United States
Environmental Protection
Agency
Health Effects
Research Laboratory
Research Triangle Park NC 27711
Hesearch and Development
EPA-600/S1-84-013 Dec. 1984
Project Summary
Recovery of Trace Organic
Compounds by the Parfait/
Distillation Method
James B. Johnston, Clarence Josefson, and Richard Trubey
A modified parfait/distillation method
was developed that recovers a wide
range of neutral, cationic, anionic, and
hydrophobic contaminants from water.
Porous polytetrafluoroethylene (PTFE,
Teflon,* duPont) was identified as an
ideal first filtering-adsorbing bed in the
parfait train. PTFE removed humic acid
and a broad range of hydrophobic
compounds. It was more easily cleaned
and contributed fewer impurities to
eluates than other porous hydrophobic
adsorbents tested. Several types of 0.2-
fjm sterilization filters were tested.
Filters containing nitrocellulose or
fiberglass appeared to adsorb appreci-
able amounts of some model
compounds; polycarbonate or polypro-
pylene filters were less adsorptive.
Various ion exchange resins were evalu-
ated for use in the parfait column. The
resins selected, Dowex MSC-1 ano
Duolite A-162, cerried strongly acidic
or basic exchange groups, were
macroporous, ha~ high exchange
capacities, and were inexpensive. An
elution protocol was developed, with
emphasis on recovery of poorly volatile,
highly water-soluble model compounds.
Finally, the modified parfait method
was tested for its ability to recover 19
model compounds in a synthetic hard
water. Poorly volatile, neutral, water-
soluble contaminants (hexoses) and
cationic aromatic compounds were
recovered with exceptional efficiency.
Humic acid was readily recovered and
did not substantially interfere with the
* Mention of trade names or commercial products
does not constitute endorsement or recommenda-
tion for use by the U.S Environmental Protection
Agency
recovery of the other model compounds.
Model amphoteric compounds were
removed from water, but were difficult
to recover from the parfait column.
Trimesic acid was readily removed from
water and was recovered selectively,
though not quantitatively.
This Project Summary was developed
by EPA's Health Effects Research
Laboratory, Research Triangle Park,
NC, 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 need to evaluate health risks
associated with organic contaminants in
surface and potable waters has prompted
basic research into methods to recover
and concentrate these substances. The
contaminants about which most is known
are hydrophobic or volatile substances,
which are readily recovered by methods
such as gas stripping and solvent
extraction, but which account for only a
minority of the organic material in a
typical water. The majority of compounds
in water are poorly understood, both
because methods to recover them have
not been developed and because
automated means of compound identifi-
cation, such as gas chromatography/
mass spectrometry, are not applicable.
The parfait system was developed with
the aim of recovering and detecting
mutagenic substances, particularly the
water-soluble, poorly volatile organic
compounds. The original parfait method
rested on the use of vacuum distillation/
-------
lyophilization to concentrate the poorly
volatile species in water. However, the
compounds obtained by this method are
in an intractable, insoluble form, because
bicarbonate dissociates under vacuum to
give metal carbonate precipitates that
can trap organic polymers and lipids. The
parfait method prevents the formation of
precipitates by removing metal ions on a
strongly acidic cation exchange bed. To
protect the bed from particulate matter
and to remove some of the hydrophobic
species, the original parfait system used a
bed of silica gel placed ahead of the cation
exchange bed. To neutralize the acid
released from the cation bed, as wel I as to
remove anions, a strongly basic anion
exchange bed was added after the cation
exchanger. In the original parfait system,
the silica gel acted as a weak cation
exchanger and leached silicic acid to the
anion exchange bed and into the final
effluent, interfering with the recovery
of compounds from these two sources.
In the first part of the study reported
here, we identified porous polytetrafluor-
oethylene (PTFE, Teflon, a registered
trademark of E.I. duPont de Nemours and
Co.) as a suitable substitute for silica gel
as the first filtering-adsorbing bed in the
parfait train. Next, we developed a proto-
col to selectively desorb organic anions
and cations from the ion exchange beds.
Finally, we evaluated the model system
by recovering 19 model compounds
added to a synthetic hard water. The
model compounds were analyzed quanti-
tatively in the presence or absence of a
humic acid supplied by the U.S.
Environmental Protection Agency's
Health Effects Research Laboratory.
Procedure
To find the best overall protocol for
recovery of the model compounds, many
variations in protocol were tried.
Whenever possible, the simplest, most
readily available analytical methods were
used; they were adequate to discern the
superiority of a new protocol but were not
always the most precise methods
possible. Because the model compounds
represented general classes of
environmental compounds, we felt that
the recovery of any model compound in a
synthetic hard water to a precision of two
significant figures would be sufficient to
indicate the probable utility of the method
for recovering similar environmental
compounds of interest. While developing
the protocol, we assayed fractions
selectively; the parfait column effluent
was not prepared and assayed in the early
experiments with single compounds
unless a compound was not adequately
accounted for in the other fractions.
Results and Discussion
Losses on Ultra filters
To test the possibility that the 0.2-/;m
ultrafilters used for sterilization could
themselves adsorb mutagens or other
substances, we exposed several ultrafil-
ters to solutions of pyrene and ethidium
bromide (EB) in high-purity water. All of
the filters adsorbed significant amounts
of pyrene, and the Millex disposable filter
unit also adsorbed >100 /yg of EB.
Fiberglass prefilters seemed to have an
unacceptably high capacity for the
solutes. Polycarbonate filters mounted in
reusable polycarbonate filter holders
seemed to be the least adsorptive.
Losses During Vacuum
Distillation and Lyophilization
The original parfait method used
vacuum distillation to concentrate
aqueous eluates from each of the parfait
beds and the aqueous effluent from the
parfait column itself. The modified parfait
method uses this method to concentrate
only the aqueous column effluent.
The potential for loss of moderately
volatile compounds during vacuum
distillation was tested. Solutions of
biphenyl, phenanthrene, and pyrene
were freeze dried or vacuum distilled for
varying times, and the amount of
compound remaining in the flask was
determined. Vacuum distillation resulted
in greater losses than did lyophilization,
and that ease of loss varied considerably.
The losses were due to volatilization
under vacuum, especially during the
degassing phase of vacuum distillation.
Vacuum distillation should be used only
to concentrate aqueous solutions of
compounds whose volatility is less than
that of pyrene, to avoid unacceptable
losses of compound.
Alternatives to Silica Gel as First
Parfait Bed
The first parfait bed filters paniculate
from the sample and adsorbs hydropho-
bic compounds that would otherwise
adhere to the polystyrene divinylbenzene
backbone of the ion exchange resins.
Various porous agents capable of filtering
an aqueous sample were tested for their
capacity to adsorb EB, to see if any
medium combined both functions. Inert
materials, including powdered cellulose,
Celite, and porous Teflon f Chromosorb T),
each adsorbed appreciable amounts of
EB; total recovery of EB was
nonquantitative with all materials except
porous Teflon. Florisil, silica gel, silanized
silica gel, and porous Teflon adsorbed
virtually all of the EB, but the compound
was not extractable with acetone from
florisil or silica gel. Our results suggested
that porous Teflon might be a desirable
first adsorbent for the parfait system,
because it would serve as an inert filtering
medium and would also reversibly bind
certain trace organic contaminants in
water.
Adsorption onto Silanized Glass
We determined the breakthrough of
caffeine, pyrene, and EB from 5-ml beds
of silanized glass wool and silanized 60-
80 mesh glass beads. Only caffeine
showed no appreciable adsorption. Both
pyrene and EB were removed essentially
quantitatively, and the adsorbed
compounds could be recovered in
acetone washings. Thus, silanized glass
and fiberglass were considered
acceptable for use in the modified parfait
method.
Characterization of Porous
Teflon
We tested the recovery of some of the
model compounds to determine whether
a detailed characterization of the
properties of porous Teflon was
warranted. In experiments with pyrene,
EB, glucose, caffeine, quinaldic acid,
glycine, and benzylamine, Chromosorb T
appeared to adsorb the larger aromatic
compounds but showed essentially no
affinity for the neutral, water-soluble
species, even when these had up to one
aromatic nucleus.
In recovery experiments, a mixture of
2,2'-dichlorobiphenyl, anthraquinone,
stearic acid, BHT, and bis(2-ethylhexyl)-
phthalate in synthetic hard water was
passed through a 50-ml bed of
Chromosorb T, and the bed eluted with
methylene chloride. At least half of each
compound was adsorbed by the Teflon
and recovered.
We next explored the possibility that
surface area limits adsorptive capacity of
porous Teflon. The manufacturers' stated
ranges of surface area are 4-7 mVg for
ChromosorbT and 2-4 mVg for Fluoropak
80 (another kind of PTFE). The morphol-
ogies of the two PTFE aggregates (as seen
in the scanning electron microscope)
account for these differences in surface
area per unit weight. Based on the
average total surface area of a 5-ml
-------
column, Chromosorb T should show
approximately 1.3 times the capacity of
Fluoropak 80 for any solute. However, the
observed ratios of the capacity of
Chromosorb T to Fluoropak 80 for three
solutes ranged from 6 to 14. The
differences in the adsorptive capacities of
Chromosorb T and Fluoropak 80 cannot
be accounted for on the basis of surface
area alone. For the present, we conclude
that columns of Chromosorb T have a void
volume that includes an internal and an
exterior volume, and that the internal
volume must be wetted with solvent if the
high adsorptive capacity of the polymer is
to be exhibited. The simplest explanation
of the data now available is the
hypothesis that water contained in the
internal volume is modified by the Teflon
to give it solvent-like properties.
The capacities of Teflon for various
organic compounds were determined
from breakthrough curves, using 5-ml
beds of Chromosorb T. Compounds that
broke through in less than one column
volume were glucose, glycine, tryptophan,
adenine, uracil, xanthine, trimesic acid,
benzylamine, and the anionic dye rose
bengal. Compounds that did not break
through in up to 1400 bed volumes, the
maximum volume tested, were 2,2'-
biquinoline and the cationic dyes crystal
violet, safranin O, and EB. The latter three
compounds are readily soluble in water,
unlike the other compounds that adsorb
to Teflon.
Seven of the compounds tested were
aromatic hydrocarbons that constitute a
series of functionally similar solutes. For
this group of compounds, capacity was
correlated with the reciprocal of solubility
in the mobile phase (r2 = 0.94) and with
the octanol-water partition coefficient
(Kow) (r2 = 0.96). Also tested were various
nitrogen-containing heterocyclic aromatic
compounds. Among the nitrogen hetero-
cyclics, the methyl-substituted xanthines
showed a tendency for capacity on Teflon
to be correlated with solubility. This was
highly unexpected, because all previously
observed correlations of capacities are
with the reciprocal of solubility. We
determined the capacities of XAD-2 and
XAD-8 for three of these compounds and
found the same unexpected tendency for
capacity and solubility to correlate. The
generality of the widely accepted inverse
correlation of solubility and capacity
factor, or direct correlation of octanol-
water partition and capacity, should there-
fore be questioned. The anomalous
behavior of the substituted xanthines may
be due to their formation of oligomers in
aqueous solution.
The other anomalous adsorption
discovered here, the tight association of
cationic aromatic dyes with PTFE, is less
straightforward. The combination of a
hydrophobic and a cationic moiety on
these water-soluble compounds suggests
a dual interaction of these solutes with
the hydrophobic and strongly
electronegative surface of Teflon. On the
other hand, trimesic acid and the anionic
aromatic dye rose bengal showed no
affinity for Teflon, suggesting that elec-
trostatic repulsion overcomes any affinity
due solely to the hydrophobic effect. The
affinity of Teflon for cationic aromatic
compounds suggests that pesticides like
diquat and paraquat may also be strongly
adsorbed on PTFE columns.
Chromosorb T is an adsorbing agent
comparable to the XAD resins. Per unit
surface area, porous Teflon has a greater
adsorptive capacity than the XADs;
however, the XADs have a much larger
surface area than Teflon per unit weight
or per bed volume. Approximately four
bed volumes of the Teflon equal the
capacity of one bed volume of an XAD-2
column, and XAD-8 and Teflon have
roughly equal capacities per bed volume.
The chemical stability, inertness, and
purity of Teflon suggest its superiority to
the XADs. Porous Teflon also differs from
the XADs in having a high capacity for
humic acid, at least for the sample used in
this study. Further development of the
parfait method employed Chromosorb T
as the first parfait bed.
Ion Exchange Resins
The ion exchange resins eventually
chosen for use in the parfait column were
selected because they carried strongly
acidic or basic exchange groups, they
were macroporous, they had relatively
higher exchange capacity than otherwise
equivalent alternatives, and they were
lower in cost. Dowex MSC-1 was finally
chosen over AG MP-50 primarily on the
basis of cost, and Duolite A-162 was
preferred over other macroporous anion
exchange resins primarily because of its
higher exchange capacity. A-162 gave
the impression of cleaner eluates than
did AG MP-1, but this was not tested
rigorously in side-by-side trials.
In choosing an eluting solvent, we
regarded liberation of contaminating
materials from the resin as the most
serious question. We expected that all
poorly water-soluble solutes would
adsorb to Teflon and that none would
reach the polystyrene-divinylbenzene
matrix of the ion exchange resins.
Therefore, the elution of the ion exchange
resins could focus exclusively on the
exchanged ionic solutes.
The approach tested was to neutralize
the Hi* or OH" counterions on the bed,
thereby converting weak organic acids or
bases to their neutral forms, and to elute
with organic solvents, in which such
neutral organics should be readily soluble.
We expected this approach to have the
advantages of exposing the resins to
lower-ionic-strength eluants, which
might reduce the liberation of
contaminants from the resin support
matrix; of selectively recovering the
organic solutes in a neutral, organic
soluble form, separate from inorganic
ions; and of recovering the eluted solutes
in any medium easily concentrated in a
Kuderna-Danish apparatus, in anticipa-
tion of analysis by gas chromatography.
Full advantage depended upon the exclu-
sion of water from the elution solvent.
A series of experiments led to the
development of a modified protocol for
cleaning the adsorbents and recovering
the solutes. The elution method
developed here is a compromise,
balancing the contamination of the eluate
against an incomplete recovery of
solutes, like trimesic acid, which tend to
be released from the resin with difficulty.
However, near-quantitative recoveries of
simpler acids, like benzoic and phthalic
acids, can apparently be expected from
this protocol. The protocol has the
advantage of selectively recovering weak
acids separately from strong anions, and
it avoids the use of high concentrations of
strong eluting ions.
The elution scheme developed here
might not be suited to all end uses of the
recovered compounds. Because the
solutes are exposed to methanol in the
presence of anhydrous HCI, methyl esters
may form. If bioassay is the intended end
use, the extent of methylation should be
considered, and alternative elutions,
perhaps using volatile buffers, might be
developed.
Recovery of Test Solutes
Experiments indicated that PTFE had a
high capacity for humic acid, and that a
50-ml Teflon bed was saturated .by
applying 16 mg of humic acid. A 150-ml
bed of Chromosorb T was sufficient to
adsorb between 90 and 95% of the 16mg
of humic acid applied. A 150-ml Teflon
bed was used in subsequent
experiments.
Humic acid did not significantly
interfere with recovery of the test solutes,
except perhaps caffeine and 2,4-
-------
dichlorophenol. On columns using a 50-
ml Teflon bed, recovery of caffeine
decreased about 20% in the presence of
the humate. However, caffeine was not
recovered from the Teflon fraction of the
columns using a 150-ml Teflon bed. The
larger amount of humic acid collected on
the 150-ml Teflon bed might have
influenced the partition of caffeine
between the water and methylene
chloride phases in the eluate; this
possibility remains to be tested.
The only other effect possibly attribu-
table solely to humic acid involved 2,4-
dichlorophenol. Humic acid apparently
decreased the ease of elution of the
chlorophenols. However, because the
reproducibility of 2,4-dichlorophenol
recovery among the various parfait
fractions was poor, differences in its
recovery may reflect the reproducibility of
the method rather than the influence of
the humate.
The porous Teflon bed gave reasonably
reproducible recoveries of test solutes.
The solutes found exclusively on Teflon
and their mean percent recoveries were
stearic acid (98.4), 2,4'-dichlorophenol
(79.1), 2,2',5,5'-tetrachlorobiphenyl
(84.7), bis(2-ethylhexyl)phthalate (94.3),
1 -chlorododecane (70.4), biphenyl (89.2),
anthraquinone (109), BHT (91.6), and
methyl isobutyl ketone (47.8). The
relatively low recoveries of 1-chlorodo-
decane and methyl isobutyl ketone were
likely due to losses during concentration
of eluates.
Glucose was recovered quantitatively
in the column effluent, whether humic
acid was present or absent. Presumably,
all other poorly volatile, neutral hydro-
philic solutes would also be recovered in
this fraction.
Among compounds that adsorbed to
Teflon and to at least one other bed,
isophorene and phenanthrene were
recovered essentially only from Teflon,
minor amounts being found in other
fractions. Caffeine was adsorbed to
Teflon and the cation exchange bed, and
was found in the column effluent. When
the larger Teflon bed and 15 other solutes
were included, recovery of caffeine was
decreased strikingly, and it was no longer
recovered from Teflon.
Three compounds recovered from
parfait columns (phenanthrene, caffeine,
and 2,4-dichlorophenol) had been tested
for breakthrough from 5-ml Teflon beds.
Their capacity factors and their recoveries
from the Teflon bed of a parfait column
showed a rough correlation. It may be
anticipated that compounds following the
inverse correlation of solubility with
*USGPO: 1984-559-111-10753
capacity factor, and having a capacity
factor greater than about 20, should be
detectably adsorbed to the Teflon bed of a
parfait column. Simply increasing the
volume of the Teflon bed may also
increase the absolute recovery of
adsorbable solutes with small capacity
factors.
Two solutes, glycine and quinaldic acid,
were not accounted for satisfactorily in
this study. Analytical problems,
impurities eluting from the ion exchange
resins, and unrecognized sources of loss
make it difficult to interpret experiments
involving these solutes.
Reasonable suggestions can be made
about the locations of the compounds not
recovered in parfait column eluates. For
example, the unrecovered methyl
isobutyl ketone, 1-chlorododecane, and
chlorobiphenyls were surely lost from the
methylene chloride eluate of the Teflon
bed by vaporization during concentration.
Trimesic acid on the other hand, appeared
to be so strongly adsorbed to the anion
exchange resin that it was incompletely
recovered in the eluate. Glycine appeared
to bind tightly to the cation exchange
resin and to elute incompletely by the
standard protocol. Both of these com-
pounds probably could be eluted with
aqueous solutions of high ionic strength,
but this would create the problem of
recovering the solute from the eluate.
Volatile buffers may offer a solution to
that problem.
Isophorone, 5-chlorouracil, and
quinoline are sufficiently water-soluble
not to adsorb to Teflon, but volatile
enough to be lost during vacuum
concentration of the column effluent.
Isophorone and 5-chlorouracil were
detected in the column effluent; at least
part of the unrecovered fraction of each of
these compounds must have been lost
during the vacuum concentration of this
effluent. Quinoline was not detected in
the column effluent, but its water
solubility and weak basicity suggest that
any of it that escaped the cation bed
would have gone to the effluent, where it
would have vaporized. Furfural is such a
reactive, volatile, and water-soluble
compound that its loss could be due to
several causes, most likely oxidation and
volatilization from eluates, especially
from the column effluent.
Conclusions and
Recommendations
A modified parfait method was
developed that is capable of recovering a
wide range of neutral, cationic, anionic.
and hydrophobia contaminants from
water. It may be the best available
method for quantitatively recovering
poorly volatile, highly water-soluble
compounds.
The success of the modified parfait
method depends heavily upon the
properties of porous PTFE. This material
removed humic acid and a broad range of
poorly water-soluble compounds from
water. It also had an anomalously high
affinity for cationic aromatic compounds,
which it adsorbed quantitatively. Porous
PTFE showed an inverse correlation of
capacity factor and aqueous solubility for
a series of hydrocarbons, and an
anomalous direct correlation of capacity
factor and solubility for a series of alkyl-
substituted xanthines. Porous PTFE was
more easily cleaned and contributed
fewer impurities to eluates than porous
hydrophobic adsorbents such as
polystyrene-divinylbenzene polymers.
PTFE may be a highly desirable
alternative to other currently available
porous hydrophobic resins, particularly in
studies concerned with the quantitative
analysis or the qualitative enumeration of
hydrophobic organic contaminants in
water. The basis for porous adsorptivity of
PTFE needs to be learned, and PTFE
should be compared in detail with resins
currently in use.
The modified parfait system removed
model amphoteric compounds from
water, but they were difficult to recover
from the parfait column by the protocols
developed for weak acids and bases, and
were not recovered quantitatively from
the adsorbent beds. Further work will be
needed to adapt the parfait method for
this purpose. Trimesic acid was readily
removed from water by the parfait
method and could be selectively recov-
ered from the anion exchange resin.
Recoveries were not quantitative; how-
ever, the selective recovery of polanions
should greatly facilitate their subsequent
isolation and analysis.
Several hydrophobic compounds were
shown to be readily lost from water
during vacuum distillation and during
freeze drying. Compounds like biphenyl
were lost in substantial amounts even in
the brief initial degassing phase of
vacuum distillation Compounds at least
as volatile as pyrene probably cannot be
reliably concentrated from water by
vacuum distillation. This behavior was of
little consequence, however, to recovery
of such compounds by the modified
parfait system; hydrophobic materials
were recovered in the first parfait bed by
adsorption onto PTFE, and materials from
-------
this bed were eluted in methylene
chloride and concentrated by solvent
evaporation in a Kuderna-Danish flask
under a three-ball Snyder column. Under
these conditions, recoveries were
satisfactory.
The humic acid used in this study was
readily recovered by the parfait method
and did not substantially interfere with
recovery of the other model compounds,
even when humate was present initially
at parts-per-million concentration. The
exceptional recovery of humic acids by
PTFE should be studied with a variety of
humic and fulvic acids, to assess the
generality of the adsorption and to deter-
mine whether PTFE could be used for the
recovery, fractionation, and characteriza-
tion of humic substances.
Certain disposable 0.2-pm sterilization
filters adsorbed appreciable amounts of
some of the model compounds from
water. Filters containing fiberglass or
nitrocellulose adsorbed more of the
solutes tested than those composed only
of polycarbonate or polypropylene; they
did not allow quantitative recoveries of
contaminants.
The parfait method does not recover
highly volatile compounds, and
contributes contaminants to eluates of
the ion exchange beds. Nevertheless, it is
very useful for the recovery of a broad
range of poorly volatile organic
contaminants. The parfait system should
be used in conjunction with short-term
lexicological assays to survey typical
drinking water supplies for mutagens and
other toxicants. It should also be used
with chemical assays when the recovery
of neutral, water-soluble, or cationic
aromatic compounds is to be emphasized.
James B. Johnston, Clarence Josefson, and Richard Trubey are with the
University of Illinois. Urbana IL 61801.
Frederick P. Kopfler is the EPA Project Officer (see below).
The complete report, entitled "Recovery of Trace Organic Compounds by the
Parfait/'DistillationMethod,"(OrderNo. PB85-127 199; Cost: $11.50, 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:
Health Effects Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
United States Center for Environmental Research BULK RATE
Environmental Protection Information POSTAGE & FEES PA
Agency Cincinnati OH 45268 EPA
PERMIT No. G-35
Official Business
Penalty for Private Use $300
------- |