Biomass and Bioenergy 90 (2016) 101e113

Contents lists available at ScienceDirect

Biomass and Bioenergy

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

Research paper

Effects of heat treatment and acid washing on properties and

reactivity of charcoal
Asem Hussein a, b, Faïcal Larachi c, Donald Ziegler d, Houshang Alamdari a, b, *
a
Department of Mining, Metallurgical and Materials Engineering, Universite Laval, 1065 Avenue De la M edecine, Quebec, QC, G1V 0A6, Canada
b
NSERC/Alcoa Industrial Research Chair MACE3 and Aluminum Research Centre e REGAL, Universit e Laval, 1065 De la Medecine Ave., Qu
ebec, QC, G1V 0A6,
Canada
c
Department of Chemical Engineering, Universite Laval, 1065 avenue de la Medecine, Quebec, QC, G1V 0A6, Canada
d
Alcoa Primary Metals, Alcoa Technical Center, 100 Technical Drive, Alcoa Center, PA, 15069-0001, USA

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

Article history: Due to its highly amorphous carbon structure and abundant minerals content, as received charcoal
Received 30 November 2015 possesses several undesirable characteristics such as low density and electrical conductivity in addition
Received in revised form to its extremely high air and CO2 reactivities. These disadvantages are the most challenging obstacle for
23 March 2016
using this material as an alternative for petroleum coke in anode manufacturing processes. In this work,
Accepted 31 March 2016
heat treatment under inert conditions was found to be a useful method to improve the molecular
structure of charcoal, during this process continuous growth of the more ordered carbon structure at the
expense of the amorphous forms was detected using XRD and Raman spectroscopy. Consequently, an
Keywords:
Wood charcoal
improvement in the physical properties and the reactivity of charcoal occurred. In addition, acid washing
Petroleum coke was employed to eliminate the inorganic minerals of the charcoal. It was found that combination be-
Anode tween acid washing and heat treatment produced charcoal having lower reactivity and better physical
Acid washing properties. The burning behavior of the pretreated charcoal samples was found to be comparable to that
Carbon structure of calcined petroleum coke. Accordingly, using the pretreated charcoal to substitute up to 10% of coke in
Reactivity the anode recipe did not show a negative effect on the anode's reactivity to both air and CO2.
© 2016 Elsevier Ltd. All rights reserved.

1. Introduction environmental interest for aluminum industry.

The anode is made by mixing petroleum coke with coal tar pitch
Wood Charcoal is an attractive alternative to petroleum coke in to form a paste, which is then consolidated and baked. It is then
production of carbon anode for the aluminum smelting process. cooled, connected to the electrical busbars, and suspended in the
Calcined petroleum coke is the major component in the anode electrolyte at high temperatures ranging between 900 and 970
C.
recipe and its consumption results in a direct Greenhouse Gases The anode is consumed in the electrolyte thus supplying part of the
(GHG) footprint for the industry [1]. Charcoal, on the other hand, is energy required for the reduction of alumina. The lifetime of an
considered as a green and abundant source of sulfur-free carbon anode in the electrolyte is around 25 days. Typically to produce one
with a massive worldwide production of more than 50 Mt per kg aluminum, 0.41 kg of carbon is required, generating about 1.5 kg
annum [2]. Using charcoal in anode production is however a big CO2 [3].
challenge since it does not meet the specifications required for The price of anode-grade petroleum coke is continuously esca-
anode making. Charcoal is characterized by its low density and high lating while its quality is incessantly degrading. Less than 20% of
electrical resistivity in addition to its extremely high air and CO2 petroleum coke production meets anode specifications. These
reactivities [1]. Any solution for these issues allowing substitution specifications are essentially based on their upper limit of sulfur,
of petroleum coke by charcoal may thus be of great economical and heavy metal contents (specially Vanadium), density, molecular
structure as well as mechanical properties (e.g. Hardgrove Grind-
ability Index (HGI)). Reflecting changes in the refinery industry [4],
* Corresponding author. Department of Mining, Metallurgical and Materials En- coke supplies are becoming higher in sulfur and heavy metals
 Laval, 1065 Avenue De la Me
gineering, Universite decine, Quebec, QC, G1V 0A6,
content with higher isotropic structure. In addition, the spectacular
Canada.
E-mail address: [email protected] (H. Alamdari). increase in aluminum production volume and almost constant level

http://dx.doi.org/10.1016/j.biombioe.2016.03.041
0961-9534/© 2016 Elsevier Ltd. All rights reserved.
102 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113

of production for petroleum coke imbalances the supply and de- leaching conditions. Afterwards, the acid-washed samples were
mand equilibrium resulting in its escalating price. calcined at a temperature similar to that for the green petroleum
An attempt to investigate the possibility of using charcoal as a coke, i.e., 1300
C. Anode samples were made by substituting a
substituent of petroleum coke in anodes for aluminum production portion of the coke fine fraction by the thus-prepared charcoal. The
was done by Monsen et al. [1]. The results showed deterioration in pretreated charcoal as well as the resulting anode samples were
both air and CO2 reactivities of the anode samples contain charcoal. characterized using helium density measurements, surface area
The porous structure together with high alkali and alkaline earth analysis, XRD, XRF and Thermogravimetric analyses (TGA) in order
metal contents (e.g. Na, K and Ca) make charcoal extremely reactive to assess their potential application in anode manufacturing.
with respect to air and CO2. They finally concluded that addition of
charcoal to the anode recipe had deleterious effect on the final 2.2. Raw materials
anode properties.
Raw charcoal has amorphous carbon structure consisting of Commercially available Maple wood charcoal was used as the
short and distorted aromatic layers. Heat treatment (HT) under charcoal source and industrial calcined petroleum coke (CPC),
inert conditions is a useful method to improve its molecular provided by Alcoa Inc., was used as the reference. Ash content and
structure. During this process continuous growth of the more or- metallic elemental composition of the wood charcoal and petro-
dered carbon structure at the expense of the amorphous forms was leum coke are shown in Table 1. In order to destroy its very porous
proven using X-ray diffraction (XRD) [5e7] and Raman spectros- structure, charcoal pieces were ground in a ball mill then sieved to
copy [8]. As the carbon structure of charcoal is continuously separate the fraction with a particle size of less than 400 mesh
improved by increasing the heat treatment temperature, remark- (<0.037 mm). CPC was also crushed and classified into different size
able changes in its physical properties occur, e.g. as real density (as fractions and the fraction with a particle size of less than 400 mesh
measured by Helium Pycnometry) and specific surface area (as (<0.037 mm) was selected.
measured by N2 or CO2 adsorption-BET). Considerable increase in
real density of charcoal occurs as the HT temperature is increased 2.3. Heat treatment
[6,7]. A maximum real density of 2.0 g cm3 was reached at
1000
C. Upon heating the cellulose carbon up to 1300
C, a sig- Raw charcoal (10 g) was loaded in a ceramic boat, placed in a
nificant reduction in its surface area was observed. This behavior tubular furnace and heated at a constant heating rate of 10 K min1
could be explained by the conversion of open micropores to closed to the target heat treatment temperature, and then kept at this
micropores as a result of the structural ordering at elevated tem- temperature for 4 h. The furnace was then switched off and the
perature [5]. sample was kept in the furnace to cool down to room temperature.
Reactivity of carbonaceous materials can be reduced by During the heat treatment process argon was delivered into the
increasing HT temperature. This is attributed to the structural mullite tube at a flow rate of 1.0 dm3 min1, thereby ensuring that
ordering during heat treatment. As HT temperature is increased, the whole process, including heating-up and cooling-down steps, is
the concentration of the more ordered and condensed carbon in- performed in an inert environment. A wide range of heat treatment
creases at the expense of the amorphous phase. Due to the struc- temperatures, covering between 600 and 1400
C, was used
tural ordering, the concentration of edge carbon atoms (high following the same procedure.
reactive form of carbon atoms) decreases, resulting in lesser
amount of active sites and thus lowering the reactivity to oxidative 2.4. Acid washing
gases [9]. In addition, surface area of the carbonaceous materials
decreases by increasing the HT temperature, contributing in a Raw charcoal was washed with various hydrochloric acid (HCl)
reduction of its reactivity [10]. concentrations (0.01, 0.1 and 1 mol dm3). For each run, a sample of
Charcoal is also known to contain a range of inorganic minerals 10 g was soaked in a volume of 100 cm3 of acid in a round bottom
such as phosphorus, potassium, sodium, calcium, magnesium and flask at 65
C for 3 h under continuous stirring. The mixture was
silicon [11]. Indeed, inorganic impurities increase charcoal reac- then filtered using a filter paper and washed with hot deionized
tivity to air and CO2 at high temperature due to their catalytic ef- water until no chlorine was detected in the filtrate using AgNO3 as
fects [12,13]. A process of demineralization of charcoal by washing an indicator. The recovered charcoal was then dried overnight at
with a chemical solution was proposed to reduce its content of 100
C. Hereafter, the raw charcoal, charcoals washed using 0.01, 0.1
inorganic minerals [14]. The samples which were washed by min- and 1 mol dm3 HCl are named as RC, HL0.01, HL0.1 and HL1,
eral acids showed lower reactivity [15]. respectively.

2. Material and methods 2.5. Combination between acid washing and heat treatment
processes
2.1. Methodology
The acid-washed samples were calcined at 1300
C in order to
This work attempts to explore the effect of heat treatment and investigate the simultaneous effect of both acid washing and HT.
acid washing on the reactivity of charcoal with the aim to elaborate That temperature was selected to be the same as that used indus-
alternative and appealing materials that approach the specifica- trially to calcine the green petroleum coke. Hereafter, charcoal
tions required for anode manufacturing. At high temperature the washed using 0.01, 0.1 and 1 mol dm3 HCl and then calcined at
structure of charcoal could be ordered to some extent. The acid 1300
C are named as HL0.01.C, HL0.1.C and HL1.C, respectively.
washing may remove the inorganic impurities, which are docu-
mented to act as catalysts for carbon oxidation reactions. Overall, it 2.6. Laboratory pilot scale anode preparation
is expected that combined heat treatment and acid washing will
result in charcoal with better properties and lower air and CO2 Laboratory scale anodes were prepared using petroleum coke
reactivities. with specific size fractions, as listed in Table 2, and coal tar pitch.
To fulfill the objective, wood charcoal samples were washed The reference anode was made completely of petroleum coke.
using different concentrations of HCl to determine the optimum Different coke fractions were individually weighed and mixed with
 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113 103

Table 1
Inorganic concentration and ash content in one kg of raw charcoal and industrial petroleum coke.

Na (mg) K (mg) Ca (mg) Si (mg) P (mg) V (mg) Fe (mg) Ni (mg) S (g) Ash (g)

Raw charcoal 257 4672 3787 388 258 0 253 10 0 19

Pet. Coke 100 27 130 120 11 360 460 250 21.3 3.5

Table 2
Size distribution of coke particles.

Particle size (US mesh) 4 þ 8 8 þ 14 14 þ 30 30 þ 50 50 þ 100 100 þ 200 200 þ 400 400
Particle size (mm) 4.76 þ 2.38 2.38 þ 1.41 1.41 þ 0.595 0.595 þ 0.297 0.297 þ 0.149 0.149 þ 0.074 0.074 þ 0.037 0.037
Mass fraction (%) 21.8 10 11.5 12.6 9 10.6 14.5 10

coal tar pitch. A pitch/coke mass ratio of 16.2/100 was used for all DubinineRadushkevich (DR) equation was used to determine the
samples. For each sample, the pitch and dry coke were placed in the specific surface area. The measurements were performed three
mixing pane and preheated at 178
C in an oven for 30 min. The times for each gas and the values for a sample did not differ by more
preheated materials were then mixed in a Hobart N50 mixer for than 4 m2 g1.
10 min to produce the green anode paste. The mixer was installed XRD patterns were recorded on a Siemens D5000 X-ray
in the preheating oven to ensure a uniform temperature. The pastes diffractometer at room temperature with Cu Ka radiation
were compacted at 150
C in a rigid cylindrical steel mold having an (l ¼ 1.54059 Å). The diffraction patterns were recorded for 2q
internal diameter of 68 mm. Compaction was performed using values comprised between 10 and 50
using a 0.02
2q step with an
60 MPa uniaxial pressure. The preheating, mixing and pressing integration time of 1.2 s.
parameters were selected based on the previous works [16,17]. X-ray fluorescence (XRF) was used to determine the chemical
To produce anodes containing charcoal, a portion of the finest composition of the charcoal samples. Elemental analysis of raw and
fraction of coke (400 mesh) was substituted with pretreated acid-washed charcoal samples was carried out according to the
wood charcoal (charcoal washed with 1 mol dm3 HCl then ASTM D4326-06 (named standard test method for major and minor
calcined at 1300
C (HL1.C)) and the anode samples were prepared elements in coal and coke ash by x-ray fluorescence). The analyses
using the same procedure. Two sets of samples were prepared, were performed by Alcoa quality control laboratories situated at
having an overall mass fraction of 5 and 10% of charcoal. Hereafter, Deschambault, Que bec.
the reference anode, the anodes containing 5% and 10% pretreated Raman spectra of the samples were recorded at 22.0 ± 0.5
C
charcoal are named as Ref A, AC 5% and AC 10% respectively. using a LABRAM 800HR Raman spectrometer (Horiba Jobin Yvon,
All the compacted anodes (green samples) were baked at the France) coupled to an Olympus BX 30 fixed stage microscope. The
same time, using a maximum baking temperature of 1100
C and a excitation light source was a l ¼ 632.8 nm red line of a HeeNe Laser
soaking time of 20 h. The baked samples were cut into small disks (Melles Griot, Carlsbad, CA). The laser beam was focused using a
and crushed in a jaw crusher into very small pieces. The crushed 100 objective. Before starting the analysis, the Raman spectrom-
anodes were fed into a ball mill to produce fine anode powder. The eter was calibrated with a silicon line at 521 cm1. Samples were
anode powder was then sieved and the fraction of particle spread on a glass slide for analysis. For each sample, about 10 mg
size 400 mesh (0.037 mm) was selected for air and CO2 reac- were randomly chosen and analyzed. The spectra were recorded in
tivity tests. the range of 800e2000 cm1, covering first-order bands. The
acquisition time was between 2 and 4 min. Quantitative parameters
from Raman spectra can be obtained from the decomposition of the
2.7. Characterization techniques spectra by fitting procedures. From the bands position, their in-
tensity and their integrated area, different ratios were calculated in
The real density of the charcoal samples was measured using order to determine the structural changes in charcoal during heat
helium pycnometer (Micromeritics, AccuPyc II 1340). Before the treatment. The Raman spectra in the range between 800 and
analysis, the samples were dried overnight at 100
C. The dried 2000 cm1 were curve-fitted with 9 Gaussian bands by the GRAMS/
samples were then kept in a desiccator to cool to room temperature 32 AI software (version 8.00). This assignment of the 9 bands was
and their volumes were measured by helium pycnometry. Helium done based on the work of Li et al. [18].
penetrates into the accessible pores and thus subtracts the volume Thermogravimetric analysis (TGA) was used to determine the air
of open pores from the calibrated volume of the empty cell. The and CO2 reactivities of charcoal, coke and anode powder samples. A
measured volume consists of the volume of the material and the TGA (NETZSCH STA 449 F3 Jupiter, Germany), equipped with a
volume of closed pores. The real density was calculated based on platinum furnace, being able to reach 1500
C was used for ana-
the ratio of the sample mass to the volume measured by the pyc- lyses. In order to compare the reactivities of all the samples, the test
nometer. The measurements were repeated three times and the conditions were fixed. Charcoal and coke particles were sieved and
values did not differ by more than 0.01 g cm3. the particles passing 400 mesh sieve (0.037 mm) were selected
To determine the specific surface area of charcoal samples, all for air and CO2 reactivity tests. Only 2 mg of the powder was spread
the samples were degassed at 300
C for 15 h under a continuous as an even layer in the alumina crucible. The air reactivity tests
flow of nitrogen. Both N2 (Praxair, purity: 99.995%, USA) and CO2 were performed at 525
C while those with CO2 were measured at
(Praxair, purity: 99.9%, USA) gases were used to determine the 960
C. The sample was heated from room temperature (RT) up to
specific surface area. In the case of N2 adsorption, a Micromeritics the desired final temperature at a heating rate of 20 K min1 under
(TriStar II) surface analyzer was used and the specific surface area N2 atmosphere with a flow rate of 100 cm3 min1 to prevent its
was calculated according to Brunauer-Emmett-Teller (BET) oxidation during the heating stage. Once the final temperature was
model at 196
C. For CO2 adsorption, the sample holder was reached, the sample was kept under these conditions for 10 min
submerged in a mixture of water and ice at 0
C and
104 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113

before switching N2 to the oxidative gas with a constant flow rate of 3.2. Raman spectroscopy
100 cm3 min1. The reaction was stopped when the sample was
completely consumed. The conversion was calculated using the Raman spectroscopy is generally used to study the effect of heat
following expression: treatment on the structural evolution of carbon materials. Usually,
two main bands are observed for the highly-ordered carbonaceous
mðtÞ materials: the G-band (Graphite band) which appears at around
XðtÞ ¼  100 (1)
mðtoÞ 1585 cm1 due to the vibrations of the sp2 bonded carbon within
the perfect crystalline carbon, and the D-band (also called disorder
where m(t) is the mass at time t and m (t0) is the initial mass. band), which appears at 1350 cm1 due to the vibrations of sp2
SEM images were taken on a JSM-840A (JEOL, Japan) instrument bonded carbon with an amorphous structure [19].
equipped with secondary and backscattered electron detectors. In The Raman spectra, obtained for the raw and the heat-treated
order to perform a non-destructive elemental analysis for localized charcoal samples at temperatures between 600 and 1400
C, are
areas, the instrument was also coupled with Energy-Dispersive X- shown in Fig. 2. Raman spectra of charcoal samples have two very
ray Spectroscopy (EDS). The working voltage was 15 KV. broad bands situated in the region of 1600 and 1335e1345 cm1,
close to the position of the previously described G and D bands.
3. Results and discussion Despite the X-ray diffraction (XRD) patterns of charcoal samples
not showing any evidence for the existence of graphitic crystalline
3.1. XRD diffraction structure (even for the sample heated at 1400
C), all the samples
had a clear band at the G position of the Raman spectra. This band
The X-Ray Diffractogram of highly-ordered carbon is known to can be assigned to the molecular vibration of the aromatic ring
present two sharp peaks. The first one (002) appears at 2q ¼ 26
; breathing. In addition, the D band refers to the presence of aromatic
being attributed to the stacking of the graphitic basal plane (Lc). clusters having 6 or more fused benzene rings [18].
The second one (10) occurs at 2q ¼ 44
; referring to the presence of The raw charcoal sample showed the highest Raman intensity.
graphene sheets within a single plane (La). This can be attributed to the presence of oxygen-containing groups,
A comparison of the X-ray diffraction patterns for the raw such as C]O, which are more sensitive to Raman vibration. As the
charcoal and the heat-treated samples at 1400
C is presented in HT temperature was increased, the smallest polyaromatic struc-
Fig. 1. In case of raw charcoal, only a very broad featureless (002) tures, which may contain aliphatic and oxygenated groups, were
peak was detected at 2q ¼ 23
indicating highly disordered carbon significantly reduced and the charcoal produced would become
structure. For the sample heated at 1400
C, the (002) band became increasingly aromatic. Both the reduction of Raman-sensitive
more defined. In addition, a new (10) peak appeared at 2q ¼ 43
. groups and the augmentation of the aromatic nature of the chars
These observations indicating continuous improvement in carbon increased their ability to absorb the Raman spectra, decreasing
structure of charcoal by HT [6]. The size of graphene sheets in- their intensity [8,18].
creases as the HTT is increased, due to the conversion of amorphous
carbon into more organized carbon structure. Although, the pres-
ence of graphene sheets was detected in the heat-treated charcoal, 3.3. Curve-fitting of Raman spectra
its XRD spectra did not show any convincing sign of the presence of
graphitic crystallite structure. The curve fitting method is usually employed to interpret the

Fig. 1. XRD patterns of raw charcoal (RC) and of charcoal heat-treated at 1400
C.
 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113 105

Fig. 2. Effect of heat treatment temperature (between 600 and 1400
C) on Raman spectra of charcoal.

results of Raman spectra. If only G and D wide bands are used in the The band positions were fixed whilst the bandwidths were
curve fitting, a large amount of spectral residue would not be adjusted to different maximum limits until a perfect fitting was
considered and thus the information about the structural evolution obtained. Examples of spectral curve-fitting of a raw charcoal and of
during heat treatment would not be very precise. Therefore, Raman charcoal heated at 1400
C Raman spectra are shown in Figs. 3 and
spectra of chars were curve-fitted with 10 bands representing the 4, respectively. This demonstrated that a Raman spectrum could be
major structures in the chars. successfully curve-fitted with these 9 bands. A similar success of
Raman spectra in the range between 800 and 2000 cm1 were curve fitting was obtained for all other charcoal samples investi-
curve-fitted with only 9 Gaussian bands. Most of the bands listed gated in this work.
elsewhere in the literature [18] were observed; however, the po- Out of the 9 bands employed in the curve-fitting of the charcoal
sition of all these bands appeared at a different Raman shift. Raman spectra of charcoal, only six of them were considered as
Therefore, the 9 Gaussian bands were located in their new positions main bands for all char samples (G, GR, VL, VR, D and S bands). The
according to the spectra of our charcoal samples. The positions of areas of GR, VL and VR bands were treated as one band representing
these bands are summarized in Table 3. the overlap between the G and D bands and thus our discussion on
In an attempt to curve-fit the Raman spectra of charcoal, the Raman data mainly focused on three groups of bands, e.g. G,
valley, which appeared between the G and D bands, was covered by (GR þ VL þ VR), and D. The other four less important bands (S, GL,
three bands, such as a GR band (standing for G right), a VR band SL and SR) were used to reduce the width of the main bands during
(standing for valley right) and VL band (standing for valley left). curve-fitting.
These three bands can represent the aromatic ring systems typi-
cally found in amorphous carbon materials, their combined in- 3.3.1. Ratio between D and G bands
tensity (GR þ VL þ VR) is believed to decrease with the aromatic The ratio between the intensity (peak area) of the D and G
ring's growth [18]. Some extra bands were used as GL (standing for bands, ID/IG, is a useful method to follow the structural ordering in
G left), S band (meaning ‘on the right side of the D band’), SL a graphitizing carbon during the heat treatment. This ratio usually
(standing for S left) and SR (standing for S right). decreases when the heat treatment temperature is increased,
implying an augmentation in the degree of graphitization. How-
ever, with charcoal, a significant increase in this ratio was observed
Table 3
by increasing the heat treatment temperature between 600 and
Positions of Raman bands used for curve fitting process.
800
C (Fig. 5). This behavior indicated an increase in the concen-
Band name Band position (cm1) tration of the condensed aromatic clusters having six or more fused
GL 1685 benzene rings. According to XRD results, the presence of condensed
G 1600 aromatic systems (graphene sheets) was detected at the same
GR 1530
temperature (800
C). Further increase in the treatment tempera-
VL 1473
VR 1412 ture resulted in little change in this ratio between 800 and 1300
C
D 1345 and in a slight reduction at 1400
C. This behavior indicated that a
SL 1255 continuous ring condensation could produce larger aromatic clus-
S 1172 ters and also eliminate various forms of the structural defects and
SR 1065
imperfections of the carbon crystallites. These results were in
106 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113

Fig. 3. Curve-fitting of a Raman spectrum of the raw charcoal sample.

Fig. 4. Curve-fitting of a Raman spectrum of the charcoal sample heated at 1400
C.

accordance with the XRD results since the (002) and (10) bands having less than 5 fused benzene rings. Their combined intensity
became more defined with increasing the treatment temperature, (GR þ VL þ VR) decreased with an increasing in the size of the
indicating that an improvement in the carbon structure occurred as aromatic clusters. By using the Raman parameter ID/
the HTT was increased. I(GR þ VL þ VR) it is possible to evaluate the molecular uniformity
of the carbon structures. As shown in Fig. 6, the continuous increase
in this ratio with an augmentation in the HT temperature indicated
3.3.2. Ratio between D and (GR þ VL þ VR) bands
a transition from small to large aromatic ring system.
Using the information extracted from the Raman spectra, three
bands (GR, VL and VR) were employed to cover the overlap be-
tween the D and G bands. They can represent aromatic ring clusters
 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113 107

Fig. 5. ID/IG band ratio as a function of heat treatment temperature.

Fig. 6. ID/I (GR þ VL þ VR) bad ratio as a function of heat treatment temperature.

3.4. Effect of heat treatment temperature on the physical properties the samples, heat-treated up to 800
C, both CO2 and N2 mea-
and morphology of charcoal surements showed an increase in the surface area due to the evo-
lution of volatiles at this temperature range. While the first
3.4.1. Specific surface area significant reduction in N2 surface area was detected at 900
C, the
Fig. 7 shows the effect of heat treatment temperature on the CO2 surface area was not affected. A relatively continuous reduction
surface area of charcoal. For the untreated sample, the surface area, of two orders of magnitude in the N2 surface area was found be-
measured by CO2 adsorption, was higher than that with N2 tween 800 and 1200
C and between 900 and 1000
C for CO2. The
adsorption; this implies that the sample had high amounts of most pronounced drop in CO2 surface area was observed at 1200
C.
microporosity, being missed by N2 adsorption [20]. The same ten- At this temperature, both N2 and CO2 measurements exhibited the
dency was observed for all samples up to 1000
C. The difference same surface area, which means that the micropores were
between both analyses at higher temperatures was negligible. For completely closed. Pronounced drop in CO2 surface area was
108 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113

Fig. 7. N2 and CO2 surface area of charcoal as a function of heat treatment temperature.

observed at 1200
C. At this temperature, both N2 and CO2 mea- the particles exhibit grooves in their inner surface. The very large
surements exhibited the same surface area, which means that the specific surface area of charcoal was attributed to the presence of
micropores were completely closed. the microporosity (pore size <2 nm) thus the contribution of the
The reduction in the surface area of charcoal at elevated tem- pores observed in Fig. 9 to the overall specific surface area of the
perature was already reported in the literature by Kercher and charcoal would be negligible. In SEM image of the sample heat-
Nagle [7]. They believed that a continuous growth of graphene treated at 1300
C, as shown in Fig. 9c, the small debris are most
sheets that occurred at elevated temperature due to the structural likely resulted from the particles fracture during ball milling. The
ordering could be responsible for the micropore coalescence and general aspects of the raw and heat-treated samples are very
the reduction in the surface area. By XRD and Raman spectroscopy similar. It seems that, in this length-scale, no significant change
we confirmed their observations. happened in the morphology of the charcoal particles upon heat
treatment. In other words, the particles of the starting material kept
the same shape even at elevated temperature.
3.4.2. Real density (helium density)
Helium pycnometry was used to determine the real density of
raw and heat-treated charcoal samples. The changes in the real
3.5. Effect of heat treatment on CO2 and air reactivities of charcoal
density of charcoal samples as a function of the final heat treatment
temperature can be followed in Fig. 8. The real density increased
In order to study the effect of HT temperature on CO2 reactivity
with the temperature and reached its maximum value of
of charcoal, the conversion percentage as a function of the reaction
2.095 g cm3 at 1000
C. This behavior, as proven by XRD and
time is shown in Fig. 10 for the charcoal samples heat-treated at
Raman spectra, can be explained by the growth of more condensed
different temperatures varying from 1000 to 1400
C. For com-
and ordered carbon structure at the expense of the amorphous
parison, the conversion profile of calcined petroleum coke (CPC)
phase.
was included in the same figure. In order to mimic the aluminum
Although the continuous structural ordering was evidenced, a
electrolysis cell conditions, CO2 reactivity of all the samples were
further increase in the treatment temperature between 1000 and
measured at 960
C. The conversion (gasification) percentage was
1400
C resulted in a reduction in the real density from 2.095 to
calculated using Eq. (1). As the samples were heated to 960
C and
1.551 g cm3. This value was lower than that of the raw charcoal
held at this temperature for 10 min before the beginning of the
sample (1.663 g cm3). Surface area measurements showed a sig-
reaction, the raw charcoal sample could then be subjected to a
nificant reduction in the charcoal samples heated at temperatures
thermal annealing and thus some structural changes could happen
higher than 1000
C (Fig. 7). Closing of some micropores during
under these conditions. Since the raw charcoal was no more
structural ordering, preventing the penetration of helium, could be
considered as a “virgin” material, therefore its CO2 reactivity pro-
responsible for the reduction in real density of charcoal at elevated
files are not included in this figure.
temperature.
It is clearly seen in Fig. 10 that the reactivity of charcoal samples
decreased progressively with an increment in the HT temperature.
3.4.3. Changes in the morphology of charcoal particles Only 5 min of the reaction were sufficient to consume all the carbon
The raw charcoal sample appeared to be a relatively porous content of the charcoal sample treated at 1000
C. In contrast, the
material as shown in Fig. 9a & b. Although charcoal samples were sample heated at 1400
C could resist the reaction conditions for
crushed to destroy their porous structure, some particles still kept 25 min before being completely burnt off. The petroleum coke
their original shape and appeared to have a very porous and or- sample (CPC) showed the lowest reactivity and was consumed after
dered network. However, as a result of the damaged pores, most of 100 min.
 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113 109

Fig. 8. Effect of heat treatment temperature on the charcoal's real density.

Fig. 9. SEM images of the raw charcoal sample (a, b) and a sample heat-treated at 1300
C.

A relationship between the conversion percentage and the re- treated at 1400
C. Again, the calcined petroleum coke sample
action time with air for the charcoal samples heat-treated at tem- showed the lowest air reactivity.
peratures comprised between 600 and 1400
C and the calcined The decreasing trend in the reactivity of charcoal to both gases,
petroleum coke is presented in Fig. 11. Although air reactivity air and CO2, can be explained by the structural changes that
generally showed a continuous decrease as the heat treatment occurred at high temperature. The previous results of XRD analyses
temperature was increased, a slight improvement was observed for revealed a continuous growth of the graphene sheets by increasing
the samples treated at temperatures between 600 and 1000
C, the heat treatment temperature. Furthermore, Raman spectra
while the most significant improvement was exhibited by those showed an increasing trend in the band intensity ratio ID/
110 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113

Fig. 10. CO2 reactivity (conversion-time data) of calcined petroleum coke (CPC) and of charcoal samples heat-treated between 1000 and 1400
C.

Fig. 11. Air reactivity (conversion percentage vs time) of calcined petroleum coke (CPC) and of charcoal samples heat-treated between 600 and 1400
C.

I(GR þ VL þ VR) implying a continuous conversion of the amor- catalysts.

phous carbon into large and more condensed aromatic ring sys-
tems. The condensed carbon structure contains less active sites,
available for the oxidative gas, than the amorphous carbon. In 3.6. Effect of acid concentration on the mineral content and
addition, the continuous reduction in surface area of charcoal reactivity of charcoal
happened as the heat treatment temperature was increased.
Consequently, the reactivity of charcoal to air and CO2 decreased With the aim at further reducing the charcoal reactivity vis-
a-vis
significantly at elevated temperature. air and CO2, the level of mineral content was reduced by acid
The lower reactivity of petroleum coke could be explained by washing prior to heat treatment. Charcoal samples were washed
the highly ordered carbon structure, lower surface area and lower with different acid concentrations (0.01, 0.1 and 1 mol dm3). A list
content of the reaction catalysts such as Ca, Na and Fe. On the other of the ash and mineral content of raw charcoal, of the samples
hand, the relatively higher reactivity exhibited by charcoal could be treated with different concentrations of hydrochloric acid is given
due to its less ordered carbon structure and to its high content of in Table 4. In general, the mineral content of the raw charcoal was
higher than that of CPC; it was richer in Ca, Na, Si and P while
 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113 111

Table 4
Inorganic concentration and ash content in one kg of charcoal samples washed with different acid concentrations.

Na (mg) K (mg) Ca (mg) Si (mg) P (mg) Fe (mg) Ni (mg) Ash (g)

HL 0.01 66 1623 2321 367 172 246 5 8.6

HL0.1 35 627 1905 346 138 90 3 6
HL1 29 800 1685 312 103 58 2 5.5

poorer in S, Ni and V elements. The concentration of all the min- reactivities of charcoal was observed for the sample heated at
erals, and consequently the ash content, was reduced as the con- 1400
C, still, its reactivity was higher than that of the calcined
centration of acid increased from 0.01 to 1 mol dm3. coke. In addition, this temperature is higher than the industrial
Fig. 12 shows air (a) and CO2 (b) reactivities of the charcoal coke calcination temperature (1300
C). Since acid washing resul-
sample washed by different concentrations of hydrochloric acid ted in a reduction on the mineral content of charcoal, the acid-
and that of the calcined coke CPC. As the acid concentration was washed samples were subjected to a further heat treatment at
increased, a reduction in both air and CO2 reactivities was noticed. 1300
C, a temperature being routinely used in aluminum industry
This could be attributed to the reduction of the minerals that are to calcine petroleum coke.
responsible for the high reactivity of the raw sample. The plots obtained for the air and CO2 reactivities of these
Since the samples were respectively heated at 525 and 960
C samples are respectively presented in Fig. 13 which also include, for
for air and CO2 reactivity measurements, a weight loss was noticed comparison, the reactivity profile of the sample only calcined at
during the heating up period before the beginning of the oxidation 1300
C and CPC. The results showed that a significant reduction in
reaction. As it can be seen in Fig. 12 (b), around 20% of the sample the charcoal reactivity was obtained by performing the acid
mass was lost at 960
C due to losing moisture and organic volatiles washing before the calcination. This could be attributed to the
at this high temperature. reduction in the content of the minerals such as Na, Ca and Fe
Although acid washing process improved the reactivity of which are known to act as catalysts for carbon gasification reactions
charcoal, its reactivity was still too high compared to that of the (Table 4). With an increment in the acid concentration an addi-
calcined coke. These results suggest that acid washing process tional reduction in the charcoal reactivity was observed. Acid
alone, even at high acid concentration, is not enough to obtain washing resulted in a significant reduction in all catalytic elements.
charcoal with appropriate properties for the anode production. However, the residual amounts have lost a lot of their catalytic
activity by interactions with other elements such as Si upon heat
treatments at high temperature [21]. Although HL1.C sample
3.7. Combined effect of acid washing and heat treatment on the
showed the least air reactivity, its CO2 reactivity was similar to that
reactivity of charcoal
of HL0.1.C sample and slightly higher than that of calcined petro-
leum coke. In addition, air reactivity profile of HL1.C became
Even if the most pronounced improvement in air and CO2

Fig. 12. Comparison between the air (a) and CO2 (b) reactivities of charcoal samples washed with different concentrations of HCl and that of raw charcoal sample (RC) and calcined
coke (CPC).
112 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113

Fig. 13. Effect of the combination of the acid washing (at different concentrations) with heat treatment on the air reactivity (a) and CO2 reactivity (b) of charcoal, for comparison, the
reactivity profiles of charcoal only calcined at 1300
C (1300) and calcined coke (CPC) are included.

comparable to that of coke proving the effectiveness of these samples, as revealed by TG analysis, is quite the same and there is
treatments. no effect of charcoal addition to the anode on its air and CO2
reactivity.
These results suggest that the pre-treated charcoal could be
3.8. Air and CO2 reactivities tests of anode powder
considered as an interesting alternative for substitution of petro-
leum coke (at least up to 10%) in anode making process. Obviously,
Since the pre-treated charcoal showed reactivity comparable or
the results presented in this work are based on the basic laboratory
better than that of petroleum coke, the pre-treated charcoal was
characterization techniques and a series of other analyses are
used in the anode recipe, substituting 5 and 10% of petroleum coke.
required before affirming that the pre-treated charcoal could be
Only the fine fraction of coke in the anode was substituted by
used for anode production. It is thus time that the pre-treated
charcoal. Fig. 14 shows the air and CO2 reactivities of the samples
charcoal is seriously considered as an alternative carbon source
with 5 and 10% of charcoal. The reference is an anode totally made
for anode production and the mechanical and electrical properties
of petroleum coke. It can be seen that the gasification behavior of all

Fig. 14. Effect of charcoal addition to the anode recipe (at different concentrations 5 (AC 5%) and 10% (AC 10%)) on its air reactivity (a) and CO2 reactivity (b) compared to that of a
reference anode (Ref A).
 A. Hussein et al. / Biomass and Bioenergy 90 (2016) 101e113 113

of the resulting anodes were further assessed. [7] A.K. Kercher, D.C. Nagle, Microstructural evolution during charcoal carbon-
ization by X-ray diffraction analysis, Carbon N. Y. 41 (2003) 15e27, http://
dx.doi.org/10.1016/S0008-6223(02)00261-0.
4. Conclusions [8] M. Asadullah, S. Zhang, Z. Min, P. Yimsiri, C.Z. Li, Effects of biomass char
structure on its gasification reactivity, Bioresour. Technol. 101 (2010)
Performing an acid washing before the calcination was found to 7935e7943, http://dx.doi.org/10.1016/j.biortech.2010.05.048.
[9] A. Arenillas, F. Rubiera, C. Pevida, C. Ania, J. Pis, Relationship between structure
be the most effective pre-treatment. For HL1.C sample a consider- and reactivity of carbonaceous materials, J. Therm. Anal. Calorim. 76 (2004)
able reduction in both air and CO2 reactivity were observed. Its air 593e602, http://dx.doi.org/10.1023/b:jtan.0000028038.34976.83.
and CO2 reaction profiles became comparable to those of the [10] L. Lu, C. Kong, V. Sahajwalla, D. Harris, Char structural ordering during py-
rolysis and combustion and its influence on char reactivity, Fuel 81 (2002)
calcined petroleum coke. The great reduction in its reactivity was 1215e1225, http://dx.doi.org/10.1016/S0016-2361(02)00035-2.
attributed to the depletion in its inorganic mineral contents by acid [11] S.G.P. Theodore, T. Kozlowski, Physiology of Woody Plants, second ed., Aca-
washing, combined with the improvement in its carbon structure, demic Press, 1996.
[12] M.P. Kannan, G.N. Richards, Potassium catalysis in air gasification of cellulosic
which became more ordered after calcination. According to reac- chars, Fuel 69 (1990) 999e1006, http://dx.doi.org/10.1016/0016-2361(90)
tivity measurements, substituting a portion up to 10% of coke with 90011-E.
pretreated charcoal in the anode recipe did not show any negative [13] T. Ganga Devi, M.P. Kannan, Calcium catalysis in air gasification of cellulosic
chars, Fuel 77 (1998) 1825e1830.
effect. [14] P. Das, A. Ganesh, P. Wangikar, Influence of pretreatment for deashing of
sugarcane bagasse on pyrolysis products, Biomass Bioenergy 27 (2004)
Acknowledgements 445e457, http://dx.doi.org/10.1016/j.biombioe.2004.04.002.
[15] E. Iniesta, F. Sanchez, A. Garcıa, A. Marcilla, Yield and CO2 reactivity of chars
from almond shells obtained by a two heating step carbonisation process.
Authors would like to acknowledge the financial support of Effect of different chemical pre-treatments and ash content, J. Anal. Appl.
NSERC and Alcoa (grant number RDCPJ47654-14). The assistance of Pyrolysis 59 (2001) 983e994. http://www.sciencedirect.com/science/article/
Alcoa team (Deschambault plant situated in Deschambault, Can- pii/S0165237000002126.
[16] F. Chevarin, L. Lemieux, D. Picard, D. Ziegler, M. Fafard, H. Alamdari, Charac-
ada) is gratefully acknowledged. The authors would also like to terization of carbon anode constituents under CO2 gasification: a try to un-
extend their appreciation to Lise Lemieux, Guillaume Gauvin and derstand the dusting phenomenon, Fuel 156 (2015) 198e210, http://
Hugues Ferland for their technical support. dx.doi.org/10.1016/j.fuel.2015.04.035.
[17] K. Azari, H. Alamdari, G. Aryanpour, D. Ziegler, D. Picard, M. Fafard,
Compaction properties of carbon materials used for prebaked anodes in
References aluminum production plants, Powder Technol. 246 (2013) 650e657, http://
dx.doi.org/10.1016/j.powtec.2013.06.025.
[1] B.E. Monsen, A.P. Ratvik, L.P. Lossius, Charcoal in anodes for aluminium pro- [18] X. Li, J.I. Hayashi, C.Z. Li, FT-Raman spectroscopic study of the evolution of char
duction, TMS Light Metals (2010) 929e934. structure during the pyrolysis of a Victorian brown coal, Fuel 85 (2006)
[2] http://faostat3.fao.org/download/F/FO/E (accessed 26.11.15). 1700e1707, http://dx.doi.org/10.1016/j.fuel.2006.03.008.
[3] H. Kvande, Production of primary aluminium, in: Fundam. Alum. Metall. Prod. [19] Y. Wang, D.C. Alsmeyer, R.L. Mccreery, Raman spectroscopy of carbon mate-
Process. Appl., Woodhead Publishing Ltd, UK, 2011, pp. 49e69. rials: structural basis of observed spectra, Chem. Mater 2 (1990) 557e563,
[4] L. Edwards, Developments in petroleum coke production and impact on http://dx.doi.org/10.1021/cm00011a018.
aluminium industry, in: Australas. Alum. Smelt. Technol. Conf. Dubai, UAE, [20] E. Cetin, R. Gupta, B. Moghtaderi, Effect of pyrolysis pressure and heating rate
2014. on radiata pine char structure and apparent gasification reactivity, Fuel 84
[5] K.J. Masters, B. McEnaney, The development of structure and microporosity in (2005) 1328e1334, http://dx.doi.org/10.1016/j.fuel.2004.07.016.
cellulose carbon, Carbon N. Y. 22 (1984) 595e601, http://dx.doi.org/10.1016/ [21] J. Xiao, F. Li, Q. Zhong, J. Huang, B. Wang, Y. Zhang, Effect of high-temperature
0008-6223(84)90095-2. pyrolysis on the structure and properties of coal and petroleum coke, J. Anal.
[6] C.E. Byrne, D.C. Nagle, Carbonized wood monolithsdcharacterization, Carbon Appl. Pyrolysis 117 (2016) 64e71.
N. Y. 35 (1997) 267e273, http://dx.doi.org/10.1016/S0008-6223(96)00135-2.
́