Chemical Engineering Journal 172 (2011) 1111–1125

Contents lists available at ScienceDirect

Chemical Engineering Journal

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

Development of magnetic activated carbon from almond shells for trinitrophenol

removal from water
Dinesh Mohan a,∗ , Ankur Sarswat a , Vinod K. Singh a , María Alexandre-Franco b , Charles U. Pittman Jr. c
a
School of Environmental Sciences, Jawaharlal Nehru University, New Delhi 110067, India
b
Departamento de Química Orgánica e Inorgánica, Universidad de Extremadura, Avda. de Elvas s/n, 06071 Badajoz, Spain
c
Department of Chemistry, Mississippi State University, Mississippi State, MS 39762, USA

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

Article history: Magnetic and non-magnetic low cost activated carbons from almond shells were prepared, characterized,
Received 20 April 2011 and used to remove 2,4,6-trinitrophenol (TNP) from water. The magnetic carbon was made by mixing
Received in revised form 19 June 2011 aqueous suspensions of activated carbon with an aqueous Fe3+ /Fe2+ solution followed by treatment with
Accepted 20 June 2011
NaOH. The morphologies and surface chemistries of both magnetic and non-magnetic activated carbons
were studied by FT-IR, X-ray, SEM, SEM-EDX, TEM, EDXRF, pHpzc , elemental analysis, surface area (SBET ),
Keywords:
density and magnetic moment determinations. The SBET areas of non-magnetic and magnetic activated
Almond shell activated carbon
carbons were 733 and 527 m2 g−1 , respectively. Batch sorption studies were performed at different tem-
Magnetic activated carbon
Low cost adsorbent
peratures, pH values and solid-to-liquid ratios. Sorption isotherms were obtained over a concentration
Picric acid range from ∼23 to 228 mg/L (1.0 × 10−4 to 1.0 × 10−3 M). As temperature increased, TNP adsorption mod-
Trinitrophenol removal estly decreased on non-magnetic activated carbon (ASAC) but slightly increased on magnetic activated
Water carbon (MASAC). Sorption performances at different temperatures were evaluated using the Freundlich,
Wastewater Langmuir, and Sips adsorption models. First and second order kinetic models were tested, confirming
second order behavior. Both ASAC and MASAC readily removed TNP in single and binary systems. MASAC
and ASAC gave similar overall adsorption results with TNP, although the lower carbon content per gram
of adsorbent and lower surface area of MASAC reduced its overall adsorption capacity per unit weight.
MASAC is easily manipulated by low external magnetic fields, permitting its use for easy separation
from aqueous-phase environmental applications. Almost complete (∼97%) desorption of TNP was read-
ily achieved by methanol and hot water. Magnetic removal would allow convenient adsorbent separation,
desorption and recycling or replacement from contaminated water.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction military purposes. The invention of efficient detonators, along with

Phenols are the most prevalent organic pollutants in wastew-
aters. Major sources of phenolic effluents include petroleum
refineries, coke ovens, synfuels production facilities, wood preser-
vatives plants and chemical manufacturing [1]. Phenols impart
disagreeable tastes and odors to drinking water, taint fish flesh,
and exert various other adverse environmental effects at very
low concentrations. Thus, stringent limits have been imposed on
the discharge of phenols into municipal sewage treatment. Phe-
nols are also on the Environmental Protection Agency’s (USEPA)
priority pollutants list [1]. By the late nineteenth century the
nitroaromatic high explosives 2,4,6-trinitrotoluene (TNT) and
2,4,6-trinitrophenol (picric acid) had been synthesized with inher-
ent stability until detonation. Their advantages over the more
shock-sensitive nitroglycerine were recognized, particularly for
∗ Corresponding author. Tel.: +91 11 26704616; fax: +91 11 26704616.
E-mail address: dm 1967@hotmail.com (D. Mohan).
refinements in production processes, allowed the incorporation of
TNT and picric acid into ordnance [2,3].
Traces of 2,4,6-trinitrophenol in water/wastewater is an envi-
ronmental concern due to its toxicity. 2,4,6-Trinitrophenol (TNP)
is acidic due to its three electron-withdrawing nitro groups which
stabilize the phenoxide ion. TNP has applications in dyes, explo-
sives, analytical reagents, germicides, fungicides, staining agents,
tissue fixatives, tanning agents, photochemicals, pharmaceuticals,
and for the oxidation and etching of iron, steel, and copper sur-
faces [4]. TNP is also a by-product in nitrobenzene synthesis [5]. The
Bureau of Indian Standards’ permissible phenol limit for drinking
water is 0.001 mg/L [6]. The World Health Organization (WHO) rec-
ommended 0.001 mg/L as the permissible phenolic concentration
in potable water [7]. USEPA issued a permissible limit of 0.1 mg/L
in wastewater [8]. Therefore, removal of phenolic impurities from
wastewater prior to discharge is necessary.
Adsorption is now evolving as a front line of defense. Selec-
tive adsorption with biological materials, mineral oxides, activated
carbons, or polymeric resins have attracted attention. Neverthe-

1385-8947/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.cej.2011.06.054
1112 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125
less, less attention has been paid to picric acid remediation versus
other nitro and chloro-phenols. Adsorbents used for various phe-
nols were reviewed by Dabrowski et al. [9].
Granular activated carbon is prepared from a variety of
carbon-containing feedstocks by pyrolysis into char and activa-
tion (physical activation) or by carbonization in the presence of the
activating agent(chemical activation). Activated carbon pore vol-
umes, pore surface areas, pore surface chemistry and the chemical
functionality present on these surfaces determine its adsorption
capacity [10]. Its chemical structure influences its interaction with
adsorbates, and its active sites determine which chemical reac-
tions occur with other molecules. All can be custom designed by
selecting the right precursor and choosing the pyrolysis and acti-
vation conditions. High volume lignocellulosic by-products may
offer inexpensive and renewable sources of special granular acti-
vated carbons [10]. This option is particularly attractive since crops
grown worldwide yield considerable quantities of agricultural by-
product wastes [10]. Their conversion into activated carbons would
add value to agricultural commodities, reduce waste disposal costs
and provide inexpensive replacement of commercial adsorbents
derived from nonrenewable sources [10].
This study focused on 2,4,6-trinitrophenol remediation from
water and wastewater using both magnetic and nonmagnetic acti-
vated carbons. Activated carbons are usually made from coal, peat,
coconut shells or wood [11–13]. Nut shells are renewable low-value
agricultural wastes that are activated carbon sources. Nutshells are
preferred precursors. They are hard and do not require a binder.
Nutshells, when activated, release very little sulfur or ash. Nutshell-
derived carbons represent a greener technology than using coal. We
have been engaged in developing low cost activated carbons using
low-value agricultural byproducts and wastes for water remedia-
tion.
Magnetic filtration is emerging as a water treatment technology
which can provide rapid, efficient contaminant removal from aque-
ous waste streams [14]. Inexpensive adsorbents could be developed
that can bind to environmental contaminants and then be magnet-
ically separated. Magnetic activated carbon adsorbents can easily
be separated from solution using a magnetic separator even if the
solution contains a significant concentration of solids. In contrast,
traditional adsorbents are removed by screening. Various mag-
netic adsorbents have been used to remove contaminants. These
include magnetic ion exchange resins, magnetic solvent extrac-
tants, magnetic zeolites, magnetic activated carbon fibers, magnetic
composites, magnetized chitosan beads, magnetic nanoparticles,
magnetic nanobowls, magnetic nanoparticles-impregnated tea
waste, and magnetic polymer adsorbents. Magnetic activated car-
bons are not fully investigated, but some synthetic routes have been
reported [15–21].
The work presented here successfully recycled almond shells
into a low cost activated carbon (ASAC) and further converted this
into magnetic activated carbon (MASAC). Both of these carbons
(magnetic and nonmagnetic) were utilized for TNP removal. ASAC
was prepared and converted into MASAC. Its magnetic properties,
surface and pore properties were characterized. Then, TNP removal
from water by both ASAC and MASAC at different pHs, tempera-
tures, and solid to liquid ratios was demonstrated. Next, sorption
kinetics were measured to establish the mechanism. Finally, these
carbon adsorbents were compared to commercially available car-
bons. Desorption studies were also carried out.
2. Reagents and equipment
All chemicals were either AR or GR-grades. Ferric chloride
(>96%), ferrous sulfate (99.5%), and sodium hydroxide (98%) were
purchased from Merck, India.
A stock solution of 2,4,6-trinitrophenol (∼228 mg/L) was pre-
pared in doubly distilled water. The pH measurements were made
using a pH meter (Model EUTECH, pH510). Test solution pHs
were adjusted using H2 SO4 (0.1N) and NaOH (0.1N). The 2,4,6-
trinitrophenol concentrations were measured using a double beam
UV–vis spectrophotometer model Perkin Elmer Lambda 35 at
356 nm.
The electrical conductivity (EC) and major anions viz. chlo-
ride, fluoride, and nitrate were analyzed using multi parameter ion
meter (model Thermo Orion 5 Star). The major cations viz. sodium,
potassium, calcium, and barium were analyzed using Flame Pho-
tometer model CL-378, Elico, India.
Agitation of adsorbent–adsorbate was carried out on a water
bath incubator shaker (models RC51000 and MSW-275). A Met-
tler Toledo balance (model AB 265-S/FACT) was used for weight
measurements. The pyrolysis of almond shells was carried out
in a Thermo Scientific Thermolyne atmosphere-controlled muf-
fle furnace [easily adjustable air flow rate (0–80 L/min); rise time:
100 min (to 750 ◦ C); temperature range: 100–975 ◦ C].
2.1. Surface area measurements
The N2 isotherms were determined in a Quantachrome sur-
face area analyzer (model Autosorb-1). Samples (0.15 g) were
out-gassed at 250 ◦ C for 12 h at a pressure <10−3 Torr, prior to con-
ducting adsorption measurements. The mercury density (Hg ) was
measured by obtaining the sample volume by exclusion of the mer-
cury volume from the volume of a previously calibrated glass cell
using a Quantachrome Poremaster mercury porosimeter. About
0.30 g of sample was used in each density test. The helium density
(He ) was measured in a Quantachrome Stereopycnometer, using
approximately 2–3 g of sample [22].
The specific surface areas (SBET ) were evaluated from the N2
adsorption isotherms by applying the BET equation [23] in the rel-
ative pressure (p/p0 ) range between 0.05 and 0.35, and taking am
(i.e., the average area occupied by a molecule of N2 in the completed
monolayer) to be equal to 16.2 Å2 . Also, the micro- and mesopore
volumes (Vmi , Vme ) were obtained from the adsorption isotherms.
Vmi was the volume of N2 adsorbed (Vad ) at p/p0 = 0.10 and Vme
the volume of N2 adsorbed (Vad ) at p/p0 = 0.95 minus the Vad at
p/p0 = 0.10. Micropore volumes (W0 ) were obtained by Dubinin-
Radushkevich equation [24,25],
 
p0
log W = log W0 − D log2 (1)
p
where W is the micropore volume that has been filled with liquid
N2 when the relative pressure is p/p0 and W0 is the total micropore
volume. D is a characteristic constant of the micropore structure of
the adsorbent. Vmi , Vme , and W0 were expressed as liquid volumes.
The total pore volume (VT ) was calculated using the expression (2):
1 1
VT = − . (2)
Hg He
2.2. Infrared spectroscopy
The FTIR spectra were recorded on a PerkinElmer spectrome-
ter model Spectrum 100, on KBr pellet samples between 4000 and
400 cm−1 , 8 scans being taken at 4 cm−1 resolution.
2.3. Zero point charge (pHpzc ) measurement
The zero charge (pHpzc ) of the magnetic and nonmagnetic acti-
vated carbons were also measured [26]. pHpzc was measured using
0.01 M NaCl aqueous solutions at pH 2, 4, 6, 8, and 10. These pH
values were fixed with either a 0.1N HCl or a 0.1N NaOH aqueous
 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125
solution. These solutions (5 mL of each) were brought into contact
with 0.01 g of sample and the system was stirred for 48 h. The super-
natant was then decanted and its pH was measured. The pHpzc value
was obtained from a plot of pH of the initial solution against pH of
the supernatant.
2.4. X-ray powder diffraction
X-ray diffraction (XRD) patterns for magnetic and nonmagnetic
activated carbons were obtained on a powder X-ray diffraction sys-
tem from PANalytical model X’Pert PRO using Cu-K␣ ( = 1.54 Å)
radiation at 45 kV and 40 mA. The samples were scanned from
5.0250 to 89.9750◦ with a scan speed of 2◦ min−1 .
2.5. Scanning electron microscopy (SEM) and energy dispersive
analysis (EDX)
The surface physical morphology of almond shell activated
carbon and almond shell magnetic activated carbon was exam-
ined using a field emission scanning electron microscope (FE-SEM)
model Hitachi S-4800II (accelerating voltage 10,000 V, working
distance 9900 ␮m and emission current 13,300 nA). In the SEM
analysis, the samples were coated with a thin layer of gold and
mounted on a copper stab using a double stick carbon tape. The
elemental composition of the sample was determined from the
SEM/EDX analysis. Energy dispersive analysis by X-ray (EDX) was
carried out on sintered pellets using a Zeiss, EVO 40 scanning elec-
tron microscope in conjunction with a BRUKER EDX system.
2.6. Energy dispersive X-ray fluorescence analysis (EDXRF)
EDXRF was carried out on energy dispersive X-ray fluorescence
Spectrometer model PANalytical Epsilon 5 to determine the pres-
ence of major elements in magnetic and nonmagnetic activated
carbons. Samples pellets were made by mixing boric acid using
an Insmart System (INSMART XRF 40) at an applied pressure of
5 tons. The prepared pellet size was 34 mm and exposure area
was 8 mm.
2.7. Transmission electron microscopy (TEM)
Magnetic and non-magnetic activated carbons were observed
by a 200-kV accelerating voltage Transmission Electron Microscope
model JEOL 2100F (Japan). TEM samples were prepared by dis-
persing the magnetic or nonmagnetic activated carbon in warm
millipore water by mixing ultrasonically for 20 min. The sample
was then deposited onto a carbon coated grid. Images showing
the morphology of the carbon were recorded using a JEOL 2100F
microscope operated at 200 keV.
2.8. Hysteresis and magnetic moment measurements
The magnetic properties of a ferromagnetic material are rep-
resented by plots of magnetization (M) against the field strengths
giving the hysteresis loop. The hysteresis measurements were car-
ried out on a Quantum Design SQUID MPMS-XL susceptometer at
5 K and 300 K in a magnetic field that varied from −7 T to +7 T.
2.9. Proximate analysis
The elemental analysis was determined in an elemental ana-
lyzer: model LECO CHNS-932. Ash content was determined by
incinerating in air ∼1 g of the sample at 650 ◦ C for 12 h in an elec-
trical furnace.
1113
3. Experimental
3.1. Pyrolysis of almond shells and activated carbon development
Almond shells were soaked for 24 h in a 1:1 wt ratio with 50 wt.%
H3 PO4 (Figure S1 in Supplementary Material) at room temperature.
The samples were placed in atmosphere-controlled muffle furnace,
flushed with nitrogen (flow rate = 0.1 m3 /h) heated to 170 ± 0.5 ◦ C
and held for 0.5 h. Then they were exposed to 450 ± 0.5 ◦ C where
they were held for 1 h, followed by cooling to room temperature
under nitrogen. The temperature and time were optimized. The car-
bon adsorbents that were exposed to pyrolysis temperatures above
or below 450 ◦ C exhibited lower adsorption capacities than those
pyrolyzed at 450 ◦ C. Next, the products were washed 20 times with
distilled water to remove extra acid until the pH was at or near 7.0
and then dried for 24 h at 100 ◦ C. The activated carbon was sieved
into >50, 50–100, 100–200, 200–300 mesh samples and stored in a
vacuum desiccator until required. These carbons were designated
as (activated carbon derived from acid treated almond shells) ASAC.
The 50–100 mesh carbon was used in both the sorption and kinetic
studies unless otherwise stated.
Filtrasorb 400 granular activated carbon provided by Calgon
Carbon Corporation was also tested to compare sorption capac-
ities. F-400 is manufactured from bituminous coal to produce a
high density, high surface area, durable granular product capable of
withstanding the abrasion and dynamics associated with repeated
reactivation, hydraulic transport, backwashing and mechanical
handling. This carbon was developed especially for organic pollu-
tant removal from wastewaters.
3.2. Development of magnetic activated carbon
A diagram for the preparation of MASAC is shown Fig. 1. MASAC
was prepared according to the method given by Nakahira et al. [27]
with slight changes. ASAC (50 g, 50–100 mesh) was suspended in
distilled water (500 mL). A ferric chloride solution was freshly pre-
pared by adding 18 g FeCl3 to distilled water (1300 mL). A ferrous
sulfate solution was prepared by adding 20 g FeSO4 to distilled
water (150 mL). Both solutions were combined and vigorously
stirred at 60–70 ◦ C. The suspension formed was then added into an
aqueous suspension of almond shell activated carbon at room tem-
perature and slowly stirred for 30 min. After mixing, 10 M-NaOH
(aqueous) was added drop wise into the suspension until the pH
raised to 10–11. During NaOH addition, the suspension became
dark brown at pH ∼ 6 and then black at pH ∼ 10. After mixing for
60 min, the suspension was aged at room temperature for 24 h and
then repeatedly washed with distilled water followed by ethanol.
The MASAC was vacuum filtered and dried overnight at 50 ◦ C in a
hot air oven.
3.3. Sorption procedure
Batch sorption studies were conducted to obtain rate and equi-
librium data at different temperatures and adsorbent doses and to
construct equilibrium isotherms.
TNP adsorption isotherms were achieved in the concentration
range of ∼23–227 mg/L (1 × 10−4 to 1.0 × 10−3 M). This range was
selected based on the phenol concentrations typically found in
water. In the equilibrium studies, solution pH and temperature
were adjusted. A known amount (1.0 g/L) of either ASAC or MASAC
(particle size 50–100 mesh) was added to each flask followed by
agitation for specified times to a maximum of 72 h. Preliminary
experiments demonstrated that equilibrium was established in
48 h. No further uptake occurred between 48 and 72 h. Solutions
were then filtered. The absorbances were recorded at a wavelength
1114 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125

Fig. 1. General scheme for the magnetic activated carbon (MASAC) preparation.

of 356 nm and the TNP removed/adsorbed was calculated using
Eq. (3):
(C0 − Ce )V
qe =
Wt
where qe is the amount (mg/g) of TNP adsorbed, C0 and Ce are the
initial and equilibrium TNP concentrations (mg/L) in solution, V is
the TNP volume (L), and Wt is the ASAC or MASAC weight (g).
The effects of contact time and adsorbate/adsorbent concentra-
tions were then studied on the adsorption process. Predetermined
amounts of activated carbon were added to stoppered flasks con-
taining 50-mL solutions of TNP at specific temperatures. The
solutions were agitated. Activated carbon was separated at spe-
cific intervals and the solutions were analyzed for TNP uptake
(Eq. (3)).
3.4. Desorption procedure
First, MASAC was loaded with TNP (8 mg/g). A known amount
(1.0 g/L) of MASAC (particle size 50–100 mesh) was added to an
aqueous solution containing 23 mg/L of TNP followed by agitation
for specified times to a maximum of 72 h. Desorption of TNP from
MASAC (1.0 g/L) was carried out with methanol (40 mL) followed
by hot water (50 mL) in different increments. The concentration of
TNP adsorbed/desorbed was estimated by recording absorbances
at a wavelength of 356 nm.
open knee exhibited by isotherms indicates a narrow porosity and
a wide range of micropore sizes. Microporosity development is sub-
stantially smaller for the MASAC. SBET and W0 are 733 m2 g−1 and
(3) 0.41 cm3 g−1 for ASAC and 527 m2 g−1 and 0.30 cm3 g−1 for MASAC
(Table 1). However, MASAC has less carbon present in it per unit
weight than does ASAC because 17% of the weight of MASAC is
due to iron oxides present in its structure. Thus, the 26–27% differ-
ence surface area between ASAC and MASAC is partially due to the
smaller fraction of carbon present in MASAC. The evolution of N2
adsorption above p/p0 = 0.10 shows mesoporosity development is
also different for the two types of samples. It is small for ASAC and
significant for MASAC.
pHpzc values are given in Table 2. The PZC is defined as the
pH where the net surface charge, resulting from the adsorption of
H3 O+ and OH− , is zero (the pH value below which the surface of
the carbon particles is positively charged). The pH of zero charge
(pHpzc ) is the pH above which the surfaces of the carbon particles
are negatively charged. The higher the surface acidity, the lower
the pHpzc will be. The pHpzc is 2.0 for ASAC versus 6.8 for MASAC.
Thus, the ASAC is an “L” type carbon while MASAC is an “H” type
(or basic) carbon. H types carbons have a lower density of oxygen-
containing functional groups and basic pHpzc values; they are more
hydrophobic and exhibit strong acid adsorption. In contrast, L (or
acidic) carbons have a higher density of oxygen-containing func-

4. Results and discussion

4.1. Characterization of magnetic and non-magnetic activated

carbons

Cumulative pore volume versus pore radius plots and N2 adsorp-

tion isotherms of MASAC and ASAC are given in Fig. 2. Cumulative
pore volume versus pore radius plots (Fig. 2a) show that MASAC
possesses a significantly less developed mesoporosity than ASAC.
The slope changes indicate that the content of narrow mesopores is
noticeably lower for MASAC. Furthermore, both ASAC and MASAC
are more mesoporous than macroporous, as shown by the Vme-p
(mesopore volume) and Vma-p (macropore volume) values. The total
pore volume VT values are 0.95 cm3 g−1 and 0.97 cm3 g−1 for ASAC
and MASAC, respectively.
The N2 adsorption isotherms of MASAC and ASAC (Fig. 2b) are
typical type I isotherms of the BDDT classification system [28],
Fig. 2. (a) Pore size distribution of ASAC and MASAC. (b) This inset shows the adsorp-
denoting microporous solids. Microporosities are well developed, tion isotherms for N2 at 77 K for ASAC and MASAC (Outgas temperature: 250 ◦ C;
as inferred from the large adsorption at very low p/p0 values. An Cross-sectional area: 16.2 Å2 /molecule; Outgas time: 12.0 h).
 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125 1115

Table 1
Surface areas, pore volumes and density values for ASAC and MASAC.

Carbon samples SBET Vma-p W0 Vme-p VT Hg Bulk density Apparent density Skeletal density
(m2 g−1 ) (cm3 g−1 ) (cm3 g−1 ) (cm3 g−1 ) (cm3 g−1 ) (g cm−3 ) (g cm−3 ) (g cm−3 ) (g cm−3 )

ASAC 733 0.06 0.41 0.10 0.95 1.84 0.67 0.67 0.74
MASAC 527 0.05 0.30 0.07 0.97 2.01 0.68 0.68 0.77

N2 isotherm at −196 ◦ C; SBET (specific surface area, BET equation, p/p0 = 0.05–0.35, am = 16.2 Å2 ), W0 (micropore volume, Dubinin-Radushkevich equation). Mercury porosime-
try: Vme-p (mesopore volume), Vma-p (macropore volume), Density measurement: VT (total pore volume) = 1/Hg − 1/He , He = helium density and Hg = mercury density.

tional groups and an acidic pHpzc ; they are hydrophilic and adsorb
bases. Oxygen, hydrogen and nitrogen can be found in the acti-
vated carbon matrix. Oxygen is the dominant heteroatom and its
presence in such surface functions as carboxyls, carbonyls, phe-
nols, enols, lactones, and quinones on activated carbon surfaces
is known [29]. The oxygen content of activated carbons may range
from 1 up to 25% by weight. Although surface functional groups may
cover a small fraction of the activated carbon surface, they signifi-
cantly influence adsorption properties and reactivities of activated
carbons.
The FTIR spectra of the ASAC and MASAC are given in Fig. 3a.
The ASAC displays absorption bands at 3420, 1710, 1548 and
1300–1000 cm−1 , assigned O–H, C O, C C and C–O bonds stretch-
ing modes, respectively. The position and shape of the band at
3420 cm−1 are compatible with the involvement of hydrogen-
bonded hydroxyl groups. Free OH groups are located at ∼3625 cm−1
for alcohols, 3605 cm−1 for phenols and 3530 cm−1 for carboxylic
acids. Based on the four types of hydrogen-bonded structures
reported previously [30], self-associated OH groups centered at
3400 cm−1 predominate in ASAC. The band at 3420 cm−1 is slightly
broader toward lower wavelengths suggesting that some OH-
ether hydrogen bonds are present. The 1710 cm−1 band intensity
denotes a high C O group content. The broad band centered at
1584 cm−1 indicates that a variety of C = C bonds exists besides
those in aromatic rings [31]. Also, water absorbs in the region
at about such a scattering angle was associated with the (002)
diffraction. It has been used to estimate the stacking number of
graphene sheets (Lc or N) [33]. The X-ray diffraction patterns for
MASAC display a number of sharp peaks which are compatible
with the presence of Fe(OH)2 and Fe(OH)3 (peaks at 2 = 39.09,
40.06 and 60.746◦ ) and of Fe2 O3 (peaks at 2 = 49.437, 55.259 and
65.910◦ ). This illustrates that domains of iron species exist which
are crystalline in the MASAC sample. These iron species are presum-
ably formed during the preparation of MASAC as a consequence of
adsorption, hydrolysis and precipitation processes. These should
typically be more favorable for Fe3+ versus Fe2+ on account of the
higher ionic potential of ferric ion. Note, the solubility product
constant is ≈10−36 for Fe(OH)3 and 10−14 to 10−17 for Fe(OH)2 .
Furthermore, Fe(OH)2 may be oxidized by dissolved oxygen (i.e.,
the oxygen concentration in water at 25 ◦ C and 1 atm is 0.267 mM)
[34] during the wet impregnation of the starting carbon:
2Fe(OH)2 + 12 O2 + H2 O → 2Fe(OH)3
Almond shell Activated Carbon (ASAC)
1710
1243 1069
1584 1165
Transmittance

around 1650 cm−1 (the bending ␯2 vibration mode). Weak bands at

3420
(a)
1300–1000 cm−1 indicate the presence of both hydroxyl and ether-
type C–O structures in ASAC [31]. MASAC exhibits a carbonyl peak
at 1710 cm−1 , which appears as a shoulder, and two additional, Almond shell Magnetic Activated Carbon (MASAC)
strong, overlapped bands at 640 and 567 cm−1 , compatible with
the presence of iron oxides in the sample [32]. In fact, bands situ- 640
567
ated close in frequency were reported before for iron minerals such
as maghemite (␥-Fe2 O3 ) [32]. 1219

MASAC will have TNP adsorption due to both carbon and iron
oxide species. Both species will contribute to TNP adsorption and to
4000 3600 3200 2800 2400 2000 1600 1200 800 400
this adsorption’s pH dependence. Thus, differentiating these effects
-1
would be very difficult. Wavenumbers/cm
Powder X-ray diffraction patterns are presented in Fig. 3b. Only
500
one broad peak centered at about 2 = 24◦ was observed in ASAC
confirming this carbon’s amorphous nature. When studying acti- Almon d shell magnetic ac tivated carbon(MASAC)
vated carbon used in double-layer supercapacitors, a peak located 400
Intensity (a.u.)

Table 2 300
Determination of point of zero charge (pHpzc ).

Points of reference ASAC MASAC 200

Almond shell ac tivated ca rbon (ASAC)
pH(initial) pH(final) pH(initial) pH(final) pH(initial) pH(final)

0 0 100 (b)
2 2 2 2* 2 4.62
4 4 4 2.54 4 6.76
6 6 6 2.55 6 6.78* 0
8 8 8 2.55 8 7.23 0 20 40 60 80 100
10 10 10 2.57 10 9.57
12 12 – – – –
2θ (in degree)

pHpzc for almond shell activated carbon and magnetized almond shell activated Fig. 3. IR spectra (a) and powder XRD (b) for almond shell activated carbon (ASAC)
carbon is 2.0, 6.8, respectively. and almond shell magnetic activated carbon (MASAC).
1116 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125

Fig. 4. SEM of almond shell carbon (ASAC) (A) ×500, (B) ×900, (C) ×1.50k, (D) ×5.0k and almond shell magnetic carbon (MASAC) (E) ×5000, (F) ×2.0k, (G) ×30.0k, (H)
×100.0k.
 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125
and also during subsequent oven-drying of the resulting hetero-
geneous mixture by the action of air:
2Fe(OH)2 + 12 O2 → Fe2 O3 + 2H2 O
These iron oxides may form via the following tentative reac-
tion paths. As pH rises during the preparation of MASAC, Fe(OH)3
must precipitate first and then Fe(OH)2 since the solubility product
constant is ≈10−36 for Fe(OH)3 versus 10−14 to 10−17 for Fe(OH)2 .
Ferrous hydroxide may be oxidized by dissolved oxygen:
2Fe(OH)2 + 12 O2 + H2 O → 2Fe(OH)3
And also during oven-drying by air:
2Fe(OH)2 + 12 O2 → Fe2 O3 + 2H2 O
Accordingly, in MASAC Fe3+ will be present as Fe(OH)3 , also
formulated as Fe2 O3 .3H2 O or as FeOOH.H2 O. Water is lost from
Fe(OH)3 only by heating above 40 ◦ C, in the interval between 40
and 170 ◦ C, 170 and 350 ◦ C and 490 and 690 ◦ C [34].
Surface morphology contributes to adsorbent–adsorbate inter-
actions. ASAC and MASAC SEM images (Fig. 4) show porous surfaces
with disorganized structural patterns that still contain some origi-
nal almond shell morphology. The macropore size distribution has
discrete groups of pore sizes rather than a continuum.
Micrographs 4(A–D) show a porous texture with small canals
for ASAC. Micrographs 4(E–H) show the morphological changes
due to iron oxide impregnation inside pores of the carbon matrix.
After iron impregnation, a spongy porous texture is observed
[Fig. 4(E–H)], suggesting the formation of well-dispersed iron
oxide particles covering ASAC. Most carbon matrix pores appear
to contain iron oxide. The surface area reduction of MASAC (from
733 m2 g−1 to 527 m2 g−1 ) after iron impregnation could be due to
some amount of pore blockage by dispersed iron-oxide particles.
Furthermore, the contributions to the surface area will be differ-
ent for the carbon and the iron oxide components of MASAC and
the MASAC has a lower carbon weight fraction. The surface area
reduction in magnetic activated carbons was also reported earlier
[17].
The SEM/EDX analysis spectra of ASAC and MASAC [Fig. 5(a and
b)] summarizes their elemental composition. ASAC contains car-
bon, oxygen, aluminum. No iron was observed. Fig. 5b confirms
the presence of iron on the MASAC surfaces. The aluminum peak,
present in ASAC, disappeared in the MASAC. The aqueous NaOH
step employed during the preparation of the MASAC likely removed
aluminum oxide from the carbon.
Transmission electron micrographs (Fig. 6) of thin carbon sec-
tions show lighter shading in the pore regions because the electron
beam passes through them, whereas pore walls look black because
of electron scattering. Quantitative analysis via image processing
gives pore size distributions and pore walls’ smoothness (fractal
dimension). Recording of higher resolution images of the carbon
might be quite difficult owing to its highly disordered structure
[35]. ASAC and MASAC shapes and sizes are shown in Fig. 6(A–D)
and (E–F), respectively. The primary carbon particles are inter-
connected with each other to form networks. MASAC primary
particles exhibited a rounded cubic shape with an almost homoge-
nous in size distribution. The iron-oxide nanoparticles were well
dispersed.
The suitability of ferromagnetic materials for application
depends on characteristics shown by their hysteresis loops [36],
obtained from plots of magnetization (M) against the field
strengths. Magnetization properties were investigated at 5 K and
300 K by vibrating sample magnetometery (VSM) which quantified
the magnetic behavior. The saturation magnetization of mag-
netic activated carbon was 4.47 and 6.54 emu/g at 300 and 5 K,
respectively, for MASAC (Fig. 7). The carbon sample without iron
1117
Table 3
Proximate analysis of ASAC and MASAC.
Samples C (wt.%) H (wt.%) N (wt.%) S (wt.%) Asha (wt.%)
ASAC 58.78 3.26 1.25 0.07 0.50
MASAC 50.48 2.63 1.24 0.03 16.93b
a
Ash content was determined by incinerating (air-treating) about 1 g of the sam-
ple at 650 ◦ C for 12 h in an electrical furnace.
b
It was observed that components of MASAC reacted with the ceramic container.
impregnation does not show any magnetic properties at 5 K and
300 K (Fig. 7) and the data overlapped each other. The attraction of
MASAC to a magnet when submerged in water is also illustrated in
Fig. 7.
The proximate analysis of ASAC and MASAC are given in Table 3.
Higher carbon content was estimated in ASAC (59 wt.%) versus
MASAC (51 wt.%). Major elements were detected in ASAC (Al, Mg,
Si, P, K, Ca, Fe) and MASAC (Na, P, Ca, Fe) by EDXRF analysis. A
high iron content was also determined in MASAC versus in ASAC
by EDXRF analysis. A 25–47 wt.% Fe2 O3 loading on activated carbon
was reported previously [17]. The aqueous NaOH step employed
during the preparation of the MASAC led to the presence of sodium
in MASAC.
4.2. Sorption studies
Adsorption of TNP on activated carbons varies significantly with
solution pH (Fig. 8). The pHPZC of ASAC was 2.0 versus 6.8 for the
MASAC (Table 2). The pKa of TNP is 0.38. At pH 2.0 (at the given
TNP concentration) the surfaces of both ASAC and MASAC are pos-
itively charged. At pH 2.0 the trinitrophenoxide concentration is
very low. Attractive electrostatic interactions favor the affinity of
the carbon surfaces for trinitrophenoxide anions. The decrease in
TNP adsorption at higher pH values could be related to the lower
net positive charge at the surfaces. As pH increases, the surface of
MASAC becomes less positive and ASAC becomes negative while
trinitrophenoxide concentration increases.
Several mechanisms have been proposed to contribute to
phenol adsorption [29]. These include (i) ␲–␲ dispersion inter-
actions between activated carbon basal planes and the phenolic
ring, (ii) electrostatic attraction–repulsion interactions, (iii) hydro-
gen bonding between adsorbate and carbon surface functions,
(iv) donor–acceptor complex formation and (v) oligomerization
of phenols on activated carbon surfaces in the presence of dis-
solved oxygen. Various combinations of i–iv type interactions may
occur with TNP. TNP is electron deficient (strong electron accep-
tor) and will resist oxidation by dissolved oxygen. This rules out
(v). The dominant interactions of TNP will be donor–acceptor
complexation with aromatic rings of the carbon basal planes, a
variety of hydrogen bonding interactions between –COOH and
–OH surface functions with the nitro groups’ oxygens (H-bond
acceptors) and the acidic –OH (H-bond donar) of TNP and also
dipole–dipole attractions (Fig. 9). TNP (a strong acid) will proto-
nate basic sites and the TNP anion will be attracted electrostatically
to the positive protonated sites (Fig. 9). All of these factors, plus
the presence of iron oxide-type surfaces in addition to activated
carbon surfaces in MASAC, make any interpretation of relative
pH dependence of TNP adsorption on ASAC versus MASAC very
problematic.
4.3. Kinetic studies and modeling
The TNP uptake rate increases with increased adsorbent
quantities between 1.0 and 2.0 g/L (see Figure S2 in Supplemen-
tary Material) for both carbons ASAC and MASAC. Furthermore
addition of 1.0 g/L adsorbent did not increase TNP uptake signif-
1118 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125

Fig. 5. EDX spectra for (a) nonmagnetic carbon (ASAC) and (b) magnetic activated carbon (MASAC).

icantly. Monolayer adsorption capacities were determined with an 4.3.1. Pseudo first order kinetic model
optimum dose of 3.0 g/L for all subsequent studies. TNP uptake on The simple pseudo-first order equation suggested by Lagergen
carbons is quite rapid. Typically, 50–65% of the ultimate adsorp- and further cited by Ho et al. [37] is
tion occurs within 4 h followed by a continuing slow approach to q − q 
equilibrium. Saturation is reached in 48 h. e t
ln = −k1 t (4)
TNP removal kinetics were also studied at 25, 35, and 45 ◦ C (see qe
Figure S3 in Supplementary Material). TNP adsorption on ASAC or
decreased upon raising the temperature, indicating the process is
exothermic, but TNP removal on MASAC increased as temperature qt = qe (1 − e−k1 t ) (Non linear form) (5)
went up, indicating an endothermic adsorption process. However,
these effects were small in the both cases. Endothermic adsorption where k1 (min−1 ) is the first order adsorption rate constant, qe , is
on MASAC might result from an activated diffusion contribution the amount adsorbed at equilibrium and qt , in the current work, is
due to pore restrictions by iron oxides species. Kinetic studies at the TNP amount adsorbed at time “t”. This Lagergren equation did
different TNP concentrations, viz., 45, 225 and 1125 mg/L were also not apply throughout the complete range of contact times [figure
studied (see Figure S4 in Supplementary Material). Higher TNP con- omitted for brevity]. The values of rate constant, k1 together with
centrations enhance the sorption rate. regression coefficients are provided in Table 4. These regression
First-order and second order kinetic models were tested to fit coefficients were not very good and qe values from plots did not
the batch experimental data. match experimental values well.
 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125 1119

Fig. 6. TEM of almond shell carbon [A, B, C, D] and magnetic almond shell activated carbon [E, F, G, H].
1120 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125

MASAC

70.12
68.96

15.53

148.85
72.06

72.06
qe calculated using
First order and pseudo-second-order rate constants with comparison of qe values calculated experimentally and by using first and second order rate equations at different temperatures and different TNP concentrations.

kinetic model
second order

71.43
71.19

15.32
69.93
72.04

162.50
(mg/g)

ASAC
MASAC

64.83

62.82

14.36
64.83
134.78
65.01
qe calculated using
first order kinetic
model (mg/g)

65.76
66.53
66.17

63.78
150.51
14.40
ASAC
MASAC

70.83
70.13
68.11

14.88
70.83
151.32
qe experimental

Fig. 7. Magnetic moment of magnetic and nonmagnetic activated carbons at 5 K and

300 K.
74.84

69.63

15.25
72.58
170.81
74.20
(mg/g)

ASAC

4.3.2. Pseudo-second-order kinetic model

A reaction involving pseudo-second-order kinetics requires that
the reaction rate is directly proportional to the number of active
sites on the adsorbent surface. The integrated rate expression for
0.9832
0.9798
0.9798

0.9832
0.9645
0.975

the pseudo-second-order reaction (Eq. (6)), discussed previously

R2

[37,38] was applied.

t 1 t
= + (6)
k2 (g mg−1 h−1 )

qt qe
rate constant,

k2 q2e
Second order

Here, t = time, k2 is pseudo-second-order adsorption rate constant

MASAC

0.049
0.004
0.007
0.008

0.004
0.003

(g mg−1 min−1 ), qe is the amount of TNP adsorbed at equilibrium

and qt the amount of TNP adsorbed at time “t”.
The product k2 q2e is the initial sorption rate represented as
0.9546
0.9629
0.9961

0.9545
0.8279
0.9530

Rate = k2 q2e (7)

R2

Rate constant (k2 ) and qe values were determined from the

plots at different TNP concentrations, temperatures and adsor-
bent concentrations (see Figure S2–S4 in Supplementary Material).
k2 (g mg−1 h−1 )
rate constant,
Second order

The correlation coefficients, R2 , and k2 values are presented in

Table 4. These correlation coefficients (R2 ) are superior to those for
0.016
0.026
0.032

0.110
0.016
0.008
ASAC

first order fits. The experimental and calculated qe values are pre-
sented in Table 4 for different TNP concentrations, temperatures
and adsorbent concentrations.
Excellent agreement exists between theoretical and experi-
0.9369
0.9262

0.9434
0.9369
0.9690

0.9079

mental qe values, suggesting that a pseudo-second-order model

R2

is operating. The pseudo-second-order model is based on the

First order rate

100 10
constant, k1

ASAC
MASAC
MASAC

Equilibrium pH with ASAC

0.312
0.396

0.471

0.333
0.209

0.210

8
(h−1 )

80 Equilibrium pH with MASAC

Percent TNP removal

Equilibrium pH

60 6
0.8847
0.9388
0.9766

0.8847
0.7438
0.9025
At different TNP concentrations(mg/L)
R2

40 4
At different temperatures (◦ C)
First order rate

20 2
constant, k1

0.848

0.988
0.832
0.872
ASAC
(h−1 )

0
1.32
1.40

0
0 2 4 6 8 10 12

Initial pH
Values
Table 4

25
35
45

45
225
1125

Fig. 8. Effect of pH on TNP removal using ASAC and MASAC at an initial TNP con-
centration of ∼125 mg/L, adsorbent dose of 3 g/L, particle size of 50–100 B.S.S mesh
and at 25 ◦ C.
 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125 1121

NO2 NO2

O 2N O Hydrogen Bonding
O2N OH Any basic- sites
O
H
eg. NO2
NO2
O

OH

O 2N NO2
Acid H+ transfer
O OH O
C OH
N O2N N O
O O

O H
H O O
C N
Carbon surface oxidizes
O O
H

COOH
O
OH
Actually each TNP could be H-bonded
several times since it has 3 NO2 groups
OH and a OH group.
H-bond acceptor
π- π donor-acceptor
complex
H-bond donor
NO2 H
O O O
O2N OH
N N
NO2
O O
Acceptor

H-bond acceptor
N

O O

Also,dipole-dipole permanent attractions and instantaneous dipole-induced dipole attraction can attract TNP to surfaces
permanent dipole/dipole attraction
Quinone function on carbon O
N

Fig. 9. Adsorption mechanism of TNP on activated carbon.

1122
80
(a) ASAC
Amount adsorbed (mg/g)
60
40
20
0 capacity. It should only be applied in the low to intermediate con-
0 5 10
Equilibrium concentration (mg/L)
80
(b) MASAC
Amount adsorbed (mg/g)
60
40
20
0 1 + bCe
0 5 10 15
Equilibrium Concentration (mg/L)
Fig. 10. Langmuir adsorption isotherms of TNP by (a) ASAC and (b) MASAC at differ-
ent temperatures [pH 2.0; adsorbent concentration of 3 g/L; particle size of 50–100
B.S.S mesh]. Solid lines represent the data fitted by the Langmuir isotherm model.
assumption that the rate-limiting step may be a chemical sorption
involving valance forces through sharing or exchange of electrons
between adsorbent and adsorbate.
Regression coefficient analysis showed that the pseudo-second-
order rate equation described the kinetic data well for ASAC (see
Figure S5 in Supplementary Material) and even better for MASAC
(see Figure S6 in Supplementary Material).
4.4. Sorption studies and modeling
Models should be capable of reliably predicting the adsorption
data for use when scaling up fixed-bed reactors to the indus-
trial scale. Various approaches for comparing adsorption data were
studied. Sorption equilibrium batch isotherms for TNP adsorption
at pH 2.0 and 25 ◦ C, 35 ◦ C, 45 ◦ C were conducted on ASAC and
MASAC and a commercial F-400 carbon as given in Fig. 10 (see
also Figures S7–S9 in Supplementary Material). These isotherms
are positive, regular, and concave to the concentration axis. TNP
adsorption on ASAC decreased upon raising the temperature, indi-
cating the process is exothermic. In contrast, TNP removal on
MASAC increased as temperature went up, indicating an endother-
mic adsorption process. Furthermore, adsorption also increases
with rise in temperature on F-400 activated carbon. Increases in
TNP adsorption with a rise in temperature were also reported on
D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125
ferrocyanides [39] and bagasse fly ash [40] while a decrease in
phenol adsorption was reported on bentonites [41].
The decrease or increase in TNP uptake with rising temperature
on ASAC and MASAC, respectively, is not very significant. There-
fore, these carbons can be used over a wide temperature range
with little or no increase/decrease in sorption efficiency. To deter-
mine the most appropriate model for the sorption equilibrium
isotherms, performances at different temperatures were evaluated
using the Freundlich, Langmuir and Sips isotherm models referred
earlier [38]. The F-400 sorption data were modeled using Langmuir
adsorption model to evaluate the monolayer adsorption capacity.
4.4.1. Freundlich isotherm
The Freundlich model does not indicate a finite sorbent uptake
15 20 25 centration range [38]. The nonlinear Freundlich equation is
1/n
qe = KF Ce (8)
where qe is the amount of solute adsorbed per unit weight of
adsorbent (mg/g), Ce is the equilibrium solute concentration in
the solution (mg/L), constant KF indicates the adsorbent’s relative
adsorption capacity (mg/g) and 1/n is the constant represent-
ing adsorption intensity. The nonlinear Freundlich adsorption
isotherms are given in Figure S7(a,b) in supplementary Material
while the parameters are reported in Table 5.
4.4.2. Langmuir isotherm
The Langmuir adsorption isotherm provides no mechanistic
information. Instead, it provides information on uptake capabili-
ties and reflects the usual equilibrium behavior [38]. The nonlinear
Langmuir equation is
Q 0 bCe
qe = (9)
20 25 30
where qe is the solute amount adsorbed per unit weight of adsor-
bent (mg/g), Ce is the solute equilibrium concentration in solution
mg/L, Q0 is the monolayer adsorption capacity (mg/g) and constant
b is related to the net enthalpy, H, of adsorption (b ∝ e−H/RT ). More
precisely, “b” is the reciprocal of concentration at which half satura-
tion is attained. These isotherms are given in Fig. 10(a and b) while
the parameters are reported in Table 5. TNP sorption data on F-400
were also modeled using the Langmuir equation (Figure S8 in sup-
plementary Material) for comparative evaluation. The Langmuir
parameters are reported in Table 5. The Langmuir model is used
to estimate maximum uptake values, where these could not be
obtained experimentally.
Langmuir isotherms can also be expressed in terms of a
dimensionless constant separation factor [RL = 1/(1 + bC0 ); RL > 1
unfavorable; RL = 1 linear; 0 < R < 1 favorable and RL = 0 irreversible],
originally defined by Hall et al. [42] and also discussed earlier [38].
RL was determined at different temperatures over a broad con-
centration range. The RL values were between 0 and 1 indicating
that TNP adsorption on ASAC, MASAC and F-400 is favorable (data
omitted for brevity).
4.4.3. Sips or Langmuir-Freundlich adsorption isotherm model
The Langmuir-Freundlich or Sip model is given in Eq. (10). At low
concentrations, this equation effectively reduces to a Freundlich
isotherm. At high sorbate concentrations, it predicts a monolayer
sorption capacity that is characteristic of Langmuir isotherm [38].
KLF CenLF
qe = (10)
1 + (aLF Ce )nLF
KLF , aLF and nLF are the Sips constants. The nonlinear Sip adsorption
isotherm parameters are reported in Table 5 from experimental
data and fits provided in Figure S9(a,b) in supplementary Material.
 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125 1123

Table 5
Adsorption isotherm parameters for 2,4,6-trinitrophenol removal from water by ASAC, MASAC and F-400 at different temperatures.

Isotherm parameters ASAC MASAC F-400

25 ◦ C 35 ◦ C 45 ◦ C 25 ◦ C 35 ◦ C 45 ◦ C 25 ◦ C 35 ◦ C 45 ◦ C

Freundlich
KF (mg/g) 30.56 30.91 26.24 23.69 22.62 20.83 85.11 116.92 96.22
1/n 0.36 0.30 0.30 0.35 0.41 0.46 0.32 0.24 0.25
R2 0.9665 0.9741 0.9854 0.9062 0.9050 0.9173 0.9223 0.8305 0.9513

Langmuir
Q0 (mg/g) 74.03 73.15 70.83 73.96 84.07 87.16 327.85 203.90 188.20
b 0.77 0.59 0.38 0.36 0.28 0.25 2.37 0.75 1.58
R2 0.9710 0.9442 0.9400 0.9831 0.9824 0.9585 0.9137 0.9507 0.9248
Q0 (mg/g) in presence of DNP 65.67 – – 61.96 – – – – –
Qmix /Q0 0.8871 – – 0.8378 – – – – –

Sips
KLF (L/g) 47.49 37.60 27.15 25.53 19.89 20.25 – – –
aLF (L/mg)a LF 0.496 0.226 0.036 0.364 0.271 0.261 – – –
nLF 0.679 0.420 0.321 1.137 1.337 1.205 – – –
R2 0.9823 0.9763 0.9855 0.9849 0.9924 0.9615 – – –

Table 6
TNP remediation of a contaminated ground watera using ASAC and MASAC (adsorbent dose 6 g/L; equilibrium time 48 h; pH 2.0; Temperature 25 ◦ C; volume of the water
sample taken 50 mL).

Parameters Values without any Values when pH adjustment was done Values after treatment Values after treatment
pH adjustment (Initial pH 7.94 which was adjusted to with ASAC with MASAC
pH-2)

pH 7.94 1.99 1.73 1.82

Cl− 67 340 249 99.96
Na+ (mg/L) 67 92 59.8 124.7
K+ (mg/L) 6.7 8.1 9.2 8.7
Ca2+ (mg/L) 15.6 17.7 13.8 22.0
TNP (mg/L) 23 23 0.20 5.38
a
Ground water sample was collected from Lutfullapur, Navada, Loni, District Ghaziabad, India and an additional ∼23.0 mg/L of TNP was spiked into this water sample.

The Sips or Langmuir-Freundlich adsorption isotherm model
best fits the experimental data with high regression (R2 ) coeffi-
cients. This means that adsorption of TNP on ASAC and MASAC
is occurring by a combined Freundlich-Langmuir (or Sips) model.
Adsorption is diffusion controlled at low TNP concentrations while
monomolecular adsorption with a saturation value takes place at
high TNP concentrations. Similar observations were reported by
Muntean and coworkers for the adsorption of dye in wood fibers
[43].
4.5. Multicomponent TNP adsorption
and the rest is desorbed in three increments of 10 mL methanol
followed by five increments of 10 mL hot water.
4.7. Applications of developed carbons to treat surface water
samples
Real water/wastewater systems contain a complex mixture of
ions. Competition for sorption sites by these ions affects overall
sorption efficacy. The efficacy of ASAC and MASAC in real water sys-
tems was investigated by collecting ground water from Lutfullapur,

Real water/wastewater systems often contain multiple phenols. 80

Competition for adsorption sites by other phenols affects overall
70
sorption efficacy. Therefore, tests were conducted in the presence
of dinitrophenol (DNP) over the concentration range from 1 × 10−4
Amount adsorbed (mg/g)

60
to 1.0 × 10−3 M. A stock solution was made by dissolving DNP and
TNP in 1:1 molar ratio and dilutions gave concentrations from 50
1 × 10−4 to 1.0 × 10−3 M. Langmuir adsorption isotherms both in
the absence and presence of DNP at 25 ◦ C are given in Fig. 11. The 40
effect of DNP on TNP adsorption was represented by the ratio of the
sorption capacity for TNP in presence of DNP, Qmix , to the sorption 30 ASAC- only TNP
ASAC-TNP+DNP
capacity for the TNP when present alone, Q0 . When Qmix /Q0 > 1, MASAC-only TNP
20
TNP sorption is promoted by DNP. When Qmix /Q0 = 1, there is no MASAC-TNP+DNP
net interaction. When Qmix /Q0 < 1, TNP sorption is suppressed by 10
DNP [44]. The Qmix /Q0 values are less than 1 (Table 5). This con-
firms that the TNP adsorption is suppressed by competition from 0
DNP. 0 10 20 30 40 50
Equilibrim concentration (mg/L)
4.6. Desorption
Fig. 11. Adsorption isotherms of TNP on ASAC and MASAC in absence and presence
of DNP. The sample with interfering DNP was in 1:1 molar ratio. Solid lines represent
Almost complete (∼97%) desorption of TNP could be achieved. the data fitted by the Langmuir isotherm mode [pH 2.0; adsorbent concentration of
The first aliquot of 10 mL methanol elutes about 45–50% of TNP 3 g/L; particle size of 50–100 B.S.S mesh].
1124 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125

Table 7
Comparison of Langmuir adsorption capacities of ASAC and MASAC for 2,4,6-trinitrophenol removal from water.

Adsorbents Type of water pH Temp. Concentration Surface area Adsorption References

(◦ C) range (mg/L) (m2 g−1 ) capacity (mg/g)

ASAC Aqueous solution 2.0 25 ∼23–230 733 74.03 This study

2.0 35 71.15
2.0 45 70.83

MASAC 2.0 25 527 73.96

2.0 35 84.07
2.0 45 87.15

F-400 (Calgon carbon) 2.0 25 327.85

2.0 35 203.90
2.0 45 188.20

Copper ferrocyanides Aqueous solution 2.0 25 ∼23–230 25.43 29.41 [39]

Zinc ferrocyanides 2.0 26.28 43.47
Molybdenum ferrocyanides 2.0 25.02 52.63
Chromium ferrocyanides 2.0 26.59 26.31
Bagasse fly ash Aqueous solution 4.0 30 ∼23–230 433 17.0 [40]
Hydrotalcite (HT) Aqueous solution 2.0 – – – 199 [45]
Calcined Hydrotalcite (HT-500) 2.0 – – – 520
Activated carbon (AC)-30 Aqueous solution – 30 1–300 650 54 [46]
Activated carbon (AC)-40 900 190
Activated carbon (AC)-50 1000 210
Activated carbon (AC)-60 1150 258
Activated carbon (AC)-70 1300 269

Amberlite IRA-67 Aqueous solution – 25 4000–14000 – 31.84 [47]

35 38.61
45 47.16

Uncalcined Mesoporous silicate, UC MCM-41 Aqueous solution 3.1 10–300 ␮g/mL 217 327.17 [48]
Activated carbon prepared from date fruit pits Aqueous solution – 25 – – 26.74 [49]

Navada, Loni, District Gaziabad, India. The physicochemical char-
acteristics of this water sample are given in Table 6. An additional
∼23.0 mg/L of TNP was spiked into this water sample. ASAC and
MASAC were then utilized to remediate TNP from this ground water
under optimum conditions previously obtained from batch sorp-
tion studies. The pH was adjusted to 2.0 before applying the carbon
sorbents. A predetermined amount of carbon (5 g/L) was added to
each sample. After equilibrating 48 h, the samples were filtered and
the TNP concentrations were measured by UV–vis spectroscopy.
It is clear from Table 6 that the final TNP concentrations were
reduced. The presence of solids and interfering ions caused only
a slight reduction in sorption efficiency. Thus, these carbons can
be applied to TNP remediation from contaminated surface/ground
water.
5. Conclusions
Almond shell non-magnetic (ASAC) and magnetic activated car-
bons (MASAC) were prepared, characterized and used successfully
for the trinitrophenol remediation from water. Almond shells are
hardwoods (high density) versus other woods. Although, porous
carbons can be made from most lignocellulosic materials, those
with hardness and high density find commercial application as
activated carbon.
Adsorption of TNP on activated carbons varies significantly with
solution pH. MASAC and ASAC performed better at low pH. Adsorp-
tion of TNP decreases on ASAC with an increase in temperature,
indicating the process is exothermic. However, TNP adsorption
increases with increase in temperature on MASAC indicating an
endothermic process. These temperature dependences are small,
so non-magnetic and magnetic almond shell activated carbons can
be used over a wide temperature range with little or no decrease
in the sorption efficiency.
The Langmuir monolayer adsorption capacity was higher for
ASAC. TNP sorption capacities of ASAC and MASAC appear higher
or comparable per unit surface area to other activated carbons but
they have lower surface areas than AC-50, 60 and 70 (Table 7). Com-
parisons with other materials reported in the literature for TNP
remediation are shown in Table 7.
The TNP adsorption rate fits a pseudo-second-order kinetic
model where the rate-limiting step is assumed to be chemical
sorption between the adsorbate and adsorbent. The MASAC parti-
cles can readily be recovered from the solution using a permanent
magnet (Fig. 7). This carbon is easily redispersed in the solution
upon removing the magnetic field. This work demonstrated that
an effective low cost MASAC exhibiting a reasonably high surface
area was synthesized which can adsorb TNP. MASAC was easily
manipulated by a low strength external magnetic field, permitting
easy recovery from environmental aqueous applications. MASAC
could be applied to the remediation of contaminated water high
in suspended solids, oil and grease, where impurities may cause
carbon fouling and require frequent separation/regeneration treat-
ments. Simple separation by a magnetic field would allow isolation
and washing followed by redispersion. Recycling of the MASAC
was demonstrated. The use and effectiveness of several desorbing
agents and their application in TNP recovery will be published in
future.
Variation of the pore surface areas and volumes should be pos-
sible in the future by tailoring the stoichiometry of Fe3+ and Fe2+
used, changing the chemical agents used in pyrolysis and adjusting
the pyrolysis time–temperature profile.
Acknowledgement
Financial support of this work provided by University Grant
Commission New Delhi, India, (PAC/SES/DM/UGC/0210113-491) is
thankfully acknowledged.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.cej.2011.06.054.
 D. Mohan et al. / Chemical Engineering Journal 172 (2011) 1111–1125
References
[1] S. Beszedits, M.D. Silbert, Treatment of Phenolic Wastewaters, B & L Information
Services, Toronto, 1990.
[2] J. Boileau, C. Fauquigonon, C. Napoly, Explosives, in: W. Gerhartz, Y.S.
Yamamoto, L. Kaudy, J.F. Rounsaville, G. Schulz (Eds.), Ullmann’s Encyclopedia
of Industrial Chemistry, VCH, Weinheim, 1987, pp. 145–172.
[3] T.A. Lewis, D.A. Newcombe, R.L. Crawford, Bioremediation of soils contam-
inated with explosives, Journal of Environmental Management 70 (2004)
291–307.
[4] USEPA, Ambient Water Quality Criteria for Nitrophenol, U.S. Environmental
Protection Agency, Washington, DC, 1980.
[5] S. Morales-Torresa, A.M.T. Silva, A.F. Pérez-Cadenas, J.L. Faria, F.J. Maldonado-
Hódar, J.L. Figueiredo, F. Carrasco-Marín, Wet air oxidation of trinitrophenol
with activated carbon catalysts: effect of textural properties on the mechanism
of degradation, Applied Catalysis B: Environmental 100 (2010) 310–317.
[6] BIS, Tolerance Limit for Industrial Effluents Discharged into Inland Surface
Waters: Coke Oven, IS 2490 (Part 1), Bureau of Indian Standards, New Delhi,
1974.
[7] R.M. Aghav, S. Kumar, S.N. Mukherjee, Artificial neural network modeling in
competitive adsorption of phenol and resorcinol from water environment using
some carbonaceous adsorbents, Journal of Hazardous Material 188 (1–3) (2011)
67–77.
[8] USEPA, Technical Support Document for Water Quality Based Toxics Control,
EPA/440/485032, United States Environmental Protection Agency, Washing-
ton, DC, USA, 1985.
[9] A. Dabrowski, P. Podkościelny, Z. Hubicki, M. Barczak, Adsorption of pheno-
lic compounds by activated carbon—a critical review, Chemosphere 58 (2005)
1049–1070.
[10] M. Ahmedna, W.E. Marshall, A.A. Husseiny, R.M. Rao, I. Goktepe, The use of
nutshell carbons in drinking water filters for removal of trace metals, Water
Research 38 (2004) 1062–1068.
[11] G. McKay, Use of Adsorbents for the Removal of Pollutants from Wastewater,
1 ed., CRC-Press, 1995.
[12] D.A.J. Wase, J. Wase, Biosorbents for Metal Ions, 1 ed., Informa Healthcare, 1997.
[13] S.J. Allen, L. Whitten, G. Mckay, The production and characterization of acti-
vated carbon—a review, Development Chemical Engineering Mineral Process
6 (1998) 231–261.
[14] J. Hristov, L. Fachikov, An overview of separation by magnetically stabilized
beds: state-of-the-art and potential applications, China Particuology 5 (2007)
11–18.
[15] A.B. Fuertes, P. Tartaj, A facile route for the preparation of superparamagnetic
porous carbons, Chemistry of Materials 18 (2006) 1675–1679.
[16] A. Nakahira, H. Nagata, M. Takimura, K. Fukunishi, Synthesis and evaluation of
magnetic active charcoals for removal of environmental endocrine disrupter
and heavy metal ion, Journal of Applied Physics 101 (2007), 09J114–109J114-
113.
[17] L.C.A. Oliveira, R.V.R.A. Rios, J.D. Fabris, V. Garg, K. Sapag, R.M. Lago, Activated
carbon/iron oxide magnetic composites for the adsorption of contaminants in
water, Carbon 40 (2002) 2177–2183.
[18] M. Schwickardi, S. Olejnik, E.-L. Salabas, W. Schmidt, F. Schüth, Scalable
synthesis of activated carbon with superparamagnetic properties, Chemical
Communication (2006) 3987–3989.
[19] G.S. Zhang, J.H. Qu, H.J. Liu, A.T. Cooper, R.C. Wu, CuFe2 O4 /activated carbon
composite: a novel magnetic adsorbent for the removal of acid orange II and
catalytic regeneration, Chemosphere 68 (2007) 1058–1066.
[20] Y. Ao, J. Xu, D. Fu, X. Shen, C. Yuan, A novel magnetically separable compos-
ite photocatalyst: titania-coated magnetic activated carbon, Separation and
Purification Technology 61 (2008) 436–441.
[21] N. Atsushi, S. Nishida, K. Fukunishi, Synthesis of magnetic activated carbons for
removal of environmental endocrine disrupter using magnetic vector, Journal
of the Ceramic Society of Japan 114 (2006) 135–137.
[22] D. Mohan, C.U. Pittman, M. Bricka, F. Smith, B. Yancey, J. Mohammad, P.H. Steele,
M.F. Alexandre-Franco, V. Gómez-Serrano, H. Gong, Sorption of arsenic, cad-
mium, and lead by chars produced from fast pyrolysis of wood and bark during
bio-oil production, Journal of Colloid and Interface Science 310 (2007) 57–73.
[23] S. Brunauer, P.H. Emmett, E. Teller, Adsorption of gases in multimolecular lay-
ers, Journal of American Chemical Society 60 (1938) 309–319.
1125
[24] M.M. Dubinin, Physical adsorption of gases and vapors in micropore, in: J.F.
Danielli, M.D. Rosenberg, D.A. Cadenhead (Eds.), Progress in Surface and Mem-
brane Science, Academic Press, Newyork and London, 1975, pp. 1–70.
[25] M.M. Dubinin, G.M. Plavnik, Microporous structures of carbonaceous adsor-
bents, Carbon 6 (1968) 183–192.
[26] S. Not, J.A. Schwarz, Estimation of the point of zero charge of simple
oxides by mass titration, Journal of Colloid and Interface Science 130 (1989)
157–164.
[27] A. Nakahira, S. Nishida, K. Fukunishi, Synthesis of magnetic activated carbons
for removal of environmental endocrine disrupter using magnetic vector, Jour-
nal of the Ceramic Society of Japan 114 (2006) 135–137.
[28] S. Brunauer, L.S. Deming, W.E. Deming, E. Teller, On a theory of the van der
waals adsorption of gases, Journal of American Chemical Society 62 (1940)
1723–1732.
[29] T.J. Bandosz, Activated Carbon Surfaces in Environmental Remediation,
Elsevier, The Netherlands, 2006.
[30] P.C. Painter, M. Sobkowiak, J. Youtcheff, Ft-ir study of hydrogen-bonding in coal,
Fuel 66 (1987) 973–978.
[31] E. Pretsch, T. Clerc, J. Seibl, W. Simons, Tablas para la Elucidación Estructural
de Compuestos Orgánicos por Métodos Espectroscópicos, Alhambra, Madrid,
1980.
[32] H.W.v.d. Marel, H. Beutelspacher, Atlas of Infrared Spectroscopy of Clay Min-
erals and their Admixtures, Elsevier, Amsterdam, 1976.
[33] D. Qu, Studies of the activated carbons used in double-layer supercapacitors,
Journal of Power Sources 109 (2002) 403–411.
[34] C.J. Geankoplis, Transport Processes and Separation. Process Principles, 4th ed.,
Prentice Hall, NJ, 2003.
[35] P.J.F. Harris, Z. Liu, K. Suenaga, Imaging the atomic structure of activated carbon,
Journal of Physics: Condensed Matter 20 (2008) 362201–362206.
[36] D. Jiles, Introduction to Magnetism and Magnetic Materials, Chapman and Hall,
1991.
[37] Y.S. Ho, G. McKay, Pseudo-second order model for sorption processes, Process
Biochemistry 34 (1999) 451.
[38] D. Mohan, S. Rajput, V.K. Singh, P.H. Steele, C.U. Pittman Jr., Modeling and eval-
uation of chromium remediation from water using low cost bio-char, a green
adsorbent, Journal of Hazardous Materials 188 (1–3) (2011) 319–333.
[39] B.B. Tewari, D. Mohan, Kamaluddin, Interaction of 2,4-dinitrophenol and 2,4,6-
trinitrophenol with copper, zinc, molybdenum and chromium ferrocyanides,
Colloids and Surfaces A: Physicochemical and Engineering Aspects 131 (1998)
89–93.
[40] S.K. Srivastava, V.K. Gupta, N. Johri, D. Mohan, Removal of 2,4,6-trinitrophenol
using bagasse fly ash: a sugar industry waste material, Indian Journal of Chem-
ical Technology 2 (1995) 333–336.
[41] S. Al-Asheh, F. Banat, L. Abu-Aitah, Adsorption of phenol using different types of
activated bentonites, Separation and Purification Technology 33 (2003) 1–10.
[42] K.R. Hall, L.C. Eagletow, A. Acrivers, T. Vermenlem, Pore and solid kinetics in
fixed-bed adsorption under constand pattern condition, Industrial Engineering
and Chemistry Fundamentals 5 (1966) 212–218.
[43] S.G. Muntean, G. Simu, P. Sfarloaga, C. Bologa, Study of a trisazo direct dye
adsorption on wood fibre using a comparison of different adsorption isotherms,
Revista de Chimie 61 (2010).
[44] T.C. Tan, C.K. Chia, C.K. Teo, Uptake of metal ions by chemically treated human
hair, Water Research 19 (1985) 157–162.
[45] M.C. Hermosín, I. Pavlovic, M.A. Ulibarri, J. Cornejo, Hydrotalcite as sorbent for
trinitrophenol: sorption capacity and mechanism, Water Research 30 (1996)
171–177.
[46] P. Aggarwal, K. Misra, S.K. Kapoor, A.K. Bhalla, R.C. Bansal, Effect of surface
oxygen complexes of activated carbon on the adsorption of 2,4,6,-
trinitrophenol, Defence Science Journal 48 (1998) 219–222.
[47] H. Uslu, G.k. Demir, Adsorption of picric acid from aqueous solution by the
weakly basic adsorbent amberlite IRA-67, Journal of Chemical and Engineering
Data 55 (2010) 3290–3296.
[48] H. Sepehrian, J. Fasihi, M.K. Mahani, Adsorption behavior studies of picric acid
on mesoporous MCM-41, Industrial Engineering and Chemistry Research 48
(2009) 6772–6775.
[49] M.A. Abdulkarim, N.A. Darwish, Y.M. Magdy, A. Dwaidar, Adsorption of phenolic
compounds and methylene blue onto activated carbon prepared from date fruit
pits, Engineering in Life Science 25 (2002) 161–165.