United States
                   Environmental Protection
                   Agency
Health Effects
Research Laboratory
Research Triangle Park NC 27711
                   Research and Development
EPA-600/S1-84-018 Jan 1985
SEPA         Project Summary
                    Isolation  or  Concentration  of
                    Organic  Substances  from
                    Water  —  An  Evaluation  of
                    Reverse  Osmosis
                    Concentration

                    S.C. Lynch, J.K. Smith, L.C. Rando, and  W.L. Yauger
                     This study describes the development
                    and evaluation of a reverse osmosis
                    (RO)XDonnan dialysis  system for
                    preparing drinking water concentrates
                    for biological testing.  Commercially
                    available RO  and Donnan membranes
                    were surveyed for applicability to the
                    RO  concentration method  and  to
                    prepare organic concentrates. Two RO
                    membranes, cellulose acetate
                    (Osmonics, Inc.)* and FT-30 (Film Tec,
                    Inc.)*, were  selected for laboratory
                    evaluation, primarily because  of their
                    purported chlorine tolerance. Two
                    Donnan  membranes,  Nafion®
                    (Dupont)* and MA 3475 (IONAC)*,
                    were chosen for further study based
                    upon their  chemical  resistance
                    characteristics (chlorine, caustic, and
                    acid tolerance). The FT-30 membrane, a
                    thin-film  composite, demonstrated  a
                    substantially greater rejection
                    efficiency than the cellulose acetate
                    membrane. The effect of humic acid, a
                    natural constituent of potable water
                    sources, and inorganic salts on the
                    rejection of model compounds by FT-
                    30  and  cellulose acetate  RO
                    membranes was also evaluated and
                    found not to be significant under the
                    test conditions.
                     As expected from the rejection data,
                    the recoveries of model  solutes were
                    much better for the FT-30 membrane
                    system  than  the  cellulose  acetate
                    'Mention of trade names or commercial products
                    does not constitute endorsement  or rec-
                    ommendation for use
system. Overall the  recovery values
(2-74%) were  disappointing for the
majority of model compounds in both
RO systems and substantially less than
anticipated based on the rejection data.

  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
  In recent years, the trace  organic
content of  potable  water supplies has
been  identified as a  potential  health
hazard  Epidemiological studies  have
suggested  a relationship  between the
ingestion of these pollutants in drinking
water and carcinogenic and teratogenic
effects. Although hundreds of organic
compounds have  been detected  and
quantified in drinking water, the majority
of organic  material, i.e., the nonvolatile
fraction,  cannot be  identified  using
currently available technology
Therefore, the direct concentration/isola-
tion of organic contaminants in aqueous
samples for biological testing offers  a
practical solution to the determination of
health risks associated with trace organic
contaminants

  The  Health Effects  Research  Lab-
oratory - Cincinnati of the U.S.  Envi-
ronmental  Protection Agency (EPA) has

-------
funded several  independent  studies to
determine the effectiveness of different
iso I at i on/con cent rat ion techniques.
Systems  or techniques investigated
include reverse osmosis (RO),  vacuum
distillation, solid  adsorbents,  and
supercritical fluid COa extraction. The
necessity of concentrating  aqueous
samples prior to biological testing stems
from the trace levels of contaminants and
the lack  of  sensitivity associated  with
existing in vitro and in vivo biological test
systems.
  The RO  concentration method has
been  used  previously to  prepare
concentrates under  several EPA tasks
(EPA Contract  Nos.  68-03-2090,  68-
03-2194, 68-03-2367, 68-03-2464, 68-
03-2550, and 68-03-2713) The method
basically involves  repetitive batch
concentrations using a RO membrane to
retain organics  and  reduce the sample
volume (with an accompanying increase
in concentration) by discarding water that
permeates the membrane The Donnan
dialysis process has been used in several
of the concentrations to minimize solids
precipitation resulting from the increase
in inorganic salt concentrations.
  Previous  studies utilized total organic
carbon   (TOC)  measurements  as  a
surrogate to  monitor the progress  or
efficiency of the concentration  Although
TOC is  a  convenient  tool,  it   is  a
nonspecific measurement and provides
little insight about either the rejection or
recovery of specific chemicals. Therefore,
a standard set of model compounds with
differing  functionalities, solubilities and
sorption  properties  were selected  to
evaluate  this  and  the  other
isolation/concentration techniques


Experimental Procedures

Preparation of Model
Compound Test Solutions
  Test solutions of  model compounds
used in  the membrane screening  runs
and in the membrane concentration runs
were prepared  by simply diluting the
required volume(s) of stock solution with
organic-free  water  containing  an
inorganic salt  matrix.  The membrane
concentration runs utilized a "IX" salt
matrix consisting of 70 ppm NaHCOa, 120
ppm CaS04 and 47 ppm CaCI2-2H20. In
the membrane screening runs,  a "5X" salt
matrix representing a 5-fold increase in
salt concentration was employed Table 1
lists the final concentration at which each
model solute was tested and the solvent
used in the preparation of individual stock
solutions
Membrane Screening Tests
  Four  membranes  were  ultimately
selected for screening  tests based on
commercial  availability,  salt rejection,
stability at pH 5, and the ability to tolerate
a low to moderate free chlorine residual
(<5  ppm)  which  is commonly
encountered in drinking water. The four
considered to be the most promising were
the cellulose acetate and FT-30 (thin-film
composite) RO membranes, the Nafion®
cation  exchange  membrane,  and  the
IONAC  MA  3475  anion   exchange
membrane.  Initial  objectives  of  the
screening tests were:

  •  To determine specific  membrane
     rejections  with selected  model
     organic solutes.
  •  To determine  whether levels  of
     humic  acid typical  of  those
     experienced  in  drinking  water
     significantly  affected the  organic
     solute rejections
  •  To determine what trace organic
     compounds  might  leach from  the
     membranes and to  determine what
     effect a free chlorine residual might
     have  on the  types of substance(s)
     leached from the membranes.
  •  To determine  whether  flushing
     regimes typical  of  those  used in
     previous RO  concentrations were
     adequate to eliminate contamination
     from leachate compounds
  • To  provide   sufficient data  to
     estimate  solute  losses  to  the
     membranes   and  other  system
     components.

  A three-stage screening protocol was
developed to  evaluate the selected RO
and Donnan membranes and to address
the  stated objectives. The three stages
were  flushing,  system  blank de-
term/nations and membrane  solute
screening. Figures 1 and2showthebasic
components of  the RO  concentration/
Donnan  dialysis system  used  for  the
preliminary screening of RO and  ion
exchanges membranes
  Because  all  the  test   runs  were
performed  using  the  same  basic
membrane screening procedure, data on
the adequacy of the flush procedure were
obtained  from  each   of  the runs on
leaching,  chlorine  exposure,  organic
model  compound  retention  and humic
exposure.
  One important aspect to be considered
in any  system designed  to  concentrate
organic  chemicals from  water  is  the
possible leaching of organic compounds
from components  of the system. While
nearly all the elements of the RO system
are  fabricated  from  glass,  Teflon,  or
Table 1.    Model Compounds and the Conditions Used in the Evaluation of the Reverse Osmosis
          Concentration System

                              Concentration (ig/L
Model
Compound
Isophorone
2, 4 -Dichlorophenol
Qumoline
1 -Chlorododecane
Biphenyl
2,4-Dichlorobiphenyl
2,2',5,5'-Tetra-
chlorobiphenyl
2,6-di-tert-butyl-
4-methylphenol
Caffeine
Anthraquinone
S tear ic Acid
bis(2-Ethylhexyl)-
phthalate
Trimesic Acid
Furfural
S-Chlorouracil
Glycine
Glucose

Chloroform
Methylisobutyl
ketone
Humic Acid
Membrane
Screening
Runs
2500
2500
2500
250
2500
250
250

2500

2500
2500
2500
2500

2500
2500
2500
2500
N.T.

2500
2500

2000
Concentration
Membrane
Runs
50
50
50
5
50
5
5

50

50
50
50
50

50
50
50
50
50

50
50

2000
Stock
Solution
Solvent
Acetone
Acetone
Acetone
Acetone
Acetone
Acetone
Acetone

Acetone

Acetone
Acetone
Acetone
Acetone

Acetone
Acetone
HzO
H20
HZO

Acetone
Acetone

NaOH
Method
of
Detection
GC. GC/MS
GC, GC/MS
GC, GC/MS
GC. GC/MS
GC, GC/MS
GC, GC/MS
GC, GC/MS

GC, GC/MS

GC, GC/MS
GC. GC/MS
GC. GC/MS
GC, GC/MS

HPLC
HPLC
HPLC
Fluorescence
Liquid
Scintillation
GC
GC, GC/MS

HPLC
N. T. = Not Tested

-------
rfi^
i
i
i
1 i — ^
HTJ

±


-•
T
Water Recycle
1



v 	
?"
1



T
i

Donnan
Membranes

Cooling
Coil 	
Strip A


Permeate $, /
Mi*
C
'


—3 c

Solution jk
Reservoir 4» CjŁ 	 &4 —
er Reject ^ _ (p335
f
x 44


>

serf Bypass "


/
Drum
f/ow- ^
rafo/1
1 	 1
1 1
&rf _.>.. l .^5


Lr~ip/y
1 	 1 Monitor
200 40°
9 ?

L-L@ ?
n








R.O.
Membrane




To Waste r;r^tlnn PumP *»««"">ly
                        Pump
            Donnan
            Circulation Loop
     Reverse Osmosis
     Circulation Loop
Figure 1.    Major components of a reverse osmosis (RO) concentration/'Donnan dialysis
           organic recovery system.
stainless steel, other components of the
concentration  system  such  as
membranes, adhesives  and seals are
made from synthetic organic materials.
Chemical analyses for organic leachables
included examination for low molecular
weight  halogenated  compounds
(primarily trihalomethanes), semivolatile
priority   pollutants  by  GC/MS  (EPA
Method   625), as  well as the  model
solutes.
  It was necessary to evaluate the effect
that low levels of disinfectant would have
on  the  membranes and other  system
components because they are exposed to
residual  disinfectant  during  the
preparation   of  drinking   water
concentrates. The effect of the chlorine
exposure was determined by dosing the
system   blank  water  with  sodium
hypochlonte   and  examining  the
recirculated water after 4 hours.
  Experiments  designed to measure the
ability  of the  membrane system  to
concentrate the model solutes were also
conducted and  represented the  single
most important aspect of the project. The
two primary  objectives  of  these
experiments were  to  determine  the
membrane's ability to retain the  model
compounds (i.e., its solute rejection) and
to determine the extent of  losses via
mechanisms  such  as   volatilization,
adsorption, or solubility limitations. Each
of the four membranes (cellulose acetate,
FT-30,  Nation®, and MA 3475)  were
evaluated in a series of membrane/solute
screening  tests. Three  samples  were
routinely collected during each  run. The
recirculating water or feed was  sampled
                                    pH monitor/
                                     Controller
    NaOH
                                 Concentrate
                                    Bath
                                                                    HCL
Figure 2.    Simplified drawing of a Donnan dialysis test apparatus.
at 5 mm (0.08 hr) and at the end of a 4-hr
period  of total recirculation (i.e., where
both the permeate and concentrate are
returned to the  process  reservoir) to
allow the calculation of solute losses to
the environment and system components
(adsorption,  volatilization)   A  third
sample, taken from the permeate stream,
was collected after 4 hr. By utilizing this
type of sampling program it was possible
to determine  solute rejections  for each
membrane  using  Equation  1  and to
estimate system losses by comparing the
initial and final concentrations of solutes
in the recirculating feed solution.

 % Rejection  = 100
              (1 -Cp/Cf)(Eq  1)
where
         Cp  = permeate cone,  of solute
         Cf  = feed cone, of solute.

  Since very  little water is transported
across  the  Donnan  (ion  exchange)
membranes a permeate sample  is not
relevant.   Instead,  the  Donnan
membranes were dialyzed against a salt,
acid, or base solution, and this "strip" or
"pump" solution  was  sampled at the
same time (4  hr) as the last sample from
the process reservoir. This allowed an
estimation of the amount of solute lost to
the chemical pump solution.
  Because humic acid comprises a major
portion of the dissolved organic material
in drinking water,  its  effect  on the
membrane  systems  and  model  solute
rejection   was  surveyed.   The
concentration of humic acid varies from
source to source but it is not uncommon
to find 1 -5  mg/L  of humic  material
present   in  surface  sources.  A
concentration of 2 mg/L was selected for
this study

Membrane Concentration  Tests
  The primary objective of the membrane
concentration tests was to evaluate the
efficiency  of the RO/Donnan dialysis
concentration system for concentrating
trace levels of model solutes (1 -50 ppb) in
500L of water.
  Figure 3  illustrates the major process-
ing steps in  the  concentration of 500
liters  of organically  spiked water. The
concentration was designed to reduce
500  liters to  10  liters and incorporates
two major types of membrane processes,
RO concentration and Donnan dialysis
First, an RO system was used to reduce
(and concentrate) three 167-liter batches
to 13.5 liters each. RO concentration was
temporarily interrupted at 12.6X volume-
tric concentration as this level of concen-
tration was found to approach the in-

-------
organic solubility threshold. At this time,
the second membrane process, Donnan
softening, was used to reduce the con-
centration of calcium  ions and thereby
avoid any inorganic salt preparation
  Following  the 4-hr Donnan softening
treatment, the RO concentration was
resumed until a 50-fold concentrate was
obtained. A slight volumetric decrease
(with  an accompanying increase in
concentration  factor)  was experienced
during  Donnan softening due to osmotic
pumping of water from the concentrate
into the salt "pump" solution.
  In Figure 3, the asterisks indicate the
samples  that were collected at  various
points  during the concentration  run to
monitor its progress. Three streams were
sampled  during each concentration run:
the concentrate (also termed "feed" and
designated F); the RO permeate, designated
P; and  the salt pump solution, designated
S.


Results and Discussion
Leaching Studies
  Leaching of organic compounds from
the membranes and system components
was studied in two series of tests. In the
first series, the samples were only
examined for model solutes. The existence
of model solutes in the system blanks was
sporadic and at trace levels and probably
represents slow,  reversible adsorption
and desorption of spike organic  solutes
from earlier exposures to the membrane.
  Potential leaching of organic compounds
from the membrane systems was examined
in more detail in the second level of tests
where  gas chromatography (GC)/mass
spectrometry (MS) was used  in addition
to  the routine  GC/flame  lonization
detector, GC/electron  capture detector,
and high performance  liquid chromatog-
raphy  methodologies  In  this series of
tests, the effect of chlorine exposure on
the types of leachable organics recovered
in the  blanks was also examined. Low
levels  of a  limited number  of  solutes
occurred sporadically for all the membrane
types (in the absence of chlorine) with no
strong  trends or consistency  of results
evident  In tests with  chlorine present,
the analysis for model solutes  did not
reveal  any significant variations from the
prior tests,  but a  greater number and
variety of compounds  were detected by
GC/MS. While tentative identifications
could  be made  only  for some  of the
compounds, it was important to note that
none of them were obvious chlormation
by-products. Several of the compounds
detected in  the FT-30 membrane blank
before  and after chlorine exposure were
nitrogen-containing compounds which
may have  originated  from the barrier
coating of the thin-film composite, FT-30
membrane.
Membrane Screening and
Concentration Studies
  Membrane rejection values from both
the screening and concentration runs are
presented  in  the  full  report  for the
cellulose acetate and FT-30 RO membranes.
Rejection values indicating the effect  of
humic acid are also provided.
  Several  items that  require  further
explanation  are  the  negative rejection
values and rejections listed as indetermi-
nate  values.  Negative values result
whenever the organic solute concentration
is higher in the permeate stream than  in
the corresponding feed solution. Rejections
listed as indeterminate  values resulted
whenever  the  sample  concentrations
were below the analytical detection
limits.
                           500 Liters
 Samples
 FCP = 1
  cf = s.'
 PCF = 8
  While membrane rejection may be the
most important single factor in  any RO
concentration system, the success  or
efficiency is  nonetheless  measured  by
the overall recovery of as many different
types of compounds as possible. Tables in
the full  report list the percent  mass
recoveries for the three concentration
runs where:

Percent recovery = 100{Cf'VfF/C,FVf)(Eq. 2)

 and Cf, Cfp represents the initial and final
          concentration of the solute in
          the feed, respectively.

     vr, VfF represents the initial and final
          volume of the feed,  respec-
          tively.

Upon inspection, the recovery values for
many of  the solutes are extremely low,
especially for the cellulose acetate
membrane. Under the experimental
conditions, losses due to adsorption were
considerable
                                      Donnan Membrane
                                      (Softening)
                                                           R.O. Permeate
                                                           to Waste
  Fee - 12 7 ;S
 Fcf , 40 •                           I	Jl	^  Ft. 0. Permeate
 Per = 40                           L                        to Waste
  FCF = so

Figure 3.    General experimental design and sampling regime for the reverse osmosil
            concentration runs.

-------
  Two  major comparisons can be made
from the recovery data listed in the full
report. First, a comparison of the recoveries
of the cellulose acetate membrane (Run
618-41) and the FT-30 membrane (Run
618-51) indicated that  in practically all
cases, organic solute recovery by the FT-
30 membrane equaled or more often far
exceeded the recovery demonstrated by
the cellulose acetate membrane. This
was not unexpected since membrane
rejections determined for the two mem-
branes in the  screening tests and the
concentration runs reflected this same
trend.
  The second comparison was between
the FT-30 tests with (618-57) and without
(618-51) humic acid in the water matrix
There are 12 pairs of data  that  can be
compared directly For comparison, three
categories were  defined   1)  higher
recovery with  humic acid  present, 2)
lower recovery with humic acid present,
and 3) nodifference. Itwasstipulatedthat
the  differences  were to  exceed  10
percent to qualify as significant
  The results of this comparison showed
that higher recovery was exhibited for
two  compounds, 1-chlorododecane and
2,2',5,5'-tetrachlorobiphenyl, with humic
acid present; and that lower recovery was
exhibited for two compounds, 5-chloroura-
cil  and methylisobutyl ketone,  in the
presence of humic acid. Eight compounds
showed no change when humic acid was
present in the test matrix.

Adsorption Studies
  The  mechanism implicated  in the
disappearance  of most of the solute
material  was adsorption, even though
limited solubility, biodegradation, and
chemical decomposition of some solutes
may be complicating factors. Estimates of
solute  losses due to adsorption on the
system components  were provided  by
examination of the extent of model solute
loss during the 4 hr of  total solution
recirculation.
  Adsorption  losses of greater than  70
percent were observed for biphenyl; 2,6-
di-tert-butyl-4-methylphenol; anthraqui-
none, stearic acid and  bis(2-ethylhexyl)
phthalate in the presence of humic acid. In
addition to determining  the  actual
adsorption losses, it was possible to make
comparisons between:  1) the two RO
systems, 2) the RO systems and the
Nafion® membrane system, and 3) solute
solutions with and without  humic acid
present. The data indicated no significant
differences in solute losses between RO
membranes;  however,  the Nafion®
system exhibited significantly less adsorp-
tion than either  RO system. Comparisons
of the effect of humic acid in the water
matrix on solute losses to adsorption in
the three membrane  systems  revealed
appreciable changes in only four instances
for the 12 solutes for which data were
available  Because of the magnitude of
scatter and the fact that none  of the
changes  were greater than  50 percent,
these changes may not be significant

Recommendations
  • The FT-30 membrane  should be
used  in  future concentration studies
because  of  better  solute  rejection,
recovery, and  chemical  resistance as
compared to the cellulose acetate mem-
brane.
  • Steps should be  taken to satisfy
adsorptive  demands  during  sample
collections using the reverse  osmosis
concentration method. Preconditioning
of this nature could involve exposing the
membranes and other system components
to water (both at ambient and concentrated
levels) being sampled. This type treatment
should help maximize recovery of organics
in any stream where significant adsorption
may occur
  • The  apparent losses to adsorption
should be investigated in further detail
under conditions more favorable to
analytical chemistry.  By removing some
of the major analytical complications (i e.,
mixed salts, acetone, and humic acid) and
spiking  at higher  solute levels,  better
accuracy will be provided  Extended test
periods and expanded sampling schedules
should be instituted to  better determine
the attainment of steady-state conditions
and characterize  the  solute sorption
phenomena.  It is expected  that more
benefit will be gained from these modifica-
tions  than will be lost in deviating from
"real world" modeling Answers to these
questions should be sought.
    1. Is the  adsorption phenomenon
       associated with the membrane or
       other system components?
    2. Is the phenomenon reversible?
    3. Can the "demand" for solute be
       satisfied?
    4. Is the adsorption a function of the
       solute concentration?
  • The  search should continue for a
better anion exchange membrane. This
membrane is needed whenever concentra-
tion procedures require the  concurrent
removal of inorganic burden. Research is
being conducted to develop improved
membranes with polyolefin backbone
structures in the laboratories of several
major membrane manufacturers.  These
membranes should possess better chemi-
cal resistance than the styrene/divinylben-
zene  based membranes currently in
production. The apparent chlorine sensi-
tivity of the MA 3475 membrane indicates
manufactures' claims may be adequate
for  electrodialysis applications, but are
less than optimum for organic sampling.

-------
     S.  C. Lynch, J. K. Smith, L C. Ftando, and W. L Yauger are with Gulf South
       Research Institute, New Orleans, LA 70126.
     H. P.,ftinghand is the EPA Project Officer (see below).
     The complete report, entitled "Isolation or Concentration of Organic Substances
       from Water—An Evaluation of Reverse Osmosis Concentration," (Order No. PB
       85-124 147; Cost: $19.00, subject to change) will be available only from:
             National Technical Information Service
             5285 Port Royal Road
             Springfield, VA22161
             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
                                                                                              --USGPO: 1985 — 559-111/10781
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
     BULK RATE
POSTAGE & FEES PAID
        EPA
   PERMIT No  G-35
Official Business
Penalty for Private Use S300
                                                    60604

-------