TCE adsorption

by GAC preloaded
with humic substances
Molecular weight makes a difference—
lower-molecular-weight substances reduce
TCE adsorption.

G
James E. Kilduf f,
Tanju Karanfil, and
Walter J. Weber Jr.

ranular activated carbon (GAC)

adsorption h as been iden tified by th e Safe Drin kin g
Water Act am en dm en ts of 1986 as th e best available
tech n ology for rem ovin g syn th etic organ ic con tam -
in an ts (SOCs) from drin kin g water su pplies.1
GAC is p a r t icu la r ly
su ited for treatin g water
A d so rp tio n o f trich lo ro e th yle n e (TCE) by activate d carbo n becau se it effectively re-
p re lo ad e d w ith h u m ic an d fu lvic acid s w as stu d ie d u n d e r se ve ral m oves a w ide ran ge of
co n d itio n s in co m p le te ly m ix e d batch syste m s. Th e au th o rs organ ic com pou n ds an d
in ve stigate d h o w m o le cu lar w e igh t an d m o le cu lar-w e igh t can protect water supplies
d istribu tio n o f p re lo ad e d h u m ic su bstan ce s affe cte d su bse qu e n t from su ch tran sien t oc-
ad so rp tio n o f TCE. Th e cap acity o f carbo n to ad so rb TCE w as cu rren ces as occasion al
m o st gre atly re d u ce d in carbo n th at w as p re lo ad e d w ith h u m ic spills an d season al load-
acid co m p o n e n ts h avin g m o le cu lar w e igh ts le ss th an abo u t 1,400 ings of toxic or potentially
g/ m o l as p o lystyre n e su lfo n ate . Th e ad so rp tio n cap acity w as carcinogenic compounds.
gre atly re d u ce d in carbo n th at w as p re lo ad e d w ith w h o le h u m ic How ever, n atu rally oc-
m ix tu re s in w h ich lo w e r m o le cu lar w e igh ts p re d o m in ate d . Th e cu rrin g, m acrom olecu lar
e n e rgy d istribu tio n s o f ad so rbe n t in d icate th at p re lo ad e d dissolved organ ic m atter
co m p o u n d s p re fe re n tially o ccu p y h igh -e n e rgy site s, m ak in g th e m
in acce ssible to su bse qu e n tly e n co u n te re d TCE. For executive summary,
see page 166.

Copyright (C) 1998 American Water Works Association

76 VOLUME 9 0 , ISSUE 5 JOURNAL AWWA
 Unused carbon as received
from the manufacturer
and carbon preloaded
with high-molecular-weight
fractions of dissolved organic
matter outperformed carbon
preloaded with smaller-
molecular-weight fractions.

t e m a t ic st u d ie s h a ve
so u gh t t o id e n t ify t h e
mechanisms of DOM pre-
loading or the factors that
con tribu te to ch an ges in
adsorben t ch aracteristics
resu ltin g from preloaded
organic matter. In a study
of preloadin g effects on
activated carbon exposed
to DOM in Hu ron River
w ater, it w as con clu ded
th at at early preloadin g
tim es, redu ction s in GAC
capacity for trich loroeth -
yle n e ( TCE) w e r e p r i-
(DOM) can greatly redu ce th e efficien cy an d com - marily a result of direct competition with the DOM for
plicate th e design of activated carbon adsorbers th at adsorption sites.13 However, at exten ded preloadin g
rem ove SOCs. DOM com pon en ts can m ove rapidly tim es, it was fou n d th at pore blockage m ay be th e
th rou gh fixed-bed adsorbers u sed in treatm en t appli- m ech an ism th at redu ces capacity th e m ost. To date,
cations, fouling the carbon before the pollutant passes n o stu dy h as addressed h ow th e m olecu lar weigh t
th rou gh . Becau se of th eir prevalen ce an d relative of DOM affects preloadin g. Molecu lar w eigh t h as
stability, humic substances are thought to be the DOM been sh ow n to in flu en ce th e adsorption of h u m ic
com pon en ts th at m ost redu ce SOC adsorption .2–4 an d fu lvic acids com posin g DOM. 20 Th is research
Th e effects of backgrou n d DOM on ad-
sorption processes an d process m odelin g
h ave been exten sively in vestigated. 3–13
A number of studies have shown that the FIG U R E 1 Adsorption of TCE by activated carbon as received
fouling or “preloading” phenomenon can from manufacturer
gr e a t ly r e d u ce bo t h t h e e q u ilibr iu m
capacity of GAC for SOCs 7,10,12–15 an d TCE adsorption data
the rate of SOC adsorption, most likely as Langmuir–Freundlich isotherm model fit
100
a result of increased resistance to diffusion
within particles.7,8,10,16 Some researchers,
h ow ever, h ave observed n o effect of
DOM on equ ilibriu m adsorption .17
TCE Adsorbed—µg/mg GAC

Variou s approach es h ave been devel-

oped to in corporate th e effects of DOM
into modeling strategies, including (1) de-
termining equilibrium and rate parameters 10
for th e h om ogen eou s su rface diffu sion
model by treatin g DOM as a system-spe-
cific but uncharacterized background com-
pon en t,6,7,9,10,16 (2) in corporatin g an ex-
pression for a tim e-variable isoth erm
capacity or rate parameters into the pore-
su rface diffu sion m odel, 8 ,1 8 an d (3)
em ployin g su ch ph ysical m odels as th e 1
rapid sm all-scale colu m n test to predict 1 10 100 1,000 10,000
adsorber breakth rou gh .4,15,19 TCE Concentration—µg/L
Alth ou gh preloadin g ph en om en a are
becom in g better u n derstood, few sys-

Copyright (C) 1998 American Water Works Association

MAY 1 9 9 8 JAMES E. KILDUFF ET AL 77
 Humic materials were dissolved in deion-
FIG U R E 2 Adsorption of TCE by a commercial activated carbonize d , d ist ille d , a n d filt e r e d w a t e r,†
preloaded with humic acid fractions and by carbon
adjusted to pH 10 (using sodium hydrox-
as received from manufacturer
ide [NaOH]), an d stirred for a m in im u m
Langmuir–Freundlich isotherm model fit for carbon of on e h ou r. Th e solu tion pH was th en
as received from manufacturer adju sted to abou t 7.0 (u sin g h ydroch lo-
3–10 K 17 mg/g >100 K 17 mg/g
10–30 K 16 mg/g >3 K 15 mg/g
ric acid [HCl]), dilu ted to a desired stock
30–100 K 18 mg/g con cen tration , an d sequ en tially filtered
100 th rou gh a 10-µ m polyeth ylen e filter, a
1-µ m glass fiber filter, an d a 0.45-µ m
polysu lfon e filter.‡ Stock solu tion s were
kept in dark refrigerated storage u n til
TCE Uptake—µg/mg GAC

use. The solution pH was checked prior to

adsorption experim en ts an d adju sted to
10
7.0 ± 0.1 with HCl or NaOH as necessary.
Stock solutions of reagent-grade (pes-
t icid e - fr e e ) TCE§ w e r e p r e p a r e d in
m eth an ol. Workin g solu tion s were pre-
pared by spikin g backgrou n d water with
stock solu tion to ach ieve th e desired in i-
tial con cen tration of solu te. Stock con -
1 ce n t r a t io n s a n d sp ik e vo lu m e s w e r e
1 10 100 1,000 10,000 design ed to m ain tain m eth an ol con cen -
TCE Concentration—µg/L trations sufficiently low (below 10 –3 mole
fraction ) to preclu de an y cosolven t effect
on TCE adsorption .21,22
Th e in organ ic m atrix for all solu tion s
in vestigated th e effects of th e m olecu lar weigh ts of consisted of 10 –3 -molar ( M) phosphate§ to buffer pH,
preloaded h u m ic an d fu lvic acids on th e su bsequ en t 0.0015- M sodiu m azide** to con trol biological activ-
adsorption of TCE by activated carbon . ity, an d sodiu m ch loride to ach ieve an ion ic stren gth
of 0.01 M. All ch em icals were reagen t grade. Experi-
Approach m en ts were con du cted at room tem peratu re (23 ±
Th e effect of m olecu lar weigh t was in vestigated 3 o C) at pH 7.0 ± 0.2; pH was adju sted by eith er HCl
by preloadin g th e carbon with differen t h u m ic frac- or NaOH as requ ired.
tion s prepared by u ltrafiltration (UF). Mixtu res of Act ivat ed carbon. Carbon obtain ed from th e
th ese fraction s were u sed to in vestigate in teraction s manufacturer†† was crushed and mechanically sieved
am on g h u m ic fraction s an d th e effects of in itial m ol- to a u n iform particle size (mean diameter of 165 µ m)
ecu lar-weigh t distribu tion . Th e effects of preloadin g u sin g US stan dard sieves. Carbon was wash ed, son -
com plex m ixtu res of h u m ic su bstan ces were in vesti- icated to rem ove fin es,23 oven -dried at 105 o C to con -
gated u sin g an adsorptive fraction ation tech n iqu e in stan t weigh t, an d stored in a desiccator u n til u se. Th e
w h ich differen t adsorben t dosages ( Do m g GAC/ L) sam e particle size (165 µ m ) was u sed for both pre-
were equ ilibrated with a ran ge of in itial h u m ic con - loadin g an d TCE isoth erm experim en ts.
centrations. In these experiments, although the initial Humic fractionation. Becau se h u m ic su bstan ces
m olecu lar-w eigh t distribu tion w as con stan t, th e occur naturally as mixtures of compounds with a dis-
adsorbed-m olecu lar-weigh t distribu tion varied. Mol- tribu tion of m olecu lar weigh ts,20 two fraction ation
ecular-weight distributions of UF fractions and whole tech n iqu es were em ployed to ch aracterize adsorbed-
h u m ic solu tion s were m easu red u sin g h igh -perfor- m olecu lar-w eigh t distribu tion s. A UF system w as
m a n ce size - e x clu sio n ch r o m a to gr a p h y. 2 0 TCE u sed to fraction ate th e m ixtu re in to distin ct m olecu -
isoth erm s w ere m easu red on preloaded carbon in lar-weight fractions, using an approach described pre-
com pletely m ixed batch reactors (CMBRs) an d com - viously.24 The UF system‡‡ incorporated hydrophilic,
pared with TCE adsorption on n on preloaded carbon . n eu tral, cellu losic-type spiral-w ou n d m em bran es
TCE isoth erm s were m odeled u sin g th e two-param e- h avin g n om in al m olecu lar-weigh t cu toffs of 100K,
ter Freundlich and the three-parameter Langmuir–Fre- 30K, 10K, an d 3K. Th e average m olecu lar weigh t of
u n dlich isoth erm s. Th e effects of preloadin g an d of each fraction w as determ in ed by size exclu sion
molecular weight were evaluated in terms of changes ch rom atograph y, calibrated w ith polystyren e su l-
in th e TCE isoth erm s an d com pu ted site-en ergy dis- fon ate (PSS) stan dards, as described in detail else-
tribu tion s. wh ere.20 Each su ccessive fraction possessed a sign if-
icantly lower polydispersity than the original solution.
Methods and materials Th e secon d approach was based on an adsorptive
Sorbates. Hu m ic an d fu lvic acids extracted from fraction ation ph en om en on in wh ich th e preferen tial
a Lau ren tian soil* w ere received as a dry pow der. adsorption of low-m olecu lar-weigh t fraction s from

Copyright (C) 1998 American Water Works Association

78 VOLUME 9 0 , ISSUE 5 JOURNAL AWWA
solu tion fraction ates h u m ic an d fu lvic
FIG U R E 3 Adsorption of TCE by activated carbon preloaded
acid m ixtu res accordin g to m olecu lar
with humic acid fractions of various molecular
weigh t.20,25 Th u s, th e m olecu lar-weigh t weights—effect on Freundlich capacity parameter KF
distribu tion of adsorbed m acrom olecu les
was con trolled by preloadin g th e carbon
u n der differen t com bin ation s of in itial 1.5
h u m ic su bst a n ce co n ce n t r a t io n a n d
adsorben t dose.20 1.4
Humic substance preloading. Most
of th e preloadin g was con du cted in 1-L

TCE Freundlich KF
1.3
am ber-colored glass bottles m ixed on a
rotary tu m bler for an operation al equ ili-
1.2
bration period of 30 days. After th e equ i-
libration period reactors w ere sam pled
with glass syrin ges, an d th e solu tion was 1.1
95 percent confidence interval
filtered th rou gh 0.45-µ m polysu lfon e fil- for carbon as received from manufacturer
ter.§§ Macrom olecu le con cen tration s 1.0
were qu an tified by total organ ic carbon
(TOC) an alysis*** an d UV spectroph o- 0.9
tom etry.††† Activated carbon was recov- 1,000 10,000 100,000
ered from preloaded reactors by vacu u m Molecular Weight of Preloaded Fraction
filtration on a 0.2-µ m filter an d wash ed. Error bars indicate 95 percent confidence intervals for preloaded carbon.
Wash in g or m ech an ically scou rin g th is
particu lar GAC did n ot in an y way m iti-
gate th e effects of preloadin g.12 Th e car-
bon w as dried u n der vacu u m at room FIG U R E 4 Adsorption of TCE by activated carbon preloaded
tem peratu re for 14 days, after wh ich th e with a <3-K humic acid fraction at various DOC loadings
m oistu re con ten t was below m easu rable
levels. Sam ples were stored in a desicca- Langmuir–Freundlich isotherm model fit
tor u n der vacu u m . 12.1 mg DOC/g 25.3 mg DOC/g 39.2 mg DOC/g
100
In a second method, some carbon was
preloaded directly in th e 250-m L bottles
u sed for th e TCE isoth erm experim en ts.
TCE Adsorbed—µg/mg GAC

Carbon as received
First, th e adsorben t was con tacted with from manufacturer
a h umic acid solu tion for a desired period
ran gin g from zero to 30 days. After th e
preloadin g period, TCE was spiked in to 10
th e reactor an d allowed to equ ilibrate for
an addition al two weeks. Preloadin g iso-
therms for dichloroethene were compared
u sin g th ese two tech n iqu es for preparin g
preloaded carbon ; th e isoth erm s w ere
iden tical regardless of wh eth er th e car-
bon was u sed wet or vacu u m -dried.12 1
Adsorpt ion isot he rms. Iso t h e r m 1 10 100 1,000 10,000
TCE Concentration—µg/L
experim en ts w ere con du cted u sin g th e
CMBR m eth od. Carbon was equ ilibrated
with TCE solutions in 250-mL amber-col-
ored glass bottles sealed with screw caps
an d PTFE-lin ed silicon e septa. Reactors were filled extended periods without causing measurable changes
com pletely an d sealed with n o h eadspace. Th e reac- in solu tion ph ase con cen tration s.
tors were kept well m ixed by tu m blin g en d-over-en d After th e equ ilibration period, carbon w as set-
on a rotary tu m bler for two weeks. Prelim in ary rate tled, reactors w ere sam pled w ith glass pipettes, an d
studies indicated that this time was sufficient to reach *Fredrik’s Research Produ cts, Am sterdam , th e Neth erlan ds
equ ilibriu m based on con stan t liqu id-ph ase TCE con - †MQ-water, Millipore In c., Bedford, Mass.
cen tration s. Th is equ ilibriu m state sh ou ld be con sid- ‡Gelm an Scien ces, An n Arbor, Mich .
§Mallin ckrodt Specialty Ch em ical, Paris, Ky.
ered operation al becau se it is likely th at tru e equ i- **Flu ka AG, Fu ch s, Switzerlan d
libriu m was n ot ach ieved. Becau se m acrom olecu le ††Calgon F400, Calgon Carbon Corp., Pittsbu rgh , Pa.
diffu sion an d su rface rearran gem en t is slow an d ‡‡Model CH2RPS, Am icon , Dan vers, Mass.
§§Su por, Gelm an Scien ces, An n Arbor, Mich .
becau se preloadin g cau ses a decrease in TCE in tra- ***TOC 500, Sh im adzu Corp., Colu m bia, Md.
particle diffusion rates, TCE transport may proceed for †††Varian Optical, Victoria, Au stralia

Copyright (C) 1998 American Water Works Association

MAY 1 9 9 8 JAMES E. KILDUFF ET AL 79
 TAB LE 1 Humic acid preloading with ultrafiltration-size fractions

Langmuir–Freundlich Isotherm Freundlich Isotherm Parameter

Preloading Conditions Parameter Values and Statistics Values and Statistics

Preloading Co qe Parameter 95 Percent Parameter 95 Percent

Fraction MW* mg TOC/ L mg/ g Parameter Estimate† CI‡ Q0 bn Parameter Estimate† CI‡

No pre lo ading 0 .0 0 .0 Q0 X 1 0 –2 1 .8 3 1 .0 7 , 2 .5 9 1 .2 8 KF 1 .2 6 1 .1 9 , 1 .3 4
b X 104 1 .7 0 0 .0 0 , 3 .4 8 n 0 .5 7 0 .5 5 , 0 .5 9
n 0 .5 7 0 .5 4 , 0 .6 0
<3 K 1 ,4 0 0 8 .0 1 2 .1 Q0 X 1 0 –2 1 .7 6 1 .4 0 , 2 .1 1 1 .0 6 KF 1 .0 4 0 .9 8 , 1 .1 1
b X 104 1 .7 5 0 .9 2 , 2 .5 7 n 0 .5 9 0 .5 7 , 0 .6 1
n 0 .5 9 0 .5 8 , 0 .6 1
3 –1 0 K 2 ,7 0 0 1 6 .9 1 6 .7 Q0 X 1 0 –2 1 .3 4 1 .0 1 , 1 .6 7 1 .2 5 KF 1 .2 3 1 .1 7 , 1 .3 0
b X 104 3 .6 4 1 .3 2 , 5 .9 6 n 0 .5 9 0 .5 7 , 0 .6 1
n 0 .5 9 0 .5 7 , 0 .6 1
1 0 –3 0 K 3 ,7 0 0 3 5 .4 1 6 .4 Q0 X 1 0 –2 1 .1 3 0 .8 9 , 1 .3 7 1 .2 4 KF 1 .2 4 1 .1 7 , 1 .3 2
b X 104 6 .0 6 2 .7 2 , 9 .4 0 n 0 .5 8 0 .5 6 , 0 .6 1
n 0 .6 1 0 .5 8 , 0 .6 3
3 0 –1 0 0 K 6 ,7 0 0 4 0 .9 1 8 .1 Q0 X 1 0 –2 1 .1 7 0 .9 1 , 1 .4 2 1 .2 7 KF 1 .2 5 1 .1 7 , 1 .3 7
b X 104 4 .9 4 2 .0 7 , 7 .8 0 n 0 .5 8 0 .5 5 , 0 .6 1
n 0 .5 9 0 .5 7 , 0 .6 2
>1 0 0 K 2 2 ,0 0 0 1 8 .9 1 6 .6 Q0 X 1 0 –2 1 .1 6 0 .8 9 , 1 .4 3 1 .4 1 KF 1 .4 0 1 .3 3 , 1 .4 7
b X 104 4 .3 2 1 .5 3 , 7 .1 2 n 0 .5 6 0 .5 3 , 0 .5 8
n 0 .5 7 0 .5 5 , 0 .5 9
>3 K >1 ,4 0 0 3 6 .9 1 4 .6 Q0 X 1 0 –2 1 .3 9 0 .9 9 , 1 .7 9 1 .2 3 KF 1 .2 0 1 .1 2 , 1 .2 8
b X 104 3 .2 3 0 .8 9 , 5 .5 7 n 0 .5 9 0 .5 6 , 0 .6 1
n 0 .5 9 0 .5 7 , 0 .6 1

* MW— mo le c ular we ight de te rmine d by s ize -e xc lus io n c hro mato graphy c alibrate d with po lys tyre ne s ulfo nate s tandards (g/ mo l as PSS)
† Is o the rm parame te rs c o rre s po nd to uptake in units o f µg/ mg and c o nc e ntratio n in units o f µg/ L.
‡ CI— c o nfide nc e inte rval

sam ples w ere extracted in to h exan e for qu an tifica- based on a sym m etric, qu asi-Gau ssian site en ergy
tion by gas ch rom atograph y u sin g electron captu re distribution, and its parameters can be shown to relate
detection .* to properties of th is distribu tion .27,28 In th is con text,
Th e gas ch rom atograph w as calibrated u sin g Q0 represen ts m axim u m adsorption capacity, b rep-
extern al stan dards prepared in aqu eou s solu tion an d resen ts an average site en ergy, an d n represen ts th e
extracted in to h exan e u sin g th e sam e procedu re as h eterogen eity of site en ergies.
for isoth erm sam ples. Precision for TCE an alysis, In m an y adsorption stu dies reported in th e liter-
in clu din g th e extraction step, was 3 percen t or bet- atu re, th e classical Freu n dlich isoth erm equ ation is
ter. Potential losses resulting from volatilization, sorp- u sed to m odel adsorption data th rou gh ou t ran ges of
tion on to reactor com pon en ts, or bacterial activity con cen tration of in terest in en gin eerin g application s:
were assessed with con trol reactors th at did n ot con -
tain adsorben t. Becau se losses in th ese experim en ts qe = KFCen (2)
were con sisten tly less th an 5 percen t, n o correction s
were m ade for th em . in wh ich KF is a u n it-capacity param eter, equ al to
Modeling adsorption equilibria. Isoth erm data th e am ou n t adsorbed at a valu e of Ce equ al to u n ity
were collected over a wide ran ge of con cen tration (in wh atever con cen tration u n its ch osen ), an d n is a
and displayed some curvature on log–log coordinates, dim en sion less param eter related to th e site-en ergy
su ggestin g th e n eed for a th ree-param eter isoth erm distribu tion .27,28 Th e Freu n dlich equ ation represen ts
m odel to accu rately describe th e data. Several su ch th e low con cen tration lim it of th e LF isoth erm equ a-
m odels were fit to represen tative experim en tal data tion ; th at is, for bCe <<1:
u sin g a n on lin ear least-squ ares regression an alysis.26
Based on a residual sum-of-squares criterion for good- qe = Q0 bn Cen = KFCen (3)
ness-of-fit, the three-parameter Langmuir–Freundlich
(LF) m odel was ch osen : The Freundlich unit capacity parameter, KF, incor-
porates param eters represen tin g m axim u m capacity,
Q0 ( bCe) n
average site en ergy, an d site-en ergy h eterogen eity.
qe = ᎏᎏ (1) It was possible to accu rately fit data in wh ich Ce <
1 + ( bCe) n 100 µ g/ L with th e Freu n dlich m odel. An advan tage
of th e Freu n dlich m odel in th is regard is th at its two
in wh ich qe is th e am ou n t adsorbed at equ ilibriu m param eters can be estim ated with greater statistical
(n orm alized to adsorben t m ass); Ce is th e equ ilibriu m sign ifican ce th an th e th ree param eters of th e LF.
solu tion ph ase con cen tration ; an d Q0 , b, an d n are Sit e-energy dist ribut ions. Approxim ate site-
em pirically determ in ed param eters. Th e LF m odel is en ergy distribu tion s were calcu lated from isoth erm

Copyright (C) 1998 American Water Works Association

80 VOLUME 9 0 , ISSUE 5 JOURNAL AWWA
data to relate observed ch an ges (resu lt-
FIG U R E 5 Adsorption of TCE by activated carbon preloaded with a
in g from preloadin g) in experim en tally <3-K humic acid fraction—effect on isotherm parameters
m easu red isoth erm s to ch an ges in th e
en ergetic ch aracteristics of sorbate–sor- 1.4 0.75
ben t in teraction s. Som e effects of pre-
loadin g m ay resu lt becau se preloaded
solu tes block access to som e m icrop- 1.2
ores. 13 Su ch ch an ges m ay be reflected 0.70
ph en om en ologically in th e adsorption

TCE Freundlich KF

TCE Freundlich n
isotherm; the effects of these changes are 1.0
th erefore in clu ded im plicitly in th e site- 0.65
en ergy distribu tion .
Th e con den sation approxim ation 29 0.8
was u sed to gen erate approxim ate site-
e n e r gy d ist r ibu t io n fu n ct io n s. Th e 0.60
m eth od resu lts in a stable, an alytical ex- 0.6
pression th at does n ot requ ire an y prior
assu m ption s abou t th e sh ape of th e dis-
0.4 0.55
tribu tion . 30 As discu ssed in detail else- 0.0 10.0 20.0 30.0 40.0
28
where, the method involves taking the Humic Acid Preloading (<3 K fraction)—mg DOC/g GAC
fir st d e r iva t ive o f iso t h e r m d a t a . A
sm ooth in g fu n ction was applied to th e
experim en tal data before differen tiatin g
to elim in ate variability th at m ay cau se
spu riou s or n egative peaks w h en derivatives are sim ilar cu rvatu re th at w as also described w ell by
taken.30 Although any function could have been cho- th e LF m odel.
sen to sm ooth th e data, th e LF m odel was ch osen Isoth erm m odel param eters are tabu lated in Table
becau se it described th e data accu rately. Specifyin g a 1. LF param eters were estim ated by fittin g th e en tire
fu n ction al form for th e isoth erm is n ot requ ired to data set, an d Freu n dlich param eters were estim ated
apply th e m eth od, an d th e u se of an isoth erm fu n c- by fittin g th e low con cen tration data (<100 µ g/ L
tion to fit th e data does n ot m ake an y prior assu m p- TCE). For com parison , th e Freu n dlich u n it capacity
tion abou t th e sh ape of th e site-en ergy distribu tion . factors calcu lated from th e th ree LF param eters ( KF =
Ho w e ve r, be ca u se t h e LF m o d e l w a s u se d a s a Q0 bn ) are also reported; th e agreem en t is gen erally
smoothing function, the approximate site-energy dis- excellen t.
tribu tion s are fu n ction s w ritten in term s of m odel Preloading with UF-size fractions. The effects of
param eters:28 m acrom olecu le m olecu lar weigh t on adsorben t pre-
loadin g an d th e su bsequ en t adsorption of TCE were
0
Q n ( bCs) n –nE* in vestigated by preloadin g a com m ercial carbon †
F( E*) = ᎏᎏ
RT 冢
exp ᎏᎏ
RT冣 with differen t UF fraction s. Th ese fraction s are des-
ign ated accordin g to th e n om in al m em bran e m olec-
(4)
– nE * –2 u lar-weigh t cu toff valu es: <3 K, 3–10 K, 10–30 K,
冢 冢
X 1 + ( bCs) n exp ᎏᎏ 冣冣
RT 30–100 K, an d >100 K.
Th e actu al cu toffs are n om in al, becau se th e m ol-
ecu lar-weigh t cu toff is based on 90 percen t rejection
in wh ich E* represen ts th e n et adsorption en ergy, Cs of globu lar protein s w ith stru ctu res differen t from
is th e sorbate solu bility, R is th e u n iversal gas con - th ose of HAs. Fu rth erm ore, th e m em bran e pore sizes
stan t, T is th e absolu te tem peratu re, an d th e rem ain - are n ot u n iform . Th erefore, m olecu les h avin g a size
in g term s are as previou sly defin ed. Th e n et adsorp- n ear th e average pore size will be presen t in m ore
t io n e n e r gy, w h ich w a s co m p u t e d u sin g a th an on e fraction . Th e fraction design ated <3 K con -
Polan yi-type expression , accou n ts for th e en ergy tain ed com pon en ts th at passed th e 3-K m em bran e.
of solven t displacem en t. Th ese valu es sh ou ld be Th e th ree fraction s design ated 3–10 K, 10–30 K, an d
view ed as approxim ate an d u sed for com parison 30–100 K con tain com pon en ts th at passed th e larger
pu rposes. m em bran e an d w ere retain ed on th e sm aller on e.
Com pon en ts retain ed on th e 100-K m em bran e are
Results and discussion design ated >100 K.
Th e adsorption of TCE on n on preloaded carbon Correspon din g m olecu lar w eigh ts are given in
w as ch aracterized to provide a basis for assessin g Table 1. Carbon w as also preloaded w ith a m ix-
ch an ges in TCE isoth erm s resu ltin g from preloadin g. tu re, design ated >3-K, prepared by rem ovin g th e
TCE u ptake sh ow s a sligh t cu rvatu re (on log–log
coordinates) that was described well by the LF model *Model 5890, Hewlett Packard, Palo Alto, Calif.
(Figu re 1). Uptake by preloaded carbon s exh ibited †F400, Calgon Corp., Pittsbu rgh , Pa.

Copyright (C) 1998 American Water Works Association

MAY 1 9 9 8 JAMES E. KILDUFF ET AL 81
 TAB LE 2 Humic acid preloading with the <3-K ultrafiltration-size fraction

Preloading Conditions Langmuir–Freundlich Isotherm Freundlich Isotherm

Preloading Co Ce / Do qe Parameter 95 Percent Parameter 95 Percent

Time mg TOC/ L g/ g mg/ g Parameter Estimate* CI† Q0 bn Parameter Estimate* CI†

Ze ro -day‡ 4 .1 NA§ NA Q0 ˘ 1 0 –2 1 .4 3 1 .1 6 , 1 .7 0 0 .7 5 KF 0 .7 5 0 .7 1 , 0 .7 9
b ˘ 104 3 .3 0 1 .8 9 , 4 .7 0 n 0 .6 5 0 .6 3 , 0 .6 7
n 0 .6 5 0 .6 4 , 0 .6 7
1 4 -day‡ 4 .1 7 6 .8 2 5 .3 Q0 ˘ 1 0 –2 1 .3 7 0 .5 0 , 2 .2 3 0 .7 5 KF 0 .7 3 0 .6 2 , 0 .8 6
b ˘ 104 2 .8 5 0 .0 0 , 6 .6 8 n 0 .6 4 0 .5 8 , 0 .7 0
n 0 .6 4 0 .5 9 , 0 .6 8
3 0 -day 4 .1 6 5 .0 3 9 .2 Q0 ˘ 1 0 –2 0 .8 8 0 .6 9 , 1 .0 8 0 .4 7 KF 0 .4 6 0 .4 4 , 0 .4 9
b ˘ 104 6 .4 8 3 .3 8 , 9 .5 9 n 0 .7 1 0 .6 9 , 0 .7 3
n 0 .7 1 0 .6 9 , 0 .7 4
3 0 -day* * 8 .0 2 .3 2 1 2 .1 Q0 ˘ 1 0 –2 1 .7 6 1 .4 0 , 2 .1 1 1 .0 6 KF 1 .0 4 0 .9 8 , 1 .1 1
b ˘ 104 1 .7 5 0 .9 2 , 2 .5 7 n 0 .5 9 0 .5 7 , 0 .6 1
n 0 .5 9 0 .5 8 , 0 .6 1

* Is o the rm parame te rs c o rre s po nd to uptake in units o f µg/ mg and c o nc e ntratio n in units o f µg/ L.
† CI— c o nfide nc e inte rval
‡ Equilibrium was no t ac hie ve d in the s ample s pre lo ade d fo r le s s than 3 0 days ; thus the c o nc e ntratio ns and ads o rbe d amo unts do no t re pre s e nt o pe ratio nal
e quilibrium value s .
§ NA— no t analyze d
* * Data fo r the 3 0 -day s ample with q e = 1 2 .1 mg TOC/ g als o appe ar in Table 1 as the <3 -K frac tio n.

<3-K fraction from th e w h ole h u m ic m ixtu re. Th e redu ced adsorption of TCE. Th e carbon preloaded
preloadin g experim en ts w ere design ed to com pare w ith th e >3-K m ixtu re, prepared by rem ovin g th e
th e effects of differen t, relatively n arrow m olecu - <3-K fraction from th e wh ole h u m ic acid, adsorbed
lar-weigh t ran ges at sim ilar TOC loadin g. An adsor- as m u ch TCE as did n on preloaded carbon , con -
ben t dose of 0.556 g/ L w as u sed for all experi- firm in g th is in terpretation .
m en ts, an d th e in itial h u m ic acid con cen tration On e explan ation for th e observed tren d is th at
w as adju sted to yield sim ilar adsorbed am ou n ts. larger humic molecules cannot physically occupy car-
Th e in itial con cen tration s for all bu t th e sm allest bon su rfaces presen t in sm all pores. Su pport for su ch
fraction were h igh er th an typically fou n d in n atu ral an explan ation follows from a com parison of carbon
system s. pore dim en sion s with th e sizes of h u m ic m olecu les.
Carbon preloaded w ith all fraction s
except th e <3-K fraction , at loadin gs
ran gin g from 14.6 to 18.1 m g TOC/ g, FIG U R E 6 Approximate site-energy distributions calculated from
adsorbed as m u ch TCE as did n on pre- TCE isotherm data—effect of preloading with a <3-K
loaded carbon . Th is con clu sion is su g- humic acid fraction
gested by visu al in spection of th e TCE
isoth erm s sh ow n in Figu re 2, an d it is Carbon as received from manufacturer
12.1 mg DOC/g 25.3 mg DOC/g 39.2 mg DOC/g
con firm ed by th e statistics for both th e
10
LF a n d Fr e u n d lich iso t h e r m m o d e l
param eters (Table 1). How ever, as in di-
ca t e d in bo t h Ta ble 1 a n d Figu r e 3 ,
w h ich sh ow s th e effects of preloadin g
on th e Freu n dlich u n it capacity factor,
KF, th e carbon preloaded w ith 12.1 m g
TOC/ g of th e <3-K fraction adsorbed a
statistically sign ifican t lesser am ou n t of
F(E*)

1
TCE. A sligh t in crease in th e Freu n dlich
n valu e w as observed for all fraction s
except th e >100-K fraction (Table 1),
alth ou gh th e in crease w as n ot statisti-
cally sign ifican t. Th e resu lts dem on -
strated tw o th in gs: (1) alth ou gh th ere
w as sign ifican t u ptake of th e h igh er-
m olecu lar-w eigh t m aterial, it did n ot 0.1
com pete effectively w ith TCE an d (2) 10 15 20 25 30 35 40
th e low -m olecu lar-w eigh t com pon en ts E*—kJ/mol
of th e h u m ic solu tion m ost effectively

Copyright (C) 1998 American Water Works Association

82 VOLUME 9 0 , ISSUE 5 JOURNAL AWWA
 TAB LE 3 Humic acid preloading—effect of initial size distribution

Preloading Conditions Langmuir–Freundlich Isotherm Freundlich Isotherm

Initial
10–30-K Mass Co Ce / Do qe Parameter 95 Percent Parameter 95 Percent
percent mg TOC/ L g/ g mg/ g Parameter Estimate* CI† Q0 bn Parameter Estimate* CI†

0 .0 ‡ 8 .0 2 .3 2 1 2 .1 Q0 X 1 0 –2 1 .7 6 1 .4 0 , 2 .1 1 1 .0 6 KF 1 .0 4 0 .9 8 , 1 .1 1
b X 104 1 .7 5 0 .9 2 , 2 .5 7 n 0 .5 9 0 .5 7 , 0 .6 1
n 0 .5 9 0 .5 8 , 0 .6 1
5 3 .9 1 6 .2 1 7 .3 1 6 .3 Q0 X 1 0 –2 1 .4 6 1 .0 7 , 1 .8 4 1 .1 3 KF 1 .1 0 1 .0 5 , 1 .1 5
b X 104 2 .7 2 0 .9 9 , 4 .4 6 n 0 .6 0 0 .5 8 , 0 .6 1
n 0 .5 9 0 .5 7 , 0 .6 1
7 3 .5 2 3 8 .4 9 .3 Q0 X 1 0 –2 1 .3 8 0 .0 4 , 2 .7 1 1 .1 1 KF 1 .0 9 1 .0 3 , 1 .1 4
b X 104 3 .0 3 0 .0 0 , 1 1 .1 n 0 .6 0 0 .5 7 , 0 .6 2
n 0 .6 0 0 .5 1 , 0 .6 8
1 0 0 .0 3 5 .4 5 1 .4 1 6 .4 Q0 X 1 0 –2 1 .1 3 0 .8 9 , 1 .3 7 1 .2 4 KF 1 .2 4 1 .1 7 , 1 .3 2
b X 104 6 .0 6 2 .7 2 , 9 .4 0 n 0 .5 8 0 .5 6 , 0 .6 1
n 0 .6 1 0 .5 9 , 0 .6 3

* Is o the rm parame te rs c o rre s po nd to uptake in units o f µg/ mg and c o nc e ntratio n in units o f µg/ L.
† CI— c o nfide nc e inte rval
‡ Data fo r 0 pe rc e nt 1 0 –3 0 -K frac tio n (i.e ., 1 0 0 pe rc e nt <3 -K frac tio n) and 1 0 0 pe rc e nt 1 0 –3 0 -K frac tio n are als o give n in Table 1 .

A large fraction of th e total su rface area of th e acti- preven tin g access to carbon su rfaces. However, th is
vated carbon u sed in th is stu dy is con tain ed in pores preloading mechanism did not appear to be operative
h avin g radiu ses less th an abou t 10 Å. Alth ou gh m ea- for th e h igh -m olecu lar-weigh t h u m ic m aterials.
su rem en ts of h u m ic su bstan ce m olecu lar weigh t an d Th e precedin g an alysis su ggests th at th e effect of
size depend in part on the analytical techniques used, preloaded n atu ral organ ic m atter on TCE adsorption
an d alth ou gh th e relation sh ip betw een m olecu lar ca n be a sse sse d by m e a su r e m e n ts o f m o le cu la r
weigh t an d m olecu lar size depen ds on solu tion ion ic w eigh t. Previou s stu dies13 also fou n d m arked pre-
stren gth , pH, divalen t m etal ion con cen tration , an d loadin g effects of Hu ron River w ater. Th e average
tem peratu re, th ere is eviden ce th at h u m ic m olecu le m olecu lar w eigh t of organ ic m atter in th is w ater
sizes are of th e sam e order of m agn itu de. w as abou t 1,000 g/ m ol as PSS, su pportin g th e rela-
Hu m ic su bstan ces separated by u ltracen trifu ga- tion sh ip between m olecu lar weigh t an d th e effect of
tion were found to have radiuses of gyration of 15, 20, preloadin g in th is case. Based on th e dem on strated
and 32 Å, which correspond
t o m o le cu la r w e igh t s o f
2,600, 4,400, an d 12,800,
respectively.31 Th e radiu ses
of gyration of h u m ic su b-
stan ces, estim ated by sm all-
an gle X-ray scatterin g to be
on th e order of 5 to 33 Å,
correspon d to a m olecu lar-
C
omponents of DOM can move rapidly
through fixed-bed adsorbers used to
treat drinking water, fouling the carbon
w e igh t r a n ge o f 5 0 0 t o
before the pollutant passes through.
>10,000 g/ m ol. 32 Radiu ses
of gyration for aqu atic fu lvic su bstan ces ran ged from reactivity of th e sm aller-m olecu lar-w eigh t m ole-
4.5 to 8.8 Å, wh ich correspon d to m olecu lar weigh ts cu les, addition al preloadin g experim en ts w ere per-
ran gin g from abou t 500 to 1,000 g/ m ol.33 Th e <3-K form ed w ith th e <3 K fraction to fu rth er evalu ate
fraction h ad a weigh t-averaged m olecu lar weigh t of th e effects of preloaded organ ic m atter on TCE
1,400 g/ m ol as PSS, determ in ed by size exclu sion adsorption .
ch rom atograph y.20 Th is m olecu lar weigh t com pares Preloading with the <3-K UF-size fraction. Four
to th at of 2,700 for th e 3–10-K fraction an d abou t preloadin g experim en ts, in clu din g th e experim en t
4,000 to 22,000 for th e larger fraction s. discu ssed in th e previou s section , u sed th e <3-K frac-
Th u s, it is likely th at larger h u m ic fraction s were tion to in vestigate th e effects of TOC loadin g an d
preven ted from occu pyin g a large percen tage of th e equ ilibration tim e. Experim en tal con dition s are
carbon su rface area an d cou ld n ot com pete directly sh own in Table 2. In two experimen ts, th e <3-K frac-
w ith TCE for adsorption sites. Based on eviden ce tion w as preloaded for a fu ll 30-day equ ilibration
th at preloaded h u m ic m olecu les can ph ysically block time. Carbon preloaded in these experiments reached
pores, 13 it wou ld be reason able to expect th at pre- an operation al equ ilibriu m con dition ( Cfinal ⬵ Ce). In
loaded h u m ic m aterial m igh t redu ce TCE u ptake by on e experim en t, design ated zero-day, th e adsorben t
Copyright (C) 1998 American Water Works Association
MAY 1 9 9 8 JAMES E. KILDUFF ET AL 83
 ter n for all Freundlich and LF isotherms
FIG U R E 7 Adsorption of TCE by activated carbon preloaded with in creased, an d th e in crease was statis-
various percentages of 10–30-K humic acid fraction tically sign ifican t for all bu t th e lowest
in mixtures also containing a <3-K humic acid fraction—
level of preloadin g. Th e average site-
effect on Freundlich capacity parameter
en ergy param eter b appeared to in -
1.5 crease som ewh at at h igh levels of pre-
lo a d in g, bu t t h e in cr e a se w a s n o t
statistically sign ifican t, an d n o stron g
1.4 95 percent confidence interval for
carbon as received from manufacturer tren d is eviden t. Th e Freu n dlich u n it-
capacity param eters com pu ted from LF
param eters an d th ose estim ated by fit-
TCE Freundlich KF

1.3
tin g th e Freu n dlich isoth erm m odel to
1.2 the low concentration data are in excel-
le n t a gr e e m e n t . Th e Fr e u n d lich K F
valu e decreased by a statistically sign if-
1.1
ican t am ou n t at all preloadin g con di-
tion s. A plot of Freu n dlich isoth erm
1.0 param eters KF an d n as a fu n ction of
h u m ic acid preloadin g is sh own in Fig-
0.9
u re 5. Th ese resu lts are con sisten t with
0 25 50 75 100 th e experim en tal resu lts reported for
Mass Percent of 10–30-K Fraction in Initial Mixture TCE adsorption on carbon preloaded
Error bars indicate 95 percent confidence intervals for preloaded carbon. with Hu ron River water an d syn th etic
polym aleic acid, 13 an d th ey provide
insight into the mechanisms responsible
for the observed changes in TCE uptake.
was exposed to TCE an d h u m ic acid sim u ltan eou sly, In creases in th e isoth erm h eterogen eity param e-
an d it equ ilibrated for two weeks. In a fin al experi- ter n reflected a decrease in th e n u m ber of h igh -
m en t th e carbon was preloaded for two weeks. en ergy sites. 28 Th is decrease in h igh -en ergy sites
TCE isoth erm s m easu red on carbon preloaded tended to make the surface less heterogeneous and to
with th e <3-K fraction are sh own in Figu re 4. Th e reduce uptake in the low-concentration region of the
sim u ltan eou s adsorption isoth erm is n ot sh ow n isoth erm . Th e decrease in u ptake was reflected in a
becau se it is sim ilar to th e isoth erm m easu red for th e decrease in th e Freu n dlich KF valu e. Site-en ergy dis-
carbon preloaded for 14 days. A sm aller redu ction in tribu tion s were com pu ted to illu strate h ow ch an ges
TCE u ptake for th e sim u ltan eou s adsorption experi- in isoth erm param eters reflect ch an ges in th e en er-
ment was expected, although it has been documented getic ch aracter of th e adsorben t su rface (Figu re 6).
that the presence of humic acid can reduce the extent At lower preloadin g con cen tration s, alm ost all of th e
o f TCE a d so r p t io n w h e n t h e t w o so r ba t e s a r e loss in capacity was th e resu lt of a redu ction in th e
adsorbed sim u ltan eou sly.13 Th e in terpretation of th is n u m ber of h igh -en ergy sites. As preloadin g con cen -
result is made more difficult because the TOC loading tration s in creased, sites w ere lost th rou gh ou t th e
in the simultaneous adsorption experiment could not en tire ran ge of site en ergies; h owever, th e greatest
be determ in ed accu rately
becau se TCE an d a sm all

T
am ou n t of m eth an ol were
presen t in th e system . It is
possible th at th e <3-K frac- he smaller-molecular-weight hydrophobic
tion contains relatively low-
m olecu lar-w eigh t com po- fraction is preferentially adsorbed
n en ts th at do n ot h ave to from mixtures.
be preloaded to com pete
effectively with TCE.
TCE capacity was greatly redu ced th rou gh ou t th e relative redu ction was still am on g h igh -en ergy sites.
en tire ran ge of con cen tration as a resu lt of h u m ic Alth ou gh th e m arked loss of low-en ergy sites at h igh
acid preloadin g (Figu re 4). However, becau se th e rel- preloadin g con cen tration s su ggests a pore blockage,
ative redu ction was greater at lower con cen tration s, th e data are n ot con clu sive in th is regard.
th e isoth erm slope as plotted on log–log coordin ates Th e data reported in Table 2 are con sisten t with
in creased. LF an d Freu n dlich isoth erm m odel fits to th e fin din gs of Carter et al,13 wh o developed em pir-
the data are tabulated in Table 2. The LF parameter Q0 ical relationships to describe how adsorption isotherm
decreased with increasing TOC loading, and the Q0 of (an d rate) param eters ch an ged with preloadin g tim e.
th e 30-day preloaded sam ple decreased by a statisti- Alth ou gh th is approach is u sefu l for dyn am ic m od-
cally sign ifican t am ou n t. Th e h eterogen eity param e- elin g, th e effects of preloadin g sh ou ld relate fu n da-

Copyright (C) 1998 American Water Works Association

84 VOLUME 9 0 , ISSUE 5 JOURNAL AWWA
 TAB LE 4 Humic acid preloading—effect of sequential loading*

Preloading Conditions Langmuir–Freundlich Isotherm Freundlich Isotherm

Preloading Co Ce / Do qe Parameter 95 Percent Parameter 95 Percent

Order mg TOC/ L g/ g mg/ g Parameter Estimate† CI‡ Q0 bn Parameter Estimate† CI‡

1 . <3 K 8 .0 6 .1 7 .8 Q0 X 1 0 –2 1 .1 4 0 .8 8 , 1 .3 9 1 .0 1 KF 0 .9 8 0 .9 2 , 1 .0 5
2 . 1 0 –3 0 K 1 9 .7 2 7 .3 8 .1 b X 104 4 .7 6 2 .0 5 , 7 .4 7 n 0 .6 1 0 .5 9 , 0 .6 4
To tal 1 5 .9 n 0 .6 2 0 .6 0 , 0 .6 4
1 . 1 0 –3 0 K 3 5 .4 5 2 .1 1 3 .1 Q0 X 1 0 –2 1 .5 4 1 .2 8 , 1 .7 9 1 .1 6 KF 1 .1 4 1 .1 0 , 1 .2 0
2 . <3 K 8 .9 6 .1 1 0 .3 b X 104 2 .4 3 1 .4 2 , 3 .4 5 n 0 .5 8 0 .5 7 , 0 .6 0
To tal 2 3 .4 n 0 .5 9 0 .5 7 , 0 .6 0

* Pre lo ading data fo r the <3 -K and 1 0 –3 0 -K frac tio ns alo ne are give n in Table 1 .
† Is o the rm parame te rs c o rre s po nd to uptake in units o f µg/ mg and c o nc e ntratio n in units o f µg/ L.
‡ CI— c o nfide nc e inte rval

m en tally to th e m ass (or m olar) loadin g of th e pre- an d sm aller fraction s did n ot in teract. Un der th ese
loaded com pou n d. Th e effects of m ass (TOC) loadin g h ypoth eses, th e effects of preloadin g sh ou ld be in de-
an d preloadin g tim e are recon ciled by recogn izin g pen den t of th e in itial m olecu lar-weigh t distribu tion
th at th e rate of m acrom olecu le adsorption can be (i.e., the fraction of the initial TOC contributed by the
slow. It may take months for macromolecules to attain 10–30-K fraction ) as lon g as th e m ass of th e <3-K
tru e equ ilibriu m , particu larly for th e particle sizes fraction per u n it m ass of adsorben t rem ain s con -
u sed in fu ll-scale adsorbers. Th e data for th e 14-day st a n t . Th is h yp o t h e sis is su p p o r t e d by t h e TCE
an d 30-day experim en ts, each of wh ich started with isoth erm data (Table 3 an d Figu re 7). Statistically
the same initial TOC concentration adsorbent dosage, sign ifican t redu ction s in th e Freu n dlich KF valu e
illu strate th at TOC loadin g can in crease sign ifican tly were observed as lon g as th e <3-K fraction was pre-
du rin g relatively lon g peri-
ods of tim e, cau sin g fu rth er

H
redu ction s in su bsequ en t
TCE adsorption .
Preloading wit h mix- umic substances are thought to be the
t ure s of UF f ra ct ions.
Carbon was preloaded with DOM components that most reduce
m ixtu res of th e <3-K an d
10–30-K fraction s to deter-
the adsorption of synthetic organic
m in e wh eth er larger-m ole- contaminants.
cu lar-w eigh t com pon en ts
cou ld in teract with sm aller,
m ore reactive com pon en ts an d to in vestigate th e sen t in solu tion , in depen den t of th e in itial m olecu -
effect of in itial m olecu lar-weigh t distribu tion . Exper- lar-w eigh t distribu tion (i.e., th e presen ce of u p to
im en tal con dition s are tabu lated in Table 3. In th is 74 percen t of th e 10–30-K fraction by m ass). Wh en
experim en t, th e m ass of th e <3-K fraction in solu - no <3 K fraction was present, however, no significant
tion (m g TOC) per u n it m ass of adsorben t w as h eld redu ction in th e Freu n dlich KF valu e was observed.
approxim ately con stan t, an d th e in itial TOC con - Th ese resu lts con firm th at th e sm aller-m olecu lar-
cen tration of th e preloadin g solu tion w as in creased weigh t fraction was preferen tially adsorbed, th at th e
by in creasin g th e TOC con tribu ted by th e 10–30-K effects of preloadin g were attribu table prim arily to
fraction. Therefore, increases in the amount adsorbed th is fraction , an d th at th e larger-m olecu lar-weigh t
cou ld on ly be du e to th e 10–30-K fraction , an d su ch com pon en ts presen t in th e 10–30-K fraction did n ot
in cr e a se s w o u ld co r r e sp o n d t o a sh ift in t h e sign ifican tly displace or oth erwise in teract with th e
adsorbed-m olecu lar-w eigh t distribu tion to h igh er sm aller com pon en ts.
m olecu lar w eigh ts. A sequential adsorption experiment further inves-
On the basis of results previously presented, it was tigated poten tial displacem en t or oth er in teraction
h ypoth esized th at redu ction s in TCE u ptake wou ld mechanisms. In this experiment, the <3-K and 10–30-
be cau sed by preloadin g of th e <3-K fraction . It was K fractions were loaded in two sequential stages (“pri-
fu rth er h ypoth esized th at th e u ptake of th is fraction m ary” an d “secon dary”) to determ in e wh eth er th e
wou ld rem ain con stan t becau se (1) th e in itial con - order of preloadin g in flu en ced su bsequ en t TCE
centration of the <3-K fraction was held constant; (2) adsorption . In on e experim en t, th e carbon was first
th e <3-K m olecu les are adsorbed preferen tially, as equ ilibrated with th e <3-K fraction , after wh ich th e
su ggested by size-exclu sion ch rom atograph y an aly- su pern atan t w as decan ted; th e carbon w as su bse-
sis of h u m ic su bstan ce adsorption ;20 an d (3) larger qu en tly equ ilibrated with th e 10–30-K fraction . In

Copyright (C) 1998 American Water Works Association

MAY 1 9 9 8 JAMES E. KILDUFF ET AL 85
 o f p r e lo a d in g by sm a lle r- m o le cu la r-
FIG U R E 8 Adsorption of TCE by activated carbon preloaded
with whole humic acid at various DOC loading rates
w eigh t com pon en ts. For fraction ation
to occu r alon g a GAC bed, organ ic m at-
Preloading by Whole Humic Acid ter with a polydisperse m olecu lar-weigh t
Loading Ce/D0 KF /KF 0 distribu tion w ou ld be requ ired. How -
mg DOC/g g/g ever, som e n atu ral w ater (e.g., Hu ron
11.8 14.6 0.76
23.7 81.0 0.51 River water) h as relatively n arrow m ol-
45.3 167.0 0.31 ecu lar-weigh t distribu tion s an d m ay n ot
100.0
exh ibit fraction ation ph en om en a.
Preloading by adsorptive fraction-
Carbon as received
from manufacturer ation. Experiments evaluated TCE uptake
TCE Adsorbed—µg/mg GAC

a ft e r ca r bo n w a s p r e lo a d e d w it h t h e
10.0 wh ole h u m ic m ixtu re. Preloadin g exper-
im en ts u sin g UF fraction s sh ow ed th at
sm aller-m olecu lar-w eigh t com pon en ts
are respon sible for redu ction s in TCE ad-
sorption . On th e basis of th ese resu lts,
1.0
Kildu ff et al h ypoth esized th at w h en a
ran ge of m olecu lar weigh ts are presen t
in solu tion , greater redu ction s in TCE
adsorption capacity wou ld be observed
0.1 when the adsorbed-molecular-weight dis-
1 10 100 1,000 10,000 tribu tion was sh ifted to sm aller m olecu -
TCE Concentration—µg/L lar weigh ts. Th is h ypoth esis was tested
by e q u ilibr a t in g ca r bo n sa m p le s t o
ach ieve a ran ge of Ce/Do ratios (from
abou t 5 to 170 m g DOC/ g), creatin g dif-
th e secon d experim en t, th e order of equ ilibration feren t adsorbed-m olecu lar-weigh t distribu tion s as a
was reversed. Experimental conditions for the sequen- resu lt of adsorptive fraction ation . Adsorptive frac-
t ia l lo a d in g e x p e r im e n t a n d LF a n d Fr e u n d lich tionation occurs because low-molecular-weight com-
isoth erm m odel fits to th e data are given in Table 4. pon en ts are adsorbed preferen tially from a m ixtu re,
Wh en th e <3-K fraction w as loaded before th e as dem on strated previou sly. 20 Large valu es of th e
10–30-K fraction , th e KF valu e for TCE was greatly Ce/Do ratio sh ift th e adsorbed-m olecu lar-weigh t dis-
reduced. In contrast, no major change occurred when tribu tion to sm aller weigh ts; sm all valu es of th is ratio
th e order w as reversed, even at a fairly h igh total sh ift th e adsorbed distribu tion tow ard th at of th e
loadin g of 23.4 m g TOC/ g. Adsorption of th e larger origin al solu tion .
fraction effectively preven ted su bsequ en t adsorption A sign ifican t advan tage of preloadin g by adsorp-
of th e <3-K fraction from com petin g with TCE. Th is tive fraction ation is th at preloaded m olecu les are
im plies th at th e <3-K fraction was u n able to occu py adsorbed from th e wh ole h u m ic solu tion ; th erefore,
th e sm aller pores th at it cou ld access in th e absen ce any competition among components will be revealed.
of th e 10–30-K fraction . Preloadin g by adsorptive fraction ation h as on e sig-
Neith er fraction appeared to sign ifican tly displace n ifican t lim itation , h owever. An in crease in th e Ce/Do
th e oth er. Con sisten t with th e fin din gs of Su m m ers ratio is accom pan ied by an in crease in th e m ass of
and Roberts,25 the lower-molecular-weight <3-K frac- th e sm all-m olecu lar-weigh t fraction relative to th e
tion may have a lower intrinsic affinity for the surface m ass of adsorben t. Th erefore, sh ifts in th e adsorbed-
th an th e 10–30-K fraction an d th erefore cou ld n ot m olecu lar-weigh t distribu tion are accom pan ied by
displace this component. Although the 10–30-K frac- ch an ges in th e TOC loadin g.
tion m ay h ave a h igh er in trin sic affin ity for th e su r- Th e design of preloadin g con dition s w as based
face, it probably could not reach surfaces on which the on h u m ic an d fu lvic acid adsorption isoth erm s, 20
<3-K fraction w as adsorbed, th ereby m akin g dis- an d a su m m ary of preloadin g con dition s for both
placem en t im possible. Th e lack of displacem en t m ay th e h u m ic an d fu lvic acids is given elsewh ere.26 Data
also relate to steric h in dran ces of both th e poten tial for TCE u ptake by preloaded carbon were fit with
displacer an d th e adsorbed species. th e LF an d Freu n dlich isoth erm s. Represen tative
Th ese fin din gs h ave im plication s for th e opera- isoth erm s w ere ch osen to illu strate th e observed
tion of fixed-bed adsorbers. If n atu ral organ ic m at- tren ds. Isoth erm s m easu red on carbon preloaded
ter is fraction ated by m olecu lar w eigh t alon g th e w ith am ou n ts of h u m ic m aterial sim ilar to th ose
GAC bed, th en sm aller, m ore adsorbable com po- u sed in th e <3-K fraction preloadin g experim en ts
n en ts w ill be rem oved earlier in th e bed. Larger were selected to provide a basis for com parin g th e
com pon en ts w ill pen etrate deeper in to th e bed, two sets of experimen ts. Isoth erm data an d LF model
loadin g th e carbon ah ead of th e sm aller com po- fits are sh own in Figu re 8, an d best-fit m odel para-
n en ts. Th is process w ill effectively lim it th e effects m eters for th e represen tative data are tabu lated in

Copyright (C) 1998 American Water Works Association

86 VOLUME 9 0 , ISSUE 5 JOURNAL AWWA
 TAB LE 5 Humic acid preloading by adsorptive fractionation

Preloading Conditions Langmuir–Freundlich Isotherm Freundlich Isotherm

Co Do Ce / Do qe Parameter 95 Percent Parameter 95 Percent

mg TOC/ L g/ L g/ g mg/ g Parameter Estimate* CI† Q0 bn Parameter Estimate* CI†

2 7 .4 1 1 .0 5 1 4 .6 1 1 .8 Q0 X 1 0 –2 1 .7 0 0 .7 3 , 2 .6 7 0 .9 5 KF 0 .9 5 0 .8 2 , 1 .0 9
b X 104 1 .8 2 0 .0 0 , 4 .2 6 n 0 .6 0 0 .5 5 , 0 .6 5
n 0 .6 0 0 .5 6 , 0 .6 4
5 5 .4 2 0 .5 3 8 1 .0 2 3 .7 Q0 X 1 0 –2 1 .1 0 0 .9 4 , 1 .2 6 0 .6 3 KF 0 .6 0 0 .5 6 , 0 .6 5
b X 104 4 .0 9 2 .6 4 , 5 .5 3 n 0 .6 7 0 .6 4 , 0 .7 1
n 0 .6 6 0 .6 4 , 0 .6 8
5 5 .4 2 0 .2 6 167 4 5 .3 Q0 X 1 0 –2 0 .7 9 0 .6 6 , 0 .9 1 0 .3 9 KF 0 .3 8 0 .3 5 , 0 .4 0
b X 104 6 .6 8 4 .3 6 , 9 .0 0 n 0 .7 3 0 .7 1 , 0 .7 6
n 0 .7 3 0 .7 0 , 0 .7 5

* Is o the rm parame te rs c o rre s po nd to uptake in units o f µg/ mg and c o nc e ntratio n in units o f µg/ L.
† CI— c o nfide nc e inte rval

Table 5. Isoth erm m odel param eters for a wide ran ge Th e sh ift in th e adsorbed-m olecu lar-weigh t dis-
of adsorptive fraction ation preloadin g experim en ts tribu tion to larger m olecu lar w eigh ts at low Ce/Do
are tabu lated elsewh ere. 26 ratios m ean s th at a large fraction of th e in itial solu -
Th e isoth erm s m easu red on carbon preloaded by tion was adsorbed, in clu din g n early all of th e low-
adsorptive fraction ation sh ow tren ds sim ilar to th ose molecular-weight fraction. The low-molecular-weight
observed on carbon preloaded with th e <3-K frac- fraction was th erefore presen t on th e adsorben t su r-
tion . Th e greatest relative redu ction in TCE capacity face at n early all valu es of Ce/Do (alth ou gh at differ-
o ccu r r e d in t h e lo w - co n ce n t r a t io n r e gio n ; t h e en t loadin gs), yet sign ifican t decreases in TCE capac-
isoth erm slope wh en plotted on log–log coordin ates ity were n oted on ly at h igh valu es of th is param eter.
sh o w s a co n co m it a n t in cr e a se . A co n sid e r a ble Two explan ation s for th is beh avior are possible. It is
decrease in capacity occu rred th rou gh ou t th e con - possible that at low Ce/Do ratios, the surface loading or
centration range examined, as was also observed with adsorption den sity was too low to affect TCE u ptake.
the <3-K fraction. The isotherms exhibit marked cur- A secon d explan ation is th at th e presen ce of larger-
vatu re on log–log coordin ates an d are well-described m olecu lar-weigh t com pon en ts redu ced th e ability of
by th e LF isoth erm . Th e LF param eter Q0 decreased low-molecular-weight fractions to compete with TCE.
with in creasin g TOC loadin g, an d at a loadin g of 45.3 The results of preloadin g experiments using mixtures
mg TOC/g, the decrease was statistically significant. All of UF fraction s su ggests th at th e first explan ation is
isoth erms in creased in th eir h eterogen eity parameter m ore likely.
n for both the Freundlich and LF isotherm models; the Th is con clu sion is su pported by th e depen den ce
in crease w as statistically sign ifican t for all bu t th e of KF/KFo on th e am ou n t of h u m ic m aterial adsorbed
low est (11.8 m g TOC/ g) TOC loadin g. Statistically (Figu re 9). Th e greatest redu ction s in TCE capacity
sign ifican t decreases in th e Freu n dlich u n it capacity occu rred at th e h igh est h u m ic loadin gs, w h ich cor-
param eter were fou n d for all TOC loadin gs. respon d to h igh Ce/Do ratios an d a sh ift in th e m ol-
The effects of preloading were expressed in terms of ecu lar-w eigh t distribu tion of adsorbed h u m ic su b-
th e relative Freu n dlich capacity param eter, KF/KFo, stan ces to sm aller m olecu lar w eigh ts. At th e h igh er
defined as the ratio of the KF parameter obtained on h u m ic loadin gs, m ost of th e preloaded m aterial was
preloaded carbon to th at obtain ed on virgin carbon . expected to be com posed of low -m olecu lar-w eigh t
This parameter can be interpreted as the fraction of the com pon en ts. Th is com position w as expected be-
original adsorbent capacity remaining after preloading. cau se th e percen tage of DOC rem oved by adsorption
Representative values and corresponding values of Ce/Do w as sim ilar to th e percen tage of <3-K com pon en ts
are shown in Figure 8. Under the hypothesis that greater com posin g th e w h ole h u m ic acid solu tion (sm aller
reductions in TCE adsorption occur when the adsorbed- com pon en ts w ere adsorbed preferen tially). Th u s,
molecular-weight distribution is shifted to smaller mol- th e effects of preloadin g by w h ole h u m ic acid an d
ecules, the relative Freundlich capacity parameter should by th e <3-K fraction w ere expected to be sim ilar,
decrease with increasing values of Ce/Do. This hypothe- an d th e data are con sisten t w ith th is expectation
sis is supported by the data depicted in Figure 8, and sim- (Figu re 9).
ilar trends were observed for the fulvic acids studied. In con trast, at lower h u m ic acid loadin gs th e per-
These results are consistent with the results of the UF cen tage of DOC rem oved was greater th an th e per-
fraction preloading experiments. Moreover, they demon- cen tage of low-m olecu lar-weigh t com pon en ts. At a
strate that molecular-weight distribution largely deter- loadin g of abou t 12 m g/ g, for exam ple, less th an h alf
mines both the extent of humic substance adsorption th e DOC rem oved from solu tion was com posed of
an d th e su bsequ en t effect on low-m olecu lar-weigh t low -m olecu lar-w eigh t com pon en ts. Th u s, it w as
synthetic organic contaminants. expected th at th e effect of preloadin g by th e <3-K

Copyright (C) 1998 American Water Works Association

MAY 1 9 9 8 JAMES E. KILDUFF ET AL 87
 decrease in h igh -en ergy sites ten ded to
FIG U R E 9 Effect of the Ce / D0 ratio on the TCE capacity remaining redu ce u ptake in th e low-con cen tration
after preloading, expressed in terms of the Freundlich
region of th e isoth erm , as reflected in a
capacity parameters before (KF0 ) and after (KF)
preloading
decrease in th e Freu n dlich KF valu e. As
preloading concentrations increased, sites
Humic acid Fulvic acid Humic acid (<3 K) were lost th rou gh ou t th e en tire ran ge of
1.2
site en ergies.
In sequential adsorption experiments,
1.0 w h en th e low -m olecu lar-w eigh t com -
ponents were loaded first, they exerted a
0.8 m arked preloadin g effect; fu rth erm ore,
th ey were n ot displaced by larger-m ole-
KFKF0

0.6 cu lar-weigh t com pon en ts presen t in th e

10–30-K fraction . How ever, w h en th e
0.4 10–30-K fraction was loaded first, it was
n ot displaced by th e <3-K fraction , an d
0.2 n o preloadin g effect was observed. Pre-
loadin g experimen ts with syn th etic mix-
0.0
tu res of UF fraction s dem on strated th at
0.0 10.0 20.0 30.0 40.0 50.0 th e effects of preloadin g were in depen -
Amount Preloaded—mg DOC/g GAC den t of th e in itial m olecu lar-weigh t dis-
An increasing Ce/D0 ratio corresponds to a shift in the adsorbed tribu tion as lon g as th e am ou n t of th e
molecular-weight distribution to smaller molecular weights. <3-K fraction per m ass of adsorben t was
h eld con stan t. Th ese resu lts sh ow th at
th e sm aller-m olecu lar-weigh t fraction is
preferen tially adsorbed from m ixtu res
fraction wou ld be greater th an th at obtain ed by pre- an d th at th e effects of preloadin g were cau sed pri-
loadin g wh ole h u m ic acid. However, th e observed marily by this fraction. Adsorptive fractionation exper-
effects were sim ilar. Th e reason s for th is beh avior are im en ts u sin g wh ole h u m ic an d fu lvic acid solu tion s
not apparent, but several possibilities can be put forth. con firm ed th e effects of preloadin g from syn th etic
It is also possible th at th ere is som e syn ergistic effect m ixtu res. Th e greatest redu ction s in TCE capacity
th at m akes th e m ixtu re m ore reactive th an th e su m occu rred at th e h igh est h u m ic loadin gs, wh ich cor-
of its parts. On e su ch effect m igh t be attribu ted to a respon d to h igh Ce/Do ratios an d a sh ift in th e m ole-
h igh e r p a ck in g d e n sit y o f h u m ic co m p o n e n t s cu lar-w eigh t distribu tion of preloaded h u m ic su b-
adsorbed from a m ixtu re as com pared with isolated stan ces to sm aller weigh ts.
fraction s. On th e oth er h an d, becau se th e effects at
low h u m ic loadin gs are n ot large, it m ay n ot be pos- Acknowledgment
sible to estim ate isoth erm param eters with su fficien t Th is work was fu n ded in part by Research Gran t
precision to clearly distin gu ish preloadin g effects of CES-8702786 from th e Nation al Scien ce Fou n dation
wh ole h u m ic solu tion s from th ose of isolated low- an d Gran t 2 P42ES04911-04 from th e Nation al In sti-
m olecu lar-weigh t fraction s. tu tes of Health . Th e au th ors th an k Pin g Yu an , Xiu
Min Yong, and Merideth Weiner for assistance in col-
Conclusions lectin g laboratory data. Th e au th ors also th an k th e
TCE adsorption on carbon preloaded with h u m ic Am erican Society of Civil En gin eers Research Fel-
acid m olecu lar-w eigh t fraction s dem on strated th e lowsh ip an d th e Tu rkish Scien tific Research Cou n cil
im portan ce of m olecu lar weigh t on th e effect of pre- for partial fin an cial su pport.
loadin g. Uptake by carbon preloaded with th e <3 K
fraction (molecular weight = 1,380 g/mol as PSS) was References
sign ifican tly redu ced com pared with adsorption by 1. Safe Drin kin g Water Act Am en dm en ts of 1986,
as-received carbon ; h owever, TCE adsorption by car- PL 99-339 (Ju n e 1986).
bon preloaded with larger-molecular-weight fractions 2. M OREL, F.M.M. Principles of Aquatic Chemistry.
(m olecu lar weigh t >3,000 g/ m ol as PSS) was statisti- Joh n Wiley an d Son s, New York (1983).
cally sim ilar. Tren ds in Freu n dlich isoth erm param e- 3. FETTIG, J. & SONTHEIMER, H. Kinetics of Adsorption
ters were consistent with experimental results reported on Activated Carbon : II. Mu ltisolu te System s.
for TCE adsorption on carbon preloaded with Hu ron Jour. Environ. Engrg ., 113:780 (1987).
River water and synthetic polymaleic acid.13 Changes 4. SUMMERS, R.S. ET AL. The Influence of Background
in com pu ted site-en ergy distribu tion s su ggest th at at Organic Matter on GAC Adsorption. Jour. AWWA ,
preloadin g con cen tration s, alm ost all of th e loss in 81:5:66 (May 1989).
capacity occu rred becau se of a redu ction in th e n u m - 5. LEE , M .C. ET AL. De sign o f Ca r bo n Be d s t o
ber of high-energy sites and is reflected in an increase Rem ove Hu m ic Su bstan ces. Jour. Environ. Engrg.
in th e isoth erm h eterogen eity param eter n . Th e Div.— ASCE, 109:631 (1983).

Copyright (C) 1998 American Water Works Association

88 VOLUME 9 0 , ISSUE 5 JOURNAL AWWA
 6. PIRBAZARI, M.P. & W EBER, W.J. J R. Rem oval of 22. CURTIS, G.P. Sorption of Halogen ated Organ ic
Dieldrin From Water by Activated Carbon . Jour. Solu tes On to Aqu ifer Materials: Com parison s
Envir. Engrg., 110:656 (1984). Between Laboratory Resu lts an d Field Observa-
7 SMITH, E.J.; TSENT, S.; & W EBER, W.J. J R. Model- tions. Doctoral dissertation, Stanford Univ., Stan-
ing the Adsorption of Target Compounds by GAC ford, Calif. (1984).
in th e Presen ce of Backgrou n d Organ ic Matter. 23. SUMMERS, R.S. ET AL. Standardized Protocol for the
Envir. Progress, 6:18 (1987). Evalu ation of GA C. AWWARF, Den ver, Colo.
8. ZIM M ER , G. ET AL. Design Con sideration s for (1992).
Fixed-Bed Adsorbers Th at Rem ove Syn th etic 24. KILDUFF, J.E. & W EBER, W.J. J R. Tran sport an d
Organ ic Ch em icals in th e Presen ce of Natu ral Separation of Organ ic Macrom olecu les in Ultra-
Organ ic Matter. Proc. 1988 AWWA An n . Con f., filtration Processes. Envir. Sci. & Technol., 26:569
Orlan do, Fla. (1992).
9. SMITH, E.H. & W EBER, W.J. J R. Modelin g Acti- 25. SUMMERS, R.S. & ROBERTS, P.V. Activated Carbon
vated Carbon Adsorption of Target Organic Com- Adsorption of Hu m ic Su bstan ces, I. Heterodis-
pou n ds From Leach ate-Con tam in ated Grou n d- perse Mixtu res an d Desorption . Jour. Colloid &
waters. Envir. Sci. & Technol., 22:313 (1988). Interface Sci., 122:367 (1988).
10. SMITH, E.H. & W EBER, W.J. J R. Evaluation of Mass 26. KILDUFF, J.E. Polyelectrolyte Adsorption by Acti-
Tran sfer Param eters for Adsorption of Organ ic vated Carbon an d th e Effects of Preloadin g on
Com pou n ds From Com plex Organ ic Matrices. Trich oroeth ylen e Adsorption . Doctoral disserta-
Envir. Sci. & Technol., 23:713 (1989). tion , Un iv. of Mich igan , An n Arbor (1995).
11. W ILMANSKI, K. & VAN BREEMEN, A.N. Com peti- 27. D ERYLO -M ARCZW SKA , A. ET AL. Heterogen eity
tive Adsorption of Trich loroeth ylen e an d Hu m ic Effects in Sin gle-Solu te Adsorption From Dilu te
Substances From Groundwater on Activated Car- Aqu eou s Solu tion s on Solids. Chem. Scripta ,
bon . Water Res., 24:773 (1990). 24:239 (1984).
12. SPETH, T.F. Evalu atin g Capacities of GAC Pre- 28. CARTER, M.C.; KILDUFF, J.E.; & W EBER, W.J. J R.
loaded With Natu ral Water. Jour. Envir. Engrg., Sorbent Heterogeneity and the Effects of Sequen-
117:66 (1991). tial Solu te Uptake. Envir. Sci. & Technol., 29:1773
13. CARTER, M.C.; W EBER, W.J. J R.; & O LMSTEAD, K.P. (1995).
Effects of Backgrou n d Dissolved Organ ic Mat- 29. CEREFOLINI, G.F. Localized Adsorption on Hetero-
ter on TCE Adsorption by GAC. Jour. AWWA , geneous Surfaces. Thin Solid Films, 23:129 (1974).
73:8:81 (Au g. 1992). 30. NEDERLOF, M.M.; VAN RIEMSDIJK, W.J.; & KOOPAL,
14. ENDICOTT, D.D. & W EBER, W.J. J R. Lu m ped-Para- L.K. Determ in ation of Adsorption Affin ity Dis-
m eter Modelin g of Mu lticom pon en t Adsorption tribu tion s: A Gen eral Fram ework for Meth ods
in the Treatment of Coal-Conversion Wastewater Related to Local Isoth erm Approxim ation s. Jour.
by GAC. Envir. Progress, 4:105 (1985). Colloid & Interface Sci., 135:410 (1990).
15. SPETH, T.F. & M ILTNER, R.J. Effect of Preloadin g 31. C AM ERON , R.S. ET AL. Molecu lar Weigh t an d
on th e Scale-Up of GAC Microcolu m n s. Jour. Sh ape of Hu m ic Acid From Sedim en tation an d
AWWA , 81:4:141 (April 1989). Diffu sio n M e a su r e m e n t s o n Fr a ct io n a t e d
16. W EBER, W.J. J R. & PIRBAZARI, M. Adsorption of Extracts. Jour. Soil Sci., 23:4:394 (April 1972).
Toxic and Carcinogenic Compounds From Water. 32. THURMAN, E.M. ET AL. Molecu lar Size of Hu m ic
Jour. AWWA , 74:201 (1982). Su bstan ces. Organic Geochem ., 4:27 (1982).
17. BALDAUF, G. Ein flu ss Natü rlich er Organ ish er 33. AIKEN, G.R. & M ALCOLM , R.L. Molecu lar Weigh t
Wasserin h altsstoffe au f die Adsorption von of Aqu atic Fu lvic Acids by Vapor Pressu re Os-
Spurenstoffen in Aktivkohlenfiltern. Vom Wasser, m om etry. Geochim. et Cosmochim. Acta, 51:2177
66:21 (1986). (1987).
18. CARTER, M.C. & W EBER, W.J. J R. Modelin g Ad-
sorption of TCE by Activated Carbon Preloaded About the authors: J ames E. Kil-
by Background Organic Matter. Envir. Sci. & Tech- duff is an assistant professor in the
nol., 28:614 (1993). Department of Environmental and
19. CRITTENDEN, J.D. ET AL. Predictin g GAC Perfor- Energy Engineering at Rensselaer
m an ce With Rapid Sm all-Scale Colu m n Tests. Polytechnic Institute, 5049 JEC, 110
Jour. AWWA , 83:1:77 (Jan . 1991). Eigh th St., Troy, N Y 12180. He
20. KILDUFF, J.E. ET AL. Adsorption of Natural Organic received his PhD from the University of
Polyelectrolytes on Activated Carbon : A Size Mich igan (A n n A rbor) an d h as
Exclu sion Ch rom atograph y Stu dy. Envir. Sci. & worked in adsorption research for the past 11 years. Tanju
Technol., 30:1336 (1996). Karanfil is an assistant professor in the Environmental Sys-
21. W AUCHOPE, R.D.; SAVAGE K.E.; & KOSKINEN, W.C. tems Engineering Department at Clemson University, 342
Adsorption –Desorption Equ ilibria of Herbicides Computer Ct., Anderson, SC 29625-6510. Walter J. Weber Jr.
in Soil: Napth alen e as a Model Com pou n d for is a professor in the Department of Civil and Environmen-
En tropy–En th alpy Effects. Weed Sci ., 31:744 tal Engineering at the University of Michigan, 2200 Bonis-
(1983). teel Ave., Ann Arbor, MI 48109-2099.

Copyright (C) 1998 American Water Works Association

MAY 1 9 9 8 JAMES E. KILDUFF ET AL 89