Upload
uhasselt
View
1
Download
0
Embed Size (px)
Citation preview
Journal of Analytical and Applied Pyrolysis 101 (2013) 199–208
Contents lists available at SciVerse ScienceDirect
Journal of Analytical and Applied Pyrolysis
journa l h o me page: www.elsev ier .com/ locate / jaap
Characterization of activated carbons derived from short rotation hardwoodpyrolysis char
Mark Stals a,c, Jens Vandewijngaarden a,c, Iwona WróbelIwaniecb, Grazyna Gryglewiczb,Robert Carleer a, Sonja Schreurs a,c, Jan Yperman a,∗
a Research Group of Applied and Analytical Chemistry, CMK, Hasselt University, Agoralaan Gebouw D, 3590 Diepenbeek, Belgiumb Institute of Chemistry and Technology of Petroleum and Coal, Wroclaw University of Technology, ul. Gdanska 7/9, 50344 Wroclaw, Polandc NuTeC, Department TIW, XIOS, Agoralaan Gebouw H, 3590 Diepenbeek, Belgium
a r t i c l e i n f o
Article history:
Received 31 August 2012
Accepted 22 January 2013
Available online 29 January 2013
Keywords:
Hardwoods
Pyrolysis char
Activated carbon
Characterization
Adsorption
a b s t r a c t
Chars from fast and flash pyrolysis of different short rotation hardwoods are activated by steam or KOH,
resulting in a variety of activated carbons (ACs). Analysis of nitrogen adsorption isotherm shows that
microporous ACs are obtained. Density functional theory calculation is applied to obtain information
concerning the micropore size distribution. Additionally, methylene blue (MB) and iodine adsorption
experiments are performed to evaluate the AC adsorptive properties. The textural and adsorptive prop
erties are compared with a commercial AC (Norit). KOH activation yields ACs with very high surface area,
micropore volume and adsorption capacity, while steam activation promotes formation of narrower
micropores but yields less pore volume. The MB adsorption kinetics can be described by the pseudo
secondorder model, and most ACs adsorb at least 70% of their maximum MB adsorption capacity within
15 min.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
Bioenergy is regarded as one of the key options to mitigate
greenhouse gas emissions and substitute fossil fuels [1]. A prime
requisite for bioenergy is the avoidance of interference with food
production [2], thus so called “second generation biofuel” from lig
nocellulosic biomass is preferred [3]. An important technique for
generating large quantities of second generation bioenergy is the
short rotation forestry of woody species [4]. Short rotation forestry
produces large amounts of lignocellulosic biomass, additionally,
the cultivation of the trees does not require intensified agricultural
methods. An interesting possibility of some hardwood species, e.g.
willow or poplar cultivars, is the removal of contaminants such as
heavy metals from the soil [5]. This use of plant species to clean up
contaminated soils is known as phytoremediation [6]. Fast or flash
pyrolysis [7,8] are techniques to convert lignocellulosic biomass
into a gaseous, liquid, and solid form, which are generally consid
ered more valuable than the biomass as such [9]. One of the key
benefits of fast pyrolysis is the possibility to produce pyrolysis oil
which contains only negligible amounts of heavy metals from heavy
metal contaminated hardwoods from phytoremediation [10,11].
∗ Corresponding author. Tel.: +32 11 268320; fax: +32 11 268301.
Email address: [email protected] (J. Yperman).
During fast or flash pyrolysis, char is generated as a byproduct,
which is usually burnt for energy. It is anticipated that large
amounts of pyrolysis char will be available [12]. Therefore, a higher
economic value than its fuel value is desired. In the present paper,
fast and flash pyrolysis chars are produced from P. nigra, S. frag
ilis and P. grimminge (from phytoremediation, except the former).
The chars are activated by steam and KOH in order to enhance the
adsorption capacity. The resultant activated carbons are character
ized by nitrogen sorption technique and density functional theory
is applied. Besides this, the adsorption isotherms and kinetics of
iodine and methylene blue (MB) are determined. A commercial
activated carbon (Norit GAC) is used as a reference. The influence of
the production route (i.e. mode of pyrolysis and mode of activation)
on the obtained mass and adsorption capacities of the activated
carbon is illustrated via mass balances.
2. Materials and methods
2.1. Hardwoods
Three shortrotation hardwood species are studied, i.e.: Wolter
son poplar (P. nigra) [4], crack willow (S. fragilis) and Grimminge
poplar (P. grimminge: P. deltoides × (P.trichocarpa × P. deltoides) [5].
All samples are chipped (< 2.0 mm) and oven dried at 110 ◦C.
The proximate and elemental analysis of these hardwood
species are given in Table 1.
01652370/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jaap.2013.01.009
200 M. Stals et al. / Journal of Analytical and Applied Pyrolysis 101 (2013) 199–208
Table 1
Proximate and elemental analysis of hardwood species for activated carbon
production.
S. fragilis P. grimminge P. nigra
Proximate analysis (%)
Volatiles 76 75 76
Fixed carbon 18 22 20
Elemental analysis (%)
C 53.5 52.9 52.8
H 6.4 6.6 6.9
N 1.2 1.0 0.5
S < < <
O 36.6 36.4 37.6
Ash 2.3 3.1 2.2
H/C 1.42 1.46 1.5
O/C 0.51 0.52 0.53
2.2. Carbonization
The carbonization step is carried out by one of the two differ
ent modes of pyrolyzing the hardwoods: (a) fast pyrolysis and (b)
flash pyrolysis. The resultant char is used as precursor material for
activation.
(a) Fast pyrolysis is done by inserting the hardwood sample (10 g)
in a quartz tube which is placed in a furnace [11] (Nabertherm).
Nitrogen is flushed at a rate of 70 ml/min through the tube to
ensure pyrolysis atmosphere. The sample is heated at 35 ◦C/min
to 450 ◦C and is then held isothermal at 450 ◦C for 15 min. After
cooling down, the char is collected and stored in a dessicator.
(b) Flash pyrolysis is done with a lab scale semicontinuous flash
pyrolysis reactor [13,14]. All pyrolysis are performed with 100 g
of biomass and at 450 ◦C pyrolysis temperature, using sand
as heat transfer medium. The biomass is injected at a rate of
40 g/min into the hot sand inside the reactor. The char is formed
in the sand matrix and after the whole system has cooled down
to room temperature, the char is separated from the sand. A
more extensive description of this setup, the studied hard
woods and the pyrolytic product streams are discussed earlier
in [14,15].
2.3. Activation
2.3.1. Physical activation with steam
The precursor material, i.e. fast or flash pyrolysis char from hard
wood biomass is milled to pass a 125 mm DIN sieve. The sieved char
is introduced into a quartz tubular reactor and the position of the
sample is fixed by quartz wool plugs. The quartz tube is flushed
with 50 ml/min N2 and inserted in the furnace which is placed in a
heated environment (110 ◦C). A thermocouple is inserted between
the tube and furnace, and the furnace is programmed via FGH
1000 controller to ramp at 35 ◦C/min to 800 ◦C. The temperature is
isothermal at 800 ◦C during activation. During activation, the Schott
TR 100 is controlled via PC to inject 22.5 ml/min liquid H2O into
the steam generator at 210 ◦C, providing 50 ml/min of steam. The
isothermal period is calculated for each precursor sample, so that
the injected mass of water corresponds to 3/2 of the precursor’s
mass. A schematic of this setup is presented in Fig. 1 [16,17].
2.3.2. Chemical activation with KOH
The precursor material, i.e. fast or flash pyrolysis char (1–2 g)
from hardwood biomass is introduced in an Erlenmeyer flask with
400 g/l KOH solution. The ratio of KOH:sample is 4:1 [18]. The sam
ple is impregnated with KOH by agitating the Erlenmeyer for 24 h in
a water bath at 20 ◦C. After impregnation, the sample is oven dried
at 110 ◦C. The sample is then introduced in a quartz tube and heated
at a rate of 10 ◦C/min to 800 ◦C, followed by 1 h isothermal period
Tubular furnace
Precursor
Steam ge nerator
MKS 247 -N2
Schott TR100
Hea ted environment
PC FGH 100 0
therm
ocouple
outlet
Fig. 1. Physical activation by steam.
at 800 ◦C. The quartz tube is continuously flushed with 50 ml/min
N2 gas, ensuring only negligible remaining oxygen during activa
tion. After activation, the excess KOH is washed from the sample
by consecutive portions of dilute (0.1 M) HCl and distilled water,
until a pH of 6.5 and 7.5 is attained. The sample is then dried and
weighed.
It is believed that activation (KOH starts to react at 700 ◦C [19]),
initially consists of a redox reaction, where carbon is oxidized to
CO or CO2, creating porosity and K2CO3 as a byproduct [20].
2.4. Characterization of activated carbons
2.4.1. Porous texture analysis
The porous texture parameters of the activated carbons (ACs)
are determined from N2 adsorption–desorption isotherms mea
sured at 77 K with a ASAP2020 sorption analyzer (Micromeritics).
Prior to measurements, the sample is outgassed for 12 h at
250 ◦C. The specific surface area (SBET) is calculated using BET
method [21]. The amount of nitrogen adsorbed at a relative pres
sure of p p0−1 = 0.98 is used to evaluate the total pore volume
(VT). The micropore volume (VDR) is determined by applying the
Dubinin–Radushkevich equation up to p p0−1 < 0.15. The Den
sity Functional Theory (DFT) is applied to assess the pore size
distribution. The average micropore size (L0) is calculated using
Dubinin–Stoeckli equation [22].
2.4.2. Methylene blue adsorption isotherm – experimental
Methylene blue (MB) adsorption is representative for the
adsorption of mediumsized molecules and mesopore volume of
the adsorbent. The MB molecules are accessible to pores with a
width larger than 1.5 nm [23]. MB with certificate of analysis is
purchased from Sigma–Aldrich. Blanks containing no MB were ana
lyzed with each series of experiments.
20 mg of AC is introduced in an Erlenmeyer flask with 20 ml MB
solution. At least six different initial MB concentrations are used for
each AC. The samples are agitated for 2 h in a water bath at 20 ◦C. The
experiments are done without pH adjustment, since the pH is only
of minor influence between pH 4 and 9 [24]. The samples are filtered
(Whatmann 40) and eventually diluted to suitable concentration
for UV–vis measurement at 665 nm. The UV–vis spectrometer is
calibrated with 6 methylene blue standards ranging from 0.5 to
5.0 mg/l MB.
2.4.3. Adsorption isotherm – models
The adsorption data are fitted with Langmuir and Freundlich
models. The Langmuir model (1) assumes a “monolayer” adsorp
tion and no change in heat of adsorption with increased surface
coverage, whereas the Freundlich model (2) assumes a “multilayer”
adsorption and a logarithmic decrease of heat of adsorption with
increased surface coverage.
qe =qmKLCe
1 + KLCe(1)
M. Stals et al. / Journal of Analytical and Applied Pyrolysis 101 (2013) 199–208 201
Table 2
Char and AC yields from different biomasses, carbonization methods and activation methods.
Hardwood Pyrolysis Char yield (wt%) Activation AC yield (wt%) Overall AC yield (g per kg hardwood)
S. fragilis Fast 29 Steam 44 128
S. fragilis Fast 29 KOH 42 122
S. fragilis Flash 21 Steam 49 103
S. fragilis Flash 21 KOH 33 69
P. grimminge Fast 31 Steam 47 146
P. grimminge Fast 31 KOH 46 143
P. grimminge Flash 19 Steam 50 95
P. grimminge Flash 19 KOH 41 78
P. nigra Fast 30 Steam 42 126
P. nigra Fast 30 KOH 49 147
P. nigra Flash 20 Steam 44 88
P. nigra Flash 20 KOH 44 88
qe = KF C1/ne (2)
qe is the amount of MB adsorbed per unit of mass of adsorbent
(mg/g) and Ce is the equilibrium concentration of MB (mg/l). KL
and qm are the Langmuir constants and KF and n are the Freund
lich constants. The Langmuir constants qm and KL are related to
the monolayer capacity of the adsorbent (mg/g) and the energy
of adsorption, respectively. Freundlich constants KF and 1/n are
related to the adsorption capacity and intensity, respectively. These
constants can be derived from the intercept and slope of their
respective linear plots. The Langmuir equation. (1) can be linearized
in different ways, for example, according to Stumm and Morgan
[25] (3) and Weber [26] (4), respectively.
1
qe=
1
qm+
1
qmKL
1
Ce(3)
Ce
qe=
1
qmKL+
Ce
qm(4)
Eq. (4) is used to calculate the Langmuir constants, since the
goodness of fit is found to be better than for Eq. (3).
The Freundlich isotherm is linearized as follows:
log qe = log KF +1
nlog Ce (5)
2.4.4. Kinetic study – experimental
Equal amounts of MB solution are added to 7 Erlenmeyer flasks
containing each 20 mg of the same type of AC. The samples are agi
tated in a water bath at 20 ◦C. The adsorption of MB is stopped after
different time intervals, i.e. after 15 min, 30 min, 1 h, 2 h, 4 h, 8 h, and
24 h. Because of the much higher MB uptake for KOH activated car
bons, the KOH AC were tested with a higher initial MB concentration
(500 mg/l) than steam AC (100–200 mg/l), see Table 5.
2.4.5. Kinetic study – model
The kinetics of MB removal is described as follows [27]
k´
S + [MB] [SMB]
k´´
←→
(6)
where S is the number of active sites on the sorbent, MB is
the concentration of methylene blue in the solution and SMB is
the concentration of MB adsorbed by the sorbent. The following
assumptions are made:
• Sorption occurs on localized sites and involves no interaction
between sorbed ions• The energy of adsorption is independent from surface coverage• Monolayer adsorption is assumed• The adsorption mechanism is governed by a secondorder rate
equation
• The desorption rate (k′′) is negligible compared to the adsorption
rate (k′).
Since the process is strongly influenced by the amount of adsor
bate on the surface and the adsorbate adsorbed at equilibrium, the
rate of reaction is proportional to the number of available sites on
the surface of the adsorbent. The pseudosecondorder rate expres
sion can be written as [12,27]
d(S)t
dt= k2[(S)0 − (S)t]
2 (7)
where (S)0 and (S)t are the number of sites available at initial
time t = 0 and t = t, respectively. Eq. (7) can be rewritten in terms
of adsorbed quantity (8) where qe is the equilibrium amount
adsorbed, and qt is the measured quantity adsorbed at time t
[28,29]:
d(q)t
dt= k2(qe − qt)
2 (8)
Integration of Eq. (8) (qt = 0 when t = 0 and qt = qt when t = t)
provides a linear equation:
t
qt=
1
k2q2e
+t
qe(9)
The reaction rate constant (k2) and equilibrium maximum
adsorption qe can be derived from the slope and intercept of a plot
of t/qt vs. t.
The experimental data are also fitted with the Lagergren pseudo
firstorder kinetics equation [30]. However, for several samples, a
lower correlation coefficient (around 0.90) is calculated. Mainly the
data points from short contact time show deviation from linearity.
Therefore, the first order model was not retained.
2.4.6. Iodine adsorption
The iodine number indicates the small pores (< 2 nm) on an AC
and reflects the ability to adsorb small molecules [31]. It is assessed
by the ASTMD4607 method. The iodine number is defined as the
mass (mg) of iodine adsorbed per gram of AC, after 30 s of con
tact time with a 0.1 N iodine solution. At least three adsorption
data points between 0.008 and 0.04 N residual iodine solutions are
recorded and log(C30 s) is plotted vs. log(q30 s). This is similar to Eq.
(5). From the regression slope and intercept, the corresponding q30 s
for a 0.02 N concentration is calculated to obtain the iodine number.
3. Results and discussion
3.1. Material balance
The three hardwoods (S. fragilis, P. grimminge and P. nigra) are
subjected to both fast and flash pyrolyses, resulting in six different
chars. Those chars are each used as precursor material for activated
202 M. Stals et al. / Journal of Analytical and Applied Pyrolysis 101 (2013) 199–208
Table 3
Characteristics of the porous texture of ACs determined by N2 sorption at 77 K.
Precursor Pyrolysis SBET(m2/g) VT(cm3/g) VDR(cm3/g) L0(nm) VDR/VT Vmez(cm3/g)
KOH activation
S. fragilis Fast 2767 1.25 0.99 1.64 0.79 0.26
P. grimminge Fast 2854 1.32 1.05 1.86 0.79 0.27
P. nigra Fast 2588 1.28 0.95 2.10 0.74 0.33
S. fragilis Flash 2227 1.03 0.85 1.68 0.82 0.18
P. grimminge Flash 2841 1.41 1.04 1.94 0.74 0.37
P. nigra Flash 2223 1.06 0.85 1.41 0.80 0.21
Steam activation
S. fragilis Fast 794 0.46 0.31 0.89 0.67 0.15
P. grimminge Fast 682 0.34 0.27 0.84 0.79 0.07
P. nigra Fast 810 0.38 0.29 0.87 0.76 0.09
S. fragilis Flash 623 0.35 0.22 1.07 0.63 0.13
P. grimminge Flash 648 0.31 0.24 1.01 0.77 0.07
P. nigra Flash 697 0.32 0.25 0.99 0.78 0.07
Reference AC 1137 0.61 0.43 1.27 0.70 0.18
carbon production by physical activation (steam) and chemical
activation (KOH), resulting in 12 different ACs. The pyrolysis char
yields and AC yields are presented in Table 2, the data are sorted
by descending AC yield per kg of biomass processed. Fast pyrolysis
yields are reproducible within 1% rsd; flash pyrolysis experiments
are reproducible within 3–5% rsd and activation (physical or chem
ical) yield within approximately 5% rsd.
From Table 2, it can be concluded that the highest AC yields are
obtained from fast pyrolysis. This is because fast pyrolysis is favor
able over flash pyrolysis, when striving for char production: a lower
heating rate results in a higher amount of char produced [32,33].
In the case of fast pyrolysis, the AC yields from steam activation
do not differ strongly from KOH activation yields. In the case of
flash pyrolysis, steam activation produces higher yields than KOH
activation, except for P. nigra.
3.2. Porous texture of ACs
The nitrogen adsorption–desorption isotherms at 77 K for the
ACs prepared by steam activation and KOH activation are shown in
Fig. 2.
Both series of ACs behave like a type I adsorbent according to
the BDDT classification, reflecting their microporous nature [21].
The nitrogen adsorption capacity of KOH ACs is significantly higher
than that of steam ACs. The profile of isotherms clearly proves that
different microporosity in terms of pore width is developed during
KOH and steam activation. A much wider knee of the isotherm of
KOH ACs compared to steam ACs proves that the latter ACs are char
acterized by narrower micropores. For a given activation method,
the nitrogen adsorption capacity of AC produced from fast pyrolysis
char is higher than that of flash pyrolysis char. The porous texture
parameters of the ACs calculated from their corresponding adsorp
tion isotherms are given in Table 3. As expected, KOH activation was
found to be a more efficient method for developing porous texture
than the activation with steam. The BET surface area is between
2223 and 2854 m2/g for KOH ACs whereas the steam ACs show the
SBET in the range of 623–810 m2/g only. The same trend is observed
for the micropore volume VDR. Considering the effect of the char
on the porous texture of resultant AC, it can be observed that fast
pyrolysis char shows a higher propensity for developing porosity
than flash pyrolysis char for both modes of activation. The con
tribution of mesopores to the total pore volume of ACs is in the
range of 0.18–0.37. Taking into account the lower total pore vol
ume of steam ACs (Table 3), it is clear that KOH ACs show 3–4
times higher mesopore volume, which can be crucial regarding
the adsorption of large molecules. A large difference in the average
micropore width between both series of ACs is observed. KOH acti
vation yields ACs with significantly wider micropores compared to
steam ACs (1.41–2.10 vs. 0.84–1.07 nm). The relatively severe con
ditions of KOH activation used in this study, i.e. the KOH/precursor
ratio of 4:1 and the temperature of 800 ◦C, promote the formation
of wider micropores [34].
Fig. 3a and b shows the pore size distribution determined from
the N2 adsorption data using the DFT method for steam and KOH
Table 4
MB adsorption isotherm models: model parameters and correlation coefficients.
Hardwood Pyrolysis Freundlich Langmuir (Stumm and Morgan) Langmuir (Weber)
Kf (mg/g (l/mg)1/n) n R2 qm (mg/g) KL (l/mg) R2 qm (mg/g) KL (l/mg) R2
Chemical activation
S. fragilis Fast 506 12.65 0.9888 637 9.41 0.7785 727 0.94 0.9996
S. fragilis Flash 563 13.12 0.9339 698 4.32 0.9234 703 4.20 0.9997
P. grimminge Fast 418 9.53 0.9440 567 5.11 0.8991 614 2.23 0.9996
P. grimminge Flash 479 9.86 0.7163 612 7.91 0.9282 616 9.04 0.9999
P. nigra Fast 478 19.10 0.9682 582 3.17 0.8184 647 0.54 0.9988
P. nigra Flash 470 11.10 0.9330 666 2.58 0.8517 710 0.69 0.9993
Physical activation
S. fragilis Fast 87 10.51 0.9672 120 1.29 0.9030 123 1.20 0.9980
S. fragilis Flash 38 4.99 0.9989 62 1.56 0.9283 67 0.88 0.9966
P. grimminge Fast 75 6.03 0.9134 158 0.22 0.7338 164 0.16 0.9965
P. grimminge Flash 170 16.24 0.9219 191 51.24 0.8850 207 3.84 0.9980
P. nigra Fast 218 21.72 0.9780 257 16.34 0.8749 266 1.94 0.9991
P. nigra Flash 156 10.43 0.8509 227 0.7 0.8814 232 0.82 0.9993
Reference AC 247 29.58 0.9725 269 3.70 0.8931 307 0.99 0.9994
M. Stals et al. / Journal of Analytical and Applied Pyrolysis 101 (2013) 199–208 203
0
200
400
600
800
1000
0 0,2 0,4 0,6 0,8 1
VS
TP, c
m3/g
p/po
0
200
400
600
800
1000
0 0,2 0,4 0,6 0,8 1
VS
TP, c
m3/g
p/po
0
200
400
600
800
1000
0 0,2 0,4 0,6 0,8 1
VS
TP, c
m3/g
p/po
0
200
400
600
800
1000
0 0,2 0,4 0,6 0,8 1
VS
TP, c
m3/g
p/po
0
200
400
600
800
1000
0 0,2 0,4 0,6 0,8 1
VS
TP, c
m3/g
p/po
a. S. f ragilis fas t pyrolysisc. P. Grimminge fast py rolysis
e. P. nigra fa st pyrolysis
b. S. f ragilis flash pyrolysis
d. P. Gri mmin ge flas h py rol ysis
f. P. nig ra flash py rolysis
0
200
400
600
800
1000
0 0,2 0,4 0,6 0,8 1
VS
TP, c
m3/g
p/po
(a) (b)
(c) (d)
(e) (f)
Fig. 2. N2 adsorption isotherms at 77 K for the ACs produced by KOH activation (N) and steam activation (d).
ACs, respectively. The histograms of KOH ACs present a similar
relatively broad pore size distribution, except the ACs derived
from flash pyrolysis char of P. grimminge and P. nigra (Fig. 3a–d
and a–f, respectively). The aforementioned samples have a pro
nounced maximum pore volume at a pore width of 1.3–1.5 nm
and 1.1–1.3 nm, respectively. The steam ACs compared to KOH ACs
show a different pore size distribution, with the predominance of
narrow micropores. It is interesting to note that the kind of biomass
and the mode of pyrolysis have a very small impact on the pore size
distribution of AC produced by activation with steam (Fig. 3b). The
maximum volume is observed at a pore width of 0.7–0.9 nm for all
samples. The experimental work also includes a comparative study
of commercial AC. As can be seen in Table 3, the values of porous
texture parameters are in the middle between steam and KOH ACs.
The N2 sorption isotherm at 77 K (a) and the pore size distribution
(b) of the reference AC are shown in Fig. 4.
The shape of isotherm clearly proves the microporous nature of
reference AC (Fig. 4a). However, in contrast to the produced ACs,
the reference AC is characterized by bimodal micropore size distri
bution (Fig. 4b). There are two maxima in the histogram, the first
maximum at a width of 0.7–0.9 nm in the range of narrow micro
pores and the second one at a pore width of 1.1–1.5 nm for wider
micropores.
The two series of ACs produced by KOH and steam activation of
flash and fast pyrolysis chars and the reference AC are character
ized by adsorption of MB and I2, two molecules of different sizes
to confirm the feasibility of these porous carbons in liquidphase
adsorption processes.
204 M. Stals et al. / Journal of Analytical and Applied Pyrolysis 101 (2013) 199–208
0
0,05
0,1
0,15
0,2
0,25
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,05
0,1
0,15
0,2
0,25
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,05
0,1
0,15
0,2
0,25
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,05
0,1
0,15
0,2
0,25
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,05
0,1
0,15
0,2
0,25
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,05
0,1
0,15
0,2
0,25
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,02
0,04
0,06
0,08
0,1
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,02
0,04
0,06
0,08
0,1
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,02
0,04
0,06
0,08
0,1
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,02
0,04
0,06
0,08
0,1
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Po
re v
olu
me
(cm
3/g
)P
ore
vo
lum
e (
cm
3/g
)
Pore diameter (nm)
0
0,02
0,04
0,06
0,08
0,1
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
0
0,02
0,04
0,06
0,08
0,1
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Pore diameter (nm)
a. S. fragilis fast pyrolysisc. P. Grimminge fast pyrolysise. P. nigra fast pyrolysis
b. S. fragilis flash pyrolysisd. P. Grimminge flash pyrolysisf. P. nigra flash pyrolysis
a. S. fragilis fast pyrolysisc. P. Grimminge fast pyrolysise. P. nigra fast pyrolysis
b. S. fragilis flash pyrolysisd. P. Grimminge flash pyrolysisf. P. nigra flash pyrolysis
a
(a) (b)b
c d
e f
a b
c d
e f
Po
re v
olu
me
(cm
3 /g
)
Po
re v
olu
me
(cm
3/g
)
Po
re v
olu
me
(cm
3/g
)
Po
re v
olu
me
(cm
3/g
)
Po
re v
olu
me
(cm
3 /g
)
Po
re v
olu
me
(cm
3/g
)
Po
re v
olu
me
(cm
3/g
)
Po
re v
olu
me
(cm
3 /g
)
Po
re v
olu
me
(cm
3/g
)
Po
re v
olu
me
(cm
3/g
)
Fig. 3. a. DFT pore size distribution of KOH ACs; 3b. DFT pore size distribution of steam ACs.
3.3. Methylene blue adsorption study
The adsorption of MB is commonly studied for the determi
nation of adsorptive properties of porous carbons [35–39]. MB
sorption experiments with pyrolysis char show that the MB adsorp
tion capacity is around 40–50 mg/g and the iodine sorption is
around 100 mg/g. These figures indicate that it is desirable to
enhance the porous texture from the char.
The MB adsorption isotherms are shown in Fig. 5: ACs from
fast pyrolysis chars left and ACs from flash pyrolysis chars right.
Each graph contains two isotherms, (i.e. for KOH and steam ACs)
together with their Langmuir approximation. All KOH ACs have
superior MB uptake compared to steam activated samples. In
case of S. fragilis and P. grimminge biomass, the carbonization
mode (fast or flash pyrolysis) is of little importance regarding
the performance of KOH ACs. For P. nigra, flash pyrolysis char
produces a slightly higher performing AC than the fast pyrolysis
char.
In the case of steam activation of P. nigra and S. fragilis, fast pyrol
ysis char is a better precursor than flash pyrolysis char. In contrast, a
(a) (b)
0
200
400
600
800
0 0,2 0,4 0,6 0,8 1
VS
TP, cm
3/g
p/po
0
0,02
0,04
0,06
0,08
0,1
0,5 0,7 1,1 1,5 1,9 2,3 2,7 3,1 3,5 3,9 4,3
Po
re v
olu
me
(cm
3 g
- 1)
Pore diameter (nm)
Fig. 4. Characteristics of porous texture of the reference AC. a, N2 adsorption isotherm at 77 K; b, pore size distribution determined by DFT method.
M. Stals et al. / Journal of Analytical and Applied Pyrolysis 101 (2013) 199–208 205
Fig. 5. MB adsorption experimental data and Langmuir model for ACs.
higher MB uptake is observed when P. grimminge is flash pyrolysed
in the carbonization step.
In Table 4, the Freundlich and Langmuir constants are pre
sented together with the model correlation coefficients. In all cases,
Weber’s approximation [26] of the Langmuir adsorption isotherm
provides the best fit. The linearization according to Stumm and
Morgan [25] produces lower correlation coefficients, therefore, the
results obtained using Weber’s linearization will be further dis
cussed. It appears that the Langmuir model describes better the
MB adsorption process than the Freundlich model for both the KOH
and steam ACs. The correlation coefficients are very high (0.999).
Since the Langmuir model best fits all the isotherm data, we can
assume that the adsorption can be described in a way similar to
a monolayer “chemisorption” process. Compared with reported
data in the literature [40,41], the ACs studied in this work have
a very large adsorption capacity. As seen from Table 4, the high
est Langmuir monolayer adsorption capacity qm is observed for
KOH AC from S. fragilis, fast pyrolysis (qm = 727 mg/g; reference
qm = 307 mg/g) and P. nigra, flash pyrolysis (qm = 710 mg/g). The low
value of qm found for S. fragilis flash pyrolysis steam AC may be
explained by insufficient and difficult removal of sand from the
char.
No obvious relation between SBET and MB adsorption was estab
lished with the obtained AC.
3.4. Methylene blue kinetic study
The sorption kinetics of ACs derived from fast pyrolysis char
are studied. Fast pyrolysis produces more char than flash pyrol
ysis, and flash pyrolysis char requires the removal of sand which
is a difficult task to perform very correctly for each sample.
Therefore, only AC from fast pyrolysis char is chosen for kinetic
studies. The first order model did not supply a satisfactory fit,
therefore the previously described pseudosecondorder model is
applied. The pseudosecond order model (7) shows a good corre
lation with the experimental data. The model data qe (mg/g), k2
(mg/mg h) and correlation coefficients are given in Table 5, together
with the experimental (measured) qe,exp. In all cases R2 is higher
than 0.999 and the model qe and experimental qe,exp are in good
agreement. The experimental data and model plots are given in
Fig. 6. The results confirm the applicability of the pseudosecond
order kinetic model on the adsorption process of MB on the studied
ACs. This is also demonstrated by other reports on the removal of
MB by biomassbased adsorbents [38–42].
A prime finding is that the adsorption rate is very high at the
beginning of the process, as is illustrated by the initial sorption
rate h which is approximately 3.5–6.0 mg/mg h for KOH ACs and
0.5–2.0 mg/mg h for steam ACs. The experiments with KOH ACs
start with a higher initial MB concentration than steam ACs. If two
equal samples are tested with a different initial MB concentration,
a higher h is expected for the sample with the highest initial MB
concentration. Thus a direct comparison between the two series of
ACs solely yields an absolute performance comparison. A relative
performance comparison can not be made from h when different
initial concentrations are applied.
A relative performance comparison can be done by comparing
the MB removal percentage after a certain time. All KOH ACs adsorb
MB in greater extent than 70% of qe within the first 15 min. The same
goes for the steam ACs samples, except for P. grimminge. All sam
ples (chemical and physical AC) reach over 80% of their qe within
206 M. Stals et al. / Journal of Analytical and Applied Pyrolysis 101 (2013) 199–208
Table 5
MB adsorption kinetic study of various AC derived from fast pyrolysis char.
Hardwood Activation C0 (mg/l) qe,exp (mg/g) qe (mg/g) k2 (mg/mg h) h (mg/mg h) R2 % of qe,exp within 15 min
S. fragilis KOH 500 756 758 10.3 5.9 0.9999 82
P. grimminge KOH 500 737 742 6.2 3.4 0.9998 74
P. nigra KOH 500 738 739 8.7 4.7 0.9996 90
S. fragilis Steam 100 168 169 20.4 0.6 0.9995 72
P. grimminge Steam 125 181 183 14.7 0.5 0.9991 57
P. nigra Steam 200 284 286 37.0 2.0 0.9994 82
Reference AC 300 323 330 24.7 2.7 0.9997 85