Author Query Sheet

Manuscript Information

Journal Acronym

TENT Volume and issue

Author name Réguer Manuscript No. (if applicable)

512297

AUTHOR: The following queries have arisen during the editing of your manuscript. Please answer the queries by making the necessary corrections on the CATS online corrections form. Once you have added all your corrections, please press the SUBMIT button.

QUERY NO. QUERY DETAILS 1 2 3 4 5 6 7 8

Please provide the location (city, country) of the supplier/manufacturer. To follow the numeric style for references, please note that [27] has been added against Chuang et al. and [28] has been added against Lorimier et al. If these are not the correct references, please change them. The reference by Qi, 2006 (legend), is not included in the reference list. Please add it to the list and provide the reference number when you return the proof, so that the figure legend can be changed. Please also make any necessary changes to the numbering of other citations and the reference list. Please note that ‘the external transfer resistance (1/kC) was lower in the fixed-bed experiment than in the batch experiments that led to a global resistance 1/k1 lower (and then a global conductance k1 higher)’ has been changed to ‘the external transfer resistance (1/kC) was lower in the fixed-bed experiment than in the batch experiments, which led to a global resistance that was 1/k1 lower (and then a global conductance that was k1 higher)’. If this is not correct, please rephrase the original text to make it clearer. Should ‘even if there is a difference’ be changed to ‘even though there was a difference’? Please provide the location (city/town) of the publisher

AQ

Artwork Query (from Artwork Dept.) Please resupply this artwork in a format suitable for printing. After resampling and resizing for typesetting, the resolution of the figure [Fig1_80, fig2_119, Fig3_143, Fig4_132] is not appropriate [the min. requirement is of 300 dpi].

Environmental Technology

Vol. 00, No. 0, Month 2010, 1–10

ISSN 0959-3330 print/ISSN 1479-487X online© 2010 Taylor & FrancisDOI: 10.1080/09593330.2010.512297http://www.informaworld.com

5

10

15

20

25

30

35

40

45

50

55

Measurement and modelling of adsorption equilibrium, adsorption kinetics and breakthrough curve of toluene at very low concentrations on to activated carbon

Anne Réguer*, Sabine Sochard, Cécile Hort and Vincent Platel

Laboratoire de Thermique, Energétique et Procédés, Université de Pau et des Pays de l’Adour, Quartier Bastillac, 65000 Tarbes, France

Taylor and Francis

(

Received 15 March 2010; Accepted 22 July 2010

)

10.1080/09593330.2010.512297

Indoor air pollution, characterized by many pollutants at very low concentrations, is nowadays known as a worryingproblem for human health. Among physical treatments, adsorption is a widely used process, since porous materialsoffer high capacity for volatile organic chemicals. However, there are few studies in the literature that deal withadsorption as an indoor air pollution treatment. The aim of this study was to investigate the adsorption of toluene onto activated carbon at characteristic indoor air concentrations. Firstly, global kinetic parameters were determined byfitting Thomas’s model to experimental data obtained with batch experiments. Then, these kinetic parameters led tothe determination of Henry’s coefficient, which was checked with experimental data of the adsorption isotherm.Secondly, we simulated a breakthrough curve made at an inlet concentration 10 times higher than the indoor air level.Even if the kinetic parameters in this experiment are different from those in batch experiments, it can be emphasizedthat the Henry coefficient stays the same.

Keywords:

indoor air; adsorption; activated carbon; Henry coefficient; Thomas’s model

Introduction

Lots of pollutants have been detected inside buildings[1–3] and in other confined environments, such as carsand aeroplane cabins [4–6], for several years now.Despite the low concentrations of all these pollutants(about one to several hundred micrograms per squaremetre), their harmful effects on health [7,8] have madeindoor air pollution a matter of great interest. Amongthese pollutants, volatile organic compounds (VOCs)are frequently encountered.

The following non-exhaustive list sets out commontechnologies applied to VOC control [9]:

●

physical technologies: absorption, adsorption,membrane processes, cryocondensation etc.

●

chemical technologies: oxidation, photocatalysis,catalytic oxidation, chemical scrubbing etc.

●

biological technologies: biofiltration, biotricklingfilters, bioscrubbers, membrane bioreactors etc.

All these common processes are generally applied towaste gas streams with high concentrations and to typi-cal industrial VOCs (toluene, methanol, hexane, methylisobutyl ketone etc.).

Among the physical treatments, adsorption is awidely used process as porous materials offer high

capacities for VOCs [10,11]. Thanks to its high porosityand high surface area, activated carbon (AC) is one ofthe most versatile adsorbents [12,13]. However, fewstudies dealing with adsorption at concentrations closeto indoor air pollution levels are available in the litera-ture, because of technological limitations of the treat-ment. Qi and LeVan [14] measured an adsorptionisotherm of toluene on AC up to indoor air pollutionlevels (41

µ

g m

−

3

). Yao

et al

. [15] estimated the adsorp-tion capacity at equilibrium of an activated carbon fibrecloth at a concentration of toluene in the gas phase of434

µ

g m

−

3

. Schindler

et al

. [16] studied the adsorptionof n-pentane on AC in the Henry’s law region (0.014

µ

gm

−

3

). The extrapolation of adsorption data obtained athigh concentrations is not valid. For this reason, studiesof VOC adsorption at concentrations close to indoor airpollution levels are crucial. Moreover, Carratala-Abril

et al

. [17] pointed out that, although mathematicalmodels to simulate and model the adsorption of aromaticcompounds are available in the literature, few publishedresults focus on modelling at very low concentrations.

Therefore, the first aim of the present study was toinvestigate the adsorption of toluene on to AC in theHenry’s law region: adsorption equilibrium and adsorp-tion kinetics were studied. Then a correlative kineticmodel, using a global parametric approach, was applied

*Corresponding author. Email: anne.reguer@univ-pau.fr

TENT_A_512297.fm Page 1 Monday, September 27, 2010 6:03 PM

2

A. Réguer

et al.

5

10

15

20

25

30

35

40

45

50

55

to kinetic experimental data. Finally, kinetic parameterswere used in a simple predictive model for the adsorp-tion column. The predictions of the model werecompared with the experimental breakthrough curve.

Materials and methods

Materials

Adsorbent

A commercial AC, made from coconut shell, was usedin this work (C1220 G90, Carbio 12). Its porous charac-teristics were measured by nitrogen adsorption at 77 K,using a Micromeritics ASAP 2010 analyser. Specificsurface area was determined from the isotherm obtainedby using the Brunauer–Emmett–Teller (BET) method.Microporous volume and median micropore diameterwere calculated by using the Hovarth–Kowazaemethod. The main properties of the adsorbent aresummarized in Table 1. No treatment was given to theAC prior to being used for these experiments.

Adsorbate

Toluene (99.99%), obtained from Fisher Scientific, wasused in this study. The main characteristics of this VOCare presented in Table 2. This compound was chosenbecause it is systematically detected in houses.

Analysis

Very sensitive analytical methods are required in orderto detect compounds at a low concentration range, as in

the Henry’s law region. It is thus necessary to precon-centrate the sample before any analysis. Cryogenicpreconcentration by the Entech 7100A () was chosen.Then, the sample was analysed by GC–MS (Trace GCTrace MS Plus, Thermofinnigan, ), which is a veryprecise analytical technique for the quantification ofwell-identified compounds. Implementations of thesetechniques are not trivial and require much attentionand validation.

Batch experiments

Adsorption isotherm and kinetics were both determinedby using a simple constant volume method [13],described below, which is in fact a mass balancemethod at constant total system volume. The experi-mental set-up represented in Figure 1 was used forbatch experiments. These experiments were performedat a regulated fixed temperature (25

±

1

°

C), by meansof a glass reactor (23 L) placed in an oven. Before eachexperiment, the reactor was flushed with high purity airproduced by a zero air generator (Air Liquide, ) untiltoluene concentration reached a value lower than thedetection limit. The characteristics of the zero air werea maximal total concentration of organo-compounds of50 ppb and a dew point of

−

40

°

C. Then the reactor wasclosed and polluted by a liquid toluene injection(ranging from 20 to 50

µ

L) through a septum. Theevaporation and the homogenization with a pump led toan initial concentration of about 0.7 to 1.9 g m

−

3

. Adetermined mass of adsorbent (close to 1 g) was thenintroduced. The concentration of toluene in the gasphase was determined using the analytical techniquesdescribed previously. The corresponding load oftoluene in the adsorbent was calculated from a massbalance on the reactor.

Figure 1. Schematic diagram of batch and fixed-bed experiments.

Since the analytical method was destructive, onlyone analysis could be done for each experiment. There-fore, each experiment, characterized by a volume ofinjected toluene and a mass of AC, was made severaltimes and analysed at different times. The experimentaldata obtained for the longest time, which should be onthe same straight line, would be considered as equilib-rium. The other experimental data gave the evolution oftoluene concentration in the gas phase with time andwould allow the determination of adsorption kinetics.

Fixed bed experiment

A fixed bed experiment was carried out at semi-pilotscale, as described by Carratala-Abril

et al.

[17]. Theexperimental set-up is shown in Figure 1. The fixed bedexperiment was carried out in a glass column having adiameter of 25 mm and a packed length of AC of1.2 cm. High purity air, produced by a zero air

Table 2. Main characteristics of toluene.

Formula C

7

H

8

Bulk density 0.865 kg L

−

1

Molar mass 92.14 g mol

−

1

Vapour molar volume 316 cm

3

mol

−

1

Vapour pressure (20

°

C) 3.80 kPaKinetic diameter 5.8 ÅDipolar moment 0.4 D

Table 1. Main characteristics of AC.

Furnished values

Values determined with

ASAP 2010

Particle size 1.4 to 4 mmBulk density 0.5 kg L

−

1

BET surface area 1000–1100 m

2

g

−

1

990 m

2

g

−

1

Microporous volume 0.393 cm

3

g

−

1

Median pore diameter 5.8 ÅPZC value 10/11

AQ1

AQ1

AQ1

TENT_A_512297.fm Page 2 Monday, September 27, 2010 6:03 PM

Environmental Technology

3

5

10

15

20

25

30

35

40

45

50

55

generator, flowed through a calibration gas generatorto produce an air polluted by toluene. The effluent fedthe fixed bed at a constant empty bed velocity of 6.3 cms

−

1

, leading to an empty bed residence time of 0.2 s. Itsconcentration was 733

µ

g m

−

3

during the first 258 hoursand was increased up to 1780

µ

g m

−

3

from this time,owing to experimental constraints. All experimentalcharacteristics are summarized in Table 3. The analyti-cal techniques used to quantify the air leavingthe adsorption column were similar to those used for thebatch experiments. The GC–MS broke down after the1800th hour leading to a lack of experimental measuresbefore the last data were obtained after its repair.

Theory

Adsorption kinetic and equilibrium

Adsorption can be described by a pseudo-second orderreaction rate law described by Thomas’s model [18,19]:

where is the average concentration of toluene in theadsorbent, C is the concentration of toluene in the gasphase, q

max

is the maximal adsorption capacity of theadsorbent, k

1

is the adsorption rate constant and k

2

is thedesorption rate constant.

It should be noted that the rate constant k

1

is not theintrinsic rate constant because most porous adsorbentssuffer from mass transfer limitations. It is a global rateconstant which contains the effects of both intrinsickinetics and internal and external mass transfers. Themodel is completed with the mass conservation equa-tion written below:

where V is the volume of the batch reactor and m

AC

isthe mass of AC introduced in the reactor.

Initial conditions for adsorption in batch mode are:

where C

0

is the initial concentration in the gas phase.Thus the global parameters k

1

and k

2

could be deter-mined by solving the model constituted by Equationsand , matching the solution to the kinetic experimentaldata: the set of equations was solved by the Runge–Kutta–Merson method, and the global parameters k

1

and k

2

were identified by minimizing a relative leastsquare criterion between experimental and calculated

dqdt

k C q q k q= − −1 2( )max (1)

q̄

dqdt

Vm

dCdtAC

= − (2)

t q and C C= = =0 0 0, (3)

Table 3. Adsorption bed and packing characteristics.

Temperature 293 KRelative humidity 0%Inside diameter of the column 25 mmGas flow rate 1.85 L min

−

1

Bed length 1.2 cmPacking amount 3.0 gBed void fraction 0.5Empty bed velocity 6.3 cm s

−

1

Empty bed residence time 0.2 sFed concentration 733–1780

µ

g m

−

3

Figure 1. Schematic diagram of batch and fixed-bed experiments.

AQ

TENT_A_512297.fm Page 3 Monday, September 27, 2010 6:03 PM

4

A. Réguer

et al.

5

10

15

20

25

30

35

40

45

50

55

data, by a Levenberg–Marquardt algorithm. It must benoted that the value of the parameter q

max

has to beknown.

At equilibrium this model leads to the well-knownLangmuir equation [20]:

where q

eq

is the concentration of toluene in the adsor-bent at equilibrium, C

eq

is the concentration of toluenein the gas phase at equilibrium and b = k

1

/k

2

.So at equilibrium Thomas’s model results in the

Langmuir isotherm, assuming a monolayer adsorptionof non-interacting particles on a homogeneous surface[20,21].

At very low concentrations, the amount of tolueneadsorbed on the AC is much lower than the maximaladsorption capacity and so can be neglected. ThusEquation leads to the following expression:

where k

1app

is the product of k, and q

max

, is the apparentglobal adsorption rate constant.

At equilibrium, this expression leads to Henry’s law(which is the limit of the Langmuir formulation whenthe gas phase concentration becomes very low):

So, experimental data, which have reached equilibriumin the Henry’s law region, have to be on a straight linewith a slope equal to the Henry coefficient (H).

Breakthrough curve

We consider an isothermal column packed with AC,through which a gas polluted by toluene flows with aconstant linear velocity. Since we have chosen to considera global concentration of the adsorbent, as we have madefor the batch experiments, the differential mass balanceon the column in the gas phase is given by [22]:

where u is the interstitial velocity, D

ax

is the axialdispersion coefficient,

ρ

AC

is the density of AC,

ε

b

is thebed void fraction, and z is the axial coordinate with theorigin at the column inlet.

The term represents the local rate of adsorption

between the gas and the adsorbent phases. The use of anequilibrium model would assume that would be linkedto C by the adsorption equilibrium isotherm (Langmuir,for example). We have preferred to assume the adsorp-tion to be a non-equilibrium process, and once again wehave chosen to describe it with the Thomas equation.The main difference between this and Equation lies inthe fact that C and are now local variables.

The variables C and were discretized and the spatialderivatives were represented by finite differences. Theset of partial differential equations was then convertedinto a set of ordinary differential equations, which weresolved by the Gear method (ode15s in Matlab software).

The initial and boundary conditions for the column,initially free of toluene and subjected to a step changein toluene concentration at the column inlet at time zero,are given by:

where C

f

is the concentration of toluene at the inlet andL is the length of the column. The axial dispersion coef-ficient D

ax

can be determined by the intermediate of thePeclet number, Pe, for isothermal operation and lowReynolds number [23]:

with

where Re is the Reynolds number and Sc is the Schmidtnumber.

Hence, adsorption isotherm experiments at highconcentrations (performed in a previous study) lead tothe maximal adsorption capacity of the adsorbent (q

max

).Adsorption isotherm experiments at very low concentra-tions, conducted in the present study, lead to the Henryadsorption coefficient (H). The kinetic parameters k

1

and k

2

are generated by fitting a kinetic expression to

q

bq C

bCeq

eq

eq

=+( )

max

1(4)

dq

dtk C k qlapp= − 2 (5)

qk

kC HCeq

lappeq eq= =

2

(6)

∂∂

= −∂∂

+∂∂

−− ∂

∂C

tu

C

zD

C

z

q

tax ACb

b

2

2

1ρ

εε

(7)

∂∂

q

t

q̄

q̄

∂∂

= − −q

tk C q q k q1 2( )max (8)

q̄

t C q= = =0 0 (9)

z

D

u

C

zC Cax

f=∂∂

= −0 (10)

z L

C

z=

∂∂

= 0 (11)

D

ud

Peaxp= (12)

PeSc

Sc

b

b= +

+

0 73 0 5

19 5

.

Re

..

Re

εε

(13)

TENT_A_512297.fm Page 4 Monday, September 27, 2010 6:03 PM

Environmental Technology

5

5

10

15

20

25

30

35

40

45

50

55

batch experimental data. Finally, q

max

, k

1

and k

2

are usedin the model of the dynamic system, and, to validate it,a comparison is made between the model and an exper-imental fixed-bed experiment.

So, the correlative modelling of batch experimentsmight ensure a predictive modelling of dynamic exper-iment if the kinetic coefficients k

1

and k

2

determinedwith batch experiments could be employed in thedynamic model.

Results and discussion

Determination of the parameter q

max

As has been already pointed out, the value of theparameter q

max

has to be known because of the choiceof Thomas’s equation to represent the adsorption–desorption kinetics. Even though this work deals withindoor air concentration levels, the determination ofq

max

needs the adsorption isotherm at 298 K withconcentrations in the gas phase at much higher valuesthan those of indoor air. Now, adsorption equilibria oftoluene on AC C1220-G90 have been previouslyconducted in our laboratory [24]; therefore q

max

hasbeen deduced from these experiments. In fact, experi-mental isotherms were obtained at four differenttemperatures (303, 308, 313 and 318 K), with concen-trations in the gas phase ranging from 300 to 7700 mgm

−

3

and they were then fitted with different models.Indeed, the Langmuir isotherm is useful to describemonolayer adsorption, but, since Langmuir assump-tions are not always fulfilled, some other empiricalformulae such as Langmuir–Freundlich or Toth equa-tions, which are generalized forms of the Langmuirequation, have been introduced [21,25]. The Tothequation, which provided the best results in the previ-ous work, is described below:

K

T

can be seen as a thermodynamic equilibriumconstant for the adsorption process and, according toTerzyk

et al.

[26], the parameter

τ

characterizes thesystem heterogeneity. The parameters q

max

and K

T

weredescribed as functions of temperature by empiricalformulations [24], and then the adjustable parameterswere identified by fitting the calculated values to the setof experimental data for the four temperatures. In viewof the good accuracy of this model, we extrapolated itto the temperature of 298 K, which led to a maximaladsorption capacity of 445 mg g

−

1

for toluene on ACC1220-G90.

For

τ

= 1, the Toth isotherm is reduced to the Lang-muir isotherm, assuming a monolayer adsorption of

non-interacting molecules on a homogeneous surface.It can be noted that the value of

τ

was 1.18. Accordingto Terzyk

et al.

[26], if this parameter is lower thanunity, the system is heterogeneous and, if the parameteris greater than unity, like in the present case, the lateralinteractions between the adsorbed molecules aregreater than the adsorptive potential. Moreover, since atindoor air concentration levels there are probably fewinteractions between the few adsorbed molecules, theuse of Thomas’s model and, implicitly, the Langmuiradsorption equilibrium is then justified at very lowconcentration levels.

Determination of kinetic parameters

The kinetic parameters k

1

and k

2

were determinedindependently by fitting the kinetic model to the exper-imental data obtained in batch experiments. The resultsare shown in Figure 2, which represents the evolutionof the concentration of toluene in the gas phase withtime, for different initial concentrations. The modelledkinetic curves follow the experimental data quite well.Values of the kinetic parameters are presented inTable 4.

Figure 2. Adsorption kinetics of toluene on AC.

In Table 4, a comparison is made with other studiesdealing with the determination of kinetic parameters. Itcan be noted that the value of the global adsorption rateconstant obtained in our study is of the same order asthe parameter obtained by Chuang

et al

. [27] for theadsorption of benzene on GAC. The global desorptionrate constant is, however, much lower in our case thanin this study.

Lorimier

et al

. [28], who studied the adsorption oftoluene on activated carbon fibre cloths and felts, deter-mined the kinetic lumped parameter of the LDF model(k

LDF

). In the Henry’s law region, this parameter is simi-lar to the desorption rate constant of Thomas’s model(k

2

). The parameter determined by Lorimier

et al

. is,however, much higher than ours.

The comparison of our study with these two studiesleads to the conclusion that, at very low concentrations,the desorption phenomenon seems to be less importantthan for high concentrations. Indeed, these studies dealtwith concentrations higher than in the present study. Thetoluene concentration range is 21 to 18,160 mg m

−

3

inthe case of Lorimier

et al

., and the benzene concentrationrange is of 624 to 11,700 mg m

−

3

in the case of Chuang

et al

.

Determination of the Henry coefficient

Lots of studies dealing with the adsorption of toluene onAC are available in the literature [13,29–32]. However,to our knowledge, very few studies have dealt withadsorption isotherms of VOCs measured in the Henry’s

q qK C

K C

T

T

=

+[ ]max

( )11

τ τ

(14)

TENT_A_512297.fm Page 5 Monday, September 27, 2010 6:03 PM

6

A. Réguer

et al.

5

10

15

20

25

30

35

40

45

50

55

law region. In fact, as we have emphasized previously,determination of the adsorption isotherm at very lowconcentrations is technically difficult.

The adsorption isotherm of toluene on AC,presented in Figure 3, was investigated at a regulated

temperature of 25 °C (±1 °C). The curve obtained forconcentrations characteristic of indoor air pollutionlevels is typical of Henry’s law.Figure 3. Adsorption isotherm of toluene on AC in the Henry’s law region.Experimental equilibrium data were compared withthe calculated equilibrium obtained with the kinetic

Figure 2. Adsorption kinetics of toluene on AC.

Table 4. Parameters determined with the kinetic model and comparisons with other studies.

Thomas model LDF model

Equations

This formulation implies a Langmuir isotherm. This formulation can be used with several models for the adsorption isotherm (Langmuir, Freundlich etc.).

In the Henry’s law region, the Thomas equation leads to a similar formulation to the LDF model:

Then kLDF can be compared to k2 in the Henry’s law region.

Kinetic results Present study (toluene, 25 °C, GAC)k1 = 3.36 × 10−5 m3 g−1 s−1

k2 = 4.19 × 10−8 s−1

Chuang et al. [27](benzene, 30 °C, GAC)k1 = 2.1 × 10−4 m3 g−1 s−1

k2 = 3.35 × 10−5 m3 g−1 s−1

Lorimier et al. [28] 2005 (toluene, 20 °C, AC fibre cloths and felts)

kLDF = 3.15 × 10−4 to 5.79 × 10−4 s−1

Results depend on the initial concentration in the batch reactor and on the medium used.

dq

dtk C q q k q= − −1 2( )max

dq

dtk q qLDF= −( )*

dq

dtk Cq k q k q q= − = −( )1 2 2max

*

AQ2

AQ

TENT_A_512297.fm Page 6 Monday, September 27, 2010 6:03 PM

Environmental Technology 7

5

10

15

20

25

30

35

40

45

50

55

parameters k1 and k2. As this kinetic is described byEquation , the curve obtained is not a straight line;indeed it is a Langmuir isotherm. However, as shown inFigure 3, there is not a great difference between thecurve and the straight line obtained by linear regression.The kinetic described by Equation leads to a similarresult, but is less accurate than Equation (1). Thisconfirms that experimental data are close to the Henry’slaw region.

The Henry coefficient obtained is equal to 356 m3

gAC−1. Our value is slightly under those of Qi and LeVan

[14] (Figure 3). Two reasons can explain this differ-ence: Qi and LeVan used a different AC, made fromBPL, and they regenerated this at 150 °C before thebeginning of the experiments.

Predictive modelling of breakthrough behaviour

When the sorption process is controlled by surface reac-tion and internal transfer, kinetic parameters determinedwith batch kinetic experiments can be used to simulatethe behaviour of a fixed-bed experiment [19,33]. So, tosimulate an experimental breakthrough curve, we usedkinetic parameters determined previously. Since therewas a significant deviation between model predictionand experimental data, we determined the best kineticparameters for the experimental breakthrough curve. It

appeared that these kinetic parameters led to the sameHenry coefficient, but they were around 24 times higherthan those determined with batch experiments. Theglobal adsorption rate constant determined for thefixed-bed experiment is however always in the range ofthose determined by Chuang et al. [27] (cf. Table 4).

Figure 4 shows different modelled breakthroughcurves obtained with the same Henry coefficient anddifferent kinetic parameters: the legend only mentionsk1, since k2 is linked to k1 by the Henry coefficient.When use is made of the parameters of the batchexperiments (k1 = 3.36 × 10–5 m3 g−1 s−1), there is animmediate leak of pollutant, because the adsorption rateis too low. By increasing k1, the modelled curveapproaches the experimental data, because the adsorp-tion rate becomes sufficiently high to allow the adsorp-tion process during the residential time in the fixed bed.When k1 becomes very high, the behaviour of the break-through curve is similar to the one obtained assumingthat instant equilibrium is reached in the bed betweenthe bulk flow and the adsorbed phase.Figure 4. Modelled and experimental behaviour of the adsorption column during the fixed-bed experiment.The deviation between the identified values of thebatch experiments and those of the dynamic experimentcan be explained in different ways.

The first explanation is that the external transferresistance (1/kC) was lower in the fixed-bed experimentthan in the batch experiments, which led to a global

Figure 3. Adsorption isotherm of toluene on AC in the Henry’s law region.AQ3

AQ

TENT_A_512297.fm Page 7 Monday, September 27, 2010 6:03 PM

8 A. Réguer et al.

5

10

15

20

25

30

35

40

45

50

55

resistance that was 1/k1 lower (and then a globalconductance that was k1 higher). Indeed, even if therewas a circulation of the air in the batch experiments, thevelocity was lower than in the fixed-bed experiment andled to a lower Reynolds number. Hence, the externalmass transfer coefficient, kC, was calculated from thefollowing correlation proposed by Wakao and Funazkri(cited by Murillo et al. [34] and Yang [22]):

The value obtained in the batch experiments was threetimes lower than in the fixed-bed experiment.

The second explanation could be that the kineticparameters determined in the batch experiments shouldnot be used for the fixed-bed experiment because ofthe difference between the concentrations studied forthe adsorption isotherm (40–150 µg m−3) and theconcentration feeding the adsorption column (1780 µgm−3). However, this explanation is not convincingbecause the Henry coefficient seems to be correct in thefixed-bed experiment. To confirm the determination ofkinetic parameters in fixed-bed experiments, it wouldbe interesting to make a breakthrough curve at a

concentration near to indoor air pollution levels.However, as was emphasized by Scahill et al. [35], tomake a breakthrough curve at those levels of concentra-tion could take thousands and thousands of hours.

Thirdly, the value of the kinetic parameters deter-mined in the batch experiments might be in doubtowing to the lack of experimental data before theequilibrium plateau (Figure 2), or perhaps the studiedrange of concentration is too large in the batch experi-ments to expect the same value of k1 from the beginningto the end of the adsorption process.

Conclusions

The first aim of this study was to determine the param-eters of Thomas’s model (the global adsorption anddesorption rate constants), by fitting the model to exper-imental kinetic data obtained in batch experiments ofadsorption of toluene, at indoor air concentration levels,on to activated carbon. This aim was reached.

These kinetic parameters led to the determination ofthe Henry coefficient. The value obtained is in accor-dance with the experimental adsorption isotherm oftoluene on AC obtained in this study at low concentra-tions (between 40 and 150 µg m−3), despite the technicaldifficulty. Moreover, even if there is a difference, which

k d

DSc

C p

m

= +2 1 1 0 6 1 3. Re . / (15)

Figure 4. Modelled and experimental behaviour of the adsorption column during the fixed-bed experiment.

AQ4

AQ5

AQ

TENT_A_512297.fm Page 8 Monday, September 27, 2010 6:03 PM

Environmental Technology 9

5

10

15

20

25

30

35

40

45

50

55

has still to be explained, between the kinetic parametersin batch experiments and in the fixed-bed experiment,the determined Henry coefficient makes the modellingof the fixed-bed experiment possible. Indeed, even ifthe fixed-bed experiment is conducted at a higherconcentration than that of the batch experiments, thisconcentration is very low in relation to concentrationsusually encountered.

To confirm this study, the experimental procedure todetermine the experimental adsorption kinetic ought tobe improved. Also, it would be interesting to make abreakthrough curve in the range of concentration of theadsorption isotherm (40–150 µg m−3).

So, Thomas’s model has been useful to model bothbatch adsorption and fixed-bed adsorption of tolueneon to activated carbon at concentrations close to thoseencountered indoors. The comparison of the break-through curve obtained with an equilibrium model hasshown that the kinetics of the adsorption process hasto be taken into account. Compared with otherdynamic adsorption models, such as the LinearDriving Force (LDF) model, Thomas’s model has theadvantage of determining both adsorption and desorp-tion parameters. This can be very useful, as far asindoor air treatment is concerned, because the knowl-edge of these parameters makes it possible to modelan adsorption column submitting to peaks of pollution(as can happen indoors). Indeed, not only the adsorp-tion phenomenon, but also the desorption which canoccur after a peak, could be modelled by Thomas’smodel.

References[1] J.F. Arenes, C. Cochet, M. Derbez, C. Duboudin, P.

Elias, A. Gregoire, B. Jedor, S. Kirchner, J.P. Lucas, N.Pasquier, M. Pigneret, and O. Ramalho, Campagnenationale logements : Etat de la qualité de l’air dansles logements français, Rapport final, OQAI,, France,2006 (In French).

[2] R.D. Edwards, J. Jurvelin, K. Saarela, and M. Jantunen,VOC concentrations measured in personal samples andresidential indoor, outdoor and workplace microenvi-ronments in EXPOLIS-Helsinki, Finland, Atmos.Environ. 35 (2001), pp. 4531–4543.

[3] C. Yu and D. Crump, A review of the emission of VOCsfrom polymeric material used in building, Build.Environ. 33 (1998), pp. 357–374.

[4] A.T. Chan and M.W. Chung, Indoor—outdoor airquality relationships in vehicle: Effect of drivingenvironment and ventilation modes, Atmos. Environ. 37(2003), pp. 3795–3808.

[5] Y. Sun, L. Fanf, D.P. Wyon, A. Wisthaler, L. Lager-crantz, and P. Strom-Tejsen, Experimental research onphotocatalytic oxidation air purification technologyapplied to aircraft cabins, Build. Environ. 43 (2008),pp. 258–268.

[6] T. Yoshida and I. Matsunaga, A case study on identifi-cation of airborne organic compounds and time courses

of their concentrations in the cabin of a new car forprivate use, Environ. Int. 32 (2006), pp. 58–79.

[7] R. Acevedo, P.G. Parnell, H. Villanueva, L.M. Chap-man, T. Gimenez, S.L. Gray, and W.S. Baldwin, Thecontribution of hepatic steroid metabolism to serumestradiol and estriol concentrations in nonylphenoltreated MMTVneu mice and its potential effects onbreast cancer incidence and latency, J. Appl. Toxicol.25 (2005), pp. 339–353.

[8] M.J. Mendell, Indoor residential chemical emissions asrisk factors for respiratory and allergic effects inchildren: A review, Indoor Air 17 (2007), pp. 259–277.

[9] B. Guieysse, C. Hort, V. Platel, R. Munoz, M. Ondarts,and S. Revah, Biological treatment of indoor air forVOC removal: Potential and challenges, Biotechnol.Adv. 26 (2008), pp. 398–410.

[10] P. Le Cloirec, Les Composés Organiques Volatils(COV) dans l’Environnement, Lavoisier Tec&Doc,Paris, 1998 (In French).

[11] P. Monneyron, M.H. Manero, and J.N. Foussard,Measurement and modelling of single and multi-compo-nent adsorption equilibria of VOC on high-silica zeolites,Environ. Sci. Technol. 37 (2003), pp. 2410–2414.

[12] F. Haghighat, C.S. Lee, B. Pant, G. Bolourani, N.Lakdawala, and A. Bastani, Evaluation of various acti-vated carbons for air cleaning – towards design ofimmune and sustainable buildings, Atmos. Environ. 42(2008), pp. 8176–8184.

[13] D. Kim, Z. Cai, and G.A. Sorial, Determination of gasphase adsorption isotherms – a simple constant volumemethod, Chemosphere 64 (2006), pp. 1362–1368.

[14] N. Qi and M.D. LeVan, Coadsorption of organiccompounds and water vapor on BPL activated carbon.5. Methyl ethyl ketone, methyl isobutyl ketone, toluene,and modelling, Ind. Eng. Chem. Res. 44 (2005),pp. 3733–3741.

[15] M. Yao, Q. Zhang, D.W. Hand, and D. Perram,Adsorption and regeneration on activated carbon fibercloth for volatile organic compounds at indoor concen-tration levels, J. Air Waste Manage. Assoc. 59 (2009),pp. 31–36.

[16] B.J. Schindler, L.C. Buettner, and M.D. LeVan, Transi-tion to Henry’s law in ultra-low concentration adsorp-tion equilibrium for n-pentane on BPL activatedcarbon, Carbon 46 (2008), pp. 1285–1293.

[17] J. Carratala-Abril, M.A. Lillo-Rodenas, A. Linares-Solano, and D. Cazorla-Amoros, Activated carbon forthe removal of low-concentration gaseous toluene at thesemipilot scale, Ind. Eng. Chem. Res. 48 (2009),pp. 2066–2075.

[18] H.C. Thomas, Heterogeneous ion exchange in a flowingsystem, J. Am. Chem. Soc. 66 (1944), pp. 1664–1666.

[19] M.A. Hashim and K.H. Chu, Prediction of proteinbreakthrough behavior using simplified analytical solu-tion, Sep. Purif. Technol. 53 (2007), pp. 189–197.

[20] I. Langmuir, The adsorption of gases on plane surfacesof glass, mica and platinum, J. Am. Chem. Soc. 40(1918), pp. 1361–1403.

[21] F. Rouquerol, J. Rouquerol, and K. Sing, Adsorption byPowders and Porous Solids: Principles, Methodologyand Application, Academic Press, London, 1999.

[22] R.T. Yang, Gas Separation by Adsorption Processes,Imperial College Press, London, 1997.

[23] J.M. Coulson and J.F. Richardson, Chemical Engineer-ing, Vol. 2, 4th ed., Butterworth-Heinemann, Oxford,1996.

AQ6

TENT_A_512297.fm Page 9 Monday, September 27, 2010 6:03 PM

10 A. Réguer et al.

5

10

15

20

25

30

35

40

45

50

55

[24] N. Fernandes, Contribution à l’étude de l’adsorptionsur charbon actif d’un mélange gazeux decomposés organiques volatils (COV), Ph.D. thesis,Université de Pau et des Pays de l’Adour, 2005 (InFrench).

[25] F. Brouers, O. Sotolongo, F. Marquez, and J.P. Pirard,Microporous and heterogeneous surface adsorptionisotherms arising from Levy distributions, Physica A349 (2005), pp. 271–282.

[26] A.P. Terzyk, J. Chatlas, P.A. Gauden, G. Rychlicki, andP. Kowalczyk, Developing the solution analogue of theToth adsorption isotherm equation, J. Colloid InterfaceSci. 266 (2003), pp. 473–476.

[27] C.L. Chuang, P.C. Chiang, and E.E. Chang, ModelingVOCs adsorption onto activated carbon, Chemosphere53 (2003), pp. 17–27.

[28] C. Lorimier, A. Subrenat, L. Le Coq, and P. Le Cloirec,Adsorption of toluene onto activated carbon fiber clothsand felts: Application in indoor air treatment, Environ.Technol. 26 (2005), pp. 1217–1230.

[29] J. Benkhedda, J.N. Jaubert, D. Barth, and L. Perrin,Experimental and modelled results describing theadsorption of toluene onto activated carbon, J. Chem.Eng. Data 45 (2000), pp. 650–653.

[30] M.A. Lillo-Rodenas, D. Cazorla-Amoros, and A.Linares-Solano, Behaviour of activated carbons with

different pore size distributions and surface oxygengroups for benzene and toluene adsorption at lowconcentrations, Carbon 43 (2005), pp. 1758–1767.

[31] M.A. Lillo-Ródenas, A.J. Fletcher, K.M. Thomas, D.Cazorla-Amorós, and A. Linares-Solano, Competitiveadsorption of a benzene-toluene mixture on activatedcarbons at low concentration, Carbon 44 (2006),pp. 1455–1463.

[32] J.H. Yun, D.K. Choi, and S.H. Kim, Equilibria anddynamics for mixed vapors of BTX in an activatedcarbon bed, AIChE J. 45 (1999), pp. 751–760.

[33] H. Bak, O.R.T. Thomas, and J. Abildskov, Lumpedparameter model for prediction of initial breakthroughprofiles for the chromatographic capture of antibodiesfrom a complex feedstock, J. Chromatogr. B 848 (2007),pp. 131–141.

[34] R. Murillo, T. Garcia, E. Aylon, M.S. Callen, M.V.Navarro, J.M. Lopez, and A.M. Mastral, Adsorption ofphenanthrene on activated carbons: Breakthroughcurve modelling, Carbon 42 (2004), pp. 2009–2017.

[35] J. Scahill, E.J. Wolfrum, W.E. Michener, M. Berg-mann, D.M. Blake, and A.S. Watt, A new method forthe rapid determination of volatile organic compoundbreakthrough times for a sorbent at concentrationsrelevant to indoor air quality, J. Air Waste Manage.Assoc. 54 (2004), pp. 105–110.

TENT_A_512297.fm Page 10 Monday, September 27, 2010 6:03 PM