Journal of

ARTICLE pubs.aip.org/aip/jap
Applied Physics

Atomistic simulation on the deposition

behavior of cold spray
Cite as: J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416
Submitted: 11 May 2024 · Accepted: 23 August 2024 · View Online Export Citation CrossMark
Published Online: 9 September 2024

Jianrui Feng,1,a) Erfeng An,2,a) and Wensen Zhao3

AFFILIATIONS
1
Beijing Key Laboratory of Microstructure and Property of Solids, College of Materials Science and Engineering,
Beijing University of Technology, Beijing 100124, China
2
State Key Laboratory of Explosion Science and Technology, Beijing Institute of Technology, Beijing 100081, China
3
School of Materials Science and Engineering, East China University of Science and Technology, Shanghai 200237, China

a)
Authors to whom correspondence should be addressed: [email protected] and [email protected]

ABSTRACT
Cold spray is an effective method for surface coating, which has been applied in various engineering areas. However, it is difficult to directly
observe the dynamic deformation process in experiments. This paper applies the molecular dynamics simulation to model the deposition of
a monocrystalline Cu particle onto a Cu substrate and, subsequently, carries out a systematic study on the deposition mechanism and micro-
structure evolution. The results indicate that the deposition process consists of an impact stage and a relaxation stage. It is mainly the high
speed collision and the friction following the collision that lead to particle deposition, which, under different circumstances, can be defined
as surface deposition or penetration deposition. Two methods, namely, drastic shear deformation and cooling in the relaxation stage, can
help form nanocrystallines. Jetting and melting are not the necessary factors for the deposition of nano-sized particles. The formation of dis-
location lines is influenced by impact velocities. At lower impact velocities, the dislocation lines are mainly distributed near the contact
surface. However, when the impact velocity is higher, dislocation lines are almost uniformly distributed in the particle.

© 2024 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution-NonCommercial-
NoDerivs 4.0 International (CC BY-NC-ND) license (https://creativecommons.org/licenses/by-nc-nd/4.0/). https://doi.org/10.1063/5.0218416

I. INTRODUCTION Different from experimental methods, molecular dynamics

Cold spray, as a solid-state deposition process, can be directly (MD) simulation can reveal atomic-scale structure evolution and
used to deposit micro-to-nano-sized particles onto a substrate.1,2 interpret the relevant experiments at the microscopic level and,
therefore, is an effective method to clarify the dynamic
During the cold spray process, particles are accelerated to impact
process.14–30 Joshi and James14,15 simulated the deposition process
the substrate at velocities ranging from 300 to 1500 m/s through a
of the Cu nanoparticle onto the Cu substrate. They proposed that
converging–diverging nozzle.3–5 The high speed collision results in
the bonding mechanism is a combination of adiabatic softening,
plastic deformation in both the particles and the substrate to adiabatic shear instabilities, and interfacial jetting. Reddy et al.16,17
acquire a uniform coating.6,7 Such a special coating technique can identified the bonding mechanisms as a combination of metallurgi-
be used to modify, repair, or even strengthen the surface of metallic cal, mechanical, and grain boundary interlocking at the interface
materials in the industry.8–11 Until now, however, it is still extraor- region. Zhao et al.18 systematically investigated the influence of par-
dinarily difficult to obtain complete dynamic microscopic informa- ticles’ crystal orientation on cold spray. They found that crystal ori-
tion of a deposition process through experiments, as the particles entation affects the interactions between the particle and substrate
are quite small and the time span of the cold spray process is very dramatically. Rahmati et al.19 investigated the deformation behavior
short (in the order of 10−9 s).12,13 Nonetheless, it is still quite nec- of the Cu particle accelerated against a Cu substrate. Temitope
essary to get a better understanding of the deposition behavior as it Oyinbo and Jen20–22 investigated the dislocation plasticity mecha-
can help improve the dynamic particle deformation process and, nism of nanoscale ductile materials. In particular, their findings
thus, optimize cold spray. indicate that the initial velocity of particles significantly influences

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-1

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

their interactions with the substrate. Wang et al.23 examined the

deposition behavior, morphologies, energetic, and interfacial con-
figuration of palladium clusters deposited on the heterogeneous
substrates. Zhang et al.24 studied the effects of motion behavior and
the arrangement of clusters on deposition characteristics between
the Pt clusters and Ni substrate. Daneshian and Assadi25 examined
the impact of single-crystalline brittle nanoparticles on a rigid sub-
strate. Their results demonstrate that the deformation, bonding,
fracture, fragmentation, and rebounding are influenced by impact
velocities and particle’s size. Jami and Jabbarzadeh26,27 systemati-
cally investigated the effects of particle’s size and shape on mechan-
ics of impact in the deposition of the titanium nanoparticle on a
titanium substrate. They also investigated the high-velocity deposi-
tion of titanium dioxide and hydroxyapatite nanoparticles on the
titanium substrate.28,29
Nevertheless, some theoretical aspects regarding particle depo-
sition still remain the subject for discussion. First, during particle
deposition, neither the dynamic deformation process nor the depo-
sition mechanism has been systematically studied. Second, the
atomic structure evolution, including the formation of stacking
faults and nanocrystallines, is yet to be thoroughly discussed.
Third, the dislocation line evolution and distribution still need
further exploration. The relation between the dislocation line and
impact velocity has not been discussed.
This paper, aiming to better understand the deposition behav-
ior of cold spray, carries out MD simulations of the deposition of a FIG. 1. Schematic illustration of cold spray. In this paper, the yellow and green
Cu nanoparticle onto a Cu substrate, discusses the influence of atoms, respectively, represent the particle and substrate.
impact velocities on cold spray, and conducts a systematical study
on the deposition process and microstructure evolution based on
the simulation results.
particle translate toward the substrate. The substrate is divided into
three layers: the Newton layer, thermostat layer, and fixed boundary
II. METHODOLOGY layer.18 Periodic boundary conditions are applied to the x and z
directions, and the free boundary condition is applied to the y
The large-scale atomic/molecular massively parallel simulator
direction. To investigate the influence of impact velocities on cold
(LAMMPS) is used to simulate the deposition process of cold
spray, impact velocities (Up), ranging from 250 to 1500 m/s with an
spray.30 Simulations are performed by modeling the impact of a Cu
increment of 250 m/s, are applied to the particle successively. For
nanoparticle on a Cu substrate. An embedded atom method poten-
all simulations, a constant integration time step of 1 fs is used. It
tial developed by Mishin et al.31 is used to describe the interatomic
takes approximately 2 days to finish one simulation by using the
interactions. Using this potential, the lattice constant of Cu is
workstation with 64 cores. It should be mentioned that, in experi-
3.615 Å at 300 K. The particle is simplified as a spherical single
ments of cold spray, the initial temperature of the Cu particle is
crystal particle, which consists of 1 180 251 atoms with a diameter
higher than 300 K. The higher temperature will enlarge the kinetic
of 30 nm. The substrate contains 19 440 000 atoms with dimensions
energy of the particle, which is more helpful to tightly join the par-
of 65 × 54 × 65 nm3. It should be mentioned that, in these simula-
ticles and the substrate together.
tions, the maximum deformation of the deposited particle changes
After simulations, atomic deformation is visualized through
from 30 nm in diameter to approximately 52 nm in the horizontal
OVITO software.32 Von Mises shear strain is used to characterize
direction, smaller than the size of the substrate (65 × 65 nm2). The
the shear flow for each atom between the present and the initial
periodic boundary conditions in the lateral direction have little
configuration.33 Common neighbor analysis (CNA) is adopted to
influence on the simulation results. Both the particle and the sub-
investigate the evolution of micro-structures in the particle and the
strate are, respectively, aligned along the [100], [010], and [001]
substrate.34 Additionally, the dislocation extraction algorithm
crystallographic directions in the x, y, and z directions. The initial
(DXA) is applied to classify the dislocation patterns in crystals and
distance between the particle and the substrate is 1 nm. The config-
output the dislocation lines.35
uration of the initial model is presented in Fig. 1.
Before cold spray, the initial model is first equilibrated at
300 K and zero pressure for 1000 ps in the constant-pressure–tem-
perature (NPT) ensemble. Then, the microcanonical (NVE) ensem- III. RESULTS
ble is used to simulate the collision process. All the atoms in the The MD simulations of the deposition processes of cold spray
particle are given the same velocity along the y axis to make the turn out to be successful. The results demonstrate that the particle

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-2

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

can be deposited onto the substrate under the impact velocities direction is a regular quadrilateral. At the impact velocity of 1250
ranging from 250 to 1500 m/s. or 1500 m/s, a part of the particle penetrates into the substrate
The compression strain (ε) of the deposited particle,26 which [Figs. 3(i) and 3(k)], but the particle almost remains in the circle
is defined by the ratio of the height in the center region to the shape in the top view [Figs. 3( j) and 3(l)]. In these conditions, the
initial diameter, is first calculated. Figure 2 illustrates the compres- particle changes into a crescent shape in the cross-sectional view,
sion strain curve under the impact velocities from 250 to 1500 m/s. and a few atomic inter-diffusions take place across the contact
As can be seen, in each situation, due to the drastic collision surface between the particle and the substrate [Figs. 3(i) and 3(k)].
between the particle and the substrate, the compression strain In cold spray, the flattening ratio (FR) is a critical measure-
increases dramatically at first. At this moment, the compression ment to evaluate the particle deformation and coating quality.36 It
strain rates (dε/dt) are, respectively, 3.49 × 109, 1.04 × 1010, can be calculated by the following equation:
1.55 × 1010, 2.13 × 1010, 2.17 × 1010, and 2.25 × 1010 s−1. Then, com-
pression strain gradually reaches a stable value. The final compres- FR ¼ d/h, (1)
sion strain values (t = 200 ps) are, respectively, 0.12, 0.32, 0.49, 0.59,
0.55, and 0.56. It is quite interesting to note that compression
where d represents the diameter of the deposited particle and h rep-
strain increases by raising the impact velocity from 250 to 1000 m/s
resents the height of the deposited particle.
but decreases under the impact velocities of 1250 and 1500 m/s.
For monocrystalline nanoparticles, under the impact velocities
To clearly observe the particle deformation, morphologies
from 500 to 1000 m/s, the particle’s final shape in the top view is a
from both the cross-sectional view and top view are analyzed.
regular quadrilateral. To accurately analyze the flattening ratio, d is
Figure 3, respectively, illustrates the morphologies from cross-
acquired by calculating the maximum area of the deposited particle
sectional view and top view at 200 ps in each situation. It should be
in the horizontal direction. Figure 4(a) demonstrates the flattening
mentioned that, in all the cross-sectional views of this paper
ratio in each situation at 200 ps. As can be seen, the flattening ratio
(Figs. 3, 5–7), a slice in 1 nm thickness, which is through the center
almost increases linearly when the impact velocity rises. In the cold
of the impact zone (x–y surface), is cut from the deposited particle
spray experiments of titanium particles to alumina substrates per-
and substrate. The deposition behavior is influenced by the impact
formed by Imbriglio et al.,37 they also identified the linear relation
velocities. When Up = 250 m/s, a flat interface is formed at the
between FR and impact velocities.
contact surface [Fig. 3(a)], and the particle almost remains in the
Figure 3 also shows that, in each situation, the particle and the
spherical shape [Figs. 3(a) and 3(b)]. Under the impact velocities
substrate are joined together at atomic scale. It means that, at the
from 500 to 1000 m/s, a concave interface is generated [Figs. 3(c),
contact surface, the particle and the substrate are tightly joined
3(e), and 3(g)], and the particle changes into a square shape in the
together. When 250 m/s ≤ Up ≤ 1000 m/s, in the horizontal direc-
top view [Figs. 3(d), 3(f ), and 3(h)]. In the MD simulations per-
tion, the area of the contact surface is lower than the maximum
formed by Zhao et al.,18 they also found that, at the impact velocity
area of the post-impact particle [Figs. 3(a), 3(c), 3(d), and 3(g)].
of 1000 m/s, the final shape of the particle with the [100] lattice
When Up = 1250 and 1500 m/s, in the horizontal direction, the area
of the contact surface is equal to the maximum area of the post-
impact particle [Figs. 3(i) and 3(k)]. To evaluate the deposition
behavior, the effective coating rate (η) in each situation is calcu-
lated. It is defined as the ratio of the area of the contact surface to
the maximum area of the deposited particle in the horizontal direc-
tion. Then, the effective coating rate can be expressed in the follow-
ing equations:

η ¼ D2 /d2 , (2)

where D represents the diameter of the contact surface between the

particle and the substrate and d represents the maximum diameter
of the deposited particle in the horizontal direction.
The effective coating rate is equal to or less than 1. If it is
equal to 1, the particles and the substrate can be tightly joined
together without any pores. However, if it is less than 1, micro-
pores may be formed at the contact surface between the particles
and the substrate. Figure 4(b) demonstrates the effective coating
rate in each situation at 200 ps. At the impact velocities of 250, 500,
750, and 1000 m/s, the effective coating rates are 0.48, 0.78, 0.81,
and 0.88, respectively. However, at the impact velocities of 1250
and 1500 m/s, the effective coating rate is equal to 1. It indicates
FIG. 2. The compression strain curves of the particle under the impact veloci-
ties from 250 to 1500 m/s.
that when Up ≤ 1000 m/s, micro-pores may be formed, but the size
tends to decrease when the impact velocity increases.

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-3

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

FIG. 3. The morphologies of the deposited particle and substrate at 200 ps, respectively, from the cross-sectional view and top view under the impact velocities of 250
(a) and (b), 500 (c) and (d), 750 m/s (e) and (f ), 1000 (g) and (h), 1250 (i) and ( j), 1500 m/s (k) and (l).

To analyze the deposition process, the von Mises shear strains deformation takes place in both the particle and the substrate. In
of the particle and the substrate are calculated. Figure 5, respec- these situations, the maximum shear deformation region occurs at
tively, illustrates the von Mises shear strain distribution from cross- the contact surface [t = 10 ps in Figs. 5(e) and 5(f )], because the
sectional view in each situation at 10, 20, 30, and 200 ps. As can be particle penetrates into the substrate. The shear strain at the edge
seen, the shear deformation initiates at the periphery of the contact region of the contact surface is much higher than that at the center
surface [t = 10 ps in Figs. 5(a) and 5(b)] and transmits into the par- region [t = 10 ps in Figs. 5(e) and 5(f )]. In Fig. 5, it can be also
ticle and substrate with 45° inclined angle [t = 20 ps and t = 30 ps in found that, in each situation, particle deformation mainly takes
Fig. 5(a)]. Under the impact velocities from 250 to 1000 m/s, shear place in the first 30 ps. At the time periods from 30 to 200 ps, the
deformation results in the slip deformation along the {111} surface morphology in both the particles and the substrate does not display
in the particle [Figs. 5(a)–5(d)]. As a result, the particle, in the top any significant change.
view, gradually changes into the square shape [Figs. 3(d), 3(f), and To analyze the atomic structure deformation, the CNAs of the
3(h)]. At the impact velocity of 1250 or 1500 m/s, drastic shear particle and the substrate are calculated. Figure 6 illustrates the

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-4

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

some disordered atoms as grain boundaries are formed in the

deposited particle [t = 200 ps in Figs. 6(d)–6(f )]. It indicates that
nanocrystallines are generated in the deposited particle. What is
more, a few nanocrystallines are also formed near the contact
surface of the substrate [t = 200 ps in Figs. 6(d)–6(f )]. The quantity
of nanocrystallines tends to increase by raising the impact velocity
[t = 200 ps in Figs. 6(c)–6(f )]. At the impact velocity of 1250 or
1500 m/s, nanocrystallines are almost uniformly distributed in the
deposited particles [t = 200 ps in Figs. 6(e) and 6(f )].
To analyze whether melting takes place during particle deposi-
tion, the temperatures in the particle and the substrate are calcu-
lated. To accurately compute the temperature distribution, a slice in
2 nm thickness, which is through the center of the impact zone (x–
y surface), is first cut from the deposited particle and substrate.
Then, the section is divided into 1419 bins in the yz plane (33 bins
in the y-direction and 43 bins in the z-direction). The dimensions
of each bin are 2 × 2 × 2 nm3. The temperature can be calculated by
the following equation:

1 X3 X N
j 3
Ek ¼ mi (vi  vj )2 ¼ kB T, (3)
2N j¼1 i¼1 2

where Ek is the average kinetic energy per atom, kB is Boltzmann’s

constant, T is the temperature, N is the number of atoms, mi is the
molar mass, vj is the j component of the average velocity in each
j
bin, vi is the j component of the peculiar velocity of the atoms i,
and 3 is the degrees of freedom per atom.
Figure 7 shows the temperature distributions in the deposited
particles under the impact velocities from 250 to 1500 m/s at 10,
20, 30, and 200 ps. Under the impact velocities from 250 to
1000 m/s, the high temperature mainly distributes in the shear
band region [t = 20 ps in Figs. 7(a)–7(d)], and the temperature
gradually rises when the impact velocity is increased [t = 30 ps in
Figs. 7(a)–7(d)]. Under the impact velocities of 1250 and 1500 m/s,
at first, the temperature at the edge of the contact surface is the
highest [t = 10 ps in Figs. 7(e) and 7(f )]. Then, the high tempera-
ture gradually distributed in the shear band region [t = 30 ps in
Figs. 7(e) and 7(f )]. It can be found that, at the impact velocity of
1500 m/s, the maximum temperature at the shear band region is
FIG. 4. The flattening ratio (a) and effective coating rate (b) in the deposited approximately 750 K, which is much lower than the melting point
particle under the impact velocities from 250 to 1500 m/s at 200 ps. of copper. It indicates that melting does not take place in the
deposited particle and the substrate. As heat conduction occurs, the
temperature in both the particle and the substrate gradually
decreases (t = 200 ps in Fig. 7).
atomic structure from the cross-sectional view in each situation at To analyze the dislocation evolution, the DXAs of the particles
10, 20, 30, and 200 ps, respectively. When Up = 250 m/s, a few are calculated. Figure 8 shows the dislocation distributions in the
stacking faults with HCP structures are generated in both the parti- deposited particles under the impact velocities from 250 to 1500 m/s
cle and substrate [Fig. 6(a)]. When the impact velocity rises, many at 10, 20, 30, and 200 ps. To clearly observe the dislocation lines,
more stacking faults are generated. In the shear deformation we reduce the diameter of the atoms to 0.01 nm with black color.
regions and the edge of the contact surface, some atoms change As can be seen, the dislocation lines initiate at the contact surface
into a disordered state [t = 20 ps in Figs. 6(b)–6(f)]. In each situa- and transmit toward the upper surface. When Up = 250 m/s, the
tion, due to the relaxation in the particle and the substrate, most dislocation lines are mainly distributed at the contact surface
atoms with HCP and disordered structures gradually change into [t = 200 ps in Fig. 8(a)]. With an increase in the impact velocity,
an FCC structure again at 200 ps. When 250 m/s ≤ Up < 1000 m/s, many more dislocation lines are produced in the deposited parti-
the atoms at the contact surface mainly arrange in the FCC struc- cles. When Up = 1250 and 1500 m/s, the dislocation lines are
ture [t = 200 ps in Figs. 6(a)–6(c)]. However, when Up ≥ 1000 m/s, almost uniformly distributed in the particle [t = 200 ps in Figs. 8(e)

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-5

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

FIG. 5. The von Mises shear strain of the deposited particle and substrate from the cross-sectional view under the impact velocities of 250 (a), 500 (b), 750 (c), 1000 (d),
1250 (e), and 1500 m/s (f ).

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-6

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

FIG. 6. The CNA of the deposited particle and substrate from the cross-sectional view under the impact velocities of 250 (a), 500 (b), 750 (c), 1000 (d), 1250 (e), and
1500 m/s (f ).

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-7

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

FIG. 7. The temperature distribution in the deposited particle and substrate from the cross-sectional view under the impact velocities of 250 (a), 500 (b), 750 (c), 1000 (d),
1250 (e), and 1500 m/s (f ).

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-8

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

FIG. 8. The dislocation distributions in the deposited particles under the impact velocities of 250 (a), 500 (b), 750 (c), 1000 (d), 1250 (e), and 1500 m/s (f ).

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-9

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

and 8(f )]. In these conditions, the dislocation lines are not have increased in the deposited particle. The dislocation segment
observed at the middle region of the particle [t = 20 and 30 ps in and average dislocation length in the deposited particle at 200 ps
Figs. 8(e) and 8(f )], as drastic shear deformation has changed are also calculated, as shown in Fig. 9(b). As the impact velocity
the atoms into the disordered state [t = 20 and 30 ps in Figs. 6(e) increases, the dislocation segment in the deposited particle
and 6(f )]. increases, but the average dislocation length decreases, because
The dislocation densities as a function of simulation time are nanocrystallines have formed in the particle.
also calculated. It is defined as the ratio of the total length of the
dislocation line to the particle volume. As can be seen in Fig. 9(a), IV. DISCUSSION
in each situation, at first, the dislocation density increases very fast
and then begins to decrease. When 250 m/s ≤ Up ≤ 1000 m/s, the A. Deposition mechanism
dislocation density finally reaches a constant value. However, when Based on the simulation results, the deposition process of cold
Up = 1250 or 1500 m/s, the dislocation density increases again at spray can be divided into two stages: the impact stage and the
approximately 24–32 ps and finally arrives at a constant value. At relaxation stage. There are three steps in the impact stage. In step
t = 200 ps, the dislocation densities in each situation are, respectively, one, the collision between the particle and substrate joins the
4.31 × 10−18, 3.14 × 10−17, 4.09 × 10−17, 4.58 × 10−17, 5.02 × 10−17, center region of the contact surface between the particle and the
and 4.61 × 10−17 m−2. The dislocation density decreases under the substrate. In step two, the shear deformation (in the deposited par-
impact velocities from 1250 to 1500 m/s, because disordered atoms ticle) enlarges the contact surface of the particle. In step three, the
friction between the particle and the substrate tightly joins the
periphery region of the contact surface. The bonding strength at
the edge of the contact surface is higher than that at the center
region.38 In the relaxation stage, the morphology of the particles
and the substrate does not show any change. However, some stack-
ing faults and disordered atoms disappear, and the atomic struc-
tures of these atoms change to the FCC structure again.
The deposition of Cu nanoparticles can be defined as surface
deposition or penetration deposition at different impact velocities.
At lower impact velocities (250–1000 m/s in this paper), the parti-
cle deposits onto the substrate surface. It is mainly the high pres-
sure that binds the particle and the substrate together. However, in
such situations, micro-pores may be formed among each deposited
particle. At higher impact velocities (1250–1500 m/s in this paper),
a part of the particle penetrates into the substrate. It is mainly the
drastic shear deformation that joins the particle and the substrate
together. In such conditions, micro-pores can be effectively reduced.
The experimental results also indicate that increasing impact velocity
is a straightforward way to reduce micro-pores.4,39,40
In experiments41–48 and finite-element modeling,2,4,49,50 jet can
be directly observed at the edge of the contact surface. However, in
the simulations of this study, jet is not observed, because the particles
as the object of the study are nano-sized. The jetting in cold spray is
similar to that in other areas, such as explosive welding, micro-
droplet impact, shaped charges, and asteroid strikes.49 In our previ-
ous works, we have observed the jet ejection in explosive welding by
using the finite-element modeling51 and MD simulation.52 We
found that jet is not ejected immediately under drastic collision. In
fact, it is gradually generated as the collision point moving forward.
In our cold spray simulations, if we enlarge the particle size or
increase the impact velocity, the jet will be ejected at the edge of the
contact surface. Therefore, we conclude that the jet is not a necessary
factor for cold spray. The nano-sized particles can be directly depos-
ited onto the substrate without jet ejection.

B. Microstructure evolution
FIG. 9. The dislocation density curves of the particles under the impact veloci- The simulation results demonstrate that microstructure evolu-
ties from 250 to 1500 m/s (a) and their dislocation segments and average dislo-
cation lengths (b).
tion is influenced by impact velocities. Under the impact velocities
from 250 to 500 m/s in this paper, shear deformation in the

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-10

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

deposited particle results in the formation of a few stacking faults. (3) It is mainly the shear deformation that leads to the atomic
Under the impact velocities from 750 to 1500 m/s in this paper, the structure evolution. Two pathways, namely, drastic shear defor-
shear deformation in both the particle and the substrate leads to mation and the cooling process, contribute to the form of the
the generation of nanocrystallines. In the cold spray experi- nanocrystallines.
ment,53,54 recrystallization can be also observed in the deposited (4) The dislocation distribution in the deposited particles is influ-
particles. Based on the simulations, we conclude that there are two enced by impact velocities. At lower impact velocities, the dis-
ways to form the nanocrystallines. One is the drastic shear defor- location lines are mainly distributed near the contact surface.
mation in the particle and the substrate. The other is the cooling At higher impact velocities, the dislocation lines are almost
process in the relaxation stage that changes the disordered atoms uniformly distributed in the particle. When the impact velocity
into the crystalline structure. is up, the dislocation segment in the deposited particle
It should be mentioned that, in these simulations, the atoms increases, but the average dislocation length decreases.
in the particle and the substrate are arranged with the same lattice
direction, and only one particle deposition is considered. The inter-
actions between each neighboring particle are not considered. In ACKNOWLEDGMENTS
experiments, because of the different lattice directions between the This work was supported by the National Natural Science
particles and the substrate, the grain boundary will be formed at Foundation of China under Grant Nos. 12102053 and 12472360.
the contact surface. What is more, the deposited particles will also
extrude and rotate with each other, resulting in the formation of AUTHOR DECLARATIONS
fine grains near the contact surface among each neighbor Conflict of Interest
particle.55
The dislocation distribution in the deposited particles is also The authors have no conflicts to disclose.
influenced by impact velocities. Under the impact velocities of
250–1000 m/s, in the impact stage, collision results in the formation Author Contributions
of dislocation line near the contact surface. The dislocation density Jianrui Feng: Formal analysis (lead); Investigation (lead);
in the deposited particles tends to increase when the impact veloc- Methodology (lead); Project administration (lead); Software (lead);
ity rises. However, in the relaxation stage, the dislocation density Supervision (equal); Writing – original draft (lead); Writing – review
decreases and gradually reaches a constant value. Under the impact & editing (equal). Erfeng An: Supervision (equal); Writing – review
velocities of 1250–1500 m/s, in the impact stage, the dislocation & editing (equal). Wensen Zhao: Methodology (equal).
lines are almost uniformly distributed in the deposited particle.
The drastic shear deformation decreases the dislocation density. DATA AVAILABILITY
However, in the relaxation stage, the dislocation density increases
again and gradually arrives at a constant value. The increase in the The data that support the findings of this study are available
dislocation density is due to the cooling process that changes the from the corresponding authors upon reasonable request.
atoms from the disorder state to the crystalline structure. With an
increase in the impact velocity, the dislocation segment in the REFERENCES
deposited particle increases, but the dislocation length decreases. 1
A. Moridi, S. M. Hassani-Gangaraj, M. Guagliano, and M. Dao, Surface Eng.
Generally, the increase in the dislocation density will cause the 30, 369 (2014).
work hardening in cold spray.56 2
M. Grujicic, C. L. Zhao, W. S. DeRosset, and D. Helfritch, Mater. Des. 25, 681
(2004).
3
T. Schmidt, F. Gartner, H. Assadi, and H. Kreye, Acta Mater. 54, 729 (2006).
4
V. CONCLUSIONS H. Assadi, F. Gartner, T. Stoltenhoff, and H. Kreye, Acta Mater. 51, 4379
(2003).
In this paper, the particle deposition of cold spray is investi- 5
A. Ganesan, M. Yamada, and M. Fukumoto, J. Therm. Spray Technol. 22, 1275
gated through MD simulation. The conclusions can be summarized (2013).
6
as follows: G. Bae, Y. Xiong, S. Kumar, K. Kang, and C. Lee, Acta Mater. 56, 4858 (2008).
7
Y. Xiong, K. Kang, G. Bae, S. Yoon, and C. Lee, Appl. Phys. Lett. 92, 194101
(1) The deposition process consists of an impact stage and a relax- (2008).
8
ation stage. In the impact stage, the particle deformation can S. Yin, P. Cavaliere, B. Aldwell, R. Jenkins, H. Liao, W. Li, and R. Lupoi, Addit.
be divided into three steps: collision, shear deformation, and Manuf. 21, 628 (2018).
9
friction. In the relaxation stage, the morphology of the particles M. Walker, Mater. Sci. Technol. 34, 2057 (2018).
10
and substrate does not change, but some stacking faults and F. Gartner, T. Stoltenhoff, T. Schmidt, and H. Kreye, J. Therm. Spray Technol.
disordered atoms disappear. 15, 223 (2006).
11
Y. Xie, C. Chen, M. Planche, S. Deng, R. Huang, Z. Ren, and H. Liao,
(2) The particle deposition can be defined as surface deposition or
J. Therm. Spray Technol. 28, 769 (2019).
penetration deposition at different impact velocities. At lower 12
S. Yin, X. Suo, Y. Xie, W. Li, R. Lupoi, and H. Liao, J. Mater. Sci. 50, 7448
impact velocities, the particle deposits onto the substrate (2015).
surface. At higher impact velocities, the particle penetrates into 13
T. Schmidt, H. Assadi, F. Gärtner, H. Richter, T. Stoltenhoff, H. Kreye, and
the substrate. Compared with the surface deposition, penetra- T. Klassen, J. Therm. Spray Technol. 18, 794 (2009).
14
tion deposition is an effective method to reduce porosity. A. Joshi and S. James, J. Manuf. Processes 33, 136 (2018).

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-11

© Author(s) 2024
 Journal of
ARTICLE pubs.aip.org/aip/jap
Applied Physics

15 40
A. Joshi and S. James, Procedia Manuf. 26, 190 (2018). G. Bae, S. Kumar, S. Yoon, K. Kang, H. Na, H. J. Kim, and C. Lee, Acta Mater.
16
C. D. Reddy, Z. Q. Zhang, S. Msolli, J. Guo, and N. Sridhar, Comput. Mater. 57, 5654 (2009).
41
Sci. 192, 110363 (2021). A. A. Tiamiyu, Y. Sun, K. A. Nelson, and C. A. Schuh, Acta Mater. 202, 159
17
C. D. Reddy, Z. Q. Zhang, S. Msolli, J. Guo, and N. Sridhar, Surf. Coat. (2021).
42
Technol. 433, 128085 (2022). M. Hassani-Gangaraj, D. Veysset, K. A. Nelson, and C. A. Schuh, Scr. Mater.
18
P. Y. Zhao, Q. Zhang, Y. B. Guo, H. Liu, and Z. Q. Deng, Comput. Mater. Sci. 145, 9 (2018).
43
184, 109859 (2020). Y. Sun, D. Veysset, K. A. Nelson, and C. A. Schuh, Appl. Phys. Lett. 117,
19
S. Rahmati, A. Zúñiga, B. Jodoin, and R. G. A. Veiga, Comput. Mater. Sci. 134105 (2020).
44
171, 109219 (2020). M. Hassani-Gangaraj, D. Veysset, K. A. Nelson, and C. A. Schuh, Appl. Surf.
20
S. Temitope Oyinbo and T. C. Jen, Coatings 10, 1079 (2020). Sci. 476, 528 (2019).
21 45
S. T. Oyinbo, T. C. Jen, S. A. Aasa, O. O. Abegunde, and Y. D. Zhu, Manuf. A. A. Tiamiyu, X. Chen, E. L. Pang, Y. Sun, J. Lienhard, J. M. LeBeau,
Rev. 7, 29 (2020). K. A. Nelson, and C. A. Schuh, Appl. Surf. Sci. 574, 151673 (2022).
22 46
S. T. Oyinbo, T. C. Jen, Y. Zhu, J. S. Ajiboye, and S. O. Ismail, Vacuum 182, J. Lienhard, C. Crook, M. Z. Azar, M. Hassania, D. R. Mumm, D. Veysset,
109779 (2020). D. Apelian, K. A. Nelson, V. Champagne, A. Nardi, C. A. Schuh, and
23
K. Wang, J. F. Liu, and Q. H. Chen, Appl. Surf. Sci. 376, 105 (2016). L. Valdevit, Acta Mater. 197, 28 (2020).
24 47
X. Q. Zhang, X. J. Li, and T. Z. Shi, Coatings 12, 142 (2022). Y. Sun, D. Veysset, K. A. Nelson, and C. A. Schuh, J. Appl. Mech. 87, 091002
25
B. Daneshian and H. Assadi, J. Therm. Spray Technol. 23, 541 (2014). (2020).
26 48
H. Jami and A. Jabbarzadeh, Appl. Surf. Sci. 489, 446 (2019). J. Lienhard, K. A. Nelson, and C. A. Schuh, Surf. Coat. Technol. 432, 128053
27
H. Jami and A. Jabbarzadeh, Surf. Coat. Technol. 394, 125880 (2020). (2022).
28 49
H. Jami and A. Jabbarzadeh, Appl. Surf. Sci. 542, 148567 (2021). M. Hassani-Gangaraj, D. Veysset, V. K. Champagne, K. A. Nelson, and
29
H. Jami and A. Jabbarzadeh, Surf. Coat. Technol. 382, 125173 (2020). C. A. Schuh, Acta Mater. 158, 430 (2018).
30 50
S. Plimpton, J. Comput. Phys. 117, 1 (1995). I. Dowding, M. Hassani, Y. Sun, D. Veysset, K. A. Nelson, and C. A. Schuha,
31
Y. Mishin, M. J. Mehl, D. A. Papaconstantopoulos, A. F. Voter, and Acta Mater. 194, 40 (2020).
51
J. D. Kress, Phys. Rev. B 63, 224106 (2001). J. Feng, P. Chen, Q. Zhou, K. Dai, E. An, and Y. Yuan, Int. J. Multiphys. 11,
32
A. Stukowski, Modell. Simul. Mater. Sci. Eng. 18, 015012 (2010). 315 (2017).
33 52
F. Shimizu, S. Ogata, and J. Li, Mater. Trans. 48, 2923 (2007). J. R. Feng, R. Liu, K. Y. Liu, Q. Zhou, R. J. Yang, and P. W. Chen, J. Appl.
34
X. Fan, D. Pan, and M. Li, J. Phys.: Condens. Matter 31, 095402 (2019). Phys. 131, 025903 (2022).
35 53
A. Stukowski, V. V. Bulatov, and A. Arsenlis, Modell. Simul. Mater. Sci. Eng. A. A. Tiamiyu, T. Lucas, E. L. Pang, X. Chen, J. M. LeBeau, and C. A. Schuh,
20, 085007 (2012). Mater. Today 72, 71 (2024).
36 54
A. A. Tiamiyu and C. A. Schuh, Surf. Coat. Technol. 403, 126386 (2020). A. A. Tiamiyu, E. L. Pang, X. Chen, J. M. LeBeau, K. A. Nelson, and
37
S. I. Imbriglio, M. Hassani-Gangaraj, D. Veysset, M. Aghasibeig, R. Gauvin, C. A. Schuh, Nat. Mater. 21, 786 (2022).
55
K. A. Nelson, C. A. Schuh, and R. R. Chromik, Surf. Coat. Technol. 361, 403 (2019). Z. Liu, H. Wang, M. J. Haché, X. Chu, E. Irissou, and Y. Zou, Acta Mater. 193,
38
V. Panova and C. A. Schuh, Acta Mater. 276, 120105 (2024). 191 (2020).
39 56
D. Goldbaum, R. R. Chromik, S. Yue, E. Irissou, and J. G. Legoux, J. Therm. Y. Feng, W. Li, C. Guo, M. Gong, and K. Yang, Mater. Sci. Eng. A 727, 119
Spray Technol. 20, 486 (2011). (2018).

J. Appl. Phys. 136, 105901 (2024); doi: 10.1063/5.0218416 136, 105901-12

© Author(s) 2024