Gas-phase NMR technique for studying the thermolysis of materials: Thermal decomposition of ammonium perfluorooctanoate

Article abstract of DOI:10.1021/ac049667k

The kinetics of the thermal decomposition of ammonium perfluorooctanoate (APFO) has been studied by high-temperature gas-phase nuclear magnetic resonance spectroscopy over the temperature range 196-234 °C. We find that APFO cleanly decomposes by first-order kinetics to give the hydrofluorocarbon 1-H-perfluoroheptane and is completely decomposed (>99%) in a matter of minutes at the upper limit of this temperature range. Based on the temperature dependence of the measured rate constants, we find that the enthalpy and entropy of activation are ΔH? = 150 ± 11 kJ mol^-1 and AS? = 3 ± 23 J mol^-1 deg^-1. These activation parameters may be used to calculate the rate of APFO decomposition at the elevated temperatures (350-400 °C) at which fluoropolymers are processed; for example, at 350 °C the half-life for APFO is estimated to be less than 0.2 s. Our studies provide the fundamental parameters involved in the decomposition of the ammonium salt of perfluorooctanoic acid and indicate the utility of gas-phase NMR for thermolysis studies of a variety of materials that release compounds that are volatile at the temperature of decomposition and that contain an NMR-active nucleus.

Full text of DOI:10.1021/ac049667k

Anal. Chem. 2004, 76, 3800-3803
Ga s -P h a s e NMR Te c h n iq u e fo r S t u d yin g t h e
Th e rm o lys is o f Ma t e ria ls : Th e rm a l De c o m p o s it io n
o f Am m o n iu m P e rflu o ro o c t a n o a t e
Pa ul J . Krus ic a nd D. Chris tophe r Roe *
Central Research and Development, E. I. Du Pont de Nemours & Company, Wilmington, Delaware 19880-0328
6
The kinetics of the thermal decomposition of ammonium
perfluorooctanoate (AP FO) has been studied by high-
temperature gas-phase nuclear magnetic resonance spec-
troscopy over the temperature range 1 9 6 -2 3 4 °C. We
find that AP FO cleanly decomposes by first-order kinetics
to give the hydrofluorocarbon 1 -H-perfluoroheptane and
is completely decomposed (>9 9 %) in a matter of minutes
at the upper limit of this temperature range. Based on
the temperature dependence of the measured rate con-
stants, we find that the enthalpy and entropy of activation
film forming foams and is environmentally significant because
the measurements were conducted after 5 or more years of
inactivity at these sites. PFOS and PFOA have also been detected
and quantified in surface water collected from the Tennessee River
7
near a fluorochemical manufacturing site, and it was suggested
that the effluent from this site was a likely source of organic
fluorochemicals in the river.
Ammonium perfluorooctanoate (APFO) is the carboxylate
surfactant most commonly used as a processing aid in the
production of fluoropolymers and fluoroelastomers. For most
8
q
-1
q
-1
are ∆H ) 1 5 0 ( 1 1 kJ mol and ∆S ) 3 ( 2 3 J mol
commercial products, fluoropolymers are heated in excess of 350
-
1
deg . These activation parameters may be used to
calculate the rate of AP FO decomposition at the elevated
temperatures (3 5 0 -4 0 0 °C) at which fluoropolymers are
processed; for example, at 3 5 0 °C the half-life for AP FO
is estimated to be less than 0 .2 s. Our studies provide
the fundamental parameters involved in the decomposi-
tion of the ammonium salt of perfluorooctanoic acid and
indicate the utility of gas-phase NMR for thermolysis
studies of a variety of materials that release compounds
that are volatile at the temperature of decomposition and
that contain an NMR-active nucleus.
°C, and often above 400 °C, to allow melt processing or sintering
8
of the resin particles. Because of this high-temperature processing
of fluoropolymers, it is relevant to understand the rates of thermal
decomposition of APFO at these elevated temperatures.9
-11
APFO is unique among the salts of PFOA in being the most
thermally unstable.12 This result parallels the work of LaZerte et
al. on the thermal stability of perfluorobutanoate salts.13 The latter
workers also established that ammonium perfluorobutanoate was
unique in providing nearly stoichiometric yields of the hydrofluo-
rocarbon 1-H-perfluoropropane (CF
dium and potassium salts gave high yields of the olefin perfluo-
ropropene (CF CFdCF ), and other salts gave the olefin along
3 2 2
CF CF H), whereas the so-
3
2
Certain perfluorinated surfactant anions such as perfluoro-
octane sulfonate (PFOS) and perfluorooctanoate (PFOA) have
become an environmental concern because they are widely
distributed in trace quantities and persist in the environment.1
Low but measurable amounts of PFOS have been detected in
samples obtained from wildlife in remote marine locations such
as the Arctic and North Pacific Oceans and at higher levels from
more urbanized areas in North America and Europe.2,3 PFOA and
PFOS have been detected and quantified in groundwater con-
taminated by fire-fighting foams at a number of fire-training
areas.4,5 The occurrence of perfluorinated surfactants in fire-
training areas is due to the role of these surfactants in aqueous
with carbonyl-containing compounds. These studies12,13 give a
sense of the thermal stability of the perfluorocarboxylate salts but
say nothing about the kinetics of their thermal decomposition. In
view of the high-temperature processing that is used for fluo-
ropolymers that are made using APFO, we have investigated the
kinetics of thermal decomposition of this surfactant using high-
temperature gas-phase NMR.
Although gas-phase NMR is well established for physical
studies,14 it has been frequently overlooked by chemists as a tool
(
(
6) Moody, C. A.; Field, J. A. Environ. Sci. Technol. 2 0 0 0 , 34, 3864-3870.
7) Hansen, K. J.; Johnson, H. O.; Eldridge, J. S.; Butenhoff, J. L.; Dick, L. A.
Environ. Sci. Technol. 2 0 0 2 , 36, 1681-1685.
(
8) Kissa, E. Fluorinated Surfactants and Repellents, 2nd ed.; Marcel Dekker:
*
Corresponding author. Phone: (302)695-3593. Fax: (302)695-9799. E-mail:
New York, 2001.
chris.roe@usa.dupont.com.
(9) Scheirs, J. Modern Fluoropolymers: High Performance Polymers for Diverse
Applications; Wiley: New York, 1997.
(10) Ebnesajjad, S. Non-Melt Processible Fluoroplastics: The Definitive User’s Guide
and Databook; PDL: New York, 2000.
(11) Ebnesajjad, S. Melt Processible Fluoroplastics: The Definitive User’s Guide
and Databook; PDL: New York, 2003.
(
1) Schultz, M. M.; Barofsky, D. F.; Field, J. A. Environ. Eng. Sci. 2 0 0 3 , 20,
87-501.
4
(
(
2) Giesy, J. P.; Kannan, K. Environ. Sci. Technol. 2 0 0 1 , 35, 1339-1342.
3) Kannan, K.; Koistinen, J.; Beckmen, K.; Evans, T.; Gorzelany, J. F.; Hansen,
K. J.; Jones, P. D.; Helle, E.; Nyman, M.: Giesy, J. P. Environ. Sci. Technol.
2
0 0 1 , 35, 1593-1598.
(12) Lines, D.; Sutcliffe, H. J. Fluorine Chem. 1 9 8 4 , 25, 505-512.
(13) LaZerte, J. D.; Hals, L. J.; Reid, T. S.; Smith, G. H. J. Am. Chem. Soc. 1 9 5 3 ,
75, 4525-4528.
(
(
4) Moody, C. A.; Field, J. A. Environ. Sci. Technol. 1 9 9 9 , 33, 2800-2806.
5) Moody, C. A.; Hebert, G. N.; Strauss, S. H.; Field, J. A. J. Environ. Monit.
2
0 0 3 , 5, 341-345.
(14) LeMaster, C. B. Prog. Nucl. Magn. Reson. Spectrosc. 1 9 9 7 , 31, 119-154.
3800 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004
10.1021/ac049667k CCC: $27.50 © 2004 American Chemical Society
Published on Web 05/11/2004

to study gas-phase reactions. Recent studies have shown that it
is a powerful tool for quantitative kinetic studies of homoge-
neous15,16 and heterogeneously catalyzed chemical processes17,18
in the vapor phase at temperatures up to 400 °C. We have
previously demonstrated that this technique can be applied to a
variety of organofluorine reactions15,16 and have shown that when
compared with other techniques, such as gas chromatography,
NMR has advantages for such studies in that the observed peaks
provide structural information in addition to molar concentrations,
all the volatile organofluorine species may be observed, and the
data are acquired in situ. Furthermore, the automation of modern
NMR spectrometers makes acquisition and processing of kinetic
data very efficient and cost-effective.
relatively uniform excitation over the wide 19F NMR spectral range.
Since 19F spin-rotation relaxation is very efficient in the gas phase,
50 ms recycle delays were appropriate and permitted rapid signal
averaging (e.g., 512 scans in less than 1 min). For the same
reason, the NMR lines are broader in the gas phase than in
solution and magnetic field homogeneity is less critical than for
solution NMR; consequently no field lock was used and the
ampule was not spun. 19F chemical shifts were measured relative
to CF
4
3
at -63.5 ppm on the CFCl (F11) scale.
Thermal losses in the probe necessarily lead to sample
temperatures that are lower than the variable temperature (VT)
controller set point. Temperature calibration was performed with
a thin thermocouple positioned in the center of a dummy ampule
identical to those used for the actual kinetic runs except for a
small hole at the end of the 5 mm stub. The internal temperature
in the ampule depends on the flow rates of the nitrogen gas whose
temperature is being regulated by the VT controller and on that
used for cooling of the peripheral space around the probe dewar.
For the flow rates used in this work, Tdelta (°C) ) 8.39-0.0609 ×
EXPERIMENTAL SECTION
Standards and Reagents. Commercial grade APFO (Fluorad
FC-143, >90% purity) containing small amounts of branched
isomers was obtained from 3M Company. Linear APFO was
prepared by treating linear PFOA (Lancaster Synthesis Inc.) with
ammonium hydroxide in diethyl ether, and the APFO product was
further recrystallized to yield the final sample. APFO was dried
in a vacuum oven and kept in a nitrogen glovebox. 1-H-Perfluo-
roheptane (1-H-pentadecafluoroheptane) was purchased from
Lancaster Synthesis Inc., and tetrafluoromethane was prepared
in our laboratories.
T
set (°C) where Tdelta is the difference between the VT controller
set point Tset and the temperature measured with the thermo-
couple in the center of the dummy ampule. Corrected tempera-
tures are used in this work and were also used in the calculation
of the kinetic activation parameters for APFO.
The gas-phase NMR ampules (sodium borosilicate glass),
made from sections of standard thin wall 10 mm o.d. NMR tubing,
were kept in a vacuum oven before use. In a nitrogen glovebox
the desired milligram quantity of dry solids was introduced into
the ampule through its 5 mm o.d. neck. An O-ring vacuum adapter
was attached to the neck, and the ampule was transferred to a
vacuum line equipped with a precision pressure transducer. After
High-Temperature Gas-P hase Nuclear Magnetic Reso-
nance (NMR) Spectroscopy. The detailed features of the
experimental approach have been presented previously1
5-18
and
are summarized here. The reaction vessel consists of a 10 mm
o.d. glass ampule of about 4 mL internal volume (after sealing)
with a 5 mm o.d. extension to facilitate attachment to a vacuum
system. The length of the ampule was chosen so as to restrict
the sample to the thermostated region of the NMR probe and to
minimize temperature gradients. Micromolar quantities of solids
or liquids to be studied are weighed into the ampule, and any
4
evacuation, micromolar amounts of CF internal standard were
condensed from a bulb of known volume into the ampule by
keeping the latter immersed in liquid nitrogen. After sealing the
neck with a propane torch, the ampule was attached to a sample
holder by means of a short piece of plastic sleeve (Figure S-1).
With the NMR probe at the desired temperature, the tube
assembly was lowered into the probe and the automatic acquisition
of spectra started. Because of the low heat capacity of the loaded
ampule, temperature equilibration took place in about two minutes,
representing a kinetic dead time.
4
gases involved, such as CF which serves as a chemical shift and
mass reference, are transferred via quantitative vacuum line
techniques prior to sealing the ampule with a torch. Kinetic runs
are performed by equilibrating the NMR probe at the desired
temperature, and upon sample insertion, automated data acquisi-
tion is begun according to a schedule compatible with the kinetic
time scale involved. The fastest kinetics which may be observed
(
i.e., the highest temperatures employed) is limited by the time
RESULTS AND DISCUSSION
required for the sample to reach thermal equilibrium, which is
less than two minutes. First-order reactions with half-life times
approaching 1 min are the fastest that can be studied kinetically
by this technique.
As anticipated from the work of Lines and Sutcliffe (see Figure
1 in ref 12), we find that the kinetics of decomposition of APFO
can be followed at temperatures near 200 °C. In our preliminary
kinetic studies, we used the commercial grade APFO (FC-143)
that was widely used industrially for decades. Because of the
electrochemical process used for its manufacture, FC-143 neces-
sarily contains small amounts of branched isomers. Typical
samples consisted of 67 mg of FC-143 (156 µmol) along with 79
The 19F NMR spectra were obtained using a 300 MHz wide-
1
9
bore Oxford magnet ( F at 282.538 MHz) and a GE NMR
Instruments Omega console. The high-temperature 10 mm probe
and variable temperature (VT) controller have an upper temper-
ature rating of 400 °C and were purchased from Nalorac Corpora-
tion. A small flip-angle pulse was used (≈15°) in order to achieve
µmol of the CF
4
internal reference. At temperatures below 190
°C, the only gas-phase NMR signal observed is that of the CF
4
internal reference since the ammonium salt has insufficient vapor
pressure for NMR detection at this temperature. However, at 196
°C new NMR lines grow in (Figure 1A), where the major product
approximately 90-95%) is linear 1-H-perfluoroheptane, as antici-
pated by analogy with the product of ammonium perfluorobu-
tanoate decomposition.13 The identity of this product may be
(
(
15) Krusic, P. J.; Roe, D. C.; Smart, B. E. Isr. J. Chem. 1 9 9 9 , 39, 117-123.
16) Shtarov, A. B.; Krusic, P. J.; Smart, B. E.; Dolbier, W. R., Jr. J. Am. Chem.
Soc. 2 0 0 1 , 123, 9956-9962.
(
(
17) Kating, P. M.; Krusic, P. J.; Roe, D. C.; Smart, B. E. J. Am. Chem. Soc. 1 9 9 6 ,
(
1
18, 10000-10001.
18) Roe, D. C.; Kating, P. M.; Krusic, P. J.; Smart, B. E. Top. Catal. 1 9 9 8 , 5,
33-147.
1
Analytical Chemistry, Vol. 76, No. 13, July 1, 2004 3801

Fig u re 2 . First-order kinetic plot for the production of 1-H-perfluo-
roheptane corresponding to the loss of APFO at 196 °C (circle), 206
°
C (triangle up), 215 °C (diamond), 224 °C (square), and 234 °C
(triangle down).
assigned to 1-H-5-trifluoromethylperfluorohexane. The occurrence
Fig u re 1 . (A) Gas-phase 19F NMR spectrum of 1-H-perfluorohep-
of other methine CF resonances (-182.0, -183.7, and -186.8 ppm,
tane (and CF4 reference) evolved in the thermolysis of commercial
APFO (FC-143) at 196 °C. The more intense CF3 and CF2 resonances
are from the linear 1-H-perfluoroheptane, and the characteristic
3
data not shown) and CF resonances (-69.5, -79.1, and -80.2
ppm, Figure 1A) is consistent with the presence of additional
branched isomers, as is the observation that the methine and
-
CF2H resonance is displayed in the right expansion. Minor NMR
3
branched CF peaks are completely absent in the sample of linear
lines belong to the branched isomers of 1-H-perfluoroheptane. (B)
Gas-phase 19F NMR spectrum of authentic linear 1-H-perfluorohep-
APFO derived from linear PFOA.
tane at 196 °C.
3
The rate of appearance of the distinct CF signals belonging
to the branched isomers of 1-H-perfluoroheptane occurs by first-
order kinetics and is in each case within ≈10% of the rate of
appearance of the fluorine resonances belonging to linear 1-H-
perfluoroheptane (Figure S-2). Integration over all the 1-H-
perfluoroheptane fluorine resonances after 168 min at 196 °C, and
2
judged by the characteristic doublet of the terminal -CF H
fluorine resonance at -135.8 ppm due to the 2-bond 1H-19F
coupling constant of 52.3 Hz (Figure 1A, right inset) and by
comparison with the gas-phase NMR spectrum for authentic linear
4
comparison with the integral for the CF mass standard, indicates
1
1
-H-perfluoroheptane obtained under similar conditions (Figure
B).
that 88% conversion of the starting APFO has taken place. Since
the formation of linear and branched 1-H-perfluoroheptane is not
yet complete at this point in the kinetics (Figure S-2), we conclude
Also visible in the spectrum of Figure 1A are some small peaks
that are attributed to the branched isomers of 1-H-perfluorohep-
tane. No peaks other than those attributed to the linear and
(
i) that the thermal decomposition of FC-143 goes to completion
with the formation of linear 1-H-perfluoroheptane, CO and NH
eq 1) together with minor amounts of branched isomers and (ii)
2
3
branched isomers of 1-H-perfluoroheptane (and the CF
standard) are observed. The most significant of these minor peaks
is the CF resonance at -71.1 ppm which may reasonably be
assigned to the terminal (magnetically equivalent) CF ’s of 1-H-
CF H) by
chemical shift reported for this com-
4
internal
(
that its rate of decomposition can be determined by following the
rate of formation of 1-H-perfluoroheptane.
3
3
5
-trifluoromethylperfluorohexane ((CF
3
)
2
CFCF
2
CF
2
CF
2
2
k1
comparison with the CF
3
CF (CF ) CF CO NH 98 CF (CF ) CF H + CO + NH
3 2 5 2 2 4 3 2 5 2 2
3
pound in deuterioacetone solvent19 using trifluoroacetic acid (TFA)
(
1)
as an external chemical shift reference. If the chemical shift of
TFA relative to F11 is taken to be -78.5 ppm,20 the CF
chemical
3
shift for 1-H-5-trifluoromethylperfluorohexane in deuterioacetone
solvent may be estimated as -74.5 ppm, in fair agreement with
our measurements relative to F11. Exact agreement cannot be
expected considering the changes in NMR chemical shifts in going
from liquid phase to vapor phase. Four methine CF resonances
are observed in the region -182 to -188 ppm (data not shown),
and of these four, the most intense signal at -182.9 ppm is
The kinetics of the thermal decomposition of pure linear APFO
(68 mg) was subsequently studied in the temperature range from
196 °C to 234 °C by following the formation of linear 1-H-
perfluoroheptane as described above. The formation of the latter,
and consequently the disappearance of APFO, follows rigorously
first-order kinetics as indicated by the linearity of the curves in
the logarithmic plot shown in Figure 2. The rate constants k at
1
each temperature were obtained by directly fitting the single-
exponential growth of 1-H-perfluoroheptane and are presented in
(
(
19) Yu Huai-Gen Youji Huaxue (Organic Chemistry) 1 9 9 0 , 10, 517-520.
20) Berger, S.; Braun, S.; Kalinowski, H.-O. NMR Spectroscopy of the Non-Metallic
Elements; John Wiley & Sons: New York, 1997.
Table 1 together with the calculated half-lives (t1/ 2 ) ln(2)/ k )
1
3802 Analytical Chemistry, Vol. 76, No. 13, July 1, 2004

In accord with the first-order nature of the decomposition, similar
results were obtained with 32 mg of APFO, and the activation
parameters obtained from the latter experiments are essentially
identical to those just presented.
Ta b le 1 . Ra t e Co n s t a n t s fo r De c a y (a n d Ha lf-Live s ) o f
Lin e a r AP FO a s a Fu n c t io n o f Te m p e ra t u re a s
Me a s u re d b y t h e Fo rm a t io n o f 1 -H-P e rflu o ro h e p t a n e
k1 (min-1)a
t1/ 2 (min)a
temp (°C)
The activation parameters obtained above may be used in the
Eyring equation to calculate the rate of APFO decomposition at
temperatures where the rate would be too fast to measure by gas-
phase NMR. For example, at the elevated temperatures specified
to process fluoropolymers, the half-lives for APFO are extrapolated
to be 0.14 (+0.11, -0.06) s at 350 °C and 0.015 (+0.017, -0.008)
s at 400 °C. At a given temperature, the extrapolated half-life is
associated with uneven 95% confidence limits (given in parenthe-
ses) because of the exponential temperature dependence of the
rate constant.
1
2
2
2
2
96
06
15
24
34
0.0158 (0.0002)
0.0388 (0.0007)
0.0787 (0.0020)
0.152 (0.0028)
0.312 (0.0101)
43.8 (0.66)
17.8 (0.32)
8.8 (0.22)
4.6 (0.08)
2.2 (0.07)
a
Numbers in parentheses refer to (95% confidence limits.
In an attempt to simulate the effect of steam on the thermolysis
of APFO, we also studied the kinetics of thermal decomposition
of linear APFO (32 mg) in the presence of a 10-fold molar excess
of water (13 µL), after ensuring that the glass ampules withstood
the increased internal pressure at the temperatures needed for
decomposition. No hydrolysis of APFO to perfluorooctanoic acid
was observed in the vapor phase, and the observed rates were
not affected strongly by the excess of water. The kinetic data are
q
well represented by the Eyring parameters ∆H ) 167 ( 6 kJ
-
1
q
-1
-1
-1
mol and ∆S ) 38 ( 12 J mol deg (E
a
) 172 ( 6 kJ mol
and log(A/ s- ) ) 15.4 ( 0.6). The net effect of these changes is
to make the rates of decomposition of APFO in the presence of
steam slightly higher above 200 °C and slightly lower below 200
1
°
C compared to APFO in the absence of steam. Under these
Fig u re 3 . Eyring plot for the temperature dependence of the first-
conditions, the half-lives and 95% confidence limits for APFO are
order rate constants for the decomposition of APFO.
extrapolated to be 0.060 (+0.020, -0.015) s at 350 °C and 0.005
(
+0.002, -0.001) s at 400 °C.
1 1
0.693/ k ). The value for k obtained at 196 °C from the linear APFO
component in the commercial FC-143 sample described above was
within 11% of that reported in Table 1 for the linear APFO derived
from linear PFOA.
CONCLUSIONS
The experiments described above demonstrate that ammonium
perfluorooctanoate decomposes rapidly at the elevated tempera-
tures which are most often used to melt process or sinter
fluoropolymers. Our studies provide the fundamental parameters
involved in the decomposition of APFO and illustrate the utility
of gas-phase NMR for studying the thermal behavior of fluoro-
chemicals.23
The data in Table 1 are well described by the transition state
(
Eyring) treatment of the temperature dependence of reaction rate
constants21 as judged by the linearity of the corresponding Eyring
q
q
plot in Figure 3. The activation enthalpy (∆H ) and entropy (∆S )
were determined by direct nonlinear least-squares fitting22 of these
parameters to the Eyring equation (k
exp(-∆H / RT)) using the temperature and rate constant data in
q
1
) (k
B
T/ h) exp(∆S / R)
q
ACKNOWLEDGMENT
We are grateful to Dr. Ming H. Hung for supplying a sample
-1
Table 1 (k
K); k is Boltzmann’s constant, h is Planck’s constant, and R is
the universal gas constant). This fitting procedure gave the
1
converted to units of s and T to absolute temperature
of linear ammonium perfluorooctanoate and for useful advice, to
Dr. Bruce E. Smart for insightful discussions, to Alexander A.
Marchione and Rita S. McMinn for valuable technical assistance,
and to Steve A. Hill for skillful and dedicated experimental work.
(
B
q
-1
q
parameter estimates ∆H ) 150 ( 11 kJ mol and ∆S ) 3 ( 23
J mol deg (the corresponding Arrhenius parameters are E
1
-
1
-1
a
)
54 ( 11 kJ mol- and log(A/ s-1) ) 13.6 ( 1.2). The stated
1
SUPPORTING INFORMATION AVAILABLE
uncertainties represent 95% confidence limits and were obtained
using an estimated error in the rate constants of 10% and an
estimated error in temperature of 2 °C. The major sources of error
in these parameter estimates are the uncertainty in the temper-
ature and the relatively narrow temperature range in which the
rate constants can be measured; the latter limitation in particular
contributes to the relatively large error in the entropy of activation.
Diagram of the gas-phase NMR ampule attached to a vacuum
system before being sealed off with a propane torch (Figure S-1)
and kinetics of formation of both linear and branched 1-H-
perfluoroheptane from the thermolysis of commercial APFO
Figure S-2). This material is available free of charge via the
Internet at http:/ / pubs.acs.org.
(
(
(
(
21) Marshall, A. G. Biophysical Chemistry; John Wiley & Sons: New York, 1978.
22) Conway, G. R.; Glass, N. R.; Wilcox, J. C. Ecology 1 9 7 0 , 51, 503-507.
23) Ellis, D. A.; Martin, J. W.; Muir, D. C. G.; Mabury, S. A. The Analyst 2 0 0 3 ,
Received for review March 2, 2004. Accepted April 1,
2004.
1
28, 756-764.
AC049667K
Analytical Chemistry, Vol. 76, No. 13, July 1, 2004 3803