Immobilization of Bacillus sp. in mesoporous activated carbon for degradation of sulphonated phenolic compound in wastewater

Materials Science and Engineering C xxx (2012) xxx–xxx

MSC-03647; No of Pages 11

Contents lists available at SciVerse ScienceDirect

Materials Science and Engineering C

j ourna l homepage: www.e lsev ie r .com/ locate /msec

Immobilization of Bacillus sp. in mesoporous activated carbon for degradation ofsulphonated phenolic compound in wastewater

G. Sekaran a,⁎, S. Karthikeyan a, V.K. Gupta b,c, R. Boopathy a, P. Maharaja a

a Environmental Technology Division, Council of Scientific & Industrial Research (CSIR), Central Leather Research Institute (CLRI), Adyar, Chennai-600 020, Indiab Department of Chemistry, Indian Institute of Technology Roorkee, Roorkee-247 667, Indiac Department of Chemistry, King Fahd University of Petroleum & Minerals, Dhahran 31261, Saudi Arabia

⁎ Corresponding author. Tel.: +91 44 24452491; fax:E-mail addresses: [email protected], vinod

(G. Sekaran).

0928-4931/$ – see front matter © 2012 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.msec.2012.10.026

Please cite this article as: G. Sekaran, et al., M

a b s t r a c t

Article history:Received 28 June 2012Received in revised form 26 September 2012Accepted 29 October 2012Available online xxxx

Keywords:ImmobilizationBacillus sp.Activated carbonXenobiotic compoundsSulphonated phenolic compound (SP)

Xenobiotic compounds are used in considerable quantities in leather industries besides natural organic andinorganic compounds. These compounds resist biological degradation and thus they remain in the treatedwastewater in the unaltered molecular configurations. Immobilization of organisms in carrier matrices pro-tects them from shock load application and from the toxicity of chemicals in bulk liquid phase. Mesoporousactivated carbon (MAC) has been considered in the present study as the carrier matrix for the immobilizationof Bacillus sp. isolated from Effluent Treatment Plant (ETP) employed for the treatment of wastewatercontaining sulphonated phenolic (SP) compounds. Temperature, pH, concentration, particle size and massof MACwere observed to influence the immobilization behavior of Bacillus sp. The percentage immobilizationof Bacillus sp. was the maximum at pH 7.0, temperature 20 °C and at particle size 300 μm. Enthalpy, free en-ergy and entropy of immobilization were −46.9 kJ mol−1, −1.19 kJ mol−1 and −161.36 J K−1 mol−1 re-spectively at pH 7.0, temperature 20 °C and particle size 300 μm. Higher values of ΔH0 indicate the firmbonding of the Bacillus sp. in MAC. Degradation of aqueous sulphonated phenolic compound by Bacillus sp.immobilized in MAC followed pseudo first order rate kinetics with rate constant 1.12×10−2 min−1.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

Leather manufacturing industry uses a multitude of xenobiotic com-pounds of varied functional characteristics in considerable quantities toimpart the basic properties to leather [1–3]. Amongst the syntheticchemicals, sulphonated mononuclear, dinuclear and trinuclear aromaticcompounds are used for the conversion of putrescible collagenfibers intonon-putrescible leather. Sulphonated phenolic (SP) compounds lack bio-degradability due to the molecular steric factor, resonance effect andhigh solubility in water [4]. Thus, sulphonated phenolic compounds inwastewater escape anaerobic and aerobic biological treatment unitswith unaltered molecular configurations contributing towards substan-tial amount of residual chemical oxygendemand in the treatedwastewa-ter [5–8]. Reasons for failure of anaerobic/aerobic biological treatmentunits to remove the SP could be due to inadequate microbiological pop-ulation density or absence of susceptible biodiversities for the degrada-tion of condensed phenolic molecular moieties in wastewater. Theconventional suspended growth biological treatment systems are inef-fective for the treatment of wastewater containing xenobiotic com-pounds due to toxicity of them. The poor efficiency of organisms in the

+91 44 [email protected]

rights reserved.

ater. Sci. Eng., C (2012), http

suspended state for metabolism of xenobiotic compounds was mainlyattributed to irreversible binding of the compounds on their extra cellu-lar biopolymers rendering active/passive transport processes becomedifficult [9]. Additional reasons for incomplete catabolism of xenobioticcompounds might be due to specific transport and induction mecha-nisms or toxic effects caused by the substrate itself or its metabolites[10]. Immobilization of organisms in suitable carriermatrices for biodeg-radation of xenobiotic compounds was suggested by many researchers[11,12] as the microorganisms were protected from shock load applica-tion and toxic effect of xenobiotic compounds [13–20]. The organic com-pounds in wastewater first adsorb onto the surface of the carrier matrixin which the organism was immobilized and then gradually penetratethrough the pores of it. Hence, organisms have sufficient contact timeto release extra cellular enzymes for pre hydrolysis of organic com-pounds and to transport the fragmented organic compounds throughcellular membrane for oxidation. Thus, the degradation of toxic com-pounds by bacterial cultures immobilized in activated carbon has beenreported as a combination of physical adsorption and biological degrada-tion [21–24]. Carrier matrices such as sodium alginate, calcium alginate,polyurethane, polyacrylamide gel, granulated clay, activated carbon andtransition metal activated pumice stone have been reported in the liter-ature for immobilization of organisms [25–27].

The present investigation is focused on three aspects i) isolationof microorganisms from the ETP treating SP containing wastewater

://dx.doi.org/10.1016/j.msec.2012.10.026

2 G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

ii) immobilization of the isolated bacterial culture in mesoporousactivated carbon and iii) degradation of SP in minimal medium byimmobilized organism.

2. Materials and methods

Rice husk, the solid waste in agricultural industry was washedwith water for several times and dried at 110 °C in hot air oven for6 h. The dried samples were then sieved to 600-micron size and thesame was used as the rawmaterial for the preparation of mesoporousactivated carbon.

2.1. Preparation of mesoporous activated carbons

Mesoporous activated carbon (MAC) was prepared through twosteps: pre-carbonization at 400 °C and chemical activation. In theprecarbonization process the rice husk was heated at 400 °C at the rateof 10 °C/min for about 4 h under N2 atmosphere and cooled down toroom temperature at the same rate. The precarbonized carbon wassubjected to chemical activation using phosphoric acid. In chemicalactivation process 50 g of the precarbonized material was agitatedwith 250 g of aqueous solution containing 85% H3PO4 by weight. Theratio of chemical activating agent to precarbonized carbon was fixed at4.2 (w/w). The chemical reagent and precarbonized carbon werehomogeneously mixed at 85 °C for 4 h in a mixer. After mixing, thechemical impregnated precarbonized carbon slurry was dried undervacuum at 110 °C for 24 h. The dried precarbonized samples werethen activated in a vertical cylindrical furnace under N2 atmosphere ata flow rate of 100 mL/min. The heat treatment was optimized byheating at three different temperatures of 700, 800 and 900 °C, at aheating rate of 5 °C/min in a temperature programmable furnace andmaintained at constant temperature for 1 h and it was followed byslow cooling. After cooling the activated carbon was washed severaltimes with hot water until the pH was neutral and finally washedwith cold water to remove the excess phosphorous compounds. Thewashed samples were dried at 110 °C in hot air oven to get thefinal activated carbon samples. The samples heated at activationtemperatures 700, 800 and 900 °C were labeled as MAC700, MAC800

and MAC900. The dried samples were stored over fused CaCl2 in adesiccator until for further use.

2.2. Characteristics of mesoporous activated carbon

2.2.1. Chemical characteristics of activated carbonThe chemical characteristics of the MAC are presented in Table 1.

They are carbon, 48.45%; Hydrogen, 0.7%; Nitrogen, 0.1%; Ash content,42.65%; bulk density, 0.402 g/cm3 and particle size, 600 μm.

Table 1Characteristics of rice bran based MAC.

S. no Parameters Values

1 Carbon (%) 48.452 Hydrogen (%) 0.73 Nitrogen (%) 0.14 Ash content (%) 42.615 Bulk density (g/mL) 0.4056 Moisture content (%) 3.87 Ash content (% by mass) 408 Matter soluble in water (%) 0.4289 Matter soluble in acid (%) 3.90810 pH 6.6611 Decolorizing power (mg/g) 5212 Phenol number (mg/g) 3.1813 Ion exchange capacity (meq/g) 0.01514 Surface area (BET), m2/g 420

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

2.2.2. N2 adsorption–desorption isothermsThe N2 adsorption–desorption isotherms of the MAC samples were

determined using an automatic adsorption instrument (QuantachromeCorp. Nova-1000 gas sorption analyzer) for the determination of surfacearea and the total pore volume. Prior to placing in the adsorption instru-ment the carbon sampleswere degassed at 150 °C overnight. The nitro-gen adsorption–desorption data were recorded at a liquid nitrogentemperature of 77 K. The surface area of the samples was calculatedusing the BET equation, the most widely used model for determiningthe specific surface area. In addition, the t-plot method was applied tocalculatemicropore volume, external surface area andmesoporous sur-face area. The total pore volume was estimated as the liquid volume ofadsorbate at a relative pressure of 0.99. All surface area measurementswere calculated from the nitrogen adsorption isotherms by assumingthe area of the nitrogen molecule to be 0.162 nm2.

2.3. X-ray diffraction technique

The X-ray diffraction experiment was carried out with PhilipsX'pert diffractometer for 2θ values from 10 to 80° using Cu-Kα radia-tion at λ=1.54 Å. The other experimental conditions included 1/2°divergence slits and 5 s at each step and intensity was measured incounts.

2.4. FT-IR studies

The surface functional groups of the MAC samples were deter-mined through Fourier transform infra red spectrometer (PerkinElmer, German). The carbon samples were mixed with KBr of spec-troscopic grade and made in the form of pellets at a pressure ofabout 1 MPa. The pellets were about 10 mm in diameter and 1 mmin thickness. The samples were scanned in the spectral range of4000–400 cm−1.

2.5. Surface morphology

Surface morphology of the MAC900 samples was carried out usinga Leo-Jeol scanning electron microscope. The MAC900 samples werecoated with gold by a gold sputtering device for the clear visibilityof the surface morphology.

2.6. Isolation and identification of Bacillus sp.

The contact filter media used in facultative lagoon of ETP for thetreatment of wastewater containing SP was collected for the isolationof the bacterial species that have remarkable catabolic activity on SP.The adhered biofilms from the contact filter media were removedmanually and resuspended in phosphate buffer solution at pH 7.2.The suspended culture sample was serial diluted and were platedon nutrient agar medium. The plates were incubated at 37 °C and ex-amined for colonies after 24 h of incubation. Microorganisms wereidentified by conventional methods from analysis of their morpholog-ical, physiological and biochemical characteristics (Bergy's manual).They were Desulfovibrio salexigens, Desulfovibrio sp., and Bacillus sp.Screening of Bacillus sp., was carried out and cultured in minimal me-dium of composition K2 HPO4, 1.5 g/L; KH2PO4, 0.5 g/L; MgSO4.7H2O,0.5 g/L; CaCl2.2H2O, 0.002 g/L; FeSO4.7H2O, 0.002 g/L; pH 7.0. The utili-zation of SP as carbon sourcewas studied by batch cultivation of Bacillussp. in liquid minimal medium. The isolated bacteria were used for fur-ther studies.

2.7. Immobilization experiment

50 mL of media containing Bacillus sp. culture (of volume 1 mL) inthe logarithmic phase of growth having pre determined concentra-tion was taken in each of the two dry well stoppered bottles. About

://dx.doi.org/10.1016/j.msec.2012.10.026

3G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

1 g of accurately weighed MAC900 (dried over H2SO4) was added intothe bottle (sample) while the other bottle containing the cell suspen-sion without MAC900 served as control. The bottles under study wereplaced in a thermostatic bath and they were gently agitated at lowrpm. In batch adsorption studies the temperature of the thermostaticbath was set at 20–37 °C. The bottles were then removed from theconstant temperature bath and the solution was filtered through thestainless steel wire gauze to separate the MAC900 from the cell sus-pension. The optical density of the filtrate was determined using aUV–Visible spectrophotometer at λmax 600 nm. The cell density wasobtained from the calibration graph relating cell density and opticaldensity. From the concentration of cell before and after immobiliza-tion using MAC900, the number of cells immobilized per gram ofMAC900 was calculated.

XM

qð Þ; cellsg−1 ¼ C0−Ctð ÞVM

ð1Þ

C0 initial cell concentration, cells cm−3

Ct cell concentration at any time t, cells cm−3

V volume of cell suspension, cm3

M mass of MAC, gX/M (q), number of cells immobilized per gram of MAC, cells g−1

3. Results and discussion

3.1. Characteristics of MAC

3.1.1. N2 adsorption–desorptionFig. 1 presents nitrogen adsorption/desorption isotherms of

precarbonized carbon (PCC) at 77 K and MAC (C700, C800, and C900),and the isotherms of the PCC are of type I as the adsorption and desorp-tion branches remain nearly horizontal and parallel over awide range ofrelative pressures, which is the characteristic behavior of micro porousmaterials. However, in the isotherms of MAC700, MAC800, and C900, theknee becomes rounded. The hysteresis effect and the slope of the pla-teau increased to yield type IV isotherms with a significant increase in

Fig. 1. a. Nitrogen adsorption/desorption isotherms at 77 K a) PCC, b) MA

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

the nitrogen uptake through the entire pressure range, indicating thepresence of mesoporous matrix. This increase in uptake of nitrogen inthe samples with increasing temperature is as a result of the major in-crease in porosity created from the carbon and silica components.

3.1.2. Pore size and pore distributionFig. 1a shows the pore size distribution of activated carbon sam-

ples. The average pore size distribution is dependent mainly on theconcentration of chemical impregnation and the heat treatment tem-perature. The pore diameter was observed to increase with activationtemperature up to 800 °C and the same was reversed with further in-crease in temperature as shown in Table 2. The activated carbon wasobtained at 700 °C characterized by an average pore diameter of 38.8 Å,and a marginal increase in pore diameter of 39.36 Å at 800 °C. The nar-row increase in pore diameter suggests that up to 800 °C certain unor-ganized carbons or residual tar materials were expelled by opening ofclosed pores [28] and existing pores are widened into larger pores ofsmall magnitude through gasification of carbons in the pore walls hav-ing labile carbon structure [29]. But the decrease in the pore diameter inMAC900 is due to the suppression of the pore widening of stable carbonstructure formed during pre carbonization process.

3.1.3. X-ray studiesIt is a general fact that the creation of porosity is considerably

influenced by various factors such as clustering or fusion of activesites, structure of carbon, inorganic impurities and diffusion of gases.The internal structure of carbon is considered to be the most importantamong these factors. The X-ray diffractograms shown in Fig. 2 of thetwo samples PCC and MAC900 reveal a broad peak near 2θ≈23° forPCC and accompanied with several other noticeable small peaks. TheXRD pattern of the MAC900 corresponding to 2θ≈17° (200), 19(201),26(002), 27(331), 29 (110), 30(002) can be attributed to different hklplanes in graphitic microstructure. In addition to these peaks there isalso an additional peak at 22°,whichmay be considered due to the crys-tallites of SiO2. The high temperature employed for the preparation ofactivated carbonmay be considered as a reason for the formation of gra-phitic structures. It also clearly suggests that precarbonization followedby chemical activation and heat treatment produced relatively well or-ganized aromatic carbons with sp2 bonding character that is more sta-ble than the amorphous-like carbons of sp3 bonding character [30].

C700, c) MAC800 d) MAC900. b. Pore size distribution in MAC samples.

://dx.doi.org/10.1016/j.msec.2012.10.026

Fig. 3. FT-IR spectrum of a) PCC, b) MAC700, c) MAC800, and d) MAC900 carbon samples.

Table 2Average pore diameter, mesoporosity (mesopore volume Vmeso/total volume Vtot) andproduction yield of mesoporous activated carbon.

Sample Average pore diameter Å Vmeso/Vtot (%) Production yield of carbon %

C700 38.82 66.78 40.66C800 39.36 68.41 39.19C900 35.28 69.33 37.69

4 G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

3.1.4. FT-IR spectraThe infrared spectroscopic data presented in Fig. 3 provides infor-

mation on the functional groups in activated carbon samples. All theMAC samples exhibit a wide band at about 3350–3425 cm−1. Thismay be attributed to the O\H stretching mode of hexagonal groupsand adsorbed water molecules. The position and asymmetry of thisband at lower wave numbers indicate the presence of strong hydrogenbonds. Aweak band at 3780–3786 cm−1may be assigned to the isolatedO-Hgroup. TheMACsamplesMAC700,MAC800 andMAC900 have recordedthe absorption bands due to aliphatic C\H at 2920 cm−1 and this wasfound to be low in the precarbonized sample (PCC). A very small peaknear 1700 cm−1 is assigned to C_O stretching vibrations of ketones, al-dehydes, lactones or carboxyl groups. The weak intensity of this peak inall the MAC samples at 1700 cm−1 may be assigned to carboxyl group.The band near 1615 cm−1 in PCC is due to aromatic stretching vibrationC_C. This was absent in MAC800 and MAC900 at high temperature. Abroad band between 1250 and 1000 cm−1 is observed. The broad peakat about 1095 cm−1 in the PCC sample and a broad peak shouldered at1180 cm−1 in MAC700, MAC800 and MAC900 indicate the presence ofphosphorous and oxygen content of the samples. The peak at1190 cm−1 is assigned to phosphate ion in MAC samples arisingdue to phosphoric acid activation and the band at 1203 cm−1 isdue to phosphoric acid esters [31,32]. The appearance of bands at900–1300 cm−1 could be due to phosphorous species generated dur-ing phosphoric acid activation of PCC. The peak at 1180 cm−1 may beassigned to the stretching mode of hydrogen-bonded P_O to O\Cstretching vibrations in P\O\C (aromatic) linkage and to P_OOH.The shoulder at 1080–1070 cm−1 may be ascribed to ionized linkageP+\O− in acid phosphate esters and to symmetrical vibration in achain of P\O\P (polyphosphate).

Fig. 2. XRD curves for PCC and MAC900 carbon samples.

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

3.1.5. Surface morphologyFig. 4a shows the surface morphology of the MAC samples before

and after phosphoric acid activation. The precarbonized sample(PCC) does not contain evidence for the pore formation in therice husk indicating mere carbonization of the raw material with-out creation of pores. Fig. 4b illustrates the formation of pores inthe MAC900 sample due to chemical activation and heat treatment.The opening of the pores in the rice husk matrix should be due tothe volatilization of some materials, e.g. lignins and other organiccomponents from the rice husk during impregnation process suchthat to create the micro and meso pores. The diameter at themouth of the pores was larger than at the interior of the pores.The diameter at the mouth of the pores conformed to macropores and the diameter at the interior of the pores assumed thedimensions of meso and micro pores. Fig. 4c showed scanningelectron micrograph of the Bacillus sp. immobilized in mesoporousactivated carbon (MAC900).

3.2. Effect of contact time

The effect of contact time on immobilization of Bacillus sp. isshown in Fig. 5a. The figure illustrates that the immobilization pro-cess progressed with the passage of contact time and it attained equi-librium after 7 h irrespective of the bulk concentration of the cellsuspension. The initial linear portion of the curve increased up to acertain maximum qmax value and was followed by a smooth curveto reach the plateau value after the lapse of contact time of 4 h. Thelinear portion of the curve could be attributed to the migration of bac-terial cells from the bulk cell suspension to the particle surface of theMAC900. The plateau portion of the curve corresponds to diffusion ofBacillus sp. from the bulk solution into the pores and condensationon the walls of them. Fig. 5a illustrates the strong influence of concen-tration of cell suspension on the quantity of cells immobilized inMAC900. The quantity of bacterial cell immobilized from the bulkcell suspension increased with its concentration at any given time.The concentration of bacterial cells in suspension is considered to bethe major factor for offering driving force for cell immobilization fromthe bulk cell suspension i.e. the concentration gradient establishedacross the plane of MAC900 and cell suspension determined the rate ofdiffusion of the latter from the bulk phase.

://dx.doi.org/10.1016/j.msec.2012.10.026

Fig. 4. a) Scanning electron micrograph of the precarbonized sample (MAC). b) Activated carbon sample (MAC900). c) Bacillus sp. immobilized in the mesoporous activated carbon(MAC900).

5G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

3.2.1. Rate constant for immobilization Bacillus sp. in MAC900The rate constant for immobilization of Bacillus sp. cells in MAC900

was calculated using Lagergren rate Eq. (2)

log qe−qtð Þ ¼ logqe−Kimmt

2:303ð2Þ

where,

qe is the amount of cells immobilized in the MAC900 at equilib-rium, cells g−1

qt is the amount of cells immobilized in MAC900 at any time, t,cells g−1

t is time(min).

The rate constants for surface immobilization of cells (Kimm) at dif-ferent set conditions were calculated from the slope of the plot log(qe−qt) versus time. The plot of log (qe−qt) versus time presents astraight line, indicating that Lagergren's equation was applicable to

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

the immobilization of Bacillus sp. in MAC900 and the immobilizationprocess followed pseudo first order rate kinetic process. The datacontained in Table 3 suggest that an increase in the initial cell con-centration leads to an increase in the immobilization rate constantand percentage immobilization Bacillus sp. in MAC900 up to a con-centration 4.5×107 cells/mL and thereafter it decreased drastically.This could be due to the overproduction of extra cellular polymersin the medium at high concentration which promoted cell aggrega-tion leading to enhanced rate of immobilization.

3.2.2. Intra particle diffusionActivated carbon contains outer pore surface area and longitudinal

pores of varying degrees of diameter. The diameter of pores in MAC900used in the present study belongs to transitional pores of size 15 nm(micro pores have size range of 0–2 nm, meso pores in the size rangeof 2–50 nm, macro pores have size in the range above 50–500 nm).The outer pore surface area of activated carbon contains a random dis-tribution of active sites imparted with varying degrees of activation en-ergy. The microorganisms that migrate to the carbon surface from the

://dx.doi.org/10.1016/j.msec.2012.10.026

Fig. 5. a) Effect of contact time on immobilization of Bacillus sp. in MAC. b) Intra particle diffusion behavior of Bacillus sp. in MAC900. c) Effect of particle size on immobilization ofBacillus sp. in MAC900. d) Variation of immobilization capacity of Bacillus sp. at different pH of cell suspension.

6 G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

bulk phase are first held at the active sites distributed across the outerpore surface area. After exhaustion of the active sites at the surfacethe microbes tend to diffuse into the pores and condense on to thewalls of the pores at rates determined by the variables such as concen-tration, pH and temperature. The intra particle diffusion is limited bybacterial layer thickness which tends to be formed at the mouth of thepores and in turn is controlled by concentration of the bulk solution.The intercept on√t axis from the plot of the amount of cell immobilizedversus √t gives the external layer thickness and slope of the plot sig-nifies intra particle diffusion rate constant. Fig. 5b presents intra particlediffusion behavior of Bacillus sp. at different concentrations at 20 °C. Thefigure suggests that immobilization at the outer pore surface area was3.0×107 cells g−1 and immobilization due to pore diffusion was1.2×107 cells g−1 at concentration 4.4×107 cells cm−3, at pH 7.0and at temperature 20 °C. Ratio of immobilization of Bacillus sp. inpores to outer pore surface area was 0.4. The intra particle diffusion

Table 3Rate constants for surface and pore immobilization of Bacillus sp. at different concentration

Concentration(×10−7 cells cm−3)(co)

Concentration(×10−7cells cm−3)(ct)

qe(×10−7cells g−1)

2.2 2.17 1.023.6 3.54 2.894.5 4.43 3.467.3 7.25 2.28

Experimental conditions; pH, 7.0; mass of MAC900, 1 g; particle size, 600 μm, temperature,

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

rate constant at different temperatures is given in Table 4. The kineticdata was tested for pore diffusion using the following equation

Di ¼ 0:03r20t1=2

ð3Þ

where,

t1/2 is the time for half value equilibrium immobilization, hr0 is the radius of the MAC900 particles,Di is the diffusion coefficient, m2/s.

The diffusion coefficient of the Bacillus sp. from the bulk phase to thesurface of the MAC900 was 4.2×10−13 m2/s. This is in close agreement

s in MAC900.

Immobilization, % Kimm

min−1Kp

(×10−7cells g−1 min−0.5)

46.22 0.0024 0.005760.17 0.0028 0.004076.88 0.0030 0.002831.23 0.0019 0.0020

20 °C.

://dx.doi.org/10.1016/j.msec.2012.10.026

Table 4Rate constants for surface and pore immobilization of Bacillus sp. in MAC900 at different temperatures.

Temp °C Concentration(×10−7 cells cm−3)(ct)

qe(×10−7cells g−1)

Percentage Immobilization Kimm

min−1Kp

(×10−7cells g−1 min−0.5)

20 4.342 2.90 76.31 0.0025 0.004025 4.342 2.88 75.78 0.0028 0.006030 4.343 2.82 74.21 0.0031 0.007637 4.345 2.74 72.12 0.0033 0.0097

Experimental conditions: pH, 7.0; concentration of cells, 4.4×107 cm−3; mass of MAC900, 1 g and particle Size, 600 μm.

7G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

with the values generally reported for the diffusion of solutes throughporous adsorbents.

3.2.3. Effect of particle sizeFig. 5c shows the variation of immobilization tendency with the

particle size (300, 600 and 1000 μm) of the MAC900 for a fixed concen-tration of cell suspension. The immobilization of Bacillus sp. from bulkcell suspension was relatively higher with particle size of 300 μm com-pared to other sizes of 600 μm and 1000 μm. Similar observation wasrecorded by Ehrhardt et al. [33]while using activated carbon as a carriermatrix for immobilization of Pseudomonas putida. The rate constant forimmobilization of Bacillus sp. decreasedwith the increase in diameter ofthe particle of MAC900 as governed by the equation.

Kimm ¼ 0:0095 Φ−0:1653 ð4Þ

where Φ is particle size expressed in micrometer.The pores in the activated carbon are classified into macro pores,

transitional pores and micro pores in the decreasing order of pore di-ameter. The macro pores can contain the microbes more strongly. Inmacro pores the Bacillus sp. develops cohesion with the walls of thepores through extra cellular fibrils produced by the bacteria so thatthey are difficult to be dislodged from the pores. The microbes held inmacro pores resist mechanical attrition force of fluids that might occurin a plug flow type reactor for removal of pollutants from wastewateror air stream using immobilized organism. This fact was confirmed bywashing the MAC900 immobilized with Bacillus sp. using buffer solutionof pH 7.2 at high flow rate for 6 h. The wash water contained negligibleconcentration of Bacillus sp. (Table 5).

3.2.4. Effect of pHFig. 5d shows the magnitude of immobilization of Bacillus sp. in

MAC900 at different pH of cell suspension. The immobilization startedto increase with increase in pH and reached the maximum value at6.6. Beyond 6.6 the intensity of immobilization decreased with fur-ther increase in pH. It is known that PZC of Bacillus sp. and MAC900are 4.5 and 6.66 respectively. Thus, the initial rise in immobilizationis due to electro static attraction and electrostatic repulsion may bethe factor for the decrease in immobilization beyond the pH 6.6. Atthe optimum pH the MAC900 was expected to carry weakly negativecharge. At pH beyond 6.6, lateral expansion of biopolymer or lateralrepulsion between the similarly charged exocellular membrane andparticle surface could be dominating the immobilization process leadingto poor immobilization. The extracellular matrix, a polymer matrix inwhich cells are embedded, accounts for the majority of the biofilm

Table 5Rate constants for surface and pore immobilization of Bacillus sp. in MAC900 at different pa

Particle Size Concentration(×10−7 cells cm−3)(ct)

qe(×10−7cells g−1)

Pe

300 4.318 4.06 92600 4.335 3.25 731000 4.356 2.12 48

Experimental conditions: pH, 7.0; concentration of cells, 4.4×107 cm−3; mass of MAC900, 1

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

volume. The extracellular matrix generally consists of chelating ligands;polysaccharides up to 65%while proteins usually comprise up to 10–15%of its total biomass. Most of these bio polymers are anionic in aqueousmedium. The negatively charged extracellular matrix has tendency tobind with positively charged MAC900 in aqueous medium and renderingthe surface to acquire positive charge. Hence, immobilization increasedwith increase in pH of the medium up to 6.6 and thereafter decreasedin immobilization. The trend continued up to pH 9.0 (Table 6).

3.2.5. Effect of temperatureThe effect of temperature on immobilization of Bacillus sp. onMAC900

can be explained in accordance with adsorption isotherm findings. InFig. 6 the adsorption isotherm of Bacillus sp. on MAC900 at different tem-peratures has been compared with each other, at pH 7.0. The immobili-zation capacity was observed to decrease significantly for a givenconcentration of cell suspension with rise in temperature. This sort ofresult is generally expected for the adsorption process accompaniedwith evolution of heat.

Kimm ¼ 0:0226T0:3984 ð5Þ

The experimental data for the immobilization of Bacillus sp. inMAC900 at different temperatures and pHs have been followed inthe rearranged Langmuir isotherm model.

Ce

qe¼ 1

Q0bþ 1Q0Ce

ð6Þ

The liner plots of Ce/qe against Ce (Fig. 7a) at different tempera-tures suggest the applicability of Langmuir model, indicating the for-mation of monolayer coverage of Bacillus sp. on the surface of MAC900.The constants Q0 and b have been determined from the slope and inter-cept respectively at temperatures of 20, 25, 30 and 37 °C. The validity ofthe above model was retested by the regression analysis for the resultsderived at pH 7 as shown below.

Ce

qe¼ 0:2676þ 0:1708Ce r2 ¼ 0:9913

Q0 ¼ 3:73b ¼ 1:56ð7Þ

Ce

qe¼ 0:3076þ 0:2141Ce r2 ¼ 0:9874

Q0 ¼ 3:25b ¼ 1:44ð8Þ

rticle sizes.

rcentage immobilization Kimm

min−1Kp

(×10−7cells g−1 min−0.5)

.27 0.0025 0.0040

.86 0.00277 0.0060

.18 0.00305 0.0076

g and temperature, 20 °C.

://dx.doi.org/10.1016/j.msec.2012.10.026

Table 6Rate constants for surface immobilization and pore immobilization of Bacillus sp. in MAC900 at different pH conditions.

pH Concentration(×10−7cells cm−3)(ct)

qe(×10−7cells g−1)

Percentage immobilization Kimm

min−1Kp

(×10−7cells g−1 min−0.5)

6.0 4.348 2.58 58.59 0.00271 0.00156.6 4.335 3.22 73.18 0.00276 0.00427.0 4.336 3.20 72.26 0.00275 0.00408.0 4.353 2.31 52.56 0.00254 0.00289.0 4.361 1.93 43.86 0.00196 0.0016

Experimental condition: concentration of cells, 4.4×107 cm−3; mass of MAC900, 1 g; particle size, 600 μm and temperature, 20 °C.

8 G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

Ce

qe¼ 0:3692þ 0:3359Ce r2 ¼ 0:9899

Q0 ¼ 2:71b ¼ 1:10ð9Þ

Ce

qe¼ 0:4108þ 0:3778Ce r2 ¼ 0:9973

Q0 ¼ 2:43b ¼ 1:09ð10Þ

The values of Q0 and b were also calculated from Eqs. (3), (4), (5)and (6) and the values were coinciding with the graphical values asshown in Table 7, indicating the fitness of the above model to thepresent study. The adsorption capacity or monolayer coverage Q0, ofMAC900 was the maximum at pH 7.0 and thereafter it decreasedwith increase in pH. Q0 and the affinity factor b, decreased with in-crease in temperature (Table 7). The affinity factor b is related tonet enthalpy of adsorption ΔH0 in the following equation as

b ¼ b′e−ΔH0RT : ð11Þ

The value of ΔH0 was obtained from the plot of log b against 1/T(Fig. 7b) and it was −46.9 kJ/N cells (N=Avogadro number) at pH7.0. The negative value of ΔH0 reflects the exothermic nature of theimmobilization process and relatively higher values suggest the firmbonding nature of Bacillus sp. with the active sites of MAC900.

3.3. Thermodynamic parameters of immobilization

The free energy change due to transfer of 1 mol of Bacillus sp. frombulk phase of cell suspension to theMAC900 surface has been calculatedusing the Eq. (12)

ΔG0 ¼ −RT lnb: ð12Þ

Fig. 6. Effect of temperature on immobilization of Bacillus sp. in MAC900.

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

The entropy of immobilization of Bacillus sp. was calculated usingEq. (13)

ΔS0 ¼ΔH0−ΔG0

� �

T: ð13Þ

Thermodynamic data for the immobilization of Bacillus sp. inMAC900 are given in Table 8. The greater negative free energy valuesat pH 6 and pH 7 imply the favorable conditions for the immobiliza-tion process to occur and the same was retarded at pH 8 and 9 as re-vealed by positive values of ΔGo values. The enthalpy of immobilizationwas−46.90 kJ/mol at pH 7.0, indicating the firm bonding of the organ-ism, probably mediated by electrostatic and hydrogen bonding betweenbacteriophage and the active sites of MAC900. The negative values forentropy of immobilization are the measure of preferred orientation ofBacillus sp. onto the surface of MAC900 and also indicating the faster in-teraction with the active sites of the MAC900. The cellular componentssuch as polysaccharides bind with the active sites of MAC forming an ac-tivated complex. The complex is typically noncovalently bound. Theminimum value of entropy at pH 9.0 indicates that the degrees of free-dom are restricted.

3.4. Effect of mass of MAC900

The effect of varying themass ofMAC900 on immobilization of Bacillussp. cells from bulk cell suspension concentration of 5.1×107 cells cm−3

at pH 7.0 and at temperature of 20 °C is shown in Fig. 7c. The immobili-zation of cells was found to increase linearly with increase in mass ofMAC900 equilibrated with cell suspension. The probable explanationcould be that as themass of MAC900 was increased both the available ex-ternal surface area and internal pore surface area were increased andconsequently the number of active sites available for immobilization ofBacillus sp. was also increased proportionately. The rate constant andthe mass of MAC900 for immobilization of Bacillus sp. in activated carbonwere controlled by the equation

Kimm ¼ 0:00105M0:6607: ð14Þ

3.5. Confirmation of immobilized Bacillus sp. in MAC900

The immobilization of bacterial species in activated carbon wascarried at optimum pH 7.0, temperature 20 °C and bacterial cell con-centration 3.2×107 cells/mL. The organisms in the immobilized statewere confirmed by extracting the immobilized carbon in 1 M NaOHand the extractant was analyzed for protein by the Lowry method.

3.6. Batch study on degradation of sulphonated phenolic compound inaqueous solution using immobilized Bacillus sp.

Bacillus sp. was precultured in nutrient broth medium for twodays and then transferred to SP containing minimal medium to induce

://dx.doi.org/10.1016/j.msec.2012.10.026

Fig. 7. a) Langmuir isotherms for the immobilization of Bacillus sp. in MAC900 at different temperatures. b) Variation of Langmuir constants with inverse temperature. c) Effect ofmass of activated carbon on immobilization of Bacillus sp. in suspension. d) The degradation profile of aqueous sulphonated phenol (SP) by free and immobilized Bacillus sp.,expressed in terms of COD.

9G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

its activity. The pure culture was washed twice with sterile water andre-suspended in minimal medium of volume 50 mL at pH 7, and then1 g of MAC900 was added to the above cell suspension. The mixturewas incubated at temperature of 20 °C for 24 h. The MAC900 was sepa-rated from themedium after incubation by filtration. The concentrationof Bacillus sp. in the cell suspension after immobilization was deter-mined by plating technique. The cell density of Bacillus sp. in MAC900was observed to be 3.2×107 cells g−1. The Bacillus sp. immobilized inMAC900 was used for the degradation of SP compounds (300 mg/L) inminimal medium. The Bacillus sp. in free suspension state was used ascontrol. The MAC900 immobilized with Bacillus sp. and free cell suspen-sion of same cell density were added tominimal medium containing SPin two separate Erlenmeyer flasks. Themixturewas agitated at 100 rpmfor 24 h in a shaker. Aliquots of samples were withdrawn at differenttime intervals and they were analyzed for COD. The microbial growthin the suspensionwas recorded at λ600 nm. The rate constant for degra-dation of SP, expressed in terms of COD, was obtained from the plot ofLog (COD)o/(COD)t against time (Fig. 7d). The determination of phenol

Table 7Langmuir isotherm parameters for adsorption of Bacillus sp. on MAC.

Temperature °C pH

6.0 7.0 8.0 9.0

Q0 b Q0 b Q0 b Q0 b

20 2.559 2.116 3.412 1.809 2.518 0.905 2.374 0.68925 2.465 1.202 3.129 1.615 2.337 0.794 2.010 0.78030 2.355 0.724 2.450 1.041 2.254 0.733 2.008 0.73437 2.139 0.559 2.373 0.662 2.129 0.577 1.985 0.655

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

content of the SP degraded sample confirmed that the complete removalof SP did occur after 28 h of incubation. Sulphonated phenolic compoundis expected to be mineralized through the sequential steps consisting ofdesulphonation of compounds by organisms, hydroxylation, opening ofaromatic rings and oxidation of simple organics into CO2 and water.Desulphonation of SP is considered to be the rate limiting step. This isperformed by extra cellular enzymes followed by hydroxylation. The en-zyme secretion is controlled by the diffusion of organic molecules intothe exocellular membrane. The exploration of the enzyme required forhydrolysis of sulphonated phenolic molecules into fragmented organicmolecule is under way. As a consequence of the rate limiting step suchas desulphonation of SP inmineralization leads to non linear growth pat-tern. The COD is an indicative parameter for the degradation of SP in theminimal medium as shown in Fig. 7d. As an evidence to the growth pat-tern of Bacillus sp. on degradation of SP compounds, reduction in CODwas also non linear with time. These rate constants for the degradationof aqueous SP by free and immobilized Bacillus sp. were 0.00408 h−1

Table 8Thermodynamic parameters for immobilization of Bacillus sp. in MAC900.

pH Enthalpy, ΔH0

(kJ mol−1)Free energy, ΔGo

(kJ mol−1)Entropy, ΔS0

(J K−1mol−1)

6 −58.49 −0.45 −197.807 −46.90 −1.18 −161.408 −19.54 +0.57 −63.659 −3.22 +0.61 −8.74

Experimental conditions: concentration of cells, 4.4×107 cm−3; mass of MAC900, 1 g;particle size, 600 μm, temperature, 20 °C.

://dx.doi.org/10.1016/j.msec.2012.10.026

10 G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

and 0.32 h−1 respectively in 28 h at the same concentration of aqueousSP (30 mg/100 mL).

The same experiment was repeated for several cycles, it was ob-served that the immobilized Bacillus sp. was able to present the consis-tency in the removal of SP for about 25 cycles, thereafter thematrixwasagain re-immobilized with Bacillus sp. The reimmobilized matrix alsorecorded the similar COD removal pattern.

3.7. Effect of glucose on degradation of phenolic syntan

Sulphonated phenolic compoundwas considered to be the refracto-ry organics which require the release of enzymes for breakdown of themolecules into simpler fragments by hydroxylation, desulphonation,ring opening and oxidation to yield CO2 and H2O. The objective of thisstudy on the effect of glucose on degradation of aqueous SPwas to illus-trate that the immobilized organism was compound specific in nature.It is clearly shown in the table that COD of the minimal mediumcontained fixed quantity of SP 30 mg/100 mL and varied concentrationof glucose (0.025, 0.075, 0.1, 0.125 g/100 mL). The COD remaining after24 h was almost constant irrespective of the glucose concentration forthe fixed quantity of SP. The eliminated COD was calculated from thevalues of initial COD and final COD obtained. Amongst the intermediatesteps in oxidation of sulphonated phenolic compound into the endproducts, hydroxylation step is the rate limiting step for aromatic ringopening and subsequently to oxidation of organics. Hydroxylation ofsulphonated phenol is the endergonic reaction, required external ener-gy for the bond formation. This was facilitated by the oxidation of glu-cose in the medium. This is clearly shown in Table 9 that increasein glucose increased the COD elimination (through the removal ofsulphonated phenolic compounds). This supports the view that elimina-tion of SP or recalcitrant molecule requires the addition of easily assimi-lable carbon source like glucose. The sustainable growth of Bacillus sp. ondegradation of sulphonated phenolic compounds did occur at the opti-mum concentration of glucose (0.075%). The slope of the initial linearline of the plot represents the rate constant for the degradation of SPby the immobilized Bacillus sp. in MAC900.

4. Conclusion

Bacillus sp. isolated from an ETP treating the wastewater containingsulphonated phenolic compounds was immobilized in MAC900 at opti-mum conditions pH, 7.0; temperature, 20 °C; particle size, 600 μm andmass of MAC900, 1 g. The rate of immobilization increased with increasein concentration up to 4.5×10−7 cells cm−3 and thereafter declinedwith further rise in concentration. The rate constant for immobilizationdecreased with increase in temperature from 20 to 37 °C. The rate con-stant for immobilization increased with decrease in particle diameterfrom 1000 to 300 μm. The rate constant for immobilization was themaximum at pH 7.0 and it acquired a less value at pH 6.0, 8.0 and 9.0.Enthalpy of immobilization was −46.9 kJ/mol at pH 7.0, indicatingthe exothermic nature and firm bonding of Bacillus sp. with carbonmatrix. The immobilized Bacillus sp. in MAC900 was able to degradesulphonated phenolic compounds in minimal medium with the rateconstant of 0.32 h−1. The degradation of sulphonated phenolic com-pound increased with the addition of glucose.

Table 9Effect of glucose on elimination of COD of aqueous sulphonated phenolic compound.

Glucose concentration(g/100 mL)

Initial CODmg/L

Final CODmg/L

Eliminated CODmg/L

0.025 308 260 480.05 373 265 1080.075 404 288 1200.1 454 302 1520.125 486 307 179

Sulphonated phenolic compound, 30 mg/100 mL; NH4Cl, 0.025%; temperature, 20 °C.

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

Acknowledgement

The authors are thankful to the Council of Scientific and IndustrialResearch (CSIR) and the Central Leather Research Institute, India, forproviding all the facilities needed to carry out this work.

Appendix A

NomenclatureC concentration of cell suspension, cells/cm3

X number of cells immobilized, cells/gM mass of activated carbon, gKimm rate of immobilizationCe concentration of Bacillus cells at equilibrium, cells cm−3

Co initial concentration of Bacillus cells, cells cm−3

Ct concentration of Bacillus cells at any time, cells cm−3

qe Bacillus cells immobilized in activated carbon at equilibri-um, cells g−1

qt Bacillus cells immobilized in activated carbon at time,cells g−1

b Langmuir constant, intensity factor cm3 cells−1

Q0 Langmuir constant, monolayer capacity, g cm−3

Kimm rate constant for surface immobilization of Bacillus sp. in ac-tivated carbon, min−1

Kp rate constant for pore immobilization of Bacillus sp. in acti-vated carbon, min−0.5

T temperature, °CR gas constant, J deg−1 mol−1

Di diffusion coefficient of Bacillus cells through carbon matrix,m2 s−1

ro radius of the carbon particle, mt1/2 time for half immobilization, sϕ diameter of carbon particle, mΔH0 enthalpy of immobilization of Bacillus sp., kJ mol−1

ΔGo free energy of immobilization of Bacillus sp., kJ mol−1

ΔSo entropy of immobilization of Bacillus sp., J K mol−1

References

[1] G. Sekaran, S. Karthikeyan, K. Ramani, B. Ravindran, A. Gnanamani, A.B. Mandal,Chem. Lett. 9 (2011) 499–504.

[2] Gholamreza Moussavi, Maryam Mahmoudi, Behnam Barikbin, Water Res. 43(2009) 1295–1302.

[3] Zhiquan Liu, Fuyi Cui, Hua Ma, Zhenqiang Fan, Zhiwei Zhao, Zhenling Hou,Dongmei Liu, Water Res. 46 (2012) 2290–2298.

[4] H.H. Locher, Bacterial Ph.D. thesis on 9434, Zurich (1991).[5] S. Karthikeyan, A. Titus, A. Gnanamani, A.B. Mandal, G. Sekaran, Desalination 281

(2011) 438–445.[6] T. Reemtsma, J. Jochimsen, M. Jekel, Vom. Wasser. 81 (1993) 353–363.[7] T. Reemtsma, O. Fiehn, G. Kalnowski, M. Jekel, Environ. Sci. Technol. 29 (1995)

478–485.[8] W.B. Hugo, A.D. Russel, New York. (1977) 202–208.[9] M. Okada, H. Morinaga, W. Nishijima, Water Sci. Technol. 42 (12) (2000) 149–154.

[10] F. Westmeier, H.J. Rehm, Appl. Microbiol. Biotechnol. 26 (1987) 78–83.[11] V.K. Gupta, Arshi Rastogi, Arunima Nayak, J. Colloid Interface Sci. 342 (2010)

135–141.[12] G. Sekaran, S. Karthikeyan, C. Nagalakshmi, A.B. Mandal, Environ. Sci. Pollut. Res.

(2012), http://dx.doi.org/10.1007/s11356-012-0891.[13] V.K. Gupta, A. Rastogi, Colloids Surf. B 64 (2008) 170–178.[14] V.K. Gupta, A. Rastogi, J. Hazard. Mater. 154 (2008) 347–354.[15] V.K. Gupta, Suhas, I. Ali, V.K. Saini, Ind. Eng. Chem. Res. 43 (2004) 1740–1747.[16] V.K. Gupta, A. Mitta, V. Gajbe, J. Mittal, Ind. Eng. Chem. Res. 45 (2006) 1446–1453.[17] V.K. Gupta, A. Mittal, L. Kurup, J. Mittal, J. Colloid Interface Sci. 304 (2006) 52–57.[18] V.K. Gupta, R. Jain, A. Mittal, M. Mathur, S. Sikarwar, J. Colloid Interface Sci. 309

(2007) 464–469.[19] V.K. Gupta, R. Jain, S. Varshney, J. Hazard. Mater. 142 (2007) 443–448.[20] Hardy Temmink, Katja Grolle, Bioresour. Technol. 96 (2005) 1683–1689.[21] B. Antizar-Ladislao, N.I. Galil, Water Res. 38 (2004) 267–276.[22] I. del Castillo, P. Hernández, A. Lafuente, I.D. Rodríguez-Llorente, M.A. Caviedes, E.

Pajuelo, Water Res. 46 (2012) 1723–1734.[23] L.J. Kennedy, K. Mohan das, G. Sekaran, Carbon 42 (2004) 2399–2407.[24] Shengbing He, Gang Xue, J. Hazard. Mater. 178 (2010) 895–899.

://dx.doi.org/10.1016/j.msec.2012.10.026

11G. Sekaran et al. / Materials Science and Engineering C xxx (2012) xxx–xxx

[25] A. Morsen, H.J. Rehm, Appl. Microbiol. Biotechnol. 33 (1990) 206–212.[26] N.K. Pazarlioglu, A. Telefoncu, Process Biochem. 40 (2005) 1807–1814.[27] S.E. Agarry, M.O. Aremu, Br. Biotechnol. J. 2 (1) (2012) 26–48.[28] O. Gyu Hwan, C.R. Park, Fuel 81 (2002) 327–336.[29] D. Yulu, W.P. Walawender, L.T. Fan, Bioresour. Technol. 81 (2002) 45–52.[30] Y. Chang Hun, Y.H. Park, C.R. Park, Carbon 39 (2001) 559.

Please cite this article as: G. Sekaran, et al., Mater. Sci. Eng., C (2012), http

[31] A. Manivannan, M. Chirila, N.C. Giles, M.S. Seehra, Carbon 37 (1999) 1741–1747.[32] A.M. Puziy, O.I. Poddubnaya, Martinez-Alonso, F. Suarez-Garcia, J.M.D. Tascon,

Carbon 40 (2002) 1493–1505.[33] H.M. Ehrhardt, H.J. Rehm, Appl. Microbiol. Biotechnol. 30 (1989) 312–317.

://dx.doi.org/10.1016/j.msec.2012.10.026