Toward wearable sensors: Fluorescent attoreactor mats as optically encoded cross-reactive sensor arrays

Article abstract of DOI:10.1002/anie.201105629

Ultrasmall fluorescent sensors and their application for fabrication of wavelength-addressable virtual sensor arrays capable of sensing metal cations are described. The fluorescent probes generated in situ within the fiber mats can be used as wearable sensors for identifying heavy metal ions in a qualitative and quantitative fashion. Copyright

Full text of DOI:10.1002/anie.201105629

Angewandte
Chemie
DOI: 10.1002/anie.201105629
Sensor Arrays
Toward Wearable Sensors: Fluorescent Attoreactor Mats as Optically
Encoded Cross-Reactive Sensor Arrays**
Pavel Anzenbacher Jr.,* Fengyu Li, and Manuel A. Palacios
Herein we demonstrate the fabrication of high-density
attoreactor sensor arrays for the detection of metal ions.
Recently, we developed a method of synthesis of ultra-small
amounts of products, down to zeptomol (zmol, 10ꢀ21 mol)
amounts confined in junctions of polymer nanofibers.[1] This
method is well suited for the synthesis of fluorescent
materials. The electro-spun nanofibers[2] give porous mats
that comprise fluorescent attoliter-sized junctions termed
attoreactors that give off bright fluorescence (Figure 1), and
form almost conformal coating on various surfaces. More-
over, the highly porous electrospun nanofiber mats facilitate
sensor–analyte mass exchange, which results in short response
times. Such mats with embedded in situ formed fluorescent
“attosensors” generate high-density and parallel sensor
arrays.[3] Furthermore, the key feature of attosensor array
mats, when compared to sensor films based on bulk material,
is that approximately a 1000 of molecules can be positioned in
a small volume (attoliters), thus concentrating the fluores-
cence of probe molecules that are easy to detect with optical
microscopy.[1]
Cation recognition and detection, particularly of heavy
metals,[4] has an enormous potential environmental and
health-related impact.[5] Herein, we explore the possibility
of synthesizing fluorescent sensors within one nanofiber mat,
which would then serve as an optically encoded (wavelength-
addressable) cross-reactive attosensor array. We demonstrate
this approach on sensing a benchmark group of ten metal ions
(Al3+, Fe3+, Co2+, Ni2+, Cu2+, Zn2+, Hg2+, Cd2+, Ca2+, Mg2+).
This method can be, however, used for other analytes as well.
First, we show the ability of a single attoreactor to act as
a fluorescent sensor. Here, we use fluorophore-polyamines
(e.g. dansyl-polyamines)[6] that bind metal ions while display-
ing a change in fluorescence. Figure 2 shows three polyur-
ethane nanofibers: Two are doped with triethylenetetramine
(4N) and one with dansyl chloride (bottom left!top right),
forming two fluorescent attoreactors. One attoreactor (lower
left) serves as a reference while the other is used as a sensor.
The images show the attosensors before and after adding
a droplet of aqueous CoCl2, and after washing away the CoCl2
with water. The fluorescence from the reference reactor did
not change appreciably throughout the experiment. The
exposure of the attosensor to CoCl2 resulted in fluorescence
Figure 1. Overlap of two types of polymer nanofibers (300ꢁ50 nm
diam.), each carrying a different reagent, for example, dansyl chloride
and polyamine (A), to produce fluorescent products (B) in the fiber
junctions, attoliter volume reactors comprising zeptomol amounts of
fluorescent products (C, D) in the form of a highly porous mat (E) that
can be deposited by a shadow mask (F) on a variety of surfaces (G).
[*] P. Anzenbacher Jr., F. Li, M. A. Palacios
Department of Chemistry, Bowling Green State University
Bowling Green, OH 43403 (USA)
Figure 2. Self-referencing attosensor: One reactor (lower left) is used
as a reference, while the other is used as Co2+ ion sensor. Top: Bright-
field images before (A) and after (B) addition of CoCl2, and after
regenerating the sensor with nanopure water (C). Corresponding
fluorescence images (D–F). Bottom: the 3D representation of the
integrated fluorescence intensity (G–I) corresponding to D–F.
E-mail: pavel@bgsu.edu
[**] P.A. acknowledges support from BGSU and the NSF (CHE-0750303,
DMR-1006761).
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201105629.
Angew. Chem. Int. Ed. 2012, 51, 2345 –2348
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
2345

.
Angewandte
Communications
quenching. The fluorescence was recovered upon washing
with water. This observation confirms the ability of the
attoreactors to act as sensors that can be generated in situ and
bind metal ions in a reversible fashion (4 cycles).
In the future, the analytical potential of the attosensor
mats could be expanded to become a massively parallel
sensing array by reading one or few individual attoreactors
(from a mat comprising approximately 107 attosensors).
While this qualitative test was performed at a high analyte
concentration (200 mm, 200 nL, pH 5), it can be performed at
a concentration as low as 5 ppm. One of the advantages of the
present approach is the possible use of small samples (down
to pL) obtained by concentrating diluted samples. In any case,
owing to the density of the nanofiber mat, there will be
approximately 104 individual attoliters sensors under every
10 nL droplet. Importantly, nanofiber mats can comprise
more than two reagents. We tested a nanomat composed of
four different nanofibers: A fiber doped with an amine such
as tetraethylene pentamine (5N), another one with dansyl
chloride (DS), one with fluorescamine (FL, Fluram), and one
with 7-chloro-4-nitrobenz-2-oxa-1,3-diazole (NBD).
None of these four chemicals are fluorescent, but the
products of the reaction of the fluorophore precursors (DS,
FL, NBD) with the amine are. Thus, the dansylated pentam-
ine displays cyan emission (lmax = 480 nm, t = 7.85 ns), the
Fluram-amine product shows blue emission (lmax = 465 nm,
t = 2.80 ns), and the NBD-amine a green emission (lmax
525 nm, t = 3.87 ns), Figure 3A–E.
=
The in-situ-synthesized probes may be generated in the
nanofiber mat to match the analyte composition. In an even
more general sense, the amounts of the label–receptor probes
may be adjusted on demand to satisfy the requirements for
excitation/emission wavelengths, amount, and concentration
of the analyte, etc. Figure 3F shows nine probes (3 ꢀ 3) with
DS–5N, FL–5N, and NBD–5N as well as zoomed-in images of
each junction type to show the difference in fluorescence
output. The three probes, DS–5N, FL–5N, and NBD–5N, have
different affinity to metal ions. This is because, for example,
FL–5N comprises a carboxylate group that may participate in
the metal ion binding, while the nitrogen atom in the
oxadiazole ring of NBD–5N may participate in the formation
of the probe–ion complex, or because the sulphonamide
oxygen in the DS–5N may interact with oxophilic cations, etc.
Similarly, a varying of the ethylene amines (4N, 5N, 6N)
introduces a variable likely to induce a different response to
different metal ions.
In the attosensor mat, the amine-bearing and fluoro-
phore-precursor-doped nanofibers are deposited at an angle
of approximately 908 to form a rectangular grid (Fig-
ure 3A,F). The location of the respective DS, FL, NBD
fibers is, however, random. Thus, the location of the in-situ
generated probes within the grid is random as the junctions
with blue/cyan/green emission are not formed in an exact pre-
defined location. This is, however, not a hindrance because
the probes are optically encoded with their specific emission
profiles.
Figure 3. The reaction of fluorophore precursors (DS, FL, NBD) with
tetraethylene pentamine in situ (i.e. in the nanofiber junctions) results
in fluorescent products. The color of emission (fluorescence image is
superimposed on a bright-field image) corresponds to the images
above (A). The emission can be observed as an image based on
fluorescence lifetime (B), so called fluorescence lifetime (FLIM) image.
The virtual sensor array comprising in situ prepared fluorescent probes
DS–5N, FL–5N, and NBD-5N (F). The three probes, FL–5N (C), NBD–
5N (D), and DS–5N (E), were synthesized from pentamine-doped
polyurethane fibers (vertical) and the fluorophore precursors in the
horizontal direction. The images are bright-field images with super-
imposed confocal fluorescence images.
ing all the probes (Figure 3). The metal sensing data were
acquired from analyte droplets (0.2 mL) deposited on the mat
together with a reference sample (water, pH 5). The drops
were allowed to dry prior to fluorescence measurement using
a UV/vis scanner (excitation at ca. 365 nm) that records
While the individual attosensors operate independently
(on a microscopic level, Figure 2), in practice, the fluores-
cence is recorded from the whole mat (12 ꢀ 12 mm) compris-
2346 www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 2345 –2348

Angewandte
Chemie
fluorescence response using four emission channels (l1 =
440 nm, l2 = 480 nm, l3 = 535 nm). The last channel used
excitation at 430 nm and emission was recorded at 535 nm.
As the probes are structurally similar, also their metal ion
affinities are not expected to lead to a very analyte-selective
response, but rather a cross-reactive one.[7] In the response
data (Figure 4A), the dark colors (purple and black) in the
heatmap correspond to negative intensity values and indicate
fluorescence quenching while the lighter colors (yellow and
white) indicate amplification. Testing the discriminatory
capability of the attosensor array was performed using
statistical multivariate methods, such as linear discriminant
analysis (LDA) (Figure 4B).[8,7] These approaches are rou-
tinely used to interpret and evaluate the responses from cross-
reactive sensor arrays, providing a graphical output useful to
gain an insight into the clustering of the response data, and to
calculate classification accuracy.[8,9] LDA show 100% correct
classification and reflect analyte-specific quenching amplifi-
cation and absorption–emission shifts due to varying metal
electropositivity.[9]
Finally, we deposited a conformal coating of the attosen-
sors mat onto a nitrile glove using a shadow mask (Figure 5).
When it was exposed to a solution of Co2+ ions (20.0 mm), the
sensor mat showed an immediate quenching of the fluores-
cence. This result illustrates the scalability of the described
electrospun attosensor mats including mechanical strength
and a fast attosensor response, thus suggesting a potential
utility in fabrication of wearable sensors that could be, for
example, part of a hazmat suit.
Figure 5. Attoliter reactor mats as wearable sensors: Nanofiber mat
shadow mask deposited on a glove (A), its fluorescence under black
light (365 nm; B), exposure to a solution of Co2+ ions (20.0 mm; C)
resulting in fluorescence attenuation as seen under black light (D).
In summary, we have shown a simple method for
fabrication of ultrasmall fluorescent probes and their appli-
cation as wavelength-addressable sensor arrays capable of
qualitative and quantitative sensing of metal cations. The
probes generated in situ from three label precursors and three
amines were capable of identifying eleven analytes in both
a qualitative and a quantitative fashion. Varying the probe
concentration in each attoreactor will directly affect the
dynamic range of the response of each attoreactor. The latter
opens a possibility for tuning the dynamic range of response
of the sensors to meet the profile of the target analyte. The
sensors could be deposited on various devices and objects
including protective safety wear (wearable sensors). Quanti-
tative studies and tests of complex analytes will be performed
in the near future.
Figure 4. The response to metal ions recorded from the nanofiber
arrays. A) Response heatmap: The fluorescence changes induced by
the metal ions in four emission channels show the evolution of the
signals (quenching/amplification), suggesting high variance in the
response. B) The linear discriminant analysis (LDA) shows a clear
clustering of the analytes.
The attosensor mats also provide quantitative information
regarding the metal ion content. In our preliminary experi-
ments, we used the 12 ꢀ 12 mm mats and deposited 200 nL
droplets of analyte of increasing metal concentration. The
changes in fluorescence were recorded as described above to
give intensity vs. concentration graphs (response isotherm,
see Supporting Information for examples). For example, in
the 200 nL samples, the lowest limit of detection for Co2+ ions
was 2.0 ppm and for Zn2+ ions it was 3.0 ppm (in water, pH 5).
Received: August 8, 2011
Revised: October 28, 2011
Published online: January 3, 2012
Angew. Chem. Int. Ed. 2012, 51, 2345 –2348
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
2347

.
Angewandte
Communications
[7] a) J. J. Lavigne, E. V. Anslyn, Angew. Chem. 2001, 113, 3212 –
3225; Angew. Chem. Int. Ed. 2001, 40, 3118 – 3130; b) A. P. Umali,
E. V. Anslyn, Curr. Opin. Chem. Biol. 2010, 14, 685 – 692; c) P.
Anzenbacher, Jr., P. Lubal, P. Bucek, M. A. Palacios, M. E.
Kozelkova, Chem. Soc. Rev. 2010, 39, 3954 – 3979; d) P. Anzen-
bacher, Jr., Y. Liu, M. E. Kozelkova, Curr. Opin. Chem. Biol.
2010, 14, 693 – 704.
Keywords: attoreactors · metal ions · nanotechnology · sensors ·
supramolecular chemistry
.
[1] P. Anzenbacher, Jr., M. A. Palacios, Nat. Chem. 2009, 1, 82 – 88.
[2] a) D. Li, Y. Xia, Adv. Mater. 2004, 16, 1151 – 1170; b) W.-E. Teo,
R. Inai, S. Ramakrishna, Sci. Technol. Adv. Mater. 2011, 12,
013002.
[3] a) F. S.-J. Kim, G.-Q. Ren, S. A. Jenekhe, Chem. Mater. 2011, 23,
682 – 732; b) B. Ding, M. Wang, X. Wang, X. J. Yu, G. Sun, Mater.
Today 2010, 13, 16 – 27.
[8] R. G. Brereton, Applied Chemometrics for Scientists, Wiley,
Chichester, 2007.
[9] a) C. H. Chen, J. Shi, Coord. Chem. Rev. 1998, 171, 161 – 174;
b) M. A. Palacios, Z. Wang, V. A. Montes, G. V. Zyryanov, P.
Anzenbacher, Jr., J. Am. Chem. Soc. 2008, 130, 10307 – 10314;
c) Z. Wang, M. A. Palacios, P. Anzenbacher, Jr., Anal. Chem.
2008, 80, 7451 – 7459; d) Z. Wang, M. A. Palacios, G. Zyryanov, P.
Anzenbacher, Jr., Chem. Eur. J. 2008, 14, 8540 – 8546; e) M. A.
Palacios, Z. Wang, V. A. Montes, G. V. Zyryanov, B. J. Hausch, K.
Jursꢁkovꢂ, P. Anzenbacher, Jr., Chem. Commun. 2007, 3708 –
3710.
[4] C. A. Pasternack, Ed. Monovalent Cations in Biological Systems,
CRC, Boca Raton, FL, 1990.
[5] a) T. R. Crompton, Toxicants in Terrestrial Ecosystems, Springer,
Berlin, 2006; b) T. R. Crompton, Toxicants in Aqueous Ecosys-
tems, Springer, Berlin, 2006; c) R. Dulbecco, Encyclopedia of
Human Biology, 2nd ed., Academic Press, New York, 1997.
[6] L. Prodi, M. Montalti, N. Zaccheroni, Helv. Chim. Acta 2001, 84,
690 – 706.
2348
www.angewandte.org
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2012, 51, 2345 –2348