 7.

Formation of faulted loops on the side of lock 1: (a) the faulted loops begin to
nucleate; (b) fault loops grow and come to a stop.

When the contact area becomes larger as the loading continues, the increase in
local stress forces the unlocked non-edge segments ab and cd to bow out, expanding
the faulted area on the (111) plane (gure 9 (a)). The resulting slip allows the local
stress in the region above to relax. This starts the minor yield at point 3. On the other
side of lock 1, away from the bow-out, the local stress continues to increase, causing
the nucleation of a pair of Shockley partials that produces a LomerCottrell lock
(lock 2), in a similar way to the formation of lock 1 (gure 9 (a)). Subsequent to the
nucleation and growth of lock 2, the stress increase on the diametrically opposite side
across lock 1 causes the nucleation of a similar pair of partials, eventually forming
lock 3 (gure 9 (b)). Both plastic processes contribute to the minor yield at point 3.
The displacement at point 3 is 0.13 A, twice that of point 1 (0.06 A). The larger
slip distance can be seen from the larger sizes of lock 2 and lock 3, compared with
lock 1. After the formation of lock 2 and lock 3, the yielding is interrupted and the
loaddisplacement curve resumes its quasi-elastic behaviour between points 3 and 4.
At this point, the microstructure beneath the indenter is made up of three locks,
namely lock 1, lock 2 and lock 3, forming a cross. At the same strain rate, dierent
parts of the surface deform via dierent mechanisms, some elastically, some via the
dislocation bow out on the (111) plane and some via the nucleation and growth of
Shockley partials, at high stress concentration points. The resultant eect produces
the quasi-elastic behaviour on the loaddisplacement curve between points 3 and 4.

3619

Initial dislocation nucleation during nanoindentation

(a)

fold

(b)

[001]

[001]

[110]

[110]

(c)

[001]

[110]
Figure 8. (a), (b) The cross-sections of the 1 10 plane across the indenter axis, corresponding
to gures 5 (b) and 7 (b) respectively; (c) cross-section of the 1 10 plane across the
chain line l l0 in gure 6 (c), snapshot between points 3 and 4 in gure 2.

Beyond point 4, the expansion of the contact area drives the non-edge segments
on lock 2 and lock 3 to bow out on the 1 11 faulted plane, producing the major
plastic deformation at point 5. The large excursions beyond point 5 correspond to the
usual cold-working process, in which dislocations are generated (load drop), interact
and form locks (load raise). We also note that, up to point 5 where major plasticity
starts to occur, the locks remain intact and act as dislocation sources.
} 3. Experimental interpretation
Feng and Ngan (2001) and Chiu and Ngan (2002) observed the time-delayed
occurrence of pop-in below the spontaneous pop-in load during their nanoindentation experiments (gures 10 (a) and (b)). In the following, we argue that the formation of the LomerCottrell lock prevents the further slip of the partial dislocations
on the {111} planes and may thus oer a better explanation for the experimental
observation.
In these experiments, the indentation was performed with a large and wellannealed Al crystal using a Berkovich tip. The load P was rst ramped to 120 mN
and held for 10 min, followed by unloading (gure 10 (a)). No plastic deformation
could be detected from the displacement response after this cycle. The load was then
reapplied at a higher level of 200 mN. This time a pop-in occurred after a holding
time of about 400 s, as shown by the occurrence of a load drop and a surge in
indenter displacement h. The loading and displacements as functions of time are
shown in gure 10 (a), and the corresponding loaddisplacement relation is shown
in gure 10 (b).

3620

H. Y. Liang et al.

(a)

(b)

(c)

[001]

Lock-3

Lock-2

(d)

(e)

[001]

[110]

[001]

[1 1 0]

Lock-2
Figure 9.

[1 1 0]

Lock-3

The formation process of lock 2 and lock 3: (a)(c) the nucleation and growth of
lock 2 and lock 3; (d ), (e) side views of lock 2 and lock 3.

The time-delay occurrence of pop-in cannot be explained by homogeneous

shear loop nucleation, because the activation energy involved would be much
too large (Chiu and Ngan 2002). Some dislocations might have been generated
during the holding time before pop-in, and these were either locked or existed in
some form of stable state maintained by the load. Using line-tension arguments,
Chiu and Ngan (2002) proposed that a small shear loop of up to a few Burgers
vectors in size could be maintained in stable equilibrium by the load. However, the
line tension model suggested is not likely to be suciently accurate at the small
dimensions involved.
The simulation results here suggest the stair-rod formation to be a more plausible
mechanism for the existence of a stable dislocation structure. Thus, at a small load,
a sizeable pop-in does not occur spontaneously because the free loops are held back
by the stair-rod dislocations, as shown in gures 4 (b)(h). Upon removal of the load,
the image force may very probably unzip the locked dislocation structure, leaving
little or no residual deformation or dislocation debris. However, if the load is
held for an extended time period, as the stair-rod dislocation is of an edge type,
it may climb by pipe diusion or other means. The proximity to the free surfaces
and the large stress levels involved should greatly enhance the climb rate compared
with those under the bulk condition. The dislocation structure may therefore
expand, albeit slowly, with time, and a critical conguration may be surpassed so
that an avalanche of dislocation slip and/or cross-slip activities may follow, yielding
a sizeable pop-in.

Initial dislocation nucleation during nanoindentation

3621

P (N)

(a) 200
150
100
50

h (nm)

111
108
18
15
12
9
6
3
0
0

400

800

1200

Time (s)
(b) 200

P (N)

150

100

50

0
0

20

40

60

80

100

120

h (nm)

Figure 10. Constant-load experiment in Al at loads below the spontaneous pop-in value:
(a) load schedule (upper curve) and displacement response (lower curve); (b) the
corresponding loaddisplacement curve.

} 4. Summary and conclusions

In this paper, the microstructure origin of the elasticplastic deformation due to
nanoindentation on the (100) surface of a Cu substrate is studied using MD simulation. For small indentation depths, the response is elastic but is found to deviate
from the Hertzian solution. This has been shown to be mostly due to the small tip
radius used in the simulation. For larger indentation depths, the deformation is
mechanistically complex, and the ensuing dislocation structure is inhomogeneous.
Even with the same strain rate, dierent mechanisms are found to be responsible for
the deformation of dierent parts of the contact surface, some elastically, some via
the dislocation bow-out and some via the nucleation and growth of Shockley partials
and their subsequent interaction to form locks. The combined eect accounts for the
observed complexity of the quasi-elastic behaviour, with interspersed minor yields on
the loaddisplacement curve.
Experimentally, the stair-rod lock formation may provide a more satisfactory
explanation to the observed time-delayed occurrence of pop-in below the spontaneous pop-in load.
Acknowledgement
The authors are grateful for funding support from the Hong Kong Research
Grant Council PolyU5167/01E, PolyU5177/02E.

3622

Initial dislocation nucleation during nanoindentation

References

Brostow, W., Chybicki, M., Laskowski, R., and Rybicki, J., 1998, Phys. Rev. B, 57,
13 448.
Chiu, Y. L., and Ngan, A. H. W., 2002, Acta mater., 50, 1599.
Clarke, A. S., and Jonsson, H., 1993, Phys. Rev. B, 47, 3975.
Corcoran, S. G., Colton, R. J., Lilleodden, E. T., and Gerberic, W. W., 1997, Phys. Rev.
B, 55, 16 057.
Christopher, D., Smith, R., and Richter, A., 2001, Nanotechnology, 12, 372.
Doyama, M., and Kogure, Y., 1999, Comput. Mater. Sci., 14, 80.
Feng, G., and Ngan, A. H. W., 2001, Scripta mater., 45, 971.
Fraxedas, J., Garcia-Manyes, S., Gorostiza, P., and Sanz, F., 2002, Proc. natn. Acad. Sci.
USA, 99, 5228.
Gane, N., and Bowden, F. P., 1968, J. appl. Phys., 39, 1432.
Gannepalli, A., and Mallapragada, S. K., 2002, Phys. Rev. B, 66, 10 4103.
Gerberich, W. W., Tymiak, N. I., Grunlan, J. C., Horstemeyer, M. F., and Baskes, M. I.,
2002, J. appl. Mech., 69, 433.
Gouldstone, A., Koh, H. J., Zeng, K. Y., Giannakopoulos, A. E., and Suresh, S., 2000,
Acta mater., 48, 2277.
Kelchner, C. L., Plimpton, S. J., and Hamilton, J. C., 1998, Phys. Rev. B, 58, 11 085.
Kiely, J. D., and Houston, J. E., 1998, Phys. Rev. B, 57, 12588.
Kiely, J. D., Hwang, R. Q., and Houston, J. E., 1998, Phys. Rev. Lett., 81, 4424.
Kiely, J. D., Jaraush, K. F., Houston, J. E., and Russell, P. E., 1999, J. Mater. Res., 14,
2219.
Kracke, B., and Damaschke, B., 2000, Appl. Phys. Lett., 77, 361.
Kramer, D. E., Yoder, K. B., and Gerberich, W. W., 2001, Phil. Mag. A, 81, 2033.
Krenn, C. R., Roundy, D., Cohen, M. L., Chrzan, D. C., and Morris, J. W., Jr, 2002,
Phys. Rev. B, 65, 134 111.
Kulp, D. T., Ackland, G. J., Sob, M., Vitek, V., and Egami, T., 1993, Modelling Simulation
Mater. Sci. Engng, 1, 315.
Li, J., Van Vliet, K. J., Zhu, T., Yip, S., and Suresh, S., 2002, Nature, 418, 307.
Liu, Y., and Ngan, A. H. W., 2001, Scripta mater., 44, 237.
Minor, A. M., Morris, J. W., Jr, and Stach, E. A., 2001, Appl. Phys. Lett., 79, 1625.
Rodri guez de la Fuente, O., Zimmerman, J. A., Gonza lez, M. A., de la Figuera, J.,
Hamilton, J. C., Pai, W. W., and Rojo, J. M., 2002, Phys. Rev. Lett., 88, 36 101.
Swadener, J. G., George, E. P., and Pharr, G. M., 2002, J. Mech. Phys. Solids, 50, 681.
Zimmerman, J. A., Kelchner, C. L., Klein, P. A., Hamilton, J. C., and Foiles, S. M., 2001,
Phys. Rev. Lett., 87, 165 507.