You are on page 1of 11

PII:

Wat. Res. Vol. 32, No. 11, pp. 34143424, 1998 # 1998 Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain S0043-1354(98)00109-2 0043-1354/98 $19.00 + 0.00

RELATIONSHIP BETWEEN CHEMICAL AND PHYSICAL SURFACE PROPERTIES OF ACTIVATED CARBON


M M F. JULIEN*, M. BAUDU* and M. MAZET*

Laboratory of Aquatic and Environmental Sciences, 123 avenue Albert Thomas, 87060 Limoges, France (First received January 1997; accepted in revised form February 1998) AbstractThe relationships between structural (chemical or physical) and more general properties such as the surface electrokinetic potential (zeta potential or isoelectric point) or spreading pressure of water were investigated for three powder activated carbon (PAC) samples (manufactured from woods, coconut shell and coal). The PAC samples were characterized by physical parameter measurements (such as zeta potential, specic surface area, spreading pressure) and chemical parameters; surface functional groups and structural mineral elements. Chemical studies show that all PAC samples contain large quantities of mineral ions, mostly calcium, sulfate and phosphate ions. These groups inuence surface properties in the same way as do acidic or basic organic surface functional groups. The thermal treatment of activated carbon decreases acidic surface functional groups and this reaction accompanies an increase in the electrokinetic potential. The ``grafting'' of a cationic polymer adds positive charge to activated carbon surfaces, but the quantity of surface functional groups is also modied by this process. Adsorption isotherms of water in activated carbon enable us to estimate initial polar adsorption sites. The localised adsorption of water on activated carbon is a function of the number of mineral ions and surface functional groups. Quantitative adsorption of water is related to the microporous volume. # 1998 Published by Elsevier Science Ltd. All rights reserved Key wordspowder activated carbon, chemical structure, surface functional groups, zeta potential, spreading pressure

INTRODUCTION

The ability of activated carbon to adsorb micropollutants is often estimated experimentally without anticipating a relationship between the chemical structure and adsorption capacity. Specic surface area, adsorption tests, are commonly used to characterize a given activated carbon. However, a knowledge of the activated carbon structure is necessary if we wish to adapt the product for micropollutant adsorption. The structure of commercial activated carbon products can be modied by chemical or physical processes. These procedures change the physicochemical structure of activated carbon, i.e. the specic surface area, porosity and surface functional groups (nature and number). The quality and quantity of ions released into water by powder activated carbon (PAC) can modify the adsorption mechanism. Indeed, some ions can coadsorb when they encounter organic molecules near the surface of the carbon. For example, calcium ions modify the adsorption of some organic compounds (Lafrance et al., 1988; Cannon et al., 1993). Changes in the structure of activated carbon can be evaluated by measuring global physical parameters such as zeta potential (Wnek and Davies,
*Author to whom all correspondence should be addressed.

1987) or spreading pressure (Dorris and Gray, 1979; Papirer et al., 1991). Zeta potential is a physical parameter which enables us to quantify the electrical potential of the solid particle surface. It is determined using the equation of Helmotz Smoluchowski and enables us to determine the acidic or basic character of the activated carbon surface (Snoeyink and Weber, 1967) and to study the molecule adsorption mechanism by analysis of surface potential variations. A decrease in the carbon/water interaction potential may also be indicated by drop in the surface energy expressed by the spreading pressure (p0). By measuring the spreading pressure, we can quantify the hydrophilic character of the activated carbon surface. Papirer et al. (1989) have shown that the quantity of adsorbed water on grafted silica is a function of the number of hydroxyl groups present on the surface. Oxidative chemical treatment of activated carbon by nitric acid signicantly increases the specic surface area of activated wood carbon (Mazet et al., 1994). Barton et al. (1984) reported that oxidation increases the surface reactivity and the adsorption of water vapor. Noh and Scharwz (1990) claim that nitric acid treatment increases the quantity of acidic surface functional groups to the detriment of basic functions and the zeta potential evolves with conditions of the process.

3414

Surface properties of activated carbon

3415

The thermal process is essentially marked by a decrease in the number of acidic surface functional groups in relation to a release of CO and CO2. Mazet et al. (1994) observed both an increase in the zeta potential and a modication of the chemical structure of activated carbon particles during thermal treatment. Gajardo et al. (1991) and Preston et al. (1990) grafted a cationic polymer onto glass particles to obtain a positive surface charge. We applied three treatments to industrial activated carbon, thermal treatment, chemical oxidation and grafting of polymer, to obtain activated carbon with dierent characteristics in order to study the relationships between isoelectric point (I.E.P.), spreading pressure and the chemical structure.
EXPERIMENTAL

positive charges on the nitrogen atom (-CH2CH2N+H2, pK = 8.0) (Preston et al., 1990; Gajardo et al., 1991). The binding (or grafting) must be done with a large enough quantity to obtain a homogeneous distribution of the polymer chains on the material surface. 10 g of activated carbon B1 (coconut) were put in contact with 100 ml of a 1 or 2% (in mass) polymer solution in an agitated bottle for 48 h. PAC (labeled B1/grafted) was then rinsed several times in distilled water (minimum volume 4 l) then ltered, dried (1051158C) and stored. Physicochemical structure of activated carbon General structure. The specic surface area of PAC was measured with a discontinuous volumetric measurement apparatus, Micrometrics Models 2000M V1.01, using nitrogen or argon adsorption at 77.3 K (temperature of liquid nitrogen). Outgassing was done under primary dynamic vacuum at 1508C for 12 h during which time the mass decreased approximately by 4%. Adsorbed quantities were calculated using the mass measured before outgasing. Ash content was determined by gravimetric analysis. 10 g of PAC were heated to 7008C for 17 h and ashes were weighed after cooling with an accuracy of 102 mg. For a better knowledge of the nature of the mineral components, an XRD study of PAC ashes was done before and after treatment at 10008C for 24 h. The powder was crushed and placed on a aluminum plate. The surface was leveled and XRD spectra were obtained for angles 2y from 8 to 708, with a 0.048 step. X-rays were produced by a tube with a copper anticathode (l = 1.54 A). The ionic conductivity and pH were measured in activated carbon rinse water obtained after ltration (0.3 mm membrane) of a suspension of 2 g l1 of PAC stirred for 6 h. The quantitative determination of the main mineral elements making up PAC was done by ionic chromatography (Dionex DX 100) on two types of samples: PAC rinse water (free elements) and solutions of mineralised ashes of PAC. The free element determination was done by leaving a suspension of PAC in distilled water (4 g l1) under vigorous agitation of 24 h. After settling, samples were ltered through a 0.3 mm membrane and analysed. To determine the amount of each element in the structure of activated carbon, 10 g of PAC ashes were mineralized, treated with 50 ml of a chlorhydric and nitric acid mixture and diluted by half. The solution was ltered and then diluted in 100 ml of distilled water. The pH was brought to 2 by adding HCl so as to avoid metallic/metal hydroxide precipitation. Metallic and non-metallic elements were determined by ion chromatography (anions: Cl, NO3, PO3 4 and SO2 and cations: Na+, K+, Mg2+, Ca2+, Mn2+, 4 2+ Fe , etc.). Quantitative determination of metallic cations implies a complexation. The eluant for manganese was PAR (4pyridyl(2) azo resorcine mononitrium monohydrate) and oxalic acid. For the other metals the eluant was a mixture of PAR and PDCA (acid 2.6 pyridine dicarbox-

Choice and modications of raw material Among the many activated carbons manufactured, we chose three commonly used raw activated carbon samples of dierent origin, coconut, wood and fossil coal. Powder activated carbons have a grain size of between 0.4 and 50 mm, with an average diameter of 1035 mm. All experiments presented in this work have been carried out using the same pool of activated carbon. The main characteristics of these PAC are given in Table 1. Oxidation by nitric acid was obtained by putting 50 g of PAC in contact with 400 ml of an aqueous solution of nitric acid 1/1 (Mazet et al., 1994). The mixture was then heated while stirring for 1 h after the appearance of the rst nitrous fumes. After cooling and settling of the PAC, the excess acid solution was eliminated by ltration. The residue was rinsed with a soda solution (0.1 N) until the distilled rinse water had a pH of 56. The PAC was then dried (105 1158C) for 72 h and stored in closed bottles (Julien et al., 1994). For the heat treatment, 10 g of PAC were placed for 1 h in a metallic reactor heated to 10008C. The temperature must not exceed 11008C because above this temperature the specic surface area can decrease signicantly (Rao et al., 1992). The metallic reactor was swept by a nitrogen ow (40 l h1) during heating and cooling. Modied PAC were stored in closed bottles. In order to study the inuence of the PAC surface electrical potential on adsorption of organic molecules, a cationic polymer was grafted on the surface by adsorption. The polymer, polyethylenimine (Sigma, MM 50000), has

Table 1. Physicochemical characteristics of non-treated activated carbon Characteristics Sample (non-treated) A1 wood water vapor 640 2 14 <50 3.9 770 B1 coconuts shell water vapor 947 2 24 <50 10.8 1042 Pore volume (cm3 g1) 0.2295 0.2948 0.0336 0.0795 6.1 6.9 2.2 3.5 1.56 1.20 0.01 0.25 C1 fossil coal water vapor 1047 2 5 <50 8.3 1336 0.2889 0.2373 6.6 3.5 0.61 0.23

Reference Origin Activation Specic surface (argon) (m2 g1) Size (mm) Total ash (%) Iodine ASTM (mg g1) 010 A >10 A Average pore width (A) I.E.P. Basic groups (meq g1) Acid groups (meq g1)

3416

F. Julien et al. lowering of surface energy as a result of water vapor adsorption (Boehm et al., 1966). It is determined by integration of Gibbs' equation: RT 1 a dPaP0 1 p0 gs gsv SBET 0 PaP0 with p0 as the spreading pressure of the adsorbate (mJ m2); gsv as the surface free energy of the solid/vapor interface (mJ m2); R as the ideal gas constant (8.314 J K1 mol1); T as the absolute temperature (293 K); SBET as the specic surface area of PAC (m2 g1); a as the quantity of adsorbed water onto PAC (mmol g1) at P and P/P0: the relative vapor pressure of water. The initial step of water molecule adsorption onto activated carbon implies H bonding with specic adsorption sites. It is possible to determine the initial site inuence on adsorption and thereby the polarity of the accessible surface of activated carbon by using the equation proposed by Dubinin et al. (1955). a a0 c h 1ch 2

ylique). The detection was done at 520 nm with a special column [IPIC CS5 PINE 37028 (precolumn CG5)]. Aluminum in the structure or on the surface of activated carbon was measured using a colorimetric method with Chromazurol S which forms a red complex in acetate/acetic buer (pH = 4.75) in the presence of aluminum. Ascorbic acid must be added to avoid interference with Fe2+. The measurement was done using UV-visible spectrophotometry (Shimadzu UV 160) at l = 547 nm. The content of the surface functional groups was studied by thermogravimetric analysis and IR analysis of gas emitted by activated carbon and enabled us to determine the consequences of a thermal treatment on the structure of the activated carbon. Thermogravimetric analyses were done with a Perkin Elmer (TGA7) system. After placing the PAC sample in the oven, all of the apparatus was swept with nitrogen. The temperature gradient was 108C per min and isothermal phases were maintained periodically in order to proceed to the emitted gas I.R. analysis (TR61R interface 1700X Perkin Elmer). Quantitative determination of the acidic surface functional groups of the various activated carbons was done according to the protocol established by Boehm et al. (1966). 1 g of dried PAC was put into 50 cm3 and of basic solutions of 0.1 N: NaHCO3, Na2CO3, NaOH, NaOC2H5 and stirred for 72 h. The suspension was then ltered and the excess basic solution was titrated with a HCl 0.1 N solution. All acid titrations were done with an automatic Schott Gerate TR 154 titrator and compared to reference solutions. Titration of the basic surface functional groups, chromene and pyrone (Garten et al., 1957; Boehm et al., 1966) was done by placing 1 g of PAC into 50 ml of 0.1 N HCl solution under inert atmosphere (nitrogen). After 72 h of stirring and then a ltration, the excess of acid was titrated. In these two cases, a distinction was made between surface functional groups xed on the activated carbon and those that could be freed by simple contact with distilled water. Estimated errors of the quantitative determination of acidic or basic surface functional groups were about 5%. Physical activated carbon properties Zeta potential. Surface functional groups were also studied by measuring the electrokinetic potential (zeta potential) of activated carbon particles. Zeta potential (z) is based on the displacement velocity of a particle in an electrical eld. A Pen-Kem Lazer Zee Meter model 500 zetameter was used with suspensions of PAC (50 to 100 mg l1) in distilled water. 25 ml of a homogeneous suspension were used for each measurement. Tests were carried out at room temperature (18228C) but expressed at 208C. The tension applied on the electrophoretic cell (100 V) was constant (value recommended for suspensions having a conductivity lower than 1 mS cm1). The precision was about 5%. Thermodynamic constants. A thermodynamic approach to water vapor adsorption provides a better knowledge of adsorbate/adsorbent interactions by characterizing the absorbent nature of the activated carbon surface and evaluating the adsorption energy. Water adsorption isotherms (at 202 28C) were obtained by a gravimetric method. Dried PAC was placed over a saturated salt solution providing a xed water vapor pressure (Wnek and Davies, 1987). After an equilibrium time of 200 h, the rate of adsorbed water was determined by weighing. The precision of the method was accurate to within 5%. From dierent water adsorption isotherms it is possible to calculate the spreading pressure of activated carbon. Spreading pressure p0 is the decrease in the potential of interaction between PAC and water, characterized by the

with a as the quantity of water adsorbed; a0 as the quantity of water adsorbed according to Langmuir's equation with localized adsorption; P0 as the relative vapor pressure of water (20228C); c as a constant and h = P/P0 as the relative vapor pressure of water. Equation 2 applies for low pressures where h/a is proportional to the relative vapor pressure. Maximum adsorption capacities were determined for h = 1 (as).

RESULTS

Determination of the chemical and physical structure of PAC Specic surface area and pore size distribution. The specic surface area and pore volume greatly modify the adsorption capacity and mechanisms. These are studied using argon and nitrogen adsorption isotherms. Argon adsorption isotherms are performed for a relative pressure of 4106 to 0.25 atm. The adsorption isotherms for all three raw activated carbons have similar shapes. PAC made from coconut shells (B1) and coal (C1) have the highest specic surface area (Table 1). Sample A was activated with water vapor although wood carbons are usually activated in industrial processes by chemical oxidation at high temperature so as to develop a wide specic surface (1000 m2g1). This water vapor activation explains the weak specic surface area values of sample A. Concerning pore size distribution, sample C1 has more mesopores than the other samples (Table 1). Possible modications of micro and mesoporosity between raw activated carbon and heat treated PAC are therefore observed. Experiments were done with sample C1 (before heat treatment) and C2 (after heat treatment) because thermal treatment of this PAC leads to a maximum evolution of the specic surface area (Table 2). The specic surface area increase by about 10% without signicantly modifying the physical structure of the activated carbon. The thermal treatment liberates part of the porosity by combustion with residual adsorbed

Surface properties of activated carbon


Table 2. Physicochemical characteristics of activated carbon, after heat treatment Characteristics Sample (after thermal treatment) A2 wood water vapor 658 2 20 <50 4.0 3.1 1.36 0 B2 coconuts shell water vapor 956 2 20 <50 10.9 6.0 1.15 0.14 C2 fossil coal water vapor 1166 2 6 <50 8.1 4.9 0.48 0.09

3417

Reference Origin Activation Specic surface (argon) (m2 g1) Size (mm) Total ash (%) I.E.P. Basic groups (meq g1) Acid groups (meq g1)

oxygen or surface functional groups. Thermogravimetric analyses to be carried out later will enable us to conrm this hypothesis (Julien et al., 1994). Figure 1 gives pore size distributions of C1 and C2. The specic surface area used below is that measured by argon adsorption because it enables a better characterization of the microporous system (molecular surface area = 16.21020 m2). Analysis of ionic structure. XRD analyses were done on ashes obtained after calcination of samples. Qualitative analysis of the main elements contained in the PAC structure was done. The original chemical structure can, however, be modied (oxidation) during calcination. Ashes are not amorphous, but contain crystallized substances of mineral origin, Sample A1 (wood coal) contains the most crystallized elements in relation to the number and magnitude of peaks recorded. The qualitative analysis was done with DIFFRACT software which contains a large database (ASTM). Sample A1 is made up of CaO (lime), B1 of KA1Si2O6 (leucite), MgFe2O6 (magnesoferrite), Mg2P2O7 (magnesium phosphate) and C1 of Fe2O3 (hematite) and SiO2 (quartz). These results do not give a complete list of all of the elements contained in PAC, but indicate the principal crystallized compounds in activated carbon, those therefore which might be detected by chemical titration. These oxides probably develop during the activating procedure and are present on

activated carbon surfaces. If so, the presence of calcium oxide in ashes of A1 can explain the basic character of a suspension of this type of activated carbon (Table 1). This oxide was not observed in B1 and C1 but other basic oxides (K2O, Na2O, etc.) were not detected by XRD (absence of pure crystallized substances). Quantitative analysis of the PAC structure revealed high Ca concentrations in both A1 (1.85%) and B1 (1.41%) (Table 3). For PAC B1, calcium could be present in the form of CaCl2. The ionic structure of the activated carbon may correspond to localized adsorption sites with strong interaction capacity (ionic connections, covalent and electrostatic) with molecules, including water molecules, adsorbing onto the surface. The titration of ions released by activated carbon or present in the structure were necessary and complementary. Mineral components liberated in water by activated carbon were measured by conductivity and pH (Table 3). These results, when associated with the determination of the rate of ash, enabled us to estimate the mineral composition of the activated carbon. Table 4 shows the dierence between raw PAC and thermal treated samples. A reduction reaction can be proposed: Cx O H2 O D C2 2OH x

Fig. 1. Pore size distribution of raw and heat treated activated carbon (C1, C2) (N2-B.E.T.).

3418

F. Julien et al.
Table 3. Concentrations of some mineral compounds in activated carbon samples Activated carbon references

Element Ca2+ Mg2+ K+ Na+ SO2 4 PO3 4 NO 3 Cl 3+ Al Mn2+ Zn2+ Ni2+ Pb2+ Fe2+
a

A1 B1 C1 structural element concentrationsa g per 100 g of PAC 1.85 0.06 0.13 0.02 0.46 0.23 0.03 0.02 <0.01 <0.01 0 0.03 1.41 0.13 0.80 0.14 0.77 0.16 0.50 0.06 <0.01 <0.01 0 0.53 0.28 0.03 0.04 0.02 0.45 0 0.32 0.01 <0.01 <0.01 0 1.35

A1 0.23 0.02 0.04 <0.01 0 0 0 0.01 <0.01 0 0 0 0 0

B1 C1 A2 B2 release element concentrationsb g per 100 g of PAC 0.45 0.02 0.06 0.01 0 0 0 0.08 0.02 0 0 0 0 0 0.18 0.01 0.01 <0.01 0.03 0 0 <0.01 <0.01 0 0 0 0 0 0.82 <0.01 0.04 0.02 0 0 0 0.66 <0.01 0 0 0 0 0 0.33 0.03 0.10 <0.01 0.01 0 0 0.06 0.02 0 0 0 0 0

C2 0.16 0.07 <0.01 <0.01 0.01 0 0 <0.01 <0.01 0 0 0 0 0

Structural element obtained after mineralisation of PAC.

Table 4. Variation of surface functional groups contents and zeta potential after heat treatment of activated carbon Origin Treatment Total acid groups (meq g1) Evolution Zeta potential (mV) pH = 5.5 Evolution wood A1, water vapor A2, water vapor + thermal 0.01 0 22 0.014 0 mV4 21 coconuts B1, water vapor B2, water vapor + thermal 0.25 0.14 14 0.114 +30 mV4 +16 fossil C1, water vapor C2, water vapor + thermal 0.23 0.09 33 0.144 +39 mV4 +6

The Cx site designates bonds CC s or p of the carbon structure. Activated carbon electrokinetic potential and pH increase with C2+ formation x (Table 5). The process, carried out under inert atmosphere (under nitrogen ow), does not include combustion of organic structure (Tables 1 and 2). Results of the titration of the elements contained in ashes or released in solution (Table 3) indicate the presence of Ca2+ in sample A1. This Ca2+ will be partially in the calcium oxide form CaO. Sample B1 also contains calcium, but not in the oxide form. For A1, B1 and C1 the ionic scale calculation indicates that anions have not been titrated: Cl and NO ions that cannot be titrated because ashes 3 are treated in a nitric and chlorhydric acid mixture. By calculating the conductivity we see that OH,

Ca2+ and K+ ions are mostly responsible for the measured conductivity. The presence of Ca2+ modies organic molecule adsorption onto activated carbon. This ion reacts as a co-adsorbate by creating a bridge between the structure of activated carbon and adsorbed molecules (Lafrance et al., 1988; Cannon et al., 1993). Calcium ions either act as intermediates with sites of adsorption or supercially modify the ionic state, thereby improving adsorption. Lafrance et al. (1988) have shown that the adsorption of a surfaceactive compound, sodium dodecyl sulfonate (SDS), on an activated carbon (F400) is improved by the addition of some quantity of Ca2+ ions. Taking this into account, they suggest that the bonding of the SDS anion in the presence of

Table 5. Comparison of some physicochemical characteristics of non-treated (A1, B1, C1), heat treated (A2, B2, C2) and chemical treated activated carbon (A1/HNO3, B1/HNO3, C1/HNO3) Reference A1 Total ash (%) Zeta potential (mV), pH = 5.5 pH Acid groups (meq g1) Basic groups (meq g1) SBET (argon) (m2 g1) Spreading pressure (mJ m2) a0103 (mmol m3) a0103 (mmol g3) a0103 (mmol m3) as (mmol g1) 3.9 22 10.3 0.01 1.56 640 18.3 2.2 1.42 25.3 16.2 A2 A1/HNO3 4.0 21 10.6 0 1.36 658 19.3 2.09 1.38 26.7 17.6 0.9 31 0.95 1540 24.3 B1 10.8 14 10.0 0.25 1.2 947 10.1 0.8 0.76 20.4 19.4 B2 B1/HNO3 10.9 +16 10.2 0.14 1.15 956 14.0 0.66 0.63 22.4 21.4 5.4 25 1.705 1025 29.3 9.6 9.9 21.0 21.5 C1 8.3 33 8.6 0.23 0.61 1047 7.4 0.27 0.28 18.5 19.4 C2 8.1 +6 9.0 0.09 0.48 11661 9.6 0.39 0.46 20.5 23.9 C1/HNO3 5.1 43 1.250 1219 26.3

Surface properties of activated carbon

3419

Fig. 2. T.G.A. of raw activated carbon (A1, ex-wood; B1, ex-coconuts shell; C1, ex-coal).

specically adsorbed metallic ions could be partially explained by the formation of an ionic bridge between the complex adsorbate and the surface of the activated carbon macropores. During thermal oxidation, calcium can turn micropores into mesopores by a catalytic reaction (Papirer et al., 1989). Other metallic ions are present in activated carbons (Table 3) but in smaller proportions (except iron in the case of C1) and are not released in distilled water. Analysis of activated carbon surface functional groups. Results of thermo-gravimetric analysis (TGA) of raw PAC are given in Fig. 2. The rst loss of mass observed at temperatures below 1008C corresponds to the departure of water in the pores. The shape of the TGA curve is identical to that obtained for other activated carbons (Lemarchand, 1981; Farkhani, 1993). Sample A1 shows no signicant loss of mass between 100 and 9008C. Only B1 and C1 show a loss of mass beyond 4505008C. This can be attributed to the destruction of acidic functional groups by the thermal process. In order to verify that there is no combustion due to oxygen traces in the vector gas, additional experiments were done with sample B1. TGA was done for raw PAC and then repeated for the same sample cooled under a nitrogen ow (Fig. 2). The expected result was obtained because the second TGA no longer reveals a loss of mass at temperatures above 1008C. Similarly, in order to analyse the dierent losses of activated carbon mass during thermal treatment, TGA and IR detection of emitted gas was done jointly. The results indicate that for a temperature of 4008C, the CO2 departure (2350 cm1) can be attributed to the destruction of carboxylic surface functional groups. The water molecule departure is also observed during a heat rise, wave lengths ranging from 3500 to 3800 cm1. This seems to conrm the hypothesis of bound water molecules (Le

Cloirec et al., 1988). At 8008C, CO and CO2 were detected. In the case of activated carbon prepared from wood (sample A1), there was no signicant release of CO and CO2 gas during the thermal process. A direct titration of acidic and basic functional groups indicates that only PAC made from wood (A) does not contain acidic surface functional groups but, on the other hand, PAC B and C contain functions that can be eliminated with thermal treatment. During TGA, CO2 comes from both the destruction of surface functional groups and from carbon structure combustion using oxygen in the pores. Basic surface functional groups were titrated in the three samples and are slightly modied by the thermal process. Inuence of PAC structure on zeta potential The study of the stability of the zeta potential of the heat treated activated carbon in distilled water suspension shows that electrokinetic potential is steady state during 10 h of contact. This contact time is sucient to reach equilibrium and to determine organic molecule adsorption isotherms. Figure 3 shows the evolution of the zeta potential as a function of pH for the three raw PAC. The evolution of their electrokinetic potential is the same. Variation of the zeta potential of the three heat treated PAC is a function of pH (Fig. 3). The isoelectric point (I.E.P. = pH value for zeta potential = 0) of raw samples B1 and C1 is almost identical for a pH value of 3.5 and for pH 2.2 in the case of A1 (Fig. 3). Nevertheless, the three raw activated carbons have a negative zeta potential under operating conditions (pH 5.5). This negative zeta potential is usually attributed to the surface functional groups distributed on the PAC surface. However, acidic surface functions were not observed on A1 and the negative surface charge

3420

F. Julien et al.

Fig. 3. Variation of zeta potential of PAC particles with pH; A1: non-treated ex-wood PAC, B1: nontreated ex-coconuts PAC, C1: non-treated ex-coal PAC.

might be attributed partly to the presence of typical mineral groups (-SO or -PO2). 4 4 Their presence is conrmed by quantitative chemical determinations (Table 3). All the results indicate that the thermal treatment systematically shifts the P.I.E. to basic pH values, the consequence of a chemical structure evolution of the surface. I.E.P. increases by 2.5 pH units for B2 and 1 for C1. There is a correlation between acidic surface functions before and after thermal treatment and the evolution of the zeta potential (Table 4). Thus a signicant decrease in PAC acidic functional groups in samples B and C can be correlated with an increase of 30 to 40 mV of the zeta potential. Sample A which has very few acid surface groups is not modied by thermal treatment and its zeta potential remains stable. Zeta potential vs heat treatment temperature are plotted in Fig. 4. Julien et al. (1994) do not observe

variation in the surface functions of groups III and IV and basic groups during heat treatment of the B1 sample. Nevertheless, for acidic groups I and II, there is a decrease from 2003008C. These results are in good agreement with the thermogravimetric analysis of sample B (Fig. 2) that indicate a loss of mass starting from 2008C corresponding to a release of CO2. Results of the acidic function titration of surface on oxidized samples (titration by sodium hydroxide) can be compared to the zeta potential of PAC (Table 4). The hot nitric acid oxidation process causes a large increase in the concentration of acidic functional groups for all activated carbons. Nitric acid reacts with basic functional groups chromene or pyrone to form acidic functional groups by heterocycle opening. Elemental analyses show that there are no nitro xing groups on the activated carbon surface. Nevertheless, oxygen content increases and

Fig. 4. Zeta potential of PAC particles vs the temperature of the thermal treatment.

Surface properties of activated carbon

3421

Fig. 5. Variation of zeta potential (pH = 5.5) of grafted or non-grafted coconuts PAC with pH.

can be correlated to the number of acidic surface functional groups. The action of nitric acid is experimentally observed by the release of abundant nitrous fumes. This systematically decreases the values of the electrokinetic potential of particles of PAC approximately 10 mV. This is related to the increase in the number of acid surface functional groups. The grafting process of 1 or 2% the mass of polymer adds a positive zeta potential of +232 2 mV to the B1 activated carbon at pH 5.5. The results suggest that a 1% solution of polymer allows a saturation of the carbon surface. Figure 5 shows that the grafting process increases the zeta potential over the whole pH range. Compared to a raw activated carbon, the ``grafting'' operation shifts the I.E.P. to basic pH (pH 9.5). This is explained by a specic adsorption of the polymer. The adsorption of polymer is not limited to small pores, but rather to the external particle surface. The number of positive charges on the surface determines the adsorption mechanism by electrostatic attraction of ionized (benzoic and picric acids) or nonionized molecules (phenol). Results of titrations indicate a signicant decrease in the concentration in both acidic and basic functional groups (Table 6). This decrease can be explained by partial surface functional group neutralization (e.g. carboxyl, carbonyl or pyrone) by the quaternary ammoniums of the polymer. Inuence of PAC structure on thermodynamic constants of water adsorption Spreading pressure. Adsorption isotherms of water in the vapor phase are given in Fig. 6 and can be compared with literature values. Isotherms

obtained at 298 K correspond to type V in Brunauer's classication. Sample preparation diers from author to author: Freeman and Reucroft (1978) regenerated activated carbons at 3008C in an enclosure under vacuum (106 mm Hg) for 45 h. Barton et al. (1984) used an enclosure and a heating temperature of 1008C. Mahle and Friday (1988) heated their sample to 1108C under atmospheric pressure. Hassan et al. (1991) heated activated carbons to 2008C under vacuum (104 mm Hg) for 10 h. Our PAC samples were dried at 1108C for 72 h at atmospheric pressure. The various isotherms obtained (Fig. 6) have almost the same shape indicating a similar water adsorption mechanism for all PAC. Isotherms can be described by Polanyi's theory (Chiou and Manes, 1973) and the potential of adsorption is expressed by the following equation (Schenz and Manes, 1975; Rosene and Manes, 1977):   P e RT ln 3 P0 where e is the adsorption potential (J mol1 g1); R is the ideal gas constant (J mol1 K1); T is the absolute temperature of the experience (K) and P/P0 is the relative pressure. This type of potential distribution is not temperature dependent in a limited range (from 10 to 308C) and for one solute (Hassan et al., 1991). Results show that the thermal process increases the potential of adsorption. For a given adsorbed quantity (10 mmol of water adsorbed onto 1 g of PAC), the measured increase is +0.25, +0.75 and

Table 6. Comparison between chemical nature of surface functional groups and zeta potential of non-treated and grafted coconuts shell activated carbon References B1 B1/``grafted'' Acid groups (meq g1) 0.252 0.01 0.010 2 0.001 Basic groups (meq g1) 1.202 0.06 0.382 0.02 Zeta potential (MV) at pH = 5.5 14 21 +23 22

3422

F. Julien et al.

Fig. 6. Comparison of experimental water vapor adsorption isotherms with literature data at 294 K.

+0.5 kJ g1 mol1 for activated carbon types A, B, C, respectively. The potential of adsorption decreases very rapidly with the saturation of the surface for all activated carbons studied (Fig. 7). Adsorption potential is almost independent of the activated carbon specic surface area, in agreement with the work of Mahle and Friday (1988). The amounts of water adsorbed on PAC are similar in magnitude. Water adsorption is signicant for low adsorption potential (less than 3 kJ g1 mol1). The curves in Fig. 7 show that nitric acid treatment (oxidation of PAC surface) increases the quantity of water adsorbed and creates hydrophilic sites (localized adsorption sites) on the activated carbon, increasing the adsorption potential. The adsorption capacities of modied PAC is, in all cases, greater than that of raw samples. The spreading pressure, p0, is calculated from the gas adsorption isotherms (Figs 6 and 7). The calculation of p0 includes the specic surface area of the material and gives a physical quan-

tity representative of the chemical surface state of the activated carbon. Thus, p0 reveals the hydrophilic character due to surface oxygenated groups and oxides (Papirer et al., 1988; Julien et al., 1994). Nitric acid and thermal treatment increase the spreading pressure (Table 5). Moreover, thermal treatment decreases the total quantity of oxygenated functional groups (acid and basic). This seems to indicate that a large number of surface functional groups are not determined by acidobasic titration. To verify that mineral impurities are not the main source of potential adsorption sites, activated carbon (A1) was washed in a chlorhydric acid solution 1 N. The initial reading pressure (18.3 mJ m2) was slightly modied by mineral element removal (21.9 mJ m2). These results verify that water molecule adsorption occurs for the most part at specic adsorption sites. There is a correlation between p0 values and the sum of all of the acidic and basic surface functional groups, and raw and heat treated activated carbon. These adsorption

Fig. 7. Characteristic curve of water vapor adsorption onto coconuts activated carbon.

Surface properties of activated carbon

3423

Fig. 8. Examples of possible bond between water molecules and activated carbon surface functional groups.

sites imply hydrophilic interactions with polar surface groups (e.g. -COOH or -SO4H, .PO4H. The likeliest mechanism between water molecule and ionic compounds or surface functional groups is a physical adsorption by hydrogen bond with mineral anions and/or cations (Fig. 8). Despite the deciency of acidic functional groups in sample A1, this activated carbon has a very reactive surface for small polar molecules. For activated carbons A1, B1 and C1, water molecules adsorbed at rst onto specic adsorption sites and modify interactions between carbon surface and nearby water molecules (multilayer adsorption model) as indicated by the dierence between a0 and as (Table 5), a0 corresponding to specic surface functional groups and as to a parameter of the physical adsorption by capillary condensation. Therefore, PAC C1 contains very few initial adsorption sites (a0) but the large amount of adsorbed water, as, corresponds to the size of the pore volume (Table 1). Nitric acid treatment produces surface acidic functional groups on activated carbon (Noh and Scharwz, 1990; Meldrum and Rochester, 1990). In this case, oxygenated groups created during oxidation constitute water molecule adsorption sites. These groups can be produced the activated carbon surface, by heterocycles opening or p bond with carbon atoms (not aromatic carbon) that are the most easily oxidizable sites.
CONCLUSION

spreading pressure. Chemical studies show that all PAC samples contain large quantities of mineral ions, mostly calcium, sulfate and phosphate ions. Structural anions are only partially released when raw activated carbon is put in suspension in distilled water. The thermal treatment of activated carbon decreases acidic surface functional groups with a detection of a release of CO and CO2. This reaction accompanies an increase in the electrokinetic potential and positive values of +16 and +6 mV are obtained with coconut and fossil PAC, respectively. Activated wood carbon, is not modied by a thermal treatment, in agreement with the absence of carboxyl functions. Adsorption isotherms of water in activated carbon enable us to estimate initial polar adsorption sites. The localised adsorption of water on activated carbon is a function of the number of mineral ions and surface functional groups. Quantitative adsorption of water is related to the microporous volume. The ``grafting'' of a cationic polymer adds a positive charge to activated carbon surfaces, but the quantity of surface functional groups is also modied by this process.
AcknowledgementsThis work was nanced by the Limousin Regional Council and the ``Contrat de Plan Etat-Region''.

REFERENCES

This work reveals relationships between the physical and chemical structure of activated carbon and global parameters such as the zeta potential or

Barton S. S., Evans J. R., Holland J. and Koresh J. E. (1984) Water and cyclohexane vapour adsorption on oxidized porous carbon. Carbon 22, 265272.

3424

F. Julien et al. Mahle J. J. and Friday D. K. (1988). Unclassied Paper, CRDEC-TR-018. U.S. Army Chemical Research Aberdeen Proving Cround, MD. Mazet M., Farkhani B. and Baudu M. (1994) Inuence d'un traitement thermique ou des charbons actifs sur l'adsorption de composes organiques. Water Res. 28, 16091917. Meldrum B. J. and Rochester C. H. (1990) In situ infrared study of the surface oxidation of activated carbon in oxygen and carbon dioxide. J. Chem. Soc. Faraday Trans. 86, 861865. Murata T. and Imagawa M. (1976) Some considerations relating to the origin of zeta potential of carbon black particles. Denki Kagaku Oyobi Kogyo Butsuri Kagaku 44, 778784. Nakhla G., Abu-Zaid N. and Farooq S. (1992) Oxygen industrial enhancement of the adsorptive capacity of activated charcoal. Environ. Sci. Technol. 13, 181188. Noh J. S. and Scharwz J. A. (1990) Eect of HNO3 treatment on the surface acidity of activated carbons. Carbon 28, 675682. Papirer E., Balard H. and Vidal A. (1988) Inverse gas chromatography: a valuable method for the surface characterization of llers for polymers (glass bers and silicas). Eur. Polym. J. 24, 783790. Papirer E., Li S., Balard H. and Jagiello L. (1991) Surface energy and adsorption energy distribution measurements on some carbon blacks. Carbon 29, 11351143. Papirer E., Vidal A. and Balard H. (1989) Analysis of solid surface modication. Am. Chem. Soc. ACS Symp. Ser. 391, 248261. Preston D. R., Bitton G. and Farrah S. R. (1990) Enhancement of enterovirus infectivity in vitro by pretreating host cell monolayers with the cationic polymer, polyethyleneimine. Appl. Environ. Microbiol. 56, 295 297. Rao A. M., Fung A. W. P., Dresselhaus M. S. and Endo M. (1992) Structural characterization of heat-treated activated carbon bers. J. Mater. Res. 7, 17881794. Rosene M. R. and Manes M. (1977) Application of Polanyi adsorption potential theory to adsorption from solution on activated carbon. Competitive of ternary solid solutes from water solution. J. Phys. Chem. 81, 16461650. Rosene M. R., Ozcan M. and Manes M. (1976) Application of Polanyi adsorption potential theory to adsorption from solution on activated carbon. Ideal, non-ideal and competitive adsorption of some solids from water solution. J. Phys. Chem. 80, 25862589. Schenz T. W. and Manes M. (1975) Application of Polanyi adsorption potential theory to adsorption from solution on activated carbon. Adsorption of some binary organic liquid mixtures. J. Phys. Chem. 79, 604 609. Snoeyink V. L. and Weber W. J. (1967) The surface chemistry of activated carbon. Environ. Sci. Technol. 1, 228 234. Wnek W. J. and Davies R. (1987) An analysis of dependence of zeta potential and surface charge on surfactant concentration, ionic strength and pH. J. Colloid Interface Sci. 60, 361377.

Boehm H. P., Diehl E., Heck W. and Sappok R. (1966) Identication of functional groups in surface oxides of soot and other carbons. Chem. Int. Ed. 3, 669. Bradley R. H. and Rand B. (1993) The adsorption of vapours by activated and heat treated microporous carbons. Carbon 31, 269272. Cannon F., Snoeyink V., Lee R., Dagois G. and De Wolfe J. (1993) Eect of calcium on thermal regeneration of GAC. J. Am. Water Works Assoc. 84, 7380. Chiou C. and Manes M. (1973) Application of Polanyi adsorption potential theory to adsorption from solution on activated carbon. Steric factors as illustrated by the adsorption of planar and octahedral metal acetylacetonates. J. Phys. Chem. 77, 809813. Dorris G. M. and Gray D. (1979) Adsorption, spreading pressure and London force interactions of hydrocarbons on cellulose and wood bers surfaces. J. Colloid Interface Sci. 71, 93106. Dubinin M. M., Zaverina E. D. and Serpinski V. V. (1955) Investigation of the porous structure of active carbons by complex methods. J. Chem. Soc. 15, 5182. Farkhani B. (1993) Inuence de la structure physico-chimi que de materiaux solides sur l'adsorption de molecules organiques. Thesis, University of Limoges, No. 25. Freeman G. B. and Reucroft P. J. (1978) Adsorption of HCN and H2O vapor mixture by activated and impregnated carbons. Carbon 17, 313316. Gajardo R., Diez J., Jofre J. and Bosch A. (1991) Adsorption-elution with negatively and positively charged glass powder for the concentration of hepatitis A virus from water. J. Virol. Methods, 345351. Garten V. A., Weiss D. E. and Willis J. B. (1957) A new interpretation of the acidic and basic structures in carbon (1). Lactone groups of the ordinary and uoresceine types in carbons. Aust. J. Chem. 10, 295308. Hassan N. M., Ghosh T. K., Hines A. L. and Thacker W. E. (1991) Adsorption of water vapor on BPL activated carbon. Carbon 29, 681683. James R. O. and Healy T. W. (1972) Adsorption and hydrolyzable metal ions at the oxidewater interface, thermodynamic model of adsorption. J. Colloid Interface Sci. 10, 4252. Julien F., Baudu M. and Mazet M. (1994) J. Water SRTAqua 43, 278286. Lafrance P. and Mazet M. (1988) Activated carbon adsorption of humic substances in the presence of salts: its relation with electrokinetics properties of activated carbon. J. Am. Water Works Assoc. 81, 155162. Lafrance P., Yaacoubi A. and Mazet M. (1988) The inuence of metal ions released by an activated carbon on the adsorption of organics: the role of calcium ions. Water Res. 22, 13211329. Le Cloirec P., Martin G. and Gallier J. (1988) Hydrogen NMR investigation on phenol saturated and unsaturated activated carbon. Carbon 26, 275282. Lemarchand D. (1981) Contribution a l'etude des possibi lites de retention de la matiere organique en solution dans Peau potable sur charbon actif. Thesis, University of Rennes, No. 133.

You might also like