RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 2003; 17: 2364–2369
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rcm.1181
Chromatographic and mass spectral studies
of perfluorooctanesulfonate and three
perfluorooctanesulfonamides{,{
D. W. Kuehl1* and B. Rozynov2
1
US Environmental Protection Agency, Office of Research and Development, National Health and Ecological Effects
Research Laboratory, Mid-Continent Ecology Division, 6201 Congdon Blvd., Duluth, MN 55804, USA
2
Aspen Research Corp., 1700 Buerkle Rd., White Bear Lake, MN 55110, USA
Received 22 April 2003; Revised 3 August 2003; Accepted 3 August 2003
The chromatographic and mass spectral characteristics of perfluorooctanesulfonate (PFOS) and
three nitrogen-substituted perfluorooctanesulfonamides have been obtained. A methyl/phenyl
mixed-phase fused-silica capillary column was used for gas chromatographic (GC) analyses, while
a C18 reversed-phase microbore column was used for liquid chromatographic (LC) analyses. Mass
(MS) and tandem mass (MS/MS) spectra were generated using electron ionization (EI), argon CE,
methane positive and negative ion CI, and ES ionization modes. EI spectra of the amides showed
ions characteristic of both the fluorinated hydrocarbon and the sulfonamide portion of the molecules. The fragmentation pathway was studied using hydrogen/deuterium exchange, and was
thought to involve a cyclic intermediate ion. Formation of molecular ions by CE and protonated
molecule ions by CI to obtain molecular weight information was only partially successful. Negative
ion ES-MS spectra provided intense [M–H] anions for the amides, and an [M–K]anion for PFOS
from which molecular weight information could be obtained, while ES-MS/MS produced product
ions that could be used to detect the presence of these compounds in biological or environmental
samples. Published in 2003 by John Wiley & Sons, Ltd.
The form of a portion of the fluorine detected in human blood
was questioned as being other than inorganic arising from the
use of fluoride in tooth care as early as 1966.1,2 By 1976, the
source of the fluorine was thought to be perfluorooctanoic
acid.3 During the next 20–30 years, these results were generally ignored, while industrial use of fluorinated chemicals
increased greatly.4 Concern for the presence of fluorinated
chemicals in humans and the environment arose again
when, using electrospray ionization mass spectrometry (ESMS), the compound accumulating in humans was correctly
identified as perfluorooctanesulfonate (PFOS), and when it
was found to have reached ppm concentrations in some factory workers.5 This concern was further heightened when it
was announced that some products containing PFOS were to
*Correspondence to: D. W. Kuehl, US Environmental Protection
Agency, Office of Research and Development, National Health
and Ecological Effects Research Laboratory, Mid-Continent
Ecology Division, 6201 Congdon Blvd., Duluth, MN 55804,
USA.
E-mail: [email protected]
{
The information in this document has been funded wholly (or
in part) by the US Environmental Protection Agency. It has
been subjected to review by the National Health and Environmental Effects Research Laboratory and approved for publication. Approval does not signify that the contents reflect the
views of the Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation
for use.
{
This article is a US Government work and is in the public
domain in the USA.
be withdrawn from the market.6,7 A review of the chemistry
of, and reasons for concern for, the accumulation of fluorinated chemicals in humans and the environment was
published shortly thereafter.8 During that same year, a number of environmental monitoring papers were published,
indicating that PFOS was quite widespread across the
earth.9– 12 At this time we conducted a literature search to
obtain information on the chromatographic and mass
spectral characteristics of PFOS and amides of PFOS. The
amides were of interest because, like PFOS, they also were
commercial products and thought to be environmental
degradation precursors of PFOS.13 Interestingly, in spite of
a large number of these types of compounds being produced,
the only manuscript we could find which was similar to
what we were searching for was a study of the fast atom
bombardment (FAB) mass spectra of fluorinated compounds.14 We therefore initiated gas and liquid chromatography mass spectrometry (GC/MS, and LC/MS) and tandem
mass spectrometry (MS/MS) studies on PFOS, perfluorooctanesulfonamide (I), N-ethylperfluorooctanesulfonamide (II),
and N-tert-butylperfluorooctanesulfonamide (III). We wish
to report those results here.
EXPERIMENTAL
Reagents
All reagents were of the highest purity available, and were
used without further purification. Perfluorooctanesulfonate
Published in 2003 by John Wiley & Sons, Ltd.
GC/MS, LC/MS and MS/MS of perfluorooctanesulfonates
2365
Figure 1. Reconstructed total ion current GC/MS chromatogram (left) for PFOSA (I), EtPFOSA (II), and t-Bu-PFOSA (III) (5, 10, and 20 ng, respectively) using a 30 m HP-5MS
(Hewlett-Packard) column. The column was temperature programmed from 608C (5 min)
to 1508C @ 58C/min to 2008C @ 108C/min. The 70 eV electron ionization spectrum for
each compound is shown (right).
(PFOS) was obtained as the potassium salt from Fluka
(Milwaukee, WI, USA). Perfluorooctanesulfonamide (PFOSA; I), N-ethylperfluorooctanesulfonamide (Et-PFOSA; II),
and N-tert-butylperfluorooctanesulfonamide (t-Bu-PFOSA;
III) were obtained from Griffin LLC (Valdosta, GA, USA).
Perfluorododecanoic acid (PFDDA) was obtained from
Lancaster Synthesis (Pelham, NH, USA). Solvents were
from Fisher Scientific (Pittsburgh, PA, USA).
Instrumentation
model 2690 liquid chromatograph fitted with a Phenomenex
(Torrance, CA, USA) Luna C18(2) reversed-phase column
(1 50 mm, 5 m), which was interfaced to a Micromass model
QTOF-I hybrid mass spectrometer through an electrospray
ionization (ES) system. The mass spectrometer was calibrated
and tuned for optimum negative ion ES performance according to the manufacturer’s recommendation. Perfluorohexadecanoic acid was used to assign m/z values. Product ion
spectra were generated using a collision energy of 30 V. The
LC eluant gradient is given in figure captions.
GC/MS
D2O reactions
Gas chromatography/mass spectrometry (GC/MS) experiments were conducted using a Varian (Walnut Creek, CA,
USA) model 3400 gas chromatograph fitted with a HewlettPackard (Avendale, PA, USA) HP-5MS fused-silica column
(30 m), which was interfaced to a Finnigan-MAT model
TSQ-700 tandem quadrupole mass spectrometer. The mass
spectrometer was calibrated and tuned to optimize each
mode of operation according to the manufacturer’s recommendations. Mass ranges and masses for product ion analyses were selected as needed and conducted at 20–35 V
collision energy. The GC temperature program is given in
the caption to Fig. 1.
To 900 mL of an analyte in methanol (1.0 mmol/mL) were
added 100 mL of D2O. The mixture was vortexed and allowed
to warm to 408C for 10 min. Solutions were diluted with
methanol for GC/MS as needed.
LC/MS
Liquid chromatography/mass spectrometry (LC/MS)
experiments were conducted using a Waters (Milford, MA)
Published in 2003 by John Wiley & Sons, Ltd.
RESULTS AND DISCUSSION
Electron ionization
The neutral sulfonamides I, II, and III were initially analyzed
by GC/MS using electron ionization (EI); however, this method of analysis is not applicable to the analysis of an organic
anion such as PFOS. A GC chromatogram and EI spectra
for the three amides are presented in Fig. 1. It was observed
that they were easily separated, having good resolution and
peak shape. During 70 eV ionization, these compounds were
unstable and readily fragmented by rupture of the C–S bond,
Rapid Commun. Mass Spectrom. 2003; 17: 2364–2369
2366
D. W. Kuehl and B. Rozynov
so that their molecular ions (m/z 499, 527, and 555 for I, II, and
III, respectively) were not detected. When the charge
remained with the fluorocarbon portion of these molecules,
two series of ions commonly observed in the EI spectra of perfluorinated hydrocarbons were present. One series started at
m/z 69 [CF3]þ and the other at m/z 131 [C3F5]þ, and each
increased by 50 Da, i.e. m/z 69, 119, 169, 219, and m/z 131,
181, 231, and 281, respectively. The ion at m/z 69 was the second most intense ion in the spectra of I and II, and the third
most intense ion in the spectrum of III.
If, however, the charge remained on the sulfonamide
portion of the molecule after the C–S rupture, ions thought to
be [SO2NHR]þ at m/z 80 (R ¼ H), 108 (R ¼ C2H5), and 136
(R ¼ t-butyl) for I, II, and III, respectively, were formed. Upon
incubation of I with D2O, the intensity of m/z 81 increased
greatly relative to m/z 80, while m/z 82 also increased, but not
as much (D2O experimental data not shown). An increase of 1
or 2 Da suggests that the assignment of the proposed
structure of the ions was likely correct, and that hydrogen/
deuterium exchange occurred during the incubation procedure used here. Similar observations of hydrogen/deuterium exchange were observed for II and III. Further support
for the structure of m/z 108 from II was the presence of m/z 92
(loss of CH3 to [SO2NHCH2]þ), and m/z 80 (loss of C2H4 to
[SO2NH2]þ); however, additional losses from m/z 136 (III)
were not observed, perhaps because m/z 136 was already a
very small ion (<5%).
The third largest ion in the EI spectrum of I was m/z 64
(22%). This ion could either be formed from the rupture of the
S–N bond of m/z 80 [SO2NH2]þ to form [SO2]þ, or it could
have also possibly been formed during fragmentation of a
ring structure ion (m/z 480, 0.5%), which was thought to be
formed after the loss of a fluorine from the molecular ion.15 If
this happened, the structure of m/z 64 could be either [SO2]þ
or [SONH2]þ with an ion also formed at m/z 416 (15%; see
Scheme 1). It was not possible to directly determine the
structure of m/z 64 using the quadrupole mass spectrometer,
because distinguishing the small difference in mass between
the two species, 25 mmass, was beyond the capability of the
instrument. Product ion analysis of m/z 416 produced an ion
at m/z 97; however, two structures, one with N and the other
with O, could again be proposed, [. CF2CFNH2]þ and
[. CF2CFO]þ. An indication that m/z 64 was formed by S–N
rupture from m/z 80 is supplied by its presence in the
spectrum of II, i.e. the loss of [SO2]þ from [SO2NHC2H5]þ (m/z
108). However, an indication that the NH2-containing ion
could also be formed was observed by comparing the
spectrum of II to the II/D2O incubation product, where the
intensity of m/z 417 increased relative to m/z 416 from not
detected to of nearly equal intensity, respectively. Further,
product ion analysis of m/z 417 produced an ion at m/z 98 but
not m/z 97. Both of these observations indicate that hydrogen/deuterium exchange must have been able to occur and
that the fragmentation probably followed both pathways. It
was not possible to determine which carbon atom participated in the ring structure of the ion at m/z 480 of I. Usually,
thermodynamically stable five- and six- membered rings are
formed; however, because of the lack of an ion at m/z 69,
[CF3]þ, in the product ion spectrum of m/z 416, it was thought
that a cyclic intermediate may involve the terminal carbon. A
Published in 2003 by John Wiley & Sons, Ltd.
Scheme 1.
Scheme 2.
more complete study using hydrogen/deuterium exchange
and high resolution mass spectrometry to help determine the
contribution of each mechanism to the formation of m/z 64
and 416 was beyond the scope of this investigation. Further
structural information can be obtained by examining other
ions. For instance, the ions at m/z 526, 512, and 480 in the
spectrum of II each lose 64 Da to m/z 462, 448, and 416,
respectively. Product ion analysis of m/z 512 showed a loss of
64 Da to m/z 448, as did the hydrogen/deuterium exchange
product (m/z 514 to 450), but m/z 419 (additional loss of CH3N
to [C(CH3)3]þ) remained at the same value in both spectra
(see Scheme 2).
The base ion in the spectrum of III was m/z 540, [M–CH3]þ,
and the second most intense ion was m/z 57, [C(CH3)3]þ. Both
of these ions are characteristic of the EI fragmentation of
organic compounds with t-butyl groups attached. Product
ion analysis of m/z 540 showed a loss of 64 Da to m/z 476, and
then an additional loss of 20Da (HF) to m/z 456. Similarly, the
hydrogen/deuterium exchange product lost 64 and then 21
Rapid Commun. Mass Spectrom. 2003; 17: 2364–2369
GC/MS, LC/MS and MS/MS of perfluorooctanesulfonates
Scheme 3.
(DF) Da. Other ions present in the EI spectrum of III included
m/z 448 (m/z 540-[NHOCF(CH3)2]þ), and m/z 416 (m/z 540[SONHCF(CH3)2]þ). The proposed EI and product ion
fragmentation pathways for m/z 540 are shown in Scheme 3.
It is interesting to note that, similar to the product ion
spectrum of m/z 416 for I, the product ion spectra for neither
m/z 512 for II nor m/z 540 for III showed the presence of m/z 69,
[CF3]þ. This may be indirect support for a similar fragmenta-
2367
tion pathway occurring after the loss of SO2 from the
relatively stable cyclic ion formed for each compound.
Because none of the three compounds showed a molecular
ion in the EI spectrum, chemical ionization (CI) of each
analyte was investigated to determine if CI spectra would
provide molecular weight information for each compound. It
was also hoped that an intense, compound- specific ion could
be found for future GC/MS quantification of these analytes in
environmental samples. Initially, argon charge exchange
(CE) spectra were produced. Argon CE spectra are often
similar to EI spectra, but show less fragmentation. It was
hoped, therefore, that molecular weight information along
with some diagnostic ions would be produced. It was
observed, however (Fig. 2), that again the molecular ion
was not detected for any of the compounds, and that the
spectra were nearly identical to the EI spectra with only slight
variation of the relative abundance of each ion. Methane
positive ion chemical ionization was investigated next. The
protonated molecule [M þ H]þ was the base ion for both
PFOSA and Et-PFOSA, with all other ions less than 5%. The
high mass portion of the CI spectrum of t-Bu-PFOSA showed
six ions above 20%, with the [M þ H]þ ion at 22%. Other ions
were observed at m/z 540, 537, 514, 500, and 476, which were
thought to correspond to the losses of CH4, F, C3H6, C4H8, and
SO2NH2. The base ion in the spectrum was m/z 57, the t-butyl
ion. Negative ion methane CI spectra were also investigated.
The base ion for PFOSA was m/z 80, for Et-PFOSA was m/z
108, and for t-Bu-PFOSA was m/z 136 for [SO2NH2],
[SO2NH2H5], and [SO2NHC4H9], respectively. None of
the spectra showed a molecular anion; however, each
showed a weak [M-HF] ion (<5%). Only the Et-PFOSA
spectrum showed any high mass ions greater than 10% in
Figure 2. Argon CE (a), and methane positive ion (b) and negative ion (c) CI-MS spectra for
PFOSA (I), Et-PFOSA (II), and t-Bu-PFOSA (III) are shown. GC conditions were the same as
described in Fig. 1.
Published in 2003 by John Wiley & Sons, Ltd.
Rapid Commun. Mass Spectrom. 2003; 17: 2364–2369
2368
D. W. Kuehl and B. Rozynov
Figure 3. Reconstructed total ion current LC/MS chromatogram for PFOS, PFOSA (I), Et-PFOSA
(II), and t-Bu-PFOSA (III) (2.0 ng each). The column was eluted with 10% B for 1 min to 90% B in
9 min, holding 90% B until 30 min. Mobile phase A was 1% ethanol, 9% methanol, 90% water; B was
1% ethanol, 9% water 90% methanol; each containing 100 mM ammonium acetate. Also shown are
ES mass spectra (solid line), and product ion spectra (dashed line). An ionization voltage of 3.2 kV,
source temperature of 1808C, desolvation gas temperature of 2258C, and cone voltage of 50 V were
used.
abundance. In this case, m/z 483, [M–NHC2H5], was
observed at 34%. Interestingly, none of the spectra showed
any of the fluorine-containing small ion fragments which
were so abundant in the EI spectra. Figure 2 provides the
argon CE and methane positive and negative CI spectra for
PFOSA, Et-PFOSA, and t-Bu-PFOSA.
Although electron ionization and the three chemical
ionization techniques produced spectra that were very useful
for structural characterization, none of the three chemical
ionization techniques were satisfactory as a more universal
technique for routine environmental monitoring of the
fluorinated sulfonamides. Because of this, and because PFOS
could not be analyzed by GC, LC with electrospray ionization
(ES) was investigated next. It was observed that all four of the
compounds could be separated easily, and produced sharp
chromatographic peaks using a reversed-phase C18 column
with a low to high percent methanol in water gradient to elute
the analytes. Negative ion ES produced a very intense anion
at [M–H] for PFOSA, Et-PFOSA and t-Bu-PFOSA, and at
[M–K] for the potassium salt of PFOS, with less than 5%
intensity for fragment ions. A reconstructed total-ion-current
(RTIC) chromatogram for all four compounds is shown in
Fig. 3. As an example of LC/MS performance, Fig. 4 shows a
mass chromatogram of m/z 498.9297 for 10 pg (20 fmol) of
PFOS. Column performance was excellent showing a tailing
factor of 1.0, while detection of 1.0 fmol (10:1 S/N) was easily
achieved. The smaller peak with the same exact mass which
eluted just before PFOS was thought to be a small amount of
Published in 2003 by John Wiley & Sons, Ltd.
Figure 4. LC/MS chromatogram of m/z 498.9297 for 20 fmol
of PFOS (20.0 min). The peak has a width at half-height width
of 0.1 min, and a peak tailing factor of 1.0 was calculated. The
small peak preceding PFOS is thought to be an alkyl
branched PFOS.
Rapid Commun. Mass Spectrom. 2003; 17: 2364–2369
GC/MS, LC/MS and MS/MS of perfluorooctanesulfonates
2369
product ion m/z 169 in rainbow trout (Oncorhynchus mykiss)
blood serum fortified with the PFOS derivatives I, II, and III at
50 ng/mL.
CONCLUSIONS
Figure 5. LC/MS reconstructed total-ion-current (RTIC)
chromatogram of rainbow trout blood serum extract is shown
in the lower trace. The [MH] product ion (m/z 169) mass
chromatogram for fluorinated compounds I, II, and III fortified
at 50 ng/mL into the serum sample is shown in the upper
trace. LC/MS and MS/MS conditions were those previously
described.
A newly recognized class of potential environmental and
human contaminants, highly fluorinated derivatives of alkylsulfonic acid, have been studied to determine their chromatographic mass spectral characteristics. Electron ionization
fragmentation mechanisms were useful for their molecular
structure determination. Charge exchange ionization and
positive and negative ion chemical ionization were not found
to reliably produce molecular weight information. Electrospray ionization in the negative ionization mode was found
to produce intense ions characteristic of the molecular
weight of each molecule studied, and thus could be potentially used to identify and quantify them in environmental
and biological samples. Product ion spectra of the electrospray-generated anion produced ions characteristic of both
the fluorinated tail and sulfonyl portions of PFOS, and the
fluorinated portion of the sulfonamides. Product ions characteristic of the fluorinated tail of the sulfonamides may be
useful as a non-specific method to detect previously unrecognized fluorinated chemicals in environmental and human
samples.
REFERENCES
branched hydrocarbon PFOS contaminating the commercial
product. Other authors have extensively used product ion
analysis of the ES-generated anion for the identification and
quantification of PFOS. 9–12 The product ion spectrum for
PFOS (m/z 499) is also shown in Fig. 3. The product ions were
quite diagnostic of PFOS, and were m/z 80 [SO3] (83%); 99
[SO3F] (81%); 130 [CF2SO3] (100%), with minor ions at m/z
169 [C3F7];180 [C2F4SO3]; and, 230 [C3F6SO3]. Product
ions m/z 80 and/or 99 have been used to confirm the presence
of PFOS in environmental and biological samples 9 –12.
Product ion spectra of [M–H] for I, II, and III showed ions
characteristic of the fluorocarbon tail, but not for the
sulfonamide portion of the molecule. Since the ion at m/z
169, [C3F7] was common in the product ion spectra for all of
the fluorinated sulfonamides (Fig. 3), it may be useful as a
diagnostic marker for many of the fluorinated environmental
contaminants of this chemical class. Figure 5 demonstrates
this possibility by presenting a mass chromatogram for the
Published in 2003 by John Wiley & Sons, Ltd.
1. Taves DR. Nature 1966; 211: 192.
2. Taves DR. Nature 1968; 217: 1050.
3. Guy WS, Taves DR, Brey WS. J. Am. Chem. Soc. Sympos. Ser.
1976; 28: 117.
4. Key BD, Howell RD, Criddle CS. Environ. Sci. Technol. 1997;
31: 2445.
5. Olsen GW, Burris JM, Mandel JH, Zobel LR. J. Occup.
Environ. Med. 1999; 41: 799.
6. Wood A, Clarin W. Chem. Week 2000; 24(31): 9.
7. Tullo AH. Chem. Eng. News 2001; 29: 17.
8. Renner R. Environ. Sci. Technol. 2001; 35: 154A.
9. Hansen KJ, Clemem LA, Ellefson ME, Johnson HO. Environ.
Sci. Technol. 2001; 35: 766.
10. Geisy JP, Kannan K. Environ. Sci. Technol. 2001; 35: 1339.
11. Kannan K, Koistinen J, Beckmen K, Evans T, Gorzelany JF,
Hansen KJ, Jones PD, Helle E, Nyman M, Giesy JP. Environ.
Sci. Technol. 2001; 35: 1593.
12. Kannan K, Franson JC, Bowerman WW, Hansen KJ,
Jones PD, Giesy JP. Environ. Sci. Technol. 2001; 35: 3065.
13. Giesy JP, Kannan K. Environ. Sci. Technol. 2002; 36: 147A.
14. Lyon PA, Tomer KB, Gross ML. Anal. Chem. 1985; 57: 2984.
15. Dynesen E, Lawesson S-O, Schroll G., Bowie JH, Cooks RG.
J. Chem. Soc. (B) Phys. Org. 1968; 15.
Rapid Commun. Mass Spectrom. 2003; 17: 2364–2369