United States
                     Environmental Protection
                     Agency
Municipal Environmental Research    v
Laboratory
Cincinnati OH 45268
                     Research and Development
EPA-600/S2-83-047  Aug. 1983
vvEPA          Project  Summary

                     Predicting  Preferential Adsorption
                     of Organics  by Activated  Carbon
                     Georges Belfort, Gordon L. Altshuler, Kusuma K. Thallam, Charles P. Ferrick,
                     Jr., and Karen L Woodfjeld
                       Preferential adsorption of organic
                     compounds from dilute aqueous solu-
                     tions onto activated carbon (AC) was
                     studied to develop a comprehensive
                     theoretical basis for predicting adsorp-
                     tion of multicomponent solutes. The
                     research  program investigates why
                     some solutes are strong adsorbers, and
                     others weak, and why some solutes
                     displace others during aqueous phase
                     adsorption. The overall objectives were
                     to develop, test, and simplify the theo-
                     retical basis for prediction.
                       The fundamental, multidimensional
                     approach  of the solvophobic thermo-
                     dynamics theory was used to correlate
                     the extent of adsorption for the com-
                     prehensive theory with the overall stan-
                     dard free energy change for the associa-
                     tion adsorption reaction in solution,
                     and for the simplified theory with the
                     cavity surface area of the solute.
                       Experimental adsorption isotherms
                     of two homologous series (alkyl phenols
                     and alkyl alcohols) were measured and
                     used to test  the theory.  Differences
                     resulting from simplestructural modifi-
                     cations of solutes were predicted theo-
                     retically and confirmed experimentally.
                     Several experimental innovations for
                     equilibrium adsorption studies have
                     been introduced to reduce solute loss
                     by extraneous adsorption and vaporiza-
                     tion.
                       Small negative  activation  energies
                     for intraparticle pore and surface diffu-
                     sion of alkyl phenols were also calcu-
                     lated from a temperature study.

                       This Project Summary was developed
                     by EPA's Municipal Environmental Re-
                     search Laboratory, Cincinnati, OH,  to
                     announce key findings of the research
                     project that is fully documented in a
                     separate report of the same title (see
Project Report ordering information at
back).

Background
  The ability to predict the effects of even
simple structural modifications on the ad-
sorption of organic molecules from dilute
aqueous solutions onto AC (or other ad-
sorbents)  could be of great value  in the
design and operation of large-scale com-
mercial water and wastewater treatment
plants.  Structural modifications such as
those found between isomers of the same
homologous series are  weU known to
make the difference between benign and
toxic compounds. At present, most theo-
retical approaches either rely on single-
solute isotherms  originally derived from
gas and vapor phase systems to predict
mixed-solute isotherms,  or they rely on
solubility theory.   Although Traube and
others more than 90 years ago recognized
the need to include the solvent interactions,
only recently have attempts to quantitize
these effects been made.
  Recent attempts to include the solvent
effect in aqueous-phase adsorption include
a semi-empirical approach based on  partial
solubility parameters and some arbitrarily
chosen  parameters called the net adsorp-
tion energy approach.  Another approach
called the "thick compressed film theory"
(or "Polanyi adsorption potential theory,"
as it is often called) has been  used to
describe adsorption  isotherm behavior.
The problem with this three-dimensional
adsorbed film  model is the difficulty in
defining the properties of the film and the
need to use "scaling factors" for construct-
ing the so-called characteristic equation.

Objectives
  The objectives of this study were:
  1.  Develop a comprehensive formalism
     of dilute aqueous-phase adsorption

-------
     of organics,  including fundamental
     formulations of all dominant inter-
     actions between solute, solvent, and
     sorbent;
  2. Test the use of this formalism to
     predict a ranking order of adsorption
     capacity of members of two homo-
     logous series;
  3. Measure experimentally the equilibri-
     um adsorption isotherms at constant
     pH and temperature for a statistically
     significant number of members of
     two homologous series;
  4. Propose and apply certain simplifying
     assumptions to the comprehensive
     model that would result, for special
     cases,  in a simplified analytical ex-
     pression;
  5. Compare correlations of adsorption
     capacity from the comprehensive and
     simplified theories with those ob-
     tained with other independent vari-
     ables such as the molecular weight,
     density, index  of refraction,  molar
     volume,  molar  refraction, octanol-
     water partition coefficient, parachor,
     and polanzability; and
  6. Examine the possibilities of using
     the derived  theory for predicting
     multicomponent adsorption.
  Objective 1 is fundamental to the other
objectives.  The theoretical formalisms
developed during  this study were tested
both with experimental data obtained from
the literature and that suggested m Objec-
tive  3.   Objective 4 is  crucial to the
practicability  and usefulness of the sug-
gested approach. Objectives 5 and 6 were
included because the usefulness of the
theory will eventually  depend on its ability
to rank-order the adsorption of solutes
with betterpredictability(i.e., higherlinear
coefficients of correlation) than other in-
dependent variables,  and from multicom-
ponent systems, respectively.

Materials and Methods
  Crushed granular activated carbon (PAC)
(U.S.  Sieve Series No. 200 to No. 400
range and 1031  m2/g)  was the major
adsorbent used in this study. Comparison
of adsorption isotherms of five alkyl phenols
on granular activated carbon and on graph-
itized carbon (89 m2/g) was undertaken.
Cleaning procedures  of the PAC were an
important aspect  in obtaining  reproduc-
ible isotherms.
  The adsorbates used included two homo-
logous series of 19 alkyl phenols and 1 2
aliphatic alcohols all of the highest grade
available (>99% purity). Stock solutions
of the individual compounds were made
up  with 0.01m  phosphate buffer and
stored in covered bottles at pH = 7.0. From
a practical  viewpoint, both homologous
series  are  commonly found  in  surface
wastewaters, i.e., alkyl phenols are widely
found in coal-conversion process waste-
waters and alcohols are found in industrial
effluents because of  their wide use as
solvents and reactants.  To minimize ex-
traneous solute-loss and maximize solute/
sorbent contact, several innovations have
been introduced into the adsorption iso-
therm  procedures in  an attempt to im-
prove reproducibility and accuracy and to
reduce solute losses.
  Thus, completely filled and capped stain-
less steel tubes containing the adsorbate/
adsorbent mixture were rotated 360 end-
over-end at 2 rpm for 24 hours at 20 
0.5C.   The tubes were  then  ultracentri-
fuged at 20,000 rpm for 20 to 30 mm at
20C, thereby spinning down  the carbon.
The tubes  were then opened,  and the
supernatant was analyzed directly.


Theoretical

General Solvophobic Approach
  The solvophobic (c$) theory describes
the tendency of a surrounding  solvent
medium to influence aggregation or dis-
sociation of those molecules with consider-
able  microsurface areas exposed to the
solvent medium.
  In the solvophobic  treatment,  adsorp-
tion is considered as a reversible  reaction
between the adsorbate molecules, S,, and
the activated carbon,  C, to form the ad-
sorbed complex, S,C,  at the surface of the
carbon, S, + C ^ S,C. The effect of the
solvent on this reaction is obtained by
subtracting the standard free energy change
for the reaction in the gas phase from that
in the presence of the solvent (taking as
standard states Xk = 1, pk = 1 atmosphere
ideal gas).  This process results  in a net
free energy change;  AG [Solvent effect) .ex-
pressing the effect of the solvent on the
association adsorption reaction.
  Conceptually,  Sinanoglu proposed a
two-step dissolution process.  First, a hole
or cavity needs to  be  prepared in the
solvent to accommodate the solute, carbon,
or adsorbed complex "molecule."  Second,
after the "molecule"  is placed   into the
cavity, it interacts with the solvent. Quan-
titatively this process is expressed as
follows:
      or
       AG
AG
     net
     (solvent effect)
Ar assoc    Arassoc
Alj (solvent) ~AU (gas)
RT
    net         _
    '(solvent effect) ~~
      . ,, net    .  ~net     . -net
      AGJ,S,C-AGJ,S,  - AGJ,C
                                         (2)
      where kk = pk/Xk is the Henry's constant
      for the kth species, and j represents each
      type of interaction. After specifying each
      interaction such as  the cavity, van  der
      Waal's,  and electrostatic terms plus two
      correction terms for  polymer mixing and
      reduced electrostatic effects because of
      the presence of the solvent, the following
      expression is obtained from Eq. (1) and (2)
      for the overall standard free energy change,
      viz.
A f^
AG
          assoc
    (solvent) -

[AGcav + AGvdw
                      assoc
                       (gas)
             net
            ,
      AGredl S.C-S,-C - RT In (RT/P0V)
                                   (3)
(1]
where the last term is called the cratic term
and results from an entropy or free volume
reduction.    AG floiventi  is  related   to
the experimental equilibrium constant,
Ksolvent i - XS,C/Xs,X0 which itself wil1 be
related to the  experimental adsorption
capacity'p, for solute S, later in this analysis.
Each term in the square bracket in Eq. (3)
can be calculated explicitly from known
physiochemical parameters obtained from
the literature.  Explicit formulae do this and
a discussion on the relevance of each term
to the adsorption association reaction are
presented in Appendix A m the compre-
hensive report.
   For the comprehensive c<-model, each
of the terms in Eq. (3) is calculated explicitly;
for the simplified model, the "thermo-
dynamic microsurface area change of the
reaction," AA, in the cavity term  is  as-
sumed to be  proportional  to the  cavity
surface area, TSA, of the specific sorbate
and  homologous series, AA =  g TSA.
Thus in this study. In Qb is correlated with
TSA, where Qb is the  initial slope at  low
solute concentration for the Langmuirian
adsorption isotherm.  In addition, competi-
tive adsorption as related to surface tension
of a multicomponent solution, diffusional
kinetics, and  a  comparison of  various
isotherm models for use in the Ideal Ad-
sorbed Solution (IAS) theory are all included
in the comprehensive report.

Experimental Results
   In the comprehensive report, the experi-
mental batch adsorption isotherm results
are presented and discussed.  After  dis-

-------
cussing adsorption sensitivity and the
range of solution concentration used, ad-
sorption onto powdered activated carbon
(PAC) is compared with adsorption onto
graphitized carbon (GC).  Kinetic studies
are then discussed ipso facto and in con-
nection with  the  time  to reach pseudo-
equilibrium and the effects of temperature.
Thereafter, both  single and  multicom-
ponent adsorption results are presented
and analyzed with respect to the solvo-
phobic thermodynamic (c0)  approach.
Some of these results are presented below.

Kinetic  Studies
  The time to reach a plateau or pseudo-
equilibrium adsorption capacity for phenol,
2-cresol, 2-ethyl phenol, 4n-propyl phenol,
2-butyl phenol, 2 pentyl phenol, and 2
hexyl phenol is 1,  4, 8, 9, 11, 12, and 14
hours, respectively.
  Plots, showing  the  experimental data
points superimposed onto the theoretical
curves based on  the  Freundlich model
obtained from Suzuki and  Kawazoe, are
used to obtain the surface and pore diffu-
sion  models.
  The results obtained from this procedure
are summarized in Table 1; the surface
and  pore  diffusion coefficients for 4n-
propyl phenol are  unexplainably too low.
The rest of the results follow the expected
trend of  decreased diffusion  rate with
increased solute cavity surface area (TSA).
  Since De ^ D^ the De values are within
an acceptable range of values, and the Ds
values are comparable to previously re-
ported values in the literature. For example,
for phenol. Van Vliet et al., using a single
parameter approach, report Ds = 1.24  x
10'9 cm2 sec'1 in Filtrasorb400, * whereas
Peel et al., using their branched-pore kinetic
model, report Ds % 7.75 to 9.01 x 10'8
cm2 sec"1  for rapid diffusion in the macro-
pores of Filtrasorb 400.  2-Butyl phenol
was  sorbed onto PAC at four  different
temperatures, 20  C, 30 C, 40 C, and 50
C, to determine the effect of temperature
on the adsorption dynamics.  Both the
single parameter  surface and pore diffu-
sion  models to fit the experimental data,
Ds(t) and De(t) were obtained and plotted
in an Arrhenius plot to obtain the activation
energy, Ea. From the plots, Ea values equal
to-2.79 cal gmole"1 and 3.20 cal gmole'1
were obtained for surface and pore diffu-
sion, respectively. Two competing phe-
nomena can  be  hypothesized;  internal
diffusion  is expected  to  increase with
temperature, while for the  exothermic re-
actions such as  adsorption of  organics
'Mention of trade names or commercial products does
 not constitute endorsement or recommendation for
 use
Table 1.    Surface and Pore Diffusion Coefficients Obtained from Adsorption Kinetics

                                               Diffusion coefficients, x 1O6
Solute
1. phenol
2. 2-cresol
3. 4-n-propyl phenol
4. 4-ethyl phenol
5. 2-butyl phenol
6. 2-pentyl phenol
7. 2-hexyl phenol
Molecular
weight
94.1
108.2
122.3
136.4
150.2
164.3
178.4
Bulk
Db
cm2 sec'1
9.127
8.122
7.366
6.773
6.292
5.893
5.520
Surface
Ds
cm2 sec'1
5.26x1 Or4
4.62x1 0-4
1.28x1Cr4
4.36x1 0-4
3.59x1 0-4
3.21x1 Or4
2.695x1 Q-4
Port?'
De
cm2 sec'1
8.88
8.15
2.16
8.09
6.97
6.62
5.56
onto PAC the reverse is true. The linear
plots  and the small negative activation
energies that  result highlight  both  the
competition and the relative importance of
the exothermic adsorption reaction.

Single Solute Adsorption
  The major  experimental effort of this
study involved the careful measurement
and accumulation of statistically relevant
data of single solute aqueous phase ad-
sorption isotherms  for two homologous
series. The main purpose for doing this is
to establish a reliable data base for evalua-
ting the efficacy of the solvophobic thermo-
dynamic treatment in ranking the adsorp-
tion intensity of the  different members of
different homologous series. Homologous
series of 19 alkyl phenols and 12 aliphatic
alcohols were chosen to represent aromatic
and  aliphatic organic groups in  water,
respectively.  The adsorption results in
terms of In Qb versus molecular weight
and TSA are summarized in Figs 1  and 2.
  One of the major findings shown in Fig 1
is that branched  compounds  have lower
adsorbability than linear or normal com-
pounds. For the phenols, however (Fig 2),
fragmented compounds (alkyl group dis-
tributed around  the ring)  exhibit higher
adsorbability.   Also above a molecular
weight of about  1 50 D,  adsorbability is
independent of molecular weight.
  Comparison between the goodness-of-
fit for correlating  adsorbabilities with MW
and TSA (i.e.,  checking the simplified c0-
model) shows that for 11 alkyl  phenols,
rmw = 0.93(7)  is different from rTSA =
0.97(5) with a confidence of 70%.  For all
12 alcohols, rmw = 0.72 is different from
rTSA = 0.93 with a confidence of 95%.
  In  summary, the adsorption  capacity
decreases (slope  increases) for  each
isomer of a homologous series with in-
creased branching  or decreasing  cavity
surface  area.
  By measuring the gas-phase adsorption
of the same homologous series of com-
pounds as that measured from the liquid-
phase, AJ^o'vent effect) could be checked. With
the study of additional homologous series
and the knowledge of the molecular struc-
ture and volume of each member, the res-
triction of  adsorption  because of steric
hinderance, as  recently suggested  by
Benedek (slow adsorption), could be de-
termined.  Further work is also necessary
to couple the theory to Myers' new charac-
teristic dimensionless adsorption para-
meter.
  With respect to equilibrium adsorption
isotherm measurements, the limited data
base  should  be extended  to  include a
comparison of different activated carbons
with  different pore sizes and  surface
activities.  Adsorption of ionizable homo-
logous groups should also be measured.
Competitive adsorption effects should be
extended to include additional homologous
series.  Finally, we recommend that the
theory be evaluated for competitive ad-
sorption of some industrial effluent streams,
such  as coal-based effluents containing
many isomers of phenols.
  The full report was submitted in fulfill-
ment of  Cooperative Agreement  No.
CR-80664801  by Rensselaer Polytechnic
Institute  under sponsorship of the U.S.
Environmental Protection Agency.

-------
     I    3
                     Alcohols

                 slope ~  0.057
                 int    ~  -2.842
                 r     =  0.72(0)
                        2-M-1-B 
                   1-B
                   1-Hx
                                        4-E-1-Pe
                                         2-E-1-B
                                        2-M-3-Pe
                                    1-Hp
                                   2,4-diM-3-Pe
                                   3-E-3-Pe
           70
                      80
                                    1-Pe
                                                    3,3-diM-1-B
                                                    2,3-diM-2-B
                                                               = n Alcohols
                                                               = Branched
  90         100         110

     Molecular Weight
           120
   5.0
   4.0
   3.0
   2.0
   1.0
   0.0
                                                                             1-Hp
                      Alkyl Alcohols
                                                           1-Hx
                                          4-M-1Pe
                                2-E-1-B     
                                           
                            2,4-diM-3-Pe
                            2-M-3-Pe
                                                     3-E-3-Pe
         2-M-1-B   
1-Pe

  3,3-diM-1-B
                          2.3-diM-2-B
       r =  0.5264
     SL =  0.0335
    INT = -8.561
  #PTS. =  12
ST. DEV. =  0.3974
                                                                          i
     250    260     270    280    290    300    310     320    330    340     350

                                      TSA (fc)
Figure 1.    A dsorption (1nQb) versus molecular weight (MW) and total cavity surface area (TSA)
            for linear and nonlinear alkyl alcohols.

-------
          7.5
          7.0
          6.5
          6.0
      I   5.5
          5.0
          4.5
          4.0
                                               4-pentyl 
                     Alky I Phenols
                                     2-butyl 
                                     4-butyl 
                                         2,3,6-trimethyl
                                                A
                                                     2-pentyl  2-hexyl

                                                         4-tert pentyl
                                             m  4-tert butyl
            2,6-dimethyl A         $^2.3,S-trimethyl
            2,3-dwethyl34al , 4-propyl
                              3,5-dimethyl
                          2,5-dimethyl*
                                      4-ethyl   4-isopropyl
                             2-methyl
                 phenol
                                         linear
                                        A fragmented
                                        m branched


                                       r = .87(9) H9 compounds)
                                       r = .93(7) (12 compounds)
90
100
110
120
130
MW
140
150
160
170
?
c
       7.5


       7.0


       6.5


       6.0


       5.5


       5.0


       4.5
      4.0
Alkyl Phenols
                                                                       4-pentyl
                                                       2-butyl
                                                   2-pentyl
                                                                  2-hexyl
                                              '4-butyl
                                              4-tert-pentyl
                2,3,6-trimethyl A   -/"  4-tert butyl
            2,6-dimethyl  A     A/ 2,3,5-trimethyl
               2.3-dimethyl
                                 4.isopropy,
                           4-ethyl
               2-methyl
                                       linear
                                      A fragmented
                                       branched
       phenol
                                     r = .88(1) 119 compounds)
                                     r = .97(5) (12 compounds)
        220  240   260   280   300    320    340   360   380   400    420   440

                                         TSA (k)

Figure 2.     Adsorption (1nQb) versus molecular weight (MW) and total cavity surface area (TSA)
             for linear, nonlinear,  and fragmented alkyl phenols.

-------
Georges Be/fort, Gordon L Altshuler, Kusuma K. Thallam, Charles P. Ferrick, Jr.,
  and Karen L.  Woodfield are with Rensselaer Polytechnic, Troy, NY 12181.
Richard A. Dobbs is the EPA Project Officer (see below).
The complete report, entitled "Predicting Preferential Adsorption of Organ ics by
  A ctivated Carbon." (Order No. PB 83-222 778; Cost: $ 13.00, subject to change)
  will be available only from:
        National  Technical Information Service
        5285 Port Royal Road
        Springfield, VA 22161
        Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
        Municipal Environmental Research Laboratory
        U.S. Environmental Protection Agency
        Cincinnati, OH 45268
                                                                -&U. S. GOVERNMENT PRINTING OFFICE: 1983/659-095/0735

-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use S300

RETURN POSTAGE GUARANTEED
                                                                                                           Third-Class
                                                                                                           Bulk Rate
                                   LOU  f"   TJLLtY
                                            ^i  V  tKA
                                                 lL
                                   230

-------