EPA
          United States
          Environmental Protection
          Agency
             Environmental Research
             Laboratory
             Duluth MN 55804
National Effluent Toxicity
Assessment Center
          Research and Development
                        EPA/600/3-88/035 Feb. 1 989
Methods for Aquatic
Toxicity Identification
Evaluations
         Phase II Toxicity
         Identification
         Procedures

-------

-------
                                           EPA/600/3-88/035
                                              February 1989
          Methods for Aquatic

Toxicity  Identification Evaluations

   Phase II Toxicity Identification Procedures
                   Donald I. Mount
           U.S. Environmental Protection Agency
            Environmental Research Laboratory
           Office of Research and Development
                Duluth, Minnesota 55804
              Linda Anderson-Carnahan
                 U.S. EPA, Region IV
              Water Management Division
                Atlanta, Georgia 30365
               National Effluent Toxicity
                 Assessment Center
               Technical Report 02-88

-------
                                  Notice
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.

-------
                                  Foreword


Phase II  is  the  second in a  series  of three  guidance  documents for  use in
characterizing and  identifying the cause  of toxicity in  effluent  Toxicity  Identification
Evaluations  (TIEs). Phase  I  is intended to  provide  some knowledge  of  the
physical/chemical characteristics of the toxicant(s). Phase II builds on that information
and  is intended to  identify suspect toxicants. These are suspect toxicants because
evidence  has implicated them.  When evidence  exists  to implicate  one or  several
toxicant(s), Phase III confirmation should begin.

Phase II is incomplete  and does not provide methods for many constituents, such as
anionic metals and  polar organics. By using a loose-leaf binding, we hope to expedite
frequent expansions or  revisions.

The  sections of Phase I which address Health  and Safety, Quality Assurance/Quality
Control (QA/QC), Facilities and Equipment, Dilution Water, Testing, and Sampling are
applicable to Phase II.  Parts of the Introduction to Phase !  are also  relevant  to  this
manual. These  sections are not repeated in this report. We would especially  note  that
these  methods  are not intended  for chronic  toxicity  studies. We welcome  any
suggestions from users for use in future revisions.

-------
                                   Abstract
This manual  describes  test  methods that can  be  used  to  identify the  specific
chemicals(s) responsible for effluent toxicity when the source of toxicity is related to
non-polar organics, ammonia, or cationic metals.  These methods are not  intended to
be irrefutable evidence but only  to be providers of enough  evidence to suggest that
Phase III, the toxicity confirmation, should be started. Phase  III uses available methods
to provide adequate evidence that the true cause of toxicity  is consistently due to the
identified toxicants.
                                         IV

-------
                                      Contents


                                                                                   Page

Foreword	                       jjj
Abstract  	.. .1 ...............	       jv
Contents	 . . .	   v
Figures	'.'.'.'.'.'.'.'.'.'.'.'.'.	   vi
Tables   	; ; ;  '  vjj
Acknowledgments  	         vjjj

   1.   Introduction  	        ^^
         1.0  General Overview	   1_1

   2.   Non-Polar Organic Compounds	    2-1
         2.1  General Overview	   2-1
         2.2  Sample Volume	'.'.'.'.'.'.'.	   2-3
         2.3  Filtration  	'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'   2-3
         2.4  Column Size	'.'.'.'.'.'"''   2-3
         2.5  C-|8 SPE Column Conditioning  	           	   24
         2.6  Elution Blanks	.'.'.".'.'.'.'.'.'.'.'.'.'   2-4
         2.7  Column Loading  with Effluent  .  . .	          	   2-5
         2.8  Cia SPE Column Elution	    2-5
         2.9  Blank  & Effluent Fraction Toxicity  Tests	'.'.'.'.'.'.'.'.'.'.'.'.'.   2-5
         2.10 Concentration of  Toxic Fractions	                    	   26
         2.11 HPLC  Separation	. ' '.'	   2-7
         2.12 HPLC  Fraction Toxicity Tests   	'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.   2-8
         2.13 HPLC  Fraction Concentration and Toxicity Testing                            2-8
         2.14 GC/MS Analyses 	.'.".'."'.'.'.".'	   2-9
         2.15 Identifying Suspected Toxicants   	'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.'.   2-9
   3.   Ammonia	                  g «
         3.1  General Overview  	-	..'.'.'.'.'.'.'.'.	   3-1
         3.2  Equitoxic Solution Test   	'.'.'.'.'.	   3-1
         3.3  Results/Subsequent Tests	       3-3
         3.4  Zeolite Test	   3.4
         3.5  Procedure  	'.'.'.'.'.'.'.'"'   3-4
         3.6  Interpretation of Results	'.'.'.'.'.'.'.'.'.'.'.'.'.'.''   3-4
   4.   Cationic Metals   	     4_-l
         4.1  General Overview	      4-1
         4.2  Procedure	       4-2
   5.   References	

-------
                                          Figures
Number                                                                                Pa9e

2-1.     Schematic for Phase II identification of Cis SPE removed toxicants ..............   2-1
2-2.     Step 1 for concentrating the whole effluent chemicals
            on the Cis SPE column ....... ! ...................................   2-4
2-3.     Procedures for eluting the column with' a gradient of methanol/water solutions .......   2-6
2-4.     Procedure for combining and concentrating toxic fractions. Concentration
            factors are approximate based on 2 L of effluent and a 6 ml_
            combined eluate for  each methanol/water gradient ........................   2-7
2-5.     Procedure to fractionate concentrate on the  HPLC ..........................   2-8
2-6.     Procedure to concentrate toxic HPLC fractions
            (combined and individually) ..... ................................ • • -   2-8
                                               VI

-------
                                         Tables
Number
                                                                                      Page
2-1.     Comparison of Toxic Units in Each Toxic Fraction to Toxic Units
           of All Fractions Combined and Whole Effluent  	  2-3
3-1.     Percent Un-ionized Ammonia in Aqueous Ammonia Solutions
           for Selected Temperatures  and pH Values	  3-2
3-2.     Calculated Un-ionized Ammonia LC50 Values at Different
           pH Values for Ceriodaphnia Using an LC50 at pH 8.0 and 25°C of 3.0 mg/L   ....  3-2
                                            VII

-------
                            Acknowledgments
This report is a product of the effort of many individuals.  Methods for  Phase II have
been under development over the last few years. As with Phase I, throughout the text
the reference to "experience" in the approach or technique is the collective experience
of the  Environmental  Research Laboratory  (ERL-Duluth) effluent group.  That group
consists  of: Elizabeth  Durhan, Gary  Ankley,  and  Teresa  Norberg-King  of  ERL-
Duluth, Joe Amato, Larry Burkhard, Art Fenstad, Jim Jenson, Marta Lukasewycz, Greg
Peterson, Eric Robert  and Jim Taraldsen  of American Scientific International, Inc.
(AScI), Duluth.
The assistance of Dorette Gueldner (AScI) in the typing, organization, and revisions is
greatly appreciated. Larry, Teresa, Liz and Gary reviewed the various drafts to improve
the document. Final  reviews were  willingly  performed by Evelyn  Hunt (ERL-Duluth)
and Teresa. In addition, Teresa prepared the figures and made changes on drafts and
final versions.

Through  the support of Rick  Brandes (EPA, Permits Division) and Nelson Thomas,
(ERL-Duluth) the National Effluent Toxicity Assessment Center (NETAC) continues to
refine these techniques. This report  will  be updated  as methods are revised  or
developed, and we welcome suggestions.
                                      VIII

-------
                                               Section 1
                                             Introduction
 1.0 General Overview
The major  objective  of Phase II is to identify the
suspected toxicant(s). Some  general guidance may
be  furnished by the outcome of Phase I tests  (Mount
and Anderson-Carnahan,  1988; hereafter referred to
as  Phase I),,but  for many effluents, both separation
and concentration steps will be  needed in order to
achieve the objective.  For metals, Atomic Absorption
spectrometry (AA)  may be   sensitive  enough  to
measure toxic concentrations  directly in the sample,
the number of metals is small enough that toxicity can
be  attributed  without  separating one from another,
and ammonia measurements are  made without
separation or concentration.  However,  for organic
chemicals, the number is so large that  separation is
usually  necessary,  both  for analytical,  as well  as
lexicological reasons.

Because there are often so  many constituents within
the class of chemicals  identified in  Phase I, initial
efforts  are  most  productively directed  towards
separating the toxic from the  non-toxic  constituents.
The need to identify  the toxicant(s)  quickly  is a
temptation to analyze too soon. Some methods, such
as Gas Chromatography/Mass  Spectrometry (GC/MS)
for  non-polar; organics, allows  one to identify many
chemicals in a mixture. Identification must be followed
by association with toxicity, and this is very difficult to
do  on mixtures with many  constituents, for several
reasons:

•   Toxicity data for many of the  chemicals identified
    are usually not available.

•   Constituents are often not available  in pure form
    to measure their toxicity.

•   Interactions (additivity, synergism,  antagonism)
    are  not known  for the  given mixture, and one
    must know interactions to apportion toxicity.

If there is a single suspect toxicant such  as ammonia,
then separation  needs are limited  largely by the
analytical  requirements. If the  toxicity is caused by
one constituent,  the  number  of  other non-toxic
constituents  js irrelevant when  attributing  toxicity.
However,  Phase  I results do  not usually lead to a
 single suspect toxicant and, therefore, separation may
 be necessary.

 When a  method  for  separating  the  toxicant(s) is
 found, concentration may be an  inherent part of the
 procedure (as in solvent extraction) which will simplify
 the problem of finding  a  method  to concentrate the
 sample.  Through  the stages  of separation  and
 concentration,  measurement of toxicity will be  the
 only way to evaluate  the  success or  failure of the
 methods.

 The interpretation of results will often be different than
 we usually  encounter. In the  classical  research
 approach, experiments are designed to either accept
 or  reject  a hypothesis.  In Toxicity  Identification
 Evaluation (TIE) work,  an  experiment  frequently will
 allow us only to accept but not reject the hypothesis.
 For example, if ammonia is the suspect toxicant, and
 you want to know  whether there  are  additional
 toxicants, the ammonia can be removed using zeolite.
 If the effluent is still toxic, you can conclude that there
 are additional toxicants present.  If the  effluent is not
 toxic,  you  cannot conclude that  there  are  no
 additional toxicants because the  zeolite may have
 removed other toxicants in addition to the ammonia.

 The always-present  question of  more than  one
 toxicant  immensely complicates  data  interpretation.
 Phase I results are not likely to give even a clue to
 this question unless the toxicant class changes over
 time.  Phase II results are often such that one cannot
 tell whether the situation is one of partial removal of a
 single  toxicant or  toxicity  resulting  from multiple
 toxicants. Frequently the issue will not  be resolved
 until at least one toxicant  is identified and measured
 analytically.  Experience shows that the  best choice is
 to try to identify the toxicant that appears easiest to
 identify. Usually that will be a toxicant that can be
 separated  from  the  sample  (e.g.,  extracted  or
 recovered from a column that sorbs at least some of
 the toxicity) and for which there is a broad spectrum
 analytical  method. Above all, data should always be
 interpreted under  all  probable  scenarios, i.e.,  one
toxicant,  multiple  toxicants, and  even different
toxicants from sample to sample.
                                                 1 - 1

-------
As  effluent constituents  are identified,  a sorting
process begins in which a  decision must be made as
to the probability that each one identified is a cause
of toxicity to the test organism being  used.  Usually,
this is based on the estimated concentration and the
constituents' toxicity. Analytical error  in quantitation
may be large (10X or more) because recoveries and
instrument response factors probably will not yet have
been determined.  A toxicological  concern  is the
uncertainty caused  by  differences  in  species
sensitivity and  water quality effects. Error of the LC50
value will vary depending  on the  quality of the test,
the  number of  times  it  was repeated,  and  the
completeness  by which the results and  conditions
were described.  Species sensitivity frequently varies
from 100X to 1.00QX. We have found 1,OOOX to be a
minimum difference in  measured concentrations
versus literature LCSOs in  order to reject a chemical
as a suspect when the uncertainty of  the  LC50 data
is high.  If one has good  data for  the  test species
being used, then  this  difference  may  be reduced
(e.g., to  10X). Since  these decisions are always
subjective they will sometimes be wrong  no matter
how carefully they are made. Perhaps most important
is to make these decisions in an interactive process,
First evaluate  candidates  that have concentrations
higher than or closest to their LC50 values and if
these prove to be negative, then examine those that
have concentrations farther  below the LC50 values.
The suspected toxicant concentrations  at the dilution
equal to the LC50 are the important concentrations to
compare. At some point, a decision must be made
that the toxicants have  not yet been measured and
different  sample preparation or analyses must  be
used.


For some effluents, Phase I results  will  not provide
any real help in Phase II. In  these cases, more drastic
characterization steps may be helpful. Boiling, ashing,
and vacuum distillation, among  others,  may be  tried.
We have little experience upon  which to  recommend
procedures  in  these cases. The  most  important
concern is to realize that the more severe the effluent
treatment, the more  likely  it is that  artifacts will  be
created.


Phase  II  efforts  should   grade  into   Phase  III
Confirmation (Mount,  1988; hereafter referred to as
Phase  III) as soon  as good evidence is obtained that
one or more candidates are probable toxicants. The
primary product of  Phase II  is the chemical identity of
the suspected toxicants to furnish  the  basis  upon
which Phase III testing will be conducted.
                                                 1 -2

-------
                                              Section 2
                                Non-polar Organic Compounds
2.1  General Overview
A  schematic of the general procedures followed in
identifying  non-polar organic  toxicants  is  shown in
Figure 2-1.  In this  procedure, the  C-\Q Solid Phase
Extraction (SPE) column is used to extract non-polar
organic compounds from  effluent  samples.  These
compounds are then  selectively  stripped  off  the
column by eluting the GIS sorbant with  solvent/water
mixtures that are increasingly  non-polar. As a series,
the "fractions"  resulting from  column elution contain
analytes that are decreasingly polar and decreasingly
water soluble. Each fraction is tested for toxicity. The
fractions that exhibit toxicity are then diluted in water
and  a  smaller  Cis SPE  column is used to con-
centrate the toxicants. The column is dried and eluted
with  a small volume of  100% methanol.  The resulting
concentrate  is  chromatographed  using  reversed
phase High Pressure Liquid Chromatography (HPLC).
HPLC fractions are collected  and tested for toxicity.
The  toxic fractions  are again  concentrated  in 100%
methanol  and tested as before and analyzed  using
GC/MS.  Those  constituents  identified  are  roughly
quantitated,  by  assuming that the identified con-
stituents and !the  internal  standard  have the  same
response factor, and the estimated concentrations are
compared to LC50 values for  each chemical. If good
suspects  are identified,  Phase III  steps  begin.  In
addition,  mass  balance  testing  (described later)
should  be  started  to  determine  whether  more
toxicants are present.  If no candidate toxicants  are
identifiable, more separation on the HPLC may help if
there  are  many constituents present.  Additional
constituents  can  be  identified  by increasing  the
concentration factor by using larger effluent samples.
At some point,  the probability that the toxicants  are
not chromatographing  or  the  mass  spectrometer is
not detecting the toxicants, must be considered. Use
of other types of mass spectrometry, such  as liquid
chromatography/mass spectrometry (LC/MS) or direct
probe mass spectrometry should be explored.

It  is  likely that only a small  percentage  of  effluent
samples will require just one sequence  as shown in
Figure 2-1.  Many  effluent samples require  several
fractionation sequences  before good suspects  are
found. Once the toxicants are separated  into HPLC
                                                                     Effluent Sample
                    I
                                      Toxicity Test
                                      Toxicity Tests
         Concentration of Toxic Fractions*     Toxicity Tests
              Methanol Fractions
                    I
              HPLC Fractionation
                                      Toxicity Tests
                    I
      Concentration of Toxic HPLC  Fractions   Toxicity Tests
                    I
             GC/MS Identification
                    I
      Compare Concentrations to LC50 Values
  *ln rare cases, GC/MS can be useful here.

 Figure 2-1.   Schematic for Phase II identification of C18
            SPE removed toxicants.
fractions,  the  effort is largely an analytical one until
some candidates are identified.

The Cjs SPE sorption, rather than non-polar solvent
extraction, is used because there is not a good way
to determine whether the toxicant was extracted from
the  effluent.  Since  the  non-polar  solvent is  not
miscible with water, getting extracted chemicals back
into water to test  for  toxicity  is difficult, and  the
solvents  themselves  are  generally  toxic.  The
extracted  sample will also be toxic  as a result of a
small amount  of the solvent dissolved in the sample.
Residual levels of solvent can be stripped from the
extracted  effluent prior  to toxicity testing only if the
toxicants are not stripped as well. Effluent extracts will
contain more interfering constituents, emulsions may
be  a problem, and  the HPLC fractions may be less
well defined. The Cia SPE column is easier when it
works. If it does  not work, extracts can be used if the
                                                 2- 1

-------
toxicants are  non-volatile  under the  required
conditions and toxicity tracking is done.

The  sorbants recommended herein for use  in  SPE
and  HPLC columns are chemically identical.  The
column packing is composed of porous silica which
has been reacted with octadecyl groups to produce a
surface  GIB  layer one  molecule thick. The same
compounds will  partition on both columns and the
order of elution of chemicals will be the same. The
major difference  between the  SPE  and  HPLC
columns is the resolution achieved. The particle size
employed in  HPLC columns is smaller, providing  a
greater surface area and better resolution.  Despite
poorer resolution, SPE columns have the advantage
of generally  possessing  a higher  loading capacity.
The SPE column is used as a preparatory column for
sample cleanup  while the  HPLC column  gives far
more refined and controlled separation.
The  mechanism of extraction  with  Cia sorbants is
relatively simple.  Extraction  of effluent compounds
occurs  because  the GIB sorbant more  strongly
attracts for the  non-polar  compounds  than  the
surrounding water molecules.

Sorption of non-polar organics is  also  influenced by
Ionic strength, pH,  and  total  organic carbon (TOG)
levels.  The important consideration is  to  check for
toxicity breakthrough  in  the  post-column  effluent.
Even when the capacity of the column is exceeded,
the procedures described in  this subsection  may be
useful in concentrating part of the non-polar effluent
toxicants.

Just  as non-polar organic compounds are separated
from other effluent components  while passing  the
effluent  through  the column,  these  non-polar
compounds can be further segregated according to
their polarity during column elution. To elute toxicants
extracted by the C-\Q sorbant, the eluting solvent must
have a  higher  affinity than octadecyl  for  the
compound. Elution  solvent  choice  is complicated
because the identity of  the toxicants is not known.
Without that  knowledge, the degree of  solvent
strength required to elute the toxicants is unknown. It
is known that  the octadecyl group is extremely non-
polar,  and  very  few  solvents,  except  hexane
(E° = 0.0), come close to its strength.  Unfortunately,
except for its compatibility with GC/MS, hexane does
not meet  the necessary criteria  of miscibility  with
water and low toxicity. Methanol  (E°=0.73), on the
other hand,  is a far weaker  solvent than may  be
desirable but is still less polar than water  (E° >0.73).
Methanol is  also  not the solvent of choice  for  GC
analysis, but can be used even though  column life is
shortened. Since methanol has a very low toxicity apd
it will elute chemicals from Cis,  it  is an  acceptable
solvent for our purposes.
Because of the  limited  number of solvents meeting
the criteria for acceptability, different concentrations
of methanol in water are used rather  than different
solvents  to selectively  elute compounds.  Using  a
step-wise  gradient  elution entailing  sequential
column rinses with successively increasing  methanol
concentrations,  the  relatively  hydrophilic,  polar
compounds  are  eluted  first,  while  the  more
hydrophobic non-polar compounds  are  eluted last.

Given the relatively weak strength  of methanol as a
solvent for non-polar  compounds,  it  is possible  that
effluent  compounds, sparingly soluble  in water, will
not be  eluted from  the C^Q  sorbant.  If toxicity  is
extracted by the  column  but not eluted by  methanol,
less polar  solvents  must be  used.  These will  be
discussed later.  Recovery of 100% of each effluent
toxicant in the   C-\Q column fractions  may not  be
crucial, because at this  stage,  the concentration of
toxicant in the fraction and the toxicity of the fraction
are compared. Assumptions about  the concentration
of toxicant  in the whole  effluent are not made at this
point, nor is any statement made regarding recovery
of whole effluent toxicity  by  C^Q column fractions. In
later stages of  Phase II, column  recovery  must be
defined  in order  to make meaningful inferences
regarding the relationship between  the concentration
of the suspected toxicant(s) and the observed toxicity.

The efficiency of C-\Q column  recovery  consists of
extraction  efficiency (i.e.,  how  well the column
sorbant removes the  effluent components) and elution
efficiency  (i.e., how well sorbed effluent  compounds
are removed from the column by the  solvent gradient
used). For purposes of this study, "efficiency" applies
only to  recovery of  those  compounds  causing or
affecting  effluent toxicity. The question of  extraction
efficiency  can   be  determined  by  measuring  the
toxicity of  the effluent  after it passes through  the
column.  By  comparing the toxicities of aliquots
collected after different  volumes  of effluent have
passed  through  the column,  a  determination of
toxicant extraction  efficiency can be made separate
from column selectivity.  If there is toxicity, but  it  is
independent of  the  volume of  effluent previously
passed  through  the column, then the post-column
effluent  toxicity  is probably caused by  toxicants  that
are not extracted by  the column. If the  toxicity of the
post-column effluent  increases with the volume of
effluent  passed through the column, the capacity of
the column to sorb the toxicants was exceeded.

Elution efficiency can be approximated  by  summing
the amount of toxicity in the toxic  fractions. This will
usually be somewhat imprecise for several reasons. A
single  toxicant may occur  in   more  than  one
contiguous fraction, in which case  a small amount of
the toxicant in  one  fraction may  not be detectable
because  it  is present below the LC50 concentration.
(This problem  can  be  solved by  combining  all
fractions  and measuring  their total toxicity.) If more
                                                2-2

-------
than one toxicant is present, they may not be strictly
additive in  their toxicities, and  when  separated into
different fractions the sum of the fraction toxicities will
be  low even if  extraction and elution  were 100%.
Table  2-1  illustrates  a  hypothetical  example.
(Section 7 of Phase I  defines toxic units.) Frequently
the  toxicity  of the  combined  fractions  will  be
somewhat greater than  whole effluent. This may be
caused by  enhanced toxicity  due to the methanol  in
the  fractions.  To  compensate  for  toxicity
enhancement by methanol, methanol has been added
to the whole effluent, but this  may stimulate biological
growth. If this happens, the test is negated. Fractions
should  be tested  in  water with low TOC  and
suspended  solids (SS) to  lessen  effects  of the
characteristics on toxicity.

Once significant  toxicity is  found in  the C-\Q  SPE
fractions, HPLC  separation must be done to reduce
the  number of  chemicals associated with the
toxicant(s).  In the procedure described below, the
HPLC  fractions will contain methanol. Aliquots  must
therefore be diluted until  the methanol is below a toxic
level before the fractions can be tested for toxicity.
One must calculate the necessary dilution to reach a
non-toxic level of methanol and then  determine how
much  effluent must  be  concentrated on the  SPE
column (including the pre-HPLC concentration  step)
in order to have enough  toxicant in the HPLC fraction
following dilution, to detect toxicity. This, of course,
depends on the toxicity of the effluent.  In the sections
to follow, a typical set of dilution  and concentration
factors are given, but these must be altered  to  fit
each situation.

The toxic HPLC fractions are concentrated using a
small SPE column and then analyzed by GC/MS. The
estimated concentrations of  constituents in  the final
concentrate (based on an internal standard) are then
compared to their toxicity to decide which ones may
be sufficiently high in concentration to cause toxicity.
If no candidates are found, more concentrations,
other analytical  methods (e.g., LC/MS), and  better
separation are possible further steps.

     Table 2-1.    Comparison of Toxic Units in Each
                Toxic Fraction to Toxic Units of All
                Fractions Combined and  Whole
                Effluent
     	Source	Toxic Units	
     Toxic Fraction (% Methanol)
         75%                      0.5
         80%                      1.2
         85%                      0.6
         SUM                     2.3
     Combined Fractions               2.7
     Whole Effluent                  2.5
2,2 Sample Volume

The volume of effluent needed depends  on  the
toxicity of the effluent,  the toxicity of  the  chemicals
causing effluent toxicity,  and the  sensitivity  of  the
analytical  method. Since  only the first of  these  will
usually be known when Phase II begins, trial and error
will dictate needed volumes.  For  effluents with LC50
values from 25-100%,  2000 mL have usually been
adequate.

2.3 Filtration

A  1 pm glass filter should be prepared as described
in  Section  8 of Phase I.  The filter should  be rinsed
with 200  mL of dilution water,  rather  than pure
distilled water and a  sample collected after most of
the volume has  been  filtered for the filter toxicity
blank. In subsequent steps, a dilution water blank will
be  collected after passing through the SPE column.
This water must not  have sorbable organics in it or
toxicity in the fraction blanks may occur. The same
type of water should be  used for the filter blank as for
the column blank. Therefore, a  reconstituted water
should be  used for these procedures.  Formulas for
reconstituted  water  can be  found in  Peltier  and
Weber, 1985.

The volume of effluent that can be passed  through a
single filter is  sample  specific. If more  than one filter
is  needed, a separate blank  should  be prepared for
each  filter.  Whether  a  vacuum  can  be  used  for
filtering or whether a  pressure system  must be used
will be ascertained in the volatility tests of Phase I.
When uncertainty of which method to use exists, use
pressure filtration, and a  toxicity  test on the filtered
effluent sample is prudent.

2.4 Column Size

Various sizes of C-\Q SPE columns are available. We
routinely have used BakerRl  6 mL high capacity (HC)
columns,  putting  1,000  mL through  one column
(Figure  2-2).  So  long  as  the  column  does  not
become clogged and  toxicity break-through does  not
occur, the column capacity has not been reached. In
the following description, volumes  for a 6 mL HC GIB
SPE column are used.

Vacuum  manifolds  are available  for  drawing the
sample and solvents through  the column. The  use of
vacuum is  much  less controllable than pumping  the
sample and  therefore  is   discouraged  because
extraction efficiencies  may vary. It is. adequate in  the
hands of a careful  worker,  but in routine  use,  flow
rates are likely to vary greatly, and there is a danger
that the  column will  go dry when it should not. A
pump  is  more  convenient  for  the large volume
                                                       1 J.T. Baker Chemical Company, Philipsburg, NJ
                                                 2-3

-------
                                                 Step 1
                                1000 ml.
                                Effluent
                                                                     Concentration Factor
                           1X
                                                        6 mL High Capacity
                                                        dsSPE column
                              Test Post G!8
                           1X
 Figure 2-2.    Step 1 for concentrating the whole effluent chemicals on the Cia SPE column.
effluent samples where flow rate can be  better
controlled, and therefore a pump is recommended.

Whichever  system  is  used,  it must be  made of
materials that dilute acid and solvents do not destroy,
or from which  chemicals  are  not leached  that are
toxic  or interfere with analytical  measurements.
Teflon, glass, and stainless steel are acceptable.

2.5  Cis SPE Column Conditioning
The 6 mL Cis SPE columns are conditioned  by
pumping  25  mL of 100%  methanol  through the
column. Before the packing goes dry, 25 mL of high
purity distilled water must be added. As the last of
that water is passing through, 25 mL of dilution  water
is added, and the last 10 mL  is collected for a column
blank toxicity test.  If the  pH of the sample  must be
altered in order for the toxicants to sorb on Cis. then
the 25 mL aliquot of dilution water should be adjusted
to the same pH.  Pumping is continued until no water
emerges from the column.

2.6  Elution  Blanks

A three mL volume of  25% methanol  in water is
pumped through  the dry column and  collected in an
analytically  clean,  labeled vial  (since  glass  also
contains metals, the vials  should  also be  acid
leached). This procedure is  repeated  with 3  mL
volumes of the 50%,  75%, 80%, 85%, 90%,  95%
and 100% methanol in water. The column should be
allowed to dry between each elution with the different
3 mL volumes of methanol/water  mixtures. These
methanol solutions may also have to be  pH-adjusted
for better elution efficiency,  and, if so, the blanks
must also be adjusted.
                                                2-4

-------
2.7  Column Loading with Effluent

The same column is again conditioned with 25 ml_ of
100% methanol and 25 ml_ of high purity water, as
described above. Without allowing the  column to dry,
1,000 ml_  of filtered  effluent sample is pumped
through  the column at  a rate of 5 mL/min (Figure 2-
2). Three  10  ml_ samples of the post-Cis column
effluent  are collected after  25 ml, 500 ml_ and 950
mL of the  sample passes through the column.  Each
10 mL aliquot is used  in a toxicity test to determine
the presence  of toxicity  in the effluent after  it has
passed  through the column. This  information can be
used  to determine whether  toxicant  breakthrough
occurred.  As  Phase II  progresses, it can be very
helpful  to  catch 20-30 mL and  make dilutions to
obtain LC50 values. Pumping is  continued until  no
effluent  emerges from the column.

2.8  Ci 8 SPE Column  Elution

Next, a  3 mL volume of the  25% methanol/75% water
mixture  is pumped through  the column and collected
in a  labeled,  analytically  clean  vial. For  better
separation, two 1.5  mL volumes to elute the column
are recommended (for  both  blanks  and sample
fractions).  Subsequently,  3 mL volumes  of each of
the 50%, 75%, 80%,  85%,  90%, 95%  and  100%
methanol  are  pumped through  the  column and
collected in separate vials  (Figure 2-3).  The  next
elution volume should not be added until no more of
the precedinig one is emerging from the column. In
general, if the toxicants are removed by the column
best at  pH 3,  then the methanol fractions  should  be
eluted at pH 1  or pH 11 and vice versa.

This  entire  procedure  is repeated using  a  second
column  for the second 1,000 mL of filtered effluent.
The dilution water  column  blank samples  should  be
kept  separate. The corresponding  fractions  from
either the blanks or the sample from each 1,000 mL
fractionation can be combined. For example, the 3 mL
100% methanol sample fraction from the first column
and the 3  mL 100% methanol  sample fraction from
the second column are combined for a total of 6 mL.

The vials containing the methanol/water fractions are
sealed  with perfluorocarbon  or foil-lined  caps and
stored under refrigeration. These fractions represent a
"first  cut" separation  of effluent  components.
Volumes will vary  if columns of  different sizes are
used or if  the particular effluent under study  or the
research  question  being  posed  dictates method
modification.

2.9  Blank and Effluent  Fraction Toxicity
    Tests

In the  initial stages of Phase  II,  toxicity tests are
conducted  on  GIB  SPE column fractions and  blanks
to detect the  presence of toxicants,  rather than to
quantify the magnitude of the toxicity  in each. As in
the toxicity tests conducted  during Phase I, there is
no need for careful measurement and maintenance of
test  solution chemistry,  nor  is there a real need for
duplicate exposures. The  major purpose behind  this
tier  of tests  is  strictly  qualitative, i.e., to assess
whether or not acute toxicity is present or  absent in
the fractions and blanks. As suspects are  identified,
quantitative toxicity measurements are  needed  to
compare with the analytical measurements.

The  next step  is to test  the  blank and sample
fractions for toxicity. The methanol in the fractions
limits the  concentration  that can be tested, and the
amount of methanol in  the  100%  fractions dictates
the necessary dilution. We  usually inject 150  piL of
each blank and sample fraction into 10 mL of dilution
water.  This   step  will  give  a  1.5% methanol
concentration  that is  below  the  LC50  for both
Ceriodaphnia  and  fathead  minnows  in  the  100%
fraction. The  resulting methanol concentration must
be adjusted for the species used (cf., Section 8 of
Phase I  for methanol  LC50 values). Usually five
animals in each  10 mL  aliquot are  used without
duplicates. Using the above  volumes,  the  tested
solution has  a  greater  concentration  than  whole
effluent, assuming 100% extraction  and 100% elution
in one fraction. These test solutions can be diluted as
described in Phase I  to provide an  LC50  for each
sample fraction.  Usually blank fractions do  not need
to be diluted.  If dilutions are  done, the initial volumes
are best increased to 300 jiL/20 mL to  provide 10 mL
with  which to make serial dilutions.

Individual  chemicals in the  fractions could be toxic
even when they  are not toxic  in the  whole effluent,
since the concentration tested may  be as high  as 5X
whole effluent. Therefore, to be toxic at whole effluent
concentrations, an individual fraction  must have an
LC50 of 20% or  less. Since  there is no way to know
whether the toxicant(s)  eluted in  more than one
fraction  or what the percent extraction and  elution
was, fraction toxicity up  to 100% (5X  whole effluent)
should not be disregarded.

If toxicity occurs in any of the fraction blank  tests and
it is small  relative to the toxicity in the corresponding
sample fraction (e.g., 20% mortality versus 80%), the
sample  fraction  results  are  not negated.  If all
organisms die, dilutions should be  tested  to  make
sure  the sample fraction is  substantially more toxic
than  the  blank. Blanks should not, in  general, have
measurable toxicity.

In addition to concentrating column artifacts to toxic
levels,  effluent constituents  present  at non-lethal
levels may be concentrated to toxic levels in this test
if they have a  relatively  high recovery  value. On  the
other hand,  actual  effluent toxicants  with  poor
recovery may not be present  in these test solutions at
                                                2-5

-------
                                                Step 2
                       6 mL High Capacity
                       CiaSPE column
                                            Y
                                                                     Concentration Factor
                                                                              333X
                                                                              Each
                  J
J
                                  Conduct Toxicity Test on Each
                                    (150jut_in 10mL Water)
                                      5X
      Figure 2-3.   Procedures for eluting the column with a gradient of methanot/water solutions.
toxic levels. Spurious results of this  nature  will  be
identified in the later stages of Phase II and/or in
Phase III.

2.10 Concentration of Toxic Fractions

As  explained in the  General  Overview, this initial
fractionation provides only a general  separation of
non-polar organics.  Except in relatively uncom-
plicated effluents, GC/MS analysis  of toxic CIQ SPE
fractions  will  result  in extremely  complicated
chromatographs from which the causative  agents
cannot  be  distinguished  from  other  effluent
components. Rarely should GC/MS analyses be tried
on  these fractions.  In  most  cases,  a secondary
fractionation using HPLC is needed to further simplify
toxic effluent fractions  prior  to  component
identification.

Because fractionation by HPLC will further dilute the
concentration of effluent components in the C^Q SPE
fractions, these fractions must be concentrated prior
to injection into the HPLC. This concentration step
             will increase the concentrations of constituents in the
             HPLC fractions and rid the fractions of water.

             The toxic effluent SPE fractions are combined (Figure
             2-4)  and diluted 1:10 with high purity  water.  The
             corresponding  blank  fractions  are  similarly treated.
             The total volume should not exceed 100 mi. for the 1
             mL  GIB column. If the  25%  or 50% fractions  are
             toxic, then  greater than  a 1:10 dilution  may be
             needed. This  must be  determined  by  testing  for
             toxicity  loss after the concentration step.
             For this step, a 1 mL Cis column is conditioned with
             10 mL of methanol and 5 mL of water similar to the
             procedures  described  in  the  SPE   Column
             Conditioning  Section. Column blanks are advisable,
             and if done, the water volume must be increased  to
             at least 20 mL.  The diluted blank fractions  are  then
             drawn through the  1  mL Cis SPE  column under a
             pressure of 15 in Hg using a vacuum manifold. The
             solution passing  through the column cannot be tested
             for toxicity  because  of  its methanol  concentration.
                                                2-6

-------
The column is then dried for 5 min using a nitrogen
flow rate of 13 mL/sec.

After drying, the luer tip of the column is fitted with a
luer-lock needle and  100 pL of 100%  methanol is
placed  into the column using a  microliter syringe;
nitrogen is  applied to  the  column at a rate of 4
mL/sec to force the  methanol through the sorbent.
The first 100  pL aliquot of  methanol  applied  to  the
column will yield approximately 25 pL of eluate. Two
more 100  pL  aliquots  of 100%  methanol  are also
forced through the column. The final volume of eluate
collected will* be approximately 200 pL. Measure  the
exact volume  collected (using a pL syringe)  for
calculation of recoveries. For a given  total volume of
methanol, three separate smaller  elutions  are more
efficient than one large one.

The same 1 ml_ C-\Q SPE  column is reconditioned
following the; directions given above  in  the Column
Conditioning -Section.  It is then used  to concentrate
the  diluted  toxic SPE  column fractions, using  the
same sequence used for the  blank fractions (Figure
2-4).

                      Step 3
                            Concentration Factor
     Dilute Toxic Fraction(s) 1:10
            :   i                  33X


     Sorb on i mL C18 SPE Column
                  (discard
                  post-C18)
        Dry Gig Column with
        Nitrogen (optional)
               i
      Elute Column with 3-100nL
      volumes of 100% Methanol
      Collect Eluate Concentrate
               I
        Conduct Toxicity Test
            in 20 mL water)
                                10.000X
                               5X
Figure 2-4.
               Procedure for combining  and  con-
               centrating toxic fractions. Concen-
               tration factors are approximate based
               on 2L of effluent and a 6 mL combined
               eluate  for  each methanol/water
               gradient.
The  original  effluent  volume of  2,000  ml is now
represented by 200 pL or is nominally 10.000X more
concentrated,  ignoring the amount used for testing.
As work progresses and more quantitative results are
needed,  the  eluate volume must be  measured to
know the  concentration  factor. If  10  pL of  the
 concentrate is diluted to 20 mL in dilution water, the
 resulting  concentration  will be 5X whole  effluent.
 Dilutions  of this can then be  made to measure an
 LC50  and toxicity  recovery can  be calculated by
 comparing this  LC50 to  the  LC50 of the whole
 effluent. Remember that an LC50  of  20%  in  the
 concentrate test would  be equivalent to an LC50 of
 100%  in  whole effluent. One  cannot  compare  the
 concentrate toxicity to  that of the individual toxic
 fractions  because  some  of the  toxicant  may have
 been in adjacent fractions but  not detectable by the
 toxicity test.  The concentrate  toxicity may be lower
 than the whole effluent toxicity  because of  extraction
 and elution inefficiencies.

 The important concern here is not 100% recovery of
 toxicants  but  enough  recovery for  analyses to be
 done and to measure toxicity in  the HPLC fractions. If
 recovery is too low, changing the column drying time
 may help. Sometimes  recovery appears to  increase
 with drying time while other compounds are volatilized
 from  the  column  during the  drying  process.  For
 concentrates analyzed using GC, removal of water is
 critical  to  instrument performance. Where  greater
 concentration factors are desirable,  SPE fractionation
 can be repeated with additional 1000 mL volumes of
 effluent,  followed  by  combining the  3  mL toxic
 fractions before concentration. The size of the column
 used for concentrating may have to be increased with
 appropriate changes in dilution and elution volumes.

 2.11  HPLC Separation

 The same column  packing should be  used  in  the
 HPLC  column as is used in the SPE column  except
 for  particle size. At  later stages, when more is known
 about the  toxicants,  other columns may be used.

 The HPLC conditions presented  in this section  are
 general.  As  more  information on  the effluent  is
 gained, HPLC conditions should  be  modified to
 achieve better separation, and  higher concentration
 factors. We use an instrument  equipped with a 5 p
 GIB reverse phase  column (250 mm x 4.6  mm) and
 UV  detector.  An  elution   gradient  of  30%
 methanol/water to  100% methanol  over 20 minutes
 with a five minute hold  period  at 100% methanol is
 suggested.  Generally,  the  278  and  230  nm
 wavelengths  are  monitored  continuously  while
 spectral scans (210 nm-600 nm) are acquired  only
for  chromatograph peaks. First, 100 pL  of the blank
concentrate is injected  and 25-1  mL fractions are
collected  in analytically  clean glass vials (Figure 2-
5).  The same procedure  is followed using  100 pL of
the  effluent sample concentrate.  The vials must be
sealed (for example, with perfluorocarbon lined caps)
and stored under refrigeration.
                                                 2- 7

-------
2.12 HPLC Fraction Toxicity Tests

Toxicity tests on the  HPLC blank  and sample
fractions  using  non-replicated  exposures of  five
animals each, without dilution, are normally used. The
methanol in the HPLC fraction limits the concentration
that can be tested. The percent methanol varies with
the  particular fraction  so  those that  are  100%
methanol determine the  necessary dilution. Again, a
1.5%  methanol  concentration,  an  acceptable
methanol  concentration  for  many species, is sug-
gested.

If 100pLof the ~200 pL concentrate is injected on
the HPLC  (Figure  2-5),  then each  1  mL  HPLC
fraction equals 1,000  mL of effluent (assuming no
loss and  elution  in  one  fraction)  or a 1.000X
concentration. If each HPLC fraction is then diluted to
1.5% for testing  (150 pL to 10  mL)  the  resultant
concentration is 15X whole effluent concentration. Of
course, some loss of toxicant will  occur in each step
and the toxicity will usually be less.
                      Step 4
                             Concentration Factor
                 Inject 100 nL of the  10.OOOX
                 Concentrate on the
                 HPLC C18 Column
                        1
                 Collect 25-1 mL
                 HPLC Fractions
                        I
                 Conduct Toxicity
                 Test on Each (150
                 liL/10 mL water)
                  1.000X
                   15X
    Figure 2-5.
Procedure  to fractionate concentrate
on the HPLC.  Concentration  factors
are approximate.
The HPLC fractionation is precise enough that more
concentrate can be fractionated without the need  to
do toxicity tests, in order to increase sample quantity
for analyses.

The blank fractions should not be toxic.  If they are,
then dilutions will  have to be made on both  blanks
and toxic  fractions to tell if there is enough additional
toxicity in  the sample fractions to warrant analyses.

2.13 HPLC Fraction Concentration and
      Toxicity Testing
The toxic HPLC fractions  and corresponding  blanks
must be concentrated in a solvent suitable for GC/MS
or other  analytical  techniques.  The  procedure  is
identical  to  Section 2.10  Concentration  of Toxic
Fractions, and is  depicted  in Figure 2-6. Judgment
must be used to decide whether to concentrate each
                                        toxic fraction separately or to combine various toxic
                                        fractions prior to concentration. If, for example, three
                                        successive  fractions are  toxic, there is a  good
                                        probability that  the  same toxicant is in all three.  If
                                        there are three other toxic fractions separated from
                                        the first set  by  several non-toxic fractions,  there is a
                                        high  probability  that  the second three contain  a
                                        different toxicant from the first three. There is also a
                                        good probability that at least one non-toxic fraction
                                        on either side of the toxic fractions contains some of
                                        the toxicant.  The virtue of  combining fractions is to
                                        reduce the work load and increase the concentration
                                        in the  concentrate. The disadvantage  is that more
                                        constituents  that are not the toxicant  will  be included.
                                        The decision has to be  based on trial  and error and
                                        experience. As  soon as an identification is  obtained,
                                        then HPLC conditions (gradient and fraction size and
                                        number) can be optimized. Blanks corresponding to
                                        the toxic fractions are concentrated the same way.
                                                             Steps
                                                                                   Concentration Factor
                                                       Dilute Fraction(s) 1:10  10OX

                                                               i
                                                          Extract on 1 mL
                                                         C18 SPE Column
                                                                                 (discard post-Ci8)
                                           Figure 2-6.
  Dry C-jg SPE Column
      with Nitrogen
       (Optional)

         1
   Elute Column with
  3-100 pL Volumes
   of 100% Methanol

         i
     Collect Eluate
     Concentrate    5.000X

         1
   Conduct Toxicity
  Test(20nL/lOmL  10X
       water)

         I
       Analyze
    Concentrate on   5.000X
       GC/MS
Procedure to concentrate toxic HPLC
fractions (combined or individually).
Volume of eluate should be measured.
                                       The concentrate must be finally checked for toxicity
                                       before analysis, as described in Section 2.10. This is
                                       the  last opportunity  to  assure that the  analytical
                                       instrument receives the toxicant. Whether the toxicant
                                       is detected  by the analytical detector is always a
                                       question. Since GC/MS  detects only about 20%  of
                                                  2-8

-------
organic chemicals (cf.,  Phase I), even such  a broad
spectrum analytical instrument is not a certain way to
measure the toxicant.

2.14 GC/MS Analyses

Procedures and methods provided in this section are
based upon our experience  in  performing  GC/MS
analyses  on fractions  from  numerous effluents.  In
general, these  procedures are suggested and should
be used.

The  typical instrument  conditions are as follows: A
GC/MS system equipped for doing  standard residue
chemistry analyses is suggested; i.e., a 30 m capillary
column, electron  impact ionization,  scan range  of
50-500 amu, scan rate of 1 or 2  scans/sec,  a GC
temperature program of 50 to 300°C at 5°C/min, and
data system with library searching ability.

For the GC/MS analysis, the prepared blank and toxic
fractions (in ~200 pL of methanol) should be tested
for toxicity (Figure 2-6).  After verification  of the
toxicity in the methanol  concentrate, inject 1 or 2 uL of
the concentrate and collect the mass spectral  data.
Note, methanol is  not a  typical  solvent  for  GC
analysis. However, since only methanol fractions can
be tested  for  toxicity verification,  methanol solvents
must be  used. Injection of methanol  on a capillary
column will shorten its life. Thus, your routine GC/MS
QA/QC  procedures should  closely  monitor  the
performance of your column.

After collection of  the mass   spectral  data,  peak
detection,  and  integration, algorithms should be used
on the data.  All  detected  peaks  should be  library
searched;  reverse search is preferred. The EPA/NIH/
NBS library (approximately  45,000  chemicals) for
performing the  library searching has been adequate in
our laboratory.

Once the library  search results are  available,  the
search report  for each  peak must  be examined. By
this examination one must decide if the  identification
by the search  is valid and reasonable. In general, a
trained  GC/MS chemist is  required  to  do this
validation  process. Factors  considered  in our
laboratory  for  this process include:  A) are all major
ions  present in the correct proportions, B)  is this
identification consistent with other  information about
the fraction, C) do  forward and  reverse searching
provide similar fits,  and D) are  the library searching
fits  greater than  70%?  Factor  A must  be  met!
Consistency, factor  B, considers circumstances such
as "has the identified chemical  been found in vastly
different fractions,"  or  "has the same  identification
been given  to numerous  peaks in  the  same
chromatogram?" Both factors C and D are somewhat
relative and depend a lot on  the sample  and its
matrix. In addition, the toxicants are often very minor
components in the GC/MS  total  ion  chromatogram
and thus,  the quality of the mass spectral data even
after background  subtraction will lead to poor results
for factors C and D.

After examination  of the library search results, a list of
identified chemicals is assembled and evaluated using
the methods  in  the  following  section.  For  the
confirmation analyses we suggest EPA method  680
(Alford-Stevens et al.,  1985).

2.15  Identifying Suspected Toxicants

The goal of the rest of  Phase II is to determine if  all
the toxicants have been identified. Two parallel lines
of investigation should be pursued to reach that goal.
One is to develop evidence that the concentrations of
the suspects are sufficient to cause toxicity. The
second  is to develop data to estimate the proportion
of the  whole  effluent  toxicity that  is due to  the
suspected toxicants, so  a decision can be made as to
whether other toxicants need to be identified.

The first investigation should begin by comparing the
estimated  concentrations of  identified chemicals  in
the concentrate  at its  LC50  to their LC50 values.
Because at this stage,  the  compound quantification
will  have  been done using one  or  more  unknown
internal  standards and recoveries ,  considerable error
may be involved in  the estimate. Secondly, the LC50
data, if available,  are likely to be for  different species
than the one used in the TIE. Species differences are
usually  as  large as  100X and often, 1.000X.  Given
these  two sources of uncertainty and the chance that
they may  reinforce one another,  certainly  if  the
estimated  concentration of a chemical at the fraction
LC50  is within 1,OOOX  of the literature  LC50,  the
chemical should remain suspect.  To the extent  that
data for either compound quantitation or  compound
LC50  values are  known to  be better, concentration
differences of smaller  magnitude  may be  used  to
eliminate suspects.

Once  a  list  of  suspects is  available,  the
measurements for  both concentration  and toxicity
should be refined. This will  usually require obtaining
or making  pure compound to  make better analytical
measurements and  to establish LC50 values for the
species of concern. This is one  of the  steps  that
pushes  us to  do as much  separation as  practical
before analysis so that the list of suspects is small.

At this stage, only the concentration of the suspected
toxicant(s) in the concentrate is known; until  recovery
through all the fractionation  and concentration steps
is completed,  suspect  compound  concentrations  in
whole  effluent  are  poorly  known.  Since  the
concentrate is virtually  devoid of  suspended  solids
and much  of the effluent TOC, both of which may
dramatically affect toxicity of non-polar organics, the
toxicity of non-polar  chemicals  may  be  quite
different in the fraction tests than in  the effluent test.
                                                2-9

-------
Therefore, the toxicity of suspects in the fraction test
should be  compared  to the suspect's  toxicity in a
relative  pure water, such  as reconstituted water.  In
general, one would expect the toxicity to be lower in
the effluent.

During this same stage, the  steps leading to the final
concentrate should be better checked  for  toxicity
recovery. The objective is to place a good  estimate
on  how much  of  the whole effluent toxicity  is
contained  in the final  concentrate.  This  is best done
by  testing  the toxicity  of  the   concentrate   at
concentrations  near those of  whole  effluent,
correcting  for volume  losses due  to toxicity testing
SPE column fractions, which was previously ignored.
If the toxicity of the final concentrate is similar to that
of whole  effluent,  allowing  for  losses,  and  if the
concentration  of  the suspect(s)  in  the  test
concentration  which  produces  an  LC50 is similar
enough to the LC50 to account for the concentrate's
toxicity, it  is  time to begin Phase  III. If  multiple
toxicants occur, the toxic units of those are compared •
to the whole effluent toxic units.

If the concentrations from improved  quantitation and
toxicity measurement  are close,  Phase  III should be
started, recognizing that other toxicants may yet be;
identified.  If no  suspects are found, then  different
analytical  methods  are  called   for,  and  more
concentration, more  separation, and more  sophis-
ticated analyses must  be done. When the concentrate
is toxic,  the problem is  not  biological  but is  an
analytical one. Not  much  more  can  be  done with
toxicity testing at this  point.  In some of  the  effluents
we have tested, finding other candidates has taken up
to four months. Since few laboratories will have all the
needed analytical  equipment, contractors  may  be
used or help from  laboratories that  have the needed
equipment may be sought.
Because artifact toxicity  can be created that equals
toxicity  due  to  lost or  unidentified  toxicants,  the
suspected toxicant(s) may be found to be the wrong
ones as Phase III progresses. One purpose of Phase
III  is to identify  such errors. Should this occur, one
must go back to the beginning of Phase II, or even to
Phase  I  and  start over. If a  number of  different
effluent samples were used during Phase  II  work,
redoing Phase  I on  additional  samples  may  be time
well spent since the effluent  may have changed in the
interim.

While the distinction between  Phase II and III may
seem well defined,  as discussed here, in  practice
there is no sharp boundary.  In general, as soon as a
probable suspect  is identified,  confirmation  proce-
dures of Phase  III should start. The assumption that
one of the toxicants has been  identified when  it has
not,  can  badly  mislead  identification  of  other
suspected toxicants.

A  final suggestion  is to look at additive toxicity  of
several to many constituents,  if one or a few  seem
not to account for toxicity. Enhancement of toxicity  by
methanol should also be checked.
                                                2-1 0

-------
                                             Section 3
                                             Ammonia
3.1  General Overview
Unlike Phase II procedures  for  non-polar  organic
compounds  or  metals,  the toxicant  identification
methods described in this section are specific for
ammonia. The  procedures used in this step of the
study  assume  that Phase I  tests  have implicated
ammonia as one of the toxicant(s) (Phase I, Section
8).

These tests should  separate toxicity caused by
ammonia from  toxicity caused by  other compounds
which  also  become more toxic as  pH increases. Two
methods can  be used to add further  evidence to
implicate ammonia as  the toxicant. Equitoxic solutions
of effluent can be created through  pH adjustment and
dilution of effluent aliquots, and a zeolite resin can be
used to strip the effluent sample of ammonia. Toxicity
tests and ammonia measurements are  subsequently
performed  on pre- and post-column effluent.

Depending  on the presence of other toxicants in the
effluent,  additional  sample  manipulations  may be
needed before using these tests. For example, if toxic
oxidants  such as  chlorine are also present in  the
effluent,  sodium thiosulfate must be added to  the
sample before  conducting the Phase  II  ammonia
tests. If other toxicants can be removed  by C-| Q SPE,
then these tests can  be done  on  post-Cia   SPE
effluent.

3.2  Equitoxic Solution Test

Principles/General Discussion:

As described  in Phase I, ammonia toxicity is relatively
unique in its behavior as pH changes. For a constant .
temperature situation,  Table 3-1  shows that as pH
increases by one unit, there is a nearly  10X increase
in the percent of total ammonia present as NHs  (toxic
form)  in aqueous solutions at pH  6.0-8.5. Table  3-1
has been constructed from the dissociation values for
ammonia (Thurston, et al., 1979). Once the effect of
pH  and  temperature   on  the percentage  of  total
ammonia as NHs are accounted for, as  in Table 3-1,
there is a separate effect  of pH  on NHs toxicity as
shown  in Table  3-2, wherein  NHs  toxicity  increases
(decreasing ;LC50) with decreasing  pH.  By combining
 these  two pieces of information, radically  different
 concentrations of effluent can be  made to be equally
 toxic by pH adjustment and dilution. Reading the text
 in the  ammonia criteria document (EPA,  1985A) and
 understanding the effect of pH on  the acute toxicity of
 ammonia  will  aid  in understanding this test.  As
 discussed in the document, the slope of the  LC50-
 pH curve is  similar enough for tested species, i.e.,
 that an average slope  is  adequate for  any of the
 tested  species.  The formula for that curve is given at
 the  bottom of  Table 3-2. You  can see  from Table
 3-2  that NHa is about three  times  less  toxic  at  pH
 8.0 than at pH 7.0. On the other hand, about 10 times
 more ammonia  in the toxic form is present at pH 8.0
 than at pH 7.0  (Table 3-1). Therefore, the toxicity of
 a given amount of total ammonia is about three times
 more toxic at pH 8.0 than at  pH  7.0.  By making the
 proper  dilutions of effluent,  accompanied by the
 proper  pH  adjustments,  different effluent
 concentrations will be equally toxic if ammonia is the
 principal cause  of toxicity.  The  probability that any
 other chemical would have both the same  dissociation
 curve and the same change in toxicity  with pH is very
 small.  If other  toxicants are involved, the expected
 toxicity  will not  occur unless  they  are  removed or
 neutralized prior to this test-

 In using the  equitoxic approach,  it is necessary to
 prepare dilutions/concentrations  based  on  the
 conflicting effects of pH on ammonia toxicity.  Thus,
 when pH  is  increased,  a  dilution  factor must  be
 calculated that  takes  into  account both  increased
 amounts of  un-ionized  ammonia  present in  the
 solution, and the  decreased toxicity of the  un-
 ionized  ammonia to the  organism due to  the higher
 pH. Conversely, when pH is lowered, a concentration
 factor must be calculated that takes into account both
 decreased  amounts of un-ionized ammonia  in  the
 sample, and increased  toxicity   of  un-ionized
ammonia due to the lower pH. A descriptive example
 of this calculation follows.

 In the following  example, the effluent has  an LC50 of
50% and its pH  at the LC50 is 7.7.  Using the above
general  principle, the following calculations for raising
the pH to 8.0 are:
                                                3- 1

-------
Table 3-1. Percent Un-ionized Ammonia in Aqueous Ammonia Solutions for Selected Temperatures and pH
Values^
Temperature ("C)

















PH
6.5
6.6
6.7
6.8
6.9
7.0
7.1
7.2
7.3
7.4
7.5
7.6
7.7
7.8
7.9
8.0
15
20
0.0865 0.125
0.109
0.137
0.172
0.217
0.273
0.343
0.432
0.543
0.683
0.858
1.08
1.35
1.70
2.13
2.66
0.158
0.199
0.250
0.314
0.396
0.497
0.625
0.786
0.988
1.24
1.56
1.95
2.44
3.06
3.82
21 i
0.135 I
0.170
0.214
0.269
0.338
0.425
0.535
0.672
0.845
1.06
1.33
1.67
2.10
2.62
3.28
4.10
22
0.145
0.182
0.230
0.289
0.363
0.457
0.575
0.733
0.908
1.14
1.43
1.8
2.25
2.82
3.52
4.39
23
0.156
0.196
0.247
0.310
0.390
0.491
0.617
0.776
0.975
1.22
1.54
1.93
2.41
3.02
3.77
4.70
24
0.167
0.210
0.265
0.333
0.419
0.527
0.663
0.833
1.05
1.31
1.65
2.07
2.59
3.24
4.04
5.03
25
0.180
0.226
0.284
0.358
0.450
0.566
0.711
0.893
1.12
1.41
1.77
2.21
2.77
3.46
4.32
5.38
aThurston, et al., 1979.
Table 3-


pH
8.0
7.9
7.8
7.7
7.6
7.5
7.4
7.3
7.2
7.1
7.0
2.














Calculated Un-ionized Ammonia LC50 Values
at Different pH
an LC50 at pH
LC50 at NH3a
(mg/L)
3.00
2.84
2.68
2.50
2.30
2.10
1.88
1.67
1.45
1.25
1.07
Values for Ceriodaphnia Using
8.0 and 25° C Of 3.0 mg/L

pH
6.9
6.8
6.7
6.6
6.5
6.4
6.3
6.2
6.1
6.0

LC50 at NH3a
(mg/L)
0.90
0.75
0.62
0.51
0.42
0.38
0.30
0.24
0.19
0.15

(1) Increase
in Nhta dissociation is from 2.77% to
5.38% (Table 3-1).
Therefore,


f «i n




O.dO
	 =1.94 = dilution factor.

(2) Decrease
toxicity:

LC50
LC50
2.77
the dilution


at pH 7.7 =
at pH 8.0 =
The toxicity decrease factor




3.00

2.50

factor by


2.50 (Table
3.00 (Table
is:

= 1.20


the reduced


3-2)
3-2)




The LC50 value at pH 8.0 that is used is not important since
the ratios, not the absolute concentration, are used in the
calculations for dilution. The following formula from EPA
(1985A) was used to calculate values:
Therefore,

    1.94(1)
            = 1.6 dilution factor corrected for toxicity
        Formula 1/350=
                     (LC50[pH=8.0]) (1.25)
                         1 + 10
                              7.4-pH
    1.2(2)

(3) Effluent concentration needed at pH 8.0 is:

                 50%
                 - =3 1.25% effluent
                 1.6

For lowering the pH to 7.5:
                                                      (4) Decreased in NHs  dissociation  is  from 2.77 to
                                                          1.77  (Table  3-1).
                                                 3-2

-------
 Therefore, concentration factor

                     2.77
                          = 1.57
                      1.77

 (5) Reduce the concentration factor due to increased
    toxicity caused by reduced pH:

       LC50 at pH 7.7 = 2.5  (Table  3-2)
       LC50 at pH 7.5 = 2.1  (Table  3-2)
Toxicity correction factor
                      2.5
                      2.1
Therefore,
          concentration factor     1.57(4)
        toxicity correction factor   1.19 (5)
                                         = 1.3
(6) Effluent concentration at pH 7.5 is 50% x 1.3 =
    65% effluent.

Consideration must be given to the CO2 that will be
released when the pH of water is lowered one to two
pH units.  Of course,  the more  carbonate  and
bicarbonate alkalinity  present, the more free CO2 will
be released.  We have tested  both young  fathead
minnows and  Ceriodaphnia and they are not bothered
by 600  mg/L of  CC>2, which  is  the approximate
concentration occurring when 300 mg/L (as  CaCOs)
hardness water is lowered from a pH of 8.5 to 6.0. If
effluent/water mixtures are used with more carbonate
and bicarbonate alkalinity, the effect of the released
CO2 must  be checked  with the test species. High
CO2 concentrations may also change the  toxicity of
ammonia, so its  effect should also  be  measured
where appropriate.

Since any dilution can be made, the selection of  the
high  pH  condition   can  be  any  pH   which is
physiologically acceptable. However, the change in
ammonia toxicity  above pH  8.0  is  smaller and  the
data are less  precise. Therefore,  pH 8.0 may be  the
upper useful pH limit.

For the lower pH, some trial and error is needed.
Obviously 100% effluent is the highest concentration
obtainable,  so the necessary  concentration factor
must not exceed  the value 100%/LC50(%).  In the
above example when  the LC50 is 50%, that value is
two. It  is obvious  that, at most, we can only double
the LC50 concentration to  100%. In the  above
example, since only  65% effluent is  required at  pH
7.5, a lower pH would be better so that the  difference
in  effluent concentrations  tested would be larger. If it
were larger,  dilution  effects  on toxicants  other than
ammonia would be larger and the effect  would  be
 easier to measure. For example,  if we used a pH of
 7.2, 90% effluent would be needed.  ,

 The key problem in doing this test is to measure and
 control the pH. A 0.1  pH unit change from nominal is
 too  much variation.  To illustrate, the change in pH
 from 8.0  to 7.9 lowers the concentration  of the toxic
 form of  ammonia 20%.  Experience has  shown  that
 just any  pH  meter is not  good enough. The meter
 used must read accurately to two decimal  places.  If
 the  meter used does  not lock on  the stabilized
 reading after the rate of change has diminished to  a
 specified  rate, then pH readings must  be  made using
 a measured  elapsed time.  Readings  should be
 repeated  using constant and  reproducible  stirring
 rates that do not result in excessive loss  of CO2 and
 therefore  change pH.

 Secondly, the  ammonia concentration must be high
 enough so the  test  can be done in a work day to
 monitor the pH carefully and  frequently.  The pH of
 pH-adjusted effluents cannot  be controlled  accep-
 tably for 48  hours. As  a guide,  the ammonia
 concentration  needs to  be 2X  the 48 hour LC50 to
 produce mortality rapidly enough to complete the test
 in eight  hours. For many effluents,  ammonia  may
 need to be added to raise the concentration to  this
 range. Of course there is an upper limit  to  the
 percentage the ammonia  can  be  increased,   and
 useful data obtained.  If one  had to increase  the
 concentration  8-1 OX then the error of the method is
 too great  to use it. An increase of 2 or 3 times is a
 reasonable upper limit for spiking. A series of known
 ammonia/pH  combinations in  laboratory  water,
 duplicating the effluent  exposure should be  run in
 parallel for comparison and  as a  check  on  the
 method.

 If all air is excluded from the test chamber, the pH will
 frequently hold very  well. Since time  to  mortality is
 used,  frequent observations  are necessary and,
 therefore,  the cover used needs to allow observations
 of the test animals. If pH needs to be adjusted during
 the exposure,  then the  cage method  must  be used
 (Phase I). Proper stirring and measurement cannot be
 done if the test animals  are present.  Once the test
 organisms die, the pH can  change appreciably and
 pH measurements should  be  made  at the  time of
 mortalities. As this document goes to  press, we  are
 experimenting  with  the  use  of controlled
 concentrations  of CO2 in  a sealed  headspace over
 the sample. The approach  appears  very promising
 and simple.

 3.3  Results/Subsequent Tests

 No LC50  is calculated for the pH-adjusted  effluent
 samples used  in this  test.  Instead, a  comparison is
 made  based  on survival  and  time  to onset of
 symptoms in the three exposures. If ammonia is  not
the sole toxicant in the effluent,  the expected change
                                                3-3

-------
in toxicities will not be found. Attempts can be  made
to  neutralize  or remove  other toxicants  before
performing  this test, but  few if  any methods are
available in  which there is  certainty that additional
constituents would  not  be removed.  If,  after  such
removal, this test shows ammonia to be the toxicant,
then ammonia is a principal part of the toxicity.

3.4  Zeolite Test

Principles/General  Discussion:

Zeolites are  naturally  occurring or synthetically
created crystalline,  hydrated alkali-aluminum sili-
cates.  They  have  a general  formula  of Mn
0*Al2Oa.ySi02.zH2O; M  = group IA or IIA element, h
=  +2 for group IA, +1  for group IIA, y>2, and z  =
the number of water molecules contained in the
interconnected  voids or channels within  the zeolite
(Windholz,  et  al., 1983). When  zeolite is placed  in
aqueous solutions, the positively charged group IA or
IIA elements (M" +) of the  zeolite are mobile and can
undergo exchange with other cations in the water. As
such, zeolites  have frequently been employed as ion
exchange  resins  for removal of the ammonium  ion
NH4+  from aqueous solutions.  Because  of their
ability  to exchange other cations  such as  heavy
metals, and their use  as molecular sieves,  filter
adsorbents and catalysts,  zeolites  have not  been
suggested  for use  in  Phase I.  Zeolites can be
effective in  Phase  II, however,  if Phase I results
implicate  ammonia  as  the  causative  agent and
establish that other groups of  toxicants  (such as
organics and metals) are  playing no role in effluent
toxicity.

3.5  Procedure

In this section of Phase  II,  toxicity tests and ammonia
measurements are  made on whole  effluent and
effluent having passed  through  a  zeolite  column.
Removal of toxicity by zeolite will add to the evidence
implicating ammonia as the toxicant.

To  prepare the zeolite  column,  30  g of zeolite is
added  to  60  mL of  high-purity  water.  Zeolite
particles should be of approximately the same size
(32 to 95 mm) to insure efficient ion exchange while
preventing  channeling  (e.g.,  effluent not contacting
the  zeolite during flow  through the column)  or
excessive resistance to flow.  Removal of extremely
large or small particles can  be accomplished by
screening  the zeolite  with sieves or mesh  screens.
The zeolite slurry is poured into the column and three
bed volumes of dilution water are passed through the
column. The last 10 mL of dilution water is  collected
for use as a zeolite blank and should not be toxic.

Next, 200 mL  of 100% effluent is passed through the
column at a rate of 2 mL/min. After the dilution water
remaining  in the column  has been  flushed out by
effluent,  post-column  effluent  is  collected and
measured for ammonia and toxicity. Temperature and
pH should be recorded in order to provide the means
to calculate both total  and un-ionized  ammonia in
the sample.

Another aliquot of the  effluent sample (not having
passed  through the  zeolite  column)  is used for
ammonia analysis and a toxicity test. These  data will
be compared with the same data  for the post-zeolite
column  effluent to determine if the  post-column
reduction in effluent  toxicity is consistent with
ammonia removal through the column.

3.6  Interpretation of Results

The  control  for test organism  survival, dilution water
quality,  and other test  conditions, will  be provided
through the  preparation  of a single exposure toxicity
test  on  dilution water.  Dilution water  having  passed
through the column will act  as  a blank for  toxic
artifacts leached from the zeolite. Increased toxicity in
the  post-zeolite  effluent,  relative  to  the  whole
effluent,  indicates  the presence  of  toxic artifacts.
Since many cations will be exchanged, adding solids,
such as the YCT  food  (yeast-Cerophyl-trout food)
fed to Ceriodaphnia, is a wise precaution. Additional
measures to clean-up the  zeolite (such as  Soxhlet
extraction) or alternate uncontaminated sources may
need to be investigated. Column packing, effluent pH,
ammonia levels, and  flow  rate through the  column
can  all affect the efficiency of the  cation exchange
process.  Lowering  effluent  pH prior to   zeolite
chromatography and/or lowering flow rate through the
column  may  also  result  in  greater  removal  of
ammonia. Occluded gas  between zeolite  particles
may also impair  the  column's capacity to  remove
ammonia. If this appears to be a problem, the zeolite
solution should  be degassed using a vacuum prior to
pouring the column.

Zeolite can  be  regenerated, but fresh zeolite should
be  used to  pack  new  columns.  If the equitoxic
ammonia test and  zeolite  test are  consistent with
ammonia toxicity, confirmation should be  started  as
described in Phase III.
                                                3-4

-------
                                             Section 4
                                         Cationic Metals
4.1  General Overview
The reduction in effluent toxicity afforded by addition
of EDTA indicates that certain cationic metals may be
present in the effluent at  lethal levels. The  suspect
causative agent(s)  is chosen based on  correlation  of
effluent toxicity and metal concentrations, reference
metal toxicity data, and changes  in toxicity observed
during manipulation of water quality characteristics.

Principles/General Discussion:

The structure of EDTA (ethylenediamine tetraacetic
acid) is shown below:
 HOOC'
  HOOC
                                         COOH
                                         COOH
This compound can be thought of as a Lewis base,
having the capability of donating electrons (through
the nitrogen and  hydroxyl oxygen  atoms) to Lewis
acids (such  as  metals)  in  order  to satisfy  the
coordinate  requirements.  EDTA is a  hexadentate
ligand;  there  are six  positions on the  molecule
available to donate electrons. The  metals that are
chelated by EDTA are  a function of the  number of
electrons needed  by the cation to fill its outer orbital.
Generally, EDTA forms a 1:1  complex with  cations of
+ 2 and + 3 valences.

Simply stated,  one could predict the  degree to which
certain metals  are chelated by  EDTA  by  comparing
the stability constants (f3 = [EDTA x metal]/[metal] x
[EDTA]) for the  metal-EDTA complexes.  Such  a
comparison suggests which  metal   would  win  in  a
competition for the electrons provided  by the EDTA
ligand. One finds  that EDTA  strongly chelates Fe3 + ,
Cu2*, Ni2 + , Pb2 + , Cd2 + , C02 + , Zn2 + ,  AI3 + ,  and
Mn2 + , while Ca2 + , Mg2*,  Sr2 + ,  Ba2 + , Tl+  and
Ag2+  are only  weakly chelated by EDTA because of
their  relatively low electrophilicity  (Flaschka  and
Barnard, 1967).
 Unfortunately, the complexing effect of EDTA towards
 metals  in  a complex  aqueous  solution cannot  be
 estimated  solely on the basis of  stability  constants
 (Stumm and Morgan, 1981).  Complex  solutions
 contain other organic  and  inorganic ligands  which
 compete to different extents for  the various available
 cations. This  becomes  obvious when  studying
 apparently  supersaturated  aqueous  solutions  of
 metals  in  which  otherwise  insoluble  metals  are
 solubilized  by their association  with  ligands.

 The concentrations of other ligands, relative to that of
 EDTA, as well as their  specificity for particular metals
 will  greatly  change the  results  from  the  above
 prediction.  For example,  even  when  the stability
 constant for EDTA and a metal is much greater than
 the  stability constant for  the  metal  and a different
 ligand, if the concentration of the competing ligand is
 much  greater  than the  EDTA concentration,  no
 substantial  complexation of  the  metal by EDTA will
 take  place.  The  aquatic  organism  itself can  be
 thought of as a powerful  and very specific organic
 ligand, often acting as  a sink  for metals in aqueous
 systems. While the  oxygen donor atoms  of  EDTA
 form  general complexes,  the  ligands provided  by
 organisms  are  very selective as  a result  of  the
 geometric  arrangements formed.  This  situation  is
 exemplified  by the case of Hg2 + ; the EDTA Hg2 +
 stability constant is quite  large  (21.8) compared  to
 that of other metals chelated by  EDTA.  Despite this,
 addition of EDTA  does not  reduce mercury toxicity
 because the aquatic organisms' ability to take up the
 metal is greater.

 Solution pH, temperature and ionic strength can also
 affect the form in which a metal exists in an aqueous
 solution. At low pHs, H+ competes with the metal for
 EDTA binding sites; at  high pHs, OH" competes with
 EDTA and other ligands for a coordinative position  on
the  metal.  Time   may  be  an  important  factor  in
 determining what  EDTA chelates and the  extent  to
which chelation takes place. Because equilibria may
 be  established relatively slowly,  unstable complexes
 may form and dissociate, changing the composition of
the solution over time. Addition of EDTA can result in
a  redistribution of  many  or  all of  the metals 'in
 solution.  This can  result when EDTA binds strongly
with a certain metal, in turn freeing up other  metals in
                                                4- 1

-------
the system. It should be pointed out that unlike metal
complexes  with naturally occurring  ligands  (e.g.,
humic and  fulvic acids),  EDTA forms a very stable
complex. Finally, the degree of EDTA complexation is
also affected by dilution. Increasing dilution results in
fewer polynuclear complexes and a greater degree of
hydrolysis. This is most obvious with the less stable
EDTA complexes of unidentate metals such at Tl+.

Given the need for information on the solution's major
and minor  ions, its  oxidation/reduction  state,  the
acid/base and complexing components, and finally the
adsorbing surfaces, it becomes  clear that even  the
multiligand,  multimetal models  currently available will
not succeed  in predicting  the  cations  and  cor-
responding  quantities bound  to the  added  EDTA.
Likewise, there exists no reliable method to detect or
isolate specific EDTA chelates, particularly when the
concentration of the  metal of  concern may  be very
low. For this reason, utilization of general methods for
the detection of cations is desirable.
                                                i.
For the most part, there are two methods available for
analysis of  cationic  elements  and  inorganic
compounds. In AA analysis  a  light beam  at  the
wavelength of  the metal of interest is passed through
an atomized sample. The light energy absorbed  by
the atomized  sample provides the  measure of  the
concentration  of the metal in the sample. A flame
(direct aspiration  procedure)  or  graphite  furnace
(furnace procedure) is used to atomize the sample. In
inductively-coupled  plasma  atomic  emission
spectroscopy (ICP), nebulized sample is directed  into
a plasma torch. The resulting  atomic  line  emission
spectra  (characteristic of the metals in the  sample)
are produced via  radio  fre-quency inductively
coupled  plasma.  The  advantage to this  latter
technique is that it can  be  used to  simultaneously
determine multielements  at trace levels in aqueous
solutions,  but sensitivity  and interferences  are
sometimes  problems.  AA analyses,  on the  other
hand, require  that the analyst choose, in advance,
those elements to  be measured. The furnace  AA
technique does offer the advantage  of producing
detection levels which  are  typically an order of
magnitude lower than the levels resulting for ICP. The
need  for the  lower levels of  detection provided  by
furnace  AA will be a function  of the toxicity  and the
concentrations of metals present in the sample.

In either  case,  a measure of the dissolved metals in
the sample  (i.e.,  those  constituents which pass
through  a  0.45 iim membrane filter)  is  recom-
mended. The  "dissolved" metals  are in  no way
synonymous with  the "biologically available  metals."
Beyond the use of the aquatic organism, however,
there  is  no technique to determine the "biologically
available" fraction  of the total metal. Further, only
rudimentary techniques are  available to  specifically
identify  the individual species of a  metal  (e.g., free
charged  metal ions  [Mn + ],  inorganic ion  pairs or
complexes such  as  aquoions,  [M(H2O)n + m],  hy-
droxoions  [M(OH)pn-P + ],  oxoions  [MOnn'  2q + ],
organic  complexes and  chelates [M  x  EDTA], metal
species bound to  high molecular  weight organic
material [M  x lipid] or metal species in the form of
highly dispersed colloids or  sorbed on colloids [M x
clay]).  Stumm and  Morgan  (1981) have listed some
general  methods for assisting  in  identification of
individual species.

Given the concentration of  metals present in toxic
effluent samples, a number  of techniques, described
below, are available to implicate the toxicant(s).

4.2  Procedure

In order to compare the results of the metal analyses
and toxicity  test, sample preservation  with HNOa  is
not recommended. Instead,  immediate analysis  is
recommended.  An  effluent  LC50  should  be
determined for the test  species using whole effluent.
The dilution  water used  should be reconstituted water
of the same hardness as that of the effluent.  A 250
ml_ aliquot of the sample is filtered through a 0.45 pm
filter. The filter should  be prepared by rinsing with 200
mL of high purity water.  An appropriate quantity of the
last portion  of the  high  purity water  passing through
the filter should be collected  as an analytical blank for
metal contamination from the filter. A single exposure
toxicity  test  on 10  mL of the dilution water  having
passed through the filter should also be performed.

The filtered  sample should  be tested for  toxicity as
described  above to  measure the effect of filtration on
sample  toxicity.  The  filtrate  should be analyzed for
metals. If toxicity  is  reduced  or  removed upon
filtration (and effluent toxicity has not previously been
affected by  C^Q SPE or filtration through a  1.0 pm
glass  fiber  filter) it will be   necessary to measure
metals retained  by the 0.45 urn filter. Methods for
determination of suspended  metals are described  in
Section 200 or Method  200.7  (EPA,  1983). If the AA
analysis is used for dissolved or suspended  metals,
iron, copper, nickel, lead, cadmium,  chromium,
cobalt,  zinc, aluminum,  and manganese  should be
included in the analyses.

Initial  determination of  suspect  metals based on  a
comparison  of metal analyses and toxicity test  results
should  be  made.  This can be  followed for  the
dissolved  and/or  suspended metals, as  necessary,
based on toxicity test  results. Because  metals can be
readily purchased,  tests on  pure  metals  should be
done  rather than to use published values.  Side-by-
side tests  with  reference metal  and  effluent avoid
subtle error  and the improved data are  well worth the
effort. Literature summaries  are  also available (EPA,
1980; 1985B; 1985D; 1985E; 1985F;  1986;  1987A;
and 1987B). In addition to matching the hardness and
pH  of  the dilution  water to  the effluent  sample by
addition of  the  appropriate  ratios of  MgCC-3  and
                                                 4-2

-------
CaCOs, the wastewater total suspended solids (TSS)
and TOC can also be approximated in the water used
to test the reference metal. TOC and  SS  from  the
YCT food can be added such that the  TSS level of
the dilution water equals that of the effluent sample.
The TOC  of the  dilution  water can  be further
supplemented through the addition of  humic  acid.
Some indication  of the binding of  metals to organics
in the effluent may  be  arrived at  through hexane
extraction of an  aliquot of the sample (Stary, 1964).
Theoretically,  metals  bound to  organics soluble in
hexane should leave the effluent. The hexane can be
evaporated and the residue  reconstituted and tested
for  metals. Additionally,  the  reduction in metals  can
be estimated by repeating the metal analysis on  the
extracted  effluent. The  toxicity  attributed to metals
associated with  organics may  be estimated  by
performing  a  toxicity test on  the solvent  extracted
effluent.  Traces   of  hexane must be  removed  by
aeration from the extracted effluent by aeration prior
to toxicity testing. The effects  of aeration on sample
toxicity must also be considered in  this analysis. In
any case, metals strongly suspected of  causing or
contributing  to sample toxicity should be tested in
pure solution as  described above with  the  TIE  test
organism species.
The correlation between the toxicity of a sample and
its concentration(s) of metal(s)  over time (Phase III)
can  also  be  used  to  narrow  the  list  of  suspect
causative  toxicants prior to confirmation. The effects
of water quality characteristics on metal  toxicity must
be included over the sampling period. One solution to
this problem is to collect several samples over a short
time span  (e.g., six grab samples in 24 hours). The
correlation coefficient for sample metal concentration
(or summed  toxic units  of metals) versus  sample
toxicity is  calculated for each sampling event (i.e., 24
hour period). The population of correlation coefficients
for  such multiple  sampling events may give results
less affected  by  hardness, suspended solids  and
TOC. The  assumption in  taking  this approach is that
water quality  characteristics affecting metal  toxicity
will vary less  during short time periods.  Obviously,
metal concentration must vary  enough  to provide a
sufficient  range  for correlation.  When  one  reaches
this stage,  Phase III work is being started and Phase
III methods should be consulted. Symptoms, species
sensitivity,  spiking,  water quality  adjustments  and
correlation  are all  applicable Phase III approaches to
confirm the cause of toxicity.

Where there  is  also non-metal  toxicity, this  toxicity
identification  and  then  its confirmation must  be
integrated with the metal work data.
                                                 4-3

-------

-------
                                           Section 5
                                          References
Alford-Stevens, A.L., J.W.  Eichelberger,  and W.L.
  Budde.  1985. Method  680.  Determination of
  Pesticides and PCBs in Water and Soil/Sediment
  by  Gas  Chromatography/Mass  Spectrometry.
  Physical  and Chemical  Methods  Branch, U.S.
  Environmental Protection Agency, Cincinnati, OH
  45268.

EPA. 1980.  Ambient Water Quality Criteria for Silver.
  EPA-440/5-80-071.  Environmental  Protection
  Agency,  Environmental  Research  Laboratory-
  Duluth, Duluth, MN.

EPA. 1983.  Methods for Chemical Analysis of Water
  and  Wastes. EPA-600/4-79-020. Environmental
  Monitoring and Support Laboratory, Cincinnati, OH.

EPA.  1985A. Ambient Water Quality  Criteria  for
  Ammonia.  EPA-440/5-85-001.  Environmental
  Protection Agency, Environmental  Research
  Laboratory-Duluth,  Duluth, MN.

EPA.  1985B. Ambient Water Quality  Criteria  for
  Cadmium.  EPA-440/5-84-032.  Environmental
  Protection Agency, Environmental  Research
  Laboratory-Duluth,  Duluth, MN.

EPA.  1985C. Ambient Water Quality  Criteria  for
  Chlorine., EPA-440/5-84-030.  Environmental
  Protection Agency, Environmental  Research
  Laboratory-Duluth,  Duluth, MN.

EPA.  1985D. Ambient' Water Quality  Criteria  for
  Chromium.  EPA-440/5-84-029.  Environmental
  Protection Agency, Environmental  Research
  Laboratory-Duluth,  Duluth, MN.

EPA.  1985E. Ambient Water Quality  Criteria  for
  Copper.   EPA-440/5-84-031.  Environmental
  Protection  Agency, Environmental  Research
  Laboratory-Duluth,  Duluth, MN.
EPA. 1985F. Ambient Water Quality Criteria for Lead.
  EPA-440/5-84-027.  Environmental  Protection
  Agency,  Environmental Research  Laboratory-
  Duluth, Duluth, MN.

EPA. 1986.  Ambient Water Quality Criteria for Nickel.
  EPA-440/5-86-004.  Environmental  Protection
  Agency,  Environmental Research  Laboratory-
  Duluth, Duluth, MN.

EPA. 1987A. Ambient Water Quality Criteria for Zinc.
  EPA-440/5-87-003.  Environmental  Protection
  Agency,  Environmental Research  Laboratory-
  Duluth, Duluth, MN.

EPA. Draft 1987B. Ambient Water Quality Criteria for
  Aluminum. U.S. Environmental Protection Agency,
  Environmental Research Laboratory, Duluth, MN.

Flaschka, H.A. and A.J. Barnard, Jr.  (Eds.).  1967.
  Chelates  in Analytical  Chemistry.  Marcel  Dekker,
  Inc.,  New York, NY. 418 p.

Mount,   D.I.  1988. Methods  for Aquatic Toxicity
  Identification  Evaluations:  Phase  III  Toxicity
  Confirmation Procedures. EPA/600/3-88/036.

Mount,  D.I. and L. Anderson-Carnahan.  1988.
  Methods  for  Aquatic Toxicity  Identification
  Evaluations: Phase I Toxicity Characterization
  Procedures.  EPA/600/3-88/034.

Peltier,  W. and C.I. Weber (Eds.).  1985. Methods for
  Measuring  the  Acute Toxicity of  Effluents to
  Freshwater and Marine Organisms. Third Edition.
  U.S.  Environmental Protection Agency. Cincinnati,
  OH.   EPA-600/4-85-013.
                                             5- 1

-------
*

-------

-------
s-g
   en
75'
= • CD
cz
en
CD
CO
o
o
1
§
o
o
00
CO
o
to
' en













C O. =5
ill
Co§-
•p -i 3
3P 0 0
11*
n Z, »
O 3" IT
O - ~»
i"°
CO o (D
r* < o
to g>
01 R>
3 5-

S 8

si
O -i
- en
to O
o.m
CD ^
(/) T
w ^
|S
If?
i-8 5!
D) 3"  S
|o

w 5"

~ S-
CD
C
TJ S"
"2 S-
® tp_
                  > me
                  "32.
                  D -i' m
   CD
0)  CO
ro
o
                                  1'
                  o 5- o

                  S £tl
                  11 *
                  0) •-» Q"
                  s. o X
                  O D U
                  NJ
                  CD
                  00
   O


   1
   CD
                        CD
                        (fl
                        CD
                        0)
                        -n
                        O
                        zr
           TJ   O)

           m   ^
           33   >DD



-------