Carbon 43 (2005) 25542563 www.elsevier.

com/locate/carbon

Hydrophobisation of active carbon surface and eect on the adsorption of water

F. Cosnier a, A. Celzard
a

a,*

gin b, J.F. Mare che a, O. Barre `s , G. Furdin a, D. Be

ral, Universite Henri Poincare Nancy I, UMR CNRS 7555, BP 239, 54506 Vandoeuvre-le ` s-Nancy, France Laboratoire de Chimie du Solide Mine b riaux Surfaces et Proce de s pour la Catalyse, Universite Louis PasteurStrasbourg I, UMR CNRS 7515, Laboratoire des Mate ECPM, 25 rue Becquerel, 67087 Strasbourg, France c cole Nationale Supe ralogie, UMR 7569, E rieure de Ge ologie de Nancy, 15 avenue du Charmois, BP 40, Laboratoire Environnement et Mine ` s-Nancy, France 54501 Vandoeuvre-le Received 3 March 2005; accepted 9 May 2005 Available online 23 June 2005

Abstract A technique of surface hydrophobisation has been applied to two microporous carbonaceous adsorbents. A granular active carbon and an activated carbon bre, both formerly chemically treated in order they preferentially present hydroxyl surface functions, were modied by action of vinyltrimethoxysilane (vtmos) in liquid phase. The resulting samples were characterised using sorption of nitrogen, FTIR, XPS and 29Si MAS-NMR spectroscopy, and elemental analysis. Their stability and heat treatment have also been investigated through thermal analysis. The eciency of the hydrophobisation treatment was evaluated by static adsorption of water vapour and vapours of chlorinated volatile organic compounds (VOCs): dichloromethane and trichloroethylene. Grafting of the vtmos and development of a coating of polysiloxane onto the adsorbent induced a modication of the carbon surface but also a partial lling of the porosity. These modications accounted for a decrease of both the amounts of water and VOC adsorbed by the hydrophobised materials. However, water uptakes were found to be much lower than those of the VOCs, evidencing an enhanced selectivity of the hydrophobised adsorbents toward VOCs. 2005 Elsevier Ltd. All rights reserved.
Keywords: Activated carbon; Surface treatment; Adsorption; Surface properties

1. Introduction Activated carbons (ACs) are widely used for the removal of various pollutants from gases, and especially for adsorption of volatile organic compounds (VOCs). In standard conditions, the air to be treated contains water vapour. Because of a competition phenomenon between adsorption of VOCs and water, the presence of the latter may aect, sometimes to a large extent,

Corresponding author. Fax: +33 03 83 68 46 19. E-mail address: Alain.Celzard@lcsm.uhp-nancy.fr (A. Celzard).

adsorption kinetics, capacity and selectivity of a given AC toward a mixture of VOCs [15]. In order to limit such eects, the presence of water on the carbon surface should be avoided as far as possible. It is well known that the hydrophilic surface functions of the ACs play a major role in water adsorption, especially at low relative pressure [6]. Most of the techniques developed for lowering the anity of water for AC act on such surface functions: depending on whether the latter are removed or modied, the so-called physical or chemical methods may be used, respectively. Physical methods are the most conventional; they are usually based on a thermal treatment in a ow of hydrogen or

0008-6223/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2005.05.007

F. Cosnier et al. / Carbon 43 (2005) 25542563

2555

inert gas [7,8], but the removal of the surface functions is seldom complete. Additionally, due to the high temperatures involved, the modication of the pore texture may be a serious problem. In contrast, the chemical methods scarcely described in the literature refer to coatings of organosilicon or uorination by uorocarbon (C2F4 or C2F6) either by plasma surface treatments [9], or through chemical reaction (or impregnation) in liquid or gas phase in the case of organosilicon compounds (usually trimethylchlorosilane) [10,11]. The chlorination of ACs was also tried [12]. All these chemical methods lead to the hydrophobisation of the surface, but most of times associated with a dramatic reduction of the pore volumes, and hence to that of the adsorption properties. In the present work, the hydrophobisation method of Budarin et al. [13] was applied to microporous ACs: a granular and a brous one. The technique was formerly developed by these authors using mesoporous ACs and silica gels, for the preparation of materials having specic catalytic properties, and was found to preserve both structural and pore texture homogeneity. The materials were rst treated in order they preferentially present hydroxyl surface functions. Then, the vinyltrimethoxysilane (vtmos) organosilicon compound was grafted on these new functional groups. The resulting samples were characterised using sorption of nitrogen, FTIR, XPS, 29 Si MASNMR spectroscopy and elemental analysis. Their heat-treatment was subsequently investigated through the same techniques and also with thermal analysis. The eciency of the hydrophobisation treatment was evaluated by static adsorption of water and chlorinated VOCs: dichloromethane and trichloroethylene. Water adsorption isotherms were analysed using appropriate equations for such adsorbates (DubininSerpinsky, DubininAstakhov and modied equation of Do et al.).

present at the surface before the materials can be hydrophobised. In order to increase the density of hydroxyl groups and limit the amount of impurities, the surface of both ACs was softly oxidised by action of 15 mol.% hydrogen peroxide. The materials were next heat treated at 450 C in an inert atmosphere. Doing this, the remaining functional groups are only hydroxyls, carbonyls and related functions (quinones, ceto-enols, etc.), the carboxylic ones and their relatives (esters, lactones and possibly anhydrides) having been eliminated. The nal step of the treatment shall lead to the quasi exclusive presence of hydroxyl functions at the ACs surface. Hence, a reduction by sodium borohydride (NaBH4), allowing the selective reduction of aldehydes and ketones in alcohols, was carried out. The three aforementioned steps, namely oxidation, heat treatment and reduction, lead to the samples designated NC120R and Fe88R, deriving from the precursors NC120 and Fe88, respectively. The hydrophobisation of the ACs was obtained by action of vinyltrimethoxysilane (vtmos) in toluene at 125 C. The amount of vtmos used depends on the oxygen content of the hydroxylated materials (1.3 mol vtmos per mol of oxygen atom; the latter was estimated by elemental analysis, see below). The hydrophobised samples were dried under evacuation at room temperature, and were termed NC120H and Fe88H. 2.3. Analytical characterisation The carbon, hydrogen, oxygen and silicon contents were determined by the Central Analysis Service of the CNRS (Vernaison, France). The infrared spectra were obtained in transmission mode using a FTIR spectrometer Bruker IFS 55 equipped with a deuterated triglycine sulphate detector. The acquisition time was 2 min and the spatial resolution was 2 cm1 from 4000 to 400 cm1. One milligram of AC sample was mixed with 500 mg of KBr, and pellets were obtained by pressing the blend 5 min under 5 tons cm2, then 10 min under 10 tons cm2. The XPS analysis was carried out with a VG Scientic Multilab apparatus using Al Ka radiation (1486.6 eV). The solid-state NMR spectra of 29Si were obtained through magic angle spinning at 300 MHz (Bruker DSX300, 7.05 T). Thermal analysis was performed with a TG/DTA 9216.18 (Setaram). The heating rate was 2 K min1 from 293 K to 1273 K under argon, at a ow rate of 40 ml min1. 2.4. Adsorption studies The pore textures were investigated through the adsorption isotherms of nitrogen at 77 K using an automatic device Sorptomatic 1990 (Thermo Finnigan). Specic surface areas were derived from the BET calculation method [14]. The DubininRadushkevich

2. Experimental 2.1. Precursor materials Two kinds of AC were used. The so-called NC120 is a granular commercial AC supplied by the company Pica (Vierzon, France), prepared from coconut shells heattreated at 850 C and next activated with steam at the same temperature. The second material, termed Fe88, is an experimental tissue supplied by MessierBugatti (Villeurbanne, France), made of rayon-based carbon bres activated with orthophosphoric acid. 2.2. Modication of the surface chemistry According to the experimental protocol described by Budarin et al. [13], hydroxyl functions should rst be

2556

F. Cosnier et al. / Carbon 43 (2005) 25542563

equation [15] was used to determine the micropore volumes, whereas specic pore volumes were dened as the so-called Gurvich volume, measured at a relative pressure P/P0 = 0.95. Micropore size distributions (MPSDs) were evaluated by the HorvathKawazoe method [16]. Sorption measurements of vapours of water and chlorinated VOCs were carried out at 30 C in a manual gravimetric apparatus of the McBainBakr type, equipped with quartz glass springs. For that purpose, liquid water and VOCs were both kept inside their own vessel, and the vapour in equilibrium was let to enter the adsorption device by discontinuous introductions: the corresponding uptakes were thus measured point to point.

face groups and formation of clusters; (2) cluster growth and eventually coalescence of these clusters at low relative pressure (P/P0 < 0.3); (3) micropore lling by water. Cossarutto et al. [20] derived a simplied equation from this model:   1 x1nH2 O bx ; 3 C C ls 1 N0 1 bx Kl where C is the concentration of water adsorbed at the relative pressure x, Cls the saturation concentration in the micropores, Kl the micropore equilibrium constant, nH2 O the average number of water molecules entering the pores, N0 the amount of water adsorbed on functional groups, and b is a constant. The rst term of the equation corresponds to the lling of the microporosity by the clusters of water, and the second one (Langmuirtype) is linked to the adsorption on primary sites.

3. Theoretical Basing on dierent adsorption mechanisms, several isotherm equations have been proposed for the description of adsorption of water vapour on AC. Some of them have been used in this study and are presented below. Using the concept of the potential eld, Dubinin and Astakhov [17] have established the following semiempirical equation:   nDA  A N N 0 exp ; 1 E where N is the amount of adsorbed water (mmol g1) at temperature T and relative pressure P/P0, N0 is the maximum amount adsorbed in the micropores, A is the adsorption potential (A = RT ln (P/P0)), and nDA and E are temperature-invariant parameters characterising the system. Based on a kinetic approach, Dubinin and Serpinsky [18] assumed that water is initially adsorbed onto primary sites (oxygenated surface functions), then subsequent adsorption occurs on water molecules already adsorbed (thus behaving as secondary sites). The following relationship, holding for 0.3 < P/P0 < 0.8, was proposed: a . P =P 0 ca0 a1 ka 2

4. Results and discussion 4.1. Characterisation of reduced and hydrophobised samples The elemental analysis of various samples of the two AC materials: raw, reduced and hydrophobised samples, is given in Table 1, while Table 2 presents their pore texture parameters and Fig. 1 compares the MPSDs. The eect of the reduction method, leading to the materials which surface was hydroxylated, is rst discussed. It may be seen from Table 1 that the oxygen content of the reduced materials is greater than that in the corresponding precursors. Moreover, XPS data of O 1s and C 1s evidenced that the amount of CO groups after reduction is much higher than that of C@O [21]. These ndings support the eciency of the method of surface modication described above. The latter does

Table 1 Elemental analysis of dry carbon samples Sample NC120 NC120R NC120H NC120H200 NC120H300 NC120H400 Fe88 Fe88R Fe88H Fe88H400 Fe88H600 %C 93.0 95.2 87.7 88.1 87.2 87.2 85.3 86.6 81.1 79.2 72.8 %H 0.6 0.9 0.7 1.6 1.4 1.1 1.1 1.7 2.8 1.4 1.5 %O 2.5 3.1 n.d. n.d. n.d. n.d. 7.4 10.6 n.d. n.d. n.d. % Si 4.0 4.2 4.6 4.8 4.5 4.7 7.5

In Eq. (2), a (mmol g1) represents the amount of water adsorbed at the relative pressure P/P0, a0 (mmol g1) is the concentration of primary sites of the AC, and c is the ratio of the rate constants of adsorption and desorption. The term (1 ka) accounts for the decrease of the number of adsorption sites with increasing micropore lling, while k represents the loss of the secondary sites during adsorption. Another approach modelled by Do and Do [19] assumes that adsorption proceeds in three schematic steps: (1) adsorption of water molecules onto functional sur-

NC120 and Fe88 are raw materials, whereas the suxes R and H stand for reduced and hydrophobised samples, respectively. The numbers after the letter H indicate that the hydrophobised samples were heattreated at the temperature (C) given by the digits.

F. Cosnier et al. / Carbon 43 (2005) 25542563 Table 2 Pore texture parameters calculated from sorption of nitrogen at 77 K Sample SBET (m2/g) 2203 2251 835 993 1103 1567 2168 1353 1286 262 857 958 Total pore volume Vtot (cm3/g) 1.14 1.15 0.46 0.52 0.59 0.85 1.11 0.79 0.70 0.15 0.50 0.56 Micropore volume Vmicro/(cm3/g) 0.83 0.76 0.26 0.32 0.38 0.56 0.86 0.67 0.58 0.13 0.40 0.43 Vmicro/Vtot

2557

NC120 NC120R NC120H NC120H200 NC120H300 NC120H400 NC120H600 Fe88 Fe88R Fe88H Fe88H400 Fe88H600

0.73 0.66 0.57 0.62 0.64 0.66 0.69 0.84 0.83 0.87 0.80 0.77

cursors and reduced materials; only the narrowest micropores (<0.8 nm) really tend to be modied. Concerning the eect of the hydrophobisation treatment, the pore texture is now strongly altered. The pore volumes are drastically reduced, see Table 2. However, the micropore size distributions remain close to those of the reduced samples, except again for the narrowest pores as shown by Fig. 1. Such similar MPSDs suggest that, although leading to a decrease of total pore volume, the vtmos was uniformly grafted at the surface of the materials. Basing on TG and DSC, Budarin et al. showed that vtmos inside the porosity indeed forms a phase on the carbon surface [13,22] but a true grafting phenomenon was not really proved. In order to determine the origin of the pore lling, the surface chemistry of the hydrophobised ACs was investigated by FTIR, RMN 29Si and Si 2p XPS analyses.
1.6 1.5

Absorbance unit (a.u.)

1.4
NC120H400

1.3 1.2
NC120H200

1.1 1 0.9 0.8 0.7 0.6 4000

NC120 NC120H

3600

3200

2800

2400

2000

1600

1200

800

400

Wavenumber (cm-1)

Fig. 2. FTIR spectra of NC120 and its hydrophobised derivatives. The two small peaks pointed out on the gure are especially discussed in the text.

Fig. 1. Micropore size distributions of the two kinds of active carbons investigated: (a) NC120 and its derivatives and (b) Fe88 and its derivatives.

not strongly aect the pore texture, since both BET surface areas and total pore volumes remain almost unchanged after the reduction step (see Table 2). Only the micropore volumes are found to be slightly lower than those of the precursors. Fig. 1(a) and (b), giving the relative proportion of micropores (expressed as a percentage of the total pore volume), does not illustrate noticeable dierences between the pore textures of pre-

Fig. 3.

29

Si MASNMR spectra of NC120H, Fe88H and Fe88H400.

2558

F. Cosnier et al. / Carbon 43 (2005) 25542563

In Fig. 2 are presented the FTIR spectra of NC120 and its hydrophobised derivatives. The spectrum of NC120H comprises two small bands at 727 cm1 and

692 cm1, not observed for NC120; the same dierence was evidenced between the spectra of Fe88H and Fe88 (not presented here). Additionally, the characteristic

H3CO OCH3 OH

OCH3
H2O

H3CO O

OCH3
+ CH3OH + H2O

OH H3CO Si OCH3 OCH3

H2O

OH

OH

H3CO H3CO OH Si O OCH3 OH

and/or

Si OH O O
and/or

Si O O O

H3CO

Si OCH3 OCH3

OCH3
H2O H2O

H3CO

Si O

OCH3

H3CO

Si O

H3CO

Si O

+ 2 CH3OH + H2O

c
H3CO Si H3CO Si O Si O O OCH3 O Si OCH3 O Si O Si O O OCH3

d
Fig. 4. Reaction scheme of vtmos with a hydroxylated AC surface. (a) First condensation of the alkoxide with hydroxyl surface groups, (b) possible resultant grafted structures, (c) subsequent condensation of vtmos leading to a multilayer surface coverage and (d) resultant multilayer polysiloxane.

F. Cosnier et al. / Carbon 43 (2005) 25542563

2559

bands of vtmos are not seen in the spectra of the hydrophobised samples, suggesting that the organosilicon compound was not adsorbed. However, the two small peaks at 727 cm1 and 692 cm1 could correspond to those, shifted, of vtmos, usually occurring at 773 cm1 and 817 cm1 [23]; these bands were attributed to the vibrations of SiO bonds, and to SiC ones within the group SiCH@CH2, respectively. Such a shift of the peaks toward lower wave numbers may account for a possible increase of the rigidity of the molecule, induced by the grafting. Fig. 3 presents the solid-state 29Si NMR spectra of both hydrophobised ACs NC120H and Fe88H at 300 MHz. Due to relaxation problems, the resolution is lower with the granular carbon than with the bre; depending on the samples, one or two peaks may be observed, close to 75 ppm and 85 ppm. Such chemical shifts do not correspond to those of the silicon present in the vtmos (55 ppm) [24], thus conrming that vtmos is not adsorbed on the surface, but has reacted with it. Various grafting possibilities of the vtmos may be conjectured, as detailed in Fig. 4 [25,26]. In a silylation reaction (Fig. 4(a)), water induces the hydrolysis of the alkoxysilane molecules which can next condense with the hydroxyl groups already present. Depending on the surface density of the latter, the binding of the vtmos to the AC surface may be achieved through one, two or three bridging hydroxyls (Fig. 4(b)). If enough water is present, a second silylation reaction with the rst layer is possible (Fig. 4(c)), and even the formation of a multilayered polysiloxane (Fig. 4(d)). In our case, the values of chemical shift obtained by 29Si NMR are near those measured with vtmos grafted onto silica microparticles by two or three bridging hydroxyls [25].

Si 2p 3/2 XPS analyses of NC120H and Fe88H (not shown in the present work) evidence a peak centred on a binding energy close to 102.5 eV, directly attributable to polysiloxane [27,28], whereas vtmos is expected to lead to a peak within the range 106107 eV [27]. Finding the signature of polysiloxane rather than that of vtmos supports the mechanism shown in Fig. 4(d), according to which the condensation of vtmos continues after the rst grafted monolayer was formed. This feature probably explains the partial pore lling observed in Table 2 for the hydrophobised samples. In order to restore the initial porosity as far as possible, and also for investigating the stability of the materials, thermal treatments were carried out, as explained below. 4.2. Thermal treatment of the hydrophobised ACs Budarin et al., from TG/DSC studies [15,22], showed the possibility of thermodesorbing chemical species derived from vtmos. In the present work, the inuence of the thermal treatment on both the pore texture and the surface chemistry of hydrophobised ACs was systematically investigated. The thermogravimetric data of NC120H and Fe88H are shown in Fig. 5. The DTG curves present three well-marked weight loss regions: a rst one ranging from 100 to 400 C, a second one from 400 C to 700 C, and a third one above 700 C. The analysis of the thermodesorbed gases by mass spectrometry evidenced that the species desorbed during the rst weight loss was not the vtmos molecule, since the main peak was observed at m/z = 91 instead of 121 for the vtmos. Such a dierence may be related to the loss of two methyl groups. This is consistent with the NMR

Temperature (C)
0
0 -0.05

100

200

300

400

500

600

700

800

900

1000

Fe88H
-0.1

dTG (%/min)

-0.15
0

100

200

300

400

500

600

700

800

900

1000

-0.2
Weight loss (%)

-5 -10 -15 -20 -25 -30

-0.25

Fe88H

NC120H
-0.3 -0.35

NC120H

Fig. 5. Thermogravimetric analysis of NC120H and Fe88H.

2560

F. Cosnier et al. / Carbon 43 (2005) 25542563

chemical shifts discussed above, since two released CH3 indeed conrm the possibility of grafting the vtmos by two bridging hydroxyls. The second weight loss region corresponds to the evolution of CO2, and no other weight loss was evidenced in this temperature range. As earlier supposed by Budarin et al., carbon dioxide originates from the thermal decomposition of the remaining unreacted organic groups of the grafted alkoxide, which is an usual feature in solgel chemistry [29]. NC120H was submitted to a thermal treatment at different temperatures (200, 300, 400 and 600 C) under nitrogen ow. The corresponding samples were termed NC120H200, NC120H300, etc. According to Table 2, the thermal treatment at increasing temperatures leads to a strong increase of the pore volumes. Beyond 600 C, i.e., after the second mass loss, the pore texture parameters are close to those of the raw precursor, suggesting the destruction of the functionalisation obtained after the hydrophobisation step. Similar observations were made with the Fe88 bre, however the increase of pore volumes was lower, especially from 400 to 600 C. Thus, at 600 C, both total pore and micropore

volumes remain clearly below those of the precursor. For both materials NC120 and Fe88, a thermal treatment at 400 C did not change the MPSDs as evidenced by Fig. 1. To sum up, the MPSDs remained very similar at each step of the processing: reduction, hydrophobisation and thermal treatment. The temperature-induced changes of the surface chemistry are now discussed. The silicon content of the hydrophobised samples is close to 4 wt.%, as seen in Table 1, and increases slightly with the thermal treatment. The observed phenomenon is opposite to what was expected, since the increase of temperature was formerly suspected to decompose and volatilise a fraction of the grafted vtmos. As evidenced by the elemental analysis, no silicon is removed from the surface, however the porosity opens and develops during the heat treatment, which phenomenon was desired. EDX analysis carried out with a SEM also conrmed similar amounts of Si at the surface of the ACs before and after the thermal treatment at 400 C. As a example, the dispersion of the Si element on NC120H and NC120H400 are given for comparison in Fig. 6: no dierence could

Fig. 6. SEM and EDX images of the dispersion of Si on the surface of (a) NC120H and (b) NC120H400.

F. Cosnier et al. / Carbon 43 (2005) 25542563

2561

be found. In contrast, the chemical nature of the silicon present on the surface before and after thermal treatment is dierent. While Si 2p XPS analysis shows similar results for heat-treated and untreated samples (the peak was shifted by only 0.2 eV), their infrared and NMR spectra exhibit clear dierences. As seen in Fig. 2, the intensity of the two small FTIR peaks at 692 cm1 and 727 cm1 decrease with the heat-treatment temperature and disappeared at 400 C. Similarly, the two NMR peaks at 75 ppm and 85 ppm of the Fe88H material vanished after heat treatment at 400 C. All these results suggest that, upon thermal treatment, the grafted polysiloxane transforms in silica xerogel, thus explaining the following facts: unblocking of the porosity, very low modication of the XPS spectra (since the binding energy of Si remains nearly the same [27]), and clear changes in the NMR and FTIR spectra. 4.3. Adsorption of water vapour Fig. 7 shows the adsorption isotherm of water on precursor, hydrophobised and thermally treated (at 400 C) materials; the solid lines are ts of the modied Do model [19,20] to the experimental data points. Such isotherms clearly demonstrate the eciency of the treat50

Amount of water adsorbed (mmol/g)

45 40 35 30 25 20 15 10 5 0 0.0

NC120 NC120H NC120H400

ment of the ACs detailed in the present work. Whatever the precursor, the hydrophobisation makes the adsorption of water occur at higher relative pressures P/P0. Moreover, whatever the value of P/P0, the adsorbed amounts are lower, as detailed in Table 3. These ndings suggest both a modication of the surface chemistry, accounting for the higher P/P0 required for water adsorption, and a change of pore texture, explaining the lower adsorbed amounts at P/P0 = 1. According to Fig. 7 and Table 3, ACs thermally treated at 400 C show an adsorption behaviour intermediate between that of precursors and that of hydrophobised materials. The isotherms were modelled using the DubininSerpinsky (DS), DubininAstakhov (DA) and modied Do equations. The main parameters characterising the surface (already discussed in Section 3) are gathered in Table 4: a0 for DS, E for DA, nH2 O and Kl for Do modied. As expected, all these parameters are strongly affected by the hydrophobisation process. Thus, a clear decrease of the anity of the ACs for water is evidenced by the decrease of the number of surface functions (through a0), of the adsorption energy (E) [30] and of the adsorption kinetics of water in the microporosity (Kl). Besides, the changes of nH2 O are related to the deep modication of the pore texture. After thermal treatment, all the parameters a0, E, Kl and nH2 O tend to increase (see Table 4), evidencing that the hydrophobic character of the materials is partially lost. Consequently, an optimum between high pore volume (obtained at high heat-treatment temperature) and high hydrophobicity (obtained without thermal treatment), should exist. 4.4. Adsorption of VOCs Water adsorption capacities are more sensitive to surface chemistry at low relative pressure, and are more affected by pore texture at high P/P0. Thus, at medium relative pressures of water vapour, it is rather dicult to distinguish clearly the eect of the modication of the surface chemistry modication and that of the pore volume on the observed decrease of adsorbed amounts. In this study, our goal was to limit the harmful inuence of water on the adsorption of VOCs. In order to check the eciency of the hydrophobisation technique developed above, measuring the adsorbed amounts of VOCs before and after the treatments of the ACs was necessary. For that purpose, two very poorly hydrosoluble VOCs, i.e., whose adsorption is strongly aected by water [4], were chosen: dichloromethane (DCM) and trichloroethylene (TCE). Their adsorption was studied on the whole series of materials (precursors, hydrophobised and heat-treated at 400 C); only Fe88H, having a too low pore volume, was discarded. The corresponding isotherms (not presented here) lead to the adsorbed amounts given in Table 5, obtained at a concentration of VOCs of

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

a
30

Relative pressure (P/P0)

Amount of water adsorbed (mmol/g)

25 20 15 10

Fe88 Fe88H Fe88H400

5 0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Relative pressure (P/P0)

Fig. 7. Water adsorption isotherm on (a) NC120 and its derivatives and (b) Fe88 and its derivatives.

2562

F. Cosnier et al. / Carbon 43 (2005) 25542563

Table 3 Amounts of water qH2 O (mmol g1) adsorbed by the dierent ACs and variation of these amounts after the hydrophobisation and subsequent heat treatment, DqH2 O (%), at dierent relative pressures P/P0 P/P0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 NC120 qH2 O 0.6 0.8 0.9 1.2 2.9 11.8 29.0 42.6 43.2 NC120H qH2 O 0.1 0.2 0.3 0.4 0.4 0.6 2.1 9.2 16.6 DqH2 O 80.5 74.2 68.8 68.7 84.7 94.6 92.7 78.4 61.6 NC120H400 qH2 O 0.4 0.6 0.7 0.8 1.0 2.7 10.8 16.7 23.7 DqH2 O 30.2 28.6 28.2 36.6 66.6 77.1 62.7 60.7 45.1 Fe88 qH2 O 0.9 1.6 2.5 7.1 17.3 23.8 26.0 26.7 26.9 Fe88H qH2 O 0.7 1.1 1.4 1.9 3.7 7.2 9.9 11.0 11.4 DqH2 O 28.0 29.7 44.4 73.3 78.6 69.7 61.9 58.8 57.6 Fe88H400 qH2 O 0.6 1.1 1.5 2.4 6.4 13.5 17.7 19.1 19.7 DqH2 O 33.6 31.0 41.5 66.1 62.9 43.3 31.9 28.3 26.8

Table 4 Values of the main tting parameters of the DubininSerpinsky (DS), DubininAsthakhov (DA) and Do modied models for the adsorption isotherms of water vapour at 30 C on precursors and hydrophobised samples DS a0 (mmol/g) NC120 NC120H NC120H400 Fe88 Fe88H Fe88H400 0.60 0.20 0.39 1.38 0.86 0.81 DA E (J) 1147 724 890 1929 1299 1813 Modied Do nH2O 17.8 12.3 13.5 10.0 9.3 11.6 Kl 104.0 35.4 46.9 997.3 103.9 303.8

Table 5 Adsorbed amounts qVOC (mmol g1) of DCM and TCE (at a concentration C0 = 0.5 mol m3) on ACs, and variation DqVOC (%) of these amounts VOC DCM TCE NC120 NC120H qVOC 5.4 8.7 4.2 6.1
3

NC120H400

Fe88 Fe88H400 4.3 4.1 25.9 25.5

qVOC DqVOC qVOC DqVOC qVOC qVOC DqVOC 22.2 4.3 29.9 6.1 20.4 5.8 29.9 5.5

was demonstrated. The surface chemistry of a granular and a brous AC was thus modied through a preliminary hydroxylation and a subsequent grafting of siloxane. The latter reaction was evidenced using a number of spectroscopic techniques (IR, NMR, XPS) and TG coupled with mass spectrometry. Grafting the vtmos and developing a polysiloxane coating on the AC surface was expected to decrease the anity for water. Whatever the relative pressure, the adsorbed amounts were indeed found to be much lower for the modied samples than for the precursors, even after a thermal treatment at 400 C. This phenomenon is due to the modication of the carbon surface but also to the partial lling of the porosity by the polysiloxane. Consequently, adsorption of VOCs also decreased after hydrophobisation. However, water uptakes are now much lower than those of the VOCs. Therefore, such an enhanced selectivity should decrease the unfavourable eect of steam on the VOC adsorption in dynamic conditions.

Acknowledgements C0 = 0.5 mol m . The latter value was chosen because it corresponds to the limit above which the adsorption process is no longer used at the industrial level. Comparing the dierent materials, it may be seen that the VOC adsorption capacities of the hydrophobised ACs are lower than those of the precursors. However, the decrease of the VOC uptakes is much smaller than those of water vapour (see and compare the values of DqH2 O and DqVOC in Tables 3 and 5), especially in the range of relative pressure 0.30.7 typically encountered in the industry. Hence, the hydrophobisation of the materials indeed increased their selectivity toward VOCs. 5. Conclusions In this work, the possibility of hydrophobising ACs for increasing their selectivity toward chlorinated VOCs The authors would like to acknowledge the helpful assistance of Pr. P. Tekely from the NMR Methodology Laboratory of Nancy (France), for NMR experiments and discussion. This research work has been supported by the CNRS and by the PAN (France-Poland Twinning Carbonaceous and Catalytic Materials for Environment), and by the Region Lorraine.

References
[1] Nelson GO, Correia AN, Harder CA. Respirator cartridge eciency studies: VII. Eect of relative humidity and temperature. Am Ind Hyg Assoc J 1976;37(5):2808. [2] Hall T, Breysse P, Corn M, Jonas LA. Eects of adsorbed water vapor on the adsorption rate constant and the kinetic adsorption capacity of the Wheeler kinetic model. Am Ind Hyg Assoc J 1988;49(9):4615.

F. Cosnier et al. / Carbon 43 (2005) 25542563 [3] Yoon YH, Nelson JH. Eects of humidity and contaminant concentration on respirator cartridge breakthrough. Am Ind Hyg Assoc J 1990;51(4):2029. [4] Lodewyckx P, Vansant EF. Inuence of humidity on adsorption capacity from the WheelerJonas Model for prediction of breakthrough times of water immiscible organic vapors on activated carbon beds. Am Ind Hyg Assoc J 1999;60(5): 6127. [5] Lodewyckx P, Vansant EF. Inuence of humidity on the overall mass transfer coecient of the Wheeler-Jonas equation. Am Ind Hyg Assoc J 2000;61(4):4618. [6] Brennan JK, Bandosz TJ, Thomson KT, Gubbins KE. Water in porous carbons. Colloids Surf A 2001;187188:53968. ndez JA, Xia B, Phillips J, Radovic LR. On the modication [7] Mene and characterization of chemical surface properties of activated carbon: microcalorimetric, electrochemical, and thermal desorptions probes. Langmuir 1997;13(13):341421. [8] Polakov NS, Petukhova GA, Vnukov SP, Shevchenko AO. The hydrophobisation of active carbons. Adsorpt Sci Technol 1993;10(14):16571. [9] Wertheimer MR, Klemberg-Sapieha JE, Cerny J, Liang S. Modication of active carbon by hydrophobic plasma polymers: II Fabric substrates. Plasma Polym 1998;3(3):151 63. [10] Nakanishi Y, Honjo K, Honjo T. Surface-hydrophobic activated carbon and its manufacture. EP patent 0765840, 1997. [11] Ryan TA, Sharrock H. Activated carbon treated with siliconcontaining compounds. UK patent 2391224, 2004. [12] Hall CR, Holmes RJ. The preparation and properties of some chlorinated activated carbons, part II. Further observations. Carbon 1993;31(6):8816. [13] Budarin VL, Clark JH, Mikhalovsky SV, Gorlova AA, Boldyreva NA, Yatsimirsky VK. The hydrophobisation of activated carbon surfaces by organic functional groups. Adsorpt Sci Technol 2000;18(1):5564. [14] Gregg SJ, Sing KSW. Adsorption. Surface area and porosity. second ed. London: Academic Press; 1982. 425. [15] Gregg SJ, Sing KSW. Adsorption. Surface area and porosity. second ed. London: Academic Press; 1982. 21827. [16] Horvath G, Kawazoe K. Method for the calculation of eective pore size distribution in molecular sieve carbon. J Chem Eng Jap 1983;16(6):4705.

2563

[17] Stoeckli F, Jakubov T, Lavanchy A. Water adsorption in active carbons described by the DubininAstakhov equation. J Chem Soc Faraday Trans 1994;90(5):7836. [18] Dubinin MM, Serpinsky VV. Isotherm equation for water vapor adsorption by microporous carbonaceous adsorbents. Carbon 1981;19(5):4023. [19] Do DD, Do HD. A model for water adsorption in activated carbon. Carbon 2000;38(5):76773. [20] Cossarutto L, Zimny T, Kaczmarczyk J, Siemieniewska T, Bimer J, Weber JV. Transport and sorption of water vapour in activated carbons. Carbon 2001;39(15):233946. [21] Cosnier F. Study of adsorption and desorption of chlorinated volatile organic compounds on activated carbon. University of Nancy, France, PhD thesis, 2004. [22] Budarin VL, Clark JH, Gorlova AA, Boldyreva NA, Yatsimirsky VK. Chemical modication of activated carbons. J Therm Anal Calorim 2000;62:34952. [23] SDBS (Integrated Spectral Data Base System for Organic Compounds) databank; web page: http://www.aist.go.jp/RIODB/ SDBS/. [24] Douskey MC, Gebhard MS, McCormick AV, Lange BC, Whitman DW, et al. Spectroscopic studies of a novel cyclic oligomer with pendant alkoxysilane groups. Prog Org Coat 2002;45(23): 14557. [25] Derouet D, Forgeard S, Brosse JC, Emery J, Buzare JY. Application of solid-state NMR(13C and 29Si CP/MAS NMR) spectroscopy to the characterization of alkenyltrialkoxysilane and trialkoxysilyl-terminated polyisoprene grafting onto silica microparticles. J Polym Sci Polym Chem 1998;36(3):43753. [26] Yoshida W, Castro RP, Jou JD, Cohen Y. Multilayer alkoxysilane silylation of oxide surfaces. Langmuir 2001;17(19):58828. [27] Wagner CD, Riggs WM, Davis LE, Moulder JF. In: Muilenberg GE, editor. Handbook of X-ray photoelectron spectroscopy. rst ed. PerkinElmer Corporation (Physical Electronics); 1979. [28] Wagner CD, Passoja DE, Hillery HF, Kinisky TG, Six HA, Jansen WT, et al. Auger and photoelectron line energy relationships in aluminium-oxygen and silicon-oxygen compouds. J Vac Sci Technol 1982;21(4):93344. [29] Brinker CJ, Scherrer GW. Solgel science. Academic Press; 1990. [30] Pires J, Pinto ML, Carvalho A, De Carvalho MB. Assessment of hydrophobichydrophilic properties of microporous materials from water adsorption isotherms. Adsorption 2003;9(4):3039.