Techniques and methods for the determination of haloacetic acids in
 potable waterf



 Edward T. Urbansky

 United States Environmental Protection Agency (EPA), Office of Research and Development,
 National Risk Management Research Laboratory, Water Supply and Water Resources
 Division, 26 West Martin Luther King Drive, Cincinnati OH 45268, USA.
 E-mail: urbansky.edward@epa.gov; Fax: +1 513 5697658; Tel: +1 513 569 7655

 Received 13th April 2000, Accepted 23rd May 2000
 Published on the Web 20th July 2000
                                      Tutorial
                                      Review
 Haloethanoic (haloacetic) acids (HAAs) are formed as disinfection byproducts (DBFs) during the chlorination
 of natural water to make it fit for consumption. Sundry analytical techniques have been applied in order to
 determine the concentrations of the HAAs in potable water supplies: gas chromatography (GC-MS, GC-ECD);
 capillary electrophoresis (CE); liquid chromatography (LC), including ion chromatography (1C); and
 electrospray ionization mass spectrometry (ESI-MS). Detection limits required to analyze potable  water samples
 can be regularly achieved only by GC-ECD and ESI-MS. Without improvements in preconcentration or
 detector sensitivity, CE and LC will not find application to potable water supplies. The predominant GC-ECD
 methods use either diazomethane or acidified methanol to esterify (methylate) the carboxylic acid  moiety. For
 HAAS analytes, regulated under the EPA's Stage 1 DBF Rule, diazomethane is satisfactory. For HAA9 data
 gathered under the Information Collection Rule, acidified methanol outperforms diazomethane, which suffers
 from photo-promoted side reactions, especially for the brominated trihaloacetic acids. Although ESI-MS can
 meet sensitivity and selectivity requirements, limited instrumentation availability means this technique will not
 be widely used for the time being. However, ESI-MS can provide valuable confirmatory information when
 coupled  with GC-ECD in a research setting.
 1 Introduction

 Chlorination is  the  most common method for  disinfecting
 drinking water in the United States. During chlorination, active
 (oxidizing) compounds (HOC1, NaOCl, C12) disable micro-
 organisms present in the raw water. However,  these com-
 pounds also react with organic matter, producing  a variety of
 chlorine-containing byproducts. Bromide ion rapidly reacts
 with HOC1 or C12 to give HOBr [eqns. (1) and (2)].
HOCl+Br-->BrCl+OH-  ^ HOBr+Cr
                         -OH"

     Cl2+Br-->BrCH-Cr  ^~
                                                     (1)
                                                     (2)
In this way, bromine-containing oxidants are produced when
                           Ed Urbansky attended Allegheny
                         College  in Meadville,  PA,  and
                         Purdue University in West Lafay-
                         ette, IN. He joined EPA's National
                         Risk   Management   Research
                         Laboratory in July 1997. Although
                         he is an inorganic chemist, much of
                         his work at EPA has dealt with the
                         analytical chemistry of disinfection
                         byproducts or perchlorate. He has
                         authored or coauthored more than
                         20 papers  and  edited a  book,
                         Perchlorate in the Environment.
 bromide-containing waters are chlorinated. Thus, brominated
 and chlorinated organic compounds can result from chlorina-
 tion. Raw (untreated) source water contains a mixture  of
 natural organic matter (NOM) from the decay and breakdown
 of vegetation and animal matter as well as anthropogenic
 organic matter, such as pesticides. Some of the organic matter
 is soluble, but some of it is particulate. Regardless of its form or
 origin,  exposed organic matter can react  with bromine-  or
 chlorine-based  oxidizing agents to  produce an  array  of
 disinfection byproducts (DBFs).'
   Two classes of compounds make up most of the identifiable
 post-disinfection halogenated organic  matter: the  trihalo-
 methanes,  which account for ~20%,  and the haloethanoic
 acids, which  account for ~13%.2 In the US, the  EPA will
 regulate the haloethanoic acids (haloacetic acids  or  HAAs)
 under  the Stage  1  Disinfection   Byproducts  (DBF)  Rule
 beginning  in December  2001 ? Regulations  regarding the
 HAAs have been nicely summarized  by Pontius and Diamond4
 and a more detailed discussion is beyond the scope of this work.
 The environmentally significant HAAs are given in Table 1.
   Haloacetic  acids  are  moderately  strong  and  therefore
 dissociated nearly completely (>99.9%) to the haloacetate
 ions under typical drinking water conditions (pH > 6). The pATas
 of mono-, di-, and trichloroacetic acids are 2.86, 1.3, and 0.66,
 respectively.5 Nonetheless, the regulations are written in terms of
 the parent acids rather than the haloacetate ions. Although the
 HAAs represent a significant fraction of the halogenated organic
 matter, they are polar, hydrophilic species that readily ionize.
 This must be accounted for  when deciding on an  analytical
 approach to their measurement in finished potable water.
——	—	    2 Extraction and preconcentration
fThis is the work of a United States government employee engaged in
his official duties. As such it is in the public domain and exempt from    Extraction of an analyte into an organic solvent affords certain
copyright. © US government.                                   .    -   — . -
DOI: 10.1039/b002977g
benefits. Chief among these is that the analyte is concentrated

                /. Environ. Monit., 2000, 2, 285-291    285
                                 This journal is © The Royal Society of Chemistry 2000

-------
Table 1 HAAs to be regulated or undergoing information collection in the US
No.
Al
A2
A3
A4
A5
A6
Haloacetic acid
Chloroacetic
Dichloroacetic
Trichloroacetic
Bromoacetic
Dibromoacetic
Tribromoacetic
Chemical formula
C1CH,CO2H
C12CHCO>H
C13CCO2H
BrCH,CO2H
Br2CHCO2H
Br3CCO2H
CAS registry no.
[79-1 1-8]
[79-43-6]
[76-03-9]
[79-08-3]
[631-64-1]
[75-96-7]
Regulatory status
HAAS MCL"
DBPR1
DBPR1
DBPR1
DBPR1
DBPR1
n/a

HAA6 ICR*
Required
Required
Required .
Required
Required
Encouraged'
A7          Bromochloroacetic            BrClCHCO2H             [5589-96-3]               n/a                   Required
AS          Bromodichloroacetic          BrCl2CCO2H             [71133-14-7]              n/a                   Encouraged'
A9          Dibromochloroacetic''         Br2ClCCO2H             [7278-95-5]               n/a                   Encouraged'
"HAA5 maximum contaminant level (MCL) is to be regulated as the sum of the concentrations of 5 HAAs (A1-A5), not to exceed 60 ug L~'
under the Stage 1 DBF Rule (DBPR1); n/a=DBPRl not applicable. The sum HAAS is expressed in ugL  , rounded to 2 significant digits.
Dichloroacetic acid has a maximum contaminant level goal (MCLG) of zero and trichloroacetic acid has an MCLG of 300 ug L~'. Chlorina-
tion plants that use 4 quarterly values to compute an average and have HAA5<30ugL~' and THM4<4ugL~' are permitted to opt out of
further treatment studies. If a public water system (PWS) has an annual average ^80% of the MCL (48 ug L~'), it is required to create a disin-
fection profile under the Interim Enhanced Surface Water Treatment Rule [IESWTR, 40 CFR 9, 141, 142, Fed. Regist., 1998, 63 (241), 69478.]
DBPR1 takes effect December 2001 for surface water PWSs (serving > 10 000 persons and December 2003 for all remaining PWSs. *HAA6 =
HAAS+bromochloroacetic acid (A1-A5+A7) as specified in the Information Collection Rule (ICR). PWSs serving > 100000 persons must
comply with all parts of the ICR. THAA9 includes all 9 species. PWSs are encouraged—but not required—to monitor for A6, AS, and A9 indi-
vidually under the ICR. These species may be considered for regulation at some point in the future, but are not part of any proposed or pro-
mulgated rule for US PWSs at this time. ''A note on  nomenclature. This compound is named correctly as dibromochloroacetic acid because the
order is alphabetical based on the halogen substituents not the prefix (di) telling the number; however, it is often encountered named incorrectly
as chlorodibromoacetic acid, especially in the drinking water/environmental chemistry literature.

by the extraction procedure. A second benefit is the elimination
of potential interferents that remain in the original aqueous
phase. In order to use liquid-liquid extraction for  the HAAs,
the haloacetate ions must be protonated. Given that the pjfa of
C13CCO2H  is  0.66, only  82%  of the  trichloroacetate  is
protonated in 1.0 M acid (pH = 0), In order for liquid-liquid
extraction to be feasible, either high acidification or a favorable
solvent is required. To further  promote partitioning into the
solvent,  the  analytes  are  often  salted out  using  high
concentrations  of  an inert  salt, such  as  sodium hydrogen
sulfate.
   In  addition  to  classical  liquid-liquid extraction, newer
solid-phase extraction cartridges, microfibers, or membranes
are also available. Because of the need for preconcentration, a
considerable amount of effort has been expended on various
extraction methods. Individual approaches to extraction will
be addressed in the selections below based on which analytical
technique was subsequently used.
       0       ©             (DO        9    ffl
        Scheme 1 Resonance structures for diazomethane.


and uses  diazomethane, which  exists  in the canonical forms
(resonance structures) shown in Scheme 1.
  Diazomethane,  CH2N2, is  a powerful  methylating agent,
able  to  esterify  oxyacids under mild  conditions without
catalysts and even with small  amounts of water present in
the solution. The mechanism is  shown in Scheme 2.
  In practice, diazomethane  is generated by  the  action of
hydroxide on JV-methyl-W-nitroso-p-toluenesulfonamide (Dia-
zald®, Aldrich,  Milwaukee, WI, USA) or l-methyl-3-nitro-l-
3  Gas chromatography

Gas  chromatography  is  the  most  common  instrumental
technique used for drinking water analysis. Gas chromatography
may be coupled with either mass-selective detection or electron-
capture detection. GC-ECD is more sensitive than GC-MS by a
factor of about 1000, and it forms the basis for EPA's Methods
552s and 552.27 because of  the low concentrations  at which
haloacetates are found in potable water supplies. Methods 552
and 552.2 are both based on liquid-liquid  extraction and GC-
ECD. Both use sulfuric acid to protonate the haloacetates in the
water sample and tert-butyl methyl ether (MTBE) to extract the
protonated analytes. The major difference between the two is the
esterification step. Fig.  1 shows a typical chromatogram of a
standard run according  to Method 552.2.
  Because of their polar nature and acidity, haloacetic acids
cannot be injected directly onto a GC column. Accordingly, the
carboxylic acid functionality is typically esterified. Methylation
has dominated the determination of HAAs; however, a wide
variety of silylating and other derivatizing agents are also
available for acidic compounds.8 Method 552 was developed
specifically for HAAS (see Table 1  for HAAS identification)
             10    15   20   25   30   35   40   4.5
                  Retention time/min

Fig. 1 GC-ECD chromatogram of a haloacetic acid standard prepared
in reagent water using  EPA Method  552.2 on a  J&W Scientific
(Folsom, CA, USA) DB-5.625 column.  Concentrations: chloroacetic
(CAA) and dichloroacetic (DCAA) acids: 6.0 ug L~'; trichloroacetic
(TCAA) and dibromoacetic (DBAA) acids: 2.0 ug L"1; bromoacetic
(BAA), bromochloroacetic (BCAA), bromodichloroacetic  (BDCAA),
and 2,2-dibromopropanoic (2,2-DCPA) acids: 4.0 ugL""1; dibromo-
chloroacetic acid (CDBAA): 1.0 ug L~';  tribromoacetic acid (TBAA):
20ugL"1.  Internal standard is 1,2,3-trichloropropane. Resolution of
the peaks  for dibromochloroacetic  acid and the  surrogate  (2,3-
dibromopropanoic acid) is better  than  it  appears  to  be  in the
chromatogram (0.2 min difference in retention time). On  the recom-
mended confirmation column (J&W Scientific DB-1701), the retention
time difference is nearly 1 min. (The figure is adapted from ref. 7.)
286     J. Environ.  Monti., 2000, 2, 285-291

-------
 RC-OH + e:CH2-N=N:  RC-OMe  <-»• ROOMe
         -H*
 O

RC-OMe
Schemes Mechanism for  methylation  of a carboxylic  acid with
acidified methanol.
MTBE as the acids, the haloacetates are trapped on an am'on
exchange  resin.  They  are then simultaneously eluted  and
esterified with acidified  methanol.17'18 EPA developed Method
552.1 based on this approach; however, on-column methyla-
tion is infrequently used today. Although Method 552.1 is still
approved for use, it has  fallen out of favor due to its more
complicated sample preparation in addition to lower precision
and accuracy.19'20  For HAAS, Standard  Method  6251  and
EPA Method  552.2 are used almost  exclusively by  the
American drinking  water community—in  large part because
they are EPA-approved methods for these DBFs.
  In  addition  to  acidified  methanol  and  diazomethane,
dimethylsulfate [(H3CO)2SO2] has been used to methylate the
HAAs in  natural  waters.21  In  that  case,  no extraction-
preconcentration step was used since the airspace was aspirated
for GC-MS; however, dimethylsulfate is quite toxic  and
requires a level of care in handling.  A few other derivatives
have also been prepared.  HAA difluoroanilides were analyzed
by GC-MS with single-ion monitoring. Detection limits were
impressive (0.5-2 ug L~:), except for  tribromoacetic  acid
(100 ug L"').22'23 When this approach was  applied to environ-
mental samples, detection limits for several analytes were of the
order of lOngL"1,  but  tribromoacetic acid  was not evalu-
ated.24 Solid-phase extraction has also been used,25'26 but it has
not been incorporated into any EPA-approved method.
4  Capillary electrophoresis

Similar acidification and liquid-liquid extraction can be used
for capillary  electrophoresis; however, CE normally uses an
aqueous  running electrolyte.  Consequently, two extractions
must be performed: the first, to preconcentrate; and the second,
to take the analytes back into an aqueous solution. Without
preconcentration, CE can  only reach  concentrations of ~ 1—
lOmgL"1.27  With  a liquid-liquid extraction  into  MTBE,
detection limits  in the  1-10 ugL"1 range are  reported.26'28
Martinez et al used MTBE-extraction to preconcentrate then
evaporated to  dryness and  redissolved  the residue in a
minimum of water.29  They reported detection limits near
lOOugL"' with 2,6-napthalenedicarboxylate as  the running
electrolyte. Online preconcentration techniques were successful
with standard solutions, but not real samples, demonstrating
the profound influence of the matrix.30 Thus, it appears that
CE will require either solid-phase or liquid-liquid extraction
prior to analysis. Solid-phase extraction has been investigated
using a variety of media. Cross-linked styrene-divinylbenzene
gave the best detection limits, rivaling that of GC-ECD for
HAAS (<5 fig L"1). The method has not been tested on all of
the HAA9 analytes, however.31 Because capillary electropher-
ographs are infrequently found in most drinking water testing
laboratories,  it is unlikely that this technique will make any
headway in the near future. Although CE is a mature analytical
technique with  many  advancements  having taken  place,32
methods for  haloacetates  are  not  yet adequately  rugged or
reliable to meet the demands of trace-level quantitation of these
DBFs.  Particularly problematic is the heavy reliance on indirect
spectrophotometric detection, which is unselective and insuffi-
ciently sensitive.  On the other  hand, CE may find a niche in
research  laboratories, especially  those that already have  the
instrumentation.
5 Liquid chromatography

Like capillary electrophoresis, ion chromatography (1C) suffers
from interferences from other anions due  to  its unselective
mode  of detection (conductivity).  At concentrations  of
25 ug L~', haloacetate standards were shown to be amenable
to 1C,  and reported detection limits (based on  1-4 ugL"1
spikes) were <1 ugL"1 for all analytes except tribromoace-
                                                                             J. Environ. Monit., 2000, 2, 285-291    287

-------
    1.4


    1.2


 T  1.0
 CO
 -5-0.8
    0.6
    0.4
    0.2
    0.0
VI
                  10    15    20    25
                       Retention time/min
                                          30    35
                                                      40
Fig. 2 Ion chromatogram of a haloacetic acid standard prepared in
reagent water using a Dionex (Sunnyvale, CA, USA) DX500 system
with AG11  and ASH  columns and suppressed  conductivity
detection.  Concentrations:  chloroacetic (MCAA), dichloroacetic
(DCAA), trichloroacetic (TCAA), bromoacetic (MBAA), dibromoa-
cetic (DBAA), and  bromochloroacetic  (BCAA) acids: l.OugL"1;
bromodichloroacetic (BDCAA) and dibromochloroacetic (DBCAA)
acids: 3.13 ug L~*; tribromoacetic acid: 4.17 ug L"1. Reprinted from
ref. 30 with permission. See ref. 30 for gradient program and other
details.
tate.   However,  at more  realistic concentrations of  2—
10 ug L~' and in  actual water samples, matrix interference
became severe despite a  preconcentrating  extraction into
MTBE. Even in reagent water, 1C was limited in its ability to
separate the analytes from background anions; see chromato-
gram  in  Fig. 2.33  Even  with  preconcentration,  realistic
concentrations for tap water could not be  determined  by
1C, with the best detection limits too high by a factor of at
least  4.34 Attempts  to  find alternate  detectors,   such  as
polymer-modified microelectrodes, have led to small increases
in sensitivity,  but the detection  limits are  not low enough
without about 50-fold preconcentration.35 High performance
liquid chromatographic methods have also been attempted,
but their detection limits are at least  10 000 times greater than
that of GC-ECD  and they have no hope of application to
potable water  analysis at this time.36'37
6  Electrospray ionization mass spectrometry

This technique has two important advantages in  that it is
both  sensitive and selective. When coupled with  liquid
chromatography, ESI-MS can achieve detection  limits of
<70ngL~',  and  it  has  been  applied  to several real
matrices.38 Sample preparation requires a single extraction
step (the same as GC-ECD) and minimal sample volume.
Selectivity  improvements have been made using reagents
(perfluorinated alkylcarboxylates) that form stable associa-
tion complexes with the analytes; these complex anions are
detectable by ESI-MS.39 The disadvantage of ESI-MS is the
cost of the instrumentation and its  lack  of availability in
most testing laboratories.  However, for research labora-
tories, ESI-MS can be an ideal complement to GC-ECD for
confirmatory analyses.
7  Preservation and fixation
Not all samples require this kind of pretreatment, and it is not
intrinsically necessary for the fundamental mechanism of the
analytical procedure itself. Nevertheless, improperly preserved
and fixed samples can  lead to meaningless  analytical data.
Initially defining clear objectives for the use of data is essential
to  choosing  appropriate  means  to preserve and fix these
analytes in solution. Real samples nearly always require some
type of preservation and fixation. It is not always possible to
immediately analyze a sample of potable water; consequently,
effort should be made to preserve the sample. Consider how
analyte concentration may change:  volatility loss, chemical
reaction  (loss or gain), biodegradation loss.  It is therefore
important  to preserve analytes  from loss  and  fix  their
concentrations at the  time of sampling.
  After disinfection, residual oxidizing agents remain behind
(e.g., HOBr, C1O~). Residual disinfectants continue to provide
protection from pathogens as water navigates the distribution
system on its way to homes.! However, this additional contact
time allows  additional  reaction  and  thus  additional DBF
formation. When a sample of water is taken for analysis, the
residual oxidants must  be scavenged (quenched) to prevent
further reaction with organic matter. Active chlorine residuals
in potable water are limited by EPA to no more than 1 mg L~'
Cl2 (14 uM). Even at this low concentration, further halogena-
tion    of    organic    matter    is    possible,    e.g.,
C1CH2CO2H-»C12CHCO2H-»C13CCO2H. In bromide-con-
taining  waters,  bromine  atoms  can also be incorporated
through  formation of species such as  BrCl and  HOBr.  In
Method  552.2,"  ammonium  ion is  added  (as  NH4C1)  to
convert hypochlorite or dichlorine to chloramine (NH2C1) so as
to minimize additional DBF formation or conversion to more-
halogenated forms. Other reducing agents, such  as ascorbic
acid, have also been used.9'40 Because dihalo- and trihaloace-
tates represent oxidized species, care must be taken not to add a
reagent capable of reducing these analytes. For instance 30 uM
sulfite has been demonstrated to destroy the haloacetonitrile
analogs   of  some   HAAs,   specifically  CC13CSN  and
CHBr2C=N.41 No such effect was observed  for the HAAs
when ~ 1.4 HIM sulfite was added either immediately before
analysis or after holding 7 or 14 days in Method 552.2." Some
sulfite is rapidly  consumed by the chloramine and some is
subject to reaction with dissolved oxygen, accounting for about
~ 10-50% of the reducing capacity of the sulfite. Thus, excess
sulfite was present throughout the holding times tested.
  Many  chemical  reactions  can be  speculated;  however,
minimal  investigation has taken place  to  indicate which
reactions are responsible for  the  loss of these analytes. For
example, hydrolysis-decarboxylation of trihaloacetic acids to
give H2CO3/CO2 and CHX3 (the corresponding haloform) are
known. Other hydrolyses may also be possible. For example,
nucleophilic attack by water at the trichloromethyl group could
convert  trihaloacetates to  ethanedioate  (oxalate),  which
actually  has been  detected  by  GC-MS  after  NOM  is
chlorinated.1  To  minimize  hydrolyses,  acidification  with
sulfuric acid is normally prescribed. Any nucleophilic species
(not just water) may in fact be able to substitute for a chloride
or  bromide; thus,  exposure  to  nucleophiles  should  be
minimized. Although chloramine  and ammonia (from excess
ammonium cation) are somewhat nucleophilic,42 this does not
appear to pose a problem in Method 552.2.">43
  Biodegradation is also a potential problem that has been
reported.44 When all residual oxidant is quenched, microbial

fAlthough a more  detailed explanation is beyond the scope of this
work, it is worth noting that sewage and potable water lines are often
run  near each  other. When breaks or leaks occur, positive pressure
normally  forces water  out of a supply line, thereby preventing
contamination. However, if supply pressure drops below ~ 140 kPa
(~20 Ib in~2),  sewage-tainted water  can seep  into a  supply line,
establishing a cross-connection. Residual disinfectant minimizes the
health impacts of cross-connections  by  inactivating microbes that
infiltrate the supply line.
288    J. Environ. Monit., 2000, 2, 285-291

-------
activity may Yead to analyte loss. Method 552.2 calls for the
addition of 0.27 M  copper(n) sulfate, which could act as an
antimicrobial agent (2.0 g CuSCVSt^O per 30 mL  sample).
However, Method 552.2 does not use the cupric cation for this
purpose. To minimize volatilization, Method 552.2 requires
samples to be collected in vials without airspace. Loss to the air
was adjudged to be a more serious problem than biodegrada-
tion. Nonetheless,  when an EPA-approved method is not
required, this reagent can be added at the time of collection,
especially when  biological activity is suspected to present a
problem. Refrigeration is  recommended in  all  cases  (and
required by Method 552.2) since volatilization and microbial
growth decrease  with  decreasing temperature.
  Not  only are stored samples subject to changes in analyte
concentration, but standards  themselves  can be  affected.
Unsurprisingly,  when the carboxylic acids  are prepared as
methanolic solutions, spontaneous esterification occurs.45 To
avoid  this  phenomenon, HAA9 standards  are prepared in
MTBE rather than  methanol.
  In addition to analyte loss by chemical means, physical losses
are also possible. Besides volatilization loss (as noted above),
analytes may be adsorbed to the container wall or absorbed
into the container structure. For this reason,  glass vials and
PTFE-lined caps are required in all versions of EPA Method
552. Because the samples are acidified, the neutral acids (rather
than the anions) will be present at sufficient concentrations that
they could  migrate  into some  materials, especially low grade
plastics. Likewise, a poor choice of container can lead to the
introduction of  contaminants that  overlap  with  the analyte
peaks in the chromatogram or otherwise skew results.
  In general, preservation and fixation  require balancing the
objectives of the  research project with possible interferences or
adverse effects. An investigator involved in HAA research must
be aware that adding or omitting reagents or practices may lead
to unexpected and  undesirable  consequences.  Vigilance and
scrupulous  evaluation of data cannot be overemphasized.
8  Quality assurance

Once the effort has been expended to analyze a sample, we
would like to have some assurance that the correct answer has
been obtained. To this end, a number of actions can be taken to
ensure that the data are  of high quality. Those actions most
commonly taken by practitioners in the field using GC-ECD
will  be summarized here.§ Use of high  quality  reagent
chemicals, appropriate glassware and measuring devices, and
good laboratory technique are important in any experiment.
There are five actions commonly taken: (1) adding a surrogate
to test the analyst, (2) adding an internal standard to test the
instrument, (3) analyzing several replicates to ensure precision,
(4) running calibration check standards to ensure accuracy, (5)
analyzing  spiked  samples to  ensure the matrix  does not
interfere (method ruggedness).
  Part of the process requires knowing what the final objectives
for the data are. Is a false positive or a false negative of greater
concern? False positives are uncommon with most chromato-
graphic methods. However, suppose that no peaks are seen in a
chromatogram. Does that mean  that the analytes were not
present? Even the best analyst will forget  a step occasionally.
Or a reagent will deteriorate. Every step in a  method is a place
for something  to go  wrong.  To detect such problems, a
surrogate analyte is often added. The surrogate is added to all
real samples at the beginning of the procedure. Surrogates must
have  several properties. (1) They must not interfere with the
determination of the actual analytes.  Overlapping chromato-
graphic peaks  is  one such  interference. (2)  A surrogate's
§These actions apply to many analytical methods and reflect standard
practices of quality assurance in environmental chemistry.
concentration must be at least as difficult to determine as that
of the actual analytes. For example, a surrogate that is much
easier to methylate than the HAAs cannot test the effectiveness
of the methylation step. (3) Surrogates must not be present in
real  samples.  Common  surrogates for HAA determination
include  fluorinated  carboxylic  acids,  which neither  occur
naturally  nor result  from  the  disinfection  process. If the
performance of the surrogate is acceptable,  the analyst can
have reasonable confidence that every step of the procedure
was successful.^
  Haloalkanes, such.as  1,2,3-trichloropropane or 1,2-dibro-
mopropane, are used  as internal standards (dissolved into the
MTBE extraction solvent). They verify that the instrument is
functioning properly (i.e., correct injection volume, retention
time, and detector response). If the internal standard peak is as
expected, but the surrogate peak is missing, we can infer that
either the extraction or the methylation failed. If the internal
standard and surrogate peaks are half of  what would be
expected, a clogged needle may be to  blame. If the internal
standard retention time shifts, the column may be deteriorat-
ing.
  One of the ways to deal with random errors  is to analyze
replicate samples. In practice, large numbers of samples taken
for many studies preclude each being run 3-5 times due to cost
and time limitations. Nevertheless, it is common to include
duplicates  at a convenient ratio,  such as 1 every 5  or 10
samples.  Experienced  analysts will  establish   criteria for
duplicate  analyses based  on  careful studies  with  many
replicates. For  Method  552.2, duplicate analyses will usually
agree within 20%. The precision needed will vary with the
intended  use  of  the data.  Measuring a 30% change in
concentration will require greater precision than demonstrating
that an  analyte is undetectable.
  Accurate measurement is of obvious importance. It is worth
noting that there can be no assurance of accuracy without
adequate precision, especially when  the number of replicate
measurements is small. Ensuring accuracy begins with calibra-
tion of the instrument, including running sufficient standards
over the linear dynamic  range. This  begins with establishing
minimum acceptable values for the correlation coefficient (e.g.,
r2>0.95), standard error of the slope, value of the ^-intercept
(and its standard error), in other words, a set of quality control
criteria. When  calibration data fall outside  this range, the
calibration must be repeated. For a chromatographic method,
the y-intercept of a calibration  curve  should be statistically
indistinct  from zero (no analyte should mean  no peak). To
ensure that the overall method response remains constant for
the analytes, calibration check standards near  the expected
concentrations should be periodically run, such as every 5-10
samples. The blank  or  zero-concentration point is usually
deionized water for this method; it should have no measurable
peaks at the methylated analyte retention times. Although a
full discussion is beyond the scope of this work, peak shape and
signal-to-noise  ratio should be incorporated into considera-
tions  of data quality, especially near the limit  of detection.
Integrating a noisy signal can  result in poor precision, and
baseline  noise  can lead  to unsatisfactorily high  limits of
detection.
  Fortifying (spiking) a  known  amount of analyte into a
sample and measuring the recovery is  a way to see if the matrix
interferes.  For example,  if  a  sample  is found to contain
10 ng L~' tribromoacetic acid, we might spike it with 7 ug L~'
more. Ideally, we should measure a concentration of 17 ug L~'
in the spiked sample. Suppose we detect  15 ug L~' in the spiked
sample. There are two ways to calculate the recovery [eqns. (4)
and  (5)].  Eqn. (4) is prescribed  by  EPA  Method  552.2.

fin direct injection 1C,  the same species  can serve as surrogate and
internal standard because there is only one step. In this case, it is not
normally thought of as a surrogate.
                                                                              J. Environ. Monit., 2000, 2, 285-291    289

-------
Table 2. Summary of analytical techniques useful for HAA determination in potable water
Technique    Characteristic
GC-ECD
GC-MS
CE
1C
Sensitive (LOD< 1 ug L~')
Moderately selective
High preparation; extraction and esterification required
Long instrumental analysis time (>55 min)

Moderately insensitive (LODas 100 ugL"1, too high for potable water)
Higher sensitivity reported for 2,4-difluoroanilide derivatives
Selective based on mlz
High preparation; extraction and esterification required

Moderately insensitive for tested analytes (LODaj 100 ug L~', too high for potable water)
High sensitivity reported for some analytes with appropriate preparation
Moderate preparation; double extraction required
Fast instrumental analysis time (< 10 min)

Moderately sensitive (LOD« 1 ug L~' in deionized water, but 10-100 x higher in real water samples); double extraction required
Matrix-dependent sensitivity; not very rugged
Moderately fast instrumental analysis time (<40min)
ESI-MS      Sensitive (LOD =S 1 ug IT')
             Selective, especially when complexation is used
             Minimal sample preparation, only one extraction required
             Fast instrumental analysis time (<10min)
"Abbreviations: GC = gas chromatography; ECD = electron capture detection, MS = mass spectrometry, CE=capillary electrophoresis, IC=ion
chromatography, ESI=electrospray ionization, LOD=(lower) limit of detection.
Recoveries  of  ±20% are  common  when calculated using
eqn. (4); the allowed variability is ±3%, so this recovery would
be acceptable using Method 552.2.
         recovery (1) =
          measured change due to spike
              known value of spike
(4a)
Acknowledgements

The assistance of Jennifer Heffron, Raymond A. Hauck,  and
Betty L. Merriman in gathering materials for this  article  is
noted. Mention of specific brand names or products should not
be construed as endorsement by the US government.
         recovery (1) =
                                -=71%
recovery (2) =
  measured concentration in spiked sample
 calculated concentration in spiked sample
         recovery (2) =
(4b)
(5a)
                                            (5b)
The term quality assurance has become an anathema in many
circles, but at its basic level it means making sure the analytical
measurement is in fact the right answer. It is the responsibility
of the analyst  to choose appropriate QA criteria to meet the
overall objectives of the investigation.
9  Conclusions

For the present,  the most widespread technique for HAA
determination  is  GC-ECD.  Characteristics of the available
techniques are summarized in Table 2. Although diazomethane
can  be used satisfactorily to methylate the carboxylic acid
moiety for HAAS, acidified methanol gives better results for
HAA9 analytes. LC methods are not sufficiently robust at this
time  to meet  drinking  water  needs.  Instrument cost and
availability prevent ESI-MS  from becoming more prominent
even though it meets sensitivity and selectivity requirements.
Because all EPA-approved methods for compliance monitoring
are  based  on  GC-ECD  and  this technique  meets  both
sensitivity and selectivity needs, it can be expected to dominate
drinking water analysis in the US for the forseeable future.
Sample preservation and fixation are complicated  and must be
attended to if usable data are to be obtained.
References

 1  S. D. Richardson, in Encyclopedia of Environmental Analysis and
    Remediation, ed. R. A. Meyers, Wiley, New York, 1998, vol. 3,
    pp.1398-1421.
 2  H. Weinberg, Anal.  Chem., 1999, 71, 801A.
 3  US Environmental Protection Agency, Fed. Regist.,  1998,63(241),
    69390.
 4  F. W. Pontius and W. R. Diamond, J. Am. Water  Works Assoc.,
    1999, 91, 16.
 5  D. C. Harris,  Quantitative Chemical Analysis,  Freeman, New
    York, 2nd edn., 1987, Appendix G.
 6  J. W. Hodgeson, 3. Collins and R.  E. Barth,   Method 552.
    Determination of Haloacetic Acids and Dalapon in Drinking Water
    by Liquid-Liquid Extraction,  Derivatizalion, and Gas Chromato-
    graphy  with Electron Capture Detection,  In Methods for the
    Determination of Organic Compounds in Drinking Water, Supple-
    ment 7, U.S. Environmental Protection Agency, Cincinnati, OH,
    July 1990, EPA/600/4-90/200.
 7  3. W. Hodgeson, D.  J. Munch, J. W. Munch and A. M. Pawlecki,
    Method 552.2. Determination  of Haloacetic Acids and Dalapon in
    Drinking Water by Liquid-Liquid Extraction, Derivatizalion, and
    Gas Chromatography with Electron Capture Detection, Rev.  1.0. In
    Methods for the Determination of Organic Compounds in Drinking
    Water, Supplement III, U.S.  Environmental Protection Agency,
    Cincinnati, OH, Aug. 1995, EPA/600/R-95/131.
 8  K. Blau and J. Halket, Handbook of Derivatives for Chromato-
    graphy, Wiley, Chichester, 2nd edn., 1993, passim.
 9  E. T. Urbansky, D. M. Freeman and F. J.  Rubio, J. Environ.
    Monit., 2000, 2, 253.
10  U.S. Environmental  Protection Agency, Fed. Regist., 1996, 61(94),
    24354.
11  A. M. Pawlecki-Vonderheide, D. J. Munch and J. W. Munch,
    J. Chromatogr. Sci.,  1997, 35, 293.
12  M. J. LaGuardia, Proc.  Water Qual. Technol.  Con/, Am.  Water
    Works Assoc., 1997,  256.
13  F. J. Rubio, E. T. Urbansky and M. L. Magnuson, J. Environ.
    Monit., 2000, 2, 248.
14  Y..Xie, D. A.  Reckhow and D. C. Springborg, J. Am.  Water
    Works Assoc., 1998,  90, 131.
15  M. J. LaGuardia,  Proc. Water Qual. Technol. Con/, Part I, Am.
    Water Works Assoc., 1996, 167.
290     J. Environ. Monit., 2000, 2, 285-291

-------
16  Standard Methods for the Examination of Water and Wastewaler,
    ed. A. D. Eaton, L. S. Clesceri and A. E. Greenberg, American
    Public Health Association/American Water Works  Association/
    Water Environment  Federation, Washington,  DC, 19th edn.,
    1995, Method 6251, pp. 6-67-6-76.
17  J. W. Hodgeson and D.  Becker, Method 552.1.  Determination of
    Haloacetic Acids and Dalapon in Drinking Water by Ion-Exchange
    Liquid-Solid Extraction and Gas Chromatography with an Electron
    Capture Detector, Rev. 1, In Methods for  the Determination of
    Organic  Compounds  in  Drinking   Water,   Supplement  II,
    U.S. Environmental Protection  Agency, Cincinnati, OH, Aug.
    1992, EPA/600/R-92/129.
18  D. Benanou,  F. Acobas and P. Sztajnbok,  Water Res., 1998, 32,
    2798.
19  H. L. Shorney  and S. J. Randtke, Proc.  Water Qual. Technol.
    Conf., Part 1, Am.  Water Works Assoc., 1995, 453.
20  S. Seckinger,  C. R. Ashe and J. F. Occhialini, Proc.  Water Qual.
    Technol. Conf., Am. Water Works Assoc., 1997, 9/A.
21  P. Neitzel, W. Walther and W. Nestler, Fresenius' J. Anal. Chem.,
    1998, 361, 318.
22  H. Ozawa and T. Tsukioka, Analyst, 1990,  115, 1343.
23  H. Ozawa, /.  Chromatogr., A, 1993, 644, 375.
24  B. F. Scott and M. Alaee, Water Qual. Res. J. Can., 1998,33, 279.
25  L. J. Yoo, Y.  Shen and S.  Fitzsimmons, Proc.  Water  Qual.
    Technol. Conf., Am. Water Works Assoc., 1991, 913.
26  L. J. Yoo, Y. Shen, S. Fitzsimmons and M. Rigby,  Proc. Water
    Qual. Technol. Conf., Part 2, Am. Water Works Assoc., 1993,1335.
27  Y. Xie and J.  P. Romano, Proc. Water Qual. Technol. Conf., Am.
    Water Works Assoc.,  1997, 5A6/1.
28  Y.  Xie,  H.  Zhou and J. P. Romano,  Abstracts  of papers
    (Environmental Division),  217th American Chemical  Society
    Meeting, Anaheim, CA,  USA, 1999, abstract 092.
29  D.  Martinez, J. Farre,  F. Borrull, M. Calull, J.  Ruann and
    A. Colom, J. Chromatogr., A, 1998, 808, 229.
30  D. Martinez, F. Borrull and M. Calull, /.  Chromatogr., A, 1999,
    835, 187.
31  D. Martinez, F. Borrull and M. Calull, /.  Chromatogr., A, 1998,
    827, 105.
32  D. Martinez, F. Borrull and M. Calull, TrA C, Trends Anal. Chem.,
    1999, 18, 282.
33  V. Lopez-Avila, Y. Liu and C. Charan, J. AOAC Int., 1999, 82,
    689.
34  C. Sarzanini, M. C. Bruzzoniti and E. Mentasti, J. Chromatogr., A,
    1999, 850, 197.
35  P. Akhtar, C. O. Too and G. G. Wallace, Anal.  Chim. Ada, 1997,
    341, 141.
36  R. Vichot and K. G. Furton, J. Liq. Chromalogr., 1994J17, 4405.
37  H. Carrero and J. F. Rusling, Talanla, 1999, 48, 711.
38  S. Hashimoto and A. Otsuki, J. High Resolul. Chromatogr., 1998,
    21, 55.
39  M. L. Magnuson and C. A. Kelty, Anal. Chem., 2000, 72, 2308.
40  Y. Takahashi and M. Morita, Kankyo Kagaku, 1997, 7, 495.
41  J.-P. Croue and D. A. Reckhow, Environ. Sci. Techno!., 1989, 23,
    1412.
42  E.  J.  Pedersen III,  E. T.  Urbansky,  B. Y.  Marinas  and
    D. W. Margerum, Environ. Sci. Technol., 1999, 33, 4239.
43  E. E. Chang, Y.-W. Ko, P.-C. Chiang and C.-F. Wang, Environ.
    Sci. Res., 1996, 51, 409.
44  A.  L. Weightman, A. J. Weightman  and J.  H. Slater,  World
    J. Microbiol.  Biotechnoi, 1992, 8, 512.
45  Y. Xie, D. A. Reckhow and R. V. Rajan, Environ. Sci. Technol,
    1993, 27, 1232.
                                                                                     J. Environ. Monit., 2000, 2, 285-291     291

-------