Está en la página 1de 10

JTTEE5 23:541–550

DOI: 10.1007/s11666-013-0019-4
1059-9630/$19.00 Ó ASM International

Peer Reviewed
Impact Behavior of Intrinsically Brittle
Nanoparticles: A Molecular Dynamics
Perspective
B. Daneshian and H. Assadi

(Submitted August 16, 2013; in revised form October 2, 2013)

Impact behavior of intrinsically brittle materials at the nanoscale is a topic of growing interest, for
instance, in aerosol deposition and cold spraying of ceramic materials. In this work, we examine the
behavior of single-crystalline brittle nanoparticles upon impact on a rigid substrate, within the framework
of a molecular dynamics model. The model is based on Lennard-Jones formulation, where brittleness is
brought about by using a relatively small cut-off interaction distance. Simulations have been carried out
for different values of particle size and velocity. The results show that despite the induced brittleness,
particles start to deform without breaking into fragments, as the particle size falls below a critical value.
They also indicate that the deformation of particles can be accompanied by poly-crystallization and
bonding to the substrate. The necessary conditions for deformation and bonding are also predicted in
view of an analytical model of impact and fracture, considering the effects of particle size and surface
energy. The results are summarized into a parameter selection map, providing an overview of the
conditions for successful deposition of intrinsically brittle materials, in terms of particle size and velocity.
The predictions are interpreted with respect to the results of the relevant kinetic spraying studies.

Keywords aerosol deposition, cold spray, molecular dynam-


deformation and adiabatic shear instability at the inter-
ics, nanoparticles, particle bonding acting surfaces (Ref 9). Based on this view, material
ductility is considered as a prerequisite for bonding
mechanism in CS. This is in contrast with the observation
that ceramic materials, which exhibit no ductility at low
temperatures, can in fact be deposited via AD or CS. The
1. Introduction mechanism through which intrinsically brittle ceramic
particles may deform and get accommodated in a dense
Ceramic materials are often viewed as being com- and pore-free AD or CS deposit has thus remained an
pletely brittle—i.e., lack the capacity to deform plasti- open question.
cally—at room temperature. Recent studies on solid-state An interesting aspect of AD is that it works only if
coating processes, namely, aerosol deposition (AD) particles are smaller than a certain size; much smaller than
(Ref 1-5) and cold spraying (CS) (Ref 6-8) appear to the typical particle size as used in CS (Ref 1-5). In case of
contradict this view. By using AD, for instance, dense cold spraying of brittle materials, also, deposition is found
layers of intrinsically brittle ceramic materials can be to be feasible only when especially tailored spray materi-
successfully deposited at room temperature. In this als—often in the form of agglomerates of sub-micron
method, particles of ceramic materials with a diameter of particulates (Ref 6)—are used. Although the length scale
approximately 100-300 nm are accelerated to relatively seems to play an important role in these solid-state
high velocities of 100-500 m/s before impinging a substrate spraying processes, the influence of particle size on the
in a partially evacuated chamber (Ref 1-5). AD may be deformation behavior of intrinsically brittle particles
regarded as a close variation of CS. Both processes work during high-velocity impact is yet to be fully understood.
on the basis of high velocity impact and bonding of solid On the other hand, the effect of size on the deformation
particles. In conventional CS, nevertheless, particles are behavior of materials, in general, is a well-established
typically metallic, hence deformable, have a relatively subject. It has long been known that cracking in brittle
larger diameter of 5-50 lm, and are accelerated to mark- materials is scale-dependent (Ref 10-13) and that it may
edly higher velocities of up to 1200 m/s. Bonding and be suppressed completely if the sample is sufficiently small
deposition mechanisms in CS are well studied and rela- (Ref 12, 13). More than three decades ago, Kendall
tively well understood. A main supposition in CS is that (Ref 13) demonstrated that crack propagation and frag-
material deposition takes place essentially via plastic mentation of ceramic particles under quasi-static com-
pressive loading is possible only when the particle size is
B. Daneshian and H. Assadi, Department of Materials larger than a certain value. Below this critical particle size,
Engineering, Tarbiat Modares University, Tehran, Iran. the material has no ‘‘option’’ but to deform plastically
Contact e-mail: ha10003@modares.ac.ir. without fracture. Further to KendallÕs finding, it has been

Journal of Thermal Spray Technology Volume 23(3) February 2014—541


shown more recently that pillars of intrinsically brittle adjust the mechanical response of the model system, to
Peer Reviewed

materials can be deformed plastically at room temperature match that of an intrinsically brittle material, the following
when their thickness falls below a critical value (Ref 14- potential function was considered
16). In view of these studies, it seems reasonable to con- ( h  i
12  6
ceive that a critical particle size must also exist in kinetic e rr  rr if rr <rc
U¼ ðEq 1Þ
spraying processes, such as AD and CS, marking the r
0 if r  rc
transition between deformation and fracture.
In this context, the present study aims to provide a where r is the atomic distance between two atoms, rc is a
computational basis for understanding the effect of size on dimensionless cut-off radius, and r and e are the charac-
the impact behavior of nanoparticles of intrinsically brittle teristic length and energy parameters, respectively. In this
materials. Computer models of particle impact allow way, brittle behavior can be induced and adjusted to the
detailed analysis of various physical phenomena at very desired level by choosing an appropriate value of rc. It
short time and length scales, which cannot be monitored should be noted that a lower value of rc leads to more
experimentally. There are several methods that can be prominent brittle behavior, but at the same time, it could
used for the simulation of particle impact, deformation, result in unrealistically low values of strength. Therefore,
and fracture at the continuum level. These include ele- rc should be carefully adjusted to represent realistic
ment vanishing technique (Ref 17), mesh splitting method material behavior. Quantitative modeling of brittle
(Ref 18), node duplicating technique (Ref 19), and behavior in real ceramic systems can be more appropri-
material point method (Ref 20, 21). In order to simulate ately pursued using more elaborate and accurate
the mechanical behavior of brittle materials at the nano- approaches, such as those using Ewald summation in three
scale; however, one should also take into account the dimensions. However, as will be demonstrated later, our
surface energy effects, which become particularly promi- MD model seems to be intricate enough to capture main
nent at nanoscale. In this respect, molecular dynamics features of particle impact that are relevant for the pur-
(MD) (Ref 22, 23) may be regarded as a suitable method pose of the present study. The equations of motion were
of simulation. In this method, the movements and inter- integrated using the Velocity Verlet algorithm. In all MD
actions of atoms or molecules are calculated based on the simulations, a microcanonical ensemble (NVE) with the
Newtonian equations of motion. The method is particu- time step of 5 9 1015 s was used. Main simulations were
larly suited for the study of physical and mechanical performed with r = 0.23 nm and e = 0.96 eV. Additional
properties of materials at small length scales (Ref 23, 24). simulations were also performed with e = 0.32 eV, but
There are various examples of particle-based simulation of these resulted in qualitatively similar results. The initial
cluster deposition in the literature (Ref 25-32). The most configurations corresponded to an initial temperature of
relevant study, which employs MD and focuses particu- about 360 K.
larly on the impact behavior of brittle nanoparticles, has Prior to the impact simulations, the mechanical
been carried out by Ogawa (Ref 33-35). Ogawa explored behavior of the model system under rapid loading was
the influence of various parameters, such as particle examined and adjusted via simulated tensile tests. The
velocity and crystallographic orientations, on the impact objective was to mimic mechanical behavior of real cera-
behavior of nanoparticles in AD. mic materials, which commonly exhibit a combination of
Here, we explore the problem further by focusing both high strength and negligible ductility. The atoms
particularly on the effect of particle size on the impact were initially arranged in a two dimensional close-packed
behavior of nanoparticles. In order to illustrate general grid—reminiscent of the (111) plane in fcc structure—and
features of a brittle-ductile transition, we make use of a cut into the shape of a prenotched tensile test specimen
basic MD model, where brittleness is mimicked simply by (Fig. 1). The specimen, oriented horizontally, had three
incorporation of an adjustable cut-off radius into the parts: one active zone in the middle, one fixed zone
Lennard-Jones potential function. Simulations of particle positioned at the left, and one moving zone at the right
impact are then carried out for various values of impact side, traveling in the x-direction with a fixed velocity of
velocity and particle size. Deformation and fracture of 20 m/s. The tensile-test simulations were performed with
particles are investigated with respect to the structural different values of rc, ranging from 1.2 to 2.5, in order to
changes of the impacted particles. The overall effect of the identify an appropriate cut-off radius for subsequent im-
particle velocity and particle size on the impact behavior pact simulations. As mentioned before, very small values
of particles is subsequently discussed in view of a criterion of rc would be expected to result in rupture at very low
for deformation without fracture, taking into account the stresses, while very large values would be expected to
surface energy effects. result in ductile behavior. Clearly, neither extreme would
represent a real ceramic material, and so, be of interest for
the purpose of the present study. Therefore, the following
criterion was conceived for the selection of cut-off radius:
2. Method The largest rc resulting in the highest strength with no
plastic deformation at fracture.
A molecular dynamics model based on the Lennard- Subsequently, impact simulations were carried out for
Jones potential (Ref 23) was used to study the impact single-crystalline closed-packed nanoparticles of about
behavior of nanoparticles in two dimensions. In order to 10-50 nm in diameter, dp, using the same cut-off radius as

542—Volume 23(3) February 2014 Journal of Thermal Spray Technology


grows from the notch at an early stage, leading to rupture

Peer Reviewed
with negligible ductility. As shown in Fig. 2b, the peak
strength of the material in the latter case is nevertheless
high, almost the same as in the former case with noticeable
ductility. Further decrease of rc resulted in a decrease of
the peak strength, as shown in Fig. 2c. According to the
criterion for the selection of rc (Sect. 2), therefore, the cut-
off radius of 1.5 was selected. Note that this value is the
upper limit of rc that can be used to simulate tensile testing
with no plastic deformation at fracture.
The significant effect of rc on the deformation and
fracture behavior can be interpreted as follows. For the
system to deform plastically, attractive interatomic forces
should sustain while interatomic distances increase, during
deformation and slip of the atomic planes. For the
examined model system, this condition is fulfilled only
when the cut-off radius is noticeably greater than 1.5. For
smaller values of rc, the interatomic forces drop signifi-
cantly during slip, as the interatomic distances increase,
albeit temporarily, to values comparable to the cut-off
radius. As a result, plastic deformation via slip becomes
impossible at lower values of rc, and the material ruptures
with no indication of slip or ductility.

3.2 Impact Simulations: Structural Features


In view of these results, the cut-off radius was fixed to
1.5 throughout the subsequent impact simulations. This
allowed the model material to exhibit an intrinsically
brittle behavior with high strength, in consistence with
high-strength ceramic materials. Figure 3 shows the
structural changes during the impact of a single-crystalline
brittle particle. Here, a most prominent feature is the
formation of discontinuities or cracks, both internally and
Fig. 1 The model setup and the initial configurations in the MD at the surface of the particle. With the progress of impact,
simulations of tensile testing (a) and particle impact (b)
some of these cracks close partially or completely, while
others propagate further and result in fragmentation.
obtained from the previous step, i.e., the value that cor- Another interesting feature is the formation of grain
responded to the samples showing brittle behavior and boundaries, especially near the particle/substrate contact
high strength in the simulated tensile tests. In this way, the area.
model was tuned so that all of the nanoparticles consid- Formation of grains and grain boundaries can also be
ered for the impact simulations would be intrinsically observed in other particles of different size and impact
brittle. The nanoparticles were given initial velocities, vp, velocity. An example is shown in Fig. 4. The observed
of 300-750 m/s before impacting a rigid substrate at right poly-crystallization appears to result from simultaneous
angle. In order to make the substrate rigid, the position of reorientation of small atomic clusters near the contact
atoms within the substrate were fixed, though the corre- area. The corresponding rotations are nevertheless small,
sponding interatomic potentials and forces were calcu- so that only low-angle grain boundaries formed. This is
lated, and hence active, throughout the simulations. also indicated by the calculated diffraction pattern, Fig. 4.
Figure 1 shows the model setup and the initial configura- Dislocation formation is also evident in this example,
tions in the corresponding simulations. though this is less common a feature, as compared to grain
boundaries and cracks.

3. Results 3.3 Impact Simulations: The Effect of Impact


Velocity and Particle Size
3.1 Simulated Tensile Testing: Calibration of rc Figure 5 shows snapshots of MD simulations after
Figure 2 shows the results of simulated tensile tests. impact, for various particles and impact velocities. At
According to the snapshots, Fig. 2a, there is a substantial relatively high impact velocities, the particles bond
change in the deformation behavior of the specimen when initially to the substrate (Fig. 5a). This stage may or may
rc is reduced from 2.5 to 1.5. In the latter case, a crack not be followed by cracking and fragmentation of the

Journal of Thermal Spray Technology Volume 23(3) February 2014—543


Peer Reviewed

Fig. 2 The results of MD simulation of tensile testing, showing (a) snapshots of the specimens, (b) the corresponding stress-strain
curves, and (c) the strength for various values of cut-off radius. The case with rc = 1.5 is taken to represent brittle behavior with the highest
strength

particle. Note that fragmentation is undesirable, as it is the particle remains almost intact and un-deformed as it
associated with unsuccessful or inefficient deposition. The detaches from the substrate after impact.
resulting effect of impact at high velocities can, thus, be
classified into two categories: (i) bonding, and (ii) frag-
mentation. An important finding is that fragmentation
becomes more likely when the particle size or the impact 4. Discussion
velocity is increased. There is also a third behavior, that is
(iii) rebounding of the particle from the substrate, as The results of simulations indicate that brittle behavior
shown in Fig. 5b. Rebounding occurs at relatively low can be mimicked by imposing a cut-off radius on the
impact velocity, here at 300 m/s and lower, in which case Lennard-Jones potential. This behavior is manifested as

544—Volume 23(3) February 2014 Journal of Thermal Spray Technology


Peer Reviewed
Fig. 3 The results of MD simulation for a 46-nm brittle particle impacting a rigid substrate at a velocity of 425 m/s, showing the
development of various microstructural features during impact

Fig. 4 Snapshot of the atomic positions and its respective Fourier transform (calculated diffraction pattern) for a 34.5-nm brittle particle
impacting a rigid substrate at a velocity of 375 m/s. Formation of sub-grains, labeled by letters, and dislocations are evident. The grain
structure is better visible in the inset showing the inverse Fourier transform of the filtered diffraction pattern

the lack of ductility via slip mechanism, and formation of for certain particle velocities and particles sizes, and is
cracks in both tensile testing and particle impact simula- always associated with a structural change, reminiscent of
tions. In the impact simulations, however, there is evi- nanocrystallization, e.g., via dynamic recrystallization.
dence for deformation without fracture. This happens only This feature of the MD simulations is consistent with the

Journal of Thermal Spray Technology Volume 23(3) February 2014—545


Peer Reviewed

Fig. 5 Snapshots of particles of various diameters after impact at various velocities, showing (a) deformation with or without facture,
and (b) rebounding from the substrate with little deformation

existing experimental studies. For instance, TEM obser- impact in AD, calculable via a simple energy balance, is
vations on the AD ceramic coatings have also indicated not nearly enough to bring about plasticity in typical
formation of nanocrystals, which were one order of mag- ceramic materials such as alumina. This is also indicated
nitude smaller than the original grain size (Ref 6, 7). by MD simulations, where the maximum homologous
Formation of nanocrystals during impact appears to be a temperature is predominantly below 0.5.). Moreover, such
key to understanding the deformation mechanism of structural changes are expected to have a significant effect
brittle nanoparticles (Ref 32). Further to the high-velocity on the physical and mechanical properties of the deposit,
impact problem, understanding these microstructural and hence, are important from the viewpoint of applica-
changes could also provide an insight to the room-tem- tions for kinetic spraying processes, such as AD and CS.
perature deformation mechanism of intrinsically brittle Nevertheless, detailed study of deformation mechanism
materials at nanoscale, which is yet to be fully understood and recrystallization during impact calls for more realistic
(It should be noted that temperature rise due to particle and elaborate MD models (see, e.g., Ref 33-35) than what

546—Volume 23(3) February 2014 Journal of Thermal Spray Technology


is employed in the present work. The present study is,

Peer Reviewed
therefore, confined to the general aspects of impact and
deposition behavior, as explained below.
The main aspect of the impact behavior as observed in
the present MD simulations is that smaller particles are
harder to break. That is, the minimum impact velocity
required to break the particle into fragments increases
with decreasing particle size (Fig. 5a). This general
observation is consistent with previous works concerning
static loading, e.g., by Kendall (Ref 13), demonstrating
that below a certain dimension fracture would not occur in
intrinsically brittle materials. It is, therefore, reasonable to
assume that a similar critical size, marking the transition
between fracture and plastic deformation, should also
exist in the case of particle impact.
In order to provide an analytical basis for the estima-
tion of the critical particle size, we consider that there
exists a minimum kinetic energy, Ek, which would be Fig. 6 Snapshots of the fragments of broken particles, showing
a rough an irregular fracture surface
necessary to bring about fracture in a defect-free particle.
This kinetic energy can be conceived to consist of two
parts as follows
Ek ¼ Eel þ Efr ðEq 2Þ
where Eel is the elastic energy stored in the particle, up to
the point of fracture/deformation, and Efr is the fracture
energy. According to this conjecture, fracture, and frag-
mentation will occur, only if the kinetic energy of the
particle upon impact is greater than the value obtained
from Eq 2. Likewise, it means that when fracture does
occur, it ‘‘consumes’’ a part of the kinetic energy that
would otherwise be used for deformation and bonding.
This is based on the presumption that fracture and frag-
mentation counteract bonding and deposition, as men-
tioned before. Each one of the energy terms in Eq 2 can
be written as a function of material and process parame-
ters as follows
Fig. 7 A schematic representation of the non-linear correlation
1 ptd2p 2 between the crack length and the diameter of the region of
Ek ¼ mv2p ¼ qvp ðEq 3aÞ interest, corresponding to a fractal dimension of 1.5
2 8
sion, which is not necessarily an integer number; it is
m Y 2 ptd2p Y 2 between 2 and 3 in a 3D space and between 1 and 2 in a
Eel ¼ ¼ ðEq 3bÞ
2q E 8 E 2D space. For the present 2D model of particle impact, we
assume in a first approximation that the fractal dimension
Efr ¼ stc ðEq 3cÞ of the cracks is 1.5. Figure 7 shows, schematically, an
example of a crack-like geometrical feature associated
where q is the density, Y is the stress at yield/fracture, E is
with a fractal dimension of 1.5.
the YoungÕs modulus, s is the crack length, t is the thick-
Assuming that the example shown in Fig. 7 is roughly
ness of the 2D model system (which has an arbitrary
representative of the crack geometry in the present MD
value), and c is the interfacial energy. The crack length, s,
simulations, the following relationship between the crack
correlates with the particle size, though the correlation is
length and the particle diameter can be conceived
not straightforward. This is so because the crack surface
 n
has a complex geometry, generally characterized as being s dp
rough and irregular (Fig. 6). A description of the crack ¼ ðEq 4Þ
a a
geometry follows.
The fracture surfaces or the cracks in materials are where a is a characteristic length, relating to the length
known to have fractal geometry (Ref 36). In fact, scale of the microstructure, and n is the fractal dimension.
description of the fracture geometry has been an early Combining Eq 2, 3, and 4, with n = 1.5, and rearranging for
application of the fractal concept, and indeed, the reason dp, results in the following relation between the particle
for the selection of the term fractal (Ref 36). According to diameter and the particle impact velocity at the threshold
the fractal concept, cracks are assigned a fractal dimen- of fracture

Journal of Thermal Spray Technology Volume 23(3) February 2014—547


because the particles break into fragments. The deposition
Peer Reviewed

window is, on the other hand, limited by a minimum impact


velocity that is needed to incur plastic deformation and
bonding. This second limit signifies the lower bound of the
kinetic energy, below which deposition in not viable be-
cause the particles bounce back. There is so far no analytical
model to describe the lower bound of the kinetic energy. In
view of the MD simulations, however, one could assume
that the minimum impact velocity depends on material
properties, but it is independent of particle size; for the
present simulations it is roughly around 300 m/s for
e = 0.32 eV, and 375 m/s for e = 0.96 eV. This supposition
has an important consequence. As shown in Fig. 8, there
would be a crossover of these two limiting boundaries of the
deposition window at a particle diameter of about 260 nm.
This means that deposition would never be viable if the
particle size is larger than 260 nm, regardless of the mag-
Fig. 8 An overview of the results of MD simulations, in com- nitude of the impact velocity.
parison with the analytical model of fracture, showing the win- The latter finding may explain why AD works only when
dow of deposition (gray area) for intrinsically brittle
nanoparticles particles are smaller than a certain size. Interestingly, the
critical particle size as obtained from Fig. 8 is also quanti-
  2 tatively consistent with the typical particle size in AD (100-
1 Y2 p 300 nm). This quantitative agreement should nevertheless
dp ¼ qv2p  ðEq 5Þ
a E 8c be interpreted with caution, since the present MD model is
only qualitative, and the analytical model is for a 2D system.
In view of the main energy balance, Eq 2, it can be
On the other hand, the agreement between the MD simu-
shown that fracture will be viable only if the particle
lations and Eq 5 may be taken as a general verification of
diameter is greater than that given by Eq 5. If the particle
the proposed analytical model. It can also be shown that the
diameter is smaller than this value, then the kinetic energy
extension of the analytical model to three dimensions lead
is not sufficient for the creation of new crack surfaces, and
to an almost identical expression for the critical particle size
hence, the particle will either bounce back from the sur-
(see Appendix). Thus, the analytical model could serve as a
face or deform plastically. In principle, no other option
basis for the prediction of the critical particle size in AD of
will be conceivable. One may also rearrange the above
real materials.
relation for vp, and in doing so conceive a critical particle
The above analysis may also shed light on CS of brittle
velocity for a given particle diameter, that would equally
materials. So far, CS of ceramics has been successful only
mark the transition between fracture and deformation.
when the feedstock is in the form of agglomerates of sub-
The critical condition for fracture in fact links the particle
micron particulates. Because of the bow shock that forms
diameter and particle velocity; either parameter could be
in front of the substrate, it is not feasible to use nano-
taken as the relevant critical variable interchangeably.
particles in cold spraying. This is so because smaller par-
The above analysis can be assessed with respect to the
ticles are more effectively decelerated in the bow shock
results of MD simulations, Fig. 8. In the present study, q
region; and hence, sub-micron particles would effectively
and c are 8980 kg/m3 and 2.5 J/m2, respectively (The sur-
be flown away and not reach the substrate, under typical
face energy is obtained by examining the potential energy
cold spraying conditions. According to the above analysis,
of a block as it is halved into two pieces, with flat surfaces on
on the other hand, larger ceramic particles would be
the planes of highest atomic density—e.g., 0, 60, or 120°
expected to fragment upon impact on the substrate,
orientations in Fig. 1a.). The values of Y and E are worked
resulting in negligible deposition efficiency. That is per-
out with respect to the strain-stress curves as obtained from
haps the reason why agglomeration of nanoparticles can
the tensile test simulations (Fig. 2b) to be 6 and 120 GPa,
be used as a workaround. In this way, the bow shock effect
respectively. The only unknown parameter is the charac-
can be alleviated, while the size of the particulates, which
teristic length, a, which is taken here as an adjustable
impinge the substrate at sufficiently high velocities,
parameter. A good agreement between the analytical
remains under the critical particle size.
model and the simulation is obtained by taking a = 0.6 nm,
which is about 2.5 times of the interatomic distance (Fig. 8).
Figure 8 also provides an overview of the window of
deposition in AD, in analogy with the parameter selection 5. Conclusions
map in CS (Ref 37). This window is, on one hand, limited by
fracture, which occurs as the particle diameter and/or par- The present molecular dynamics simulations indicated
ticle impact velocity exceeds the respective value given by three distinct behaviors for the impact of intrinsically
Eq 5. The fracture limit corresponds to the upper bound of brittle nanoparticles: (i) deformation and bonding, (ii)
the kinetic energy, beyond which deposition is not viable fracture and fragmentation, and (iii) rebounding. These

548—Volume 23(3) February 2014 Journal of Thermal Spray Technology


behaviors were shown to depend on two key factors: that the critical particle size, below which deposition is

Peer Reviewed
particle impact velocity and particle size. Increasing the viable, increases with increasing YoungÕs modulus and
impact velocity or the particle size beyond a limit resulted surface energy, or with decreasing particle impact velocity,
in fragmentation, while impact below a minimum velocity density and strength.
resulted in rebounding. The results of simulations have
been interpreted with respect to an analytical model, in
which the kinetic energy of particles was linked to the
References
fracture energy. Within the framework of the present
analysis, and for the specific material model as considered 1. J. Akedo, Aerosol Deposition Method for Fabrication of Nano
for the study, it was shown that bonding and deposition Crystal Ceramic Layer, Mater. Sci. Forum, 2004, 449, p 43-48
2. J. Akedo, Aerosol Deposition of Ceramic Thick Films at Room
would be possible only when the particle size is below
Temperature: Densification Mechanism of Ceramic Layers, J.
0.3 lm, regardless of the value of impact velocity. This Am. Ceram. Soc., 2006, 89, p 1834-1839
finding is generally consistent with the observations in 3. J. Akedo, Room Temperature Impact Consolidation (RTIC) of
aerosol deposition and cold spraying of ceramic materials, Fine Ceramic Powder by Aerosol Deposition Method and
in which the particle size, or the size of particulates within Applications to Microdevices, J. Therm. Spray Technol., 2008, 17,
p 181-198
agglomerates, are generally in the sub-micron range. The 4. J. Akedo, Aerosol Deposition Method (ADM) for Nano-Crystal
simulations also indicate poly-crystallization of the parti- Ceramics Coating Without Firing, MRS Symp. Proc., 2003, 778, p
cles especially near the contact area, which is also con- 289-296
sistent with the existing experimental observations. 5. J. Akedo, J. Park, and H. Tsuda, Fine Patterning of Ceramic
Thick Layer on Aerosol Deposition by Lift-Off Process Using
Photoresist, J. Electroceram., 2009, 22, p 319-326
6. S. Ravanbakhsh, H. Assadi, H. Nekoomanesh, A. Hassanzadeh,
Acknowledgments and E. Taheri-Nassaj, Cold Spraying of Ceramic Coatings—A
Feasibility Study, Proceedings of ITSC 2011, 27-29 September,
B.D. is thankful to Professor Michael P. Marder of Hamburg, Germany
University of Texas at Austin for valuable advice on the 7. N. Tjitra Salim, M. Yamada, H. Isago, K. Shima, H. Nakano, and
M. Fukumoto, The Understanding on Adhesion Mechanism of
calculation of surface energy.
Cold Sprayed TiO2 Coating, Proceedings of ITSC 2011, 27-29
September, Hamburg, Germany
8. M. Yamada, M.E. Dickinson, K. Shima, N. Tjitra Salim, H.
Nakano, and M. Fukumoto, Deposition Behavior and Adhesion
Strength of Cold-Sprayed TiO2 Particles, Proceedings of ITSC
Appendix: Extension of the Analytical 2011, 27-29 September, Hamburg, Germany
9. H. Assadi, F. Gärtner, T. Stoltenhoff, and H. Kreye, Bonding
Model to Three Dimensions Mechanism in Cold Gas Spraying, Acta Mater., 2003, 51, p 4379-
4394
For a 3D system, the kinetic, the elastic, and the frac- 10. A. Carpinteri, B. Chiaia, and P. Cornetti, A Scale-Invariant
ture energy terms can be rewritten as follows Cohesive Crack Model for Quasi-Brittle Materials, Eng. Fract.
Mech., 2002, 69, p 207-217
1 pd3p 2 11. A. Carpinteri, P. Cornetti, F. Barpi, and S. Valente, Cohesive
Ek ¼ mv2p ¼ qv ðEq 6Þ Crack Model Description of Ductile to Brittle Size-Scale Tran-
2 12 p sition: Dimensional Analysis vs. Renormalization Group Theory,
Eng. Fract. Mech., 2003, 70, p 1809-1839
m Y 2 pd3p Y 2 12. C. Gurney, and J. Hunt, Quasi-Static Crack Propagation, Pro-
Eel ¼ ¼ ðEq 7Þ ceedings of the Royal Society of London. Series A. Mathematical
2q E 12 E and Physical Sciences, Vol 299, 1967, p 508-524
13. K. Kendall, The Impossibility of Comminuting Small Particles by
Efr ¼ sA c ðEq 8Þ Compression, Nature, 1978, 272, p 710-711
14. J. Michler, K. Wasmer, S. Meier, F. Ostlund, and K. Leifer,
where sA is the surface area of the crack. In analogy with Plastic Deformation of Gallium Arsenide Micropillars Under
the example shown in Fig. 7, the crack can be assumed to Uniaxial Compression at Room Temperature, Appl. Phys. Lett.,
have a fractal dimension of n = 2.5 in the 3D space, so that 2007, 90, p 043123
one may conceive the following relationship between the 15. F. Ostlund, K. Rzepiejewska-Malyska, K. Leifer, L.M. Hale, Y.
Tang, R. Ballarini, W.W. Gerberich, and J. Michler, Brittle-to-
crack surface area and the particle diameter Ductile Transition in Uniaxial Compression of Silicon Pillars at
 2:5 Room Temperature, Adv. Funct. Mater., 2009, 19(15), p 2439-
sA dp 2444
¼ ðEq 9Þ
a2 a 16. F. Östlund, P.R. Howie, R. Ghisleni, S. Korte, K. Leifer, W.J.
Clegg, and J. Michler, Ductile-Brittle Transition in Micropillar
This leads to the following relationship for the velocity- Compression of GaAs at Room Temperature, Phil. Mag., 2011,
dependent critical particle size in a real 3D system 91, p 1190-1199
17. V. Tvergaard, Influence of Void Nucleation on Ductile Shear
  2 Fracture at a Free Surface, J. Mech. Phys. Solids, 1982, 30, p 399-425
1 2 Y2 p
dp ¼ qvp  ðEq 10Þ 18. X.P. Xu and A. Needleman, Numerical Simulations of Fast Crack
a E 12c Growth in Brittle Solids, J. Mech. Phys. Solids, 1994, 42, p 1397-
1434
The above relation is almost identical to Eq 5, with the 19. R.C. Batra and M.H. Lear, Simulation of Brittle and Ductile
only exception that the numerical coefficient in the right- Fracture in an Impact Loaded Prenotched Plate, Int. J. Fract.,
hand side is changed from 8 to 12. This equation shows 2004, 126, p 179-203

Journal of Thermal Spray Technology Volume 23(3) February 2014—549


20. F. Li, J. Pan, and C. Sinka, Modelling Brittle Impact Failure of 29. K.H. Muller, Cluster-Beam Deposition of Thin Films: A Molec-
Peer Reviewed

Disc Particles Using Material Point Method, Int. J. Impact Eng., ular Dynamics Simulation, J. Appl. Phys., 1987, 61, p 2516-2521
2011, 38, p 653-660 30. J. Kraft, O. Rattunde, O. Rusu, A. Häfele, and H. Haberland,
21. D. Sulsky, S.J. Zhou, and H.L. Schreyer, Application of a Parti- Thin Films from Fast Clusters: Golden TiN Layers on a Room
cle-in-Cell Method to Solid Mechanics, Comput. Phys. Commun., Temperature Substrate, Surf. Coat. Technol., 2002, 158, p 131-135
1995, 87, p 236-252 31. H. Haberland, Z. Insepov, and M. Moseler, Molecular-Dynamics
22. D. Frenkel and B. Smit, Understanding Molecular Simulation: Simulation of Thin-Film Growth by Energetic Cluster Impact,
From Algorithms to Applications, Academic Press, London, 2001 Phys. Rev. B, 1995, 51, p 11061
23. J.M. Haile, Molecular Dynamics Simulation: Elementary Meth- 32. D.M. Chun and S.H. Ahn, Deposition Mechanism of Dry
ods, Wiley, Weinheim, 1992 Sprayed Ceramic Particles at Room Temperature Using a Nano-
24. M.P. Allen and D.J. Tildesley, Computer Simulation of Liquids, Particle Deposition System, Acta Mater., 2011, 59, p 2693-2703
Oxford University Press, Oxford, 1989 33. H. Ogawa, Molecular Dynamics Simulation on the Single Particle
25. P. Jensen, Growth of Nanostructures by Cluster Deposition: Impacts in the Aerosol Deposition Process, Mater. Trans., 2005,
Experiments and Simple Models, Rev. Mod. Phys., 1999, 71, 46, p 1235-1239
p 1695 34. H. Ogawa, Atomistic Simulation of the Aerosol Deposition Method
26. R. Biswas, G.S. Grest, and C.M. Soukoulis, Molecular-Dynamics with Zirconia Nanoparticles, Mater. Trans., 2006, 47, p 1945
Simulation of Cluster and Atom Deposition on Silicon (111), 35. H. Ogawa, Molecular Dynamics Simulation on the Modification
Phys. Rev. B, 1988, 38, p 8154 of Crystallographic Orientation in Fragmented Particles in the
27. H. Haberland, M. Karrais, M. Mall, and Y. Thurner, Thin Films Aerosol-Deposition Process, Mater. Trans., 2007, 48, p 2067-2071
from Energetic Cluster Impact: A Feasibility Study, J. Vac. Sci. 36. M.M. Mandelbrot, D.E. Passoja, and A.J. Paullay, Fractal
Technol. A, 1992, 10, p 3266-3271 Character of Fracture Surfaces of Metals, Nature, 1984, 308,
28. H. Haberland, Z. Insepov, M. Kurrais, M. Mall, M. Moseler, and p 721-722
Y. Thurner, Thin Films from Energetic Cluster Impact; Experi- 37. H. Assadi, T. Schmidt, H. Richter, J.O. Kliemann, K. Binder, F.
ment and Molecular Dynamics Simulations, Nucl. Instrum. Gartner, T. Klassen, and H. Kreye, On Parameter Selection in
Methods Phys. Res. Sect. B, 1993, 80, p 1320-1323 Cold Spraying, J. Therm. Spray Technol., 2011, 20, p 1161-1176

550—Volume 23(3) February 2014 Journal of Thermal Spray Technology

También podría gustarte