Journal of Alloys and Compounds 749 (2018) 85e102

Contents lists available at ScienceDirect

Journal of Alloys and Compounds

journal homepage: http://www.elsevier.com/locate/jalcom

Optimization of supported bimetallic (Fe-Co/CaCO3) catalyst synthesis

parameters for carbon nanotubes growth using factorial experimental
design
M.T. Bankole a, e, I.A. Mohammed b, A.S. Abdulkareem b, e, J.O. Tijani a, e, *, S.S. Ochigbo e,
O.K. Abubakre d, A.S. Afolabi c
a
Chemistry Department, Federal University of Technology, PMB 65, Minna, Nigeria
b
Chemical Engineering Department, Federal University of Technology, PMB 65, Minna, Nigeria
c
Department of Chemical, Metallurgical and Materials Engineering, Botswana International University of Science and Technology (BIUST), Plot 10071,
Buseja Ward, Palapye, Botswana
d
Mechanical Engineering Department, Federal University of Technology, PMB. 65, Minna, Nigeria
e
Nanotechnology Group, Centre for Genetic Engineering and Biotechnology (CGEB), Federal University of Technology, PMB 65, Minna, Nigeria

a r t i c l e i n f o a b s t r a c t

Article history: The FeeCo bimetallic catalyst supported on CaCO3 was prepared by a wet impregnation method. The
Received 29 September 2017 interactive effects among the key synthesis parameters such as drying time, calcination temperature,
Received in revised form stirring speed, mass of CaCO3 support on the yield and quality properties of the catalyst were investi-
27 February 2018
gated using 24 Factorial design of experiment. Additionally, the catalyst obtained under the optimal
Accepted 12 March 2018
Available online 21 March 2018
conditions was used to prepare Multi-walled carbon nanotubes (MWCNT) via chemical vapour deposi-
tion (CVD) of acetylene gas in a tubular horizontal reactor. The quality of the prepared materials was
examined for their morphology, microstructure, elemental composition, surface chemical properties,
Keywords:
Optimization
thermal stability, surface area, and mineralogical phase by HRSEM, HRTEM, EDS, TGA, FTIR, BET and XRD.
Factorial design The results revealed that maximum catalyst yield of 99.6% obtained at the drying time of 12 h, calcination
Bimetallic (Fe-Co) catalyst temperature of 100
C, stirring speed of 1000 rpm and mass of CaCO3 support of 10 g favoured the for-
Calcium carbonate mation of MWCNT at 700
C and 60 min reaction time. The results of statistical analysis demonstrated a
Carbon nanotubes direct relationship and synergetic effect between stirring speed and mass of support and the two factors
exerted highest impact on the yield of catalyst than other parameters. Microscopy analysis revealed
successful dispersion of Fe and Co particles onto CaCO3 support while XRD patterns confirmed the
presence of highly active crystalline mixed oxide (CoFe2O4) as the dominant phase. The HRSEM/XRD
analysis displayed the formation of tubular networks of graphitic carbon materials with few traces of by-
products. The study established the absorption and not diffusion mechanism of Fe and Co onto CaCO3 in
the reaction sequence produced CoFe2O4 during the wet impregnation process employed for catalyst
preparation.
© 2018 Published by Elsevier B.V.

1. Introduction electrical conductivity [1,2]. Different synthetic methods namely;

In the last couple of years, the number of publications and
patents on carbon nanotubes (CNTs) and its technological appli-
cations have increased rapidly due to its exceptional properties
such as high mechanical strength, capillary properties and
* Corresponding author. Chemistry Department, Federal University of Technol-
ogy, PMB 65, Minna, Nigeria.
E-mail address: jimohtijani@futminna.edu.ng (J.O. Tijani).
arc discharge laser ablation [3], electrolysis, catalytic chemical
vapour deposition (CCVD), plasma-enhanced, thermal or catalytic
have been employed to synthesis CNTs [4]. While methods such as
arc discharge laser ablation, electrolysis have certain shortcomings,
catalytic chemical vapour deposition (CCVD) stands out as an ideal,
simple and versatile method of obtaining high quality and large
quantity CNTs at low temperature and pressure [1e3]. Additionally,
catalysts preparation especially with supports for carbon nano-
tubes production have attracted attention among researchers due
to the impact of support materials on the properties and activity of

https://doi.org/10.1016/j.jallcom.2018.03.150
0925-8388/© 2018 Published by Elsevier B.V.
86 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102
the catalyst mixture [4]. Transition metals namely; Fe, Co, Ni, Mo
and their combinations on support materials such as SiO2, Al2O3,
MgO, zeolite, CaCO3, TiO2, kaolin amongst others in form of catalyst
have been used to produce carbon nanotubes of different
morphology and diameter size [5,6]. These supports were selected
due to their high surface area, high thermal stability and low cost,
ease of dispersion accompanied by electronic interactions and
migration of metallic particles. On the contrary, there is difficulty in
the removal of these supports from CNTs after synthesis due to
their insolubility in mineral acids such as HCl, HNO3 or H2SO4. It is
noteworthy to mention that, Fe, Co and Ni are mostly preferred
transition metals due to high solubility and diffusion of carbon onto
metallic component at higher temperatures which in turn
enhanced the yield of CNT. Additionally, Fe, Co and Ni particles
produced most active and effective phase required for CNTs growth
compared to other transition metals [11].
Several supported bimetallic catalysts have been produced
however Fe-Co/CaCO3 is superior in terms of preparation and effi-
ciency relative to other types of metal-support formulations
[7e12]. CaCO3 is preferable to other supports due to the following
uniqueness namely; non-porosity, abundance nature, readily
available and favoured suppression of amorphous carbon during
CNT growth [12]. However, there is disparity of opinion as regards
the exact quantity of active phase of Fe or Co that must formed the
Fe-Co catalyst which directly determines the CNT yields and qual-
ity. Although, bimetallic catalysts such as Fe-Co supported on
CaCO3 produced high quality and yields of MWCNT than other
catalyst supported on different materials [13]. This is because
CaCO3 used during CNTs growth is cheap and ease to remove from
the as-synthesized MWCNT via mild acid treatment. The higher
CNTs yield from bimetallic Fe-Co catalyst supported on CaCO3
relative to others was ascribed to stronger metal-support interac-
tion, smaller particles size and less particles aggregation [14].
In view of the aforementioned exceptional qualities of CaCO3
relative to other supports, Magrez et al. [7] and Li et al. [13]
investigated the effects of variations in Fe and Co compositions of
bimetallic Fe-Co catalyst supported on CaCO3 on CNTs growth and
found Fe2Co as the catalytically active phase. Mhlanga and Coville
[8], observed that bimetallic Fe-Co catalyst supported on CaCO3
prepared by wet impregnation gave better yield of MWCNTs with
near 100% selectivity and excellent reproducibility than deposition-
precipitation and reverse micelle impregnation methods. In the
same vein, Mhlanga et al. [9] investigated the dependence of CNTs
yield and quality on the method of preparation of a similar catalyst
and found the stoichiometric amount of Fe and Co salts favoured
excellent CNTs. Surprisingly, in most of these studies, catalyst
preparation parameters involving the use of statistical design of
experiment that will show the interactive mechanism among the
synthesis parameters was not explored. In view of the influence of
the synthesis parameters on the catalysts growth, the idea of an
experimental approach involving variation of one variable at a time’
while keeping others constant may therefore not guarantee the
desired local optimum [14]. Besides not achieving the required
optimum, this approach is time consuming and generate large
number of experiments. Thus, these number of experiments can be
reduced by applying design of experiment (DOE) either central
composite or full factorial [15,16]. Nevertheless, the statistical DOE
allows the screening and identification of factors that are important
for explaining process variation. It also provides the systematic
investigation of the process and the parameters that influence the
quality of final product, thereby showing the direction of
improvement to enhance manufacturability, reliability and quality
of the product [14]. Despite several successes recorded in the field
of catalyst development, there is limited information of the roles
played by the parameters that influence the growth process [14]. In
addition, the mechanism of catalyst formation is still not clear and
require further investigations. To the best of our knowledge, there is
paucity of information regarding the optimization of effects of
synthesis parameters on the yield and properties of Fe-Co/CaCO3
catalyst.
The present study focuses on optimization of synthesis param-
eters such as drying time, temperature, stirring speed, and mass of
CaCO3 on the yield and quality of bimetallic Fe-Co catalyst sup-
ported on CaCO3 used for carbon nanotubes growth. More so, sta-
tistical approach of DOE based on factorial design was employed in
catalyst synthesis to establish the parameters with most significant
effects on the quality of the Fe-Co/CaCO3 catalysts.
2. Materials and methods
2.1. Materials
The chemicals such as calcium trioxocarbonate (IV), cobalt (II)
trioxonitrate (V) hexahydrate, and iron (III) trioxonitrate (V) non-
ahydrate used in this study were obtained from Sigma Aldrich.
Liquid nitrogen, acetylene, argon and nitrogen gases were pur-
chased from British Oxygen Company/Brin's Oxygen Company
(BOC Gases Plc, Lagos, Nigeria). All the chemicals used in this study
are of analytical grade with percentage purity in the range of
98e99.99%.
2.2. Fe-Co/CaCO3catalyst preparation
Optimization was systematically done by varying the following
parameters; drying time, the drying temperature, stirring speed
and mass of CaCO3 support. The wet impregnation method
described by Ref. [10] was strictly followed with little modifica-
tions. In this study, 3.62 g and 2.47 g of nitrate salts of Iron and
Cobalt compounds, Fe(NO3)3$9H2O and Co(NO3)2$6H2O respec-
tively were weighed and dissolved in 50 cm3 of distilled water
followed by the addition of 10 g of CaCO3 under continuous stirring
at 2000 rpm for 60 min. The resulting slurry was then allowed to
dry at room temperature and later dried in an oven at 120
C for
12 h, cooled to room temperature, ground and finally screened
through a 150 mm sieve. The final powder was then calcined at
400
C for 16 h. The dried catalyst was ground to avoid agglomer-
ation. Levels of the factors varied for optimization of preparation
parameters were chosen by considering the operating limit,
referred to as lower and upper levels. The lower levels () indicate
variable values that were considered as lower limits, and the high
levels (þ) indicate variable values that were considered as upper
limits as shown in Table 1. Levels of factors were chosen following
preliminary studies by Ref. [10]. After drying at 100/120
C, all
sixteen samples prepared according to the 24 factorial design of
experiment were subsequently calcined at the same temperature of
400
C for 16 h. The yields of bi-metallic catalysts prepared in this
present study were calculated after drying (YieldDried), and after
calcination (YieldCalcined) according to Equations (1) and (2).
Mass of catalyst after Drying
YieldDried ð%Þ ¼  100% (1)
Initial mass of mixture
Mass of catalyst after calcination
YieldCalcined ð%Þ ¼  100% (2)
Mass after Drying
Statistical analysis using Minitab ANOVA, and response opti-
mization was conducted on the factorial experimental design and
the single catalyst with the best yield from optimization of four
catalyst synthesis parameters was chosen for carbon nanotubes
production.
 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102 87

Table 1
Levels of factors considered in 24factorial of catalyst preparation.


Drying time (hr) Drying temp. ( C) Stirring speed (rpm) Mass of support (g)

Upper (þ) level 12 120 2000 10

Lower () level 8 100 1000 8

2.3. Production of carbon nanotubes
Carbon nanotubes were synthesized by the decomposition of
acetylene in a CVD reactor. A known weight (0.5 g) of the Fe-Co
catalyst on CaCO3 support was placed in the ceramic boat, which
was inserted in the horizontal quartz tube of the CVD furnace and
heating was done at 10
C/min. The heating commenced and argon
gas was flown over the catalyst at a flow rate of 30 mL/min to purge
the system of air. Once the set temperature was attained (700
C),
the argon flow was adjusted to the required flow rate (230 mL/min)
and acetylene gas was introduced at its required flow rate (190 mL/
min). The process was allowed to proceed until the reaction time
(60 min) was reached after which the flow of acetylene was
stopped. The furnace was allowed to cool to room temperature
under continuous flow of argon. The ceramic boat was then
removed and weighed to determine the quantity of CNTs produced.
Percentage of CNTs yield was determined using the relationship
provided by Yeoh et al. [17], and Taleshi [18], as presented in
Equation (3);
MTotal  MCatalyst
CNTs yield ð%Þ ¼  100 %
MCatalyst
where MTotal is the total mass of the final catalyst and carbon
products after CVD reaction process and MCatalyst is the initial mass
of Fe-Co/CaCO3 catalyst.
2.3.1. Purification of as-synthesized CNTs
The purification of the as-synthesized CNTs was achieved via an
acid treatment of 30% HNO3 and H2SO4. This was done to remove
residual Fe, Co, CaCO3, including amorphous carbon that serve as
impurities introduced into the samples during the catalyst pro-
duction. About 1 g of as-synthesized CNT sample was washed in
100 mL of 30% HNO3 and H2SO4 by continuous stirring for 1 h. The
mixture was then refluxed and washed with distilled water until a
pH of 7 was achieved. The mixture was later filtered out to obtain
the wet CNT residue which was dried mildly at 120
C for 12 h.
primary axis while the derivative percentage weight loss with
respect to time is on the secondary axis.
The TGA curves in Fig. 1 indicates that the CaCO3 decomposed in
a single step between 630
C and 850
C to form stable oxide (CaO)
and CO2 gas. Additionally, percentage weight loss of 44.89% which
corresponds to the stoichiometry thermal decomposition of CaCO3
shown in Equation (4) was observed. The weight loss may be
attributed to partial destruction of the intramolecular framework of
CaCO3 the temperature increase from 630
C to 850
C. resulting to
the formation of CaO and 44% mass of CO2 gas (see supplementary
information, Table S2). Using the single-step thermal decomposi-
tion curve of Fig. 1, the chemistry of CaCO3 thermal decomposition
reaction can be explained as follows:
CaCO3 ðsÞ / CaOðsÞ þ CO2 ðgÞ (4)
|fflfflfflfflfflffl{zfflfflfflfflfflffl} |fflfflfflffl{zfflfflfflffl} |fflfflfflffl{zfflfflfflffl}
MCaCO3 ¼ 100:1 g=mol MCaO ¼ 56:1 g=mol MCO2 ¼ 44 g=mol
Based on stoichiometry of the decomposition of CaCO3 repre-
sented by Equation (4), the mass changes calculations also show %
weight loss of CaCO3 to be 44% (Equation (5)),
(3)
MCO2 44
m ð%Þ ¼  100 ¼  100 ¼ 44 % (5)
MCaCO3 100:1
Furthermore, it can be seen that the decomposition of CaCO3
occurred at a temperature lower than CVD reaction temperature of
about 700
C, which suggest that when CaCO3 is used as catalyst
support, the CNTs could grow over the two different supports
(CaCO3 and CaO). This implies that CNTs was either grown on Fe-
Co/CaCO3 or Fe-Co/CaO which is responsible for the formation of
CNTs with different diameter. Another implication of Equation (4)
in the TGA analysis is that, acid treatment resulted to the removal
of residual catalyst particles, CaCO3 and CaO from as-grown CNTs as
shown in Equation (6a)e(6d);
H2 SO4 ðaqÞ þ CaCO3 ðsÞ/CO2 ðgÞ þ CaSO4 ðsÞ þ H2 OðlÞ (6a)
|fflfflfflfflfflffl{zfflfflfflfflfflffl}
Precipitates out

2.4. Characterization of samples H2 SO4 ðaqÞ þ CaCO3 ðsÞ/CO2 ðgÞ þ CaSO4 ðaqÞ þ H2 OðlÞ (6b)

Catalysts, as-synthesized and purified CNTs samples were H2 SO4 ðaqÞ þ CaCO3 ðsÞ/CaSO4 ðsÞ þ H2 CO3 ðaqÞ (6c)
characterized for their morphologies, microstructures, elemental |fflfflfflfflfflffl{zfflfflfflfflfflffl}
Precipitates
composition, crystallinity, specific surface areas, functional groups,
particles size by High Resolution Scanning Electron Microscope
(HRSEM), (High Resolution Transmission Electron Microscope H2 SO4 ðaqÞ þ CaOðsÞ/CaSO4 ðsÞ þ H2 OðaqÞ (6d)
(HRTEM), Energy-dispersive X-ray spectroscopy (EDX). (Thermal |fflfflfflfflfflffl{zfflfflfflfflfflffl}
Precipitates
gravimetry analysis (TGA 4000 model), X-ray diffraction (XRD),
6000 model), Brunauer-Emmet-Teller (BET) analysis (NOVA 4200e The insoluble CaSO4(s) (slightly soluble, Equation (6b)) pre-
model), Fourier Transform Infrared (FTIR) (Perkin-Elmer 4000 cipitates out of solution and is easily decanted away from top layer
model) and Zetasizer Nano-S series). with some light carbonaceous substances. The previous studies did
not provide step by step reaction mechanism shown in Equations
3. Results and discussion (1)e(6d).

The thermal profiles of the CaCO3 as a support material for CNTs 3.1. BET analysis of Fe-Co/CaCO3
growth at CVD reaction temperature of 700
C under N2 atmo-
sphere and heating rate of 10
C/min was investigated using TGA. The BET analysis shows that the specific surface area of the
Fig. 1 shows the percentage weight loss of the pure CaCO3 on the CaCO3 sample is 3.94 m2/g, which however slightly increased to
88 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102
Fig. 1. TGA and DTG curves of CaCO3 obtained at 10
C/min heating rate.
3.8 m2/g when the metallic particles filled the pores of CaCO3. The
surface area of 3.8 m2/g obtained in this study is however lower
than 13.2 m2/g reported by Chiwaye et al. [12] for CaCO3, the reason
being that the CaCO3 used by Chiwaye and colleague was in
nanoscale form while conventional CaCO3 was used in this study.
The decrease in the surface area of Fe-Co/CaCO3 compared to CaCO3
may be attributed to the blockage of the pores of CaCO3 by the
metallic particles. Using the CaCO3 true density of 2.71 g/cm3, the
BET average particle size was estimated to be 0.583 mm by Equation
(7) assuming spherical geometry of particles [19];
6000
DBET ¼
ðSBET  rÞ
where DBET is the estimated BET-particle size, SBET is BET-specific
surface area (m2/g), and r is true density of CaCO3 (g/cm3).
Despite the unrestricted mono-multilayer adsorption in phys-
isorption, the CaCO3 has low monolayer capacity of 0.873 cm3/g as
calculated from Equation (8), depicting a characteristic of a non-
porous solid material [20].
SBET  MV
Vm ¼
LAV  AN2
where Vm is monolayer capacity (cm3/g), SBET is BET-specific surface
area (m2/g), MV is the molar volume of nitrogen gas (22.414 cm3) at
STP, LAV is Avogadro's number (6.022  1023). AN2 is the cross-
sectional area of adsorbed nitrogen molecule (the probe mole-
cule) representing the area occupied by a single adsorbed nitrogen
gas molecule (16.2 Å2). Previous BET analysis of CaCO3 powder
revealed poor pore properties of CaCO3 toward the nitrogen probe
molecule than common porous materials such as activated clay or
zeolite [21]. Consequently, the wet impregnation technique was
chosen to deposit ions of catalyst active components on the CaCO3
support because it allows proper control of distribution of the
metallic particles and high dispersions are possible; and works best
if ion/solid interactions are involved [22].
3.2. The yield of Fe-Co catalyst supported on CaCO3
In order to investigate the effect of preparation parameters on
the quality of the Fe-Co/CaCO3 catalyst, the following synthesis
parameters namely; drying temperature, stirring speed, drying
time, and mass of CaCO3 support were varied at lower and upper
level and their influence on the yield of Fe-Co/CaCO3 catalyst was
observed. Fig. 2 shows the effect of temperature between 100 and
120
C on the performance of Fe-Co/CaCO3 catalyst. Growth tem-
perature influence metal-support interaction and stability of the
support which in turn determine the activity of the catalytic metals
and its yield. In our case, we noticed no clear cut difference in the
(7) catalyst yield with respect to temperature increment and other
applied conditions in Fig. 2. For instance, the catalyst yields initially
increases at 100
C and thereafter reduces at the same temperature
but different stirring speed, dying temperature and mass of sup-
port. Similar trend was observed at 120
C. Comparatively, when
the drying temperature was increased from 100 to 120
C, it was
found that the yield of the catalyst decreased due to the loss or
evaporation of more moisture. This however contradicts the find-
ings of McCarty et al. [23] who opined that the interactions be-
tween support and active material at high temperature may
resulted to high catalyst yield. Besides, the pre-calcination process
(8)
may promote an increase in crystallite sizes of catalyst particles due
to restructuring of the materials and by extension slightly increase
the catalyst yield [24]. Comparatively, the yield of the catalyst ob-
tained at drying time of 12 h was slightly higher than the corre-
sponding 10 h possibly due to enhanced metal-support interactions
at a longer holding time leading to formation of high catalyst yield.
It can be summarized that the deposited catalyst yield is directly
proportional to the drying time and independent of drying tem-
perature. The observed trend suggests high catalytic conversion of
Fe-Co/CaCO3 particles to carbon nanotubes following acetylene
decomposition as function of reaction time.
Stirring speed guarantees homogeneous mixing and dispersion
of the metal particles on the surface of support materials during
catalyst preparation process. Fig. 3 shows the yield-dependence of
 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102 89

Fig. 2. Effect of drying temperature on catalyst yield at (a) 8 h, and (b) 12 h drying time.

Fig. 3. Effect of stirring speed on catalyst yield at (a) 8 g, and (b) 10 g CaCO3 support.

Fe-Co/CaCO3 catalysts as a function of variation of stirring speed.
According to Fig. 3, there is a direct linear relationship between
the catalyst yield and stirring speed only at 1000 rpm while no
specific trend was observed at stirring speed of 2000 rpm. This
means that lower stirring speed of 1000 rpm favours high catalyst
yield than at 2000 rpm. This suggests that the adsorption of
metallic particles during impregnation was faster than moisture
diffusion noticed during stirring at 1000 rpm than 2000 rpm. The
stirring speed of 1000 rpm caused excellent dispersion of metal salt
solutions onto the CaCO3 support resulting in improved chemically
bound that are not easily removed during the mild drying treat-
ment [25]. Conversely, Aliyu et al. [26], found that increase in
stirring speed from 1200 to 2400 rpm favoured high yield of cata-
lyst due to filtration and deposition of the active metal phase onto
the support. At stirring speed of 2000 rpm, the collision of the
metallic and support particles with wall of the container was rapid
however splash and loss of active phase of the catalysts become
obvious. Also in Fig. 3, it was noticed that the yield of catalyst
samples generally decreased with increase in stirring speed after
calcination. This decrease in yield irrespective of the stirring speed
after calcination process may be ascribed to loss of adsorbed
moisture and decrease in inter and intra molecular attractive forces
as evident in the formation of mixed oxides such as Fe2O3 and CoO
observed in the XRD result shown in Fig. 12.
Fig. 4 shows the variation of drying time on catalyst yield from 8
to 12 h at constant temperature.
Studies have shown that the rate of drying of active catalyst
components impregnated on the CaCO3 support may be affected by
two factors such as adsorption of solutes on the support and pre-
cipitation (diffusion) of unabsorbed solute from the entrapped
liquid in the solid pores during the drying process [27]. Again,
drying time determines the transportation and adsorption mech-
anism, and subsequently influence the distribution of final catalyst
obtained. Accordingly, in Fig. 4, it was noticed that the yield of
catalyst is directly proportional to the drying time, which suggest
ease of decomposition of acetylene gas over Fe-Co/CaCO3 particles
to produce carbon nanotubes. The observed trend can be explained
in terms of high rate of evaporation and transfer rate between the
metallic particles and the support resulting to formation of mixed
oxides, bond breakage and consequently loss of CO2. Other possible
reason may be linked to the blockage of the agglomerated solutes
on the surface sites, thereby suppressing the transportation rate
and by and large affecting the equilibrium between the adsorption
and transport mechanism during drying. Nevertheless, it was
difficult to explain the reaction mechanism observed at the
following applied conditions of (100
C, 2000 rpm, 8 g) and (120
C,
1000 rpm, 8 g) probably due to the interactive and over bearing
influence of the other three parameters on the catalyst yield.
90 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102

Fig. 4. Effect of drying time on catalyst yield at (a) 100
C, and (b) 120
C temperature.

Fig. 5 reveals the effect of mass of support (CaCO3) on catalyst powder in water followed by stirring, drying and calcination in the
yield at constant stirring speed. furnace caused some physico-chemical changes. In the present
When the mass of CaCO3 increased from 8 to 10 g, the adsorp- study, two schemes were proposed for possible chemical reactions
tion rate was fast due to increase in the binding site for impreg- involved in the catalyst formation. The first scheme demonstrated
nation than diffusion rate during drying process [28]. The variation parallel reactions between iron and cobalt nitrates resulting to
in the yield did not follow a specific trend, implying the existence of formation of an anhydrous salt due to the removal of hygroscopic
equilibrium between the solute adsorption and liquid diffusion. water (or water of crystallization). Secondly, there was a conversion
Generally, the chemistry of Fe-Co/CaCO3 synthesis reactions that of the nitrates into corresponding oxides as shown in Equations (9)
occurred during the catalyst preparation involved an and (10) [25].
impregnation-oxide formation process that comprises impregna-
tion, drying and oxide formation steps. While all nitrate salts are FeðNO3 Þ3 :9H2 O þ water/FeðNO3 Þ3 þ [9H2 O þ [water
soluble in water, calcium carbonate powder is almost insoluble. |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
Drying removes excess water
Hence, the CaCO3 support did not undergo chemical reaction(s)
with other components in the catalyst mixture. This is because the (9a)
thermally decomposition temperature of 400
C employed for post-
calcination is far below CaCO3 decomposition temperature of CoðNO3 Þ2 :6H2 O þ water/CoðNO3 Þ2 þ [6H2 O þ [water
650
C as earlier presented in Fig. 1. However, particle size attrition |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
was observed on its surface due to collision during stirring and Drying removes excess water

mechanical grinding. (9b)

Furthermore, there have been conflicting reports on whether or
not reactions occur between CaCO3 and the iron and cobalt metals 2FeðNO3 Þ3 /Fe2 O3 þ 6NO2 þ 1:5O2 (10a)
in the catalyst mixture during heat treatment [7,12,13,29]. None- |fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl}
theless, the dissolution and mixing of the metallic salts Calcination removes nitrates
(Fe(NO3)3$9H2O, Co (NO3)2$6H2O salts) and the support (CaCO3)

Fig. 5. Effect of mass of support on catalyst yield at (a) 1000 rpm, (b) 2000 rpm stirring speed.
 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102
3CoðNO3 Þ2 /Co3 O4 þ 6NO2 þ O2
|fflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflffl}
Calcination removes nitrates
It should be mentioned that the calcination process was
responsible for the given off O2 and NO2 and subsequently
contributed to the conversion of the anhydrous nitrates into oxides,
which in turn increases surface area. The iron and cobalt exist in the
as-prepared catalyst as binary metal oxides (Fe2O3 and Co3O4). The
second scheme involves a chemical reaction between the two
anhydrous salts, resulting in the formation of a ternary metal oxide,
cobalt ferrite (CoFe2O4) during calcination. The formation of a
mixed oxide (CoFe2O4) is in agreement with the XRD results shown
in Fig. 12. The conversion of iron and cobalt nitrate into cobalt iron
oxide was accelerated by the thermal treatment at 400
C under air
for 16 h. Very importantly, the second scheme was more likely to
occur and the overall reaction including impregnation, drying and
calcination steps is represented in equation (11).
CoðNO3 Þ2 :6H2 O þ 2FeðNO3 Þ3 :9H2 O/CoFe2 O4 þ 8NO2 þ 2O2
þ 24H2 O
The ternary (or multi-metal) oxide, containing more than one
type of metal cation, is characterized by superior properties as
compared to the binary systems [25]. Therefore, the proposed re-
action mechanism shows that the final catalyst composite is either
a mixture of CoFe2O4 or (Fe2O3, Co3O4, CoFe2O4) evenly distributed
on the surface of CaCO3 matrix. This is contrary to the outcome of
Chiwaye et al. [12], who established the growth of Ca2FeCoO5 on
the surface of CaCO3 prior to heating. The difference in catalyst
preparation steps could be responsible for the different phases
noticed in the both studies.
Besides the formation of mixed oxides, there may be reduction
of the catalyst oxides during the CNT growth in CVD to metallic Fe
and Co by the hydrogen from acetylene gas as evidence in the
HRSEM/EDS results shown in Fig. 8. This reduction phenomenon is
normally responsible for the catalyst activation [12]. The catalytic
reduction follows the chemical reactions presented in Equations
(12)e(14);
Fe2 O3 þ H2 /2FeO þ H2 O; FeO þ H2 /Fe þ H2 O
Co3 O4 þ H2 /3CoO þ H2 O; CoO þ H2 /Co þ H2 O
CoFe2 O4 þ H2 /CoO þ 2FeO þ H2 O;
CoO þ FeO þ 2H2 /Co þ Fe þ 2H2 O
This is contrary to the findings of Surowiec et al. [29] who
established in the temperature programmed reduction (TPR) that
complete reduction of FeO to the metallic Fe hardly occur in the
CVD reaction temperatures of 700
C to 750
C (see equation (15)).
3Fe2 O3 þ H2 /2Fe3 O4 þ H2 O; Fe3 O4 þ H2 / 3FeO þ H2 O
Additionally, Chiwaye et al. [12], earlier noticed that reduction of
Ca2FeCoO5 when equal ratio of Fe and Co salt were mixed in the
presence of hydrogen gas produced a mixture of Fe2Co, Co and not
Fe alone.
Therefore, the correct and balanced chemical reaction schemes
involved in the Fe-Co/CaCO3 catalyst can be investigated further by
XRD analysis of the catalyst and as-synthesized CNTs. Additionally,
the formation of various oxides due to metal-support interactions
91
after calcination process suggest both tip and base-growth mech-
(10b) anism for the catalyst, an indication of strong and weak interactions
[8,9].
For the purpose of optimization process, the responses obtained
on the effects of synthesis parameters on the yield and qualities of
catalyst were used in carrying out a factorial design analysis. This
involved an analysis of variance (ANOVA) to generate empirical
models for main and combined effects in post-calcination yields.
The factorial design of experiment and its analysis was carried out
in Minitab statistical analysis software. The concept of p-
values < alpha-value of 0.05 were used to identify factors that have
significant effect on catalyst yield. The overall evaluation of the
statistical factorial analysis revealed that stirring speed, and its
interaction with mass of support were significant and positive in
catalyst preparation. This is in agreement with the findings of
Haber et al. [22] and [27]. Morbidelli et al. [24] demonstrated that
two mechanism governing the catalyst distribution and yield are:
adsorption of metal particles onto the catalyst support at impreg-
nation and the transport (or diffusion) during calcination. However,
the statistical analysis under the influence of stirring speed
revealed that the final yield of catalyst was governed by adsorption
mechanism and not transports type that occurred during impreg-
(11)
nation stage. It can be concluded that there was an existence of
synergy between the stirring speed (p-value, 0.061) and mass of
support (p-value, 0.076) given rise to a significant effect with an
alpha-value of 0.10 (corresponding to 90% confidence level). The
effects of drying time, drying temperature, and mass of support on
the catalyst yield can also be seen in the Normal plot in Fig. 6. But,
the factorial design failed to uncover the statistical significance of
other factors because the sample size was probably too small. A
confidence level of 90% shows that at constant catalyst dosage,
stirring speed and mass of support exhibited synergetic effects on
catalyst yield.
The regression model in Equation (16) was generated using the
general linear model in Minitab ANOVA;
Yieldð%Þ ¼ 23:49 þ 1:81312A  0:234125B þ 0:0182250C
þ 8:85125D  0:00051562AC þ 0:000290875BC
 0:00568125CD
(16)
(12)
where A ¼ time (hr), B ¼ temperature (oC), C ¼ stirring speed (rpm),
and D ¼ mass of support (g). The other factors such as temperature
(13)
and drying time that were statistically non-significant in ANOVA
were observed to exert influence on the catalyst yield as revealed
(14a) by the surface plots of effects of these factors and their interactions
shown in Fig. 7.
(14b) The numerical optimization of Equation (16) for the yield
otherwise called response optimization was conducted under the
following optimal conditions: 8 h drying time, 100
C drying tem-
perature, high stirring speed, and 10 g mass of support; having
desirability of 1.000. Under these conditions, the maximum catalyst
yield was 88.61%. The catalyst sample with optimum yield at the
stated optimum conditions was used in the production of CNTs via
the CVD method.
(15)
3.3. Characterization of Fe-Co/CaCO3 catalyst
The morphology and elemental composition of the as-prepared
Fe-Co/CaCO3 catalyst obtained under the optimum conditions of
stirring speed, mass of support, drying time and temperature was
examined using high resolution scanning electron microscopy
(HRSEM) coupled with Energy-dispersive X-ray spectroscopy
(EDS). The micrographs and the EDS spectrum are presented in
92 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102

Fig. 6. Normal plot showing stirring speed and its interaction with mass of support as significant at a ¼ 0.10.

Fig. 7. Bi-factor yield-control of the catalysts at low/high settings of two other factors held constant at: (a.) 8 h, 1000 rpm, (b.) 12 h, 2000 rpm, (c.) 100
C, 1000 rpm, (d.) 120
C,
2000 rpm, (e.) 1000 rpm, 8 g mass of CaCO3, (f.) 2000 rpm, 10 g mass of CaCO3.
 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102
Figs. 8 and 9.
The HRSEM image at low and high magnification revealed the
presence of spherical, clustered solid catalyst material incorporated
onto the support material. The level of agglomeration noticed in the
HRSEM images could be linked to the calcination process employed
during sample preparation. At higher magnification of 49.63 KX
(Fig. 8b), micropores can be identified in the catalyst, a property
indicating that there was proper dispersion of Fe and Co nano-
particles on the CaCO3 support material with minute spaces within
the composite, which suggest a tip growth mechanism. The CaCO3
matrix sites were uniformly occupied by the oxides of Fe and Co
that were formed during calcination of the catalyst sample. This
observation was further confirmed by BET surface area results
showing the micro porous radius in the order of (0.184e1.4 nm),
which are not characteristics pores of the catalyst components but
lattice-fringe patterns in the catalyst sample [28]. The sizes of the
Fe-Co nanoparticles could influence the diameter of CNTs. In the
present study, a high resolution SEM image of the catalyst pre-
sented in Fig. 8 shows that the Fe-Co nanoparticles that were
dispersed on the CaCO3 support material are sufficiently less than
50 nm in size. The dispersion of the Fe-Co nanoparticles on the
CaCO3 is also supported by the accompanying EDS spectrum
showing that Fe, Co, Ca, C and O as the dominant element present in
the sample. The EDS spectrum qualitatively confirmed the chemical
components of the Fe-Co/CaCO3 catalyst. The results also showed
that the nitrates of Fe and Co salts have been converted to their
respective metallic oxides during calcination process. The per-
centage elemental compositions of the original catalyst mixture
revealed that all particles contained both Fe and Co in the ratio of
1:1 mass % as used in catalyst preparation. Fig. 9 also showed high
oxygen content which could be as a result of iron and cobalt oxides
formation (CoFe2O4, Fe2O3 and Co3O4) during calcination process.
Another important qualitative analysis from the EDS spectrum of
Fig. 9 is the use of standard line energy values at corresponding
wavelengths, and the result of each element (see supplementary
information, Table S3). The elements such as carbon and oxygen
have low-energy (0.277 and 0.525 keV) Ka lines. Because oxygen is
possibly in the mixed oxides form of the following elements; Ca, Fe
and Co, while the heavy elements can be found on the low-energy
(<0.8 keV) Ka lines [30]. It can be inferred that CaCO3 support and
Fe-Co oxides are present in the final catalyst sample. The occur-
rence of the low-energy representations of the heavy elements
follow the order C<Ca<O<Fe<Co as seen in Fig. 8. Thus, the pres-
ence of CaCO3 and CoFe2O4 were speculated based on the relative
closeness of these elements in line energies around the usual po-
sition of oxygen in the spectrum.
Surface area analysis was carried out on the Fe-Co/CaCO3 cata-
lyst obtained at the optimum conditions using BET method. It was
Fig. 8. HRSEM image (a) low magnification (b) high magnification showing nanoparticles of Fe and Co dispersed on CaCO3 support.
93
found that the Fe-Co/CaCO3 catalyst has a specific surface area
almost equal to that of the support material. The BET specific sur-
face area of catalyst was found to be 3.8 m2/g while the pure CaCO3
used as support possessed BET specific surface area of 3.9 m2/g.
There was no significant change in surface area because no thermal
decomposition of the major catalyst component occurred instead
CaCO3 decomposed at the treatment temperature of 400
C. The
dispersion of Fe and Co nanoparticles in the support matrix could
be responsible for the slight decrease in surface area. Also, the
catalyst has specific pore volume of about 0.002 cm3/g, and pore
radius of between 0.184 and 1.4 nm. The measured pores are lattice-
fringe patterns in the catalyst sample, demonstrated that the
nanoparticles are highly crystalline in nature.
Dynamic Light Scattering (DLS) technique was used to obtain
the sizes of the as-synthesized catalysts in terms of particle diam-
eter, assuming spherical particle geometry. The samples particle
sizes ranges between 163.2 and 775.7 nm (average of 409.7 nm)
(Fig. S10(a), see supplementary information). The estimated BET
average particle size for pure CaCO3 support material was earlier
calculated to be 583 nm. After drying at 100e120
C, the catalyst
mixture particle sizes become smaller than that of the CaCO3
support material; this was an evidence of conversion of the
amorphous support material into a crystalline form. Another
possible reason could be that the growth of CaCO3 crystals was
suppressed by the Fe and Co loaded onto the CaCO3 [31]. Again,
some of the dried catalyst particles are larger than particles of
CaCO3 with the biggest reaching about 775.7 nm due to agglom-
eration of particles. Drying was necessary to fix the metal catalysts
onto the CaCO3 support matrix however it was noticed that high
temperature calcination does not cause a significant crystal defor-
mation, instead the drying of catalysts promoted the formation of
agglomeration and sintering of catalyst particles. When such
catalyst is introduced into the reactor, only the outer surface of the
large agglomerates is exposed to the reaction gas hence the catalyst
efficiency will be strongly limited. To address this problem in the
present study, mechanical grinding of catalyst samples was
employed after calcination to prevent the catalyst particles from
clustering. After calcinations at 400
C for 16e20 h, followed by
grinding of the dried catalyst, the particle sizes of the catalyst
samples determined and found to be in the range of 97.4e110 nm as
shown in Fig. S10b.
Fig. S10 (a) and (b) (see supplementary information) showed
that catalyst particles reduced after calcination because the dried
catalysts were ground after calcination. Hence, the DLS size analysis
showed that smaller particle sizes of the catalysts with narrow size
distribution were obtained after calcination, and the surface areas
of catalyst particles have increased. It is important to reduce the
particle size of the catalyst so as to obtain a high yield of CNTs [4].
94 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102

Fig. 9. The EDS elemental compositions of Fe-Co/CaCO3 catalyst sample.

Thermogravimetric analysis (TGA) was used to investigate the
thermal profiles of the catalyst substance within temperature range
of the CVD reaction. In CVD method, the catalysts for CNT pro-
duction are usually exposed to high reaction temperature of about
700
C or above and the result is presented in Fig. 10. The samples
showed variation in percent weight loss at different temperatures,
which means that the samples have different heat content. The
variation in TGA profile could be due to the different CaCO3 com-
positions used in preparation and moisture content of the samples.
Results as presented in Fig. 10 indicate that there were four regimes
of weight losses in the TGA curve; at 45
C, 50e180
C, 180e380
C,
and 380e550
C. At approximately 45
C in all samples, weight loss
can be attributed to loss of absorbed moisture. Upon heating, water
of crystallization in the trioxonitrate (V) salts causes these salts to
crack and dissolve. The salts only decomposed after all the water
has been driven off. The most probable reaction during the TGA
heat treatment is the formation of Ca2FeCoO5. The initial mass
losses prior to 380
C on the TG curve are due to the elimination of
volatile products and of the crystal water of the metal nitrates. The
slope between 380 and 550
C is attributable to formation of
Ca2FeCoO5 by oxidative decomposition of iron and cobalt nitrates
mixture with elimination of O2 and NO2 [12,30]. Above 550
C, the
samples mass remained constant because CaCO3 has not decom-
posed at that temperature. This scheme also proposes that the
CaCO3 reacts with iron and cobalt liberating CO2 in the process. The
reaction during TGA thermal analysis, where the Fe-Co/CaCO3
catalyst was exposed to further heat treatment, are different from
those in the heat treatment calcinations of catalyst preparation that
occurred around 400
C. Rather than CoFe2O4, the formation of
Ca2FeCoO5 is most likely during TGA. In order to investigate the

Fig. 10. TGA graph of Fe-Co/CaCO3 catalysts from 24 factorial design of experiment, obtained at 10
C/min in N2 environment.
 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102
effect of the CaCO3 support on thermal behaviour of the supported
catalyst, a comparison between their TGA curves was made (see
Fig. 11).
As presented in Fig. 11, it was noticed that the Fe-Co/CaCO3ca-
talyst had an improved residual weight % at 850
C higher than that
of the support material (pure CaCO3) probably due to the iron and
cobalt metals loaded onto the catalyst. Weight loss also occurred at
temperatures of 150
C, 550
C and 810
C. The loss in mass that
occurred at 150
C was attributed to loss of absorbed moisture.
Again, the weight loss at 550
C was attributed to the formation of
Ca2FeCoO5 [12]. This loss was due to CO2 evolution when CaCO3
reacted with the iron oxide and cobalt oxide. The third slope at
810
C represents the decomposition of CaCO3 to CaO and CO2. The
weight losses were compared as summarized (See supplementary
information, Table S4).
A wide-angle X-ray diffraction (XRD) technique was used to
identify phases that are present in the catalyst sample, its texture
analysis and crystallite size measurement. The phases identified in
the XRD analysis of the Fe-Co/CaCO3 catalyst at room temperature
are shown in Fig. 12.
The XRD peaks in Fig. 12 showed that the occurrence of a sharp
and intense diffraction peak at 2q of ` e 36
with a corresponding
reflection plane (311), which has been reported connotes CoFe2O4
[30]. While a short and less intense diffraction peak at 2q of 29e30

corresponds to CaCO3 peak [28]. The low and high intense peaks
observed at 2 thetha value of 30e40
are assigned to mixed oxides
active phase such as Fe2O3, CoO including CaCO3 and CoFe2O4
respectively. The presence of Fe2O3 and CoO may be attributed to
decomposition of their respective salts during the calcination
process. The presence of active phases closely agrees with the
findings of Chiwaye et al. [12], who also identified CaCO3 and
CoFe2O4 as the main phases in the Fe-Co/CaCO3 catalyst and suggest
the crystalline nature of the catalyst. The crystallite sizes of the
catalyst calculated according to the Debye-Scherrer Equation (17) is
in the range of 26.81e121.65 nm [19].
Fig. 11. Comparing Thermal behaviour of catalyst with that of the CaCO3 support.
95
Kl
SizeScherrer ¼ (17)
ðFWHMÞ  CosðqÞ
where FWHM is the full width at half maximum of a peak in the
XRD pattern (in radians), K is shape factor called Scherrer constant
with value between 0.2 and 2.0, l is the X-ray wavelength of
0.154 nm used in the X-ray diffraction, and q was evaluated from
the 2q values of the XRD pattern.
The average mean crystallite size was 61.92 nm, which showed
that the catalyst particle is at nano level and a characteristic feature
of highly crystalline material. The different sizes of metals in the
catalyst alloy are responsible for diameter distributions of nano-
tubes [24]. The average crystallite size is 10 times greater than the
value of 6.25 nm reported by Ref. [8]. The differences in the particle
size may be linked method employed for catalyst preparation,
while [8] used a reverse micelle impregnation method, this study
utilized only impregnation method. Mhalanga and Coville [8],
found that no CNTs were formed due to the small size nature of the
catalyst particles. Whereas in this study, high yield of CNTs were
obtained, which implies that the crystal sizes of the catalyst particle
are large (see supplementary information, Fig. S14) and hence
determine the yield of the carbon nanotubes [32].
In addition to the HRSEM/EDS analysis, the catalyst particles
microstructures were equally examined using high resolution
transmission electron microscopy (HRTEM). In the same vein, the
corresponding selected area electron diffraction (SAED) pattern
was collected for possible determination of different phases pre-
sent in the sample. Fig. 16 reveals the presence of highly crystalline
metallic particles dispersed onto the whitish CaCO3 support matrix.
The presence of a crystalline phase in the HRTEM may be attributed
to the possible formation of CoFe2O4 in the final catalyst shown in
Fig. 13. The wet impregnation method used in catalyst preparation
proved to be effective in the dispersion of the metallic nano-
particles onto the support matrix. This was enhanced by stirring of
the slurry mixture prior to heat treatment. Although, there were
some isolated zones where metal particles and CaCO3 support are
96 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102

Fig. 12. XRD pattern of Fe-Co/CaCO3 catalyst.

not in properly mixed (Fig. 13), excellently dispersion of crystals of
the metal particles on the support matrix are illustrated in Fig. 14
(a) and (b). The regions of metal particles are the active part or
sites of the catalyst on which CNTs would grow, while the support-
dominated regions (white bulk regions) may be referred to as the
inactive sites.
It was however, difficult to determine the exact size and lattice
structures of the crystals of active components and that of the
CaCO3 support from the HRTEM images. The active sites identified
in Fig. 14 (b) showed that crystals of catalyst active components are
in the order of 1 nm. Furthermore, the presence of lattice fringes in
the HRTEM micrographs further confirms the crystalline nature of
the as-synthesized catalyst. The HRTEM/EDS in Fig. 15 shows
similar qualitative elemental microanalysis to that obtained by
HRSEM/EDS. But, the HRTEM/EDS shows presence of Cu which
represents the Cu grid used as sample holder in HRTEM analysis.
The EDS qualitatively confirms CaCO3 as part of the catalyst
solid, and the combined presence of Co, Fe and O at the line energy
position was a strong evidence of possible formation of a ternary
oxide (CoFe2O4) as shown in Fig. 15. The Cu detected in the sample
originated from the holey copper grid used during the HRTEM
analysis of the sample. The corresponding SAED pattern from the
catalyst HRTEM analysis is presented in Fig. 16.
The corresponding d-spacing calculated from the reciprocal
lattice spacings (1/d) in the SAED are as follows: 2.0, 3.2, 3.7, 4.5, 5.8,
6.0, 6.5, 7.3, 9.0 nm1. In order to identify phases that correspond to
spinel crystal structure, the lattice constant, a, was calculated using
Equation (18),
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
a ¼ dhkl h2 þ k2 þ l2 (18)
Equivalent 2q peak position was obtained from Equation (19),
where l ¼ 0.154 nm.

Fig. 13. HRTEM images of the catalyst showing a clear view of coloured metal ions; (light brown Fe3þ, dark/black Co2þ) and white CaCO3 bulk with modified crystal morphologies.
(For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102 97

Fig. 14. (a) Iron, Cobalt on CaCO3 matrix (evidence of good dispersal),and (b) active sites having regular arrangement of metal nanoparticles on the support.

Fig. 15. TEM/EDS of the Fe-Co/CaCO3 catalyst.

l ¼ 2dhkl sin qhkl (19)
The calculations revealed that the sharpest ring in the SAED
(Fig. 16) corresponds to plane (311) which may be due to CoFe2O4
that occurred at 2q peak position of 33
, which is an evidence of
good crystallinity of the Fe-Co/CaCO3 nanoparticles. In the XRD
pattern presented in Fig. 12, this reflection occurred at 36
, the
discrepancy suggests a slight shift due to stress during the HRTEM/
SAED collection. However, the SAED result is closely agree with
literature reports that have attributed the strongest peak around
35-36
to CoFe2O4 [29]. Again, the earlier presented XRD peak of
CaCO3 (2q ¼ 29
) which showed the intense diffraction peaks
indicated the relative abundance of CaCO3 support in the catalyst
solid. This does not mean that CaCO3 is more crystalline than the
CoFe2O4. Other reflections are also present but some of these are
due to CaCO3 crystal structure. The hexagonal (trigonal) calcium
carbonate crystal structure has lattice parameters/constants of
a ¼ 0.49896 nm, and c ¼ 1.7061 nm. Equation (20) gives the Bragg's
relationship for this crystal system;
 
1 4 h2 þ hk þ k2 l2
2
¼ 2
þ 2 (20)
d 3 a c
According to equation (20), the following reflections plane in
the SAED; (104), (110), (100), (101), (012), (006), (112), (114), (102)
and (011) can be due to CaCO3. Notably, the third ring formed from
reflection of spinel (311) plane can be attributed also to (112) and
(006) planes of the hexagonal CaCO3 lattice structure [12,30].
Generally, the properties of the as-synthesized catalyst showed
98 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102
Fig. 16. SAED of the Fe-Co/CaCO3 catalyst showing sharpest rings of CoFe2O4 (311) and
CaCO3 (104).
that the bimetallic (Fe-Co) was well-dispersed on the CaCO3 sup-
port with thus make it suitable for CNTs production. The HRTEM
and HRSEM images of the carbon nanotubes are presented in Fig. 17
(a, b) and (c) respectively.
The HRTEM images (Fig. 17 a, b) show that multi-walled carbon
nanotubes (MWCNTs) were produced, and the MWCNTs have an
average diameter of ~25 nm. It can be also observed from the
HRTEM micrographs that metal particle from the catalyst were
encapsulated in the tip of the MWCNTs, suggesting the occurrence
of a tip-growth mechanism [30]. The HRSEM image in Fig. 17 (c)
showed the formation of homogeneously and densely distributed
MWCNTs with little level agglomeration. More so, the BET-specific
surface area of the as-synthesized MWCNTs is 270.06 m2/g, which
is comparable with the values reported in the literature [25]. The
elemental composition of the as-synthesized MWCNTs (see sup-
plementary information, Table S5) and it was found that the major
element is carbon. While other elements such as Ca, Fe and Co
originated from the support material (CaCO3) and the catalyst
particle (Fe, Co). Table S5 (see supplementary information) also
shows that the acid purification treatment was responsible for the
increase in the carbon content and decrease in the amount of other
elements, which act as impurities. More effective purification
methods would be necessary to achieve complete removal of other
elements.
The H2SO4 acid treatment removed the residual CaCO3 and CaO
catalyst from as-grown CNTs according to Equation 6 (a-d). From
Fig. 17. (a), (b)HRTEM and (c) HRSEM images of MWCNTs produced on Fe-Co/CaCO3.
the chemical equations, formation of the insoluble calcium sul-
phate (CaSO4) can be identified as a disadvantage of H2SO4 purifi-
cation when CaCO3 is used as catalyst support. But, this was
carefully handled in the present study. The insoluble CaSO4(s)
precipitates out of solution and was easily decanted away from top
layer together with some light carbonaceous substances. The yield
of 92.29% C in the purified MWCNTs represent its purity, which is
directly proportional to the stoichiometric amount of transition
metals (Fe and Co) loaded on the CaCO3 supported under the
applied conditions. This implied that the bimetallic (Fe-Co) catalyst
provided the nuclei for CNTs to grow and as such responsible for the
high yield and diameter of the carbonaceous materials. This means
that the catalyst yield of 88.61% under the applied synthesis con-
ditions transformed to 92.29% carbon yield, an indication of ease of
diffusion of acetylene gas onto the stoichiometry amount of Fe-Co
supported on CaCO3. The carbon yield obtained in this study is
5.5 manifold greater that reported by Yang et al. [33], who utilized
Nitrogen-doped porous graphitic carbons (NPGCs) supported on
CaCO3 to synthesize graphitic CNTs. The differences in the carbon
yield may be ascribed to reaction conditions applied during the
synthesis. In this study, acetylene gas was used as a carbon source
which decomposed at temperature and reaction time of 700
C for
60 min however Yang and co-workers, utilized melamine-
formaldehyde resin as a carbon source, which takes longer time
to decompose and as such responsible for its low yield. Addition-
ally, the CNTs yield obtained in this study is 9 times higher than the
amount reported by Escobar et al. [34], who demonstrated the
synthesis of CNTs involving decomposition of benzene on the
bimetallic Fe-Co catalyst supported on MCM-48 Silica at 650
C
under nitrogen flow for 1 h. The differences in the carbon yield may
be due to the nature of support material, reduction of metal particle
melting point, enhanced carbon solubility as well as successful
dispersion of metallic clusters during CNTs growth [8,9,35]. The
percentage of the CNTs yield obtained from bimetallic Fe-Co cata-
lyst on different supports using different carbon sources previously
reported in the literature in comparative with the present study are
given in Table 2.
According to Table 2, it is obvious that several workers have
established different yield of CNTs as a function of applied experi-
ment conditions. Additionally, it was noticed that the yield of CNTs
produced using acetylene gas was higher than other carbon sources
due to its ease of decomposition at low temperature forming rad-
icals and ions, thus aid easy formation of MWCNTs. While methane
gas and other source of carbon favoured the formation of SWCNT
and well aligned DWCNTs of different length and diameter. Table 4
further revealed that the yield of CNTs depends largely on the type
 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102 99

Table 2
Comparison of the CNTs yield results of the previous and present study.

Catalyst Applied reaction conditions of CNTs growth CNTs yield (%) References

Fe-Co/zeolite-Y Mass of catalyst (50 g), reaction temperature (800
C), deposition time (2 h), argon flow rate (40 kPa), carbon source 95% SWCNTS [36]
(ethanol)
Fe-Co/ Argon flow rate (500 sccm), carbon source (CH4 þ H2), temperature (750e870
C), deposition time (10e60 min) The yield was not [37]
mesoporous determined
silica
Fe-Co/zeolite Mass of catalyst (700 mg), Reaction temperature (900
C), reaction time (30 min), argon flow rate (250 mL/min), >80% DWCNTs [38]
10 mL/min flow of acetylene gas as carbon source
Fe-Co/MgO Mass of catalyst (0.6 mg), Reaction temperature (900
C), reaction time (30 min), nitrogen flow rate (not specified), CNTs yield not [39]
1000 sccm flow rate of methane gas determined
Fe-Co/CaCO3 Mass of catalyst (0.5 or 1 g), Reaction temperature (700
C), reaction time (60 min min), 300 mL/min nitrogen flow Yield was not [40]
rate and 30 mL/min acetylene/ethyne flow rate calculated
Fe-Co/CaCO3 Mass of catalyst (50 mg), Reaction temperature (720
C), reaction time (30 min min), 70 l/h nitrogen flow rate and 30 MWCNTs [41]
10 mL/min acetylene flow rate
Fe-Co/CaCO3 Mass of catalyst (50 mg), Reaction temperature (600e850
C), reaction time (60 min), ethylene/hydrogen/nitrogen 89 MWCNTs [42]
mixture (50:1:450)
Fe-Co/USY Zeolite Pyrolysis temperature (700e850
C), carbon source (ethanol), argon flow rate (100 sccm), reaction time (120 min) 40 SWCNTs [43]
powder
Fe-Co/CaCO3 Mass of catalyst (0.5 g), temperature (700
C), the argon flow rate (230 mL/min), and acetylene gas flow rate 92.3 MWCNTs Present
(190 mL/min), reaction time (60 min) study

of the catalyst support, source of carbon, temperature, deposition
time and argon flow rate. Some conclusions can be drawn that the
yield and quality of CNTs did not only depend on metal-support
interaction alone but also on the source of carbon and its flow
rate. Thus, suggest the existence of a linearly relationship between
the carbon yield and acetylene flow rate. The flow rate of acetylene
gas used in this study is 19 and 3 times higher than the flow rate
used by Yu et al. [5], and See and Haris [44], who obtained 30% and
89% CNTs yield respectively. Furthermore, the high yield and
quality of CNTs in this study may be due to the formation of alloy
phase and existence of synergetic effect between the Fe and Co
catalyst.
The outer diameter-distribution of CNTs is important in their
consideration for different applications. The outer diameter-
distribution of MWCNTs produced in this study is presented in
Fig. 18 (a) and its corresponding selected area electron diffraction
(SAED) pattern is shown in Fig. 18 (b).
As shown in Fig. 18 (a), the average outer diameter of the carbon
nanotubes was estimated to be 25 nm which is within the nano-
scale level. It has been established that MWCNTs with a narrow
diameter distribution is more suitable for industrial mechanical
reinforcement and polymer-composite applications. The value of
external diameter (25 nm) falls within the range (20e30 nm) re-
ported by Mhlanga et al. [9], for MWCNTs despite performing the
experiment under different set of standard conditions. On the other
hand, the average outer diameter (25 nm) reported in this study is
smaller compared to average external diameters of 57.5 nm ob-
tained by Bahgat and co-workers who synthesized MWCNTs from
bimetallic Fe-Co/CaCO3 catalyst pyrolysed in horizontal tubular
reactor at 800
C for 40 min. The differences may be ascribed to the
different applied synthesis conditions. The SAED pattern suggests
that the crystals arrangements in the as-synthesized MWCNTs are
polycrystalline in nature as revealed by the presence of several
bright spots and the rings. These concentric and bright rings
correspond to the individual peaks observed in the XRD patterns.
Fig. 19 shows XRD patterns of the unpurified and purified MWCNTs,
and accordingly similar diffraction peaks were observed in both
cases. A sharp and intense peak was clearly seen at 2 thetha value of
26.5
, which correspond to the crystal plane (002). Other peaks of
low intensities were noticed at 2 thetha value of 30.5
(110),

Fig. 18. (a) Diameter-distribution, and (b) the SAED pattern of MWCNTs.
100 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102

Fig. 19. XRD patterns of purified and unpurified CNTs.

44.3
(100), 52.7
(220) and 61.8
(211). The two most dominant
peaks at 2 thetha value of 26.5
and 44.3
are attributed to a highly
graphitized CNTs, which have been widely observed by several
authors [8e15]. Surprisingly, there was no observable peak for Fe,
Co or residual CaO in the as-synthesized CNTs. This could be
attributed to their nanometer size and high dispersion of metal
particles on the surface of CaCO3. It could also be said that their
concentrations were below the detection limits of XRD or that the
residual metallic and support particles were encapsulated within
the wall of the CNTs. In spite of this, the purification step did not
destroy the CNT structure (see supplementary information,
Table S5). The TGA profile of the CNTs carried out under N2 envi-
ronment as presented in Fig. 20 shows a single-step degradation
which is typical of a pure CNTs. It was also observed that the
nanotubes are thermally stable up to 600
C, where oxidation of
carbon begins to occurs, which further suggest the suitability if the
prepared CNTs for high-temperature applications. Beyond 600
C,
only the remnant of metals particles (Fe-Co) and support materials
(CaO) still remain and were not completely removed despite the
acid purification treatment [30].

Fig. 20. TGA profile of the MWCNT.

 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102
4. Conclusion
In summary, the combined influences of key synthesis param-
eters on the yield and quality properties of Fe-Co catalyst supported
on CaCO3 for CNTs growth using Factorial design of experiment
were investigated. The findings from the study are summarized as
follows. Bimetallic (Fe-Co) catalyst supported on CaCO3 was suc-
cessfully synthesized and the optimum conditions to achieve the
highest yield of 99.6% were: 100
C drying temperature, 12 h drying
time, 1000 rpm stirring speed and 10 g mass of CaCO3. Of all the key
synthesis parameters, the study demonstrated that stirring speed
and mass of support significantly and positively influence the yield
and control the quality properties of the CaCO3 supported Fe-Co
catalyst than others. The absorption and diffusion mechanism
improved the bond rearrangement of the metallic atoms during the
wet impregnation process of catalyst preparation. CaCO3 supported
Fe-Co catalyst with thermal stability above 750
C was successfully
employed to grow multi-walled carbon nanotubes of ~25 nm
diameter and 270 m2/g BET-specific surface area.
Conflicts of interest
The authors declared no conflicts of interest.
Acknowledgements
Financial support received from Tertiary Education Trust Fund
(TETFUND) Nigeria with grant number TETF/DESS/NRF/FUTM-
2016/STI/1 is greatly appreciated. The authors are grateful to Centre
for Genetic Engineering and Biotechnology, Federal University of
Technology, Minna for the assistance rendered during sample
analysis. The authors wish to appreciate the following people that
helped in analysing the samples: Dr Remy Bucher (XRD, ithemba
Labs), Dr. Franscious Cummings (HRTEM, Physics department,
University of the Western Cape (UWC), South Africa), Andrian Jo-
seph (HRSEM, Physics department, UWC, South Africa).
Appendix A. Supplementary data
Supplementary data related to this article can be found at
https://doi.org/10.1016/j.jallcom.2018.03.150.
References
[1] A. Aqel, K.M.M. Abou El-Nour, R.A.A. Ammar, A. Al-Warthan, Carbon nano-
tubes, science and Technology part (I) structure, synthesis and characteriza-
tion, Arab. J. Chem. 5 (2012) 1e23.
[2] J. Prasek, J. Drbohlavova, J. Chomoucka, J. Hubalek, O. Jasek, V. Adam, Methods
for carbon nanotubes synthesis e review, J. Mater. Chem. 21 (2011)
15872e15884.
[3] M. Kumar, Y. Ando, Chemical vapour deposition of carbon nanotubes: a re-
view on growth mechanism and mass production, J. Nanosci. Nanotechnol. 10
(2010) 3739e3758.
[4] J. Cheng, X. Zhang, Z. Luo, F. Liu, Y. Ye, W. Yin, Carbon nanotubes Synthesis and
Parametric Study using CaCO3 nanocrystals as catalyst support by CVD, Mater.
Chem. Phys. 95 (2006) 5e11.
[5] E. Couteau, K. Hern adi, J.W. Seo, L. Thie^n-Nga, Cs. Miko , r. Gaa
l, CVD synthesis
of high-purity multi-walled carbon nanotubes using CaCO3 catalyst support
for large-scale production, Chem. Phys. Lett. 378 (2003) 9e17.
[6] A.S. Abdulkareem, I. Kariim, M.T. Bankole, J.O. Tijani, T.F. Abodunrin, S.C. Olu,
Synthesis and characterization of tri-metallic FeeCoeNi catalyst supported on
CaCO3 for multi-walled carbon nanotubes growth via chemical vapor depo-
sition technique, Arab. J. Sci. Eng. 42 (2017) 4365e4381.
[7] A. Magrez, J.W. Seo, Cs Miko , K. Hernadi, L. Forro
, Growth of carbon nanotubes
with alkaline earth carbonate as support, J. Phys. Chem. B 109 (20) (2005)
10087e10091.
[8] S.D. Mhlanga, N.J. Coville, Ironecobalt catalysts synthesized by a reverse
micelle impregnation method for controlled growth of carbon nanotubes,
Diam. Relat. Mater. 17 (2008) 1489e1493.
[9] S.D. Mhlanga, K.C. Mondal, R. Carter, M.J. Witcomb, N.J. Coville, The effect of
synthesis parameters on the catalytic synthesis of multi-walled carbon
101
nanotubes using Fe-Co/CaCO3 catalysts, S. Afr. J. Chem. 62 (2009) 67e76.
[10] A.S. Afolabi, A.S. Abdulkareem, S.D. Mhlanga, S.E. Iyuke, Synthesis and puri-
fication of bimetallic catalyzed carbon nanotubes in a horizontal CVD reactor,
J. Exp. Nanosci. 6 (3) (2011) 248e262.
[11] M.A.M. Motchelaho, H. Xiong, M. Moyo, L.L. Jewell, N.J. Coville, Effect of acid
treatment on the surface of Multi-Walled Carbon nanotubes prepared from
FeeCo supported on CaCO3: correlation with FischereTropsch catalyst ac-
tivity, J. Mol. Catal. Chem. 335 (2011) 189e198.
[12] N. Chiwaye, L.L. Jewell, D.G. Billing, D. Naidoo, M. Ncube, N.J. Coville, In situ
powder XRD and Mo €ssbauer study of FeeCo supported on CaCO3, Mater. Res.
Bull. 56 (2014) 98e106.
[13] Z. Li, E. Dervishi, Y. Xu, X. Ma, V. Saini, A.S. Biris, Effects of the FeeCo inter-
action on the growth of Multiwall Carbon nanotubes, J. Chem. Phys. 129
(2008), 074712.
[14] A. Pander, A. Hatta, H. Furuta, Optimization of catalyst formation conditions
for synthesis of Carbon Nanotubes using Taguchi Method, Appl. Surf. Sci.
(2016) 1e37.
[15] S. Sakurai, H. Nishino, D.N. Futaba, S. Yasuda, T. Yamada, A. Maigne, Role of
subsurface diffusion and ostwald ripening in catalyst formation for single-
walled carbon nanotube forest growth, J. Am. Chem. Soc. 134 (2012)
2148e2153.
[16] S. Porro, S. Musso, M. Giorcelli, A. Chiodoni, A. Tagliaferro, Optimization of a
thermal-CVD system for carbon nanotube growth, Phys. E Low-dimens. Syst.
Nanostruct. 37 (2007) 16e20.
[17] W.M. Yeoh, K.Y. Lee, S.P. Chai, K.T. Lee, A.R. Mohamed, Synthesis of high purity
multi-walled carbon nanotubes over Co-Mo/MgO catalyst by the Catalytic
Chemical Vapour Deposition of Methane, N. Carbon Mater. 24 (2) (2009)
119e123.
[18] F. Taleshi, Evaluation of new processes to achieve a high yield of carbon
nanotubes by CVD method, Int. Nano Lett. 2 (23) (2012) 1e5.
[19] Z. Zarnegar, J. Safari, Green chemistry-mediated synthesis of benzil by nano-
MgO, J. Exp. Nanosci. 10 (9) (2015) 651e661.
[20] A. Trunschke, Modern Methods in Heterogeneous Catalysis Research: Surface
Area and Pore Size Determination. Power point Presentation, 2014. Quan-
tachrome Surface area determination. Porous Material Primers - Quantach-
rome Instruments Primer Series.
[21] W.T. Tsai, Microstructural characterization of calcite-based powder materials
prepared by planetary ball milling, Materials 6 (2013) 3361e3372.
[22] J. Haber, J.H. Block, B. Delmon, Manual of methods and procedures for catalyst
characterization (Technical Report), J. Pure. Appl. Chem. 67 (1995) 1257e1306
(1995).
[23] J.G. McCarty, M. Gusman, D.M. Lowe, D.L. Hildenbrand, K.N. Lau, Stability of
supported metal and supported metal oxide combustion catalysts, Catal.
Today 47 (1999) 5e17.
[24] M. Morbidelli, A. Gavriilidis, A. Varma, Catalyst Design: Optimal Distribution
of Catalyst in Pellets, Reactors, and Membranes, Cambridge University Press,
Cambridge, U.K, 2001.
[25] N. Chiwaye, In Situ X-Ray Diffraction Analysis of Fischer Tropsch Synthesis,
MSc Thesis, University of the Witwatersrand, Johannesburg, 2012.
[26] A. Aliyu, A.S. Abdulkareem, A.S. Kovo, O.K. Abubakre, J.O. Tijani, I. Kariim,
Synthesize multi-walled carbon nanotubes via catalytic chemical vapour
deposition method on Fe-Ni bimetallic catalyst supported on kaolin, Carbon
Lett. 21 (2017) 33e50.
[27] F. Ramezanipour, J.E. Greedan, A.P. Grosvenor, J. Britten, L.M.D. Cranswick,
V.O. Garlea, Intralayer cation ordering in a brownmillerite superstructure:
synthesis, crystal, and magnetic structures of Ca2FeCoO5, Chem. Mater. 22
(2010) 6008e6020.
[28] S. Haffer, Mesoporous Spinel-type Cobalt Oxide, Cobalt Ferrite and Alumina by
Nanocasting, PhD Thesis, Paderborn University, Paderborn, 2013.
[29] Z. Surowiec, M. Wiertel, M. Budzyn  ski, W. Gac, Synthesis and characterization
of iron-cobalt nanoparticles embedded in mesoporous silica MCM-41,
Nukleonika 58 (1) (2013) 87e92.
[30] L. Chang, Effects of stirring speed on the performance of Pd/Al2O3 catalysts for
methane catalytic combustion, J. Ind. Catal. 23 (1) (2015) 45e49.
[31] V.P. Woodward, Z. Amjad, Microscopy and X-Ray-based Analytical Techniques
for Identifying Mineral Scales and Deposits, Mineral Scales in Biological and
Industrial Systems, 2014, pp. 387e405.
[32] S.A. Khorrami, Q.S. Manuchehri, Magnetic properties of cobalt ferrite syn-
thesized by hydrothermal and Co-Precipitation methods: a comparative
study, J. Appl. Chem. Res. 7 (3) (2013) 15e23.
[33] G. Yang, H. Han, T. Li, C. Du, Synthesis of nitrogen-doped porous graphitic
carbons using nano-CaCO3 as template, graphitization catalyst, and activating
agent, Carbon 50 (2012) 3753e3765.
[34] M. Escobar, M.S. Moreno, R.J. Candal, M.C. Marchi, A. Caso, P.I. Polosecki,
G.H. Rubiolo, S. Goyanes, Synthesis of carbon nanotubes by CVD: effect of
acetylene pressure on nanotubes characteristics, Appl. Surf. Sci. 254 (2007)
251e256.
[35] H. Igarashi, H. Murakami, Y. Murakami, S. Maruyama, N. Nakashima, Purifi-
cation and characterization of zeolite-supported single-walled carbon nano-
tubes catalytically synthesized from ethanol, Chem. Phys. Lett. 392 (2004)
529e532.
[36] J. Zhu, M. Yudasaka, S. Iijima, A catalytic chemical vapor deposition synthesis
of double-walled carbon nanotubes over metal catalysts supported on a
mesoporous material, Chem. Phys. Lett. 380 (2003) 496e502.
[37] T. Hiraoka, T. Kawakubo, J. Kimura, R. Taniguchi, A. Okamoto, T. Okazaki,
102 M.T. Bankole et al. / Journal of Alloys and Compounds 749 (2018) 85e102
T. Sugai, Y. Ozeki, M. Yoshikawa, H. Shinohara, Selective Synthesis of Double-
Wall Carbon nanotubes by CCVD of acetylene using zeolite supports, Chem.
Phys. Lett. 382 (2003) 679e685.
[38] B.C. Liu, B. Yu, M.X. Zhang, Catalytic CVD Synthesis of Double-Walled carbon
nanotubes with a narrow distribution of diameters over Fe-Co/MgO catalyst,
Chem. Phys. Lett. 407 (2005) 232e235.
[39] H. Kathyayini, N. Nagaraju, A. Fonseca, J.B. Nagy, Catalytic activity of Fe, Co and
Fe/Co supported on Ca and Mg oxides, hydroxides and carbonates in the
synthesis of carbon nanotubes, J. Mol. Catal. Chem. 223 (2004) 129e136.
[40] Z. Yu, D. Chen, B. Tøtdal, A. Holmen, Effect of Catalyst preparation on the
Carbon nanotubes growth rate, Catal. Today 100 (2005) 261e267.
[41] Y. Murakami, Y. Miyauchi, S. Chiashi, Maruyama, S; characterization of single-
walled carbon nanotubes catalytically synthesized from alcohol, Chem. Phys.
Lett. 374 (1e2) (2003) 53e58.
[42] T.C. Schmitt, A.S. Biris, D.W. Miller, A.R. Biris, D. Lupu, S. Trigwell, Z.U. Rahman,
Analysis of effluent gases during the CCVD growth of Multi-Wall Carbon
nanotubes from acetylene, Carbon 44 (2006) 2032e2038.
[43] Z. Balogh, G. Halasi, B. Korbely, K. Hernadi, CVD-synthesis of multiwall carbon
nanotubes over potassium-doped supported catalysts, Appl. Catal. Gen. 344
(2008) 191e197.
[44] C.H. See, A.T. Harris, CaCO3 supported Co-Fe catalysts for Carbon Nanotubes
Synthesis in fluidized bed reactors, Particle Technology and Fluidization 54 (3)
(2008) 657e664.