Chemical Engineering Journal 165 (2010) 883–890

Contents lists available at ScienceDirect

Chemical Engineering Journal

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

Optimization of preparation conditions for mangosteen peel-based activated

carbons for the removal of Remazol Brilliant Blue R using response
surface methodology
Mohd Azmier Ahmad ∗ , Rasyidah Alrozi
School of Chemical Engineering, Engineering Campus, Universiti Sains Malaysia, 14300 Nibong Tebal, Penang, Malaysia

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

Article history: This study investigates the optimal conditions for preparation of activated carbons from mangosteen peel
Received 2 September 2010 (MP) for removal of Remazol Brilliant Blue R (RBBR) reactive dye from aqueous solution. The MP activated
Received in revised form 18 October 2010 carbon was prepared using physiochemical activation method which consisted of potassium hydroxide
Accepted 18 October 2010
(KOH) treatment and carbon dioxide (CO2 ) gasification. Central composite design (CCD) was used to
determine the effects of the three preparation variables; CO2 activation temperature, CO2 activation
Keywords:
time and KOH impregnation ratio (IR) on RBBR percentage removal and activated carbon yield. Based
Mangosteen peel
on the CCD, a quadratic model and a two-factor interaction (2FI) model were respectively developed for
Activated carbon
Remazol Brilliant Blue R
RBBR percentage removal and carbon yield. The significant factors on each experimental design response
Central composite design were identified from the analysis of variance (ANOVA). The optimum conditions for MP activated carbon
Optimization preparation were obtained by using activation temperature of 828 ◦ C, activation time of 1 h and IR of 3.0,
which resulted in 80.35% of RBBR removal and 20.76% of activated carbon yield.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction to other techniques in terms of its simplicity of design, high effi-

ciency and ease of operation [8]. However, the manufacturing costs
Almost 45% of textile dyes produced worldwide belongs to of commercial activated carbons are in fact rather high. As such,
the reactive class [1]. Reactive dyes are common dyes used for there is a need to produce activated carbon with high adsorption
dyeing cellulosic fibres due to their favorable characteristics of performance from alternative material that is cheaper and read-
bright color, water-fastness, simple application techniques and low ily available. From the literature, many studies have been carried
energy consumption [2]. They are usually characterized by nitrogen out to prepare low cost activated carbons from agricultural wastes
to nitrogen double bonds (N N azo bonds) where the color of azo such as sugar beet pulp [2], bamboo [5], pomegranate peel [8], date
dyes is due to this azo bond and the associated chromophores [3]. stone [9], vetiver roots [10], corn grain [11], cotton stalk fibre [12],
The formation of a covalent bond between the dye molecule and oil palm empty fruit bunch [13], coconut husk [14] and rice straw
the fibre is much stronger than the physio-chemical bond between [15].
other classes of dyes and cellulose [4]. However, reactive dyes pose In this work, an attempt was made in preparing activated carbon
the greatest problem in textiles wastewater since they are not eas- from mangosteen peel (MP) precursor. Mangosteen (Garcinia man-
ily biodegradable, and thus color may still remain in the effluent [5]. gostana L.), a tropical evergreen tree has been planted in Malaysia
In fact, the expanded uses of azo dyes and their reaction products with the total acreage between 7000 and 8000 ha [16]. It is also
such as aromatic amines are highly carcinogenic. The discharge of being planted in other countries such as India, Thailand, Vietnam
these wastewaters into receiving streams causes damage not only and Philippines. This fruit has become an economically impor-
to aquatic life but also to human beings [6]. tant species since the demand in the domestic and export markets
In recent years, many processes have been applied for the are tremendous. However, due to the high consumption of man-
treatment of reactive dyes from wastewater including biological, gosteen’s edible part, massive amounts of the peels are disposed,
physical and chemical process [7]. Among them, adsorption process causing a severe problem in the community as they gradually fer-
with activated carbon has been proved to be superior compared ment and release off odours [17]. Therefore, by utilizing MP into
activated carbon will decrease the cost of wastes disposal and also
converted this waste into value-added product [18].
∗ Corresponding author. Tel.: +60 4 5996459; fax: +60 4 5941013. Currently no study has been done on optimization of the pro-
E-mail addresses: [email protected], [email protected] duction of activated carbon from MP using the response surface
(M.A. Ahmad). methodology (RSM) approach. RSM has been found to be a useful

1385-8947/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.cej.2010.10.049
884 M.A. Ahmad, R. Alrozi / Chemical Engineering Journal 165 (2010) 883–890

tor as in carbonization step. Once the final activation temperature

was reached, the gas flow was switched from nitrogen to CO2 at
flowrate of 150 ml min−1 for different period of time. The activated
product was then cooled to room temperature under nitrogen flow.
Then, the sample was washed with hot deionized water and HCl
(0.1 M) until the pH of the washed solution reached 6.5–7.

2.3. Adsorption studies

Fig. 1. Chemical structure of RBBR dye.
For batch adsorption studies, 300 mg of adsorbent were mixed
tool to study the interactions of two or more factors [19]. A stan- with 200 ml aqueous dye solutions of 100 mg/l initial concen-
dard RSM design called a central composite design (CCD) is suitable tration in 20 sets of 250 ml Erlenmeyer flasks. The mixture was
for fitting a quadratic surface and it helps to optimize the effec- agitated at 120 rpm at 30 ◦ C until equilibrium was reached. The
tive parameters with a minimum number of experiments, as well pH of the solution was natural without any pH adjustment.
as to analyze the interaction between the parameters [20]. RSM The concentration of RBBR dye solution was determined using a
has just recently been used for the optimization of activated car- double UV–vis spectrophotometer (UV-1800 Shimadzu, Japan) at
bon production from rattan sawdust [19], Tamarind wood [21] and maximum wavelength of 590 nm. The percentage removal at equi-
Turkish lignite [22] by chemical activation whereas oil palm empty librium was calculated by the following equation:
fruit bunch [13] and coconut husk [14] by physiochemical activa-
(Co − Ce )
tion method. The goal of this work was to optimize the preparation Removal (%) = × 100 (2)
Co
conditions of activated carbon from MP for the removal of Rema-
zol Brilliant Blue R (RBBR) dye. The effects of preparation variables; where Co and Ce are the liquid-phase dye concentrations at initial
activation temperature, activation time and chemical impregnation state and at equilibrium (mg/l), respectively.
ratio were studied simultaneously to obtain a high activated car-
bon yield and high RBBR percentage removal from aqueous solution
2.4. Activated carbon yield
using the CCD.
The activated carbon yield was calculated based on the following
2. Materials and methods equation:

2.1. Adsorbate wc
Yield (%) = × 100 (3)
wo
Remazol Brilliant Blue R (RBBR) supplied by Sigma–Aldrich (M)
where wc and wo are the dry weight of final activated carbon (g)
Sdn Bhd, Malaysia was used as an adsorbate. Deionized water
and the dry weight of precursor (g), respectively.
was used to prepare all solutions. RBBR has a chemical formula of
C22 H16 N2 Na2 O11 S3 with molecular weight of 626.54 g mol−1 . The
chemical structure of RBBR is shown in Fig. 1. 2.5. Design of experiments

2.2. Preparation of activated carbon RSM is a collection of statistical and mathematical techniques
that uses quantitative data from appropriate experiments to deter-
Mangosteen peel (MP) used for preparation of activated car- mine regression model equations and operating conditions which
bon was obtained from the local market in Nibong Tebal, Penang, are useful for developing, improving and optimizing processes [13].
Malaysia. MP was firstly washed with water and subsequently dried In this work, a standard RSM design, known as central composite
at 105 ◦ C for 24 h to remove the moisture contents. The dried MP design (CCD) was applied to study the variables for preparing the
was ground and sieved to the size of 1–2 mm before loaded in a activated carbons from MP. This method can reduce the number
stainless steel vertical tubular reactor placed in a tube furnace. Car- of experimental trials needed to evaluate multiple parameters and
bonization step was carried out at 700 ◦ C for 2 h under purified their interactions [14]. Generally, the CCD consists of 2n factorial
nitrogen (99.99%) flow at flowrate of 150 ml min−1 . The char pro- runs, 2(n) axial runs and six center runs, where n is the number of
duced was mixed with KOH pellets with different impregnation factors.
ratio (IR), as calculated using following equation: In the present study, the activated carbons were prepared using
physiochemical activation method where the variables studied
wKOH
IR = (1) were CO2 activation temperature (x1 ), CO2 activation time (x2 ) and
wchar
KOH:char IR (x3 ). These three variables together with their respec-
where wKOH is the dry weight of KOH pellets (g) and wchar is the tive ranges were chosen based on the literature and preliminary
dry weight of char (g). Deionized water was then added to dissolve studies as given in Table 1. For each categorical variable, a 23 full
all the KOH pellets. factorial CCD for the three variables, consisting of 8 factorial points,
The mixture was then dehydrated in an oven at 105 ◦ C for 24 h to 6 axial points and 6 replicates at the center points were employed,
remove moisture. The activation step was done using similar reac- indicating that altogether 20 experiments for this procedure for

Table 1
Independent variables and their coded levels for the central composite design.

Variables (factors) Code Units Coded variable levels

−˛ −1 0 +1 +˛
◦
Activation temperature x1 C 648.87 700.00 775.00 850.00 901.13
Activation time x2 h 0.32 1.00 2.00 3.00 3.68
Impregnation ratio (IR) x3 – 0.15 1.00 2.25 3.50 4.35
 M.A. Ahmad, R. Alrozi / Chemical Engineering Journal 165 (2010) 883–890 885

Table 2
Experimental design matrix for preparation of MP activated carbons.

Run Level Activated carbon preparation variable RBBR removal, Y1 (%) Activated carbon
yield, Y2 (%)

Activation temperature, x1 (◦ C) Activation time, x2 (h) IR, x3

1 −1 −1 −1 700.00 1.00 1.00 37.47 25.34

2 +1 −1 −1 850.00 1.00 1.00 39.15 21.05
3 −1 +1 −1 700.00 3.00 1.00 36.65 24.45
4 +1 +1 −1 850.00 3.00 1.00 39.51 12.56
5 −1 −1 +1 700.00 1.00 3.50 60.43 19.57
6 +1 −1 +1 850.00 1.00 3.50 83.56 19.09
7 −1 +1 +1 700.00 3.00 3.50 61.53 19.31
8 +1 +1 +1 850.00 3.00 3.50 82.20 11.38
9 −1.682 0 0 648.87 2.00 2.25 30.54 23.72
10 +1.682 0 0 901.13 2.00 2.25 80.54 17.56
11 0 −1.682 0 775.00 0.32 2.25 64.89 25.51
12 0 +1.682 0 775.00 3.68 2.25 73.49 19.84
13 0 0 −1.682 775.00 2.00 0.15 35.05 26.59
14 0 0 +1.682 775.00 2.00 4.35 79.53 19.56
15 0 0 0 775.00 2.00 2.25 76.45 21.31
16 0 0 0 775.00 2.00 2.25 77.54 20.18
17 0 0 0 775.00 2.00 2.25 77.83 20.75
18 0 0 0 775.00 2.00 2.25 77.55 21.01
19 0 0 0 775.00 2.00 2.25 76.54 20.02
20 0 0 0 775.00 2.00 2.25 77.40 20.11

each precursor, as calculated from Eq. (4) [20]: was determined using Brunauer–Emmett–Teller (BET) equation.
n 3 The total pore volume was estimated to be the liquid volume of
N = 2 + 2n + nc = 2 + 2(3) + 6 = 20 (4)
nitrogen at a relative pressure of 0.98.
where N is the total number of experiments required. The surface morphology of the samples was examined using a
The center points are used to determine the experimental error scanning electron microscope (JEOL, JSM-6460 LV, Japan). Proxi-
and the reproducibility of the data. The axial points are located mate analysis was carried out using thermogravimetric analyser
at (±˛, 0, 0), (0, ±˛, 0) and (0, 0, ±˛) where ˛ is the distance (Perkin Elmer TGA7, USA) and elemental analysis was performed
of the axial point from center and makes the design rotatable. In using Elemental Analyzer (Perkin Elmer Series II 2400, USA).
this study, ˛ value was fixed at 1.682 (rotatable). The experimen-
tal sequence was randomized in order to minimize the effects of 3. Results and discussion
the uncontrolled factor. The two responses were RBBR removal
(Y1 ) and activated carbon yield (Y2 ). Each response was used to 3.1. Development of regression model equation
develop an empirical model which correlated the response to the
three activated carbon preparation variables using a second-degree CCD was used to develop a polynomial regression equation
polynomial equation as given by Eq. (5): in order to analyze the correlation between the activated carbon
 n  preparation variables to the RBBR removal and activated car-

n
  
n−1 n
bon yield. Table 2 shows the complete design matrixes together
Y = bo + bi xi + bii xi + bij xi xj (5)
with both the response values obtained from the experimental
i=1 i=1 i=1 j=i+1 work. Run 15–20 at the center point were conducted to deter-
where Y is the predicted response, bo the constant coefficient, bi the mine the experimental error and the reproducibility of the data.
linear coefficients, bij the interaction coefficients, bii the quadratic RBBR removal and activated carbon yield were found to range from
coefficients and xi , xj are the coded values of the activated carbon 30.54 to 83.56% and 11.38 to 26.59%, respectively. According to the
preparation variables. sequential model sum of squares, the models were selected based
on the highest order polynomials where the additional terms were
2.6. Model fitting and statistical analysis significant and the models were not aliased. For response of RBBR
removal the quadratic model was selected as suggested by the soft-
The experimental data were analyzed using a statistical soft- ware. On the other hand, for activated carbon yield, the two-factor
ware Design Expert software version 6.0.6 (STAT-EASE Inc., interaction (2FI) model was the best model to correlate the data
Minneapolis, USA) for regression analysis to fit the second-degree to the response. The final empirical formula models for the RBBR
polynomial equation and also for the evaluation of the statistical removal (Y1 ) and MP activated carbon yield (Y2 ) in terms of coded
significance of the equations developed. factors after excluding the insignificant terms are represented by
Eq. (6) and (7), respectively.
2.7. Characterization of activated carbon Y1 = 77.37 + 9.70x1 + 1.01x2 + 15.36x3 − 8.64x12 − 3.81x22

The surface area, pore volume and average pore diameter of − 8.02x32 − 0.16x1 x2 + 4.91x1 x3 + 0.025x2 x3 (6)
the MP activated carbon prepared under optimum preparation
conditions were determined by using Micromeritics ASAP 2020 Y2 = 20.45 − 2.56x1 − 1.97x2 − 1.89x3 − 1.88x1 x2 + 0.97x1 x3
volumetric adsorption analyzer. Prior to analysis, the sample was
+ 0.18 x2 x3 (7)
degassed for 2 h under vacuum at 300 ◦ C. After degassing, the sam-
ple was transferred to the analysis system where it was cooled in
liquid nitrogen. A 21-point analysis was carried out at 77 K to obtain The coefficient with one factor represent the effect of the partic-
the nitrogen adsorption isotherm. The surface area of the sample ular factor, while the coefficients with two factors and those with
886 M.A. Ahmad, R. Alrozi / Chemical Engineering Journal 165 (2010) 883–890

Table 3
Analysis of variance (ANOVA) for response surface quadratic model for RBBR removal of MP activated carbon.

Source Sum of squares Degree of freedom (DF) Mean square F-value Prob. > F Comment

Model 6611.48 9 734.61 18.54 <0.0001 Significant

x1 1284.16 1 1284.16 32.41 0.0002
x2 13.83 1 13.83 0.35 0.5678
x3 3221.35 1 3221.35 81.29 <0.0001
x12 1074.97 1 1074.97 27.13 0.0004
x22 209.27 1 209.27 5.28 0.0444
x32 926.47 1 926.47 23.38 0.0007
x1 x2 0.20 1 0.20 0.005 0.9441
x1 x3 192.67 1 192.67 4.86 0.0520
x2 x3 0.005 1 0.005 0.0001 0.9913
Residual 396.26 10 39.63 – –
Lack-of fit 394.58 5 78.92 236.13 <0.0001 Significant
Pure error 1.67 5 0.33 – –

second-order terms represent the interaction between two factors

and quadratic effect, respectively. The positive sign in front of the
terms indicates synergistic effect, whereas negative sign indicates
antagonistic effect. The quality of the models developed was eval-
uated based on the correlation coefficients, R2 value. In fact, the
models developed seems to be the best at low standard deviation
and high R2 statistics which is closer to unity as it will give predicted
value closer to the actual value for the responses [19].
In this experiment, the R2 values for Eq. (6) and (7) were 0.943
and 0.831, respectively. This indicated that 94.3 and 83.1% of the
total variation in the RBBR removal and carbon yield, respectively
were attributed to the experimental variables studied. The stan-
dard deviations for the two models were 6.29 and 1.89 for Eq. (6)
and (7), respectively. The R2 values of 0.943 for Eq. (6) was consid-
ered relatively high, indicating that the predicted values for RBBR
removal of the MP activated carbon would be more accurate and Fig. 2. Predicted versus experimental RBBR removal of MP activated carbons.
closer to its actual value. The R2 of 0.831 for Eq. (7) was considered
as moderate to validate the fit, which might lead to larger variation icant [21]. The ANOVA for the quadratic model for RBBR removal
in the MP activated carbon yield predicted from the model. How- of MP activated carbon is listed in Table 3. The model F-value of
ever, the standard deviation for this model is lower than the model 18.54 and Prob. > F of less than 0.0001 implied that this model was
for RBBR removal which indicates that the predicted values for this significant. In this cases, x1 , x3 , x12 and x32 factors were significant
model is still considered as suitable to correlate the experimental
model terms whereas x2 , x22 , x1 x2 , x1 x3 and x2 x3 were insignificant
data.
to the response.
From the ANOVA for two-factor interaction model for MP acti-
3.2. Analysis of variance vated carbon yield as shown in Table 4, the model F-value of 10.66
and Prob. > F of 0.0002 revealed that the model was significant. In
The significance and adequacy of the models were further jus- this case, x1 , x2 , x3 and x1 x2 were significant model terms whereas
tified through analysis of variance (ANOVA). In the ANOVA, the x1 x3 and x2 x3 were insignificant to the response. From the statisti-
mean squares were obtained by dividing the sum of the squares cal results obtained, it was shown that the above models (Eqs. (6)
of each of the variation sources, the model and the error variance, and (7)) were adequate to predict the RBBR removal and the acti-
by the respective degrees of freedom. The fishers variance ratio F- vated carbon yield within the range of variables studied. In addition,
value = (Sr2 /Se2 ) is the ratio of the mean square owing to regression Figs. 2 and 3 show the predicted values versus the experimental val-
to the mean square owing to error. It is the measure of variation ues for RBBR removal and carbon yield, respectively. It can be seen
in the data about the mean. The ANOVA of the regression model that the models developed were successful in capturing the cor-
demonstrates that the model is highly significant as evident from relation between the MP activated carbon preparation variables to
the calculated F-value and a very low probability value. If the value the responses when the predicted values obtained were quite close
of Prob. > F less than 0.05, the model terms are considered as signif- to the experimental values.

Table 4
Analysis of variance (ANOVA) for response surface 2FI model for MP activated carbon yield.

Source Sum of squares Degree of freedom (DF) Mean square F-value Prob. > F Comment

Model 227.50 6 37.92 10.66 0.0002 Significant

x1 89.44 1 89.44 25.14 0.0002
x2 52.93 1 52.93 14.88 0.0020
x3 49.02 1 49.02 13.78 0.0026
x1 x2 28.31 1 28.31 7.96 0.0144
x1 x3 7.55 1 7.55 2.12 0.1690
x2 x3 0.25 1 0.25 0.070 0.7957
Residual 46.25 13 3.56 – –
Lack-of fit 44.81 8 5.60 19.45 0.0023 Significant
Pure error 1.44 5 0.29 – –
 M.A. Ahmad, R. Alrozi / Chemical Engineering Journal 165 (2010) 883–890 887

30 a
25
Predicted yield (%)

21.26
18.93
20
R² = 0.831 Activated carbon 16.61
15 yield 14.28
11.96
10

5
5 10 15 20 25 30
3.00
Experimental yield (%) 850.00
2.50
812.50
Fig. 3. Predicted versus experimental MP activated carbon yield. 2.00
775.00
B: Time 1.50 737.50 A: Temperature
As a special diagnostic test for adequacy of a model, lack-of-fit 1.00 700.00
present the model performance as compared to pure error which
is based on the replicate measurements [23]. F-value for lack of fit b
is calculated as the ratio between the lack-of-fit mean square and
the pure error mean square. As the statistic parameter, the F-value 24.04
is used to determine whether the lack of fit is significant or not, at a 21.75
significant level ˛ [19]. For RBBR removal, the lack-of-fit F-value of
Activated carbon 19.46
236.13 implies the lack-of-fit is significant (Table 3). There is only a yield 17.17
0.01% chance that a “lack of fit F-value” this large could occur due to 14.89
noise. For the carbon yield, the “lack of fit F-value” of 19.45 implies
the lack-of-fit is undesirable significant relative to the pure error
(Table 4). There is a 0.23% chance that a “lack-of-fit F-value” this
large could occur due to noise.
3.50
850.00
2.88
3.3. RBBR removal 812.50
2.25 775.00
C: IR 1.63 737.50
Based on the F-value as shown in Table 3, both the activation A: Temperature
temperature and IR were found to have significant effects on the 1.00 700.00
RBBR removal of MP activated carbon, whereas activation time Fig. 5. Three-dimensional response surface plot of MP activated carbon yield; (a)
showed the least significant effect on this response. Among all the effect of activation temperature and activation time, IR = 2.25, (b) effect of activation
factors being considered, IR was found to impose the greatest effect temperature and IR, t = 2 h.
on the RBBR removal, as it showed the highest F-value. Besides, the
quadratic effects of IR and activation temperature as well as the two variables studied. The results obtained were in aggrement with
interaction effects between these two variables were considered the work done by Sudaryanto et al. [24] which reported that IR and
moderate. Fig. 4 shows the three-dimensional response surfaces activation temperature gave significant effect on the pore struc-
which was constructed to show the interaction effects of the acti- ture of activated carbon produced from casava peel, compared to
vation temperature and IR on the RBBR removal. For this plot, the activation time. Sentorun-Shalaby et al. [25] also found that activa-
activation time was fixed at zero level (t = 2 h). As can be seen from tion time did not show much effect on the surface area obtained for
Fig. 4, the RBBR removal generally increased with increase in the activated carbons prepared from apricot stones using steam activa-
tion. Increasing in activation temperature and time would entail an
opening an enlargement of the pores, which enhanced the adsorp-
tion of dye [26]. However, Gratuito et al. [27] indicated that it is
not necessary to prolong activation time so much beyond the basic
90.86
requirement. As doing so would cause pores enlargement, which
78.28
may be undesireable depending on the requirement of a specific
RBBR removal 65.71 activated carbon application.
53.13 As the most prominent factor on RBBR removal of MP acti-
40.56 vated carbon, IR played a crucial role in the formation of pores
and increase the surface area as well. This phenomena is proba-
bly due to the fact that upon the impregnation of carbons with
KOH, K2 CO3 is formed with a simultaneous evolution of CO2 and CO
3.50 [28]. In addition, activation temperature above 700 ◦ C is believed
850.00 to be responsible for the drastic expansion of the carbon material
2.88
812.50 and the creation of larger surface area and high porosity due to
2.25 775.00 the intercalation of metallic potassium [29], leading towards high
C: IR 1.63 737.50
A: Temperature RBBR removal efficiency. Tan et al. [14] have found that KOH could
1.00 700.00
accelerate the reaction rate and therefore the quantity of pores on
Fig. 4. Three-dimensional response surface plot of RBBR removal (effect of activa- the activated carbon derived from coconut husk is increased cor-
tion temperature and IR, t = 2 h) of MP activated carbons. respondingly. However, El-Hendawy [30] has observed that at too
888 M.A. Ahmad, R. Alrozi / Chemical Engineering Journal 165 (2010) 883–890

Table 5
Model validation.

Model desirability Activation Activation time, IR, x3 RBBR removal (%) Activated carbon yield (%)
temperature, x1 (◦ C) x2 (h)
Predicted Experimental Error (%) Predicted Experimental Error (%)

0.800 828 1.0 3.0 83.56 80.35 3.84 21.11 20.76 1.69

high concentration of KOH, might lead to the presence of K2 CO3 at the minimum point within the range studied. This result was in
and metallic potassium that was left in the carbon and cannot eas- agreement with the work done by Ahmad et al. [19] which found
ily leached even after repeated washing. This may cause blocking of that activation temperature play an important role on the yield of
some pores leading to the observed drastic decrease in the acces- activated carbon prepared by rattan sawdust whereas activation
sible area for nitrogen molecules that finally reduced the surface time did not show much effect on the carbon yield. Overall weight
area of the activated carbon prepared. loss was found to increase with increasing temperature, resulting
in decreasing yield of activated carbon at high temperature [21].
The weight loss was due to devolatilization of the precursors, pri-
3.4. Activated carbon yield mary to increase of the pore development and create new pores,
as a result of intensifying dehydration and elimination reactions
Referring to Table 4, all the three variables were found to be sig- [29]. As temperature increasing, C–KOH and C–CO2 reaction rate is
nificant on the response for activated carbon yield, with activation increased as well, leading towards decreasing of carbon yield [26].
temperature imposing the greatest effect on it, while activa- The effect of IR on yield of activated carbon is shown in Fig. 5(b)
tion time and IR showed almost similar effect on this response, where increasing IR decreased the carbon yield and increased the
which were considered moderate. However, the interaction effects carbon burn-off. At higher IR, the weight loss of carbon increases
between the variables were all less significants. Fig. 5(a) and due to the oxidation process on the surface of carbon atoms that
(b) shows the three-dimensional response surface which were promote by KOH [24]. Shevkoplyas and Saranchuk [31] observed
constructed to show the effects of activated carbon preparation that impregnation of coal with KOH leads to the breaking of C–O–C
variables on the MP activated carbon yield. Fig. 5(a) illustrates the and C–C bonds, facilitating coal decomposition during pyrolisis and
effect of activation temperature and activation time on the MP acti- as a result decreasing the carbon yield [21].
vated carbon yield, with IR fixed at zero level (IR = 2.25), whereas
Fig. 5(b) illustrates the effect of activation temperature and IR on
the response, with activation time fixed at zero level (t = 2 h). 3.5. Process optimization
The carbon yield was found to decrease with increasing activa-
tion temperature, activation time and IR. From both figures, it can One of the main aims of this study was to find the optimum pro-
be seen that activation temperature was more influential factor on cess parameters which activated carbons produced should have a
the MP activated carbon yield as compared to the other two vari- high carbon yield and a high RBBR removal. However, it is diffi-
ables. The highest yield was obtained when all the variables were cult to optimize both these responses under the same condition

Fig. 6. Scanning electron micrographs; (a) precursor, (b) char, (c) MP activated carbon (magnifications: 500×).
 M.A. Ahmad, R. Alrozi / Chemical Engineering Journal 165 (2010) 883–890 889

Table 6
Proximate analysis and elemental contents.

Sample Proximate analysis (%) Elemental analysis (%)

Moisture Volatile Fixed Carbon Ash C H S (N + O)a

Precursor 9.49 60.13 28.48 1.90 43.87 5.45 0.46 50.22

Char 8.42 17.44 70.45 3.69 70.84 2.54 0.20 26.82
MP activated carbon 9.08 16.36 72.93 1.63 73.10 1.27 0.08 25.55
a
Estimated by difference.

because the interest region of factors is different. When adsorption cursors decreased significantly whereas the fixed carbon content
performance increases, carbon yield will decrease and vice versa. increased in activated carbons. This might be due to pyrolytic effect
Therefore, the function of desirability was applied using Design- at high temperature where most of the organic substances have
Expert software version 6.0.6 (STAT-EASE Inc., Minneapolis, USA) been degraded and discharged both as gas and liquid tars leaving
in order to compromise between these two responses. a material with high carbon purity. A reduction of hydrogen con-
In the optimization analysis, the target criteria was set as max- tent to a value less than 1.5% was due to the organic molecular
imum values for the two responses of RBBR removal (Y1 ) and chains ruptured after undergoing carbonization and physiochemi-
activated carbon yield (Y2 ) while the values of the three variables cal activation process [33]. Besides, there is also insignificant sulfur
were set in the ranges being studied. The experimental conditions content for MP activated carbon which similar observation was
with the highest desirability were selected to be verified. The pre- obtained by Tay et al. [34] in preparing activated carbon from waste
dicted and experimental results of RBBR removal and carbon yield biomass by chemical activation of K2 CO3 and KOH.
obtained at optimum conditions are listed in Table 5. The optimum
MP activated carbon was obtained by using activation tempera-
4. Conclusions
ture, activation time and IR of 828 ◦ C, 1 h and 3.0, respectively. The
optimum activated carbon showed RBBR removal of 71.35% and
Response surface methodology was successfully used to inves-
activated carbon yield of 20.76%. It was observed that the exper-
tigate the effects of activation temperature, activation time and
imental values obtained were in good agreement with the values
IR, on the percentage removal of RBBR and yield of the activated
predicted from the models, with relatively small errors between the
carbon prepared. The optimum MP activated carbon preparation
predicted and the actual values, which was only 3.84% and 1.69%,
conditions were obtained using 828 ◦ C activation temperature, 1 h
for RBBR removal and activated carbon yield, respectively.
activation time and 3.0 IR resulting in 71.35% of RBBR removal and
20.76% of carbon yield. Through analysis of the response surface,
3.6. Characterization of activated carbon prepared under activation temperature and IR were found to have significant effects
optimum condition on RBBR removal. Activation temperature was found to have the
greatest effect on carbon yield. The MP activated carbon prepared
3.6.1. BET surface area and pore volume demonstrated high surface area and well-developed porosity. MP
The BET surface area, total pore volume and average pore diame- activated carbon was shown to be a promising adsorbent for
ter of the prepared activated carbon were found to be 994.21 m2 /g, removal of Remazol Brilliant Blue R from aqueous solutions.
0.57 cm3 /g and 2.61 nm, respectively. The average pore diameter
of 2.61 nm indicates that the MP activated carbon prepared was
in the mesopores region according to the IUPAC classification [32]. Acknowledgement
The physiochemical activation process has contributed to the high
surface area and total pore volume of the prepared activated car- The authors gratefully acknowledge the financial support
bon. The activated carbon development porosity is associated with received in the form of research grants (PJKIMIA/814003 and
gasification reaction [15]. PJAWAM/814021) from Universiti Sains Malaysia.

3.6.2. Surface morphology References

Fig. 6(a), (b) and (c), respectively shows the SEM images of the
precursor, char and the MP activated carbon. The precursor’s sur- [1] O. Tunc, H. Tanac, Z. Aksu, Potential use of cotton plant wastes for the removal
of Remazol Black B reactive dye, J. Hazard. Mater. 163 (2009) 187–198.
face textures were rough, uneven, undulating and very little pores [2] Z. Aksu, I.A. Isoglu, Use of agricultural waste sugar beet pulp for the removal
were presence as observed in other biomass [14,19]. After car- of Gemazol turquoise blue-G reactive dye from aqueous solution, J. Hazard.
bonization process, some irregular holes and pores were developed Mater. 137 (2006) 418–430.
[3] A. Demirbas, Agricultural based activated carbons for the removal of dyes from
and found on the surfaces of the char due to the sudden burst of aqueous solutions: a review, J. Hazard. Mater. 167 (2009) 1–9.
the thermal expansion from pyrolysis as shown in Fig. 6(b). From [4] Y. Al-Degs, M.A.M. Khraisheh, S.J. Allen, M.N. Ahmad, Effect of carbon surface
Fig. 6(c), it can be seen that almost homogeneous type pores struc- chemistry on the removal of reactive dyes from textiles effluent, Water Res. 34
(2000) 927–935.
ture were distributed on the surface of the MP activated carbon. This
[5] A.W.M. Ip, J.P. Barford, G. McKay, Production and comparison of high sur-
result revealed that the combination activation process of KOH and face area bamboo derived active carbons, Bioresour. Technol. 99 (2008) 8909–
CO2 were effective in creating well-developed pores, resulting to 8916.
[6] P. Baskaralingam, M. Pulikesi, V. Ramamurthi, S. Sivanesan, Modified hectorites
large surface area activated carbon with good mesoporous struc-
and adsorption studies of a reactive dye—technical note, Appl. Clay Sci. 37
ture. Similar observations were reported by other researchers in (2007) 207–214.
their work of preparing activated carbons from coconut shells [14], [7] Z. Aksu, S. Tezer, Equilibrium and kinetic modelling of biosorption of Remazol
oil palm empty fruit bunch [13] and oil palm fronds [33]. Black B by Rhizopus arrhizus in a batch system: effect of temperature, Process
Biochem. 36 (2000) 431–439.
[8] N.K. Amin, Removal of direct blue-106 dye from aqueous solution using new
3.6.3. Proximate and elemental analysis activated carbons developed from pomegranate peel: adsorption equilibrium
Table 6 presents the proximate and elemental analysis of precur- and kinetics, J. Hazard. Mater. 165 (2009) 52–62.
[9] B.H. Hameed, J.M. Salman, A.L. Ahmad, Adsorption isotherm and kinetic mod-
sor, char and MP activated carbon. After undergoing carbonization eling of 2,4-d pesticide on activated carbon derived from date stones, J. Hazard.
and activation process, the volatile matter content of the pre- Mater. 163 (2009) 121–126.
890 M.A. Ahmad, R. Alrozi / Chemical Engineering Journal 165 (2010) 883–890

[10] S. Altenor, B. Carene, E. Emmanuel, J. Lambert, J.-J. Ehrhardt, S. Gaspard, Adsorp- [22] F. Karacan, U. Ozden, S. Karacan, Optimization of manufacturing conditions for
tion studies of methylene blue and phenol onto vetiver roots activated carbon activated carbon from Turkish lignite by chemical activation using response
prepared by chemical activation, J. Hazard. Mater. 165 (2009) 1029–1039. surface methodology, Appl. Therm. Eng. 2 (2007) 1212–1218.
[11] M.S. Balathanigaimani, W.G. Shim, K.H. Park, J.W. Lee, H. Moon, Effects of struc- [23] M.Y. Noordin, V.C. Venkatesh, S. Sharif, S. Elting, A. Abdullah, Application of
tural and surface energetic heterogeneity properties of novel corn grain-based response surface methodology in describing the performance of coated carbide
activated carbons on dye adsorption, Micropor. Mesopor. Mater. 118 (2009) tools when turning AISI 1045 steel, J. Mater. Process. Tech. 145 (2004) 46–58.
232–238. [24] Y. Sudaryanto, S.B. Hartono, W. Irawaty, H. Hindarso, S. Ismadji, High sur-
[12] K. Li, Z. Zheng, X. Huang, G. Zhao, J. Feng, J. Zhang, Equilibrium, kinetic and ther- face area activated carbons from olive-seed waste residue, Micropor. Mesopor.
modynamic studies on the adsorption of 2-nitroaniline onto activated carbon Mater. 82 (2006) 79–85.
prepared from cotton stalk fibre, J. Hazard. Mater. 166 (2009) 213–220. [25] C. Sentorun-Shalaby, M.G. Ucak-Astarlıoğlu, L. Artok, C. Sarıcı, Preparation and
[13] B.H. Hameed, I.A.W. Tan, A.L. Ahmad, Preparation of oil palm empty fruit bunch- characterization of activated carbons by one-step steam pyrolysis/activation
based activated carbon for removal of 2,4,6-trichlorophenol: Optimization from apricot stones, Micropor. Mesopor. Mater. 88 (2006) 126–134.
using response surface methodology, J. Hazard. Mater. 164 (2009) 1316–1324. [26] A.C. Lua, T. Yang, Effect of activation temperature on the textural and chemical
[14] I.A.W. Tan, A.L. Ahmad, B.H. Hameed, Optimization of preparation conditions properties of potassium hydroxide activated carbons, J. Anal. Appl. Pyrol. 76
for activated carbons from coconut husk using response surface methodology, (2004) 96–102.
Chem. Eng. J. 137 (2008) 462–470. [27] M.K.B. Gratuito, T. Panyathanmaporn, R.-A. Chumnanklang, N. Sirinuntawit-
[15] A.H. Basta, V. Fierro, H. El-Saied, A. Celzard, 2-Steps KOH activation of rice taya, A. Dutta, Production of activated carbon from coconut shell: optimization
straw: an efficient method for preparing high-performance activated carbons, using response surface methodology, Bioresour. Technol. 99 (2008) 4887–4895.
Bioresour. Technol. 100 (2009) 3941–3947. [28] J. Dıaz-Teran, D.M. Nevskaia, J.L. Fierro, G.A.J. Lopez-Peinado, A. Jerez, Study of
[16] M. Osman, A.R. Milan, Mangosteen-Garcinia mangostana L. Southampton Cen- chemical activation process of a lignocellulosic material with KOH by XPS and
tre for Underutilised Crops, University of Southampton, Southampton, UK, XRD, Micropor. Mesopor. Mater. 60 (2003) 173–181.
2006. [29] D. Adinata, W.M.A. Wan Daud, M.K. Aroua, Preparation and characterization of
[17] S. Okonogi, C. Duangrat, S. Anuchpreeda, S. Tachakittirungrod, S. activated carbon from palm shell by chemical activation with K2 CO3 , Bioresour.
Chowwanapoonpohn, Comparison of antioxidant capacities and cytotoxicities Technol. 98 (2007) 145–149.
of certain fruit peels, Food Chem. 103 (2007) 839–846. [30] A.N.A. El-Hendawy, An insight into the KOH activation mechanism through the
[18] R. Zien, R. Suhaili, F. Earnestly, Indrawati, E. Munaf, Removal of Pb(II), Cd(II) and production of microporous activated carbon for the removal of Pb2+ cations,
Co(II) from aqueous solution using Garcinia mangostana L. fruit shell, J. Hazard. Appl. Surf. Sci. 255 (2009) 3723–3730.
Mater. 162 (2010) 43–50. [31] V.N. Shevkoplyas, V.I. Saranchuk, The impregnation effect on low and middle
[19] A.A. Ahmad, B.H. Hameed, A.L. Ahmad, Removal of disperse dye from aqueous rank coals structure reorganization and their behavior during pyrolysis, Fuel
solution using waste-derived activated carbon: optimization study, J. Hazard. 79 (2000) 557–565.
Mater. 170 (2009) 612–619. [32] IUPAC, IUPAC manual of symbols and terminology, Pure Appl. Chem. 31 (1972)
[20] A.M.M. Vargas, C.A. Garcia, E.M. Reis, E. Lenzi, W.F. Costa, V.C. Almeida, 587.
NaOH-activated carbon from flamboyant (Delonix regia) pods: optimization of [33] J.M. Salman, B.H. Hameed, Effect of preparation conditions of oil palm fronds
preparation conditions using central composite rotatable design, Chem. Eng. J. activated carbon on adsorption of bentazon from aqueous solutions, J. Hazard.
162 (2010) 43–50. Mater. 175 (2010) 133–137.
[21] J.N. Sahu, J. Acharya, B.C. Meikap, Optimization of production conditions for [34] T. Tay, S. Ucar, S. Karagöz, Preparation and characterization of activated carbon
activated carbons from Tamarind wood by zinc chloride using response surface from waste biomass, J. Hazard. Mater. 165 (2009) 481–485.
methodology, Bioresour. Technol. 10 (2010) 1974–1982.