Chemical stability and application of a fluorophilic tetraalkylphosphonium salt in fluorous membrane anion-selective electrodes

Article abstract of DOI:10.1039/b9nj00696f

The highly fluorophilic phosphonium salt (R_f8(CH ₂)₂)(R_f6(CH₂)₂) ₃P⁺I^-, 1, was used to provide cationic sites for anion-selective electrodes based on fluorous sensing membranes. The electrodes exhibited the theoretically expected "Nernstian" response slopes, and their selectivities for different anions were determined. For environmental analysis, the selective detection of perfluorooctanesulfonate and perfluorooctanoate is of particular interest. While previously reported electrodes based on a fluorophilic methyltriarylphosphonium salt suffered from OH^- interference to the extent that OH^- selectivities could not be determined, and general base catalysis caused decomposition of the phosphonium sites in the presence of weakly basic anions, electrode membranes based on 1 are much more robust. A loss of sensor response is only observed when the fluorous membranes doped with 1 are exposed to 0.1 M hydroxide solutions for 24 h. NMR and mass spectrometry show that the fluorophilic tetraalkylphosphonium cation of 1 decomposes under these extreme conditions to give two trialkylphosphine oxides and perfluoroalkylethanes. Interestingly, this decomposition is much slower than the base catalyzed exchange of the hydrogens in α position to the phosphonium center, which in the presence of D ₂O results in the quantitative formation of the octadeuterated tetraalkylphosphonium cation. While formation of a pentacoordinated transition state in the decomposition of 1 is likely, no pentavalent complexes of the phosphonium ion with OH^- could be observed by NMR spectroscopy.

Full text of DOI:10.1039/b9nj00696f

View Article Online / Journal Homepage / Table of Contents for this issue
PAPER
www.rsc.org/njc | New Journal of Chemistry
Chemical stability and application of a fluorophilic
tetraalkylphosphonium salt in fluorous membrane
anion-selective electrodes
Li D. Chen,a Debaprasad Mandal,b John A. Gladyszb and Philippe Buhlmann*a
¨
Received (in Gainesville, FL, USA) 21st November 2009, Accepted 7th April 2010
DOI: 10.1039/b9nj00696f
The highly fluorophilic phosphonium salt (Rf8(CH2)2)(Rf6(CH2)2)3P+IÀ, 1, was used to provide
cationic sites for anion-selective electrodes based on fluorous sensing membranes. The electrodes
exhibited the theoretically expected ‘‘Nernstian’’ response slopes, and their selectivities for
different anions were determined. For environmental analysis, the selective detection of
perfluorooctanesulfonate and perfluorooctanoate is of particular interest. While previously
reported electrodes based on a fluorophilic methyltriarylphosphonium salt suffered from OHÀ
interference to the extent that OHÀ selectivities could not be determined, and general base
catalysis caused decomposition of the phosphonium sites in the presence of weakly basic anions,
electrode membranes based on 1 are much more robust. A loss of sensor response is only
observed when the fluorous membranes doped with 1 are exposed to 0.1 M hydroxide solutions
for 24 h. NMR and mass spectrometry show that the fluorophilic tetraalkylphosphonium cation
of 1 decomposes under these extreme conditions to give two trialkylphosphine oxides and
perfluoroalkylethanes. Interestingly, this decomposition is much slower than the base catalyzed
exchange of the hydrogens in a position to the phosphonium center, which in the presence of
D2O results in the quantitative formation of the octadeuterated tetraalkylphosphonium cation.
While formation of a pentacoordinated transition state in the decomposition of 1 is likely, no
pentavalent complexes of the phosphonium ion with OHÀ could be observed by NMR
spectroscopy.
developed recently. The extraction of lipids and hydrophobic
Introduction
proteins into the hydrophobic membranes of conventional
ISEs causes signal drifts and selectivity losses.17,18 Therefore,
the poor solubility of these lipophilic compounds in fluorous
sensing membranes is of considerable interest for measurements
in real-life samples such as blood or urine. Moreover, due to
the extremely poor solvation of interfering ions, fluorous
membrane ISEs have been found to exhibit greater selectivity.3,4,19
The selectivity of these ISE membranes arises from similar
equilibria between two mutually immiscible phases as they are
also observed in solvent extraction, chromatography, and
biphase catalysis. To achieve selective response to either
cations or anions, ISE membranes contain ionic functional
groups (so-called ionic sites) that are either covalently attached
to a polymer backbone or are part of a very hydrophobic salt
that readily dissolves in the hydrophobic sensing membrane.15
When a hydrophobic ISE membrane is put into contact with
an aqueous sample containing ions with a charge sign opposite
to that of the ionic sites, those sample ions can enter the ISE
membrane and create an electrical potential difference across
the sample/membrane interface. The measured potential
increases with the logarithm of the ion activity in the
sample. The selectivity of the ISE membrane for different
ions of the same charge sign is determined by the free energy
of transfer from the aqueous sample into the sensing
membrane. Therefore, ion-selective electrodes typically respond
much more strongly to weakly hydrated ions than to very
hydrophilic ions.
Because of the extremely low polarizability of the C–F bond
and the resulting weak van der Waals forces, perfluorocarbons
are among the least polar and polarizable condensed phases
known.1 This is illustrated, for example, by the value of À0.41
for perfluorooctane on the p* scale of solvent polarity,
where dimethyl sulfoxide and cyclohexane define 1 and 0,
respectively.2 Perfluorinated amines and perfluoroethers
exhibit an equally low polarity since their numerous strongly
electron-withdrawing fluorines efficiently decrease the electron
density at their nitrogen and oxygen atoms, nearly eliminating
any complexing capabilities.3–6 To emphasize their unique
properties, perfluorinated organic compounds that are not
freely miscible with hydrocarbons have been referred to as
being fluorous.7 A variety of such compounds have been
developed and are used in many application fields, such as
the electronics industry, aerospace engineering, medical
sciences, the cosmetic industry and phase transfer catalysis.1,8–12
Besides their immiscibility with typical common oils, fluorous
compounds benefit from a very high chemical inertness. These
properties are also crucial for ion-selective electrodes
(ISEs)13–16 with fluorous sensing membranes, which have been
a University of Minnesota, Department of Chemistry,
207 Pleasant St. SE, Minneapolis, Minnesota 55455, USA.
E-mail: buhlmann@umn.edu
b Texas A&M University, Department of Chemistry, PO Box 30012,
College Station, Texas 77842, USA
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010 New J. Chem., 2010, 34, 1867–1874 | 1867

View Article Online
1
The first reported ISEs with fluorous membranes incorporated
the fluorophilic tetrakis[3,5-bis(perfluorohexyl)phenyl]borate
as anionic site, providing for cation selectivity.3,4 More
recently, the selectivity of such cation-exchanger ISEs has
been modified by adding into the sensing membranes not only
ionic sites but also a fluorophilic ionophore that selectively
forms a complex with the target cation. This permitted the
fabrication of highly selective pH electrodes.19
spectrometry and 31P and H NMR spectroscopy were used
to elucidate the mechanism of the decomposition of 1 when
exposed to aqueous solutions of high pH.
To extend the fluorous sensor system to anion sensing, a
fluorophilic cationic site was needed. However, since fluorous
phases have such an extremely low polarity and polarizability,
they are inherently poor solvents for ionic compounds.
Therefore, it is not surprising that only very few such
compounds have been described in the literature.8 Importantly,
no suitable ammonium salt is currently available for applications
in ISEs. Several fluorophilic primary and secondary ammonium
salts have been reported,8,20 but the possibility of deprotonation
Experimental
+
+
of their NH3 and NH2 groups renders them inept as
cationic sites. A remarkable bicyclic compound with a
quaternary ammonium group at the spiro bridgehead
proved to be very effective as phase transfer catalyst and
would be very promising for sensing applications if its
synthesis were not based on 11 steps, limiting its availability.21
Finally, while the tris[(perfluorooctyl)propyl]methylammonium
cation has been shown to solubilize polyoxymetalates
very efficiently in perfluorohydrocarbons,22 the salts of this
cation with small singly charged anions such as chloride were
found to be poorly soluble in perfluorocarbons such as
perfluoroperhydrophenanthrene.23
Reagents and materials
(Rf8(CH2)2)(Rf6(CH2)2)3P+IÀ, 1, was synthesized by reaction
of
a fluorous tertiary phosphine, (Rf6(CH2)2)3P, with
Rf8(CH2)2I in DMF at 120 1C in a N2 atmosphere, as reported
previously.9 Perfluoroperhydrophenanthrene, 3, and the linear
perfluoropolyether
a-(heptafluoropropyl)-o-(pentafluoro-
ethoxy)-poly[oxy(1,1,2,2,3,3-hexafluoro-1,3-propanediyl)], 4,
were obtained from Alfa Aesar (Ward Hill, MA, USA).
Fluoropore membrane filters (pure polytetrafluoroethylene
(PTFE), 47 mm diameter, 0.45 mm pore size, 50 mm thick,
85% porosity) were purchased from Millipore (Bedford, MA,
USA). The potassium and sodium salts were obtained from
Mallinckrodt Baker (Paris, KY, USA). All sample solutions
were prepared using deionized and charcoal-treated water
(0.182 MO cm resistance) purified in a Milli-Q PLUS reagent-
grade water system (Millipore). Perfluorohexanes (FC-72) was
purchased from 3M (St. Paul, MN, USA) and a,a,a-trifluoro-
toluene from Oakwood products (West Columbia, SC, USA).
An interesting alternative to fluorophilic ammonium ions
are fluorophilic phosphonium cations. Fluorophilic tetraalkyl-
phosphonium and trialkylphenylphosphonium salts as well
as their application in phase transfer catalysis have been
reported.8,9,24–27 The first fluorophilic phosphonium salt used
to prepare fluorous sensing membranes for a potentiometric
anion sensor was tris{3,5-bis[(perfluorooctyl)-propyl]phenyl}-
methylphosphonium methyl sulfate, 2. The sensor membranes
based on this methyltriarylphosphonium salt exhibited the
theoretically expected (Nernstian) response to several anions
such as NO3À, IÀ, and BrÀ, and an excellent selectivity for
perfluoroalkyl-carboxylates and -sulfonates. However, notable
exceptions were observed for the basic anions FÀ, OAcÀ,
NO2À, and HCO3À, for which much slower responses and
long-term drifts occurred. This could be explained by the
decomposition of the phosphonium salt in the presence of
water according to a general base mechanism, resulting in the
elimination of one aryl group and formation of phosphine
oxide.23 For measurements in alkaline media and for the
determination of basic anions, alternative fluorophilic sites
were needed.
Electrodes
Membrane supports were prepared from Fluoropore filters
sandwiched between two note cards and cut to a 13 mm
diameter with a hole punch. Supported liquid membranes
were obtained by addition of approximately 50 mL of
In the work reported here, we used the fluorophilic tetra-
alkylphosphonium salt (Rf8(CH2)2)(Rf6(CH2)2)3P+IÀ, 1, to
provide cationic sites for ionophore-free anion-selective
electrodes. The response and selectivity of the resulting sensors
was determined in potentiometric experiments. While the
resulting sensor membranes were chemically much more
robust than the earlier reported ones based on the methyl-
triarylphosphonium derivative 2, a loss of response was
observed upon exposure to very alkaline solutions. Mass
(Rf8(CH2)2)(Rf6(CH2)2)3P+IÀ,
1 (2 mM), in perfluoro-
perhydrophenanthrene, 3, onto a stack of two porous filter
supports until the supports changed from opaque white to
translucent with a glossy surface. These supported liquid
membranes were then mounted into custom-machined
poly(chlorotrifluoroethylene) electrode bodies and sealed
with screw caps with a 8.3 mm diameter hole exposing the
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
1868 | New J. Chem., 2010, 34, 1867–1874

View Article Online
membranes to the sample solutions.4 The solution inside each
electrode was typically 1 mM KCl in contact with a AgCl-
coated Ag wire. However, for the measurements of the
selectivities of the perfluorinated organic anions, 0.1 mM
solutions of sodium perfluorooctanoate were used as inner
filling solutions. The electrodes were conditioned in
solutions of the target ion (primary ion) for 2–3 h prior to
measurements.
perfluoropolyether, 4, for the plasticization of the perfluoro-
polymer Teflon AF240034 and its use in cation-selective
electrodes is very successful and permitted long-term stabilities
of at least one month.5 Therefore, we first investigated the
possibility of preparing fluorous anion-selective membranes by
imbibing a porous inert Teflon filter with solutions of the
fluorophilic phosphonium salt, 1, in the linear perfluoropoly-
ether. Unfortunately, it was found that 1 is sparingly soluble in
the linear perfluoropolyether. The solubility was lower
than 0.25 mM. Moreover, the electrical resistance of the
approximately 50 mm thick membranes (diameter E 8.3 mm)
was 400 GO. This extremely high value results not only from
the low solubility but also from a high level of ion pairing,
which is typical for fluorous solvents.4,35 Since this high
resistance made potentiometric measurements difficult, all
subsequent experiments were performed with solutions of
the fluorophilic salt, 1, in perfluoroperhydrophenanthrene, 3,
as the fluorous sensing phase, supported again by an inert
porous Teflon filter.
Potentiometric measurements
The electrodes made in the way described above were used in
combination with a double-junction type external reference
electrode (DX200, Mettler Toledo, Switzerland; 3.0 M KCl
saturated with AgCl as inner-filling solution and 1.0 M LiOAc
as bridge electrolyte):
Ag|AgCl|3.0 M KCl, AgCl sat.||1.0 M LiOAc||sample|fluorous
membrane|inner filling solution|AgCl|Ag
Potentials were monitored with an EMF 16 potentiometer
(input impedance 10 TO) controlled with EMF Suite 1.03
software (Lawson Labs, Malvern, PA, USA). DC resistances
of the fabricated electrodes were measured by comparing
measured potentials with and without a known shunt.28,29
Calibration curves were obtained by measuring the EMF while
successively diluting a concentrated solution. Nernstian responses
were confirmed for all anions of interest in the concentration
range where selectivities were measured. Selectivity coefficients
were determined with the separate solution and fixed interference
methods30,31 and averaged for three electrodes, giving typical
deviations of Æ0.3. All EMF values were corrected with the
Henderson equation for liquid-junction potentials.32 Activity
coefficients were calculated according to a two-parameter
The solubility of 1 in perfluoroperhydrophenanthrene, 3,
was significantly higher than in the linear perfluoropolyether 4,
so that 2.0 mM solutions could be prepared without difficulties.
The resistance of the thus prepared fluorous anion-exchanger
membranes were measured to be 5.6 Æ 0.3 GO, which is
sufficiently low to permit potentiometric measurements
without complications. Interestingly, this resistance was much
lower than that of the fluorous anion selective membranes with
the same concentration of the methyltriarylphosphonium salt
2 (2 mM) in the same fluorous solvent. Indeed, sensing
membranes containing the latter could only be used when a
fluorophilic electrolyte salt was added to lower the electrical
resistance.23 Without a more thorough study of ion pair
formation in fluorous solutions of 1 and 2, the reasons
for this big difference cannot be identified with certainty.
However, charge delocalization in the methyltriarylphosphonium
unit would be expected to disfavor ion pair formation, which is
just the opposite of the observed trend. Therefore, it appears
likely that steric effects dominate, favoring ion pairs of the
sterically only poorly shielded methyltriarylphosphonium ion
2. Evidently, the tendency to form electrically neutral ion
aggregates can limit the usefulness of a fluorophilic salt just
as much as a limited solubility.
Debye–Huckel approximation.33
¨
Spectroscopy experiments
1H NMR spectra were recorded on a Varian Inova 500
spectrometer (Varian, Palo Alto, CA, USA) with D2O in a
capillary for locking and referencing purposes. 31P NMR
spectra were recorded on a Varian Inova 300 spectrometer
with 5% H3PO4/95% D2O (wt/wt) in a capillary for locking
and referencing. All spectra were recorded with a solvent
mixture of perfluoroperhydrophenanthrene and perfluorohexanes
in a volume ratio of 1 : 4.
For potentiometric measurements of chloride, electrodes
with fluorous sensing membrane containing 1 (2 mM) in
perfluoroperhydrophenanthrene, 3, were equilibrated prior
to their first use with an aqueous 0.1 M KCl solution for
2 h. This so-called conditioning removes the iodide originally
introduced into the membranes with 1 by exchange with the
chloride from the aqueous solution. Once conditioned, the
electrodes respond within seconds to different concentrations
of chloride. Calibration curves (Fig. 1) show that the electrodes
exhibited the expected theoretical (‘‘Nernstian’’) response with
a slope of À59.2 mV per ten-fold increase in the activity of
ClÀ. The detection limit was reached at 10À5.6 M, where the
potential remains unchanged upon further decreasing the ClÀ
activity.
Mass spectra were recorded on a reflectron ESI-TOF spectro-
meter (BioTOF II, Bruker Daltonics, Billerica, MA, USA).
For good sample solubility and ionization, a mixture of
methanol and a,a,a-trifluorotoluene with a volume ratio of
1 : 1 was used as the solvent.
Results and discussion
Response and selectivity of fluorous membrane ISE membranes
To prepare an ISE membrane, the salt providing the ionic sites
is dissolved in a hydrophobic inert matrix. The latter may
be a liquid supported by a highly porous inert structure
or a polymer that is typically blended with a plasticizer
to increase ion mobilities and thereby lower the electrical
resistance. As previously reported, the use of the linear
The selectivity is undoubtedly the most important property
of an ISE since it determines the reliability of the measurement
of the target ion I in the presence of interfering ions, J. It is
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010 New J. Chem., 2010, 34, 1867–1874 | 1869

View Article Online
chloride) (PVC) membranes plasticized by 2-nitrophenyl octyl
ether (o-NPOE) or chloroparaffin. Since both 1 and 2 are
poorly soluble in o-NPOE or chloroparaffin, tridodecylmethyl-
ammonium chloride was used in the PVC-based membranes as
the anion exchanger site instead of 1 or 2. For most ions, the
selectivities of the new fluorous membranes based on cationic
site 1 are similar to those of non-fluorous membranes.
As expected for ion-exchanger membranes without added
ionophore, the selectivities follow for most investigated ions
the so-called Hofmeister series, i.e., IÀ > NO3À > BrÀ > ClÀ
>
FÀ > CH3COOÀ, a sequence that is dominated by the
hydration energies of these ions in the aqueous phase. Unlike
in the case of cation-selective electrodes,3 the range of selectivities
for these common ions was not much wider for the fluorous
than for the non-fluorous membranes. However, as similarly
found for membranes based on the less stable cationic site 2,
the membranes with 1 exhibited very high selectivities for
the perfluorinated anions perfluorooctanoate (PFOÀ) and
perfluorooctanesulfonate (PFOSÀ). While the PFOSÀ selectivities
of the PVC membranes could not be determined because of the
limited solubility of PFOSÀ in water and the high detection
limit of the PVC membranes, the PFOÀ selectivities of the two
fluorous types of membranes were found to exceed those of the
PVC/chloroparaffin membranes by approximately 1.5 orders
of magnitude. Among all the investigated ions, only hydroxide
exhibited a selectivity very different from what is expected
based on the Hofmeister series. In comparison to the PVC-based
ISEs, the fluorous membrane ISE based on cationic site 1
showed an increase in hydroxide response by approximately
four orders of magnitude. The selectivities for the other anions
did not differ much from those of the PVC-based ISEs, but a
somewhat increased response to the weakly basic anions FÀ
Fig. 1 Potentiometric ClÀ response of an ISE based on a liquid
membrane containing 1 (2 mM) in perfluoroperhydrophenanthrene, 3.
pot
commonly quantified by the selectivity coefficient, K , which
I,J
for ions of the same charge equals the ratio of the activities of
ion I and ion J that produce the same potential in separate
measurements.30,31 Fig. 2 shows the selectivity coefficients for
the fluorous ISE membranes containing cationic site 1 or 2
along with corresponding values for conventional poly(vinyl
and HCO3 and a slightly higher discrimination of IÀ, BrÀ,
À
and NO3À were observed. Detection limits for different anions
of the membranes with site 1 fell in the range of 10À4 to
10À8 M and followed the same pattern as its selectivities. For
example, in the course of the determination of selectivity
coefficients with the separate solution method, detection limits
of 10À3, 10À5, and 10À5 M were observed for the highly
discriminated ions OAcÀ, FÀ, and NO2À, respectively. On
the other hand, preliminary experiments with these perfluoro-
perhydrophenanthrene-based membranes showed detection
limits lower than 10À7 M and 10À8 M for PFOÀ and PFOSÀ,
respectively. Experiments to exploit the high selectivity for
perfluoroalkyl anions in mechanically more robust perfluoro-
polymer membranes with 3-dimensionally ordered nanoporous
carbon (3DOM C)36,37 as solid contacts are in progress in our
laboratories. The sensitivities for all these ions lie in the range
of À58 Æ 2 mV per tenfold increase in anion activity, and the
response times of these electrodes are less than 10 s.
Unlike in the case of the fluorophilic methyltriarylphosphonium
cation 2, ISE membranes based on the cationic site 1 did not
exhibit short-term drifts in the responses to hydroxide and
basic anions. As reported previously, exposure of fluorous
solutions of the methyltriarylphosphonium salt 2 to OHÀ, FÀ,
Fig. 2 Logarithmic representation of the selectivities of two fluorous
anion-exchanger membranes (a: 2 mM 1 in 3; b: 2 mM 2 in 3) and two
PVC-based anion-exchanger membranes (c: PVC/chloroparaffin
membrane containing 5% wt tridodecylmethylammonium chloride;
d: PVC/o-NPOE membrane containing 5% wt tridodecylmethyl-
ammonium chloride). Selectivities for (b) from ref. 23.
À
OAcÀ, NO2À, and HCO3 in the presence of traces of water
resulted in noisy responses and long response times due to
dearylation and the formation of a diphenylmethylphosphine
oxide.23 In the case of FÀ, OAcÀ, NO2À, and HCO3À, the
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
1870 | New J. Chem., 2010, 34, 1867–1874

View Article Online
Fig. 3 Fluoride response of fluorophilic anion exchanger (panel a: cationic site 2; panel b: cationic site 1) electrodes to KF solutions of different
concentrations. Inner filling solution of ISEs: 1 mM KF. Panel (a) adapted from ref. 23.
decomposition of 2 involved general base catalysis, resulting in
the formation of OHÀ by proton transfer from water to these
weak bases. Short term drifts in the potentiometric responses
caused by this decomposition are illustrated in Fig. 3a for the
FÀ response of an electrode based on 2. As Fig. 3a shows, the
measured potentials consistently dropped when stirring was
started, and increased when stirring was stopped. Sudden
changes in the potentiometric response due to stirring were
caused by local changes in analyte activities at the sample/
membrane phase boundary, caused by the ongoing decom-
position of 2.w This was further confirmed by observation of
the phosphine oxide formation by mass spectrometry.23
The same type of stirring experiment in FÀ solutions was
also performed for ISE membranes doped with the cationic
site 1. As Fig. 3b shows, unlike the membranes doped with 2,
the membranes containing cationic site 1 did not exhibit a stir
rate dependence of the FÀ response. This shows that 1 is much
more resistant to decomposition by general base catalysis than 2.
Indeed, the above described measurements of the selectivities
of ISE membranes based on 1 for ClÀ over OHÀ were only
possible because, unlike for membranes doped with 2, the
short-term stability of the OHÀ response of these membranes
was much higher. Only when membranes doped with 1 were
stored for 24 h in 0.1 M NaOH, were potentiometric responses
to ClÀ no longer observed. Decomposition of the phosphonium
compound 2 was not observed when it was incorporated into
ISE membranes and exposed to other anions, such as IÀ, BrÀ,
and NO3À, and even the weakly basic anions FÀ, OAcÀ,
NO2À, and HCO3À, as evidenced by the continued flawless
responses for well over 24 h.
Spectroscopic confirmation of the decomposition
of (Rf8(CH2)2)(Rf6(CH2)2)3P+IÀ
The unexpectedly strong OHÀ interference of the ISE
membranes based on the cationic salt 1, the eventual loss of
anion response upon prolonged OHÀ exposure, and also the
somewhat larger than expected selectivities for the weakly
basic anions FÀ and HCO3 warranted further study. While
À
only few reports of the coordinative binding of inorganic
anions to tetraalkylphosphonium ions are known, both
À
pentavalent NO2 and FÀ adducts have been described.38–40
Therefore, it seemed possible that
a pentacoordinated
(Rf8(CH2)2)(Rf6(CH2)2)3POH species might form due to the
extremely low polarity of fluorous solvents—unfavorable to
any ionic species—and enhanced Lewis acidity due to the
strongly electron-withdrawing properties of the four perfluoro-
alkylethyl groups. Because of the oxophilicity of phosphorus,
OHÀ binding to phosphonium cation would be expected to
be stronger than coordination of other anions, offering a
plausible explanation for the larger than expected OHÀ
interference observed for ISE membranes based on the
fluorophilic phosphonium cation 1. This hypothesis was
tested by 31P NMR spectroscopy. The 31P chemical shifts of
+
symmetrically substituted phosphonium compounds, PR4
,
w Note that transmembrane ion fluxes between the inner filling
solution of the ISE and the sample can also cause similar stirring
dependencies of the potentiometric signal, but that the potential
change would be in opposite direction when the inner filling solution
has a higher or lower concentration than the sample.
can be found in the range of approximately +20 to +45 ppm,
while 31P chemical shifts of pentacoordinated phosphorus
compounds are in the range of À90 to À60 ppm.41 As
expected, 31P NMR spectroscopy showed a singlet with the
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010 New J. Chem., 2010, 34, 1867–1874 | 1871

View Article Online
chemical shift of dP +39.79 for solutions of the iodide salt 1 in
perfluoroperhydrophenanthrene/perfluorohexane (1 : 4 v/v).
oxides (experimental: C, 1217.0208 and D, 1117.0272; theoretical:
C, 1217.0213 and D, 1117.0277).
1
The H NMR spectra show which protons are exchanged
When
a
perfluoroperhydrophenanthrene solution of
1
(2 mM) was stirred at room temperature for 24 h in contact
with an aqueous 0.1 M KOH solution and then diluted with
perfluorohexane for a 31P NMR measurement, the signal at
+39.79 ppm disappeared, and a new singlet at +36.58 ppm
was observed. This is consistent with the formation of a
trialkylphosphine oxide since the 31P chemical shifts of
R3P(QO) and the corresponding tetraalkylphosphonium
species are quite similar.41 No signal that could be attributed
to a pentacoordinated phosphorus species was observed.
The 31P NMR data, along with the fact that the electrodes
with the perfluoroperhydrophenanthrene solution of 1 lost
response to ClÀ after KOH/H2O exposure, lead to the
conclusion that in extremely alkaline conditions such as
pH = 13, the phosphonium cation (Rf8(CH2)2)(Rf6(CH2)2)3P+
slowly decomposes to form two phosphine oxides, i.e.,
(Rf8(CH2)2)(Rf6(CH2)2)2PQO and (Rf6(CH2)2)3PQO (see
Fig. 4). This interpretation was further substantiated by high
resolution mass spectrometry.
for deuterons, confirming the structures shown in Fig. 4.
Before exposure to alkaline solutions, the perfluorohexane
solution of the fluorophilic salt 1 showed peaks at 2.63–2.92
ppm (8H, m, CH2P) and 3.18–3.44 ppm (8H, m, CF2CH2)
(Fig. 5a). Both peaks were relatively broad, which appears to
be the result of strong ion aggregation. This is not surprising
since the extremely low polarity of fluorous solvents is
known to result in exceptionally strong ion pairing.3,4
After exposure of these solutions to aqueous KOH solution,
the two broad signals disappeared, and instead two pairs of
signals were observed (see Fig. 5b), i.e., 2.30–2.34 (6H, m,
CF2CH2) and 1.78–1.84 (6H, m, CH2P), and 1.86–2.00
(2H, m, CF2CH2) and 0.98–1.00 (3H, m, CH3). The first
pair of signals arises from the protons of the two phosphine
oxides. The latter two signals correspond to the protons of the
two perfluoroalkylethanes CH3CH2Rf8 and CH3CH2Rf6,
which are formed in the hydroxide-induced decomposition
of (Rf8(CH2)2)(Rf6(CH2)2)3P+IÀ, 1 (see Fig. 4), and indeed
match well with the literature values for perfluorohexylethane
at 2.0 and 1.1 ppm.42 When NaOH/D2O was used instead of
the KOH/H2O, only the six protons of the three methylene
groups adjacent to Rfn moieties of the phosphine oxide
functionality and the three protons of the methyl group in
the two perfluoroalkylethanes were observed (Fig. 5c), which
also simplified the coupling patterns. The formation of
RfnCH2CD3 shows that the proton–deuteron exchange at
the phosphorus-bound methylene groups occurs prior to
phosphine oxide formation. In other words, the proton–
deuteron exchange is initiated by deprotonation of the tetra-
alkylphosphonium salt, which is not only stabilized by ylide
formation but also the electron-withdrawing perfluoroalkyl
group on the carbon atom adjacent to the ylide. Since the
two hexadeuterated phosphine oxides and the trideuterated
perfluoroalkylethanes RfnCH2CD3 dominate entirely over
species of lesser degree of deuteration, it is evident that the
Electrospray ionization mass spectra of the tetraalkyl-
phosphonium salt 1 show the tetraalkylphosphonium ion at
m/z 1519 as the only positively charged ion with an appreciably
intense signal. However, after a solution of 1 in perfluoro-
(perhydrophenanthrene), 3, was equilibrated with an 0.1 M
KOH aqueous solution for 24 hours, the signal at m/z 1519
was much less intense. The two signals at 1226.9571 and
1126.9635 that were observed instead agree within error with
the K+ adducts of the two trialkylphosphine oxides shown
in Fig. 4 (theoretical: A, 1226.9576; B, 1126.9640). Using
NaOH/D2O instead of KOH/H2O shifted the two most
intense peaks by 10 u, which cannot be explained by the mass
difference of K+ and Na+ alone but suggests the exchange of
six protons of the trialkylphosphine oxides for deuterons.
Indeed, high resolution MS confirms the two peaks to be
within error identical with the theoretical values expected for
the two sodium ion adducts of two hexadeuterated phosphine
Fig. 4 Decomposition for the fluorophilic cation of salt 1 when exposed to KOH/H2O or NaOH/D2O.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
1872 | New J. Chem., 2010, 34, 1867–1874

View Article Online
the tetraalkylphosphonium cation could not be observed by
31P NMR. Hence, this cannot account for the larger than
expected OHÀ interference of membranes containing 1. Instead,
the OHÀ interference may be related to the accumulation of
phosphine oxides in membranes exposed to strongly alkaline
solutions and may be the result of dipole–anion interactions
between phosphine oxides and anions, as they are similarly
known, e.g., for anion–dimethyl sulfoxide adducts.43,44
Indeed, the slightly higher discrimination of IÀ, BrÀ, and
NO3 with reference to ClÀ, as observed in the selectivity
À
experiments, may be the result of an interaction between ClÀ
and a phosphine oxide. Further studies with ISEs doped with
phosphine oxides will be needed to confirm this hypothesis.
For environmental analysis, the high selectivity of the
fluorous membranes for perfluorooctanesulfonate and perfluoro-
octanoate as shown in this contribution is of particular
interest.45–47 Work to exploit the high selectivity in mechanically
more robust perfluoropolymer membranes doped with these
fluorophilic cationic sites is currently in progress in our
laboratories. However, fluorophilic tetraalkylphosphonium
ions may be used not only in ion-exchanger electrodes as
discussed in this work, but also in ISE membranes that contain
in addition to ionic sites an ionophore that selectively binds
specific ions.13–16 Without ionic sites, ISEs based on electrically
neutral ionophores do not show any ionic response.48 Also,
the ratio of ion exchanger sites and ionophore plays a crucial
role in the control of the selectivity of ion-selective electrodes
based on electrically neutral and charged ionophores.49
Indeed, uses of ionic sites in conventional non-fluorous
ionophore-doped membranes by far outnumber the uses of
ionophore-free membranes containing only ionic sites. For
example, for the years 1999 to 2009, the Science Citation Index
lists 197 publications on ISE membranes comprising the
common tetrakis(p-chlorophenyl)borate as ionic site (introduced
as potassium salt), and 184 of these 197 publications discuss
ionophore-based sensors. Similarly, the vast majority of the
applications of 1 and related tetraalkylphosphonium salts in
the still new field of fluorous membrane ISEs are expected to
be in combination with fluorophilic ionophores. Moreover,
the results presented here will also be relevant to the use of
these fluorophilic salts in biphasic catalysis and applications
beyond chemical sensing.
Fig. 5 1H NMR spectra of a solution of the fluorophilic salt 1 in
perfluoroperhydrophenanthrene, 3, directly after preparation (a) and
after equilibration for 24 h with 1 M KOH in H2O (b) or 1 M NaOH in
D2O (c). All solutions were diluted five times with perfluorohexanes
immediately prior to measurement.
proton–deuteron exchange is much faster than the decomposition
of the tetraalkylphosphonium cation to the two phosphine
oxides.
Acknowledgements
This project was supported by the National Science Foundation
(Grant CTS-0428046) and the National Institute of Health
(Grant 1R01 EB005225-01), and the Welch Foundation
(Grant A-1656).
Conclusions
While fluorous membrane ISEs based on the methyltriaryl-
phosphonium salt 2 and ISEs based on the tetraalkyl-
phosphonium salt 1 are suitable for measurements of anions
such as IÀ, NO3À, BrÀ, and ClÀ, and the environmental
pollutants perfluorocarboxylates and perfluorosulfonates,
only the latter are stable when exposed to anions that are
weakly basic. Extended exposure of membranes based on
1 to solutions of pH 13 eventually leads to irreversible
damage to the sensors, but brief exposure to highly concentrated
solutions of OHÀ does not impair the anionic response.
Formation of a pentavalent species by addition of OHÀ to
References
1 J. A. Gladysz, D. P. Curran and I. T. Horvath, Handbook of
´
Fluorous Chemistry, Wiley/VCH, Weinheim, 2004.
2 J. E. Brady and P. W. Carr, Anal. Chem., 1982, 54, 1751–1757.
3 P. G. Boswell and P. Buhlmann, J. Am. Chem. Soc., 2005, 127,
¨
8958–8959.
4 P. G. Boswell, E. C. Lugert, J. Rabai, E. A. Amin and
P. Buhlmann, J. Am. Chem. Soc., 2005, 127, 16976–16984.
¨
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010 New J. Chem., 2010, 34, 1867–1874 | 1873

View Article Online
´
5 C.-Z. Lai, S. S. Koseoglu, E. C. Lugert, P. G. Boswell, J. Rabai,
T. P. Lodge and P. Buhlmann, J. Am. Chem. Soc., 2009, 131,
1598–1606.
28 O. Dinten, U. E. Spichiger, N. Chaniotakis, P. Gehrig,
B. Rusterholz, W. E. Morf and W. Simon, Anal. Chem., 1991,
63, 596–603.
¨
6 C.-Z. Lai, M. E. Reardon, P. G. Boswell and P. Buhlmann,
¨
J. Fluorine Chem., 2010, 131, 42–46.
´ ´
7 I. T. Horvath and J. Rabai, Science (Washington, D. C.), 1994,
266, 72–75.
29 D. Ammann, E. Pretsch, W. Simon, E. Lindner, A. Bezegh and
E. Pungor, Anal. Chim. Acta, 1985, 171, 119–129.
30 E. Lindner and Y. Umezawa, Pure Appl. Chem., 2008, 80,
85–104.
8 D. Mandal and J. A. Gladysz, Tetrahedron, 2010, 66, 1070–1077.
9 D. Mandal, M. Jurisch, C. S. Consorti and J. A. Gladysz,
Chem.–Asian J., 2008, 3, 1772–1782.
10 Y. Yang, N. Vaidyanathan and S. G. Weber, J. Fluorine Chem.,
2009, 130, 1022–1027.
11 J.-M. Vincent, J. Fluorine Chem., 2008, 129, 903–909.
12 G. Pozzi, S. Quici and R. H. Fish, J. Fluorine Chem., 2008, 129,
920–929.
31 E. Bakker, E. Pretsch and P. Buhlmann, Anal. Chem., 2000, 72,
1127–1133.
32 L. J. Henderson, J. Biol. Chem., 1921, 46, 411–419.
33 P. C. Meier, Anal. Chim. Acta, 1982, 136, 363–368.
¨
34 E. C. Lugert, T. P. Lodge and P. Buhlmann, J. Polym. Sci., Part B:
¨
Polym. Phys., 2007, 42, 4915–4924.
35 E. J. Olson, P. G Boswell, B. L. Givot, L. Yao and P. Buhlmann,
¨
J. Electroanal. Chem., 2010, 639, 154–160.
13 W. E. Morf, The Principles of Ion-selective Electrodes and of
Membrane Transport, Elsevier, New York, NY, 1981.
14 R. D. Johnson and L. G. Bachas, Anal. Bioanal. Chem., 2003, 376,
328–341.
36 C.-Z. Lai, M. A. Fierke, A. Stein and P. Buhlmann, Anal. Chem.,
¨
2007, 79, 4621–4626.
37 M. A. Fierke, C.-Z. Lai, P. Buhlmann and A. Stein, Anal. Chem.,
¨
2010, 82, 680–688.
38 J. H. Clark and D. J. Macquarrie, J. Chem. Soc., Chem. Commun.,
1988, 229–230.
15 E. Bakker, P. Buhlmann and E. Pretsch, Chem. Rev., 1997, 97,
3083–3132.
¨
16 P. Buhlmann, E. Pretsch and E. Bakker, Chem. Rev., 1998, 98,
39 S. J. Brown, J. H. Clark and D. J. Macquarrie, J. Chem. Soc.,
Dalton Trans., 1988, 277–280.
¨
1593–1687.
17 P. Buhlmann, M. Hayakawa, T. Ohshiro, S. Amemiya and
¨
Y. Umezawa, Anal. Chem., 2001, 73, 3199–3205.
40 H. Schmidbaur, K.-H. Mitschke and J. Weidlein, Angew. Chem.,
Int. Ed. Engl., 1972, 11, 144–145.
18 P. Upreti, L. E. Metzger and P. Buhlmann, Talanta, 2004, 63,
139–148.
19 P. G. Boswell, C. Szıjjarto, M. Jurisch, J. A. Gladysz and J. Rabai,
Anal. Chem., 2008, 80, 2084–2090.
41 E. Pretsch, P. Buhlmann and M. Badertscher, Structure
¨
¨
Determination of Organic Compounds, Springer, New York, 2009.
42 B. Boutevin, F. Guida-Pietrasanta and A. Ratsimihety, J. Fluorine
Chem., 1993, 60, 211–223.
20 L. Mercs, G. Pozzi and S. Quici, Tetrahedron Lett., 2007, 48,
3053–3056.
43 T. F. Magnera, G. Caldwell, J. Sunner, S. Ikuta and P. Kebarle,
J. Am. Chem. Soc., 1984, 106, 6140–6146.
21 S. Shirakawa, Y. Tanaka and K. Maruoka, Org. Lett., 2004, 6,
1429–1431.
44 R. W. Taft and F. G. Bordwell, Acc. Chem. Res., 1988, 21,
463–469.
22 G. Maayan, R. H. Fish and R. Neumann, Org. Lett., 2003, 5,
3547–3550.
45 G. W. Olsen, J. M. Burris, J. H. Mandel and L. R. Zobel, J. Occup.
Environ. Med., 1999, 41, 799–806.
23 P. G. Boswell, A. C. Anfang and P. Buhlmann, J. Fluorine Chem.,
2008, 129, 961–967.
46 S.-K. Kim and K. Kannan, Environ. Sci. Technol., 2007, 41,
8328–8334.
¨
24 L.-N. He, H. Yasuda and T. Sakakura, Green Chem., 2003, 5, 92–94.
25 C. Emnet, K. M. Weber, J. A. Vidal, C. S. Consorti, A. M. Stuart
and J. A. Gladysz, Adv. Synth. Catal., 2006, 348, 1625–1634.
47 J. W. Martin, M. M. Smithwick, B. M. Braune, P. E. Hoekstra, D.
C. G. Muir and S. A. Mabury, Environ. Sci. Technol., 2004, 38,
373–380.
26 G. Vla
´
d, F. Richter and I. T. Horva
´
th, Org. Lett., 2004, 6,
48 P. Buhlmann, S. Yajima, K. Tohda and Y. Umezawa, Electrochim.
¨
Acta, 1995, 40, 3021.
4559–4561.
27 G. Vla
´
d, F. Richter and I. T. Horva
´
th, Tetrahedron Lett., 2005, 46,
49 S. Amemiya, P. Buhlmann, E. Pretsch, B. Rusterholz and
¨
Y. Umezawa, Anal. Chem., 2000, 72, 1618.
8605–8608.
ꢀc
This journal is The Royal Society of Chemistry and the Centre National de la Recherche Scientifique 2010
1874 | New J. Chem., 2010, 34, 1867–1874