A small molecule sensor for fluoride based on an autoinductive, colorimetric signal amplification reaction

Article abstract of DOI:10.1039/c2ob25363a

This article describes a small molecule reagent that is capable of detecting fluoride down to 0.12 mM (2.3 ppm) in water. The reagent reveals this level of fluoride through a novel autoinductive signal amplification reaction that produces an unambiguous colorimetric readout.

Full text of DOI:10.1039/c2ob25363a

View Article Online / Journal Homepage / Table of Contents for this issue
Dynamic Article Links
Organic &
Biomolecular
Chemistry
Cite this: Org. Biomol. Chem., 2012, 10, 3595
www.rsc.org/obc
COMMUNICATION
A small molecule sensor for fluoride based on an autoinductive, colorimetric
signal amplification reaction†
Matthew S. Baker and Scott T. Phillips*
Received 20th February 2012, Accepted 11th March 2012
DOI: 10.1039/c2ob25363a
reaction and provides a time-dependent colorimetric readout that
reveals the level of fluoride in a sample of water. We also
describe an efficient one-step synthesis of 1, as well as the devel-
opment and optimization of assay conditions using reagent 1 that
enable both semi-quantitative and quantitative measurements of
aqueous fluoride down to 2.3 ppm (0.12 mM), which is within
the relevant window of the US Environmental Protection
Agency’s (EPA) recommended fluoride concentration in water
(2 ppm)1 as well as the EPA’s mandated upper limit (4 ppm).1
Reagent 1 is unique compared with other reagents for detect-
ing fluoride because of a combination of three features: (i) it
detects fluoride in an aqueous sample,2 (ii) it detects fluoride
derived from inorganic salts rather than from fluoride bound to a
phase-transfer catalyst such as tetrabutylammonium,3 and (iii) it
amplifies signal for the detection event to increase the sensitivity
of the assay.4 These capabilities arise from a designed reaction
cascade and network of reactions. Specifically, the reactions are:
fluoride-mediated cleavage of the silyl ether group in 1 that
induces a cascade elimination reaction (Fig. 1). This cascade
reaction releases 4-aminobenzaldehyde (which produces a bright
yellow-colored solution) and two new fluoride ions. These two
fluoride ions then propagate the signal amplification reaction in a
reaction network by reacting with more of 1 to increase the sen-
sitivity of the assay. This approach is complimentary to a recent,
independent study conducted by Shabat et al.5 in the area of
fluoride detection.
This article describes a small molecule reagent that is
capable of detecting fluoride down to 0.12 mM (2.3 ppm) in
water. The reagent reveals this level of fluoride through a
novel autoinductive signal amplification reaction that pro-
duces an unambiguous colorimetric readout.
Inexpensive and operationally straightforward methods for
detecting fluoride in water are needed for measuring the quality
of drinking water in resource-limited regions. Herein we describe
reagent 1 as a unique sensor for aqueous fluoride (Fig. 1).
Reagent 1 functions via an autoinductive signal amplification
Rationale for developing a fluoride detection reagent. Studies
now indicate that long-term ingestion of water containing more
than 4 ppm of fluoride can cause dental and skeletal fluorosis, as
well as osteoporosis.6 Recent estimates suggest that approxi-
mately 200 million people from among 25 nations face hazar-
dous levels of naturally occurring fluoride in drinking water.7
Many of these people live in resource-limited regions where
advanced water purification facilities are lacking8 and where
standard analytical equipment9 (such as ion chromatography, gas
chromatography, capillary zone electrophoresis and radio-
analysis) are not available or cannot be used to measure the
levels of fluoride in drinking water.
Fig. 1 Pathway by which reagent 1 detects aqueous fluoride, amplifies
4-aminobenzaldehyde as a colorimetric readout, and releases two more
equivalents of fluoride to propagate the autoinductive signal amplifica-
tion reaction.
As a first step towards mitigating this problem, sources of
drinking water that contain excess fluoride must be identified
before an appropriate local treatment can be applied. Moreover,
once a source of water is treated to remove fluoride, the effective-
ness of the treatment must be evaluated. Both of these situations
Department of Chemistry, The Pennsylvania State University, 104
Chemistry Building, University Park, PA 16802, USA. E-mail: sphil-
lips@psu.edu; Fax: +814 865 5235; Tel: +814 867 2502
†Electronic supplementary information (ESI) available: Experimental
procedures, tables of data, figures providing additional results. See DOI:
10.1039/c2ob25363a
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3595–3599 | 3595

View Article Online
Scheme 1 Synthesis of detection reagent 1.
Fig. 2 Schematic of the assay procedure for detecting fluoride colori-
metrically. Step 1 involves diluting a sample that contains aqueous
fluoride into a solution of 1. After a period of time required for the auto-
inductive signal amplification reaction to proceed, photographs are taken
(step 2) and the colorimetric response is measured using image-proces-
sing software.
require an inexpensive, yet sensitive and selective sensor for
measuring relevant levels of fluoride in water.
The ideal sensor for these environments must be capable of
operating under a stringent set of criteria: the sensor must (i)
provide an unambiguous result that is easy to interpret by an
untrained user; (ii) provide measurements without using instru-
ments or readers; (iii) remain stable for prolonged periods of
time (i.e., when transported and stored without refrigeration);
(iv) detect fluoride selectively over all other anions present in
water; and (v) measure fluoride down to at least 4 ppm in
water.10
Substantial progress has been made recently towards creating
reagents for portable fluoride sensors, some of which satisfy
several of the criteria outlined above for the ideal fluoride
sensor.11 These reagents can be classified into four main cat-
egories based on their mode of operation: (i) supramolecular rec-
ognition,12 (ii) Lewis acid–base interactions,13 (iii) hydrogen
bonding,14 and (iv) reaction-based detection.15 The reagent
described in this article falls into the latter category, but unlike
most reaction-based sensors that detect fluoride in water our
reagent amplifies signal,4 which provides a method for increas-
ing the sensitivity of the assay.
Design of reagent 1. We designed reagent 1 with four con-
siderations in mind: (i) the reagent must detect fluoride selec-
tively (the tert-butyldimethylsilyl ether group provides this
capability16); (ii) a colorimetric readout must be generated for
each reaction with fluoride to provide an unambiguous readout
(generation of 4-aminobenzaldehyde provides this function16,17);
(iii) two molecules of fluoride must be released per reaction of 1
with fluoride, both to propagate the signal amplification reaction,
and to accelerate the rate of the reaction (release of only one mol-
ecule of fluoride would propagate, but not accelerate the rate of
the reaction); and (iv) the reagent that enables all of these capa-
bilities must be accessible via an exceedingly short and inexpen-
sive synthesis that makes the reagent an attractive starting point
for use in resource-limited environments.18
Synthesis of reagent 1. The synthesis of reagent 1 is shown in
Scheme 1. Specifically, 4-(difluoromethyl)benzoic acid (2) was
coupled with p-[(tert-butyldimethylsilyl)oxy]benzyl alcohol
through an efficient Curtius rearrangement19 to afford 1 in 87%
yield. In comparison, our previous synthesis of 1 required three
steps,16 proceeded in only 40% overall yield, and required the
use of the expensive and potentially dangerous reagent diethyla-
minosulfur trifluoride.
detection of ∼2 ppm fluoride in water. The general assay design
is shown in Fig. 2. First, the aqueous sample that contains
fluoride is diluted into a solution of 1 in organic solvents. Next,
the autoinductive signal amplification reaction depicted in Fig. 1
begins, and a yellow color (from 4-aminobenzaldehyde) forms
in the solution. Over time, the quantity of 4-aminobenzaldehyde,
due to the autoinductive signal amplification reaction, increases
to improve the sensitivity of the assay.
A fixed-time assay using this procedure enables semi-quanti-
tative as well as quantitative measurements of the level of
fluoride in the sample. Semi-quantitative assays simply require
monitoring the assay solution until an obvious yellow color
forms and then comparing the time required to form this color
with calibration data relating time (to a yellow color) with
known concentrations of fluoride in water. To obtain a quantitat-
ive measurement, the assay solution is photographed using a
camera-equipped cellular phone, and the intensity of the color in
the digital image is measured using image-processing software
(this digital analysis process, in theory, could be accomplished
off-site by a trained physician through a process referred to as
Telemedicine20). Once the assay reaches 30% of the maximum
colorimetric signal (which is known (vide infra) for different
concentrations of fluoride), the time required to reach this inten-
sity of signal can be compared with a calibration curve that
relates “time to 30% of the maximum colorimetric signal” with
“concentration of fluoride in the sample” to obtain the concen-
tration of fluoride in the experimental sample.
Quantitative detection of aqueous fluoride. Our first gener-
ation assay conditions were as follows: a sample of water
(10 μL) that contains fluoride (we used CsF) was transferred to a
solution of reagent 1 (0.13 M, 95 μL) in 18 : 1 methanol–pyri-
dine. In this step, the original concentration of fluoride is diluted
10.5×. The resulting diluted solution was agitated by shaking for
5 s and then left undisturbed as the assay developed. The appear-
ance of a visible yellow color marked the end-point of the assay,
although in some experiments, we also allowed the reaction to
reach completion (i.e., the point at which the yellow color no
longer increased in intensity) to obtain full kinetics for the
response.
Design of the assay. Since reagent 1 is hydrophobic, we used
a mixed organic–aqueous solution to detect aqueous fluoride,
and then optimized the solvent conditions for the assay to enable
Both the original concentration of fluoride in water (before it
was diluted in the assay) and the assay time affect the intensity
of the colorimetric response that is obtained with this type of
3596 | Org. Biomol. Chem., 2012, 10, 3595–3599
This journal is © The Royal Society of Chemistry 2012

View Article Online
Fig. 4 A control reagent (compound 3) for determining the sensitivity
for detecting fluoride without the contribution of an autoinductive signal
amplification reaction. The assay conditions were identical to those used
in Fig. 3. (a) Reaction of reagent 3 with fluoride to produce 4-aminoben-
zaldehyde. (b) Reagent 3 provides a non-amplified response to fluoride.
intensity that corresponds to 30% completion is reached (the first
sign of a distinctly yellow color), and then compare the measure-
ment of time with the calibration curve.
The selectivity of the assay is equally as important as the sen-
sitivity, and, as expected for an activity-based detection mechan-
ism,21 reagent 1 is highly selective for detecting fluoride over
many other anions that might be expected in an aqueous sample
(Fig. 3c).
Likewise, the stability of the reagent is crucial if 1 (or, more
likely, future derivatives of 1) is ever to be used in resource-
limited environments. While we have not performed an exhaus-
tive stability study at this early stage in development, we have
found that, as a solid, reagent 1 shows no signs of decomposition
when stored open to the air at 37 °C for four weeks.
Effect of the signal amplification reaction on the sensitivity of
the assay. To test the importance of the signal amplification reac-
tion on the sensitivity of the assay, we prepared reagent 3
(Fig. 4a). Upon reaction with fluoride, reagent 3 releases 4-ami-
nobenzaldehyde (which also is released by reagent 1), but does
not release two new equivalents of fluoride, and, hence, does not
amplify signal (Fig. 4b). The graph in Fig. 4b was obtained
using a procedure analogous to that described in Fig. 2.
Comparison of Fig. 4b and 3a shows that 3 provides only a
linear response to fluoride (as expected), whereas 1 provides an
amplified response. More importantly, reagent 3 is at least 1000×
less sensitive than 1 in detecting fluoride (refer to the visual
detection limits in Fig. 3a and 4b), and is capable of detecting
only 123 mM (2300 ppm) fluoride in water, which is well
outside the EPA-recommended range.
Fig. 3 Assay results using reagent 1 to detect aqueous fluoride accord-
ing to the procedure outlined in Fig. 2. (a) Quantity of color produced
when 1 reacts with various initial quantities of fluoride via the autoin-
ductive cycle shown in Fig. 1. The solution turns a bright, visible yellow
color when 30% of the maximum possible color is produced (i.e., when
30% of 1 is consumed). The experiments were performed in triplicate
and all data are plotted on the graph. (b) A log–log calibration curve for
quantitatively measuring the level of fluoride in a sample. Overlapping
triplicate measurements are plotted on the graph. (c) Effect of anions on
the colorimetric response of reagent 1. The photographs were taken 3 h
after 1 (0.13 M in 18 : 1 MeOH–pyr) was exposed to 0.5 equiv of
various anions (10 μL from 0.62 M solutions in nanopure water).
assay (Fig. 3a). Fig. 3a reveals four key features about this first
generation version of the assay: (i) reagent 1 is consumed com-
pletely even when substoichiometric quantities of fluoride are
present in the assay; (ii) the sigmoidal response profile is charac-
teristic of an autoinductive signal amplification reaction, as
expected;16 (iii) an intense yellow color is visible after only 30%
of reagent 1 is consumed (see the solid indicator line); and (iv)
background signal in the absence of applied fluoride becomes an
issue only after long assay times (>33 h).
For quantitative assays, the calibration curve relating time (to
a visible color) with concentration of fluoride in water is shown
in Fig. 3b. To quantitatively measure the concentration of
fluoride in a sample, one must simply run the assay until the
Optimizing the assay. Given that the autoinductive signal
amplification reaction is necessary for achieving the desired sen-
sitivity, we further optimized the assay to enable detection of
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3595–3599 | 3597

View Article Online
even 2 ppm fluoride in water and to increase the rate of the
signal amplification reaction to decrease the assay time. Our
assay strategy (Fig. 2) remained nearly the same, but we
reasoned that increasing the polarity of the assay medium would
stabilize the presumed azaquinone methide transition state
(Fig. 1) that releases the first equivalent of fluoride; as a conse-
quence, the rate of the signal amplification reaction should
increase.22 In addition, we recognized that we could increase the
relative proportion (v/v) of the aqueous sample in the assay sol-
ution if different solvents were used. By increasing the volume
of the aqueous sample relative to the organic solvent, we would
decrease the extent to which the fluoride is diluted in Step 1
(Fig. 2) and further increase the rate of the detection event.
To achieve these goals we conducted two sets of solvent
screens. First, preliminary experiments showed that pyridine
accelerates the amplification reaction. More specifically, ∼1%
pyridine (relative to the total volume of solvent) is sufficient to
maintain this accelerated rate (Scheme S1†); higher concen-
trations of pyridine offered no additional benefit, and, in fact,
concentrations above 24% slowed the signal amplification reac-
tion. Second, we screened polar protic solvents (data not shown)
to identify conditions that maintain a high solvent polarity while
still effectively solvating reagent 1, even after addition of the
aqueous sample in Step 1 (Fig. 2).
The solvent ratio of 10 : 4 : 1 iPrOH–aqueous sample–pyridine
proved most effective for increasing the rate of the signal
amplification reaction when 2300 ppm (0.122 M) fluoride was
present in the aqueous sample (the assay used 0.16 M of 1 and
was conducted at 23 °C). With this solvent combination, the
aqueous sample now makes up 27% of the total assay volume,
which means that the fluoride in the aqueous sample is diluted
only 3.8× in step 1 (for comparison, the first-generation solvent
conditions resulted in 10.5× dilution). The consequence of these
changes is that an assay for 2300 ppm (0.122 M) fluoride now is
∼3.1× faster than when the original solvent conditions are used.
Likewise, this acceleration in reaction rate has the added
benefit of increasing the sensitivity of the assay – now even
2.3 ppm (0.12 mM) fluoride can be distinguished easily from the
background reaction (Fig. 5a). This increase in reaction rate and
sensitivity, however, does not impact the selectivity of the assay:
as shown in Fig. 5c, fluoride remains the only anion tested that
provides an observable signal.
Fig. 5 Optimized conditions for increasing the rate and sensitivity of
the assay for detecting fluoride. (a) Response of 1 to various quantities
of fluoride in an aqueous sample. The optimized assay conditions
involved transferring 20 μL (step 1, Fig. 2) of an aqueous sample into
55 μL (10 : 1 iPrOH–pyridine) containing 0.22 M 1 and then allowing
the color to develop (step 2, Fig. 2) at 23 °C. The assays were repeated
in triplicate and all data are shown on the graph. (b) Photographs
showing that even 2.3 ppm (0.12 mM) aqueous fluoride can be detected
easily over the background reaction. (c) Photographs showing that the
selectivity for fluoride has not been affected by the optimized assay con-
ditions. The photographs were taken 2 h after 1 was exposed to 0.2
equiv of each anion under the assay conditions described in (a).
From a practical viewpoint, to determine whether a sample
has more or less than 2 ppm fluoride (the EPA recommended
limit1), the assay solution must simply be checked to see if it
turns a visible yellow color by 9 h (Fig. 5b); if it does not, then
it likely contains less than 2 ppm fluoride. If it turns a visible
yellow color before 9 h, then it likely contains more than 2 ppm
fluoride.
of ideal detection reagent that is needed for detecting aqueous
fluoride in resource-limited regions.
Acknowledgements
In conclusion, while the reagent described in this Communi-
cation is comparable in sensitivity to other state-of-the-art
reagents that detect fluoride in water, the design of the reagent
offers two key advances in the area of fluoride detection: (i) it
provides a unique and necessary signal amplification reaction for
detecting relevant levels of fluoride; and (ii) it can be used to
measure the level of fluoride in a sample of water, both in semi-
quantitative threshold-type assays, and in quantitative assays.
Moreover, because it is accessible in only one synthetic step,
reagent 1 provides a useful starting point for designing the type
We acknowledge financial support from the Arnold and Mabel
Beckman Foundation, The Camille and Henry Dreyfus Foun-
dation, Mr Louis Martarano, and The Pennsylvania State Univer-
sity. We thank Dr Landy K. Blasdel for assistance in preparing
the manuscript.
Notes and references
1 EPA National Primary Drinking Water Standards 2009, see http://water.
epa.gov/drink/contaminants/ for more information.
3598 | Org. Biomol. Chem., 2012, 10, 3595–3599
This journal is © The Royal Society of Chemistry 2012

View Article Online
2 The vast majority of fluoride sensors are designed to detect fluoride in
organic solvents where issues related to solvent interactions with fluoride
are minimized. Several recent fluoride sensors now are capable of detect-
ing fluoride in water. These sensors are highlighted in ref. 12, 13, 15.
3 Tetrabutylammonium fluoride (TBAF) contains substantial quantities of
hydroxide.23 This hydroxide likely cleaves silyl ethers and contributes to
the overall activity-based detection event when TBAF is used as the
source of fluoride. In other words, the use of TBAF may not accurately
reflect the ability of a fluoride sensor to detect fluoride that is dissolved in
water.
(e) I. H. A. Badr and M. E. Meyerhoff, J. Am. Chem. Soc., 2005, 127,
5318; (e) M. J. Bayer, S. S. Jalisatgi, B. Smart, A. Herzog, C. B. Knobler
and M. F. Hawthorne, Angew. Chem., Int. Ed., 2004, 43, 1854;
(f) A. Oehlke, A. A. Auer, I. Jahre, B. Walfort, T. Rueffer, P. Zoufala,
H. Lang and S. Spange, J. Org. Chem., 2007, 72, 4328; (g) Y. Kim and
F. P. Gabbaï, J. Am. Chem. Soc., 2009, 131, 3363; (h) C. R. Wade, I.-
S. Ke and F. P. Gabbaï, Angew. Chem., Int. Ed., 2012, 51, 478.
14 (a) Y. Li, C. Lifeng and H. Tian, J. Org. Chem., 2006, 71, 8279; (b) D.
A. Jose, D. K. Kumar, B. Ganguly and A. Das, Org. Lett., 2004, 6, 3445;
(c) Y. Qu, J. Hua and H. Tian, Org. Lett., 2010, 12, 3320; (d) Y. Shiraishi,
H. Maehara, T. Sugii, D. Wang and T. Hirai, Tetrahedron Lett., 2009, 50,
4293; (e) C. B. Black, B. Andrioletti, A. C. Try, C. Ruiperez and J.
L. Sessler, J. Am. Chem. Soc., 1999, 121, 10438; (f) H. Miyaji, W. Sato
and J. L. Sessler, Angew. Chem., Int. Ed., 2000, 39, 1777.
4 Only two other fluoride sensors have been reported that amplify signal
for the detection event. They are ref. 15a, which detects fluoride in
organic solvent, and ref. 5, which is complimentary to the sensor
described in this Communication.
5 R. Perry-Feigenbaum, E. Sella and D. Shabat, Chem.–Eur. J., 2011, 17,
12123.
15 (a) T.-H. Kim and T. M. Swager, Angew. Chem., Int. Ed., 2003, 42, 4803;
(b) S. Y. Kim and J.-I. Hong, Org. Lett., 2007, 9, 3109; (c) C.-Q. Zhu, J.-
L. Chen, H. Zheng, Y.-Q. Wu and J.-G. Xu, Anal. Chim. Acta, 2005, 539,
311; (d) J. F. Zhang, C. S. Lim, S. Bhuniya, B. R. Cho and J. S. Kim,
Org. Lett., 2011, 13, 1190; (e) O. A. Bozdemir, F. Sozmen,
O. Buyukcakir, R. Guliyev, Y. Cakmak and E. U. Akkaya, Org. Lett.,
2010, 12, 1400.
6 (a) S. Jagtapt, M. K. Yenkie, N. Labhsetwart and S. Rayalu, Chem. Rev.,
2012, DOI: 10.1021/cr2002855; (b) M. Mohapatra, S. Anand, B.
K. Mishra, D. E. Giles and P. Singh, J. Environ. Manage., 2009, 91, 67.
7 S. Ayoob and A. K. Gupta, Crit. Rev. Environ. Sci. Technol., 2006, 36,
433.
8 Certain resource-limited regions in India provide a case-in-point. It is esti-
mated that approximately 60 million people in India are ingesting water
that contains levels of fluoride that are above the recommended limit, and
that of these 60 million, 6 million now have crippling osteoporosis as a
result.D. R. Ready, Neurol. India, 2009, 57, 7.
9 C.-Q. Zhu, J.-L. Chen, H. Zheng, Y.-Q. Wu and J.-G. Xu, Anal. Chim.
Acta, 2005, 539, 311.
10 Since the deleterious health effects associated with fluoride are the conse-
quence of long-term ingestion of water that contains high levels of
fluoride, a sensor for use in resource-limited environments does not
necessarily need to provide rapid results for an assay.
11 (a) Z.-H. Lin, Y.-G. Zhao, C.-Y. Duan, B.-G. Zhang and Z.-P. Bai, Dalton
Trans., 2006, 3678; Z.-H. Lin, S.-J. Ou, C.-Y. Duan, B.-G. Zhang and Z.-
P. Bai, Chem. Commun., 2006, 624(b) R. Hu, J. Feng, D. Hu, S. Wang,
S. Li, Y. Li and G. Yang, Angew. Chem., Int. Ed., 2010, 49, 4915.
12 (a) M. Takeuchi, T. Shioya and T. M. Swager, Angew. Chem., Int. Ed.,
2001, 40, 3372; (b) S. Guha and S. Saha, J. Am. Chem. Soc., 2010, 132,
17674; (c) S. Watanabe, H. Seguchi, K. Yoshida, K. Kifune, T. Tadaki
and H. Shiozaki, Tetrahedron Lett., 2005, 46, 8827.
13 (a) C. R. Wade, A. E. J. Broomsgrove, S. Aldridge and F. P. Gabbaï,
Chem. Rev., 2010, 110, 3958; (b) S. Yamaguchi, S. Akiyama and
K. Tamao, J. Am. Chem. Soc., 2001, 123, 11372; (c) Q. Zhao, F. Li,
S. Liu, M. Yu, Z. Liu, T. Yi and C. Huang, Inorg. Chem., 2008, 47, 9256;
16 M. S. Baker and S. T. Phillips, J. Am. Chem. Soc., 2011, 133, 5170.
17 S. A. Nuñez, K. Yeung, N. S. Fox and S. T. Phillips, J. Org. Chem.,
2011, 76, 10099.
18 We previously described reagent 1 as a general signal amplification
reagent that could be paired with a second detection reagent that released
fluoride upon reaction with a specific analyte (we demonstrated the detec-
tion of Pd(0)).16 In this article, we now describe several advances that
arise as a result of a year-long study on the mechanism and kinetics of
the reaction of 1 with fluoride. These advances include (i) the develop-
ment of a convenient assay procedure for detecting aqueous fluoride, (ii)
the development of optimized assay conditions that enable 1 to detect rel-
evant levels of fluoride in water, and (iii) the development of an exceed-
ingly efficient synthesis route that provides 1 in high yield in a single
synthetic step.
19 M. B. Smith and J. March, March’s Advanced Organic Chemistry, Wiley,
Hoboken, New Jersey, 6th edn, 2007, p. 1609.
20 A. Martinez, S. T. Phillips, E. Carrilho, S. W. Thomas III, H. Sindi and
G. M. Whitesides, Anal. Chem., 2008, 80, 3699.
21 D.-G. Cho and J. L. Sessler, Chem. Soc. Rev., 2009, 38, 1647.
22 K. M. Schmid and S. T. Phillips, unpublished work.
23 (a) D. P. Cox, J. Terpinski and W. Lawrynowicz, J. Org. Chem., 1984,
49, 3216; (b) A. S. Pilcher, H. L. Ammon and P. DeShong, J. Am. Chem.
Soc., 1995, 117, 5166.
This journal is © The Royal Society of Chemistry 2012
Org. Biomol. Chem., 2012, 10, 3595–3599 | 3599