Selective adsorption on fibrous activated carbon of organics from aqueous solution: Correlation between adsorption and molecular structure

Selective adsorption on fibrous activated carbon of organics from aqueous solution: Correlation between adsorption and molecular structure

~ Pergamon Waf. Sci. T«h. Vol. 35, No.7. pp. 251-259,1997. C 1997 IA WQ. Pubhshed by Elsevier SCIence LId PU: S0273-1223(97)QO 138-8 Pnnled In Or...

722KB Sizes 6 Downloads 17 Views

~

Pergamon

Waf. Sci. T«h. Vol. 35, No.7. pp. 251-259,1997.

C 1997 IA WQ. Pubhshed by Elsevier SCIence LId

PU: S0273-1223(97)QO 138-8

Pnnled In Oreat Britain. 0273-1223197 SI7'
SELECTIVE ADSORPTION ON FIBROUS ACTIVATED CARBON OF ORGANICS FROM AQUEOUS SOLUTION: CORRELATION BETWEEN ADSORPTION AND MOLECULAR STRUCTURE Catherine Brasquet*, Etienne Subrenat** and Pierre Le Cloirec* • Ecole des Mines de Nantes. Subatech. UMR no 6457. BP 20722. 4 rue Alfred Kastler. 44307 Nantes Cedex 3. France •• Actitex. 16 rue Trezel. 92300 Levallois Cedex. France

ABSTRACJ' In industrial processes. granular activated carbon (GAC) is generally used to remove pollutants from wastewater. Recently, a new adsorbanl has been explored. fibrous activated carbon (FAC). Experiments were carried out with two FACs having different specific surface areas (1500 and 1300 m 2.g- l ) and pore-size distributIons to study adsorption of various organic compounds from aqueous solution. Results were compared with adsorption onto one GAC with a specific surface area of about 1000 m 2.1I- 1• Classic models were applied and kinetic constants were computed. In most cases. FAC with the higher specific surface area (named CS 1501) showed beUer adsorption capacities and kinetics than the two other FACs. For example. adsorption velocily of benzaldehyde was 7.2 ~ JO-5 I.mll-I.mm-I with CS 1501 and about 3 ~ JO- 5 1.mll• I.min- I with other FACs. Furthermore, adsorption onto CS 1501 of a great number of organic compounds (aliphatic and aromatic) depended on solute molecular characteristics. For instance. solute molecular size seemed to play an important role: adsorption capacily of high molecular weight compounds (humic substances) was about 3 mg.g- I• a value much lower than those of low molecular weight compounds, which were respectively 200 mg.g- I and 400 mll.II- 1 for phenol and benzoic acid. From experimental results, a correlation of QSAR (Quantitative Structure-Activily Relationship) type has been set up. This relationship predicts the adsorbability of organics compounds onlo fibrous activated carbon from the molecular properties of these compounds. ~ 1997 IA WQ. Published by Elsevier Science Ltd

KEYWORDS Adsorption; fibrous activated carbon; humic substances; organic micropollutants; molecular connectivity indexes; QSAR; selectivity; water treatment. INTRODUCTION Granular activated carbon (GAC) is the most widely used adsorbent for removing organic pollutants from water and wastewater (Weber et al., 1964). Recently, there has been an increase in the interest in a new form of activated carbon: fibrous activated carbon (FAC). The first isotherm and kinetic studies carried out in batch systems on this new material showed that FAC seemed very effective in removing contaminants from 251

252

c. BRASQUET tt a1.

aqueous or gaseous phase (Weber et al., 1964; Cal et al., 1994; Le Cloirec et al., 1996). Adsorption capacities and velocities are often higher than those of GAC (Thwaites et al., 1993; Baudu et al.• 1991a, b). Furthermore. FAC adsorption seems selective for low molecular weight compounds (Brasquet et al., 1996). For example. Ryu (1990) compared adsorption of benzene and cyclohexane on a FAC. The molecular sieve of FAC resulted in the selective adsorption of benzene to cyclohexane. the molecular diameter of cyclohexane being higher than that of benzene. These adsorbability data generated in the laboratory are used only in evaluating the suitability of the adsorption for the chemical(s) of interest. However, considering the large number of chemicals identified as pollutants, it is interesting to develop quantitative relationships to predict the adsorption of organic compounds on activated carbon without experimentation (Nirmalakhandan and Speece. 1988). These relationships. the Quantative Structure-Activity Relationships (QSAR), were first developed to predict the toxicity and the fate of chemicals in the environment, especially in the pharmaceutical and medicinal areas. Now. they are also used to develop quantitative relationships between adsorption properties and molecular characteristics of a set of chemicals. Blum and Suffet (1989) and Blum et al. (1994) have compared four QSAR methods (the solvophobic theory, the Linear Solvation Energy Relationship LSER, molecular connectivity indexes and the Polyani potential theory) and have identified molecular connectivity as the most promising method for an adsorption QSAR for use in water treatment only. Nirmalakhandan and Speece (1990) have also preferred the connectivity model to the solvophobic theory, because it allows various classes of compounds to fit in the same equation. These molecular connectivity indexes are descriptors originally proposed by Randic as branching indexes and later extensively developed and formalized by Kier et al. (1977) and Kier and Hall (1986). These indexes, fundamental in nature. are calculated only from stuctural information and encode information relating to physicochemical properties. Correlations of high quality have been demonstrated with connectivity indexes. Kier et al. (1977) have found correlations between odor similarities to a reference standard and connectivity terms. Nirmalakhandan and Speece have correlated the Henry's constant (Nirmalakhandan and Speece, 1988a) and the solubility (Nirmalakhandan and Speece. 1988b) with molecular connectivity indexes and polarizability. This paper is a study of various FAC performances for adsorption of 17 aromatic and aliphatic compounds. It shows the selectivity of FAC for low molecular weight organics. From the data set of the 14 compounds, a relationship between adsorbability and molecular connectivity indexes is given. METHODS Actiyated carbon materials Activated carbons are commercial products. Before use. they were washed with deionized water and dried at 105°C. Their main characteristics are given in Table 1. Specific surface areas were determined by the Brunauer Emmet and Teller (BET) (Bansal et al., 1988) method by N2. Scanning Electronic Microscope pictures of cloths have been reported previously (Le Cloirec et al., 1997), showing a fiber diameter of about

lOJ.1m.

Table 1. Main characteristics of adsorbents (Actitex & Pica co., Levallois, France)

Fibrous activated carbon of organics

253

Solutes and analysis All solutes were commercially available. Their main characteristics are given in Table 2. Humic substances were extracted from peat by Aldrich, Fulvic and Humic acids were extracted from Lake Dremond by the Thurman and Malcom method (Le Ooirec et al., 1990). M is the molecular weight and S the solubility. Aromatic compounds were analysed in the UV region with a spectrophotometer UV-2I01 PC Shimadzu. Aliphatic compounds were analysed with a Shimadzu TOC-5000 A analyser. Table 2. Main characteristics of adsorbates solute toluene 10 1 phenol 2 benzaldehyde 13 benzOiC acid 4 dl- tyrosme 5 dl- phenylalanine l-ethoxf·2-terbutoxy 16 ethane 7 4-terbutyl benzoIc acid 8 vaniline aniline /9 ! 10 p- chlorophenol II p- nitrophenol 12 2-terbutyl 4-methyl phenol I3 3,)- dimetoxy benzoIc acid 14 dlethylphtalate 15 I humiC substances'" (HS) 16 fulvlc acid ** (FA) 17 humiC acid ** (HA) IT'

fonnula T6H5CH3 <::6HSOH C6HSCHO C6HSCOOH C9HII03N C9 H II02H l:SHI802 (CH3)3CC6H4cOUH 4-(OH)C6H3-3-(OCH3~OH

C6 HSNH2 CIC6H4UH N02C6H40H C5HI2C6H30H (CH30)2C6H3COOH C6H4-I,2-(C02C2HS)2

-

M (g.mol- I ) 92 94 106 122 lSI 165 146 11M 152 93 129 139 164 182 222 2000-5000

-

-

S (g.l-I) 0.47 61.00 3.30 2.80 0.41 14.20

-

:!> 0.4 ~.I-1

10.00 34.UU

27.10 16.00 S 0.4 g.l-I :!> 0.4 g.l-I S 0.4/!:.I- 1

-

AdsoJlltion procedure The study of adsorption took place at a constant temperature (20 ± I°C) under continuous stirring. Two sets of adsorption experiments were perfonned: kinetics and equilibrium. Kinetic experiments of organic micropollutants (respectively HS) were carried out with 0.5 g (resp. 0.2 g) of activated carbon in the form of granules or cloth. Solution Volume was 1 I (resp. 0.250 I) and organic concentration was 100 or 200 mg.!-! (resp. 50 mg.l- I). Samples were withdrawn at regular times to plot the concentration of organic in solution as a function of time. For the equilibrium experiments, various weights of activated carbon were stirred in 0.250 I of a solution containing a micropollutant with a concentration of about 100 or 200 mg.!-!, or HS with a concentration of 50 mg.!-!. The contact time to reach equilibrium was set at 12 hours for fibers and 48 hours for granules (Baudu et al., 1991 b). To study FAC selectivity, equilibrium experiments were also performed with CS 1501 and binary mixtures of phenol and HS (both at a concentration of 100 mg.!-I) and phenol and FA (both at a concentration of 50 mg.!-I).

C. BRASQUET el al

254

RESULTS AND DISCUSSION MonosoIute adsorption

Adsorption kinetics. From the Adams-Bohart-Thomas' equation, the initial adsorption coefficients were computed Baudu et ai., 1991a) for each organic micropollutant by:

with C: esolute concentration (mg.!-l) t: time (5) CO: initial solute concentration (mg.!-l) V: volume of solution (I) m: mass of the activated carbon in the batch reactor (g). Values are given in Table 3. Adsorption kinetics of humic materials were performed with CS 1501, RS 130 I and NC 60, and with another GAC, PICABIOL. None of the four activated carbons was effective in removing HS and HA. FA adsorption kinetic curves are given in Figure I. Initial adsorption coefficients of FA on RS 1301 and PICABIOL are given in Table 3. 60

-""

50 0

§

"

"

'0

0

~

~

30





20

«

"-

100

200

300

400

500

Tune (mm)

Figure I. Adsorption kinetic curws of FA by various aCllvatoo carbons. Co = 50 mg.l· 1, AC weight =0.2 g. V = 0.2501.

Table 3 shows that micropollutants' adsorptIOn velocity decreases generally following: CS 1501 > RS 130 I > NC 1i0. The higher adsorption rate, obtained with FAC are due to the htgh specific surface area of doths (table I) and to the fact that the micropores are directly on the external surface area of fibers (Ryu. 1990; Thwaites et al .• 1993; Daley et al .. 1996). Migration distances and mass transfer resistance are reduced compared with GAC (Pimenov et al . 1995). However. for solutes no. 3, 6. 10 and 12. initial adsorptJon coefficients were higher with RS 1301 than with CS 1501. Indeed. these solutes have a larger molecular size. and their adsorption onto RS 1301 may be facilitated by its higher mesopores concentration. a part of CS 150\ micropores being less accessible to them.

255

Fibrous activated carbon of organics

Table 3. Initial adsorption coefficients, Freundlich and Langmuir parameters of organic micropollutants adsorption onto GAC and FAC solute n°

AC

0

cs 15UI

1

2 3 4 5 6 7 8 9 10 II

12 13 14 16

RS 1301 NC60 CS 1501 RS 1301 NC60 CS 1501 RS 1301 NC60 CS 1501 RS 1301 NC60 CS 1501 RS 1301 NC60 CS 1501 RS 1301 NC60 CS1501 RS 1301 NC60 CS 1501 RS 1301 NC60 CS 1501 RS 1301 NC60 CS 150r RS 1301 NC60 CS 1501 RS 1301 NC60 COS 1501 RS 1301 NC60 CS 1501 RS 1301 NC60 CS 1501 RS 1301 NC60 US 15UI RS 1301 RS 1301 PICABIOL

'YX I()-s

I.mg·l.min- 1 7.4 6.5 2.2 4.8 4.2 1.9 7.2 3.5 2.2 4.6 5.9 1.4 5.4 3.5 0.6 3.8 2.6 0.3 T5 4.7 1.9 3.0 3.7 0.9 6.2 5.4 1.7 4.4 3.8 1.8 4.8 5.5 2.3 8.2 5.7 2.1 5.2 6.7 0.7 2.5 2.3 0.5 6.5 4.1 0.9 0.2

LangmuIr

lin

Freundlich K mg.g- I

r

b

0.182 0.204

270.4 206.4

g:~~

0.07 0.344

0.354 0.300 0.333 0.276 0.255 0.247 0.384 0.476 1.200 0.272 0.270 0.258

56.0 43.9 41.9 114.1 75.64 81.5 82.0 59.7 3.0 78.9 55.1 35.2 75.2 83.9 8.3 110.0 97.2 46.0 42.7 21.3 12.8

0.96 0.99 0.99 U.9H 0.97 0.99

O~

0.124 0.620

OTIS 0.225 0.280 0.309 0.655 0.381 0.184 0.095 0.220 0.330 0.373 0.358 0.206 0.271 0.216

162.~

188.7 60.9 47.5 43.8 28.5 13[6 141.2 89.1

O:no igr~ 0.153 0.172 0.117 0.167 0.328 0.096 0.235 0.102

-o~6

0.121 0.791

117.9 301.0 207.7 74.5 43.8 30.3 32.5 -222.3 215.1 4.0

U.~~

0.98 0.96 0.990.98 0.95 0.96 0.95 0.98 U.9~

0.95 0.98 0.95 0.98 0.95 0.93 -0.98 0.99 U.91S

0.99 0.98 0.99 0.99 0.99

g:~~

0.95

O~

0.98 0.99 0.86 0.98 0.88 0.95 0.95 0.97

qm mg.g- I 699.0 476.0

208.0 0.193 118.1 0.520 202.0 0.072 277.0 lJ:557 207.5 0.258 219.8 0.268 422.0 lY.o96 400.0 0.094 0.012 209.5 0.373 179.9 0.144 111.9 0.115 163.0 0.547 132.1 1.000 202.4 0.017 0.486 -232.6 0.32300 260.4 176.4 .080 232.6 0.073 534.8 0.Q21 86.2 0.039 5.500 -227.8 261.1 3.200 152.9 0.201 182.5 0.[35 205.3 0.111 150.6 0.065 20T9 2-:J50 233.1 2.270 197.6 0.380 2HO:9 J.39U 333.3 1.200 253.2 0.195 -0349 427:4 343.6 1.060 273.2 0.164 70.9 0.196 96.2 0.086 53.8 0.190 602.4 0.255 350.9 0.750 24l!.O 0.104

r U.89 0.98 0.92 0.94 0.99

g:~~

0.97 0.97 0.98 0.96 0.95 0.97 0.94 0.97 0.92 0.98 0.94 0.99 0.97 0.97 0.99 0.92

g:~~

0.93 0.98 0.99 0.99 0.95 0.96 0.97 0.99 0.98 0.88 0.94 0.95 0.97

¥:~~

0.85 U.99 0.93 0.95

This result seems confirmed by humic materials' adsorption. Indeed, these compounds have a very high molecular size (Christman and Gjessing, 1983), and a molecular weight decreasing in the following order: peat HA > soil HA > FA (Kilduff et al., 1996a). Neither humic nor fulvic compounds were removed by CS 1501 or NC 60, which are highly microporous. In addition. it appears in Figure 1 that FA, which are the

256

C. BRASQUET el al.

"smaller" macromolecules, are in part removed by RS 1301 and PICABIOL thanks to the macropores making up these carbons. The fact that some activated carbons act as "molecular sieves" for humic materials was developed by Kilduff and coworkers, but only with GAC (Kilduff et aI., 1996a,b). Starek and coworkers studied HS adsorption on FAC and GAC and they found a ratio 2 ~ 103 between GAC and FAC intraparticle diffusion coefficients (Starek et al., 1994). However, the GAC used in this study had a mesoporous volume much higher than those of NC 60 and PICABIOL carbons.

Equilibrium isotherms. Freundlich and Langmuir classical models (Brasquet et al., 1996) were applied to the

adsorption isotherms of pollutants on CS 1501, RS 1301 and NC 60. Relation parameters are given in Table 3. These values confirm the higher adsorption of micropollutants by FAC, because of their larger specific surface area (Baudu et al., 1991 b), which increases the concentration of micropores directly in contact with the medium (Economy and Lin, 1976). These experiments shOwed that adsorption on FAC seems to depend on the solute molecular properties (size) and on the adsorbents' characteristics. A selective adsorption of a given solute could then be made by choosing the suitable adsorbent. Bisolute adsOll!tion To confirm the selectivity of CS 150 I for micropollutants in the presence of macromolecules such as humic materials (HS, FA), phenol adsorption isotherms were performed with and without HS or FA. Results are given in Figures 2 and 3. 160

.,

140

.... tt

120

..

100

7.

l

80

!I-

D

~

0

.. .",.

•.. ,.,

'Ii'

-

0

0



e.

0

~

..

80 40

..................... .

'J •

20

'1

c

phenOl only

phenol + HS

I I

O~~~~~~~~~~~ o

10

20

30

40

50

60

70

80

C. (mgl-I)

Figure 2. Phenol adsorption with and without HS. Co 100 mg.l- i for phenol and HS, V =0.250 I, contact time = 12 h.

=

200~~~,-~'-~'-~~~~~-r~~~

o 150

.

100

• "C.. • ~

D

50

••

........ .



CJ

"'1

~ ~~~~~: ~n~~

i.

........................... '" ....... ,................

t

o~~~~~~~~~~~~~~~

o

10

20

30

40

50

Ce (mg.1-1) Figure 3. Phenol adsorption with and without FA. Co. 50 mgJ-t for phenol and HS. V =0.250 I. contact time = 12 h.

Fibrous activated carbon of organics

257

In both cases, similar curves were obtained: phenol adsorption is not affected by the presence of HS or FA. The CS 1501 could then be a way to remove selectively micropollutants like phenol in binary mixtures. Molecular Connectivity indexes to predict adsorbabiIity Because of the apparent relation between compounds' experimental adsorption and molecular properties, a QSAR using molecular connectivity was developed for aromatic compounds. As a descriptor of adsorption, "log K" was used, K being the Freundlich parameter. Various molecular connectivity indexes were calculated (Kier and Hall, 1986), the values are given in Table 4. Table 4. Molecular connectivity parameters and adsorbability of aromatic compounds rt'

solute

toluene I I phenol 2 benzaldehyde 3 I benzOiC acid 4 dl- tyrosme dl- phenylalanme i 5_ 7 I 4-terblJtyl benzOiC acld 8 vaniline 9 aniline IU . p- chlorophenol II I p- nitrophenol 12 2-terbutyl 4-methY!Rhenol 13 I 3,5- dlmetoxy benzoIc acid 14 diethylphtalate 0

Ixv 20410

IX 3.390 3.390 4.34U 4.3UO 6.090 5.7UO _6.190 5.270

4~~

OAUlS 0.408

~UO

LAju

u.4~

I.U4U l.7Sb 1.347

Z.,/sU

j.1S50

3.720

4.L)U

L.:WU 2.13~

3.3~

3.790 4.700 5.410 5.090 6.370

~

~

~

0.086 U.220 0.430 U.394

~

0.40~

Ooll!

2.2jO 4.2\0

0.816 1. 44lS. 2.002

j.4W

1~61

3.960

10gK

O.~_

2.b9~

1.394

~IOO

4Xpcv

2.04/S

0.389

u.15'} 0·]·I6 1.lju _Q.1±! 0.,19

1.7~

1.9!± 1.897 l.t!~ 1~30

2.211 1:!!...7-.!...

~.Il2.

~.2?J... ~32~

I~I ~3
IX encodes molecular size, molecular surface area and relative degree of branching, whereas Ixv encodes molecular volume and the topology of unsaturation and heteroatoms. 4Xpc carries information on the structural analysis of substituted rings, and 4Xpcv is useful to quantify the presence, position, type and length of the ring subsitutent (Kier and Hall, 1986). By using indexes IX and lxv, the correlation obtained with all aromatic compounds gave a poor correlation between "log K observed" and "log K calculated". Indeed. some of the 14 compounds have adsorbabilities which are less reliable and, for that reason, we eliminated solutes 0, 11 and 12 to obtain the following equation: log K = 1.266 + 0.863 IX - 1.166 IXV. The relation between log K observed and log K calculated is given in Figure 4.

2.e ,......~..--,-:~-'-""i~-'-'T'i-~.,--. [~"""'i~.""""'"

i. . . . . .~.~J.~;....~.~. t. ~~;.:::-::J.k........ .

2.4

..... " .....•..............

2,2

................:................ +..................

1,8

,

::.:

1.4 1,4

.,/

>·f:. ·". ·...·7......·"....·..

+. . . . . . .

.............~ ..............; ........~... ~..........

2

1.e

~· ..·1..

+......... ..

:.,/! : .: : : J.;.~.~: r.: .: : ·:r:·: : :. :.:::t .: :.' .:...:: .::': .' " i

.

!

'

2,4

1,8 10. Kf obi

Figure 4. Correlation between log Kf observed and log Kf calculated with molecular connectivity indexes.

258

C. BRASQUET el al.

The correlation coefficient between the values of log K calculated and observed was high (0.9216). To improve the quality of the correlation, we added another connectivity index to the correlation: 4Xpc or 4Xpcv. But in both cases, the correlation was not really improved, correlation coefficients becoming 0.9241 and 0.9218. These results give informations about molecular properties which influence adsorption. Indeed, it may signify that, if molecular surface area and volume, heteroatoms and insaturation playa role in adsorption process, the substitutant role is negligible. The equation obtained shows that adsorbability increases with the molecular surface area (lX) and decreases as the number of heteroatoms and the insaturation increase (lxv). The same tendencies were observed by Abe et al. (1983) with another kind of QSAR: the linear free-energy relationship. For 102 aromatic and aliphatic compounds, they obtained the follOwing equation: log K = 0.104 RM - 0.469 D • Nf - 1.450 where k is the adsorbability, RM is the molar refraction which increases with the molecular surface area (1 X> (Kier and Hall, 1986), Nf is the number of functional groups, and D is a constant (= 1 for aliphatic compounds, = 0 for aromatic compounds). This equation confrrms some results of our study: adsorbability increases with molecular surface area, and the substitutant role is minor (D = 0) for aromatic compounds. This QSAR approach must be studied thoroughly by increasing the number of solutes constituting the data set. However, the first results are promising to develop relations between adsorption on PAC and molecular structure by means of molecular connectivity indexes. CONCLUSION This study has shown the effiCiency of adsorption on fibrous activated carbon in water treatment, the performance of this new material being better in terms of adsorption kinetic and capacity. Furthermore, micropollutants adsorption by FAC seems selective, due to the high microporosity of fibers, and to the fact that micropores act as a molecular sieve for macromolecules like humic materials. These properties, capacity and selectivity, may allow the creation of new reactors which could be smaller and/or combine FAC and another treatment process such as ultraflltration to remove humic materials (Brasquet et al., 1996). Furthermore, the QSAR study will be completed with more data, to fmd an equation which could predict adsorption of a great number of Chemicals. This equation could also facilitate the creation of new reactors by avoiding laboratory experiments. ACKOWLEDGMENTS The authors thank Renee Hoeffer, Drexel University, for fruitful suggestions for the editing of this text. REFERENCES Weber.Ir. W. I .• Asce. A. M. Carrell Morris, I. (1964).J. Sanll. Engn. Div. 90(SA3),79·107. Economy. I. and Lin. R. Y. (1976). Applied Polymer Symposium. 29.199·211. Cal. M. P.• Larson, S. M. and Rood, M. J. (1994). Envlron.Progress. 13(1),26-30. Lc C\oirec. P., Brasquet, C. and Subrenat, E. (1996). Am. Chem. Soc. Congress. New Orleans. 41(\).379·384. Thwaites. M. W.• Stewart. M. L .• McNeese. B. E. and Sumner. M. B. (1993). Fuel Process. Technol .• 34,137·145. Baudu, M., Lc C\oirec. P. and Martin. O. (l99\a). Wal. Sci. Tech .. 13(7·9). 1659·1665. Baudu. M•• Le Cloirec.P. and Manin O. (l991b). Rev. Sci. Dq'..1.111·123. Brasquet. C., Roussy, I .. Subrenat, E. and Le Cloirec. P. (1996). Env. Technol., 17. 1245·1252. Ryu, S. K. (1990). High temperatures·High pressures. n. 345·354. Ninnalakhandan. N. and Speece R. (19g8a). Environ. Sci. Technol .• 12(6). 606-615. Blum. D. J. W. and Suffet, I. H. (1989). 1st Macau Workshop in Water Treatment, Macau.

Fibrous activated carbon of organics

259

Blum, D. J. W., Suffet, I. H. and Duguet J. P. (1994). Wat. Res.,lS(3), 687-699. NumaIakhandan, N. N. and Speece, R. E. (1990). Environ. Sci. Technol.. 14(4), 575-580. Kier, L. B., Di Paolo, T. and Hall, L. H. (1977). J. Theor. 8iol., 67, 585-595. Kier, L. B. and Hall, L. H. (1986). Molecular connectivity in structure-activity analysis, Research Studies Press Ltd, 162 pp. Ninnalakhandan, N. N. and Speece, R. E. (l988b). Environ. Sci. Ttchnol., 11(11), 1349-1357. Ninnalakhandan, N. N. and Speece, R. E. (1988c). Environ. Sci. Technol.,U(3), 328-338. Bansal, R. C., DoMet, J. B. and Stoeckli, F. (1988). Active Carbon, Marcel Dekker Inc.. 482 pp. Le Cloirec, P.. Brasquet, C. and Subrenat. E. (1997). Energy" Fuels, in press. Le Cloirec, P., Le Lacheur, R. M., Johnson, J. D. and Chrisbllan, R. F. (1990). Wat. Res.,14(9),1l51-1155. Daley, M A., Mangun, C. L. and Economy, J. (1996). Am. Chern. Soc. Congress. New Orleans, 41(1),326-330. Pimenov, A. V., Lieberman, A. I., Shmidt, J. L. and Cheh. H. Y. (1995). Separation Science and Technology, 30(16), 3183-3194. Christman, R. F. and Gjessing, E. T. (1983). Aqua/ic and Terrestrial Humic Materials. Ann Arbor Science Publishers, 538 pp. Kilduff. J. E., KaranfIS, T., Chin, Y. P., Weber, Jr. W. J. (1996a). Environ. Sci. Technol., 30(4),1336-1343. Kilduff. 1. E.. KaranfIS, T. and Weber, Jr. W. J. (1996). Environ. Sci. Technol., 30(4), 1344-1351. Starck, 1., Zujal, A. and Rathousky, J. (1994). Carbon, 31(2), 207-211. Abe, I., Hayashi, K., Kitagawa, M. /Uld Hirashima, T. (1983). 8ull. Chem. Soc. Jpn, 56(4),1002-1005.