Physica B: Condensed Matter 557 (2019) 132–140

Contents lists available at ScienceDirect

Physica B: Condensed Matter

journal homepage: www.elsevier.com/locate/physb

Characterization and fractal modeling of boron carbide nanoparticles T

synthesized by nanomilling
Mehrdad Khakbiza,∗, Farshad Akhlaghib, Pedram Sotoudeh Baghaa, Lida Ghazanfaric
a
Division of Biomedical Engineering, Department of Life Science Engineering, Faculty of New Sciences and Technologies, University of Tehran, North Karegar Ave., P.O.
Box 14395-1561, Tehran, Iran
b
School of Metallurgy and Materials Engineering, Center of Excellence for High Performance Materials, University of Tehran, P.O. Box 11155-4563, Tehran, Iran
c
Center for Nanotechnology in Drug Delivery, University of North Carolina, Chapel Hill, NC, USA

ARTICLE INFO ABSTRACT

Keywords: In this paper, the narrow size and spherical shape of boron carbide nanoparticles are synthesized by milling,
High energy ball milling purification and then the classification of nanoparticles by sedimentation method. In addition, the fractal be-
Nanomilling havior of particles is studied during synthesizing systematically and a fractal equation is proposed for prediction
Boron carbide nanoparticles of nanopowders size distribution plots. Milling of particles is conducted in a stirred attrition mill for different
Fractal behavior
times, and hydrochloric acid is used for solvent purification and then sedimentation process is conducted. The
Modeling
size of particles is measured by laser particle size (LPS) analyzer. For studying particles morphology, scanning
and transmission electron microscopy (SEM, TEM) are used. The crystal size of nanoparticles is calculated by X-
ray diffraction (XRD). LPS results reveal that after 110 h milling time, nanoparticles with an average diameter of
90 nm are produced. SEM images show that the final nanoparticles have equiaxed shape with iron impurity less
than 1.7 wt. % after acid washing. XRD results demonstrate that the crystalline size of boron carbide (B4C)
particles after 110 h of milling reduces from 220 to 57 nm. The fractal analysis shows that particles size dis-
tribution curves of finer particles can be predicted by proposed fractal modeling better in contrast to larger
particles.

1. Introduction
Nanoparticles indicate useful properties in electrical, optical and
chemical characteristics because of the high surface area. Ceramic na-
noparticles, which mostly consist of oxides, nitrides or carbides, have
attracted much attention. Carbide ceramics such as B4C has a high
thermal and variable electrical conductivity. Moreover, it is extremely
resistant to abrasion, corrosion, and high temperature. Boron carbide
particles could be used to develop advanced material like additive
manufacturing [1], nanoporous metal-based solids for heterogeneous
catalysis [2], material for lightening and photocatalyst [3,4], micro-
electromechanical systems (MEMS) [5], and biomedical [6].
There are different applications for B4C nanoparticles in medical
applications such as cancer treatment by boron neutron capture therapy
especially for oral squamous cell carcinoma [7,8]. Another example is
producing microlaminates for the elastomeric composite membrane to
reduce wear debris of polymeric biomaterial [9]. In orthopedic bearing
implants, high-performance Ultra High Molecular Weight Polyethylene
(UHMWPE) composite with B4C shows protection against X-ray, gamma
radiation [10].
Several methods for synthesizing of B4C nanoparticles investigated
including carbo-thermic reduction of boric acid [11], sol–gel process by
phenolic resin and boron alkoxide as precursors [12], magnesiothermic
reduction of boron oxide and porous carbons [13] and radio frequency
sputtering deposition [14].
High energy ball milling is a capable method for producing a wide
range of advanced materials such as nanocomposite and nanostructured
materials [15]. In addition, this method could produce different na-
noparticles from metallic to ceramic particles successfully [16]. Nano-
milling in stirred media mill is a promising wet grinding method for the
synthesizing of nanomaterials. Some of the main benefits of this method
in comparison to other methods include simple operation, low-cost
construction, high rate of size reduction, low contamination and
minimum particles agglomeration [17].
To the best of our knowledge, there is very limited research on
milling and production of B4C nanoparticles by stirred media mill. The

∗
Corresponding author. Division of Biomedical Engineering, Department of Life Science Engineering, Faculty of New Sciences and Technologies, University of
Tehran, Tehran , Iran.
E-mail address: khakbiz@ut.ac.ir (M. Khakbiz).

https://doi.org/10.1016/j.physb.2018.12.033
Received 29 September 2018; Received in revised form 11 November 2018; Accepted 19 December 2018
Available online 20 December 2018
0921-4526/ © 2018 Elsevier B.V. All rights reserved.
M. Khakbiz et al.
main reason is that boron carbide is the world's third-hardest materials
after diamond and cubic boron nitride which has high hardness and
high impact resistance [18]. Impurity is another issue for the produc-
tion of boron nanoparticle and milling and coating of milling media
cannot prevent this problem. In addition, agglomeration of nano-
particles is an obstacle for synthesizing nanoparticles by milling. The
wide range of synthesized (10–100 nm) nanoparticles with sharp edges
is another limitation of milling technique which restricts its application
for different industries.
On the other hand, the comminuting mechanism of nanoparticles
during milling is not well understood yet and very limited equation and
modeling were used to predict the girding behavior of nanoparticles.
Fractal modeling is a very useful method for studying of grinding of a
particle in the micrometric range [19], but, no study was conducted on
the application of this equations for predicting of nano grinding pro-
cess.
The optimization, characterization, and modeling of stirred media
milling for the preparation of nanoparticles with specific properties of
B4C seem to be lacking. The purpose of this paper is to synthesize the
B4C nanoparticles by nano milling, so a multi-step procedure was de-
veloped for synthesizing high purity and narrow range of B4C nano-
particles. The effect of process parameters and post-processing steps
was investigated on characteristics of nanoparticles such as their shape
and size. The grinding behavior was studied by fractal modeling and the
effect of process parameters was discussed on fractal parameters sys-
temically. The achievements of this work could be used for scalable and
high quantities production of B4C nanoparticles with specific char-
acteristics for different application such as biomedical application.
2. Experimental procedure
The commercial boron carbide powder used for the milling process
originally had an average particle size of 0.7 μm. Fig. 1 and Table 1
show the particle size distribution and chemical composition of the
original B4C powder, respectively. The milling was conducted in a
stirring attrition mill (Union Process Co. model S1) using a hardened
stainless steel vial and hardened steel balls with 6 mm in diameter
under argon atmosphere. Prior to the main process, boron carbide
powder was milled for 2 h to make a protective coating of B on the ball
surfaces and the container to reduce contamination. The ball-to-
powder-weight ratio was 20 and the ethanol was used as stirring media.
The milling process interrupted every 2 h to inhibit temperature in-
crease and 10 mL of ethanol added to the system. Laser particle size
analyzer (Cilas-1064) was used for measuring the size of original par-
ticles as well as their size during the milling procedure. Scanning
electron microscopy (SEM model CamScan MV2300) was used for
evaluation of particle morphology changes during milling. After milling
of the particles for 110 h, Hydrochloric acid (HCl) 2 M was used for
purification of particles. Energy dispersive X-ray analysis (EDAX,
Fig. 1. Particle size distribution of as-received B4C particles.
133
Physica B: Condensed Matter 557 (2019) 132–140
Table 1
The chemical composition of as-received B4C powder.
B C B2O3 Fe Si
Wt.% 76.10 20.30 0.73 1.96 0.91
QUANTAX EDS XFlash®, Bruker, U.S.A) was used for chemical analysis
of particle during different steps of the experimental procedure. Pur-
ified nanoparticles were poured into a measuring cylinder with a height
of 1 m for the sedimentation process. In this step, ethanol was used as
the liquid medium in the cylinder. Subsequent to drying of particles,
two forms of powder were selected from the top and bottom of the
cylinder. Transition electron microscopy (TEM) (Leo 912-AB) was used
for investigation of particle size and shape after the sedimentation
process. X-ray diffraction method (Philips PW-1730) was used for
measuring the crystal size of boron carbide nanoparticles.
Williamson-Hall method [20] was applied for calculating the crystal
size of boron carbide nanoparticle.
3. Results and discussion
3.1. Particle size
Fig. 2 shows the normal and cumulative particle size distribution
curve of B4C powders during ball milling. From Fig. 2 (a) it can be seen
that by increasing the milling time, the peak of plots moves towards
smaller particle size. In addition, by increasing the milling time the
graphs have narrower distribution, indicating that a homogeneous
distribution of particles could be produced by milling. Narrow particle
size distribution is results of inclination of fine particles to weld to-
gether and large particles to fracture under stable conditions [16].
Fig. 2 (b) shows that by milling for different times, cumulative plots
move to a finer size. It can be seen that by milling of powder for 110 h,
almost all particles are reduced in size to the nanometric range. It can
be concluded that the milling conditions in this experiment are able to
pulverize the B4C particles even though the hardness of the ball is lower
than that of B4C particles. Fig. 3 shows the effect of milling time on the
variation of the median size (D50) of particles [18]. It demonstrates that
by increasing milling time the finer particles were obtained, however,
the reduction rate of particle size decreased gradually. As can be seen,
the most particle fragmentation accrued in the first 32 h milling time
and particles reached steady state stage in the next 78 h of milling. In
this stage, the particle's size reaches a limit which further fracturing
does not occur. It was suggested that this limit could be the result of
predomination of plastically deformation behavior of fine particles ra-
ther than fragmentation. These results confirm that excessive milling
more than 110 h does not decrease particle size and longer grinding
times does not seem to be helpful for the production of nanoparticles.
As shown in Fig. 4 (a), in the first 12 h milling time the breakage
rate is 23 (nm/h) but for the next 98 h milling time, this rate reached to
3.5 (nm/h). There are three main breakage mechanisms in comminu-
tion [21]:
• Abrasion that is caused by shear stress and particles with a size close
to that of the initial particles are produced.
• Cleavage that occurs when intense compression stresses are slowly
applied to a particle so fragments sizes are smaller than that of in-
itial particles.
• Fracture is another mechanism that is the result of rapidly impact
stress and fragments distribution are between 20 and 70 vol % of the
initial particle size. Propagation of defects such as cracks results in a
fracture. Propagation cracks depend on the size of the flaw, its po-
sition in the particle and its orientation with respect to the loading
axis, but the average particle diameter (D) is influenced by fracture
toughness (KIC) (Eq. (1)) [22]:
M. Khakbiz et al. Physica B: Condensed Matter 557 (2019) 132–140

(a)80 2h 6h 12h
150
(b) 2h 6h 12h
36h 44h 64h
36h 44h 64h
80h 110h 125 80h 110h
60

Cumulative (%)
100

Volume (%)
40 75

50
20
25

0 0
0.01 0.04 0.16 0.64 2.56 0.01 0.04 0.16 0.64 2.56
Diameter (µm) Diameter (µm)
Fig. 2. Particle size distribution curves of B4C powders during ball milling, (a) normal particle size distribution and (b) cumulative particle size distribution.

800 Φ=(2a/D) < 0.5, k is constant and normally 0.25 to 0.3 and p is the
compressive stress generated by load P and m is a multiplying
700 factor > 1. Results from Fig. 4 (a) shows that particle breakage was
higher at initial milling time, but this rate decreased dramatically. This
phenomenon is attributed to different factors:
600
(i) By increasing the milling time, the finer particles are produced
which are sustainable for agglomeration which reduces the breakage
500 rate. Moreover, the ethanol is evaporated during wet milling and its
performance is reduced significantly so this type of agglomeration
D50 (nm)

400 could be related to local drilling and dry milling at the end of each
milling step. (ii) The viscosity of the slurry during milling increased
300 with increasing particle volume, which prevents further breakage of
particles. (iii) By increasing the particle number in milling media, the
efficiency of the mill decreased, because the particles obtained lower
200 energy or stress (σp) from the mill solely and according to Eq. (1) the
obtained particles with finer size reduced. (iv) The surface of the
100 grinding medium will be coated by particles which affect the surface
roughness of the ball and results in lower impact energy. (v) During the
0 milling process, the ball sizes changed, hence the compressive and
0 20 40 60 80 100 120 shearing forces/stresses were reduced during impact and sliding of
smaller balls. (vi) The finer particles have a lower chance of cracks and
Milling time (h) defaults in the volume of the particles. So finer particles breakage more
difficulty. So finer particles have lower K1C and according to Eq. (1), the
Fig. 3. Effect of milling time on average particle size (D50) [18].
milling process produces larger particles.
The factors that affect crushing powders are powder shape, the re-
KIC 2 sistance of particle to breakage such as their hardness and toughness,
D= ( )m
2k p (1) adhesion characterization of the particle which results in agglomer-
where, Φ is fraction of D (if we assume 2a as the maximum flaw size it ating, the viscosity of the slurry, the roughness of milling media and
must be less than the particle diameter, 2a < D, it was assumed that ability of the powders to be trapped among the milling balls [23]. In the

(a) 50 (b) 3
R² = 0.7535 Span D90-D10
2.5
40
D90-D10 (nm) , SPAN
Breakage rate ( nm/h)

2
30
1.5
20
1

10
0.5

0 0
2 6 12 32 36 40 44 48 56 64 80 110 0 20 40 60 80 100 120
Milling time (h) Milling time (h)
Fig. 4. Effect of milling time on (a) breakage rate of particles and (b) D90-D10 and SPAN.

134
M. Khakbiz et al.
final stage of milling, the limitation of milling is related to equilibrium
between agglomeration and breakage rate. For example, it has been
demonstrated that the harder powder is more prone to grinding than
softer particles [24]. Also, a transition in ultra-fine particles may be
obtained at high compressive stress [25]. This phenomenon occurs
when high induced tensile components (kσp) remain insufficient to
produce particle fracture, whereas the shear stress components (σp/2)
become sufficient to promote dislocation motion in the particle (plastic
deformation). This term consumes a huge amount of milling energy via
plastic deformation with minimum powder fracture.
Fig. 4 (b) shows the D90-D10 and SPAN ((D90-D10)/(D50)) index
versus different milling times. Higher (D90-D10) content results in more
different sizes of milled powder. Therefore, it can be seen that more
homogenous size of the particle was achieved by increasing milling
time. The values (D90-D10) sharply decreases during the first hours of
milling and then tends towards a limit size. It can be concluded that the
effect of milling on production homogenization distribution is more
pronounced in the first milling time. However, the SPAN index does not
show a constant behavior which is attributed to predomination of dif-
ferent fracturing or agglomeration mechanisms during milling.
3.2. Particles morphology
Fig. 5 shows scanning electron microscopy (SEM) of boron carbide
during milling. It can be seen in Fig. 5, the as-received powders are plate-
like in a broad range of sizes. After 10 h milling (Fig. 5 (b)), the content
of fine particles increased which could be due to the abrasion of large
particles. However, the large particles in this step appear to be still
platelike with smoother surfaces. Fig. 5 (c), which is milled powder after
20 h, shows that the product consists of particles with a wide range of
sizes and two main types of shapes. On the one hand, there are large
faceted particles with a size range of 1–500 μm, whereas, on the other
hand, there are relatively smaller particles with a size range of
Fig. 5. SEM images of boron carbide during different milling time, (a) as-re-
ceived, (b) 10 h, (c) 20 h, (d) 40 h, (e) 80 h, and (f) 110 h.
135
Physica B: Condensed Matter 557 (2019) 132–140
Fig. 6. SEM images of (a) B4C agglomerated particles and (b) Cold welded B4C
nanoparticle.
100–200 nm; and having an overall spherical shape. After milling for
40 h (Fig. 5 (d)), the sharp edges of both of the particles disappeared and
the faceted particles became more spherical with 400 nm particle size. A
comparison of the SEM images taken from powder milled for 40 h and
20 h show an evident increase in the proportion of the finer particles in
the range of 200–400 nm with increasing milling from 20 h to 40 h.
Further milling of powders up to 80 h (Fig. 5 (e)) resulted in more uni-
form particle size with approximately 200 nm resulting from the reduc-
tion of the larger particles due to further attrition with milling time. The
presence of agglomeration of the smaller particles into bigger clusters
could be seen at this stage. In the final stage of milling for 110 h, the
produced particles were 90 nm in size with a narrower size distribution.
In general, as the milling time increases, the smaller but also uniform and
spherical shape and severely agglomerated particles are produced in the
media, as shown in Fig. 6 (a). Fig. 6 (b) shows a typical SEM image of
agglomerated rounder particles appearing as clusters and some welded
particle with the equiaxed shape. This image is evidence that brittle
particles in their limitation fragmentation size act like malleable powder
with plastic deformation behavior. These micrographs demonstrate the
interconnection of fine agglomerates. The periodic fracture and coales-
cence events continue throughout the milling process. In the long milling
process, the particle cluster's size becomes larger than their initial size
which is due to agglomeration. These powders like to agglomerate very
easily, mainly due to the large specific surface area.
3.3. Classification by sedimentation
After 110 h milling in the wet media mill, the slurry was dispersed
in an ethanol solvent and ultrasonic vibration was used for homo-
geneous dispersion of nanoparticles. The sedimentation in a large tube,
which was controlled by the time of the process, was used in the final
step. To collect the small size nanoparticles, slurry settled down for 5 h.
The large particles sediment gradually, while smaller particles were still
floating in the solution. The ethanol was removed slowly and the
smallest floating nanoparticles in the top section of the product were
collected. The aforementioned process was repeated on the remaining
part of the particles.
Sedimentation is based on Stokes' law, according to which the set-
tling velocity of particles in a viscous medium under the gravity force
depends on particle size and the viscosity of the medium. Stokes' law,
however, is only valid for particles of solely spherical shape, while the
metal powders used in industry consist of particles of the most varied
shapes [22]. The Stokes' equation can be seen in Eq. (2) [26]:
gd 2 ( p f)
V=
18 (2)
In which V (m/s) is the velocity, d (m) is particle size, g is gravity (m/
s2), η is the viscosity (kg/m2) of the fluid, ρp, and ρf (kg/m3) are the density
of particles and fluid, respectively. Based on this equation, coarser parti-
cles could settle rapidly in the bottom of the cylinder compared to finer
particles. Fig. 7 shows the TEM micrograph of B4C nanoparticles for dif-
ferent process steps. Similar analysis using TEM confirms the characteristic
M. Khakbiz et al. Physica B: Condensed Matter 557 (2019) 132–140

(a)

Fig. 7. TEM images of B4C nanoparticles after (a) 110 milled, (b) after sedimentation, (c) after purification and (d) after sedimentation and purification.

morphology of the rounded agglomerated produced powder after 110 h

milling as described for the SEM study (Fig. 7 (a)). After the sedimentation
process (Fig. 7 (b)), a remarkable decrease in the proportion of the smaller
particles below 50 nm could be seen, however the number of agglom-
eration increases. Fig. 7 (c) is a micrograph of the B4C only after HCl
purification and it has a similar shape to Fig. 7 (a), but unlike the initial
powder, it has a larger proportion of finer rounded particles; however, the
large particles appear to be more faceted with smooth surfaces. Fig. 7 (d)
shows the milled particles after purification and sedimentation. The
spherical particles finer than 50 nm with the less agglomerated state could
be seen at this stage. In all cases, less dark particles could be seen in Fig. 7
(c) and (d) which could be attributed to a smaller the amount of iron or
iron oxide compound after purification.
A closer look at single particles by TEM is shown in Fig. 8. It can be
seen that the color of the inner part of a particle is different from the
outer surface of the same particle and it seems that this particle is
covered by layers of another phase, which could be Fe compound. The
core-shell structure of B4C particles with Fe contamination has been
reported by other researchers [27].
Fig. 9 shows the cumulative and volume percent particle size dis-
tribution curves of after different milling, sedimentation and purifica-
tion processes which were obtained by image analysis of TEM micro-
Fig. 8. TEM image of single B4C nanoparticle after sedimentation.
graph using ImageJ software. This graph demonstrates that by the
sedimentation process the cumulative and normal distribution of par-
ticle sizes shift towards smaller particles. Furthermore, the particle size 3.4. Chemical analysis
in both cases (particles size < 50 nm and particles size < 100 nm) de-
creases after purification which could be attributed to the elimination Fig. 10 (a) shows the chemical point analysis results of powder after
of Fe layer around the powders. These results are in agreement with 110 h milling times. It could be seen that by increasing the milling time
data obtained in Fig. 8. the iron peak appears clearly. Fig. 10 (b) shows the chemical point

136
M. Khakbiz et al. Physica B: Condensed Matter 557 (2019) 132–140

(a)
(a)

Diameter (nm) Diameter (nm)

30 45
(c) 110h milling/ purification (d) 110h milling/Sedimentation/ Purification
40
25 110 milling /Sedimentation
110h milling
35
20
Volume (%)
30
Volume (%)

15 25

20
10
15
5
10

0 5
27 40 46 63 72 85 104 115 132 147 17 21 22 27 31 35 45 55 56 60
Diameter (nm) Diameter (nm)
Fig. 9. Cumulative and volume percent of particles size distribution of B4C nanoparticles at different steps of the milling and post processes, respectively (a) and (c)
milling and purification (b) and (d) milling, sedimentation, and purification.

analysis of powders after acid washing. It can be seen that after acid
washing the impurity of powder reduced considerably. For example,
the Fe content reached 1.82 wt. % which is very close to the as-received
boron carbide. Fig. 10 (c) shows chemical point analysis of boron car-
bide nanoparticles after acid washing and sedimentation procedure. In
this situation, boron carbide nanoparticles have Fe impurity with
1.32 wt. % which is less than that of the as-received particles.
Fig. 11 (a) shows attached boron carbide on the top layer of the
Fig. 10. Chemical point analysis of boron carbide (a) milled for 110 h, (b)
milled for 110 h and purification and (c) 110 h milled, acid washed and after
sedimentation procedure.
milling ball that could be a good reason for reducing the rate of con-
tamination. Fig. 11 (b) shows the fracture surface of the single milling
ball in which result in considerable observed iron impurity in the
powder.
3.5. X-ray Analysis
Fig. 12 shows the XRD pattern of the as-received, milled and milled
then purified boron carbide powder. In the as-received B4C particles,
only sharp characteristic peaks of different planes of boron carbide
were observed (JCPDS 6-0555). It can be seen that after milling for
110 h, only the Fe2B crystal plane is clearly seen (JCPDS 39-1314) and
B4C planes are invisible. It seems boron carbide nanoparticles planes
were below the detection limit of the XRD instrument and their peaks
disappear in the background of the XRD pattern. After milling and
purification, B4C crystal planes appear significantly and no diffraction
peaks from iron or iron oxides were found in this pattern. The re-
storation of the B4C peaks indicates that most of the iron compounds
coated on the surface of the B4C particles were dissolved after acid
washing. By comparison of original boron carbide XRD pattern and
milled/purified particles, it can be concluded that both samples have a
similar rhombohedral crystal structure. However, the sharp peaks of the
original materials degraded significantly and the peaks broadened after
milling. The refinement of crystalline along with decreasing particle
size of milled powder results in the broadening of the XRD diffraction
peaks. Also, stacking faults introduced by cold working caused anom-
alous hkl-dependent peak broadening in the X-ray diffraction patterns

137
M. Khakbiz et al. Physica B: Condensed Matter 557 (2019) 132–140

(a) (b)

Fig. 11. SEM image of (a) attached B4C particles on the ball (dark particles) and (b) the fracture surface of the single milling ball.

0.011

0.01
y = 0.0119x + 0.0028
0.009
R² = 0.885

0.008 (b)
s.Cos( )

0.007

0.006

0.005

0.004

Sin ( )
Fig. 12. XRD pattern of (a) as received B4C particles, (b) milled for 110 h and
Fig. 13. Crystallite size estimation from XRD results for milled and purified B4C
(c) milled and purified.
nanoparticles.

[16]. Fig. 13 shows βs. Cos (θ) versus Sin (θ) in a linear graph with a
Fig. 14 shows fractal model, typical plot of log (Mr/Mt) versus log
correlation factor of 0.88. The boron carbide nanocrystal size of 57 nm
(r/M(, for the milled particles. It can be seen that a linear Eq. (4) can be
was measured by the Williamson-Hall equation [20]. This size is dif-
presented as follow:
ferent from the TEM results which is attributed to an instrumental error
of XRD analysis. log(
Mr
)= × log(
r
)+
Mt RM (4)
3.6. Fractal modeling of nanoparticles during nano milling The power– law version of Eq. (4) is:

For studying of particles breakage mechanisms during milling, M r

( r ) = 10 × ( )
fractal equation (Eq. (2)) with the power-law relationship can be used Mt RM (5)
[28]: where, δ is y-intercept and ε is slop of curves. By comparing Eq. (4) by
Mr r 3 Eq. (3), it can be seen that γ = 3-ε and, γ can be calculated from ob-
=( ) = F (r )
Mt RM (2) tained slop of plots for all samples (Fig. 14). Fig. 15 (a) shows that by
increasing the milling time the γ increases within initial milling time
where, Mr is the total mass of all particles with the size less than r, Mt is and reaches to a constant value over time. These results are in agree-
the total mass of the system, so (Mr/Mt) can be considered as the mass ment with Fig. 4 which shows decreasing breakage rate with increasing
cumulative frequency and equal to cumulative frequency of volume milling time. The effect of milling time on fractal parameters is shown
(Fr), RM is the largest size of the particle in the system, and γ is the in Fig. 15 (a). It also can be seen that by increasing the milling time, δ
fractal dimension. Eq. (3) can be obtained by double logarithmic of Eq. increased and Rm decreased, gradually. Based on the curve fitting re-
(2) in which two part of this equation has a good linear relationship. sults, the relationship between this parameter and milling time can be
Mr r calculated as follow (Eqs. (6)–(8)):
log( ) = (3 ) × log( )
Mt RM (3) = 0.082 ln(t ) + 1.3141 (6)

138
M. Khakbiz et al. Physica B: Condensed Matter 557 (2019) 132–140

2.5 Fr = 100.0041t + 2.2314 × (

r
) 0.082 ln(t ) + 1.3141
0.0105t + 3.0818 (9)

2 With Eq. (9), it is possible to plot curves of Fr versus particles size in

any milling time. Fig. 15 (b) shows a typical comparison between ex-
perimental and calculated Fr for powder milled for 2 h, 64 h, and 110 h.
1.5
These plots show there is good agreement between experimental and
predicted results, especially in finer particles size. Furthermore, Fig. 15
Log (Mr/Mt)

1 (b) shows that by increasing the milling time, predicted cumulative

distribution shifts towards finer particles size. Fig. 15 (c) shows the plot
of experimental value versus predicted data, obtained from Eq. (9) for
0.5
all samples. Results show that the regression coefficient is more than
2h 6h 12h 0.92 which demonstrate the good accuracy of the proposed equation in
0 the nano milling system of boron carbide nanoparticles. In addition, it
36h 44h 64h can be seen that the prediction validity increased for finer particles
which are in spite of agglomeration problem of nanoparticles of na-
-0.5 80h 110h nometric size. The plots of log (Mr/Mt) versus log (r/RM) for 110 h
milled boron carbide particles and then purification, sedimentation is
-1 obtained by using experimental particles size distribution curves (Fig. 9
1.25 0.25 -0.75 -1.75 -2.75 (a)) and is shown in Fig. 15 (d). γ for purified, sedimented and sedi-
Log (ro/Rm) mented/purified particles of 110 h milled are 1.1, 1.27 and 1.47, re-
spectively. The regression coefficient is more than 0.99 for all samples
Fig. 14. Comparison of the fractal model for the milled particles. that are higher than the R = 0.92 value obtained for particles milled for
110 h. These results show that the accuracy of fractal modeling for
= 0.0041t + 2.2314 (7) particles after purification and sedimentation (particles finer than
50 nm) is higher than particles finer than 100 nm. This is because of the
RM = 0.0105t + 3.0818 (8) more homogeneous distribution of particles obtained after leaching and
sedimentation process. Liu et al. [26] showed that the fractal model is
By substituting these parameters in Eq. (5), the relationship between useful to explain or estimate the particle size distribution of milled
the cumulative frequency of volume and milling time can be obtained TiH2, too. It was also used to evaluate particle size distribution of many
by Eq. (9): minerals [29,30] or multi-wall carbon nanotubes [31] which showed
the accuracy of modeling by this method.

(a) (b)
2h-Exp.
2h-Predict.
4 D m K Rm 80
64h-Exp.
64h-Predict
Cumulative (%)
Fractal parameter

y = -0.0105x + 3.0818 60 110h-Exp.

3
110h-Predict.
40
2 y = 0.0041x + 2.2314
20
1
y = -0.082ln(x) + 1.3141
0
0
0 20 40 60 80 100 120
Milling time (h) Particles Size (um)

(c) (d)
y = 1.7276x + 2.8723
120 R² = 0.9239 R² = 0.949 y = 1.9084x + 2.7475
2 R² = 0.9919
Experimental Data (Fr)

100
1.7
Log (Mr/Mt)

80

60 1.4
y = 1.5342x + 2.0731
40 1.1 R² = 0.9956

20 0.8 Milled-Sedi.-Purifiy
Milled-Sedi.
0 Milled-Purifiy
0.5
0 20 40 60 80 100 0 -0.2 -0.4 -0.6 -0.8 -1
predicted Data (Fr)
Log (ro/Rm)

Fig. 15. (a) Effect of milling time on fractal parameters, (b) experimental and predicted particle size distribution curves, (c) comparison of the fractal model for
experimental and predicted data and (d) fractal model for purified, sediment and sediment/purified particles of 110 h milled.

139
M. Khakbiz et al.
Here, we studied the milling, purification, and sedimentation of B4C
to obtain a model for further investigations. For biomedical applica-
tions, it is necessary to reduce the impurities. Indeed, using harder balls
such as WC-Co or Si3N4 with superior hardness and wear resistance
could be reducing the milling time, contamination, lower changing in
ball sizes which can affect the SPAN index.
4. Conclusion
Nano-sized B4C particles were successfully produced by nano mil-
ling of commercially available (0.7 μm) boron carbide (B4C) in a stirred
media mill. At the initial stages of milling (i.e. for the first 32 h of
milling), a sharp decrease in the particle size occurred, but this rate
decreased gradually and reached a minimum size after 110 h milling.
These results are attributed to the re-aggregation and breakage events
that take place simultaneously during milling and swapping, mostly
during milling.
Moreover, the scanning electron microscopy images revealed that
the original powders are platelike whereas the milled particles for 110 h
have equiaxed morphology. The smallest nanoparticle size (50 nm)
could be achieved using a by comminution and sedimentation process.
Furthermore, the purified nanoparticles with low Fe impurity were
purified by acid washing. The crystal size calculated from XRD results
showed that the average crystallite size of B4C was 57 nm after 110 h
milling. Results obtained in this work indicate that fractal fragmenta-
tion theory allows for quantifying fragmentation processes effectively
during comminution. The proposed equation could predict particles
size distribution of boron carbide nanoparticles with high accuracy. The
synthesized B4C nanoparticles with the high surface area and rounded
shape have potential applications in biomedical and bioscience areas.
Acknowledgment
Financial support from Iranian Nanotechnology initiative (INI) is
gratefully acknowledged.
References
[1] A. Olsson, M.S. Hellsing, A.R. Rennie, New possibilities using additive manu-
facturing with materials that are difficult to process and with complex structures,
Phys. Scripta 92 (2017), https://doi.org/10.1088/1402-4896/aa694e 053002.
[2] W. Luc, F. Jiao, Synthesis of nanoporous metals, oxides, carbides, and sulfides:
beyond nanocasting, Acc. Chem. Res. 49 (2016) 1351–1358, https://doi.org/10.
1021/acs.accounts.6b00109.
[3] B. Singh, G. Kaur, P. Singh, K. Singh, J. Sharma, M. Kumar, et al., Nanostructured
BN–TiO 2 composite with ultra-high photocatalytic activity, New J. Chem. 41
(2017) 11640–11646, https://doi.org/10.1039/C7NJ02509B.
[4] P. Singh, K. Singh, M. Kaur, H. Kaur, B. Singh, G. Kaur, et al., Preferentially grown
nanostructured MgB 2 C 2: a new material for lightening applications, Superlattice.
Microst. 103 (2017) 1–8, https://doi.org/10.1016/j.spmi.2017.01.013.
[5] M. Mahmoodi, L. Ghazanfari, Silicon Carbide: a Biocompatible semiconductor used
in advanced biosensors and bioMEMS/NEMS, Phys. Technol. Silicon Carbide
Devices, InTech, 2012, , https://doi.org/10.5772/51811.
[6] A.K. Suri, C. Subramanian, J.K. Sonber, T.S.R.C. Murthy, Synthesis and consolida-
tion of boron carbide: a review, Int. Mater. Rev. 55 (2010) 4–40, https://doi.org/
10.1179/095066009X12506721665211.
[7] T.I. Yoshie Ishikawa, Boron carbide particle as a boron compound for boron neutron
capture therapy, J. Nucl. Med. Radiat. Ther. 05 (2014), https://doi.org/10.4172/
2155-9619.1000177.
140
Physica B: Condensed Matter 557 (2019) 132–140
[8] B. Singh, G. Kaur, P. Singh, K. Singh, B. Kumar, A. Vij, et al., Nanostructured boron
nitride with high Water dispersibility for boron neutron capture therapy, Sci. Rep. 6
(2016) 35535, https://doi.org/10.1038/srep35535.
[9] Teoh Swee-hin, Engineering Materials for Biomedical Applications, World
Scientific, 2004.
[10] Steven M. Kurtz, UHMWPE Biomaterials Handbook: Ultra High Molecular Weight
Polyethylene in Total Joint Replacement and Medical Devices, third ed., William
Andrew, 2015.
[11] P. Singh, B. Singh, M. Kumar, A. Kumar, One step reduction of boric acid to boron
carbide nanoparticles, Ceram. Int. 40 (2014) 15331–15334, https://doi.org/10.
1016/j.ceramint.2014.06.101.
[12] A. Najafi, F. Golestani-Fard, H.R. Rezaie, N. Ehsani, A novel route to obtain B4C
nano powder via sol–gel method, Ceram. Int. 38 (2012) 3583–3589, https://doi.
org/10.1016/j.ceramint.2011.12.074.
[13] P. Asgarian, A. Nourbakhsh, P. Amin, R. Ebrahimi-Kahrizsangi, K.J.D. MacKenzie,
The effect of different sources of porous carbon on the synthesis of nanostructured
boron carbide by magnesiothermic reduction, Ceram. Int. 40 (2014) 16399–16408,
https://doi.org/10.1016/j.ceramint.2014.07.147.
[14] E. Pascual, E. Martı́nez, J. Esteve, A. Lousa, Boron carbide thin films deposited by
tuned-substrate RF magnetron sputtering, Diam. Relat. Mater. 8 (1999) 402–405,
https://doi.org/10.1016/S0925-9635(98)00274-X.
[15] M. Omidi, A. Khodabandeh, S. Nategh, M. Khakbiz, Wear mechanisms maps of CNT
reinforced Al6061 nanocomposites treated by cryomilling and mechanical milling,
Tribol. Int. 110 (2017) 151–160, https://doi.org/10.1016/j.triboint.2017.01.033.
[16] C. Suryanarayana, Mechanical alloying and milling, Prog. Mater. Sci. 46 (2001)
1–184, https://doi.org/10.1016/S0079-6425(99)00010-9.
[17] F. Stenger, S. Mende, J. Schwedes, W. Peukert, Nanomilling in stirred media mills,
Chem. Eng. Sci. 60 (2005) 4557–4565, https://doi.org/10.1016/j.ces.2005.02.057.
[18] M. Khakbiz, F. Akhlaghi, Synthesis and structural characterization of Al–B4C na-
no–composite powders by mechanical alloying, J. Alloys Compd. 479 (2009)
334–341 https://doi.org/10.1016/j.jallcom.2008.12.076.
[19] E. Perfect, Fractal models for the fragmentation of rocks and soils: a review, Eng.
Geol. 48 (1997) 185–198, https://doi.org/10.1016/S0013-7952(97)00040-9.
[20] G. Williamson, W. Hall, X-ray line broadening from filed aluminium and wolfram,
Acta Metall. 1 (1953) 22–31, https://doi.org/10.1016/0001-6160(53)90006-6.
[21] S.L. a. Hennart, W.J. Wildeboer, P. van Hee, G.M.H. Meesters, Identification of the
grinding mechanisms and their origin in a stirred ball mill using population bal-
ances, Chem. Eng. Sci. 64 (2009) 4123–4130, https://doi.org/10.1016/j.ces.2009.
06.031.
[22] D. Tromans, J. Meech, Fracture toughness and surface energies of minerals: theo-
retical estimates for oxides, sulphides, silicates and halides, Miner. Eng. 15 (2002)
1027–1041, https://doi.org/10.1016/S0892-6875(02)00213-3.
[23] C.C. Harris, On the role of energy in comminution: a review of physical and
mathematical principles, Miner. Process. Extr. Metall. (IMM Trans. Sect. C) 75
(1966) 37–56.
[24] J.B. Fogagnolo, F. Velasco, M.H. Robert, J.M. Torralba, Effect of mechanical al-
loying on the morphology, microstructure and properties of aluminium matrix
composite powders, Mater. Sci. Eng. 342 (2003) 131–143 https://doi.org/10.1016/
S0921-5093(02)00246-0.
[25] D. Tromans, J. a. Meech, Fracture toughness and surface energies of covalent mi-
nerals: theoretical estimates, Miner. Eng. 17 (2004) 1–15, https://doi.org/10.1016/
j.mineng.2003.09.006.
[26] R.K. Rajput, A Textbook of Fluid Mechanics, S. Chand Limited, 2008, http://books.
google.com/books?id=BNgb-5S7hVkC.
[27] W.A. Kaczmarek, A. Calka, B.W. Ninham, Evaluation of Fe contamination in ball
milling of nonmagnetic materials by VSM, Phys. Status Solidi 141 (1994)
K123–K126, https://doi.org/10.1002/pssa.2211410241.
[28] S.W. Tyler, S.W. Wheatcraft, Fractal scaling of soil particle-size distributions: ana-
lysis and limitations, Soil Sci. Soc. Am. J. 56 (1992) 362, https://doi.org/10.2136/
sssaj1992.03615995005600020005x.
[29] R. Zuo, Q. Cheng, Q. Xia, F.P. Agterberg, Application of fractal models to distin-
guish between different mineral phases, Math. Geosci. 41 (2009) 71–80, https://
doi.org/10.1007/s11004-008-9191-3.
[30] A. Taşdemir, Fractal evaluation of particle size distributions of chromites in dif-
ferent comminution environments, Miner. Eng. 22 (2009) 156–167, https://doi.
org/10.1016/j.mineng.2008.06.002.
[31] Á. Kukovecz, T. Kanyó, Z. Kónya, I. Kiricsi, Long-time low-impact ball milling of
multi-wall carbon nanotubes, Carbon N. Y 43 (2005) 994–1000, https://doi.org/10.
1016/j.carbon.2004.11.030.