Journal of Environmental Management 125 (2013) 117e125

Contents lists available at SciVerse ScienceDirect

Journal of Environmental Management

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

Determining optimal conditions to produce activated carbon from barley husks

using single or dual optimization
M. Loredo-Cancino, E. Soto-Regalado*, F.J. Cerino-Córdova, R.B. García-Reyes, A.M. García-León,
M.T. Garza-González
Universidad Autónoma de Nuevo León, UANL, Facultad de Ciencias Químicas, Av. Universidad S/N, Ciudad Universitaria, San Nicolás de los Garza, NL 66451, Mexico

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

Article history: When producing activated carbons from agricultural by-products, certain properties, such as yield and
Received 9 July 2012 specific surface area, are very important for obtaining an economical and promising adsorbent material.
Received in revised form Nevertheless, many researchers have not simultaneously optimized these properties and have obtained
31 January 2013
different optimal conditions for the production of activated carbon that either increases specific surface
Accepted 12 March 2013
Available online 4 May 2013
area but decreases yield or vice versa. In this research, the production of activated carbon from barley
husks (BH) by chemical activation with zinc chloride was optimized by using a 23 factorial design with
replicates at the central point, followed by a central composite design with two responses (the yield and
Keywords:
Barley husks
iodine number) and three factors (the activation temperature, activation time, and impregnation ratio).
Activated carbon Both responses were simultaneously optimized by using the desirability functions approach to deter-
Optimization mine the optimal conditions of this process. The findings reveal that after the simultaneous dual opti-
Central composite design mization, the maximal response values were obtained at an activation temperature of 436
C, an
Iodine number activation time of 20 min, and an impregnation ratio of 1.1 g ZnCl2/g BH, although the results after the
Yield single optimization of each response were quite different. At these conditions, the predicted values for
the iodine number and yield were 829.58  78.30 mg/g and 46.82  2.64%, respectively, whereas
experimental tests produced values of 901.86 mg/g and 48.48%, respectively. Moreover, activated carbons
from BH obtained at the optimal conditions primarily developed a porous structure (mesopores > 71%
and micropores > 28%), achieving a high surface area (811.44 m2/g) that is similar to commercial acti-
vated carbons and lignocellulosic-based activated carbons. These results imply that the pore width and
surface area are large enough to allow the diffusion and adsorption of pollutants inside the adsorbent
particles. In summary, two responses were optimized to determine the optimal conditions for the pro-
duction of activated carbons because it is possible to increase both the specific surface area and yield.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction number of lignocellulosic by-products have been tested as pre-

cursors in the production of activated carbon, including date stones
Due to its high specific surface area and the possibility of (Alhamed, 2009), coffee residue (Khenniche and Benissad-Aissani,
incorporating functional groups on the surface, activated carbon 2010), nut shells (Mohanty et al., 2005, 2006), agave bagasse
could have a strong affinity for binding organic or inorganic pol- (Nieto-Delgado and Rangel-Mendez, 2011), pharmaceutical herb
lutants from aqueous solutions. For these reasons, activated car- residue (Yang and Qiu, 2011), tamarind wood (Sahu et al., 2010),
bons are commonly used in wastewater treatment. However, coir pith (Namasivayam and Sangeetha, 2006), oreganum stalks
despite these advantages, the use of commercial activated carbon (Timur et al., 2006), pistachio nut shells (Lua and Yang, 2005), etc.
as an adsorbent is often limited for economic reasons, and many In this study, barley husks were used as precursor for activated
researchers have paid attention to the use of alternative precursors, carbon production. It was recently reported that barley production
such as agro-waste materials, for activated carbon production. A in México was about 487,448 ton/year (SIAP, 2012) and it is also
reported that around 70% of this crop is used in the beer industry
(COFUPRO, 2002). Prior to the fermentation process, barley husks
(which represent between 10 and 20% of the whole crop) are
* Corresponding author. Tel./fax: þ52 8183294000.
E-mail addresses: eduardo.sotorg@uanl.edu.mx, edsoto1962@yahoo.com.mx removed from the grain and large quantities of waste are generated
(E. Soto-Regalado). (Palmarola-Adrados et al., 2005; Krawczyk et al., 2008).

0301-4797/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jenvman.2013.03.028
118 M. Loredo-Cancino et al. / Journal of Environmental Management 125 (2013) 117e125

The manufacture of activated carbon consists of the pyrolysis of 2.2. Chemicals

the precursor material followed by a controlled oxidation stage (in
cases of physical activation) or the pyrolysis of the precursor ma- All used chemicals were of analytical reagent grade. Zinc chlo-
terial in a single step by activating agents such as KOH, K2CO3, ride, KI, I2 and Na2S2O3$5H2O were obtained from DEQ, HCl was
NaOH, ZnCl2 or H3PO4. The production of activated carbon by obtained from PQM, NaOH and NaCl were obtained from Fisher
physical activation requires high temperatures (800e1000
C), chemical, KIO3 was obtained from Merck, and nitrogen of ultra-
which involves high power consumption and a low yield of carbon high purity was obtained from AOC Mexico. Distilled water was
(Cooney, 1998; Yang, 2003). On the contrary, chemical activation used throughout this work.
requires low temperatures (400e600
C) and, consequently, the
power consumption is significantly reduced and the yield can be 2.3. Activated carbon production
increased. However, there are also some disadvantages of the
chemical activation, such as the generation of acidic wastewater The precursor was impregnated with ZnCl2 at three different
containing the residual activating agent from the washing process weight ratios: 0.5, 1.0, and 1.5 g ZnCl2/g BH. Known amounts of zinc
of the final product. For this issue, some authors have investigated chloride (5, 10 or 15 g) were first dissolved in distilled water, 10 g of
the reuse of the filtrate from the washing step as activating solu- the precursor was then added to ZnCl2 solution, mixed by a mag-
tion, with promising results (Mohanty et al., 2005; Timur et al., netic stirrer (200 min1) and heated at the boiling point for 6 h. The
2006, 2010). Depending on the conditions of the manufacturing wet mixture was then dried at 60
C for 12 h. A mass of 10 g of the
process, the typical surface areas for activated carbon vary from 500 impregnated precursor was heated at a rate of 5
C/min in an
to 1400 m2/g, although values as high as 2500 m2/g have been electrical furnace (length, 30.5 cm; inner diameter, 5 cm) to reach
reported (Efremenko and Sheintuch, 2006). the activation temperature (300, 500, and 700
C) under a nitrogen
To select the optimal conditions (e.g., the temperature, flow of 60 mL/min. Once the activation temperature was reached,
impregnation ratio, and activation time, among other factors) for the samples were kept at constant temperature for different acti-
the production of activated carbon, various researchers have car- vation times (20, 100, and 180 min). The carbonization experi-
ried out the optimization using the one-factor-at-time strategy (Lua ments were carried out in duplicate in order to provide an estimate
and Yang, 2005; Mohanty et al., 2005, 2006; Timur et al., 2006). of the repeatability of each experimental run. The BHAC was then
This single-dimensional search is laborious and time-consuming, washed with a 0.1 N HCl solution to remove the activating agent
especially for a large number of variables. In recent years, some from the porous material. The BHAC was separated from the so-
authors have used experimental designs to optimize and study the lution by filtration and was washed several times with distilled
effects of the process parameters and their possible interactions water until a non-turbid solution was obtained. To verify the
(Cerino-Córdova et al., 2011); in particular, the factors that affect complete removal of the activating agent, 10 mL of 0.1 M NaOH
the activated carbon production from lignocellulosic by-products solution were added to a 10 mL of filtrate sample for each washing
have been studied using Taguchi experimental design for the stage. Then, the BHAC was dried at 60
C for 12 h and stored in a
optimization of one response (Guijarro-Aldaco et al., 2011; desiccator.
Hernández-Montoya et al., 2012) and the response surface meth- The BHAC yield percentage (Y) was calculated by Eq. (1) as
odology when two or more responses were optimized (Sahu et al., follows:
2010; Nieto-Delgado and Rangel-Mendez, 2011; Yang and Qiu,
2011). According to recent literature, there are not reported M
Y ¼  100 (1)
studies related to simultaneous optimization of yield and iodine M0
number for the production of activated carbon from barley husks.
In this research, the production of activated carbon from barley where M (g) is the mass of BHAC and M0 (g) is the mass of BH.
husks (BH) by chemical activation with zinc chloride was simulta-
neously optimized using a factorial design with two responses (the 2.4. Materials characterization
yield and iodine number). The factors included in the experimental
design were the activation temperature, activation time, and In order to apply pyrolysis, the precursor must be properly
impregnation ratio. To establish the optimal conditions for the characterized for elemental and fixed carbon, volatile matter, ash,
production of barley husks activated carbon (BHAC), a 23 factorial and moisture content. The elemental analysis (C, H, N, and S) was
design with replicates at the central point was followed by a central performed using an elemental analyzer (2400 Series II, Perkin
composite design with axial points to build a quadratic model and Elmer). The proximate analysis was carried out according to the
to explore the response surface space. Finally, the BHAC produced method described in ASTM Standard D3172 (2007).
at the optimal conditions was characterized by N2 physisorption, The surface area is a very important property of activated
Fourier Transform Infrared (FTIR) analysis, and scanning electron carbons for adsorption processes, and it can be determined by a
microscopy (SEM). N2-physisorption method or estimated by the iodine number. The
former property is measured as the quantity of iodine adsorbed on
activated carbon (mg/g) in equilibrium with iodine solutions
2. Material and methods (0.02 N I2 e KI) according to the ASTM Standard D4607 (2011)
procedure. Although the iodine method only indicates the sur-
2.1. Precursor face area which exists in pores of widths equal to or greater than
1 nm (Cooney, 1998), the surface area is quickly determined and
Barley husks were used as the activated carbon precursor; this special equipment is not needed as for the N2-physisorption
material, a by-product of the beer industry, was obtained from a method.
local brewery. The precursor was washed five times with distilled The pore size distribution is an important parameter that is
water until a colorless solution was obtained. After that, BH was useful in determining whether an adsorbate could diffuse towards
dried at 60
C for 24 h and sieved to obtain 0.6e1 mm particle size the functional groups inside the activated carbon particle. The
(18/30 mesh screen) before zinc chloride impregnation and pore size distribution and specific surface area were obtained from
pyrolysis. the adsorption isotherm of nitrogen at 197.543
C and 32
 M. Loredo-Cancino et al. / Journal of Environmental Management 125 (2013) 117e125 119

different partial pressures (0.01e0.995), using the BET equation. X

l X
l X
l X
l X
l X
l
The adsorption isotherm of nitrogen was determined using an Y ¼ b0 þ bi xi þ < bij xi xj þ < < bijk xi xj xk
ASAP 2020 (Micromeritics) analyzer after out-gassing BHAC at i¼1 i¼1 j¼1 i¼1 j¼1 k¼ 1
500 mm of Hg for 240 min. The surface morphology of BHAC was X
l
studied by SEM (NanoSEM 200, Nova) with a 15 kV accelerating þ bii x2i þ 3 ð3Þ
voltage. i¼1
The identification of the functional groups on activated carbon is
where bii is the coefficient for the quadratic effect and represents
useful for understanding the interactions between the adsorbate
the quadratic independent variables.
and adsorbent. The surface functional groups were qualitatively
An analysis of variance was used to measure the magnitude of
identified using an FTIR spectrometer (Spectrum One, Perkin
the effects of the factors for the two studied response variables: the
Elmer) and a scanning was recorded in the range of 4000e
yield and iodine number. The adjusted determination coefficient
650 cm1.
(R2Adj) was used as a measurement of the proportion of the total
An adsorbent in an aqueous solution acquires a positive or
observed variability described by the model (Montgomery, 2010).
negative surface charge as a result of the interactions between the
Quadratic models were used to build the surfaces of the response
functional groups in the adsorbent and the Hþ or OH ions present
variables and to find the maximal value of responses by using the
in solution. When the positive and negative charges are equal, the
Design expert software (version 7.0.0, Stat-Ease, Inc., USA).
adsorbent net surface charge is neutral; this pH value is commonly
called the point of zero charge (pHPZC). The pHPZC of BHAC was
determined based on the method reported by Noh and Schwarz 2.6. Optimization method
(1989). Sixteen different volumes of 0.1 M NaOH or HCl were
placed into 25 mL volumetric flasks and filled to the mark with The iodine number and yield were simultaneously optimized by
0.1 M NaCl. Then, 50 mg of BHAC were added to the solutions and using the desirability functions approach included in the Design
stirred continuously for 5 days at room temperature. The final pH expert software. This method first converts each estimated
was measured at the end of the experiment. Blank tests were response yi (the iodine number or yield) into an individual scale-
simultaneously performed without BHAC under the same free desirability function di that varies over the range of 0e1.
conditions. When a response is to be maximized, the partial desirability of yi at
condition x* can be defined as:
2.5. Experimental design 8
> 0 if yi ðxÞ  yi;min
>
>
>
> !r
< yi ðxÞ  yi;min
The optimal conditions in the production of activated carbon
di ðxÞ if yi;min  yi ðxÞ  yi;max (4)
from BH were determined by using a 23 factorial design with >
> y i;max  yi;min
>
>
central points followed by a central composite design. Two re- >
:
sponses (the yield and iodine number) were simultaneously 1 if yi ðxÞ  yi;max
optimized by studying three factors (the activation temperature,
activation time, and impregnation ratio) at different levels as where yi,min and yi,max are the lower and upper acceptability bounds
shown in Table 1. for response i, respectively, and r is the weight. The lower and upper
A linear model (Eq. (2)) was initially used to represent the acceptability bounds were set at the extreme values of the indi-
experimental data obtained from the full 23 factorial design: vidual estimated response, since the response surface had been
reasonably explored in the range of studied variables. The weights
X
l X
l X
l X
l X
l
for the response variables were set to one, because both variables
Y ¼ b0 þ bi xi þ < bij xi xj þ <
were considered equally important.
i¼1 i¼1 j¼1 i¼1 j¼1
The design variables are then chosen to maximize the overall
X
l
desirability D, which is another value between 0 and 1 and is
< bijk xi xj xk þ 3 (2)
defined by combining di using a weighted geometric mean as
k¼1
shown in Eq. (5).
where Y is the independent variable or response, b0 is a constant  1
coefficient, bi, bij, bijk are the coefficients for the linear, double, and D ¼ dr11  dr22 .  drnn Sri (5)
triple interaction effects, respectively, xi, xj, xk are the independent
variables or factors, and 3 is the random error. where the ri are the relative importance among the n variables and
After the linear model was applied, replicates of the central responses with target i ¼ 1, 2, ., n. The objective is then to find the
point were added to full 23 factorial design to test curvature on input variable setting x* which maximizes the value of D (Kim and
response surface. Subsequently, axial points were added to the Lin, 2000;Montgomery, 2010). The overall desirability function was
existing data to build the quadratic model (Eq. (3)) shown below:

Table 2
Table 1 Proximate and elemental analyses of barley husk.
The factors and levels used in the experimental design.
Analysis Content Weight percentage (%)
Independent variable Coded factor Coded levels
Proximate Moisture 8.86
1 0 1 Ash 6.78
Volatile matter 75.02
Actual levels Fixed carbon 18.19
Activation time (min) A 20 100 180 Elemental C 41.16
Temperature (
C) B 300 500 700 H 6.83
Impregnation ratio C 0.5 1.0 1.5 N 0.045
(g ZnCl2/g BH) S 0.65
120 M. Loredo-Cancino et al. / Journal of Environmental Management 125 (2013) 117e125

Table 3
The design matrix and experimental responses.

Run Coded values Iodine number (mg/g) Yield (%)

A B C Replicate Average Predicted Replicate Average Predicted

1 2 1 2
1 1 1 1 221.70 374.63 298.16 293.83 65.00 59.79 62.40 60.11
2 þ1 1 1 336.48 313.48 324.98 293.83 54.19 62.62 58.40 60.11
3 1 þ1 1 270.25 309.83 290.04 247.56 30.79 31.52 31.16 29.25
4 þ1 þ1 1 288.09 110.31 199.20 247.56 29.91 27.11 28.51 29.25
5 1 1 þ1 452.54 452.54 452.54 513.23 51.51 48.53 50.02 49.25
6 þ1 1 þ1 537.66 541.97 539.82 513.23 46.77 48.57 47.67 49.25
7 1 þ1 þ1 736.10 670.72 703.41 757.23 11.89 11.77 11.83 10.33
8 þ1 þ1 þ1 788.38 682.81 735.60 757.23 10.28 6.92 8.6 10.33
9 1 0 0 1011.06 e 1011.06 854.74 49.02 e 49.02 42.82
10 þ1 0 0 885.37 e 885.37 854.74 38.65 e 38.65 42.82
11 0 1 0 400.74 e 400.74 403.53 56.66 e 56.66 60.26
12 0 þ1 0 665.06 e 665.06 502.40 19.48 e 19.48 25.37
13 0 0 1 448.66 e 448.66 672.47 41.19 e 41.19 44.68
14 0 0 þ1 1091.54 e 1091.54 1037 31.88 e 31.88 29.79
15 0 0 0 712.74 813.86 844.65 854.74 42.80 41.54 43.75 42.82
870.83 730.10 42.90 45.25
1009.96 871.21 47.29 41.65
1019.30 792.19 45.62 42.95

maximized using the Design expert software by a penalty function yield are reported in Tables 6 and 7, respectively. According to the
approach in a downhill simplex search (StatEase, 2005). ANOVA data for the iodine number, the most significant individual
factors were the temperature and impregnation ratio, the most
significant double interaction was between the temperature and the
3. Results and discussion
impregnation ratio, and the most significant quadratic term was for
temperature (Eq. (6)). Similar results were reported by Nieto-
3.1. Precursor properties
Delgado and Rangel-Mendez (2011). Because the activation time
was not a significant factor for the iodine number, it was not included
Several factors could be taken into account in the selection of
in final coded model (Eq. (6)). The individual factors and the double
raw material for the manufacturing of activated carbon, for instance
interaction showed a positive effect on the response of the iodine
high carbon content, low inorganic content (ash), sufficient volatile
number in contrast with the quadratic term for the temperature.
matter content, high local availability, low cost, and large quanti-
Based on the ANOVA data for the yield response (Table 7), the
ties, among others. Therefore, barley husks can be considered an
most significant individual factors were the temperature and
interesting alternative for the production of activated carbon
impregnation ratio, the most significant double interaction was
because of the aforementioned properties (Table 2).
between the temperature and the impregnation ratio, and the most
significant quadratic term was for the impregnation ratio (Eq. (7)).
3.2. Statistical analysis The above-mentioned factors showed a negative effect on the yield
response. Additionally, the activation time did not have a significant
Table 3 shows the conditions and the experimental results for effect on the yield response.
the production of BHAC. The analysis of variance (ANOVA) used to The adjusted determination coefficient (R2adj) for the iodine
determine the significance of the curvature in the iodine number number and the yield was 0.8597 and 0.9566, respectively. This
and yield responses at a confidence level of 99% are shown in result indicates that the models could represent approximately 86%
Table 4 and Table 5, respectively. In both cases, the curvature is and 96% of the observed variability for the iodine number and yield
significant, which implies that there is an inflection point on the responses, respectively.
studied range of variables. Therefore, the linear model is not
appropriate to represent the design space and, for this reason, a Iodine number ¼ 854:74 þ 49:43B þ 182:27C þ 72:57BC
central composite design was used to fit the quadratic model
 401:77B2
considering the 18 experimental runs (Table 3).
After discarding the insignificant terms, the ANOVA data for the (6)
coded and hierarchical quadratic models of the iodine number and 2
Yield ¼ 42:82  17:44B  7:45C  2:02BC  5:58C (7)

Table 4 Table 5
The analysis of variance data for iodine number determined from a factorial design The analysis of variance data for yield determined from a factorial design with
with central points. central points.

Variation Sum of Degree of Mean F-value p-value Variation Sum of Degree of Mean F-value p-value
source square freedom square source square freedom square
Model 4.349  105 1 4.349  105 35.74 <0.0001 Model 58.25 3 19.42 276.25 <0.0001
Curvature 8.946  105 1 8.946  105 73.51 <0.0001 Curvature 2.2 1 2.2 31.34 <0.0001
Residual 2.555  105 21 0.122  105 Residual 1.34 19 0.070
Lack of fit 1.263  105 6 0.211  105 2.44 0.0751 Lack of fit 0.39 4 0.097 1.54 0.2416
Pure error 1.292  105 15 0.086  105 Pure error 0.95 15 0.063
Total 1.585  106 23 Total 61.79 23
 M. Loredo-Cancino et al. / Journal of Environmental Management 125 (2013) 117e125 121

Table 6 The mathematical models for the iodine number (Eq. (6)) and
The analysis of variance data for iodine number determined from a central com- the yield (Eq. (7)), were used to build response surfaces and contour
posite design.
plots, as well as to determine the optimal conditions of the process.
Variation source Sum of Degree of Mean F-value p-value For all of the above-mentioned purposes, an activation time of
square freedom square 20 min was selected because this factor was not significant in either
Model 1.89  106 4 4.721  105 45.41 <0.0001 response (Fig. 1). Moreover, for practical and economic reasons, it is
B 4.399  104 1 4.399  104 4.23 0.0503
desirable to choose the lowest activation time in the studied range
C 5.980  105 1 5.980  105 57.51 <0.0001
BC 8.426  104 1 8.426  104 8.10 0.0087 to reduce the energy consumption in the pyrolysis process. Fig. 1 a
B2 1.162  106 1 1.162  106 111.78 <0.0001 and b shows that the iodine number increased with increasing
Residual 2.599  105 25 1.040  104 impregnation ratio when temperature is fixed, and a maximal
Lack of fit 1.307  105 10 1.307  104 1.52 0.2254 iodine number is observed in the range of 500e600
C regardless of
Pure error 1.292  105 15 8.614  103
Total 2.148  106 29
the impregnation ratio. Fig. 1 c and d shows that when increasing
both the temperature and impregnation ratio, the yield response
decreases.
To determine the optimal conditions, Eqs. (6) and (7) were used
Table 7 to maximize the iodine number and yield, respectively. The
The analysis of variance data for yield determined from a central composite design.
maximum iodine number (1046.26  105.52 mg/g) was predicted
Variation source Sum of Degree of Mean F-value p-value to occur at 530
C and an impregnation ratio of 1.5 g ZnCl2/g BH. The
square freedom square maximal yield (61.58  3.32%) was predicted to occur at 300
C and
Model 6763.75 4 1690.94 160.72 < 0.0001 an impregnation ratio of 0.76 g ZnCl2/g BH.
B 5476.72 1 5476.72 520.56 < 0.0001 The effect of temperature on the iodine number and yield could
C 997.73 1 997.73 94.83 <0.0001
BC 64.99 1 64.99 6.18 0.0200
be explained because an increase in the carbonization temperature
C2 224.31 1 224.31 21.32 0.0001 increases the release of volatile matter from the precursor, causing
Residual 263.02 25 10.52 an increase in the dimensions of the pores and, therefore causing an
Lack of fit 168.32 10 16.83 2.67 0.0422 increase in the iodine number up to the optimal conditions but a
Pure error 94.70 15 6.31
decrease in the yield. When the temperature is greater than 530
C,
Total 7026.77 29
a decrease in the iodine number could be associated with the sin-
tering effect of the volatiles and the shrinkage of the carbon

Fig. 1. The surface response and corresponding contour plot for the iodine number (a and b) and the yield (c and d).
122 M. Loredo-Cancino et al. / Journal of Environmental Management 125 (2013) 117e125

Fig. 2. The surface response and contour plot for the desirability function for the simultaneous optimization of the iodine number and yield.

structure, resulting in the narrowing and closing-up of the pores considered equally important, and thus the weights for the
(Lua and Yang, 2005; Yang and Lua, 2006), as well as the widening response variables were set to one.
of the pores, which means that many micropores were enlarged to The surface response and the corresponding contour plot for
mesopores (Demiral and Gündüzog lu, 2010). At temperatures desirability can be observed in Fig. 2 a and b. The maximum value of
greater than 600
C, a decrease in the iodine number is caused by the desirability function (0.779) was obtained at a temperature of
both the temperature effect (as described above) and the progres- 436
C, an activation time of 20 min, and an impregnation ratio of
sive evaporation of ZnCl2, which leads to a reduction in the 1.1 g ZnCl2/g BH. At these conditions, the predicted responses for
degradation of the tars that are responsible for the shrinkage of the the iodine number and yield were 829.58  78.30 mg/g and
pores (Temdrara et al., 2008). With respect to the effect of the 46.82  2.64%, respectively. To validate the predicted responses,
impregnation ratio on the iodine number, in the range of studied experiments were conducted at the optimal conditions, obtaining
temperatures, the ZnCl2 assumes a role as a dehydration agent an iodine number of 901.86 mg/g and yield of 48.48%. The iodine
during activation, inhibiting the formation of tars and other liquids number obtained in this research is similar to or greater than those
that could clog up the pores of the carbons. In this way, the reported by other researchers studying lignocellulosic precursors
movement of the volatiles through the pores will be enhanced, (Table 8).
increasing iodine number, but the yield is decreased.
According to the results, the maximum iodine number 3.3. Physical chemical properties
(1046.26  105.52 mg/g) and maximum yield (61.58  3.32%) were
found at different optimal conditions, and both responses depend 3.3.1. Surface area and pore volume
on the temperature and impregnation ratio. For instance, when Fig. 3 shows the isotherms of the adsorption/desorption of ni-
using the optimal conditions for the iodine number (530
C and trogen at 77 K on the BHAC produced at different conditions. These
impregnation ratio of 1.5 g ZnCl2/g BH), a low yield isotherms exhibit typical Type IV adsorption/desorption isotherm
(26.84  3.05%) was predicted by Eq. (7). In contrast, if the optimal shape with hysteresis loop and steep rises at low relative pressure
conditions for the yield (300
C and impregnation ratio of 0.76 g indicating the presence of both mesopore and micropore structures
ZnCl2/g BH) were used, a low iodine number (350.87  105.7 mg/ (Sing et al., 1985), and at the highest impregnation ratio, the hys-
g) was predicted by Eq. (6). Therefore, both responses were opti- teresis effect is more evident, indicating an increase in the per-
mized simultaneously by using the desirability functions approach centage of mesopores. Additionally, an increase in the
included in the Design expert software. All variables were impregnation ratio results in an increase in the adsorbed volume;

Table 8
Comparison of the activated carbon properties and production conditions with literature data.

Precursor Impregnation ratio Temperature Activation time Iodine Surface area Yield Reference
(g ZnCl2/g BH) (
C) (min) number (mg/g) (m2/g) (%)
Dates stones 0.5 700 180 e 951 e Alhamed, 2009
Coffee residue 0.25 110 1440 e 520 e Khenniche and
Benissad-Aissani, 2010
Nuts of Terminalia arjuna 2 500 60 e 1260 e Mohanty et al., 2005
Nutshells of Sterculia alata 1 500 60 1122 712 46 Mohanty et al., 2006
Agave bagassea 1.08 456.1 23.8 e 1593.4 e Nieto-Delgado and
Rangel-Mendez, 2011
Herb residuesa 1.105 467 60 896 1125 34 Yang and Qiu, 2011
Tamarind wooda 2.95 439.76 40.26 e 1322 45.26 Sahu et al., 2010
Coir pith 0.5 700 e 203 910 e Namasivayam and
Sangeetha, 2006
Oreganum stalks 1 600 120 e 944 36 HTimur et al., 2006
Pistachio-nut shell 0.75 400 60 e 1635.37 e Lua and Yang, 2005
This research 1.1 436 20 901.86 811.44 48.48
a
Optimal production conditions determined using the response surface methodology.
 M. Loredo-Cancino et al. / Journal of Environmental Management 125 (2013) 117e125 123

Fig. 3. The adsorption/desorption isotherms of N2 at 77 K on BHAC produced under Fig. 6. FTIR spectra of BH and BHAC produced under optimal conditions (activation
optimal conditions (activation time of 20 min, temperature of 436
C, and impreg- time of 20 min, temperature of 436
C, and impregnation ratio of 1.1 g ZnCl2/g BH).
nation ratio of 1.1 g ZnCl2/g BH) and under different impregnation ratios (with acti-
vation time of 100 min and temperature of 500
C).

for example, at a relative pressure of 1, the adsorbed volume in-

creases almost two-fold when the impregnation ratio shifts from
1.0 to 1.5 g ZnCl2/g BH (Fig. 3).
The specific surface area was calculated using the BET model.
The activated carbon produced under the optimal conditions has a
surface area of 811.44 m2/g, which is similar to the commercial and
lignocellulosic activated carbons shown in Table 8. Additionally, the
surface areas of carbons produced at different impregnation ratios
(0.5, 1.0, and 1.5 g ZnCl2/g BH, at 500
C and 100 min) were
calculated to validate the assumption that the iodine number could
be used as an indicator of the surface area. Fig. 4 shows that surface
area and iodine number follow the same trend, where both increase
as the impregnation ratio increases. Nevertheless, the surface area
of BHAC produced at the optimal conditions is not as high as the
BHAC produced with an impregnation ratio of 1.5 g ZnCl2/g BH
(1445.52 m2/g). However, the last conditions were not selected
because of the low activated carbon yield (31.88%), while optimal
conditions were determined by simultaneously taking into account
the surface area and activated carbon yield.
Fig. 4. The effect of the impregnation ratio on the iodine number (mg/g) and surface Fig. 5 shows the pore size distribution of the BHAC produced
area (m2/g) (with activation time of 100 min and temperature of 500
C).
under different conditions. Activated carbon has a significant
fraction of the volume corresponding to micropores (<2 nm) and

Fig. 5. The pore size distribution (a) and cumulative pore volume (b) for BHAC produced under optimal conditions (activation time of 20 min, temperature of 436
C, and
impregnation ratio of 1.1 g ZnCl2/g BH) and under different impregnation ratios (with activation time of 100 min and temperature of 500
C).
124 M. Loredo-Cancino et al. / Journal of Environmental Management 125 (2013) 117e125

reactions between the activating agent and precursor or the pre-

viously mentioned hydrolysis products.

3.3.3. FTIR analysis

Infrared analysis was used to qualitatively identify the func-
tional groups on the surface of BH and BHAC. Fig. 6 shows the FTIR
spectra for BH and for BHAC produced under optimal conditions
(436
C, 1.1 g ZnCl2/g BH, and 20 min), and Fig. 7 shows BHAC
produced under different temperatures (300, 500, and 700
C) but
at a fixed impregnation ratio and time (1.0 g ZnCl2/g BH and
100 min). The main differences between the BH and BHAC spectra
are the decrease in the wide absorption band from 3000 to
3500 cm1, corresponding to the O e H stretching mode of the
hydroxyl group, and the decrease in the bands with maxim at
2935 cm1 and 1422 cm1, corresponding to the C e H stretching
mode of the aliphatic groups. The decrease in the band from 3000
to 3500 cm1 suggests the elimination of the OH groups from the
Fig. 7. FTIR spectra for BHAC produced at different temperatures (with activation time
of 100 min and impregnation ratio 1.0 g ZnCl2/g BH).
biopolymers that compose BH during the activation process.
Regarding the temperature effect on the BHAC surface chemistry, it
can be observed that an increase in the temperature leads to an
enhancement in the elimination of hydrogen, which is reflected in
the elimination of the peak at approximately 1422 cm1.
mesopores (2e50 nm). Fig. 5 shows the pore size distribution of
BHAC produced at different impregnation ratio. The average pore 3.3.4. Morphology of BHAC
increased as the impregnation ratio increased, suggesting that Scanning electron microscopy (SEM) was employed to observe
there was a widening of existing pores into larger pores through the the surface morphology of BHAC. Fig. 8 shows micrographs of
gasification of carbon in the pore walls. BHAC produced under different conditions. As observed in these
micrographs, chemical activation followed by pyrolysis was useful
3.3.2. Surface charge distribution in developing pore structure within the carbon. At the lowest
A solid adsorbent in aqueous solution acquires a positive or impregnation ratio (0.5 g ZnCl2/g BH), the activated carbon shows
negative charge balance as a result of the interactions between the a smooth surface (Fig. 8a), whereas at the highest impregnation
ions in the solution and the functional groups on the surface of the ratio (1.5 g ZnCl2/g BH), a highly porous surface was developed
adsorbent, such as amines, pyrones, ethers, carbonyls, carboxyl (Fig. 8b). These results are related to those reported in Fig. 5, where
acids, phenols and lactones (Snoeyink and Weber, 1967; Cooney, when the impregnation ratio increased from 0.5 to 1.5 g ZnCl2/g
1998; Yang, 2003; Leyva Ramos, 2007). The electrostatic charges BH, the pore volume rose from 0.072 to 0.728 cm3/g. In addition,
on the activated carbon surface can be elucidated by determining this increase in the impregnation ratio decreased the quantity of
the pH at which the carbon surface is neutral; this pH value is called micropores by a factor of 2.6, whereas the quantity of mesopores
point of zero charge (pHPZC). The pHPZC values for raw BH and BHAC increased by a factor of up to 1.4 (Fig. 5). Similarly, the surface area
(produced under optimal conditions) were 4.25 and 1.9, respec- increased by nearly a factor of 2 when the impregnation ratio
tively. According to these values, raw BH and BHAC can be increased by a factor of 3 (Fig. 5). The BHAC obtained under the
considered acidic. Some authors (Boudrahem et al., 2011) proposed optimal conditions shows a porous surface (Fig. 8b) with a pore
that the formation of acidic groups on activated carbon developed volume of 0.263 cm3/g and principally a mesoporous structure
from lignocellulosic wastes activated with zinc chloride are due to (Fig. 5), resulting in a high surface area (811.44 m2/g). It is
the hydrolysis of the precursor under acidic conditions, which leads important to note that these values fall in the middle of the values
to the formation of carboxylic groups of different strengths and to obtained at the lowest and highest impregnation ratio.

Fig. 8. SEM micrographs of BHAC at (a) a low impregnation ratio (0.5 g ZnCl2/g BH; activation time of 100 min and temperature of 500
C), (b) optimal conditions (activation time of
20 min, temperature of 436
C, and impregnation ratio of 1.1 g ZnCl2/g BH) and (c) a high impregnation ratio (1.5 g ZnCl2/g BH; activation time of 100 min and temperature of
500
C).
 M. Loredo-Cancino et al. / Journal of Environmental Management 125 (2013) 117e125 125

4. Conclusions metals from water using commercial carbons modified with egg shell wastes.
Ind. Eng. Chem. Res. 50 (15), 9354e9362.
Hernández-Montoya, V., Ramírez-Montoya, L.A., Bonilla-Petriciolet, A., Montes-
This study demonstrates the usefulness of a central composite Morán, M.A., 2012. Optimizing the removal of fluoride from water using new
design to determine the optimal conditions to improve both the carbons obtained by modification of nut shell with a calcium solution from egg
surface area and yield in the production of activated carbon from shell. Biochem. Eng. J. 61 (1), 1e7.
Khenniche, L., Benissad-Aissani, F., 2010. Adsorptive removal of phenol by coffee
BH. Dual optimization is better than single optimization because residue activated carbon and commercial activated carbon: equilibrium, ki-
both the yield and surface area are very important parameters in netics, and thermodynamics. J. Chem. Eng. Data 55 (11), 4677e4686.
manufacturing activated carbon from lignocellulosic precursors. Kim, K.J., Lin, D.K.J., 2000. Simultaneous optimization of mechanical properties of
steel by maximizing exponential desirability functions. Appl. Statist. 49 (3),
After the simultaneous dual optimization by quadratic models, 311e325.
the optimal conditions were obtained at 436
C, an activation time Krawczyk, H., Persson, T., Andersson, A., Jonsson, A.-S., 2008. Isolation of hemi-
of 20 min, and an impregnation ratio of 1.1 g ZnCl2/g BH. At these celluloses from barley husks. Food Bioprod. Process. 86 (1), 31e36.
Leyva Ramos, R., 2007. Importancia y Aplicaciones de la Adsorción en Fase Líquida.
conditions, the predicted responses for the iodine number and In: Piraján, J.C.M. (Ed.), Sólidos porosos. Preparación, caracterización y aplica-
yield were 829.58  78.30 mg/g and 46.82  2.64%, respectively, ciones, Ediciones Uniandes. Bogotá, Colombia, pp. 155e211.
that are similar to the experimental results (901.86 mg/g and Lua, A.C., Yang, T., 2005. Characteristics of activated carbon prepared from
pistachio-nut shell by zinc chloride activation under nitrogen and vacuum
48.48% for the iodine number and yield, respectively). The BHAC conditions. J. Colloid Interface Sci. 290 (2), 505e513.
produced under the optimal conditions determined by the simul- Mohanty, K., Das, D., Biswas, M., 2006. Preparation and characterization of activated
taneous dual optimization showed a porous surface and had a carbons from Sterculia alata nutshell by chemical activation with zinc chloride
to remove phenol from wastewater. Adsorption 12 (2), 119e132.
surface area of 811.44 m2/g, which is similar to commercial acti-
Mohanty, K., Jha, M., Meikap, B.C., Biswas, M.N., 2005. Preparation and character-
vated carbons and lignocellulosic-based activated carbons. Lastly, ization of activated carbons from terminalia arjuna nut with zinc chloride
the physicochemical characterization and iodine number of BHAC activation for the removal of phenol from wastewater. Ind. Eng. Chem. Res. 44
are indicatives of its potential use as a good adsorbent to remove (11), 4128e4138.
Montgomery, D.C., 2010. Diseño y Análisis de Experimentos, second ed. Limusa
pollutants with similar properties of iodine. Wiley, México.
Namasivayam, C., Sangeetha, D., 2006. recycling of agricultural solid waste, coir
Acknowledgements pith: removal of anions, heavy metals, organics and dyes from water by
adsorption onto ZnCl2 activated coir pith carbon. J. Hazard. Mater. B 135,
449e452.
Loredo-Cancino thanks the Facultad de Ciencias Químicas de la Nieto-Delgado, C., Rangel-Mendez, J.R., 2011. Production of activated carbon from
Universidad Autónoma de Nuevo León and CONACYT for funds organic by-products from the alcoholic beverage industry: surface area and
hardness optimization by using the response surface methodology. Ind. Crop
provided through doctorate grant number 60916. The authors Prod. 34 (3), 1528e1537.
thank Dr. Enrique M. López-Cuellar for providing assistance with Noh, J.S., Schwarz, J.A., 1989. Estimation of the point of zero charge of simple oxides
SEM and Gregorio Rosas for providing assistance with elemental by mass titration. J. Colloid Interface Sci. 130 (1), 157e164.
Palmarola-Adrados, B., Galbe, M., Zacchi, G., 2005. Pretreatment of barley husks for
and proximal analysis. bioethanol production. J. Chem. Technol. Biotechnol. 80, 85e91.
Sahu, J.N., Acharya, J., Meikap, B.C., 2010. Optimization of production conditions for
References activated carbons from tamarind wood by zinc chloride using response surface
methodology. Bioresour. Technol. 101 (6), 1974e1982.
Alhamed, Y.A., 2009. Adsorption kinetics and performance of packed bed adsorber SIAP, 2012. Anuario Estadístico de la Producción Agrícola. Servicio de Información
for phenol removal using activated carbon from dates’ stones. J. Hazard. Mater. Agroalimentaria y Pesquera.
170 (2e3), 763e770. Sing, K.S.W., Everett, D.H., Haul, R.A.W., Moscou, L., Pierotti, R.A., Rouquerol, J.,
ASTM Standard D3172, 2007a, 2007. Standard Practice for Proximate Analysis of Siemieniewska, T., 1985. Reporting physisorption data for gas/solid systems
Coal and Coke. ASTM International, West Conshohocken, PA. with special reference to the determination of surface area and porosity. Pure
ASTM Standard D4607, 1994, 2011. Standard Test Method for Determination of Appl. Chem. 57 (4), 603e619.
Iodine Number of Activated Carbon. ASTM International, West Conshohocken, Snoeyink, V.L., Weber, W.J., 1967. The surface chemistry of active carbon: a dis-
PA, 2007. cussion of structure and surface functional groups. Environ. Sci. Technol. 1 (3),
Boudrahem, F., Soualah, A., Aissani-Benissad, F., 2011. Pb(II) and Cd(II) removal from 228e234.
aqueous solutions using activated carbon developed from coffee residue activated StatEase, Inc, 2005. Design Expert 7 User’s Guide, Minneapolis.
with phosphoric acid and zinc chloride. J. Chem. Eng. Data 56 (5), 1946e1955. Temdrara, L., Khelifi, A., Addoun, A., Spahis, N., 2008. Study of the adsorption
Cerino-Córdova, F.J., García-León, A.M., García-Reyes, R.B., Garza-González, M.T., properties of lignocellulosic material activated chemically by gas adsorption
Soto-Regalado, E., Sánchez-González, M.N., Quezada-López, I., 2011. Response and immersion calorimetry. Desalination 223 (1e3), 274e282.
surface methodology for lead biosorption on Aspergillus terreus. Int. J. Environ. Timur, S., Ikizoglu, E., Yanik, J., 2006. Preparation of activated carbons from Ore-
Sci. Technol. 8 (4), 695e704. ganum stalks by chemical activation. Energy Fuels 20 (6), 2636e2641.
COFUPRO, 2002. Identificación de las Demandas Tecnológicas de la Cadena Agro- Timur, S., Kantarli, I.C., Onenc, S., Yanik, J., 2010. Characterization and application of
alimentaria de Cebada. Coordinación Nacional de las Fundaciones Produce, A. C. activated carbon produced from oak cups pulp. J. Anal. Appl. Pyrolysis 89 (1),
Cooney, D.O., 1998. Adsorption Design for Wastewater Treatment. Lewis Publishers, 129e136.
Florida. Yang, J., Qiu, K., 2011. Experimental design to optimize the preparation of activated
Demiral, H., Gündüzog lu, G., 2010. Removal of nitrate from aqueous solutions by carbons from herb residues by vacuum and traditional ZnCl2 chemical activa-
activated carbon prepared from sugar beet bagasse. Bioresour. Technol. 101 (6), tion. Ind. Eng. Chem. Res. 50 (7), 4057e4064.
1675e1680. Yang, R.T., 2003. Adsorbents: Fundamentals and Applications. John Wiley & Sons,
Efremenko, I., Sheintuch, M., 2006. Predicting solute adsorption on activated car- New Jersey.
bon: phenol. Langmuir 22, 3614e3621. Yang, T., Lua, A.C., 2006. Textural and chemical properties of zinc chloride activated
Guijarro-Aldaco, A., Hernández-Montoya, V., Bonilla-Petriciolet, A., Montes- carbons prepared from pistachio-nut shells. Materials Chemistry and Physics
Morán, M.A., Mendoza-Castillo, D.I., 2011. Improving the adsorption of heavy 100 (2e3), 438e444.