Bioresource Technology 101 (2010) 3474–3483

Contents lists available at ScienceDirect

Bioresource Technology
journal homepage: www.elsevier.com/locate/biortech

Treatment of dairy wastewater by commercial activated carbon and bagasse fly

ash: Parametric, kinetic and equilibrium modelling, disposal studies
Jai Prakash Kushwaha, Vimal Chandra Srivastava, Indra Deo Mall *
Department of Chemical Engineering, Indian Institute of Technology, Roorkee, 247667 Uttarakhand, India

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

Article history: Present study reports treatment of synthetic dairy wastewater (SDW) in terms of chemical oxygen
Received 12 August 2009 demand (COD) removal by means of adsorption onto activated carbon-commercial grade (ACC) and
Received in revised form 25 November 2009 bagasse fly ash (BFA). Optimum conditions for SDW treatment were found to be: initial pH  4.8, adsor-
Accepted 1 January 2010
bent dose of 20 g/l for ACC and 10 g/l for BFA and contact time 8 h. Pseudo-second-order kinetic model
Available online 25 January 2010
was found to fit the kinetic data and Redlich–Peterson isotherm model was generally found to best rep-
resent the equilibrium data for SDW treatment by ACC and BFA. The change in entropy and enthalpy for
Keywords:
SDW adsorption onto ACC and BFA were estimated as 125.85 kJ/mol K and 91.53 kJ/mol; and 25.71 kJ/
Dairy wastewater
Adsorption
mol K and 17.26 kJ/mol, respectively. The negative values of change in Gibbs free energy indicate the fea-
Activated carbon sibility and spontaneous nature of the adsorptive treatment.
Bagasse fly ash Ó 2010 Elsevier Ltd. All rights reserved.
Disposal

1. Introduction Dairy wastewaters are generally treated by physico-chemical

and biological methods. Among biological treatment processes,
Dairy industries have grown in most of countries of the world treatment in ponds, activated sludge plants and anaerobic treat-
due to the steady rise in demand of milk and milk products. Among ment are commonly employed for dairy wastewater treatment
various nations, India is projected to become one of the largest pro- (Bangsbo-Hansen, 1985). The high energy requirement of aerobic
ducers of milk and dairy products in the world (Gupta, 1997). The treatment plants is a significant drawback of these processes.
dairy industry is the highly polluting not only in terms of volume of Longer startup time, need for the alkalinity addition and further
effluent generated (Vourch et al., 2008) as well as in terms of its treatment for effluent are the drawbacks of the anaerobic treat-
characteristics too. The dairy industry generates strong wastewa- ment. Tertiary treatment methods consist of membrane separation
ter characterized by high chemical oxygen demand (COD) and bio- methods like nano-filtration and/or reverse osmosis, adsorption
chemical oxygen demand (BOD) (Omil et al., 2003) and by high and coagulation and flocculation.
concentrations of nutrients, organic contents and pathogens Among various tertiary treatment processes, adsorption has
(USDA-SCS, 1992) along with considerable variation in pH (4.2– been found to be attractive for the removal of organic compounds
9.4), relatively large load of suspended solids (0.4–2 g/l), and large from wastewaters. Activated carbon (AC) is generally used as an
variations in wastewater supply (Kosseva et al., 2003). It is esti- adsorbent for the treatment of various types of wastewaters. How-
mated that about 2% of total milk processed is wasted into drains ever, many investigators have utilized several low-cost adsorbents
(Munavalli and Saler, 2009). This milk get mixed with waters pro- like coal fly ash, rice husk ash, bagasse fly ash (BFA), etc. for the
duced in the starting, equilibrating, stopping and rinsing of various treatment of a wide variety of wastewaters.
processing units as well as with water generated during cleaning Rao and Bhole (2002) used some low-cost adsorbents along
of floors and other day-to-day operation of the plant. The amount with powdered activated carbon (PAC) for the treatment of dairy
of wastewater generated is estimated to be in the range of 0.2–10 l wastewater. PAC was found to be better in lowering total dissolved
of effluent (average = 2.5) per litre of processed milk (Ramasamy solids (TDS) than other pretreated adsorbents like bagasse, straw
et al., 2004). dust, saw dust, coconut coir and fly ash. Sarkar et al. (2006) em-
ployed coagulation by chitosan followed by adsorption with PAC
as pretreatment steps before treating the dairy wastewater by
membrane separation method. During adsorption step, PAC
* Corresponding author. Tel.: +91 1332 285319(O), 285106 (R); fax: +91 1332
showed a maximum COD removal of 68% at pH 4, adsorbent dose
276535/273560.
E-mail addresses: [email protected] (V.C. Srivastava), id_mall2000@yahoo. of 1.5 g/l, contact time of 1.5 h for dairy wastewater having an ini-
co.in (I.D. Mall). tial COD of 2000 mg/l.

0960-8524/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biortech.2010.01.002
 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483 3475

No other study is available in literature except these two stud- were determined as 599.32 kg/m3 and 18.81 MJ/kg, respectively.
ies for the treatment of dairy wastewater by adsorption. Moreover, For BFA, respective values were 185.51 kg/m3 and 19.23 MJ/kg,
in the above two studies, mechanism of adsorption is lacking; ki- respectively.
netic, isotherm and thermodynamics aspects have not been dealt The d-spacing values provided by the XRD spectra of ACC re-
with which are important for the design of any adsorption unit. flected the presence of Moganite (SiO2), Akdalaite ((Al2O3)4H2O),
Disposal aspects of the spent adsorbents were also not studied in Tamarugite (NaAl(SO4)26H2O) Fersilicate (FeSi) and Majorite
the previous studies. In the present study, the synthetic dairy (Mg3(Fe,Al,Si)2(SiO4)3) whereas Silica (SiO2), Wollastonite (CaSiO3),
wastewater (SDW) was treated with activated carbon-commercial Aragonite (CaCO3) and Akdalaite ((Al2O3)4H2O) were the major
grade (ACC) and a low-cost adsorbent, BFA to derive experimental components identified in BFA.
data for the removal of COD. Investigations were carried out with The BET surface area of ACC and BFA were 336.6 and 168.39 m2/
ACC and BFA by varying parameters such as initial pH (pHi), adsor- g, respectively. The analysis of the BJH adsorption pore distribution
bent dose (m), contact time (t), initial COD concentration (C0) and for ACC showed that the micro-pores (d < 20 Å) had a total pore
temperature (T). Adsorption kinetics has been studied by using area of about 24%, meso-pores accounted for about 76% of the total
pseudo-first-order and pseudo-second-order kinetic models. Fre- pore area. The analysis of BFA showed that the meso-pores
undlich, Langmuir and Redlich–Peterson (R–P) isotherm models (20 Å < d < 500 Å) had a total pore area of about 99% and that the
were used to represent the adsorption equilibrium data. Thermo- macro-pores accounted for about 1% of the total pore area. The
dynamics of adsorption process have been studied and the change BET average pore diameter of ACC and BFA were found to be
in Gibbs free energy and the enthalpy, and heat of adsorption have 31.03 and 25.54 Å, respectively. Thus, BFA exhibits higher meso-
also been determined. Possibility of disposal of the spent adsor- porous nature among the two adsorbents.
bents has also been studied.
2.3. Chemicals and analytical measurements
2. Methods
All the chemicals used in the study were of analytical reagent
2.1. Wastewater (AR) grade. COD was measured using digestion unit (DRB 200,
HACH, USA) and double beam UV–visible spectrophotometer
SDW was generated in the laboratory by dissolving 4 g of milk (HACH, DR 5000, USA) and BOD was measured using OX Direct,
powder (Amulya brand, manufactured by Banaskantha District Aqualytic, Germany. The chloride content was determined by stan-
Cooperative Milk producer’s Union Ltd., Palanpur, Uttarakhand, In- dard titrimetric Volhard method and Total Kjeldahl Nitrogen was
dia) per litre of distilled water in order to generate a constant determined by using the standard Kjeldahl method.
wastewater composition throughout the experiments. Several
investigators used same method for making SDW (Balannec 2.4. Experimental programme
et al., 2005; Leal et al., 2006; Ramasamy et al., 2004). The charac-
teristics of the SDW used in the present study are presented in For each experiment, a known amount of the adsorbent was
Table 1. The synthetic wastewater was prepared freshly whenever introduced into 100 ml stoppered conical flasks in which 100 ml
required and concentration was maintained uniform throughout of the SDW of known COD and pHi was already present. This mix-
the study. ture was kept in a temperature-controlled shaker at a constant
speed of 150 rpm at a pre-decided constant temperature for 8 h
2.2. Adsorbents and their characterization to attain the equilibrium. The adsorbent was separated from the
wastewater after 8 h and analysed for COD. The percentage re-
Coconut-based ACC was supplied by Zeo Tech Adsorbents Pvt. moval of COD was calculated using the following relationship:
Ltd., New Delhi, India. It was used as procured, except for the re-
moval of very fine particles by sieving. BFA was collected from ðC 0  C e Þ100
Percent COD removal ¼ ð1Þ
the particulate collection device attached to the flue gas line of a C0
bagasse-fired boiler stack from a nearby sugar mill (Deoband sugar
where, Ce is the equilibrium COD concentration (mg/l).
mill, UP, India) and used as an adsorbent without any pretreat-
Dosage study was carried out by varying the dosages in the
ment. Detailed physico-chemical characteristics of the ACC
range of 0.5–25 g/l for both the adsorbents at the optimum pHi
(Srivastava et al., 2008) and BFA (Srivastava et al., 2007) have
and 303 K. The adsorption of SDW by the adsorbents was studied
already been presented elsewhere. Important characteristics of
over a pHi range of 3–10 at 303 K. The pHi of the adsorbate solu-
ACC and BFA are summarized as follows:
tions was adjusted using 1 N aqueous solution of either H2SO4 or
The average particle size of ACC and BFA were 1.671 mm and
NaOH. The m value was kept constant at 20 g/l for ACC and 10 g/l
0.167 mm, respectively. Bulk density and heating value of ACC
for BFA.
Kinetic parameters were evaluated at various C0 values (390–
Table 1 3900 mg/l) at optimum m of ACC and BFA at optimum pHi. The
Characteristics of synthetic dairy wastewater. amount of adsorbate adsorbed, qt (mg/g), at any time t was calcu-
Parameters Range lated as:
pH 6.3–6.8 ðC 0  C t Þ
COD (mg/l) 3900 qt ¼ V ð2Þ
BOD (mg/l) 2300
w
COD/BOD 1.69 where, Ct is the COD concentration (mg/l) at time t, V is the volume
Total solids (mg/l) 3090
Volatile solids (mg/l) 2810
of the solution (l) and w is the mass of the adsorbent (g).
Total fixed solids (mg/l) 280 Isothermal experiments were performed at 283, 293 and 303 K
Turbidity 1744 with C0 values of 39–3900 mg/l at optimum pHi. The adsorbents
Conductivity (ls/cm) 220 were separated from the wastewater after 8 h and analysed for
Chloride (mg/l) 31
Ce. The equilibrium adsorption uptake, qe (mg/g), were calculated
Total N (mg/l) 113.18
using the following relationship:
3476 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483

ðC 0  C e Þ 3. Results and discussion

qe ¼ V ð3Þ
w
3.1. Effect of adsorbent dosage (m)
2.5. Adsorption kinetic theory
The effect of m on the adsorption of SDW by ACC and BFA was
2.5.1. Pseudo-first-order and pseudo-second-order model studied at C0 = 3900 mg/l. An increase in ACC and BFA dosage re-
The COD removal from SDW by the ACC and BFA at any instant sulted in increased COD removal up to a certain value and thereaf-
(t) can be represented by pseudo-second-order model which is gi- ter the removal efficiency remained almost constant. An increase
ven as: in the removal with the increase in m can be attributed to greater
surface area and the availability of more adsorption sites. At
qt ¼ qe ½1  expðkf tÞ ð4Þ m P 20 g/l for ACC and at m P 10 g/l for BFA, the incremental
COD removal become very less. This is due to the fact that for
where, kf is the pseudo-first-order rate constant. m P optimum dosage (moptimum), the removal efficiency more de-
The pseudo-second-order model is represented as (Blanchard pends upon the concentration of the solution and less depends
et al., 1984; Ho and McKay, 1999): upon the m. Thus, the moptimum for COD removal from SDW by
tkS q2e ACC and BFA is 20 g/l and 10 g/l, respectively.
qt ¼ ð5Þ
1 þ tkS qe
The initial adsorption rate, h (mg/g min), at t ? 0 is defined as 3.2. Effect of initial pH (pHi)

h ¼ kS q2e ð6Þ The pH value of the aqueous solution is an important control-

ling parameter in the process of adsorption. Fig. 1 represents the
Marquardt’s percent standard deviation (MPSD) error function
effect of pHi on the COD removal of SDW by ACC and BFA. For both
(Marquardt, 1963) was employed to find out the most suitable ki-
the adsorbents, highest removal efficiency of 68.7% for ACC and
netic model to represent the experimental data. This error func-
26% for BFA was obtained at pHi = 4.8. The system pH changes dur-
tions is given as:
ing the adsorption process. For the experiments with C0 = 3900 mg/
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
!2
u l, the final pH values (pHf) of the adsorption process are shown as a
u 1 X n
qt;i;exp  qt;i;cal
MPSD ¼ 100t ð7Þ function of pHi in Fig. 1. For ACC, the pHf values are lower than the
nm  np i¼1 qt;i;exp pHi values for pHi > 5.5. However, for some experiments (pHi < 5.5)
the pHf values are greater than pHi. However, for BFA, the pHf val-
In this equation, the subscript ‘exp’ and ‘cal’ represent the ues are greater than the pHi values in full studied range except for
experimental and calculated values, nm is the number of measure- pHi > 9.8. It may be observed in the figure that as the pHf values in-
ments, and np is the number of parameters in the model. creases, the COD removal efficiency decreases for both the adsor-
bents. Since the rise in pH is lower for ACC, its COD removal
2.5.2. Diffusion study efficiency is higher. In case of BFA, pHf increases faster as well as
The possibility of intra-particle diffusion was explored by using it shows positive change in pH. Thus its removal efficiency is lower.
the intra-particle diffusion model (Weber and Morris, 1963). It may also be inferred that the COD removal efficiency is higher if
qt ¼ kid t 1=2 þ I ð8Þ pHi 6 4.8 or pHf 6 5.5. In further studies, pHi = 4.8 was taken as
optimum pHi.
where, kid is the intra-particle diffusion rate constant, and values of The FTIR spectra of the blank and SDW loaded ACC and BFA are
I give an idea about the thickness of the boundary layer. shown in Fig. 2. All spectra show sharp peaks at 1635 and
In order to check whether surface diffusion controls the adsorp- 1385 cm1. The peak around 1635 cm1 may be due to the CO
tion process, the kinetic data were further analysed using Boyd ki- group stretching from aldehydes and ketones and the band around
netic expression which is given by (Boyd et al., 1947): 1385 cm1 in both the adsorbents may be attributed to the car-
6 boxyl–carbonate structures. These FTIR spectra also show trans-
F ¼1 expðBt Þ or Bt ¼ 0:4977  lnð1  FÞ ð9Þ mittance around 1110 cm1 region due to the vibration of the CO
p2
group in lactones (Davila-Jimenez et al., 2005). A broad band
where, F(t) = qt/qe is the fractional attainment of equilibrium at time
t, and Bt is a mathematical function of F.
100 12
ACC BFA
2.6. Adsorption equilibrium study ACC -pHf BFA -pHf
80 10
Equilibrium adsorption equations are required in the design of
Percent removal

an adsorption system and their subsequent optimization. There-

fore, it is important to establish the most appropriate correlation 60 8
pHf

for the equilibrium isotherm curves. Srivastava et al. (2007) have

discussed the theory associated with the most commonly used 40 6
isotherm models. We used three isotherm equations namely
Freundlich (1906), Langmuir (1918) and Redlich and Peterson
20 4
(1959) to fit the experimental data for SDW adsorption onto ACC
and BFA at various temperatures.
The Chi-square error analysis function was used to find out best 0 2
fit isotherm model. It is given as: 2 4 6 8 10 12
pHi
Xn
ðqe;i;exp  qe;i;cal Þ2
CHI2 ¼ ð10Þ Fig. 1. Effect of pHi on the COD removal of SDW by ACC and BFA (T = 303 K, t = 8 h,
i¼1
qe;i;exp
C0 = 3900 mg/l, m = 20 g/l for ACC, m = 10 g/l for BFA).
 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483 3477

Fig. 2. FTIR of blank and SDW loaded ACC and BFA.

between 3100 and 3700 cm1 in BFA is indicative of the presence of nesium, 2.50% and 0.14% aluminum and 5.72% and 2.86% silica,
both free and hydrogen bonded OH groups on the BFA surface. This respectively. Thus, SDW loaded ACC and BFA contain higher
stretching is due to both the silanol groups (SiOH) and adsorbed amount of carbon as compared to blank ACC and BFA. This may
water (peak at 3400 cm1) on the surface (Abou-Mesalam, 2000). be due to the loading of SDW onto blank ACC and BFA.
SDW loaded ACC and BFA show shifting of peaks from about
1635, 1385 and 1110 cm1. This means that the functional groups 3.3. Effect of contact time
at these wave numbers participate in the SDW treatment. The
shifting occurs both to higher and lower wave numbers. Although SDW having C0 = 390–3900 mg/l and pHi = 4.8, were kept in
some inference can be drawn about the surface functional groups contact with the ACC (m = 20 g/l) and BFA (m = 10 g/l) for 8 h.
participating in the adsorption, the weak and broad bands do not Fig. 4 shows the effect of contact time (by data points) on adsorp-
provide any authentic information about the nature of the tive COD removal of SDW by ACC and BFA. Very rapid uptake can
adsorption. be seen during first 15 min for both ACC and BFA. After 3 h, COD
The morphologies of blank and SDW loaded ACC and BFA have uptake nearly reached the equilibrium value. This rapid uptake of
been examined by SEM and EDX analysis. The SEMs of the blank COD and attaining of equilibrium in a short period indicates the
and SDW loaded ACC and BFA are shown in Fig. 3. This figure re- efficacy of the adsorbents used for its use in dairy wastewater
veals surface texture and porosity of the blank and SDW loaded treatment. At the later stages, there is no significant COD removal
adsorbents. SEM micrographs at lower magnification (50) does due to the lesser availability of vacant surfaces on the solid surface
not show much difference in the texture of the blank and SDW and due to the large resistance offered to the SDW components in
loaded ACC. It only shows that the ACC is crystalline in nature the bulk phase by the SDW components already adsorbed in the
and with varying particle size. However, at higher magnification initial stages of adsorption. Contact time of 3 h can be assumed
(5000) some difference in the surface texture of the blank and to be equilibrium time for the adsorption of SDW onto ACC and
SDW loaded ACC can be envisaged. Blank ACC shows porous struc- BFA.
tures with pores of varying sizes. Numbers of pores observed in the
loaded ACC are less as compared to that in the blank ACC. SEM of 3.4. Kinetics of adsorption
BFA at 50 magnification shows its fibrous and porous structure.
After the treatment the fibres not only get reduced in size but also Two kinetic models, namely, pseudo-first-order and pseudo-
their pores get filled-up by SDW molecules. It can be inferred from second-order have been used to test their validity with the kinetic
these figures that the surface texture of the ACC and BFA changes experimental adsorption data. The best-fit values of parameters
after the SDW components loading. The change in surface texture along with the correlation coefficients and MPSD values are given
of ACC is less as compared to that of BFA. in Table 2. It may be seen that the adsorption kinetics can be
EDX was conducted to study the distribution of the elements in satisfactorily and adequately represented by the pseudo-second-
blank and SDW loaded ACC and BFA. EDX analysis showed the order model using non-linear regression fit. The fit of pseudo-
presence of 72.01% and 82.79% carbon, 0.69% and 0.81% magne- second-order model is shown in Fig. 4 by solid line for COD
sium, 1.01% and 0.22% aluminum, and 2.33% and 3.30% silica in removal of SDW by ACC and BFA.
blank ACC and BFA, respectively. SDW loaded ACC and BFA were From Table 2, it is observed that the h and qe values increase
found to contain 77.32% and 86.85% carbon, 0.35% and 0.32% mag- with an increase in C0, whereas, kS decreases with an increase in
3478 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483

Fig. 3. SEM of blank and SDW loaded ACC and BFA.

C0. The rate of uptake is limited by the C0 and the components transfer of SDW components between the aqueous and the solid
affinity to the adsorbent, diffusion coefficient of the adsorbate in phases. The increase in C0 also enhances the interaction between
the bulk and solid phases, the pore size distribution of the adsor- SDW components and the adsorbent. The rate of adsorption also
bent, and the degree of mixing (Zogorski et al., 1976). They provide increases with the increase in C0 due to increase in the driving
the necessary driving force to overcome the resistances to the mass force.
 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483 3479

150 Table 2
Kinetic parameters for the COD removal of SDW by ACC and BFA (t = 8 h, C0 = 3900–
390 mg/l, m = 20 g/l for ACC, m = 10 g/l for BFA).
125 Co(mg/l)
3900 C0 (mg/l) qe,exp (mg/g) qe,cal (mg/g) kf (min1) R2 MPSD
100 1950 Pseudo-first-order model
qt (mg/g)

975 ACC
75 390 390 13.3 12.0 0.437 0.974 10.85
975 32.3 30.0 0.808 0.987 11.06
50 1950 65.0 60.5 0.415 0.976 13.83
3900 134.0 129.5 0.499 0.987 9.62
BFA
25 390 15.0 9.0 0.219 0.737 50.22
975 36.5 32.4 0.237 0.918 25.34
0 1950 57.0 56.0 0.258 0.982 15.65
0 100 200 300 400 500 3900 101.5 93.1 0.226 0.989 11.29
Time (min)
(a) ACC C0 (mg/l) qe,cal (mg/g) ks (g/mg min) h (mg/g min) R2 MPSD
Pseudo-second-order model
120
ACC
390 12.3 0.104 15.8 0.984 9.28
100 975 31.6 0.063 62.3 0.993 6.56
Co (mg/l)
3900 1950 60.6 0.048 175.5 0.985 9.98
80 1950 3900 129.8 0.017 279.9 0.992 7.78
BFA
qt (mg/g)

975
390 390 11.6 0.013 1.7 0.889 36.14
60
975 32.9 0.013 14.2 0.959 17.06
1950 55.0 0.010 31.0 0.995 5.73
40 3900 98.4 0.004 35.0 0.996 8.33

20 C0 kid,1 (mg/ I1 (mg/ R2 kid,2 (mg/ I2 (mg/ R2

(mg/l) g min1/2) g) g min1/2) g)
0 Intra-particle diffusion
0 100 200 300 400 500 ACC
Time (min) 390 0.415 9.545 0.827 0.810 11.30 0.692
975 0.225 28.19 0.922 0.650 30.64 0.7
(b) BFA 1950 0.463 54.99 0.812 0.435 55.45 1
3900 1.317 115.5 0.979 0.054 122.8 1
Fig. 4. Effect of contact time on the COD removal of SDW by ACC and BFA.
Experimental data points given by the symbols and the lines predicted by the BFA
pseudo-second-order model. T = 303 K, m = 20 g/l for ACC, m = 10 g/l for BFA. 390 0.550 5.326 0.945 0.217 10.22 1
975 1.199 20.67 0.916 0.418 27.56 0.919
1950 4.887 29.67 1 0.333 49.56 0.772
3900 16.49 26.11 1 0.577 90.35 0.877

3.5. Controlling mechanism

The transport of SDW components from the solution phase into Fig. 5 shows a representative qt versus t0.5 plot for SDW adsorp-
the pores of the adsorbent particles may be controlled either by tion onto ACC and BFA for C0 = 390, 975, 1950 and 3900 mg/l at
one or more of the following steps: film or external diffusion, pore 303 K and pHi = 4.8. In this figure, the plots are not linear over
diffusion, surface diffusion and adsorption on the pore surface. It is the whole time range, implying that the more than one process
necessary to calculate the slowest step involved among these steps is controlling the adsorption process. The first portion (line not
to identify the controlling step during the adsorption process drawn for the clarity of picture) gives the diffusion of SDW compo-
(Srivastava and Srivastava, 2009). It is expected that the intra- nents through the solution to the external surface of adsorbent or
particle diffusion may be the rate controlling-step during SDW boundary layer diffusion. Further, two linear portions depict intra-
components adsorption onto ACC and BFA. This is due to the fact particle diffusion. The second linear portion is attributed to the
that intra-particle diffusion controls the adsorption process for sys- gradual equilibrium stage with intra-particle diffusion dominating.
tems with good mixing, large particle sizes of adsorbent and high The third portion is the final equilibrium stage for which the intra-
concentration of adsorbate (Aravindhan et al., 2007). If the particle diffusion starts to slow down due to the extremely low
Weber–Morris plot of qt versus t0.5 satisfies the linear relationship adsorbate concentration left in the solution (Crank, 1965). The
with the experimental data, then the adsorption process is as- slope of the linear portions (kid,1 and kid,2) defined as rate parame-
sumed to be controlled by intra-particle diffusion only. However, ters and are characteristics of the rate of adsorption in the region
if the data exhibit multi-linear plots, then two or more steps influ- where intra-particle diffusion is rate controlling. Extrapolation of
ence the overall-adsorption process. the linear portions of the plots back to the y-axis gives the inter-
External mass transfer is characterized by the initial solute up- cepts that provide the measure of the boundary layer thickness.
take. In the present study, it was assumed that the external mass It seems that the intra-particle diffusion of SDW components into
transfer occurred in the first 15 min and that the relationship be- meso-pores (third portion) is the rate-controlling step in the
tween C/C0 versus time for first 15 min was linear. The initial adsorption process. Slopes of second and third portions (kid,1 and
adsorption rates (Ks) (min1) were quantified as (C15 min/C0)/15. kid,2) are higher for higher C0, which corresponds to an enhanced
The calculated Ks values for an initial SDW concentration of 390, diffusion of SDW components through meso- and micro-pores.
975, 1950 and 3900 mg/l were found to be 0.028, 0.030, 0.027 This is due to the greater driving force at higher C0. The deviation
and 0.026 min1; and 0.053, 0.049, 0.051 and 0.051 min1 for of straight lines from the origin indicates that the pore diffusion
ACC and BFA, respectively. is not the sole rate-controlling step. Therefore, the adsorption
3480 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483

140 on the diffusing components decrease, thereby increasing the

adsorptive capacity of adsorbent. It is known from the previous
120 section that the diffusion of SDW components into pores of the
100 Co (mg/l) 390 adsorbent is not the only rate-controlling step, and the diffusion
975 1950 3900
process could be ignored with adequate contact time. Therefore,
qt (mg/g)

80 the increase in adsorption capacity with an increase in tempera-

ture may be attributed to chemisorptive nature of adsorption.
60

40 3.6.2. Isotherm modelling

The parameters, R2 and CHI2 values for the Freundlich, Lang-
20 muir and R–P isotherms fitting to the experimental data are listed
0 in Table 3. It may be concluded by comparing the CHI2 and R2 val-
0 5 10 15 20 25 ues, that R–P isotherm generally best-fits the equilibrium adsorp-
Time 0.5 (min0.5 ) tion of SDW components onto both ACC and BFA at all
temperatures except at 283 K for BFA where Freundlich isotherm
(a) ACC
best-fit the isotherm data. The CHI2 values are similar for R–P
120 and Freundlich isotherm and any of the two can be used for iso-
Co (mg/l) 390 975 1950 3900 therm modelling. The fitting of R–P isotherm on the experimental
100 data points is shown in Fig. 6 by solid lines. KF and 1/n indicate the
adsorption capacity and adsorption intensity, respectively. Higher
80 value of 1/n indicates the higher affinity between the adsorbate
qt (mg/g)

and adsorbent, and heterogeneity of the adsorbent sites. It may

60 be concluded from Table 3 that the ACC has greater heterogeneity
for SDW components. The KF value is considered an indicator of the
40 adsorption capacity of ACC and BFA. The magnitude of KF also
showed the higher uptake of SDW components at higher tempera-
20 ture for both the adsorbents indicating endothermic nature of
adsorption. The qm value indicates the affinity of the SDW compo-
0 nents to bind with the adsorbents. A high qm value indicates a
0 5 10 15 20 25
higher affinity. The data in Table 3 indicate that values of qm
Time0.5 (min0.5 )
(b) BFA
Table 3
Fig. 5. Weber–Morris plot for the treatment of SDW by ACC and BFA. t = 8 h, Isotherm parameters for the treatment of SDW by ACC and BFA (t) = 8 h, pHi = 4.8,
C0 = 390–3900 mg/l, m = 20 g/l for ACC, m = 10 g/l for BFA. m = 20 g/l for ACC, m = 10 g/l for BFA).

Freundlich qe ¼ K F C 1=n
proceeds via a complex mechanism consisting of both surface e
1/n
adsorption and intra-particle transport within the pores of ACC T (K) KF ((mg/g)/(l/mg) ) 1/n R2 CHI2
and BFA. ACC
The multi-phasic nature of intra-particle diffusion plot confirms 283 K 0.070 0.999 0.975 10.090
the presence of both surface and pore diffusion. In order to predict 293 K 0.148 0.915 0.991 7.149
303 K 0.152 0.959 0.993 4.540
the actual slow step involved, the kinetic data were further ana-
lysed using Boyd kinetic expression. Eq. (8) was used to calculate BFA
283 0.102 0.811 0.9885 1.356
Bt values at different time t. The linearity of the plot of Bt versus 293 0.103 0.859 0.9752 4.139
t was used to distinguish whether surface or intra-particle trans- 303 0.217 0.790 0.9926 7.141
port controls the adsorption rate. It was observed that the relation
between Bt and t (not shown here) was non-linear (R2 = 0.63–0.97) Langmuir qm K L C e
qe ¼ 1þK L Ce
at all concentrations for both the adsorbents confirming that
surface diffusion is the not the sole rate-limiting step. Thus, both T (K) KL (l/mg) qm (mg/g) R2 CHI2
surface and pore diffusion seem to be the rate-limiting steps in ACC
the adsorption process and the adsorption proceeds via complex 283 0.00006 1327.978 0.980 9.792
mechanism. 293 0.00006 1408.473 0.989 8.373
303 0.00007 1596.736 0.993 11.149
BFA
3.6. Adsorption isothermal study
283 0.0002 208.613 0.9833 2.520
293 0.0002 248.021 0.9894 2.181
3.6.1. Effect of temperature (T) 303 0.0002 278.497 0.9939 4.988
T has a pronounced effect on the adsorption capacity of the
adsorbents. Adsorption of SDW onto ACC and BFA were studied Redlich–Peterson qe ¼ K R Ce
1þaR C be

at different T ranging from 283 K to 303 K. The absorptivity of T (K) KR (l/g) aR (l/mg)1/b b R2 CHI2
the SDW components was found to increase with an increase in
ACC
T. Adsorption is generally an exothermic process. However, if the 283 0.074 0.0001 1.000 0.980 9.792
adsorption process is controlled by the diffusion process (intra- 293 0.117 0.0007 0.910 0.994 5.479
particle transport-pore diffusion), the adsorption capacity will 303 0.106 0.0007 0.000 0.994 8.926
show an increase with an increase in temperature. This is basically BFA
due to the fact that the diffusion is an endothermic process (Sri- 283 0.032 0.0006 0.8256 0.9885 2.407
vastava et al., 2007). With an increase in temperature, the mobility 293 0.051 0.0002 0.9990 0.9893 2.182
303 0.057 0.0002 0.9990 0.9937 3.356
of the SDW components increases and the retarding forces acting
 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483 3481

200 Table 4
Thermodynamics parameters for the adsorption of SDW by ACC and BFA (t) = 8 h,
283 K 293 K 303 K
C0 = 3900–390 mg/l, m = 20 g/l for ACC, m = 10 g/l for BFA).

150 Temperature K  103 (l/ DG0 (kJ/ DH0 (kJ/ DS0 (kJ/
(K) kg) mol) mol) mol K)
qe (mg/g)

ACC
100 283 57.94 9.55 25.72 125.86
293 132.34 11.90
303 117.98 12.02
50 BFA
283 37.67 8.54 17.26 91.53
293 55.46 9.79
0 303 60.94 10.35
0 500 1000 1500 2000
Ce (mg/l)
(a) ACC
SDW components onto ACC and BFA is neither fully physical nor
200 fully chemical and some complex mechanism dictates the adsorp-
283 K tion process. The positive value of DS0 suggests increased random-
293 K
ness at the solid/solution interface with some structural changes in
150
the adsorbate and adsorbent, and an affinity of the ACC and BFA to-
303 K
wards SDW components. Also, positive DS0 value corresponds to
qe (mg/g)

an increase in the degree of freedom of the adsorbed species.

100
DG0 values are negative indicating that the adsorption process
led to a decrease in Gibbs free energy. Negative DG0 indicates
50 the feasibility and spontaneity of the adsorption process.

0 3.8. Disposal of ACC and BFA

0 1000 2000 3000 4000
Ce (mg/l) The exhausted adsorbents pose problems to their disposal. The
thermal degradation characteristics of the SDW loaded ACC and
(b) BFA
BFA were studied by thermogravimetric (TG) instrument to inves-
Fig. 6. Equilibrium adsorption isotherms at different temperature for the treatment tigate the possible oxidation behavior of loaded ACC and BFA if
of SDW by ACC and BFA. Experimental data points given by symbols and the lines they are burned in oxidizing environment in any boiler or furnace
predicated by R–P isotherm model. t = 8 h, C0 = 39–3900 mg/l, m = 20 g/l for ACC, as fuel. The thermo gravimetric analysis (TGA), differential thermal
m = 10 g/l for BFA.
analysis (DTA) and differential thermal gravimetry (DTG) curves of
loaded ACC and BFA for the heating rate of 10 °C/min are shown in
increased with increase in temperature confirming endothermic Fig. 7.
nature of overall-sorption process for all SDW-adsorbent systems. The TG traces shows 10.8% loss of moisture and some volatile
It also shows that ACC has higher adsorption capacity as compared matter evolution from 24 °C to 500 °C for loaded ACC; and 20%
to BFA at all temperatures. This may be due to the higher surface moisture and volatile matter loss from 24 °C to 350 °C in loaded
area of ACC (336.6 m2/g) as compared to that of BFA (168.39 m2/g). BFA. Higher temperature drying (>100 °C) occurs due to loss of
the surface bound water of the particles. Adsorbents do not show
any endothermic transition in the first oxidation zone indicating
3.7. Estimation of thermodynamic parameters
the lack of any phase change during the heating process (Ng
et al., 2002). The weight loss 41.9% occurred between tempera-
Classical thermodynamics gives Eq. (11) for the calculation of
ture 500 °C and 645 °C for loaded ACC, and for loaded BFA
the Gibbs free energy change (DG0), the entropy change (DS0)
weight loss of 71.6% occurred from 334 °C to 556 °C. The strong
and heat of adsorption (DH0) from variation of adsorption equilib-
exothermic peak for loaded ACC centered between 400 °C and
rium constant (KD) with T
700 °C and for BFA between 400 °C and 550 °C shows the degrada-
DG0 DS0 DH0 1 tion of sample by oxidation. Generally, when the carbonaceous
ln K D ¼ ¼  ð11Þ compounds are heated at higher temperatures, the surface groups
RT R R T
decompose, producing CO (200–600 °C), CO2 (450–1000 °C), water
where, R is the universal gas constant (8.314 J/mol K). KD is found vapor and free hydrogen (500–1000 °C) (Mall et al., 2006). There-
from the slope of ln qcee versus qe (Srivastava et al., 2007). fore, the weight loss in second oxidation zone may be due to the
The adsorption of SDW components onto ACC and BFA is endo- evolution of CO2 and CO. A third oxidation zone with weight loss
thermic in nature, giving a positive value of DH0 (Table 4). The po- <1% can be envisaged at higher temperatures up to 1000 °C. Overall
sitive DH0 value confirms the endothermic nature of the overall- SDW loaded ACC and BFA showed total weight loss of 53.3% and
adsorption process. In physisorption, the bond between adsorbate 85.6%, respectively. The higher oxidation of BFA may be due to
and adsorbent is van der Waals interaction and adsorption energy its higher carbon content as shown by EDX analysis earlier.
is typically 5–10 kJ/mol. In the case of chemisorption, a chemical Heating value of blank ACC and BFA were determined as
bond is formed between molecules and the surface; the adsorption 18.81 MJ/kg and 19.23 MJ/kg, respectively. Thus, the loaded ACC
energy is comparable to the energy of a chemical bond. The chemi- and BFA can be dried and used as a fuel in the boilers/incinerators,
sorption energy is, generally, 30–70 kJ/mol for (Murzin and Salami, or can be used for the production of fuel-briquettes. The bottom
2005). In the present study, DH0 value is 25.72 kJ/mol for ACC and ash may be blended with clay to make fire bricks, or with ce-
17.26 kJ/mol for BFA. Therefore, it seems that the adsorption of ment-concrete mixture to make colored building blocks thus
3482 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483

200.0
300.0 548Cel
0.58mg/min

0.50
250.0 180.0
54Cel
0.07mg/min
200.0 542Cel
166.0 uV
0.00
160.0
150.0
-0.50
100.0 140.0

DTG mg/min
DTA uV

50.0 -1.00

TG %
120.0
0.0
27Cel -1.50
100.0% 199Cel
399Cel
-50.0 96.6%
94.8% 100.0
500Cel
89.2%
80Cel
-100.0 97.4% -2.00
80.0
-150.0
-2.50
-200.0 645Cel 1000Cel 60.0
47.3% 800Cel 46.7%
46.7%
-250.0 -3.00

100 200 300 400 500 600 700 800 900 1000
Temp Cel
(a) Loaded - ACC

600.0 200.0
2.0

180.0
400.0 500Cel 0.0
301.4 uV
160.0
-2.0
200.0
329Cel 519Cel 140.0
44.4 uV 102.4 uV
-4.0
DTA uV

DTG mg/min
0.0 -162uJ/mg -8227uJ/mg 120.0

TG %
24Cel
99.9% 200Cel
92.5% 100.0 -6.0
-200.0 450Cel
99Cel 72.5%
93.4% 334Cel 80.0 -8.0
86.8%
-400.0
60.0 -10.0

-600.0 493Cel 40.0

37.1% -12.0
556Cel 1004Cel
15.1% 14.4%
20.0
-800.0 -14.0
100 200 300 400 500 600 700 800 900 1000
Temp Cel

(a) Loaded - BFA

Fig. 7. TGA-DTA graphs of SDW loaded ACC and BFA.

disposing of SDW loaded ACC and BFA through chemical and blank ACC and BFA were determined as 18.81 MJ/kg and
physical fixation. 19.23 MJ/kg, respectively. Thus, the loaded ACC and BFA can be
The adsorption capacities of adsorbents are given in Table 4. It is dried and used as a fuel in the boilers/incinerators, or can be used
observed that the adsorption capacity of ACC is higher for SDW for the production of fuel-briquettes.
adsorption than that of BFA. However, BFA is available in abun-
dance as a waste from the sugar mills at no cost except the han-
dling charges for the collection and transportation. Thus, the
References
total cost of BFA will be extremely low in comparison to those
for ACC. Hence, treatment of SDW with BFA can be a viable alter- Abou-Mesalam, M.M., 2000. Adsorption kinetics of copper, zinc, cadmium and
native for ACC. nickel ions on synthesized silico-antimonate ion exchanger. Colloid Surf. A:
Physicochem. Eng. Aspects 225, 85–94.
Aravindhan, R., Rao, J.R., Nair, B.U., 2007. Removal of basic yellow dye from aqueous
4. Conclusions solution by adsorption on green algae: Caulerpa scalpelliformis. J. Hazard. Mater.
142, 68–76.
Balannec, B., Vourch, M., Rabiller-Baudry Chaufer, M.B., 2005. Comparative study of
Present study shows that ACC is better adsorbent than BFA for different nanofiltration and reverse osmosis membranes for dairy effluent
the removal of COD from SDW. Optimum conditions for SDW treat- treatment by dead-end filtration. Sep. Purif. Technol. 42, 195–200.
ment were found to be initial pH  4.8, contact time 8 h, adsor- Bangsbo-Hansen, D.I., 1985. Treatment of dairy wastewater in the developing
countries – the Danish experience. Ind. Environ. 8, 10–12.
bent dose = 20 g/l for ACC and 10 g/l for BFA. The overall-sorption Blanchard, G., Maunaye, M., Martin, G., 1984. Removal of heavy metals from water
process was found to be endothermic in nature. Heating value of by means of natural zeolites. Water Res. 18, 1501–1507.
 J.P. Kushwaha et al. / Bioresource Technology 101 (2010) 3474–3483 3483

Boyd, G.E., Adamson, A.W., Meyers, L.S., 1947. The exchange adsorption of ions from Omil, F., Garrido, J.M., Arrojo, B., Mendez, R., 2003. Anaerobic filter reactor
aqueous solution by organic zeolites. II Kinetics. J. Am. Chem. Soc. 69, 2836– performance for the treatment of complex dairy wastewater at industrial
2848. scale. Water Res. 37, 4099–4108.
Crank, J., 1965. The Mathematics of Diffusion, first ed. Oxford Clarendon Press, Ramasamy, E.V., Gajalakshmi, S., Sanjeevi, R., Jithesh, M.N., Abbasi, S.A., 2004.
London. Feasibility studies on the treatment of dairy wastewaters with upflow
Davila-Jimenez, M.M., Elizalde-Gonzalez, M.P., Pelaez-Cid, A.A., 2005. Adsorption anaerobic sludge blanket reactors. Bioresour. Technol. 93, 209–212.
interaction between natural adsorbents and textile dyes in aqueous solution. Rao, M., Bhole, A.G., 2002. Removal of organic matter from dairy industry
Colloid Surf. A: Physicochem. Eng. Aspects 254, 107–114. wastewater using low-cost adsorbents. J. Indian Chem. Eng. Sect. A 44 (1),
Freundlich, H.M.F., 1906. Over the adsorption in solution. J. Phys. Chem. 57, 385– 25–28.
471. Redlich, O., Peterson, D.L., 1959. A useful adsorption isotherm. J. Phys. Chem. 63,
Gupta, P.R., 1997. Dairy India, fifth ed. National Dairy Development Board, Anand. 1024–1026.
Ho, Y.S., McKay, G., 1999. Pseudo-second order model for adsorption processes. Sarkar, B., Chakrabarti, P.P., Vijaykumar, A., Kale, V., 2006. Wastewater treatment in
Process Biochem. 34, 451–465. dairy industries – possibility of reuse. Desalination 195, 141–152.
Kosseva, M.R., Kent, C.A., Lloyd, D.R., 2003. Thermophilic bioremediation strategies Srivastava, A., Srivastava, V.C., 2009. Adsorptive desulfurization by activated
for a dairy waste. Biochem. Eng. J. 15, 125–130. alumina. J. Hazard. Mater. 170, 1133–1140.
Langmuir, I., 1918. The adsorption of gases on plane surfaces of glass, mica and Srivastava, V.C., Mall, I.D., Mishra, I.M., 2007. Adsorption thermodynamics and
platinum. J. Am. Chem. Soc. 40, 1361–1403. isosteric heat of adsorption of toxic metal ions onto bagasse fly ash (BFA) and
Leal, M.C.M.R., Freire, D.M.G., Cammarota, M.C., Anna, G.L., Sant, Jr., 2006. Effect of rice husk ash (RHA). Chem. Eng. J. 132 (1–3), 267–278.
enzymatic hydrolysis on anaerobic treatment of dairy wastewater. Process Srivastava, V.C., Mall, I.D., Mishra, I.M., 2008. Adsorption of toxic metal ions onto
Biochem. 41, 1173–1178. activated carbon. Study of sorption behaviour through characterization and
Mall, I.D., Srivastava, V.C., Agarwal, N.K., 2006. Removal of orange-G and methyl kinetics. Chem. Eng. Process. 47, 1275–1286.
violet dyes by adsorption onto bagasse fly ash – kinetic study and equilibrium US Department of Agriculture-Soil Conservation Service (USDA-SCS). Agricultural
isotherm analyses. Dyes Pigment 69, 210–223. Waste Management Field Handbook. Washington, DC, 1992.
Marquardt, D.W., 1963. An algorithm for least-squares estimation of nonlinear Vourch, M., Balannec, B., Chaufer, B., Dorange, G., 2008. Treatment of dairy industry
parameters. J. Soc. Ind. Appl. Math. 11, 431–441. wastewater by reverse osmosis for water reuse. Desalination 219, 190–202.
Munavalli, G.R., Saler, P.S., 2009. Treatment of dairy wastewater by water hyacinth. Weber Jr., W.J., Morris, J.C., 1963. Kinetics of adsorption on carbon from solution. J.
Water Sci. Technol. 59 (4), 713–722. Sanitary Eng. Div. ASCE 89, 31–59.
Murzin, D., Salami, T., 2005. Chemical Kinetics. Elsevier, Amsterdam. Zogorski, J.S., Faust, S.M., Hass, J.H., 1976. The kinetics adsorption of phenols by
Ng, J.C.Y., Cheung, W.H., McKay, G., 2002. Equilibrium studies of the adsorption of granular activated carbon. J. Colloid Interface Sci. 55 (2), 329–341.
Cu (II) ions chitosan. J. Colloid Interface Sci. 255, 64–74.