A rapid and efficient way to dynamic creation of cross-reactive sensor arrays based on ionic liquids

Article abstract of DOI:10.1002/chem.201300789

Based on the simple counterion exchange of ionic liquids, a rapid, facile, and efficient strategy to create a cross-reactive sensor array with a dynamic tunable feature was developed, and exemplified by the construction of a sensor array for the identification and classification of nitroaromatics and explosives mimics. To achieve a good sensing system with fast response, good sensitivity, and low detection limit, the synthesized ionic liquid receptors were tethered onto a silica matrix with a macro-mesoporous hierarchical structure. Through the facile anion exchange approach, abundant ionic-liquid-based individual receptors with diversiform properties, such as different micro-environments, diverse molecular interactions, and distinctive physico-chemical properties, were easily and quickly synthesized to generate a distinct fingerprint of explosives for pattern recognition. The reversible anion exchange ability further endowed the sensor array with a dynamic tunable feature as well as good controllability and practicality for real-world application. With the assistance of statistical analysis, such as principal component analysis (PCA) and linear discrimination analysis (LDA), an optimized-size array with a good resolution was rationally established from a large number of IL-based receptors. The performed experiments suggested that the ionic-liquid-based sensing protocol is a general and powerful strategy for creating a cross-reactive sensor array that could find a wide range of applications for sensing various analytes or complex mixtures. Chemical track hound: A cross-reactive sensor array based on an ionic liquid (IL) was developed for nitroaromatics and explosives mimics. Based on the simple counterion exchange, abundant IL-based individual receptors with diverse properties are easily and quickly synthesized. The reversible anion exchange ability further endowed the sensor array with a dynamic tunable feature as well as good controllability and practicality for real-world applications (see figure). Copyright

Full text of DOI:10.1002/chem.201300789

FULL PAPER
DOI: 10.1002/chem.201300789
A Rapid and Efficient Way to Dynamic Creation of Cross-Reactive Sensor
Arrays Based on Ionic Liquids
Wei Zhu,[a] Weina Li,[a] Haowei Yang,[a] Yin Jiang,[a] Chen Wang,[a] Yu Chen,[b] and
Guangtao Li*[a]
Abstract: Based on the simple counter-
ion exchange of ionic liquids, a rapid,
facile, and efficient strategy to create a
cross-reactive sensor array with a dy-
namic tunable feature was developed,
and exemplified by the construction of
a sensor array for the identification
and classification of nitroaromatics and
explosives mimics. To achieve a good
sensing system with fast response, good
sensitivity, and low detection limit, the
synthesized ionic liquid receptors were
tethered onto a silica matrix with a
macro-mesoporous hierarchical struc-
ture. Through the facile anion ex-
change approach, abundant ionic-
liquid-based individual receptors with
diversiform properties, such as differ-
ent micro-environments, diverse molec-
ular interactions, and distinctive physi-
co-chemical properties, were easily and
quickly synthesized to generate a dis-
tinct fingerprint of explosives for pat-
tern recognition. The reversible anion
exchange ability further endowed the
sensor array with a dynamic tunable
feature as well as good controllability
and practicality for real-world applica-
tion. With the assistance of statistical
analysis, such as principal component
analysis (PCA) and linear discrimina-
tion analysis (LDA), an optimized-size
array with a good resolution was ra-
tionally established from
a
large
number of IL-based receptors. The per-
formed experiments suggested that the
ionic-liquid-based sensing protocol is a
general and powerful strategy for cre-
ating a cross-reactive sensor array that
could find a wide range of applications
for sensing various analytes or complex
mixtures.
Keywords: cross-reactive · dynamic
creation · explosive detection · ionic
liquids · sensors
Introduction
protocols that nicely classify various metal ions.[5] And quite
recently, the combinational strategy for creating synthetic
receptors or using gold nanoparticles and conjugated poly-
mers as sensing elements had been separately reported by
Hamilton et al. and Rotello et al.[6] In another approaches,
Severin and Buryak[7] highlighted the remarkable power of
differential sensing by combining simple transition metals
with dyes and also Suslick et al. targeted the optoelectronic
nose by using a series of cross-responsive dye molecules.[8]
Nevertheless, there still exist some factors that limit fur-
ther applications of the sensor array format. One of the
major challenges besetting the development of an array-
based sensing protocol is the requirement of rapidly and ef-
ficiently collecting/synthesizing a large number of receptors
with abundant diversities.[9] This is the crucial point for cre-
ating high-dispersion sensor arrays. When an array owns a
high-dispersion property, it not only offers a high level of di-
mensionality, which is beneficial to improve the discrimina-
tion capability, but this also helps to obtain a high resolution
sensor array.[5] However, the preparation of the individual
receptor is always a time- and resource-intensive process
and the creation of an array of numerous receptors by ra-
tionally designing would multiply the difficulties.[10] In this
context, the design and development of a general, rapid,
and efficient way to create receptors with abundant diversi-
ties in a cross-reactive sensor array is highly desirable.
Inspired by the mammalian olfactory system, a cross-reac-
tive sensor array has become a powerful tool for the identi-
fication and classification of analytes based on the formation
of distinct patterns generated through the interaction with
non-specific, but semi-selective receptors.[1] Up to now, there
are numerous examples in the literature on sensor arrays for
differential sensing of various analytes by using various syn-
thetic receptors.[2] For example, Anslyn et al. exploited di-
versiform synthetic receptors, which are composed of combi-
natorial peptidic arms, various metal ions, and numerous in-
dicators, to rapidly construct sensor arrays.[3] Shimizu et al.
demonstrated the utility of molecularly imprinting processes
for effectively creating differential receptors in arrays.[4] An-
zenbacher et al. described the rational design of different
classes of fluorescent receptors to create differential sensing
[a] W. Zhu, W. Li, H. Yang, Y. Jiang, C. Wang, Prof. Dr. G. Li
Department of Chemistry, Tsinghua University
Beijing, 100084 (P. R. China)
E-mail: lgt@mail.tsinghua.edu.cn
[b] Prof. Dr. Y. Chen
Department State Key Laboratory of Heavy Oil Processing
College of Chemical Engineering
China University of Petroleum
Beijing, 102249 (P.R. China)
Another limitation of the array-based sensing protocol is
the lack of a dynamic tunable feature, which would be bene-
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201300789.
Chem. Eur. J. 2013, 19, 11603 – 11612
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
11603

ficial for the prescreening of individual receptors to effi-
ciently achieve the ultimate application of a sensor array for
the identification and quantification of the target analytes.
This dynamic feature not only focuses on dynamically
tuning the individual receptors in the aspects of their prop-
erties and thus their sensing capabilities, but also the overall
characteristic and performance of the array. As mentioned
above, currently, numerous methods have been developed
to synthesize a large number of receptors for sensor array
construction. However, once these individual receptors have
been synthesized, it is not easy to further adjust their prop-
erties and sensing behaviors. As a result, further optimizing
the characteristic and performance of the created array
would often need to rebuild the whole array. Obviously, this
reconstruction would be a time-consuming and labor-inten-
sive process. Thus, for an ideal sensor array, the introduction
of the dynamic tunable feature would endow the array with
better controllability and practicality for real-world applica-
tion.
the reversible counterion exchange capability of ILs also
provides a huge room and useful means to post-modify the
sensing properties of an individual element for further cor-
recting or optimizing the sensing performance of the formed
sensor array, and endows the array with a desirable dynamic
tunable feature. Indeed, a series of sensor elements was fac-
ilely produced by the exchange of the anion of the synthe-
sized fluorescent ionic liquid with six different counterions
À
such as ClÀ, NO3 , and Tf2NÀ (Tf=triflate) ions, and used as
an array for nitroaromatics and explosives mimics detection
and discrimination. The described dynamic tunable feature
allows a rapid and efficient optimization of the formed
array. Finally, based on the evaluation by using statistical
analysis such as principal component analysis (PCA) and
linear discrimination analysis (LDA), a cross-reactive sensor
array with good discrimination ability and a high resolution
was established.
Ionic liquids (ILs), a class of compounds consisting of or-
ganic cations and anions, exhibit a series of unique proper-
ties including negligible vapor pressure, high ionic conduc-
tivity, a wide electrochemical window, and in particular, a
facile ion exchange capability.[11] By virtue of simple ex-
change of the counterion, the IL-based moiety would proc-
ess abundant functionalities, different microenvironments
(the local hydrophilicity and hydrophobicity), and diverse
molecular interactions as well as distinctive physicochemical
properties.[12] These distinct properties and easy synthesis
makes ILs ideal candidates for sensing application. So far,
there are some reports on the creation of molecular recep-
tors by just simply utilizing different ionic liquids.[13] Howev-
er, the exploration of the excellent counterion exchange ca-
pability of ILs for a rapid, efficient, and dynamic synthesis
of diversiform synthetic receptors for the creation of sensor
array has not been reported.
Results and Discussion
Creation of the IL-based sensing elements with numerous
diversities: A three-step approach was employed to synthe-
size the ionic-liquid-functionalized silane, as shown in
Scheme 1. To achieve a good sensing system with a fast re-
sponse, good sensitivity, and low detection limit, the synthe-
sized fluorescent IL sensing element was tethered onto a
silica matrix with a macro-mesoporous structure.[14] Fig-
ure 1A shows the well-defined inverse replication of the
formed polystyrene (PS) colloidal crystal arrays on a glass
substrate (Figure S1 in the Supporting Information). The
macropore size is about 180 nm with a wall thickness of
20 nm. XRD pattern displays a broad diffraction peak at
0.828, which is typical for a worm-like mesostructure (Fig-
ure 1C), which was confirmed by TEM measurement (Fig-
ure 1B). Clearly, the mesochannels are distributed within
the walls of the 3D-ordered macroporous silica and connect
the adjacent macropores, which would facilitate the rapid
diffusion of analyte molecules. The data of the Brunauer–
Emmett–Teller (BET) surface area (198 m2gÀ1) and the pore
size (10 nm) from N2 adsorption–desorption measurements
further confirmed the hierarchical pore structure of the pre-
pared silica films.
The correct incorporation of the synthesized ionic liquid
into the silica matrix was proven by fluorescence measure-
ment and FTIR spectroscopy. As shown in Figure 1D, the
characteristic emission peaks of the PAI cation core in the
silica film at wavelengths of l=345, 360, and 389 nm were
observable comparable to those of the IL monomer in etha-
nol solution (Figure S2 in the Supporting Information). Also
the characteristic FTIR absorptions at n˜ =3061, 3025, 1602,
and 1581 cmÀ1 confirmed the presence of the used ionic
liquid moieties in the silica films (Figure S3a in the Support-
ing Information).
Herein, based on the unique counterion exchange proper-
ty of ionic liquids, we presented a new strategy for rapidly
and efficiently creating cross-reactive sensor arrays with dy-
namic tunable feature. To achieve a good sensing system
with a fast response, good sensitivity, and low detection
limit, the synthesized IL receptors were tethered onto a
silica matrix with a macro-mesoporous hierarchical struc-
ture. As a demonstration, an ionic liquid with a fluorescent
phenanthroACHTUNGTRENNUNG
[9,10-d]imidazolium (PAI) cation core and IÀ as
counterion was synthesized as a primary sensor element
(Scheme 1), in which the cationic part acted as a core-signal-
ing unit, and the described strategy was exemplified by the
formation of a sensor array for nitroaromatics and explo-
sives mimics identification and classification. It was found
that the LUMO–HOMO energies, the excitation and emis-
sion spectra, the fluorescent lifetimes as well as the sensing
capability of the synthesized IL could be significantly influ-
enced and tuned by coupling different counterions, indicat-
ing that from one single IL sensor element a big sensor ele-
ment pool with rich diversities could be conceivably created
by a simple exchange process of the counterions. Moreover,
The anion exchange possibility of the immobilized IL
moieties in silica films was firstly examined by soaking for
2 h. In our case, NaCl, NaBr, NaNO3, NaPF6, and LiTf2N
11604
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 11603 – 11612

Cross-Reactive Sensor Arrays Based on Ionic Liquids
FULL PAPER
mation, that the HOMO and LUMO energies as well as the
corresponding band gaps could be greatly changed through
the introduction of different counterions. This result implied
that the counterion exchange of the synthesized IL sensing
element offered a feasible way for effectively tuning its sens-
ing properties.
To experimentally confirm the possibility of creating vari-
ous sensing elements by a simple exchange of the counter-
ion, the optical properties of the prepared silica films with
different counterions were also measured as shown in
Figure 3. All emission spectra showed three primary charac-
teristic emission peaks at l=345, 360 and 389 nm but obvi-
ous differential intensities at each peak. Similarly, although
the excitation spectra were clearly divided into three zones:
l=240–270, 270–308, and 308–360 nm, the PAI core showed
different excitation patterns with different anions. Also the
induced diversities by coupling different counterions could
be found in the aspect of excited-state lifetimes. As shown
in Table S2 and Figure S5 in the Supporting Information, all
six decay profiles were fitted by tri-exponential decay kinet-
ics to show three nanosecond time domains. However, due
to the variation of the counterions, these three time domains
in each sample took different percentages.
All results described above proved the feasibility that a
simple counterion exchange approach could be exploited for
generating receptors or sensor elements with rich diversities.
Virtually, ionic liquids offer an unlimited tunability. In addi-
tion, based on the concept of “task-specific” ionic liquids,
functional groups as anions can also be introduced into IL
units, leading to ILs with specific functions. Clearly, these
fascinating features may provide an attractive opportunity
and assurance for rapidly and efficiently creating a big
sensor element pool with rich diversities from one basic IL
sensor element.
Scheme 1. Top) Three-step approach to synthesize the PAI-based ionic
liquid silane core for the counterion exchange. Middle) Six different
anions used to create diversiform sensing elements for a sensor array
construction. Bottom) Structure of the six nitroaromatics and explosives
mimics for identification and classification. TNT=trinitrotoluol, DNT=
dinitrotoluol, DN=dinitrobenzene, CB=chlorobenzene, NQ=nitroqui-
noline, PETN=pentaerythritol tetranitrate.
were used. As shown in Figure S3 in the Supporting Infor-
mation, the characteristic FTIR absorptions of the corre-
sponding anions appeared at n˜ =1385, 1356, and 1681 cmÀ1
separately, confirming the successful introduction of the dif-
ferent anionic groups. In consistent with the FTIR results,
the change in the contact angle on the resulting silica films
further proved the occurrence of the counterion exchange
(Figure S4 in the Supporting Information). These results
clearly indicate that the unique counterion exchange feature
of ILs allows for rapidly accessing a series of derivates from
one basic structure.
Ionic liquids own a binary structure with cationic and
anionic parts. From previous studies, it is known that the
coupling of a cation core with different counterions would
afford ionic liquids with totally different properties, such as
different micro-environments, diverse molecular interac-
tions, and distinct physicochemical properties. In our case,
the influence of the coupled counterion on the energy of the
HOMO and LUMO as well as on the band gap (eV) of the
PAI cation core was firstly calculated by using density func-
tional theory at the B3LYP/6-31G(d) level.[15] It was found,
as shown in Figure 2 and Table S1 in the Supporting Infor-
Construction of a cross-reactive sensor array for nitro-
AHCTUNGTREGaNNUN romatics and explosives mimics detection: Rapid detection
and discrimination of hidden explosives are very important
aspects concerning homeland security as well as environ-
mental and humanitarian safety.[16] To demonstrate the great
potential of the strategy described above in the construction
of a cross-reactive sensor array, the sensor elements with
À
five different counterions (ClÀ, BrÀ, NO3 , AlaÀ, and GlnÀ)
produced by a simple counterion exchange of the synthe-
sized fluorescent IL 3 were used as an initial array for the
discrimination of six nitroaromatics and explosives mimics
(see Scheme 1). In our work, the analyte sensing was done
by following the same method described by Swager et al.,
which was also employed in most of reported work in the
explosive detection field. Actually, due to the macro-meso-
porous hierarchical structure in our system, the detection
limit could be far less than the concentration at the saturat-
ed vapor pressure.
The differential sensing capability of each sensor element
was evaluated, and the sensing measurement was repeated
five times for each analyte. After measurement of their ini-
tial fluorescent intensity, each sensor element was exposed
Chem. Eur. J. 2013, 19, 11603 – 11612
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11605

G. Li et al.
Figure 1. A) SEM (scale bar=300 nm) and B) TEM (scale bar=100 nm) images of the IL-doped silica film with a macro-mesoporous structure. C) XRD
pattern and N2 adsorption–desorption isotherms (insert) of the hybrid silica film. D) Fluorescence excitation and emission spectra of the PAI-based ionic
liquid molecules in the formed silica film.
analyte varied in a wide range, indicating that counterion
exchange is a powerful mean for effective adjustment of the
sensing properties of the IL sensor element.
The observed recognition patterns of the IL-based sensor
array toward six different samples were subjected to a prin-
cipal component analysis (PCA) to generate a clustering
map for more practical visualization. PCA is a non-super-
vised mathematical method that can transform a number of
correlated variables into uncorrelated variables called prin-
cipal components (PCs).[5] The first principal component
(PC1) contained the highest degree of variance, and the
other PCs followed in the order of decreasing variance. In
this way, the PCA concentrates the most significant charac-
teristics of the whole data into a lower dimensionally space
without losing much information. In Figure 5, the PCA of
the data obtained from six samples on a 5-IL sensor array is
shown in a 3D-Euclidean space, which contained 98.69% of
the information. The 3D-PCA plot contains points for each
of the nitrated samples measured five times. The responses
for each sample were clustered into six distinct groups, dem-
onstrating a good reproducibility. This result suggests that
the initial IL sensor array fabricated with randomly selected
counterions, already shows a good detection and discrimina-
tion ability for the analytes.
Figure 2. Band gap energies of the fluorescent PAI-based ionic liquid
sensor element with different counterions, which were calculated by
using density functional theory at the B3LYP/6-31G(d) level for the opti-
mized geometries.
to six different nitroaromatics and explosives mimics, and
their fluorescence intensity was scanned again. Figure 4
shows the plot of the change in the fluorescence intensity as
a percentage of the original (((F0ÀF)/F0)100%), which gives
each detected analyte a 5D fingerprint. Taking a close com-
parison in the fingerprints, we found that the nitroaromatics
TNT, DNT, and DN have a better quenching efficiency, and
the quenching reached 60% in 120 s, which is indicative of a
good sensitivity. Furthermore, the best sensing performance
for each analyte was irregularly hopping between the sen-
sors in a 5-IL array system. To better illustrate the differen-
tial sensing properties of the IL sensor elements, the maxi-
mum and minimum quenching efficiencies are displayed in
Figure 4F. Clearly, dependent on the coupled counterion,
the quenching efficiency of the IL sensor element for each
Sensor array with a dynamic tunable feature: The introduc-
tion of dynamic adjustability in a sensor array could endow
the array a better controllability and practicality. Herein,
the dynamic feature focused on the aspects of the properties
and sensor capability, and the overall performance of the
11606
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 11603 – 11612

Cross-Reactive Sensor Arrays Based on Ionic Liquids
FULL PAPER
efficiently change the sensing
properties of an individual
sensor element, providing
a
desirable and dynamic means
to turn the performance of the
established sensor array.
As mentioned above, an
array with four IL sensor ele-
ments (named array 1) was
firstly created with randomly
selected counterions (ClÀ, BrÀ,
AlaÀ, and Tf2NÀ), which shows
a modest detection and dis-
crimination for six nitrated
samples, see Figure S6 in the
Supporting Information. Based
on the simple anion displace-
ment, the arrays 2–4 could be
facilely derived from the estab-
lished array 1, as shown in
Figure 7. In array 1, the four
analytes (TNT, DNT, DN, and
CB) clustered in close proximi-
ty. Particularly, the data be-
tween TNT and DNT had
some degree of overlap for
hard classification. In array 2,
which was produced by replac-
ing BrÀ with PFÀ6 , no clear im-
provement of the discrimina-
tion was achieved. However,
in array 3 the responses for
each analyte were clustered
into six distinct groups, sug-
gesting a good discrimination.
The average Euclidean dis-
tance between the three
Figure 3. Fluorescence excitation spectra (red line, lEm =387 nm) and emission spectra (black line, lEx
260 nm) of the PAI-based ionic liquid sensor element with six different counterions in the formed silica films.
=
array. In our case, due to the reversible anion exchange abil-
ity, it offered the possibility to own this dynamic feature in
our ionic-liquid-based arrays.
classes of TNT, DNT, and DN is 0.39. Fortunately, after
carefully screening the counterions in array 1, array 4 with a
good discrimination was established. The calculated average
Euclidean distance between the three classes of TNT, DNT,
and DN increased from 0.39 to 0.98. The “jack-knife” analy-
sis showed 100% correct classification in the formed sensor
array when all three roots were taken into consideration.
These results demonstrate that the dynamic tunable feature
in the IL-based sensor arrays is very useful for convenient
and efficient optimization or extension of the formed sensor
array.
In the above discussion, we have proven that the individu-
al receptors, which were generated by simple counterion ex-
change, had different properties. Due to the reversible anion
exchange, these properties were dynamically tunable. To fur-
ther demonstrate the dynamic adjustment in the sensing
properties of the sensor element, we explored the time-de-
pendent fluorescence quenching curves upon exposure to sa-
turated dinitrobenzene and 2,4-dinitrotoluene vapors, which
are shown in Figure 6. In the case of the IL sensor with
AlaÀ as counterion, only 28% of the initial fluorescence was
quenched by DN after 60 s exposure (Figure 6A). However,
the displacement of the counterion with PFÀ6 led to a re-
markable improvement of the quenching efficiency of the
IL sensor. A fluorescence quenching of 56% occurred after
60 s, and 70% quenching was detected after 120 s. A similar
phenomenon was also observed for the case of DNT (Fig-
ure 6B). Obviously, the displacement of the counterion can
Rational design of an IL-based sensor array: The anion ex-
change method was a powerful strategy to create a large
amount of sensor elements for array construction. However,
it was not true that the more sensor elements, the better
sensing performance. During the development of array-
based detection systems, it had been proven that increasing
the number of receptors would increase the overall sensing
performance. This was attributed to the increase in the di-
Chem. Eur. J. 2013, 19, 11603 – 11612
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11607

G. Li et al.
versity of receptors, which
leaded to an increase in the re-
sponsive information.[6] How-
ever, after a certain point, in-
creasing the number of recep-
tors may lead to a little effect
suggesting
an
optimized
number for analyte discrimina-
tion.
To demonstrate the similar
property in the IL-based
sensor array, we first construct
a primary array A with seven
individual sensor elements.
Following the similar data
processing method described
in the work of Anzenbacher
et al„[5] arrays B (five ele-
ments) and C (four elements)
were sequentially derived from
the primary array A (Figure 8).
During the minimization pro-
cedure, the correct classifica-
tion and good resolving power
were the two most concerned
factors. With the assistance of
LDA and the “jack-knife”
analysis based on three roots,
all the raw array A as well as
the reduced arrays B and C
showed 100% correct classifi-
cation in all 30 cases. Although
it was notable that even after
excluding the two (S1, S6) or
three (S1, S6, S7) out of the
raw array A, the PCA score
Figure 4. A–E) Fingerprints for the five nitroaromatics and explosives mimics in the five-element of the PAI-
based ionic liquid sensor arrays. F) Maximum and minimum fluorescence quenching of the PAI-based ionic
liquid sensor elements with five counterions upon exposure to the six analytes.
plot in array B and
C still
showed good clustering with
no evident overlap between the samples. Further compari-
son of the resolving power of each array was performed
only between the arrays B and C for the smaller array size.
As shown in Figure 8, the average Euclidean distances be-
tween the samples were 2.88 for array B and 2.53 for
array C, respectively. The increased ratio was about 14%,
indicating that array B had a better discriminatory perform-
ance and resolving power. These observations suggest that
by optimizing the number and diversity of the receptors in
the array, we could really obtain a good resolution array
with optimized size.
The results described above were encouraging. Based on
the anion exchange approach, a rapid, facile, and efficient
way to create a cross-reactive sensor array with dynamic ad-
justability was realized for the identification and classifica-
tion of nitroaromatics and explosives mimics. Herein, the
anion exchange is the key feature for the IL-based sensor
array. In our work, although the hypothesis where different
counterions affect the response to different analytes has
Figure 5. PCA score plot of the three PCs describing approximately
98.7% of the total variance. The PCA score plot shows six clustering for
all samples.
11608
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 11603 – 11612

Cross-Reactive Sensor Arrays Based on Ionic Liquids
FULL PAPER
synthesized IL could be significantly influenced and tuned
by coupling different counteranions, indicating that from
one single IL sensor element a big sensor element pool with
rich diversities could be conceivably created by simple ex-
change process of counteranions. Moreover, the reversible
counterion exchange capability of IL also provides a huge
room and useful means to post-modify the sensing proper-
ties of individual element for further correcting or optimiz-
ing the sensing performance of the formed sensor array, and
so endows the array with desirable dynamic tunable feature.
With the assistance of statistical analysis such as principal
component analysis (PCA) and linear discrimination analy-
sis (LDA), an optimized-size array with high resolution can
be rationally established from a large number of the IL-
based receptors. Although PAI-based luminescent ionic
liquid was used for present study, these results are expected
to be universal for other types of ionic liquids. In addition,
based on the concept of “task-specific” ionic liquids, func-
tional groups as anions can also be introduced into IL units,
leading to ILs with specific functions. Thus, we believed our
developed strategy offered a powerful method for facilely
and efficiently constructing sensor array, which could find a
wide range of applications.
Figure 6. Time-dependent fluorescence quenching of the PAI-based
sensor elements with PFÀ6 (top curves) and ClÀ (bottom curves) as coun-
terions upon exposure to A) DN and B) DNT.
Experimental Section
The general methods used for compound identification as well as pur-
chase information are given in the Supporting Information.
been proven by a series of experiments, however, the corre-
sponding mechanism at the molecular lever is still not clear.
In the future we will focus on the investigation of the mech-
anism both in a theoretical and experimental way to gain
deep understanding of this kind of system and thus to afford
high-performance sensor array devices. In addition, although
a PAI-based luminescent ionic liquid was selected for sens-
ing, actually other types of luminescent ionic liquids with
different excitation wavelengths could also be used to broad
the sensor array application. Recently, many ionic liquids
had also been found to selectively enrich analytes. For ex-
ample, in the work of Tao et al., [BMIM]PF6 (BMIM=1-
butyl-3-methylimidazolium) was used for selectively enrich-
ing the aromatic explosives.[16] Thus, this property of ionic
liquids would add a new spatial dimension in the IL-based
array to further increase the discriminatory performance
and resolving power of the sensor array.
Synthesis of 1H-phenanthroACHTNUTRGNE[NUG 9,10-d]imidazole (1): 9,10-Phenanthrene-
quinone (5.0 g, 0.024 mol) was dissolved in formaldehyde (3.85 mL,
37 wt%) containing glacial acetic acid (93.33 mL, 1.63 mol) and ammoni-
um acetate (38.45 g, 0.5 mol). The resulting reaction mixture was heated
to reflux for 4 h. After cooling, the reaction mixture was diluted with
water (100 mL) and neutralized with concentrated aqueous ammonia
(30 wt%) to pH 7 to get a light cream precipitate. After filtering, the ob-
tained solid was washed with water, acetone, dichloromethane, and ethyl
ether, and then dried to afford the desired product. 1H NMR (400 MHz,
[D6]DMSO): d=8.85 (d, J=8 Hz, 2H; Ar-H), 8.55–8.35 (m, 3H; NH,
Ar-H), 8.32 (s, 1H; N=CH), 7.70 (t, J=14 Hz, 2H; Ar-H), 7.62 ppm (t,
J=14 Hz, 2H; Ar-H); MS (ESI+): m/z: 437.2 [2M+H]+.
Synthesis of 1-butylphenanthroACHTUNTRGNEU[GN 9,10-d]imidazole (2): Compound 1 (4.0 g,
18.4 mmol) and sodium hydroxide (0.8 g, 20 mmol) were resolved in
DMSO (40 mL). The reaction mixture was stirred at room temperature
for 2 h. After addition of butyl bromide (2.0 mL, 18.4 mmol), the reaction
solution was stirred at room temperature for further 25 min, then the
temperature was increased to 378C. After overnight reaction under stir-
ring, the solvent was removed and the residue was taken up in acetoni-
trile (30 mL). After filtration, the solvent was removed and the residue
was further purified by trituration with ethyl ether to give the desired
product. 1H NMR (400 MHz, [D6]DMSO): d=8.94 (d, J=9 Hz, 1H; Ar-
H), 8.85 (d, J=9 Hz, 1H; Ar-H), 8.53 (d, J=9 Hz, 1H; Ar-H), 8.35 (d,
J=9 Hz, 1H; Ar-H), 8.26 (s, 1H; N=CH), 7.80–7.60 (m, 4H; Ar-H), 4.71
(t, J=7 Hz, 2H; N-CH2), 1.93–1.83 (m, 2H; NCH2CH2), 1.40–1.30 (m,
2H; CH2CH3), 0.92 ppm (t, J=9 Hz, 3H; CH3); MS (ESI+): m/z: 275.2
[M+H]+.
Conclusion
In summary, based on the simple counterion exchange of
ionic liquid, a rapid, facile and efficient strategy to create
cross-reactive sensor array with dynamic tunable feature
was developed, and exemplified by the construction of
sensor array for six nitroaromatics and explosives mimics
identification and classification. It was found that the
LUMO-HOMO energies, excitation and emission spectra,
fluorescent lifetimes as well as the sensing capability of the
Synthesis of 1-butyl-3-(3-trimethoxysilyl)propyl)phenanthroACTHNUGRTNEUNG[9,10-d]imi-
dazol-3-ium iodide (3). Compound 2 (0.5 g, 1.8 mmol) and (3-iodoprop-
yl)-trimethoxysilane (2.4 g, 8.3 mmol) were added into a pressure flask.
The flask was vacuumed, and the reaction mixture was stirred at 1008C
for 48 h. After removal of unreacted (3-iodopropyl)trimethoxysilane, the
red pure product was obtained. 1H NMR (300 MHz, [D6]DMSO): d=
Chem. Eur. J. 2013, 19, 11603 – 11612
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11609

G. Li et al.
Figure 7. Schematic illustration of the reversible anion exchange process to demonstrate the dynamic tunable feature of the ionic-liquid-based sensor
arrays. Array 1 is firstly created with four randomly selected counterions (ClÀ, BrÀ, AlaÀ, and Tf2NÀ). Arrays 2–4 are further facilely derived from the es-
tablished array 1 by simple anion displacement. PCA and LDA are used for statistical analysis. Among these four arrays, array 4 has the best accuracy
~
!
&
*
and resolving power. =TNT, =DNT, =DN, =CB, 3 =NQ, " =PETN.
9.79 (s, 1H; N-CH), 9.10 (d, J=8.7 Hz, 2H; Ar-H), 8.54 (d, J=6.9 Hz,
2H; Ar-H), 7.95–7.85 (m, 4H; Ar-H), 4.95–4.85 (m, 4H; N-CH2), 3.46 (s,
6H; OCH3), 2.07–1.99 (m, 4H; NCH2CH3), 1.43–1.53 (m, 2H; CH2CH3),
0.97 (t, J=6.7 Hz, 3H; CH3), 0.83–0.66 ppm (m, 2H; Si-CH2); MS
(ESI+): m/z: 437.3 [MÀIÀ]+.
Formation of the ionic-liquid-doped silica films with macro-mesopore
structure: The PAI-bridged silane 3 (50 mg, 8.86ꢁ10À2 mmol), tetraethyl
orthosilicate (TEOS) (1.04 g, 5 mmol), and ethanol (5.82 g) were dis-
solved in water (0.73 g, 41 mmol) containing 1mHCl (105 mL), and the re-
sulting reaction mixture was heated to reflux at 608C for 30 min. Then,
surfactant F127 (0.26 g) and anhydrous ethanol (5.82 g) were added for
further 30 min stirring. The final molar ratio of the reactants was 1
TEOS/PAI/EtOH/H2O/HCl/F127=1:1.78ꢁ10À2:50.6:9.28:2.1ꢁ10À2:4.12ꢁ
10À3. Then the synthesized precursor was infiltrated into PS colloidal
crystal voids until they became transparent. To increase the stability of
the formed porous framework, the obtained films were aged in oven
(608C) for eight days. Finally, the used templates (PS spheres and F127)
were extracted by toluene and ethanol to form the PAI-functionalized
silica films with macro-mesoporous structure.
Anion exchange procedure: The synthesized silica films were soaked into
0.2m aqueous solutions of the corresponding salt, such as NaCl, NaBr,
NaNO3, NaPF6, LiTf2N, NaAla, and NaGln. After 2 h anion exchange re-
action, the films were washed thoroughly with water and dried in air.
Before the sensing experiments, all the prepared films were vacuumed at
808C for 2 h.
Acknowledgements
We gratefully acknowledge financial support from the National Science
Foundation of China (No. 50873051, 205333050, 21273284), MOST
(2007AA03Z307), and Transregional Project (TRR61).
11610
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 11603 – 11612

Cross-Reactive Sensor Arrays Based on Ionic Liquids
FULL PAPER
Figure 8. Schematic illustration of the rational process for optimizing the number of sensor elements in an array. A) PCA for all sensors (S1–S7) shows
that the main contributors for the dispersion are S2–S5, and S7 on the PCs. B) Sensor elements S1 and S6 are excluded for an analysis again with PCA.
PCA shows that the main contributors are S2–S5. C) S7 is excluded and PCA is used for further statistical analysis. Cross-validated LDA shows that all
the sensor arrays A–C have a 100% accurate classification.
J. T. McDevitt, Angew. Chem. 2005, 117, 6533–6536; Angew. Chem.
[1] a) K. J. Albert, N. S. Lewis, C. L. Schauer, G. A. Sotzing, S. E. Stit-
zel, T. P. Vaid, D. R. Walt, Chem. Rev. 2000, 100, 2595–2626; b) F.
Rçck, N. Barsan, U. Weimar, Chem. Rev. 2008, 108, 705–725.
Int. Ed. 2005, 44, 6375–6378; Angew. Chem. 2005, 117, 6533–6536;
c) A. T. Wright, E. V. Anslyn, J. T. McDevitt, J. Am. Chem. Soc.
2005, 127, 17405–17411; d) T. Zhang, N. Y. Edwards, M. Bonizzoni,
E. V. Anslyn, J. Am. Chem. Soc. 2009, 131, 11976–11984.
[2] a) T. A. Dickinson, J. S. White, J. S. Kauer, D. R. Walt, Nature 1996,
382, 697–700; b) M. N. Stojanovic, E. G. Green, S. Semova, D. B.
Nikic, D. W. Landry, J. Am. Chem. Soc. 2003, 125, 6085–6089; c) E.
Green, M. J. Olah, T. Abramova, L. R. Williams, D. Stefanovic, T.
Worgall, M. N. Stojanovic, J. Am. Chem. Soc. 2006, 128, 15278–
15282; d) A. Duarte, A. Chworos, S. F. Flagan, G. Hanrahan, G. C.
Bazan, J. Am. Chem. Soc. 2010, 132, 12562–12564; e) L. D. Bonifa-
cio, D. P. Puzzo, S. Breslav, B. M. Wiley, A. McGree, G. A. Ozin,
Adv. Mater. 2010, 22, 1351–1354; f) F. Samain, S. Ghosh, Y. N. Teo,
E. T. Kool, Angew. Chem. 2010, 122, 7179–7183; Angew. Chem. Int.
Ed. 2010, 49, 7025–7029; g) S. S. Tan, S. J. Kim, E. T. Kool, J. Am.
Chem. Soc. 2011, 133, 2664–2671; h) M. E. Germain, M. J. Knapp, J.
Am. Chem. Soc. 2008, 130, 5422–5423; i) Y. L. Liu, M. A. Palacios,
P. Anzenbacher, Jr., Chem. Commun. 2010, 46, 1860–1862.
[4] a) N. T. Greene, S. T. Morgan, K. D. Shimizu, Chem. Commun. 2004,
1172–1173; b) N. T. Greene, K. D. Shimizu, J. Am. Chem. Soc. 2005,
127, 5695–5700.
[5] a) M. A. Palacios, Z. Wang, V. A. Montes, G. V. Zyryanov, P. Anzen-
bacher, Jr., J. Am. Chem. Soc. 2008, 130, 10307–10314; b) M. A.
Palacios, R. Nishiyabu, M. Marquez, P. Anzenbacher, Jr., J. Am.
Chem. Soc. 2007, 129, 7538–7544; c) G. V. Zyryanov, M. A. Palacios,
P. Anzenbacher, Jr., Angew. Chem. 2007, 119, 7995–7998; Angew.
Chem. Int. Ed. 2007, 46, 7849–7852; Angew. Chem. 2007, 119, 7995–
7998.
[6] a) U. H. F. Bunz, V. M. Rotello, Angew. Chem. 2010, 122, 3338–
3350; Angew. Chem. Int. Ed. 2010, 49, 3268–3279; b) C. C. You,
O. R. Miranda, B. Gider, P. S. Ghosh, I. B. Kim, B. Erdogan, S. A.
Krovi, U. H. F. Bunz, V. M. Rotello, Nat. Nanotechnol. 2007, 2, 318–
323; c) M. De, S. Rana, H. Akpinar, O. R. Miranda, R. R. Arvizo,
U. H. F. Bunz, V. M. Rotello, Nat. Chem. 2009, 1, 461–465; d) L.
[3] a) S. C. McCleskey, M. J. Griffin, S. E. Schneider, J. T. McDevitt,
E. V. Anslyn, J. Am. Chem. Soc. 2003, 125, 1114–1115; b) A. T.
Wright, M. J. Griffin, Z. L. Zhong, S. C. McCleskey, E. V. Anslyn,
Chem. Eur. J. 2013, 19, 11603 – 11612
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
11611

G. Li et al.
Baldini, A. J. Wilson, J. Hong, A. D. Hamilton, J. Am. Chem. Soc.
2004, 126, 5656–5657; e) H. Zhou, L. Baldini, J. Hong, A. J. Wilson,
A. D. Hamilton, J. Am. Chem. Soc. 2006, 128, 2421–2425; f) D. Mar-
gulies, A. D. Hamilton, Angew. Chem. 2009, 121, 1803–1806;
Angew. Chem. Int. Ed. 2009, 48, 1771–1774.
[14] a) A. Stein, F. Li, N. R. Denny, Chem. Mater. 2008, 20, 649–666;
b) F. Li, Z. Y. Wang, N. S. Ergang, C. A. Fyfe, A. Stein, Langmuir
2007, 23, 3996–4004; c) Z. Y. Yuan, B. L. Su, J. Mater. Chem. 2006,
16, 663–677; d) M. Antonietti, G. A. Ozin, Chem. Eur. J. 2004, 10,
28–41.
[7] a) A. Buryak, K. Severin, Angew. Chem. 2004, 116, 4875–4878;
Angew. Chem. Int. Ed. 2004, 43, 4771–4774; b) A. Buryak, K. Sever-
in, J. Am. Chem. Soc. 2005, 127, 3700–3701; c) A. Buryak, K. Sever-
in, Angew. Chem. 2005, 117, 8149–8152; Angew. Chem. Int. Ed.
2005, 44, 7935–7938.
[8] a) S. H. Lim, L. Feng, J. W. Kemling, C. J. Musto, K. S. Suslick, Nat.
Chem. 2009, 1, 562–567; b) N. A. Rakow, K. S. Suslick, Nature 2000,
406, 710–713; c) C. Zhang, K. S. Suslick, J. Am. Chem. Soc. 2005,
127, 11548–11549.
[9] A. P. Umali, E. V. Anslyn, Curr. Opin. Chem. Biol. 2010, 14, 685–
692.
[10] H. S. Hewage, E. V. Anslyn, J. Am. Chem. Soc. 2009, 131, 13099–
13106.
[11] N. V. Plechkova, K. R. Seddon, Chem. Soc. Rev. 2008, 37, 123–150.
[12] E. Borzin, A. Shemesh, C. H. Ronen, Y. Gerchikov, N. Tessler, Y.
Eichen, J. Phys. Org. Chem. 2010, 23, 1108–1113.
[13] a) X. X. Jin, L. Yu, D. Garcia, R. X. Ren, X. Q. Zeng, Anal. Chem.
2006, 78, 6980–6989; b) K. Y. Hou, A. Rehman, X. Q. Zeng, Lang-
muir 2011, 27, 5136–5146; c) C. H. Xiao, A. Rehman, X. Q. Zeng,
Anal. Chem. 2012, 84, 1416–1424.
[15] a) A. D. Becke, J. Chem. Phys. 1993, 98, 5648; b) C. Lee, W. Yang,
R. G. Parr, Phys. Rev. B 1988, 37, 785.
[16] a) S. W. Thomas, G. D. Joly, T. M. Swager, Chem. Rev. 2007, 107,
1339–1386; b) D. T. McQuade, A. E. Pullen, T. M. Swager, Chem.
Rev. 2000, 100, 2537–2574; c) J. C. Sanchez, W. C. Trogler, J. Mater.
Chem. 2008, 18, 3143–3156; d) M. E. Germain, M. J. Knapp, Chem.
Soc. Rev. 2009, 38, 2543–2555; e) T. Naddo, Y. K. Che, W. Zhang,
K. Balakrishnan, X. M. Yang, M. Yen, J. C. Zhao, J. S. Moore, L.
Zang, J. Am. Chem. Soc. 2007, 129, 6978–6979; f) Y. Salinas, R. M.
MꢂÇez, M. D. Macos, F. Sancenꢃn, A. M. Costero, M. Parra, S. Gil,
Chem. Soc. Rev. 2012, 41, 1261–1296; g) A. Ponnu, N. Y. Edwards,
E. V. Anslyn, New J. Chem. 2008, 32, 848–855; h) A. D. Hughes,
I. C. Glenn, A. D. Patrick, A. Ellington, E. V. Anslyn, Chem. Eur. J.
2008, 14, 1822–1827.
[17] E. S. Forzani, D. L. Lu, M. J. Leright, A. D. Aguilar, F. Tsow, R. A.
Lglesias, Q. Zhang, J. Lu, J. H. Li, N. J. Tao, J. Am. Chem. Soc. 2009,
131, 1390–1391.
Received: February 28, 2013
Published online: July 19, 2013
11612
www.chemeurj.org
ꢀ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 11603 – 11612