Influences of metal cations on the determination of the
a-oxocarboxylates as the methyl esters of the 0-(2,3,4,5,6-
pentafluorobenzyl)oximes by gas chromatography: the importance of
accounting for matrix effectsf J
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, Cincinnati, OH45268,, USA. E-mail: urbansky.edward@epa.gov; Fax: +1513569
7658; Tel: +1 513 569 7655
Received 17th April 2000, Accepted 9th May 2000
Published on the Web 9th June 2000
The a-oxocarboxylates (a-ketocarboxylates) and the corresponding a-oxoacids (a-ketoacids) have been reported
as disinfection byproducts of ozonation of potable water supplies. In this analytical method, the oxo moiety is
derivatized with O-(2,3,4,5,6-pentafluorobenzyl)oxylamme (PFBOA) to form an oxime which is then extracted
into tert-butyl methyl ether. The carboxylic acid moiety is esterified (methylated) with diazomethane. In this
study, five analytes were investigated: oxoethanoate (glyoxylate), 2-oxopropanoate (pyruvate), 2-oxobutanoate
(2-ketobutyrate), 2-oxopentanoate (2-ketovalerate), and oxopropanedioate (ketomalonate, mesoxalate). The
influence of Lewis acid metal cations in the water matrix was evaluated for the gas chromatographic method
commonly used for the quantitation of these analytes at concentrations ^ISOngmL"1. Tested metals included
Ca(n), Mg(ii), Fe(iK), Cu(n) and Zn(n). At typical concentrations, calcium, in particular, can have profound
impact, especially on oxoethanoate quantitation. Oxopropanoate experiences an increase in recovery in the
presence of metal cations. 2-Oxobutanoate and 2-oxopentanoate are the most resistant to these effects, but
2-oxopentanoate shows increased recoveries at higher concentrations when assayed in the presence of calcium
ion. Oxopropanedioate generally shows poorer precision and recovery when determined in solutions containing
metal ions. This investigation demonstrates the significance of metal effects in the quantitative determination of
these analytes and further emphasizes the importance of thorough matrix characterization and careful recovery
studies with fortified (spiked) samples and blanks.
1 Aim of investigation
The a-oxocarboxylates are routinely determined by gas
chromatography using a two-step process.1 First, the a-oxo
moiety is derivatized with O-(2,3,4,5,6-pentafluorobenzyl)ox-
ylamine (PFBOA). Second, the carboxylic acid moiety is
esterified or silylated to increase volatility. Oximation has two
advantages: first, it permits extraction and therefore concen-
tration of hydrophilic analytes into an organic solvent; second,
it incorporates the pentafluorobenzyl moiety, thereby permit-
ting electron-capture detection. Several investigators have thus
relied on this procedure for measuring the concentrations of
ozonation byproducts (OBPs).2""4 Under the dilute concentra-
tions (< 100 |iM) that these species occur in post-ozonation
drinking water systems, they exist >99.9% as the ionized
anions rather than the parent carboxylic acids; therefore, they
can also be determined by ion chromatography.5 These species
are listed in Table 1. Structures of the analytes are shown in
Fig. 1.
This method has been incorporated into a number of
schemes for the identification and quantitation of OBPs;
however, there is a minimum of information on the ruggedness
tThis paper is the work product of United States government
employees engaged in their official duties. As such, it is in the public
domain and exempt from copyright restrictions. © US government.
|This paper is dedicated to the memory of Professor Jerry March,
whose book Advanced Organic Chemistry has become an invaluable
reference across the disciplines of chemistry and remains one of the
most authoritative and comprehensive works in the field.
of this method or its susceptibility to matrix effects, such as
metal cations. Carboxylates form coordination complexes with
alkaline-earth and transition metals. A number of metal
cations likely to be encountered in potable water supplies
were selected to test for matrix effects at concentrations that
could occur. Our aim was to show that these metal cations
could in fact have a significant impact on the measured
concentrations of the analytes that must be addressed through
proper characterization of the matrix constituents and
duplication of the matrix for use in preparing calibration
graphs.
2 Experimental
2.1 Metal cation solutions
Stock solutions of complexable (Lewis acid) metal cations were
made by dissolving the analytical reagent grade perchlorate
salts (obtained from GFS, Columbus, OH, USA) into doubly
deionized water. Ca2+and Mg2"1" stocks were prepared at
10.0 mgmL"1, while Mn2+, Zn2+, Cu24Y and Fe3+ stocks
were prepared at l.OOmgmL"'). Portions of these solutions
were added to test solutions to determine the effects of the
metal cations. This was done by keeping metal cation
concentration constant and varying the analyte concentration.
2.2 Analytes and test solutions
An aqueous standard was prepared at 1000 ugmL"1 in each
analyte by dissolving the commercially available reagents into
334 J. Environ. Monit., 2000, 2, 334-338
This journal is © The Royal Society of Chemistry 2000
DOI: 10.1039/b0030651
-------
Table I a-Oxocarboxylatcs examined in this study and chromatographic retention tiroes for the doubly derivatized analytes
Analyte synonyms
Formula of parent acid
Acid CAS registry no.
Retention times for oxime Me ester isomers/min
Oxocihanoa'.c
Glyoxylate
Formylformate
2-Oxopropanoate
2-McthylgIyoxylate
Pyruvate
2-Oxobutanoate
2-Ketobutyrate
a-Ketobutyrate
2-Oxopcntanoate
2-Ketovalerate
cc-Ketovaterate
Oxopropanedioate
Kctomalonate
Oxomalonate
Mesoxalate
HC(0)C02H
CH3C(O)CO2H
C2H5C(0)C02H
C3H,C(0)C02H
HO2CC(COC02H
563-96-2°
113-24-6*
600-18-0°
13022-83-8''
7346-13-6'
12.3, 13.1
12.6, 14.3
14.4, 15.6
16.5, 17.5
19./
dium salt can be made anhydrous, both anionic forms and the acid exist as gem-diols at carbon 2. Thus, oxopropanedioic acid actually exists
mostly in the form of dihydroxypropanedioic acid: C(OH)2(CO2H)2. The same would be true for the deprotonated anions in aqueous solution.
'Carbon 2 of Oxopropanedioate is not chiral; therefore, only one geometric isomer is formed upon oximation. .
doubly deionized water (see Table 1). Volumes of this stock
standard were diluted 1:49 to produce a working standard
20.0 ugmL"1 in each analyte. Both stock and working
standards were kept in polypropylene bottles under refrigera-
tion at 4±1°C. The working standard was injected via
microliter syringes or Eppendorf® pipettor (Brinkmann,
Westbury, NY, USA) into 20.0 mL portions of doubly
deionized water to produce test solutions. Blanks (no analytes
added) were also prepared. Test solutions were prepared
directly in new 40 mL borosilicate glass vials with screw-caps
and PTFE-lined septa obtained from Supelco (Bellafonte, PA,
USA) or Nalge Nunc (I-Chem International, Rochester, NY,
USA).
This work is divided into three parts. (1) Test solutions were
prepared by holding an individual metal cation concentration
constant (one metal at a time) and then varying the analyte
concentrations to produce matrix-dependent calibration
curves. (2) Test solutions were prepared by holding the analyte
concentration at SOngmL"1 and then varying the metal
concentration over a typical range. (3) Test solutions were
prepared to contain a synthetic drinking water matrix of
several metal cations at fixed concentrations and then varying
the analyte concentrations to produce matrix-dependent
calibration curves. Post-mixing concentrations for all species
are reported in the tables (see below).
2.3 Sample preparation
2.3.1 Oximation and extraction. Aqueous solutions of
O-(2,3,4,5,6-pentafluorobenzyl)oxylamine were prepared
fresh daily at lOmgmL"1 of the hydrochloride salt,
PFBOA-HC1 (Sigma, St. Louis, MO, USA) in a manner
similar to EPA Method 556.6 Portions of 1.0 mL each of
PFBOA solution and I.OM phosphate buffer (0.50 M
NaH2PO4+0.50 M Na2HPO4, Fluka, Buchs, Switzerland)
were added to each 20.0 mL of test solution. The vials were
placed into a forced air oven thermostated to 45±2°C for
90 min. Afterwards, test solutions were chilled in an ice bath in
order to quench the oximation reaction. A 1.0 mL portion of
9.0 M H2SO4(aq) (prepared from 98% w/w H2SO4, J.T. Baker,
Phillipsburg, NJ, USA) was added to protonate the carb-
oxylate groups so that they might be extracted. The test
solutions were then extracted with 4.0 mL of pesticide residue
analysis (PRA) grade tert-butyl methyl ether (MTBE) (Aldrich,
Milwaukee, WI, USA). Addition of 2 drops of 2.5mgmL~'
FD&C Blue No. 1 aqueous solution (Warner Jenkinson, St.
Louis, MO, USA) was used to improve the visibility of the
phase separation. The extracts were transferred to 4.0 mL glass
vials to which ~0.01 g of Na2SO4 had been added to remove
residual water dissolved in the MTBE (EM Science Tracepur®,
Gibbstown, NJ, USA). Sample preparation is summarized in
Scheme 1.
O O
ii n
H-c-c-cr
O O
II II
CH3—c-c-cr
o o
CHjCH2CH2-C-C-C
?> ff
o o
n n
CH3CH2—C-C-Q-
o
II
o
II
o
OH
5b
Fig. 1 Structures of et-oxocarboxylates: 1, oxoethanoate (glyoxylate);
2, oxopropanoate (pyruvate); 3, 2-oxobutanoate (a-ketobutyrate); 4, 2-
oxopentanoate (a-ketovalerate); 5a, Oxopropanedioate (ketomalonate);
and 5b, the f«m-diol of Sa, dihydroxypropanedioate.
2.3.2 Esterification. Methylation of the carboxylic acid
functionalities was done using a flowing stream of diazo-
methane (CHkNJ in argon. This minimizes the risks associated
with concentrated CH2N2 solutions and the uncertainty
associated with adding a volume of diazomethane solution to
a volume of the recovered MTBE phase. Using a gas stream
eliminates the need for measuring a specific amount of the
MTBE phase as the dissolved diazomethane does not affect the
volume of the ethereal solution.
A solution of Af-methyl-Al'-nitroso-.p-toluenesulfonamide [80-
111-5] (Diazald®, Aldrich) was prepared fresh daily by
combining 3.0 g AT-methyl-A/'-nitroso-p-toluene-sulfonamide
and 30 mL of mixed solvent (8 mL USP EtOH + 22 mL PRA
grade MTBE, scaled as needed). A solution of 33% w/w sodium
hydroxide was prepared by diluting 50% w/w solution obtained
from Fisher Scientific (Pittsburgh, PA, USA).
The apparatus used was similar to that shown in EPA
Method 552;7 however, the diazomethane gas was dispensed
directly into the recovered extract rather than being collected in
J. Environ. Monit., 2000, 2, 334-338 335
-------
Oximatewith 0-(2,3,4,5,6-
pentafluorobenzyljoxylamine
(PFBOA) to Introduce chromolag;
substitute at the a-oxo moiety.
Acidify with H2S04 to protonate the
anion and extract the parent acid
into (ert-butyl methyl ether.
ft ft PFBOA
H—C-C-O"
S0-CH2
Esterify the carboxylic acid with
dlazomethane (CH2N2).
CH2N2
H-C-C-OH,
o
II
H C C—OCHs
Analyze by gas chromalography
with electron capture detection.
Scheme 1 Flow chart for determination of the ot-oxocarboxylates; oxoethanoate (glyoxylate) is used as the example analyte. Only the analytically
important reactants and products are shown. For instance, protonation of oxime nitrogens is not shown.
ethyl ether. Immediately preceding the methylating apparatus,
the argon stream was saturated with MTBE vapor by passing it
through a sintered glass dispersion tube immersed in MTBE;
this prevents evaporative loss. Extracts were transferred to
16 mm x 100 mm disposable borosilicate glass test tubes and
bubbled with the CH2N2-Ar stream for 30-45 s. When a
pronounced straw yellow tint (due to the CHjNjj) was visible,
bubbling was stopped. Tubes were stoppered and the reaction
was considered to be complete after 30-45 min had elapsed.
The methylated extracts were transferred to autosampler vials
and stored in a freezer at -15°C prior to GC-ECD analysis.
2.4 GC-ECD analysis
Oximated and methylated extracts were analyzed on a Hewlett-
Packard (Palo Alto, CA, USA) 6890 GC-ECD system
equipped with an HP 7673 autoinjector. Using splitless
injections, volumes of 2.0-4.0 uL were loaded onto a J&W
Scientific (Folsom, CA, USA) DB-5MS column
(30 m x 250 um id x 0.25 urn film) at a constant (high purity)
helium flow of 1.0 mL min"1; inlet and detector temperatures:
270 "C. Temperature program: hold at 60 °C for 2.0 min; ramp
at 20.0°Cmin~
4.0 °C min"1
4.0 °C min"1
to
to 120 °C, hold for 1.0 min; ramp at
130°C, hold for 2.0 min; ramp at
to 150°C; ramp_at 5°Cmin~' to 200 °C, hold
for 1.0 min; ramp at 20 °C min"1 to 260 °C. Retention times for
the doubly derivatized analytes (two geometric isomers for
each analyte except oxopropanedioate) are given in Table 1.
3 Results and discussion
3.1 Possible modes of interaction
It is possible to postulate a number of mechanisms by which
metal cations could interact in this procedure. Carboxylates
can act as Lewis bases to form dative bonds with transition and
alkaline-earth metal cations. In addition, Mn(RCO2)2 pre-
cipitation is possible. Even at concentrations of 150 ng mL"! of
each analyte, the total ligand (analyte) concentration is
<10uM. Consequently, concentrations of di-coordinated
metal cations are extremely small, even if mixed ligand
complexes are allowed. Strong coordination complexes, of
the form [(RCO2)Mn]+, reduce the concentration of the parent
acid and thus also reduce extraction of the oximated analyte
into the MTBE phase. Although the parent carboxylic acids are
weak and the solutions are acidified with sulfuric acid, strong
Lewis acids such as Fe(m) can successfully compete with H"1"
for these species.
In addition to the carboxylate functionality, it is also
possible for the oxo group to be involved in coordination; thus,
electropositive metal cations may promote the nucleophilic
substitution required for oximation by increasing the polarity
of the oxo group. They may also stabilize the transition state by
allowing a hydroxide ion to act as a leaving group that is
already complexed to the metal. In this fashion, they weaken
the carbon-oxygen bond. Such effects may be primarily kinetic
in nature. Regardless, the end result is increased oximation and
therefore increased signal for a particular analyte.
Table 2 Percentage difference in analyte calibration curve slope relative to metal-free control caused by an individual metal cation at a specific
concentration"
Ajialyte
3.7mMCa2+ (ISOugmLT1)
2.0 HIM Mr* (SOugmL"1)
0.16 mM Cu2+ (10 ug mL"1)
Oxoethanoate
2-Oxopropanoate
2-Oxobutanoate
2-Oxopentanoate
Oxopropanedioate
-20±7
+ 12±4
+ 5±3
+9±6
(-10±20)*
(-3±6)
(-4±4)
(~4±4)
(-4+4)
-10±7
-12±6
-10±3
(-3 + 3)
"Relative difference=(mmclll|-mclriymclrl, expressed as a percentage; values in parentheses are not statistically distinct from zero. Intercepts dif-
fered from their standard errors by < 10%. Each curve was generated from eight standard solutions containing concentrations of 0 (blank) 10
20^40, 60, 80, 100, or ISOngmL"1 of all five analytes. Kr>0.91 except as noted. Injection volume was 3.0(0,. *.R2=0.78 for Ca2* data'
CR-=0.3& for Cu-+ data, which suggests a total failure of the method for this analyte in the presence of cupric ion.
336 J. Environ. Afonit., 2000, 2, 334-338
-------
3.2 Observed effects
As shown by Table 2, calcium exerts an effect under test
conditions, especially on oxoethanoate (reduced by 20%) and
2-oxopropanoate (increased by 12%). Calcium has a less
pronounced effect on the other three analytes. The effect on
Oxopropanedioate is somewhat debatable because of problems
with data quality; this particular analyte is known to suffer
from reproducibility problems.8 At analyte concentrations of
100 and 150 ng mL , precipitation is observed. Perhaps this
accounts for the loss of oxoethanoate.
Magnesium at the test concentration produced no adverse
effects except for Oxopropanedioate, and it is not possible to
state with certainty that this effect is due to the magnesium.
Therefore, it is concluded that this method tolerates up to
2.0 HIM Mg2"1" in solution. Curiously enough, the data in
Table 3 suggest that even low concentrations of magnesium
affect the results. Since the Table 3 data are based on a
relatively high analyte concentration, it may be that this effect
is not sufficiently pronounced at the lower analyte concentra-
tions used to produce the data in Table 2.
Cupric ion seems to have affected the smaller analytes the
most, as demonstrated by the results in Table 2. Neither
2-oxobutanoate nor 2-oxopentanoate showed statistically
significant effects from cupric ion. Both oxoethanoate and
Oxopropanedioate showed about 10% less response in the
presence of copper(ii). Again, Oxopropanedioate behaved so
poorly that it is difficult to attribute an effect to the copper(n)
ion. However, it seems reasonable to postulate that copper(n)
binds sufficiently with dihydroxypropanedioate (the actual
form) to interfere with the derivatization process.
Oxopropanoate experiences an increase in signal, possibly
due to increased oximation as suggested above. This phenom-
enon is observed regardless of the metal in Table 3. At
SOngmL"', 2-oxobutanoate and 2-oxopentanoate suffer
almost no effects when subjected to variable concentrations
of iron(w) and magnesium, as shown by Table 3. Calcium, on
the other hand, produces an increase in 2-oxopentanoate, but
not in 2-oxobutanoate. Table 2 shows that high calcium
concentrations increase the response for 2-oxopropanoate,
2-oxobutanoate, and 2-oxopentanoate. All metals caused a
drop for Oxopropanedioate. Furthermore, calcium and
copper(n) induce variability in Oxopropanedioate quantitation;
linearity of response for calibration standards was poor as
indicated by the least squares correlation coefficients in Table 2
(see footnotes to table).
(a)
(b)
9-2
OH
Fig. 2 Possible structures of a dihydroxypropanedioato-metal complex
where the metal has valence q. Note that the ligand is capable of
forming (a) a five- or (b) six-membered cycle with the metal cation. It is
possible that the coordinating oxygen site on carbon 2 is actually
deprotonated as can occur with citrate, owing to the Lewis acidity of
the metal ion. Oximation is blocked by the steric bulk of the metal and
its coordination sphere or by the two hydroxy groups. Oximations are
believed to occur via nucleophilic attack at the carbonyl carbon, as
opposed to a concerted SN2 reaction with hydroxide as a leaving group.
Formation of a gem-diol that is further stabilized by complexing with
metal cations makes the concentration of the 2-keto species very small,
thereby reducing the reaction rate and probably the thermodynaraic
favorability of oximation as well.
Oxopropanedioate is a particularly refractory species for this
double-derivatization GC analysis. This particular species has
a- and (3-carbonyl functionalities (the oxo moiety and the
second carboxylic acid). Both a- and (3-ketocarboxyIic acids are
known to decarboxylate readily; this is often used synthetically
with malonic acid derivatives.9'10 Both charged and uncharged
oxygen sites are capable of forming dative bonds with a metal
cation as shown in Fig. 2. Such complexes may be resistant or
sluggish to oximation.
3.3 Multiple metal cations
In the presence of a mixture of metal cations, significant
deviation from the metal-free control group is observed, as
indicated by the results in Table 4, with slopes increasing by
30% (2-oxopentanoate) or decreasing by as much as 60%
(2-oxopropanoate). The result for 2-oxopropanoate is some-
' what surprising based on Tables 2 and 3, where the signal was
increased. This suggests two opposing mechanisms. Most
Table 3 Percentage recovery as a function of variable metal cation concentration relative to metal-free controls
Metal concentration/ Oxoethanoate,
ugmL"
Oxopropanoate,
~
2-Oxobutanoate,
2-Oxopentanoatc, Oxopropanedioate,
50ngmL~'=b.68HM 50ngmL~'=0.57uM 50ngmL~'=0.49uM 50ngmL~!=0.43 UM 50ngraL~'=0.42uM
[Ca(u)J—
25, 620
50, 7250
75, 18SO
100, 2500
125, 3120
fMg(il)]—
10,410
20,520
30, 1240
40, 1650
50,2050
2,o! 36
4.0, 72
6.0, 770
8.0, 140
10, 780
64
63
62
44
52
72
59
72
74
71
109
84
72
76
73
143
135
134
135
151
123
133
134
136
130
117
130
124
134
130
100
97
96
92
100
93
93
94
97
101
99
98
93
98
94
'Data were acquired using mixed standards that contained all the analytes at 50 ng mL~'.
183
180
182
172
275
106
107
108
110
116
104
110
103
109
106
70
71
70
36
55
81
62
74
78
81
99
90
72
70
68
'Concentrations expressed in UM are italicized.
J. Environ.
Momr.,2000, 2, 334-338 337
-------
Tahlt 4 Percentage difference in analyte calibration curve slope
relative to metal-free control caused by a combination of metals
cations at specific concentrations"
Analyte
Metals mix* Metals mix+50 mM EDTAC
Oxoethanoate
2-Oxopropanoate
2-Oxobutanoate
2-Oxopentanoate
Oxopropanedioate
-27±4
-60±20
+3±2
+30±10
-21±19
+ 13±7
-6±3
+5±2
+40±10
+44±33
"Relative difference=(mmc,a|-mciri)/'"cirii expressed as a percentage.
Intercepts were statistically indistinct from zero. Injection volume
was 3.0 uL. *Metals mix contains 2.5 mM (100(igmL~') Ca2+,
1.2mM (30ugmL"') Mr+, 107HM (6.0ugmL~') Fe3+, 63 UM
(4.0ugmL-') Cu2+, 46uM (S.OugmL-') Zn2+, 9.1 nw
(O.SOugmL"1) Zn2+. TEDTA was added as a solution of O.lOn
Na2HiEDTA(aq) (1.00 mL to 20.0 mL sample).
likely, the very high metal cation concentrations encountered in
Table 4 conditions are responsible for sequestering the 1-2
carbon analytes in the aqueous phase through the formation of
complex ions of the form [(RCO2)Mn]+ . The result for
2-oxopentanoate is consistent with what was previously
observed at high calcium and analyte concentrations in
Table 3.
Although adding EDTA as a chelant appears to have
mitigated this effect for oxoethanoate, oxopropanoate, and
2-oxobutanoate, it seems to have adversely affected the
quantitation of Oxopropanedioate. With EDTA, the calibra-
tion curve slope exceeds that of the control by 44%. The effect
on 2-oxopentanoate (+40%) was also substantial, but not
especially important in terms of application to OBP determina-
tion since this species is rarely observed as an OBP. Overall, it is
clear that this method is not very rugged when a mixture of
common metal cations is present at concentrations that might
be encountered in real water supplies.
4 Conclusions
Given the magnitude of the effects that metal cations can have
on the determination of these species, it is imperative that
investigators carefully measure recoveries in the matrix of
interest by using fortified samples and blanks. In addition, a
thorough characterization of the matrix is highly recom-
mended. This allows the investigator to produce an equivalent
synthetic matrix that incorporates the same interfering metal
cations. Because the ct-oxocarboxylates compete with inorganic
and other organic ligands for the metal cations, it is not
possible to draw simple conclusions about the magnitude of
effects. As shown by the EDTA data above, it is not possible to
simply add a large bolus of strong chelating agent so as to
negate these effects.
Although interactions between metal cations and ligands can
be effectively modeled using software packages, e.g.,
MINEQL+, this requires a complete knowledge of the
constituents in the sample, including major organic ligands
and inorganic ligands, e.g., carbonate, phosphate, and sulfate.
Moreover, modeling requires accurate and precise values for
the stability constants of the metal-ligand equilibria at
conditions near zero ionic strength. At present, the stability
constants for common metals with the oc-oxocarboxylates have
not been measured and it appears that this deficiency will not
be addressed in the immediate future. Some values exist for
oxoethanoate and oxopropanoate, but not for 2-oxobutanoate,
2-oxopentanoate, and dihydroxypropanedioate. Consequently,
the current state of knowledge precludes prediction of metal
cation effects using multiple simultaneous equilibrium calcula-
tions.
Acknowledgements
Warner Jenkinson Co., Inc., is acknowledged for providing a
complimentary sample of FD&C Blue No. 1 for use in' this
study. EPA technician Kenneth Kropp is noted for his
assistance with data analysis. Mention of specific brand
names or manufacturers should not be construed as an
endorsement of the products or companies by the United
States government.
References
1 K. Kobayashi, E. Fukui, M. Tanaka and S. Kawai,
J. Chromatogr., 1980, 202, 93.
2 Y. Xie and D. A. Reckhow, Ozone Sci. Eng., 1992,14, 269.
3 Y. Xie and D. A. Reckhow, Proc. - Annu. Conf., Am. Water
Works Assoc., 1992, Water Quality, 251.
4 J. F. Garcia-Araya, J. P. Croue, F. J. Beltran and B. Legube,
Ozone Sci. Eng., 1995, 17, 647.
5 C.-Y. K.UO, J. Chromatogr., A, 1998, 804, 265.
6 J. W. Munch, D. J. Munch, S. D. Winslow, S. C. Wendelken and
B. V. Pepich, Method556. Determination ofCarbonyl Compounds
in Drinking Water by Pentafluorobenzylhydroxylamine Derivatiza-
tion and Capillary Gas Chromatography with Electron Capture
Detection, Rev. 1.0, United States Environmental Protection
Agency, Cincinnati, OH, June 1998.
7 J. W. Hodgeson, J. Collins and R. E. Earth, Method 552.
Determination of Haloacetic Acids in Drinking Water by Liquid-
liquid Extraction, Derivatization, and Gas Chromatography with
Electron Capture Detection, United States Environmental Protec-
tion Agency, Cincinnati, OH, July 1990; note that this is not the
most current version (552.2), which uses acidified methanol instead
of diazomethane.
8 E. T. Urbansky and W. J. Bashe, J.. Chromatogr., A, 2000, 867,
143.
9 F. A. Carey and R. J. Sundberg, Advanced Organic Chemistry,
Part B: Reactions and Synthesis, Plenum, New York, 3rd edn.,
1990, p. 15, and references cited therein.
10 J. March, Advanced Organic Chemistry: Reactions, Mechanisms,
and Structure, Wiley, New York, 4th edn., 1992, p. 627-628, and
references cited therein.
338
/ Environ. Monit., 2000, 2, 334-338
-------
------- |