United States
Environmental Protection
Agency
Environmental Research
Laboratory
Duluth MN 55804
o'V/,
.>• '•' -s
/1 v
Research and Development
EPA-600/S3-83-088 Dec. 1983
&EPA Project Summary
Adsorption and Desorption
of Hexachlorobiphenyl
D.M. DiToro, LM. Horzempa, and M.C. Casey
The experimental and theoretical re-
sults that lead to the development mod-
el for the analysis of adsorption and
desorption of hexachlorobiphenyl from
suspended and sedimented particles
are discussed.
The reversibility of the adsorption
reaction between dissolved organic
chemicals and naturally occurring soils.
sediments, and suspended particles is
of" fundamental importance in the
understanding of the fate of these
chemicals in the environment. The issue
of reversibility becomes critical if the
adsorption-desorption behavior of a
chemical is to be expressed quantitative-
ly within the framework of mass
balance equations. In the formulations
used to date, with a notable exception
to be discussed below, the formulations
used to express the adsorption and
desorption reactions assume reversible
behavior, that is, at equilibrium, the
same isotherm applies for adsorption
and desorption.
The difficulty with this assumption is
that for many organic chemicals and
many naturally occurring adsorbents,
laboratory adsorption and subsequent
desorption experiments demonstrate
only partially reversible behavior.
In the experiment described in Part A
of the final report, this nonsingular
behavior was confirmed and, using
various experimental procedures, it was
found to persist, which suggests that it
is necessary to account for this behavior
in a quantitative and consistent way.
In Part B of the final report, a
framework, is presented within which
this nonsingular behavior can be analyzed
in a manner that can be easily incorporated
into mass balance calculations.
This Project Summary was developed
by EPA's Environmental Research
Laboratory, Duluth. MN, 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).
Introduction
The purpose of the final report,
summarized herein, is to present the
experimental and theoretical results that
lead to the development model for the
analysis of adsorption and desorption of
hexachlorobiphenyl from suspended and
sedimented particles. In contrast to the
desorption reaction, a large body of
information already exists for the adsorp-
tion reaction. Several authors have re-
ported relationships that relate the extent
of adsorption of organic chemicals to
their characteristics such as aqueous
solubility and adsorbent properties such
as specific surface area and organic
carbon content.
This is not the case, however, for the
desorption reaction. The available informa-
tion, to be discussed in more detail below,
indicates that for a great many organic
adsorbent systems the desorption reaction
is not completely or even moderately
reversible. As a consequence, the assump-
tion of reversible behavior is neither
justified nor realistic, and it is not possible
to directly apply the large body of
adsorption theory and data to describe
desorption since, for nonreversible
systems, it is not the same reaction.
This is unfortunate since it is not clear
how to incorporate nonreversible behavior
into modeling frameworks that have
been, and are being, developed by EPA
and other groups for the computation of
the fate of toxic chemicals in natural
waters. If the adsorption were either
completely reversible, or completely
-------
irreversible so that no desorption occurred,
then it would be straightforward to
include such behavior in a fate model.
What has been found experimentally,
however, is that some desorption takes
place. The amount is variable and
depends on the details of the situation,
such as the mass of adsorbent and the
adsorbate-adsorbent pair involved.
The use of a desorption "partition
coefficient" in a way that is analogous to
the use of the adsorption partition
coefficient in fate computations, is not a
solution to the problem since the actual
quantity of chemical desorbed when
exposed to lower aqueous concentration
is not directly related to only the
desorption partition coefficient but also to
the quantity of chemical previously
adsorbed. As it happens, for the desorption
model described below, the desorption
partition coefficient does have a specific
meaning, which the model clarifies, but it
cannot be used directly in fate computations.
Without a specific model for nonreversible
desorption, it is not surprising that this
mechanism has not been explicitly
included in fate computations.
The nonreversible behavior of adsorption
and desorption can have important
consequences for the fate of chemicals in
natural waters. As inputs of toxic
chemicals are reduced, the desorption of
already existing toxic chemical from
suspended solids and sediments will
constitute the major inputs of dissolved
toxicants into the water column. The
magnitude and extent of this reaction can
control the environmental distribution
and the exposure level for the biota. If the
quantity of chemical desorbed is much
less than the quantity initially adsorbed
then assuming completely reversible
behavior can significantly overestimate
the dissolved chemical in the water
column. This overestimate may translate
into an underestimation of the impact of
remedial measures such as discharge
reductions via treatment of effluents.
Hense, a quantitative understanding of
the factors that influence the behavior of
the desorption reaction is an essential
component for understanding the fate of
toxic chemicals in natural waters and the
consequences of remedial actions.
Summary
As part of the effort at Manhattan
College to formulate and test mathematical
models of the fate of RGB's in the Great
Lakes, a series of experiments have been
conducted using tritiated hexachlorobi-
phenyl (abbreviated as HCB) as the
adsorbate and natural sediments and
inorganic clays as the adsorbents. The
experiments concentrated on the desorp-
tion behavior as well as conventional
adsorption tests. Nonreversible desorp-
tion occurred, and an effort was made to
formulate a model which explained the
data.
It was assumed that the adsorbed HCB
was made up of two components: an
exchangeable component which readily
and reversibly desorbs and readsorbs
depending upon aqueous phase concentra-
tion, and a second component, which was
termed nonexchangeable, which resisted
desorption until very low(or possibly zero)
aqueous concentrations. This idea is
often used to explain nonreversible
behavior in qualitative terms, e.g., physi-
cal versus chemical adsorption. Methods
were developed for calculating the
quantity of the exchangeable and nonex-
changeable components from the experi-
mental adsorption and desorption data.
This is the unique feature of the model
since it gives quantitative estimates of
the magnitudes of these components. An
analysis of the individual behavior
suggested that each was describable in
terms of (distinct) linear isotherms. This
regular behavior, for both natural sediments
and inorganic clays, represents a significant
simplification and codification of a large
quantity of adsorption and desorption
data in terms of distinct partition coefficients
for the exchangeable and nonexchangeable
components. Subsequent consecutive
adsorption experiments confirmed the
distinct behavior of the two components
and supported the validity of the model.
The fact that two distinct isotherms are
found for the adsorption and desorption
data indicates that the desorption is not
completely reversible. Consider a single
pair of points corresponding to a single
adsorption-desorption experiment. If it
is assumed that continued desorption
cycles follow a straight line, then the
intersection of this line and the ordinate
defines the particulate concentration
which is nonexchangeable (since it
remains on the particles even at zero
aqueous concentration). Once the nonex-
changeable component concentration, r0,
has been found, the differences between
this concentration and that found at
adsorption and desorption equilibria
must be the exchangeable component
since two components are assumed to be
present. The fact that it responded to the
decrease in aqueous concentration that
occurred from adsorption to desorption
equilibrium supports its exchangeability.
Note that two exchangeable component
data points result: at adsorption equili-
brium, rxa, and that desorption equilibrium,
rXd. These correspond to the two aqueous
concentrations ca and Cd, respectively. If
this analysis is repeated for the remaining
two adsorption-desorption data pairs, the
result is six pairs of exchangeable
component-aqueous concentration data.
The validity of this analysis depends
upon the observation that all the exchange-
able component data conform to a single
isotherm. The same isotherm applies to
all exchangeable component data, regard-
less of whether they correspond to the
quantity of exchangeable component that
is present at adsorption, rxa, in equilibrium
with aqueous concentration, C* or at
desorption, rxd, in equilibrium with
aqueous concentration, cd. That is, the
exchangeable component is behaving in
accordance with classical reversible
adsorption-desorption theory.
The three nonexchangeable component
concentrations calculated from the data
analysis also have been found to follow
one isotherm. They are a linear function
of the adsorption aqueous concentration.
Part B of the final report contains
additional results of the isotherm analysis
for HCB and a full discussion of the
development of the proposed adsorption-
desorption model. Part A of the final
report presents further data and the
results of experimental modifications
designed to eliminate experimental
artifacts as the cause of the nonreversible
behavior.
A second focus of the experiments
conducted with HCB was the effect of the
mass of adsorbent on the partition
coefficients. It had been observed from an
analysis of published data that adsorption
partition coefficients decrease as adsor-
bent mass increases. This phenomenon
was investigated for HCB adsorption and
also for desorption. It was found to occur
for both reactions. If the data is interpreted
in terms of exchangeable and nonexchange-
able components, the nonexchangeable
partition coefficient is essentially inde-
pendent of adsorbent mass, whereas the
exchangeable partition coefficient is
inversely proportional to adsorbent mass.
The adsorption and desorption partition
coefficients are seen to decrease as
adsorbent mass, m, increases. Note that
the extent of irreversibility increases as
mass increases. That is, the desorption
partition coefficient becomes increasingly
larger than the adsorption partition
coefficient as adsorbent mass increases.
The exchangeable partition coefficient is
seen to be inversely proportional to mass
whereas the nonexchangeable partition
coefficient is independent of mass. This |
-------
suggests a definition of an exchangeable
distribution coefficient: vx= ir^m which is
also mass independent.
The variation of these mass-indepen-
dent parameters as functions of sediment
properties has been examined. The
details, together with additional data and
a more complete description of the
adsorbent mass effects are given in Parts
A and B of the final report.
The result of this combined experimental
and modeling program is a quantitative
framework within which it is possible to
predict the adsorption, and more signifi-
cantly, the desorption behavior of hexa-
chlorobiphenyl as a function of adsorbent
mass and its characteristics.
The importance of the interaction
between modeling analysis and experimen-
tal design in such an investigation cannot
be overemphasized. The exchangeable-
nonexchangeable model was formulated
as a consequence of the need to organize,
synthesize, and understand the experimen-
tal data. Once the hypothesis was
formulated in quantitative terms, it was
possible to design experiments to test the
model predictions and illucidate the
relevant features. The dual capability to
do the experiments and formulate the
models within a cooperating group is
essential if progress is to be made in
formulating and testing rational, quantita-
tive descriptions of complex phenomena
such as the desorption reaction.
Implications for Receiving
Water Fate Models
The use of models for the computation
of toxic chemicals exposure levels in
natural waters is currently an important
component of rational toxic chemical
regulation and control. The development
of EXAMS by EPA models for PCB, radionu-
clides and toxic heavy metals in the Great
Lakes by Manhattan College, and other
investigations, are currently in progress.
These models have a common approach
in dealing with the adsorption-desorption
reaction.
The mass balance equations are
written in terms of total chemical, CT, with
the transport and kinetic terms suitably
modified with the fraction of chemical in
the dissolved, fd, or particulate fp, form
depending on whether the terms in the
equation apply to particulate or dissolved
phases. As an example, consider a two-
layer segmentation representing the
water column of depth Hi, and an active
sediment layer of depth H2. These
interact via vertical mixing of the aqueous
phases, with mass transfer coefficient KL;
and settling and resuspension of the
particulate.phases, with velocities wa and
wrs respectively. The governing mass
balance equations are:
dt
Waf plCTI + Wref p2CT2 + W
= KL(fdCT,-fd2CT2)
dt
Wafp|CT|-Wrefp2CT2
(2)
where CTI and cT2 are the total chemical
concentrations in the water column and
sediment layers respectively, and W is
the input mass loading rate (M/L2/T).
Note that the central roles of the
dissolved fdi and fd2 are particu late (f Pi and
fp2> fractions in the water column and
sediment segments, respectively. They
directly affect the magnitudes of the mass
transfer coefficients and, therefore, the
fate of the chemical. A more complex fate
computation would include terms for
outflow, the various appropriate decay
mechanisms, and sedimentation losses.
However, the principle is still the same.
Once the total concentration is computed,
the dissolved water column concentration
is given by: cdi = fdiCri, with analogous
expressions for the particulate concentra-
tion. Again, the particulate and dissolved
fractions play a central role, and these
fractions are a direct result of the
adsorption-desorption model employed.
For completely reversible adsorption-
desorption and a linear isotherm, the
dissolved and particulate fractions are
given by:
1
PITT
(3)
(4)
where TT is the reversible partition
coefficient and m is the adsorbent
concentrations. The subscripts 1 and 2 in
equations (1) and (2) refer to evaluating
these fractions using the appropriate
adsorbent concentration in segments 1
and 2.
For the HCBC exchangeable-nonex-
changeable component model of adsorp-
tion-desorption, these fractions depend
upon the model parameters: TTO, the
partition coefficient for the nonexchange-
able component; and i/*, the distribution
coefficient for the exchangeable compon-
ent; and the maximum dissolved aqueous
concentration to which the particle has
been exposed: Cmd. This latter concentra
tion sets the magnitude of the nonexchange-
able component. It can be shown that the
dissolved and particulate fractions are
given by the expressions:
fd =
1
, _
Tp -
1 + Vx + m TTofCmd/Cd)
Vx +
1 + Vx +
(5)
(6)
where cd is the current dissolved aqueous
phase concentration. The conventional
expression, assuming reversible behavior
is also shown. There is a significant
difference between the conventional
reversible formulation and the exchange-
able - nonexchangeable model. The
particulate fraction is always a substan-
tial portion of the total chemical concen-
tration, even at low suspended solids
concentrations that are characteristic of
most receiving waters (10-100 mg/l).
This suggests that fate computations
using the exchangeable-nonexchangeable
model will give quite different results
which emphasize the importance of
particle transport.
-------
D. M. Di Toro, L. M. Horzempa, and M. C. Casey are with Manhattan College,
Bronx, NY 10471.
W. R. Richardson is the EPA Project Officer (see below).
The complete report, entitled "Adsorption andDesorption ofHexachlorobiphenyl:
A. Experimental Results and Discussions; B. Analysis of Exchangeable and
Nonexchangeable Components," (Order No. PB 83-261 677; Cost: $25.00,
subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, V'A 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Large Lakes Research Station
Environmental Research Laboratory-Duluth
U.S. Environmental Protection Agency
9311 Groh Road
Grosselle,MI48138
•frUS GOVERNMENT PRINTING OFFICE 1983-659-017/7236
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
------- |