Available online at www.sciencedirect.

com

ScienceDirect
Available
Available online
online at at www.sciencedirect.com
www.sciencedirect.com
Structural Integrity Procedia 00 (2016) 000–000

ScienceDirect
www.elsevier.com/locate/procedia
ScienceDirect
Procedia Structural
Structural IntegrityIntegrity
Procedia2 (2016) 1351–1358
00 (2016) 000–000
www.elsevier.com/locate/procedia

21st European Conference on Fracture, ECF21, 20-24 June 2016, Catania, Italy

Defect sensitivity of single-crystal nano-sized Cu beams

XV Portuguese Conference on Fracture, PCF 2016, 10-12 February 2016, Paço de Arcos, Portugal
Solveig Melin*, Per Hansson, Aylin Ahadi
Thermo-mechanical modeling of a high pressure turbine blade of an
Division of Mechanics, Lund University,P.O. Box 118, 22100 Lund, Sweden
airplane gas turbine engine
Abstract
P. Brandãoa, V. Infanteb, A.M. Deusc*
Molecular
a
dynamics simulations of nano-sized beams with square cross sections of single-crystal Cu, solid as
Department of Mechanical Engineering, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisboa,
well as holding defects and loaded in displacement controlled tension until rupture have been performed. The defects
Portugal
are beither
IDMEC,edge crack-like
Department or through-the-thickness
of Mechanical voids.Técnico,
Engineering, Instituto Superior ThreeUniversidade
different decross section
Lisboa, sizesPais,
Av. Rovisco and two different
1, 1049-001 Lisboa,
crystallographic
c
orientations have been investigated. As expected,
Portugal both geometry and crystal orientation influence
CeFEMA, Department of Mechanical Engineering, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisboa,
the mechanical behavior. The strain at plastic initiation was, however, found almost independent of cross section size
Portugal
at given geometry and crystal orientation. Not surprisingly, the presence of defects lowers the strain at both plastic
initiation and at rupture. More surprising is that embedded through-the-thickness voids under some circumstances
might close to heal the beam cross section. The healing process seems to strengthen the beam and delay the final
Abstract
rupture. This applies for thin enough beams, allowing strong enough interatomic forces over the void, assisted by
During slip
pertinent theirevents.
operation, modern aircraft engine components are subjected to increasingly demanding operating conditions,
especially the high pressure turbine (HPT) blades. Such conditions cause these parts to undergo different types of time-dependent
degradation,
Copyright © 2016one
TheofAuthors.
which Published
is creep. Abymodel
Elsevierusing
B.V. the
Thisfinite element
is an open method
access article (FEM) was
under the CCdeveloped,
BY-NC-NDinlicense
order to be able to predict
© 2016 S Melin, P Hansson, A Ahadi. Published by Elsevier B.V.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
the creep behaviour of HPT blades. Flight data records (FDR) for a specific aircraft, provided by a commercial aviation
Peer-reviewunder
Peer-review under responsibility of Scientific
the Scientific Committee of ECF21.
company, wereresponsibility of the
used to obtain thermal and Committee of ECF21.
mechanical data for three different flight cycles. In order to create the 3D model
needed for the FEM analysis, a HPT blade scrap was scanned, and its chemical composition and material properties were
Keywords:
obtained.Defect
The nano-beams,
data that wassingle-crystal
gathered wasCu, void closure
fed into theinFEM
nano-beams,
model tensile nano-beams
and different simulations were run, first with a simplified 3D
rectangular block shape, in order to better establish the model, and then with the real 3D mesh obtained from the blade scrap. The
overall expected behaviour in terms of displacement was observed, in particular at the trailing edge of the blade. Therefore such a
1. model
Introduction
can be useful in the goal of predicting turbine blade life, given a set of FDR data.

©Nano-sized
2016 The Authors. Published
components arebycontained
Elsevier B.V.in a wide verity of everyday products; medical as well as functional
Peer-revieware
applications under responsibility
today commonly of the Scientific
found on theCommittee
market. ofSince
PCF 2016.
it is well known that the mechanical properties of
engineering materials at the nano-scale differ from corresponding macroscopic ones, a new awareness of the
Keywords: High Pressure Turbine Blade; Creep; Finite Element Method; 3D Model; Simulation.
necessity of revising traditional dimensioning concepts, established for structures at the macro-scale, is emerging in
the engineering community. A size effect is obviously at hand, cf. eg. Ahadi and Melin (2016) and Olsson et al.
(2007).

* Corresponding author. Tel.: +46 46 223037.

E-mail address: [email protected]

2452-3216 © 2016 S Melin, P Hansson, A Ahadi. Published by Elsevier B.V.

Peer-review underauthor.
* Corresponding responsibility
Tel.: +351of218419991.
the Scientific Committee of ECF21.
E-mail address: [email protected]

2452-3216 © 2016 The Authors. Published by Elsevier B.V.

Peer-review under responsibility of the Scientific Committee of PCF 2016.
Copyright © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer review under responsibility of the Scientific Committee of ECF21.
10.1016/j.prostr.2016.06.172
1352 Solveig Melin et al. / Procedia Structural Integrity 2 (2016) 1351–1358
2 S Melin, P Hansson, A Ahadi/ Structural Integrity Procedia 00 (2016) 000–000

From experiments it has been concluded that it is the relative number of surface atoms as compared to number of
bulk atoms that influences the mechanical properties, cf. e.g. Hommel and Kraft (2001) or Schweiger et al. (2003a)
and Schweiger and Kraft (2003b). This is due to that surfaces imply the absence of atomic bonds, leaving surface
atoms in energetic states deviating from that of bulk atoms. The redistribution of the electron density close to a
surface affects the inter-atomic bonding forces and, thereby, the local mechanical response.
Another crucial factor for the mechanical response on the nano-scale is the crystallographic orientation due to that
the electron distribution close to a surface is orientation-dependent to its nature, cf. i.e. Zhou and Huang (2004). The
large impact from orientation was demonstrated by Hansson and Jansson (2013) and Hansson (2015) for nano-
indentation in thin Cu coatings, were both the elastic and the plastic responses were observed for different
crystallographic orientations. Since, at the atomic scale, all geometrical features play a significant role, cracks or
other defects that introduce free surfaces should significantly influence the mechanical behavior and collapse
mechanisms.
This paper demonstrates the sensitivity to small defects in nano-sized fcc single-crystal Cu beams. To accomplish
this, 3D molecular dynamics simulations, using the free-ware LAMMPS, have been performed. Beams of square
shaped cross sections of three different sizes and of two different crystallographic orientations have been loaded in
displacement controlled tension until final rupture. The beams are either solid or hold defects; edge crack-like from
the surface, or embedded through-the-thickness. As expected, both geometry and crystal orientation influence the
mechanical behavior.

2. Statement of the problem

2.1. Geometrical conditions

Beams of single crystal fcc Cu, loaded under displacement controlled tension in their length directions x are
considered, cf. Fig. 1. The coordinate system (x,y,z) has its origin at the center of a beam. Each beam is built from
the repetition of Cu unit cells with lattice parameter a0 = 3.615Å. The beam length L equals L=100a0 and three
different square cross section sizes s×s, with s = 6a0, 12a0 and 18a0, have been considered. The beams are either
solid, Fig. 1a), or hold defects. The defect is either edge crack-like, Fig. 1b), or a through-the thickness defect,
symmetrically placed with respect to the coordinate system origin, Fig. 1c). The defects have the width w in the x-
direction and the height h in the y-direction, and here the choice w×h = a0×2a0 is made.

Fig. 1 Beam configurations: a) solid beam, b) beam with edge crack-like defect, c) beam with through-the thickness void.

Two different crystallographic orientations have been investigated. For the first orientation, referred to as the [100]-
orientation, we designate the coordinates (x,y,z) to the crystallographic orientations as x = [100], y = [010] and z =
[001]. For the second, referred to as the [110]-orientation, x = [110], y = [-110] and z = [001].

2.2. Molecular dynamics

For the simulations the molecular dynamics free-ware LAMMPS has been employed and the atomic images are
produced using OVITO, developed by Stukowski (2010).
 Solveig Melin et al. / Procedia Structural Integrity 2 (2016) 1351–1358 1353
S Melin, P Hansson, A Ahadi / Structural Integrity Procedia 00 (2016) 000–000 3

The interaction between the Cu atoms is described by an EAM-potential, giving the potential energy of an atom.
It consists of one pair-wise repulsive part and one N-body attractive part with a cut-off radii, cf. Holian and Ravelo

(1995). The potential energy, Ei , of atom i of type α is given by Eq. (1):
(1)
  1
E i  F     ( rij )      ( rij )
   i j  2 i j

where α and β are two types of atoms, rij is the distance between atoms i and j,  is a pair-wise potential function,
 is the contribution to the electron charge density from atom j of type at the location of atom i, and Fα is an
embedding function that represents the energy required to place atom i of type  into the electron cloud. Here only
one type of atoms is present so that α and β are equal. For the present study the potential file named Cu_u3.eam,
provided by LAMMPS and developed by Foiles et al. (1986), has been used.
For the simulations, a NVT-ensemble held at a constant temperature of 0.01 K by a Nosé-Hoover thermostat as
found in Ellad and Miller (2011) is generated. Initially the atomic ensemble constituting the beam is relaxed to its
equilibrium state for 5000 time steps, corresponding to 25ns. Thereafter an axial elongation, resulting in the axial
strain εx, is effectuated by applying a constant velocity of a0/200/ps in the +x- and ―x-directions to the atoms within
four unit cells at each end of the beam, thus mimicking clamped ends. The atoms in between these end cells are free
to move without constraints. The displacement controlled load is applied with time step Δt = 5ps.
The results are evaluated using the Centro-Symmetry Parameter CSP according to Kelchner et al. (1998), as being
a measure of the instantaneous lattice disorder, i.e. the instantaneous plasticity. The CSP for an atom is defined
according to Eq. (2),
N /2

 R R
2
CSP
 i i N /2 (2)
i 1

where the N is the number of nearest neighbors in the surrounding lattice, equal to 12 for a fcc material. Ri and Ri+N/2
are the vectors corresponding to pairs of opposite nearest-neighbors in the lattice. The value of the CSP signals
whether an atom is part of a perfect lattice, a local defect (a vacancy, partial dislocation or a stacking fault), or part
of a free surface. Through the definition in Eq. (2) the CSP is zero for a perfect fcc lattice and non-zero otherwise.
Commonly used CSP values for different situations in fcc lattices are shown in Table 1, cf. Liang et al. (2006) for
the values marked by * in Table 1. However, for atoms situated along edges or at corners, the CSP values reach
much higher values. For the geometries and crystallographic orientations studied here, the pertinent CSP values for
surface atoms together with edge- and corner atoms are inserted in Table 1.
In Fig. 2 the geometry in Fig. 1b) and s = 12 a0 is used to visualize the atoms for which CSP > 21 for the [100]-
orientation; Fig. 2a) shows the situation at zero applied strain, directly after relaxation, and Fig. 2b) at an axial strain
of εx = 0.075. All atoms colored red have their CSP ≤ 21, for the rest of the atoms, 21 < CSP < 60. The highest CSP
values are found for atoms at the edges of the defect, followed by atoms at corners. As seen from Fig. 2b), also edges
formed by slip events attain such high CSP.

Table 1 CSP values for fcc lattices; * after Liang et al. (2006).
Lattice structure CSP
*Ideal fcc structure CSP<3
*Partial dislocation 3<CSP<5
*Stacking fault 5< CSP<9
*Surface atoms 9<CSP<20
Surface atoms [100] 9<CSP<21
Surface atoms [110] 9<CSP<25
Edge- and corner atoms [100] CSP>21
Edge- and corner atoms [110] CSP>25
1354 Solveig Melin et al. / Procedia Structural Integrity 2 (2016) 1351–1358
4 S Melin, P Hansson, A Ahadi/ Structural Integrity Procedia 00 (2016) 000–000

Fig. 2 Atoms with CSP ≤ 21 red; rest of the atoms have 21 < CSP < 60. [100]-orientation and s = 12a0. a) εx = 0, b) εx = 0.075.

3. Results and discussion

Beams with geometries according to Fig. 1 have been loaded until rupture and the strains at onset of plasticity, εi,
εie and εiv for geometries Fig. 1 (a-c), respectively, as well as rupture strains, εf, εfe and εfv Fig. 1 (a-c), respectively,
are recorded for each geometry. Further, in the case of void closure for geometry Fig. 1c, the closure strain εcv is
determined.
The results are presented in Table 2 and depicted in Fig. 3, where lines connect recorded values, for the three
cross section sizes and the two crystallographic orientations. From the results the following observations according
to the sub-sections below can be made:

Fig. 3 Strain at plastic initiation: εi, εie, εiv for geometries Fig. 1 (a-c), at rupture: εf, εfe, εfv for geometries Fig. 1 (a-c) and closure strain εcv for
geometry Fig. 1c). a) [100]-orientation, b) [110]-orientation.
 Solveig Melin et al. / Procedia Structural Integrity 2 (2016) 1351–1358 1355
S Melin, P Hansson, A Ahadi / Structural Integrity Procedia 00 (2016) 000–000 5

Table 2 Strain at plastic initiation: εi, εie, εiv for geometries Fig. 1 (a-c), at rupture: εf, εfe, εfv for geometries Fig. 1 (a-c), and
closure strain εcv for geometry Fig. 1c.
s[orient] 6a0 [100] 12a0 [100] 18a0 [100] 6a0 [110] 12a0 [110] 18a0 [110]
Solid Beam
εi 0.092 0.089 0.087 0.064 0.063 0.064
εf 0.270 0.555 0.378 0.235 0.353 0.578
Beam with edge defect
εie 0.050 0.055 0.053 0.030 0.027 0.027
εfe 0.135 0.228 0.248 0.148 0.275 0.265
Beam with through void
εiv 0.053 0.068 0.065 0.030 0.040 0.035
εcv 0.068 - - 0.060 - -
εfv 0.125 0.203 0.295 0.103 0.298 0.565

3.1 Solid beams

Plasticity is forced along {111}-planes as expected, and rupture occurs through necking somewhere along the
beam. From Table 2 and Fig. 3, solid lines, it is noticed that the strain at plastic initiation, εi, seems practically
independent of cross section size for each orientation. But the [110]-orientation initiates first, in the interval 0.70 <
εi[110] /εi[100] < 0.74.
Also failure occurs first for the [110]-orientation for the two smallest cross section sizes; with εf[110]/εf[100] ≈ 0.87
for s = 6a0 and εf[110]/εf[100] ≈ 0.64 for s = 12a0. The exception is the case of s = 18a0 in the [100]-orientation which
ruptures before s = 18a0 in the [110]-orientation, with εf[110]/εf[100] ≈ 1.5. A closer investigation shows that this is due
to the placement of the necking region. All beams, with the exception of the case s = 18a0 in the [100]-orientation,
necks away from the clamped ends of the beams. For s = 18a0 in the [100]-orientation, however, the neck forms
close to one end where the constraint influences its vicinity and speeds up the necking process.

3.2 Beams with edge crack-like defects

By introducing an edge defect the beam is weakened in the sense that both the strain at plastic initiation, εie, and at
rupture, εfv, are lowered for both orientation, cf. Table 2 and Fig. 3, dashed lines. Again, initiation of the first slip
event seems independent of s. The [110]-orientation initiates plasticity first, with 0.49 < εie[110]/εie[100] < 0.60, like for
the solid beams. Here, however, the [100]-orientation ruptures first, with 1.1 < εfe[110]/εfe[100] < 1.2, opposite to what
applies to solid beams that necked away from the clamped ends. The invers rate becomes εfe[100]/εfe[110] ≈ 0.87, so that
the orientations now behave conversely.
Comparing to the solid beams, the strain at initiation drops more for the [110]-orientation than for the [100]-
orientation, with 0.43 < εie[110]/εi[110] < 0.47 and 0.54 < εie[100]/εi[100] < 0.62, respectively. The opposite applies to the
failure strain, where 0.46 < εfe[110]/εf[110] < 0.78 and 0.41 < εfe[100]/εf[100] < 0.50 with exclusion of the case s = 18a0,
respectively.
The events during deformation are illustrated in Fig. 4 by snapshots showing the CSP for the case s = 6a0 and
both orientations. Slip occurs along {111}-planes first emerging from the defect corners, and rupture eventually
occurs through necking in this, initially defect, region. Final failure takes place due to sliding along the weakest slip
planes. Thus, the edge crack-like defects do not in any of the cases show a mode I crack-like behavior, breaking
atomic bonds. Only shear mechanisms are active.
As seen from Figs 3a),d), showing the states after relaxation but prior to load application, the cavities are
somewhat expanded due to lack of interatomic bonds across the free defect surfaces, causing the beams to bend
slightly. This is most pronounced for the [100]-orientation. This is due to that, for the [110]-orientation, the surfaces
are somewhat rougher than for the [100]-orientation, giving higher attraction over the void in the [110]-orientation
due to slightly shorter interatomic distances.
1356 Solveig Melin et al. / Procedia Structural Integrity 2 (2016) 1351–1358
6 S Melin, P Hansson, A Ahadi/ Structural Integrity Procedia 00 (2016) 000–000

Fig. 4 CSP for beams with s = 6a0. a),b),c) [100]-orientation d),e),f) [110]-orientation. a),d) εx = 0, b),e) εx = 0.075, c),f) εx = 0.115.

3.3 Beams with through-the-thickness voids

Next, consider beams with centrally placed, through-the-thickness voids according to Fig. 1c. The strains at
plastic initiation, εiv, at rupture, εfv, together with strain at eventual closure of the void, εcv, are seen in Table 2 and
Fig. 3, dotted lines. Also for this geometry the voids expand somewhat due to relaxation because of lack of
interatomic bonds over the voids.
Initiation of plasticity is, again, in practice independent of s. For beams with through voids, the [110] orientation
initiates plasticity first, as do the solid beams and beams with edge defects. Here 0.57 < εiv[110]/εiv[100] < 0.59, showing
that the initiation occurs at only about half strain for the [110]-direction, similar to for the geometry with an edge
defects. As for final rupture, εfv increases markedly with size. For the smallest beams (s = 6a0) the [110]-orientation
fails first, with εxei[110] /εxei[100] ≈ 0.82, opposite to the beams with edge defects but in agreement with the solid beams.
One noticeable thing for this geometrical configuration with s = 6a0 is that the voids eventually close for both
orientations. Figure 5 shows the CSP through snapshots at different stages of loading for beams with s = 6a0, Figs
5a)-d), and s = 12a0, Figs 5e)-h), in the [100]-orientation. Plasticity initiate along {111}-planes from the void
corners. For the beams with s = 6a0, the shearing causes the voids to fill and close completely, and ruptures occur
through necking of the now healed mid-sections.
The beam with s = 6a0 oriented in the [100]-orientation fails after that of the [110]-oriented beam. This might be
due to the fact that the defect closes first for [110]-orientation which leads to onset of the necking development
earlier for the [110]-oriented beam. Necking thus seems to strengthen the beam and prolong the final failure.
Increasing s to s = 12a0 instead causes the void to expand so that two beam ligaments, each necking individually,
are formed, cf. Fig. 5 (e-h). This change in rupture behavior probably depends on that the void in a s = 6a0-beam is
surrounded by only two unit cells of atoms above and below it. This makes it possible to establish enough atomic
interaction across the void to bend the beam ligaments towards each other, eventually allowing for shear-assisted
 Solveig Melin et al. / Procedia Structural Integrity 2 (2016) 1351–1358 1357
S Melin, P Hansson, A Ahadi / Structural Integrity Procedia 00 (2016) 000–000 7

closure. Increasing the beam size to s = 12a0 adds to the number of unit cells above and below the void, increasing
the bonding to the beam ligaments for the atoms at the void surfaces, and no void closure will occur.

Fig. 5 CSP for beams with a)-d) s = 6a0 and e)-h) s = 12a0 for the [100]-orientation. a) εx = 0, b) εx = 0.060, c) εx = 0.075, d) εx = 0115, e) εx = 0, f)
εx = 0.115, g) εx = 0.170, h) εx = 0.183.

Further investigations show that void closure also occurs for the [100]-orientation when increasing s to s = 8a0
and s = 10a0. For the [110]-orientation, however, no void closure occurs for s = 8a0 or s = 10a0.
For the two largest beams, for which no void closure occurs, the [100]-orientation ruptures first, with εfv[110]
/εfv[100] ≈ 1.5 for s = 12a0 and ≈ 1.9 for s = 18a0. This, with first rupture for the [100]-orientation, agrees with what
applies to beams with edge defects but disagrees with what applies to solid beams which ruptures first in the [110]-
orientation. Thus the final strength is much higher for the [110]-direction for this geometry and the ratios of εfv[110]
/εfv[100] much higher than those for the beams with edge defects, where it was 1.1 - 1.2. Obviously purely geometrical
effects play a role here.
The lowering of the strain at initiation of plasticity for this geometry, in comparison to solid beams, is less than
for beams with edge defects; here 0.47 < εiv[110]/εi[110] < 0.64 and 0.58 < εiv[100]/εi[100] < 0.76. Thus, even if the
1358 Solveig Melin et al. / Procedia Structural Integrity 2 (2016) 1351–1358
8 S Melin, P Hansson, A Ahadi/ Structural Integrity Procedia 00 (2016) 000–000

remaining ligaments are of the same size, the geometry influences the behavior. As regards rupture strains, the
[100]-orientation gains in comparison with the [110]-orientation; 0.84 < εfv[110]/εf[110] < 0.98 for the largest beams.
Thus, the beam with s = 18a0, holding a relatively speaking smaller defect, reaches 98% of the strength of a solid
beam. This is not the case for the beam of the same size holding an edge for which εfe[110]/ εf[110] ≈ 0.46, only. Again a
geometry influence is at hand. For the [100]-orientation only one comparison can be made between the two defect
beams due to void closure and influence from the clamed ends; the one for s = 12a0. For that case εfe[100] is higher
than εfv[100].

4. Conclusions

Beams of single-crystal Cu, of dimensions 100a0×s×s, s = 6a0, 12a0 and 18a0 along the (x,y,z) directions, with a0
denoting the lattice parameter, have been loaded in displacement controlled tension along x until rupture. Two
crystallographic orientations have been considered; the [100]-orientation for which we designate crystallographic
orientations according to (x,y,z) = ([100], [010], [001]), and the [110]-orientation for which (x,y,z) = ([110], [-110],
[001]). Solid beams have been compared to beams with edge crack-like defects or centrally placed through-the
thickness voids. The defect sizes were of width a0 along x and height 2a0 along y.
The introduction of a defect markedly lowers the strain at plastic initiation; in general to about one half of what
applies to solid beams. The strain at initiation was, in practice, found to be independent of cross section size. Also a
loss in rupture strain is at hand due to a defect. Even if the introduced defect is small, still the defect corners acts as
stress raisers, triggering slip along {111}-planes. Generally, a beam with a through void is seen to initiate plasticity
later than a beam with an edge defect.
It turns out that the crystallographic orientation plays a crucial role. For the two orientations considered here, the
[110]-orientation always initiates plasticity before the [100]-orientation. This can be explained by the rougher
surfaces for the [110]-orientation, making it easier to initiate slip along preferred slip directions.
For beams holding through-the thickness voids, void closure provoked by different slip events occurred if the
cross section size was small enough. These beams thereafter fail through necking of the now healed cross section.
Comparison between when the void closure was completed indicated that void closure delays the final rupture and,
thus, strengthens the beam. For larger cross section sizes no void closure occurred. Instead the beams split in two
ligaments, each necking independently.

References

Ahadi, A. Melin, S., 2016. Size dependence of the Poisson’s ratio in single crystal fcc copper nanobeams. Computational Materials Science 111,
322-327.
Ellad, B. T., Miller, R. E., 2011. Modeling Materials Continuum, Atomistic and Multiscale Techniques. Cambridge University press.
Foiles, S. M, Baskes, M. I., Daw, M. S., 1986. Embedded-atom-method functions for the fcc metals Cu, Ag, Au, Ni, Pd, Pt, and their alloys
Physical review B 33(12).
Hansson, Jansson, 2013. Nanoindentation of thin copper coatings Materials Structure & Micromechanics of Fracture VII, 417-429.
Hansson (2015), Influence of the crystallographic orientation and thickness of thin copper coatings during Nanoindentation, Eng. Frac. Mech.
150, 143-152.
Holian, B.L., Ravelo, R. 1995. Fracture simulations using large-scale molecular-dynamics. Phys. Rev. B 51 17 11275-11288.
Hommel, M., Kraft, O., 2001. Deformation behavior of thin cupper films on deformable substrates. Acta Mater. 49 3935-47.
Kelchner, CL, Plimpton, SJ, Hamilton JC, 1998. Dislocation nucleation and defect structure during surface indentation Phys. Rev. B 58:11085-8
LAMMPS; http://lammps.sandia.gov
Liang. H. Y, Liu, G R, Han, X, 2006, Atomistic simulation on the stiffening and softening mechanism of nanowires Computational Methods,
1667-1672.
Olsson, PAT, Melin, S, Persson, C, 2007, Atomistic simulations of tensile and bending properties of single-chrystal bcc iron nanobeams. Phys.
Rev. B 76 224112.
Schweiger, R, Dehm, G, Kraft, O, 2003a. Cyclic deformation of polycrystalline Cu films. Phil. Mag. 83, 693-710.
Schweiger, R, Kraft, O, 2003b. Size effects in the fatigue behavior in thin Ag films. Acta Mater. 51, 195-206.
Stukowski, A, 2010. Visualization and analysis of atomistic simulation data with OVITO–the Open Visualization Tool. Modelling Simul. Mater.
Sci. Eng. 18.
Zhou, LG, Huang, HC, 2004. Are surfaces elastically softer or stiffer? Appl. Phys. Lett. 84, 1940 .