amino acid-based fluorescent chiral ionic liquid for enantiomeric recognition

Amino Acid-Based Fluorescent Chiral Ionic Liquidfor Enantiomeric Recognition

David K. Bwambok, Santhosh K. Challa, Mark Lowry, and Isiah M. Warner*

Department of Chemistry, Louisiana State University, Baton Rouge, Louisiana 70803

We report on the synthesis and characterization of a newfluorescent chiral ionic liquid (FCIL), L-phenylalanineethyl ester bis(trifluoromethane) sulfonimide (L-PheC2-NTf2), capable of serving simultaneously as solvent,chiral selector, and fluorescent reporter in chiralanalytical measurements. Enantiomers of differentanalytes, including fluorescent and nonfluorescentcompounds, with a variety of structures were shownto induce wavelength- and analyte-dependent changesin the fluorescence intensity of this FCIL. This systemmay provide both chemo- and enantioselectivity towardmultiple analytes simultaneously. The newly synthe-sized FCIL, derived from commercially available L-phenylalanine ethyl ester chloride and lithium bis(tri-fluoromethane) sulfonamide, was obtained as liquid atroom temperature and is stable to thermal decomposi-tion up to 270 °C. Absorption and fluorescenceproperties of neat L-PheC2NTf2 were complex. Whilethe absorption properties were similar to phenylalaninewith a weakly absorbing tail extending beyond 400 nm,multiple excitation and emission bands were observedin its Excitation-Emission Matrix (EEM). A prominentexcimer emission displayed the greatest intensity of allemission bands, and a long-wavelength emission shiftedtoward the red with increasing excitation wavelength.These different spectral regions were shown to responddifferently toward several analytes, including sugarssuch as glucose and mannose, making this an idealsystem to exploit the multidimensional properties offluorescence. The unique properties of L-PheC2NTf2combined with EEMs resulted in reliable identificationof different enantiomers and measurement of enantio-meric composition. Importantly, the choice of excita-tion and emission wavelength regions was an importantvariable shown to improve prediction of enantiomericcomposition.

Chiral analysis continues to receive considerable attention.Enantiomers of chiral drugs often differ in their pharmacologicalactivity and pharmacokinetic profile because the molecules theyinteract with in biological systems are also optically active. Oneenantiomer may have the desired medicinal activity, while theother may be inactive or show a qualitatively different effect or

quantitatively different potency and/or lead to toxic effects.1 Withinthe past decade, the marketing of single enantiomers of chiraldrugs has become more common, with single enantiomers nowbeing the leading component of drugs approved today.2,3 This isin part due to the Food and Drug Administration’s increasedregulation following their 1992 policy statement regarding thedevelopment of new stereoisomeric drugs,4 combined with thepharmaceutical industry’s use of the “chiral switching” strategyto manage drug life cycles and remarket established racematesas single enantiomers.2,5 Thus, methods to quantitatively deter-mine enantiomeric purity have become highly desirable.

Various chiral selectors have been developed for enantiomericrecognition. These include cyclodextrins,6-12 antibodies,13-15

antibiotics,16,17 molecular micelles,18-21 polysaccharides,22-26 and

* Corresponding author. E-mail: [email protected].

(1) Islam, M. R.; Mahdi, J. G.; Bowen, I. D. Drug Saf. 1997, 17, 149–165.(2) Agranat, I.; Caner, H.; Caldwell, A. Nat. Rev. Drug Discovery 2002, 1, 753–

768.(3) Caner, H.; Groner, E.; Levy, L.; Agranat, I. Drug Discovery Today 2004, 9,

105–110.(4) FDA’s Policy Statement for the Development of New Stereoisomeric Drugs;

FED REGISTER, 1992. http://www.fda.gov/cder/guidance/stereo.htm.(5) Agrawal, Y. K.; Bhatt, H. G.; Raval, H. G.; Oza, P. M.; Gogoi, P. J. Mini-

Rev. Med. Chem. 2007, 7, 451–460.(6) Kano, K. J. Phys. Org. Chem. 1997, 10, 286–291.(7) Fanali, S. J. Chromatogr. A 2000, 875, 89–122.(8) Fillet, M.; Hubert, P.; Crommen, J. J. Chromatogr. A 2000, 875, 123–134.(9) Chankvetadze, B. Chem. Soc. Rev. 2004, 33, 337–347.

(10) Dodziuk, H.; Kozminski, W.; Ejchart, A. Chirality 2004, 16, 90–105.(11) Juvancz, Z.; Kendrovics, R. B.; Ivanyi, R.; Szente, L. Electrophoresis 2008,

29, 1701–1712.(12) Fakayode, S. O.; Brady, P. N.; Pollard, D. A.; Mohammed, A. K.; Warner,

I. M. Anal. Bioanal. Chem. 2009, 394, 1645–1653.(13) Got, P. A.; Scherrmann, J. M. Pharm. Res. 1997, 14, 1516–1523.(14) Hofstetter, H.; Cary, J. R.; Eleniste, P. P.; Hertweck, J. K.; Lindstrom, H. J.;

Ranieri, D. I.; Smith, G. B.; Undesser, L. P.; Zeleke, J. M.; Zeleke, T. K.;Hofstetter, O. Chirality 2005, 17, S9–S18.

(15) Hofstetter, H.; Hofstetter, O. TrAC, Trends Anal. Chem. 2005, 24, 869–879.

(16) Ward, T. J.; Farris, A. B. J. Chromatogr. A 2001, 906, 73–89.(17) Ilisz, I.; Berkecz, R.; Peter, A. J. Sep. Sci. 2006, 29, 1305–1321.(18) Shamsi, S. A.; Palmer, C. P.; Warner, I. M. Anal. Chem. 2001, 73, 140A–

149A.(19) Fakayode, S. O.; Williams, A. A.; Busch, M. A.; Busch, K. W.; Warner, I. M.

J. Fluoresc. 2006, 16, 659–670.(20) Williams, A. A.; Fakayode, S. O.; Alpturk, O.; Jones, C. M.; Lowry, M.;

Strongin, R. M.; Warner, I. M. J. Fluoresc. 2008, 18, 285–296.(21) Willlams, A. A.; Fakayode, S. O.; Lowry, M.; Warner, I. M. Chirality 2009,

21, 305–315.(22) Okamoto, Y.; Yashima, E. Angew. Chem., Int. Ed. 1998, 37, 1020–1043.(23) Tachibana, K.; Ohnishi, A. J. Chromatogr. A 2001, 906, 127–154.(24) Yashima, E. J. Chromatogr. A 2001, 906, 105–125.(25) Stringham, R. W. In Advances in Chromatography; CRC PRESS-TAYLOR

& FRANCIS GROUP: Boca Raton, 2006; Vol. 44, pp 257-290.(26) Ali, I.; Aboul-Enein, H. Y. Curr. Pharm. Anal. 2007, 3, 71–82.

Anal. Chem. 2010, 82, 5028–5037

10.1021/ac9027774 2010 American Chemical Society5028 Analytical Chemistry, Vol. 82, No. 12, June 15, 2010Published on Web 05/19/2010

crown ethers.27,28 However, their use is often hindered byproblems such as low solubility of either the analyte or selectorin the solvent of choice, difficult multistep syntheses, and instabil-ity at high temperatures. Complex solvent mixtures may some-times be required to dissolve the analyte and the chiral selectorwhen they are not both soluble in the same solvent. Solubility isan important factor in the formation of diastereomeric complexes.In previous work using a phenylalanine-based molecular micelleas the chiral selector, we observed that the addition of methanolto aqueous solutions of analyte and selector improved thediscrimination toward nonpolar analytes.21 Chiral ionic liquidsshow discrimination through the solvating power of ionicliquids.29-32 Therefore, the multiple solvation interactions providedby ionic liquids (ILs) may minimize solubility problems.33 Forexample, ILs dissolve a variety of polar and nonpolar analytes,while chiral ionic liquids (CILs) may at the same time induce therequired chiral selectivity.31 Moreover, unlike most other chiralselectors, CILs do not require inclusion of the analyte into theselector’s cavity to provide discrimination, thus adding to theiruniversality. In addition, ILs may be relatively simple to prepare.Many one-step syntheses of ILs, including CILs, from com-mercially available reagents have been reported. They are alsoknown to be stable at high temperatures as evidenced throughtheir use in gas chromatography.34,35 Therefore, the manybeneficial properties of CILs may minimize some of the problemsencountered with other chiral selectors.

Chiral ionic liquids belong to a class of compounds generallycalled ionic liquids or room temperature ionic liquids. Thepotential of ILs in analytical chemistry,36-38 spectroscopy,31

extractions and separations,39-42 synthesis or catalysis,43 andbeyond44 has been widely reviewed. Chiral ionic liquids are ILsin which either the cation, or the anion, or both may be chiral.One such example of a series of CILs possessing either a chiralcation, chiral anion, or both was recently reported.45 The cationswere derivatives of an imidazolium group, while the anions werebased on a borate ion with spiral structure and chiral substituents.The growing interest in CILs is demonstrated by the increasing

number of new classes of CILs, including the amino acid-basedCILs used in this work, reported in the literature. Synthesis andapplications of various CILs have been thoroughly described inseveral recent reviews.46-50

Chiral ionic liquids have recently gained popularity in partbased on their many potential applications including their usesas chiral stationary phases in gas chromatography34,35 and asadditives for enantiomeric separations in micellar electrokineticelectrochromatography51 or high performance liquid chromatog-raphy.35 The development of ILs based on amino acids that arechiral, biocompatible, and biodegradable, and also display reducedtoxicity, has also increased the popularity of CILs. Ohno and co-workers reported amino acid-based ILs composed of imidazoliumcations and amino acid anions and speculated as to their potentialutility.52 Subsequently, several other reports of amino acid-basedILs have appeared. Amino acid-based ILs were recently re-viewed,53,54 and it was predicted that this area of research is poisedfor rapid development and expansion.54

Despite numerous reports on the synthesis of CILs, a surpris-ingly limited number of studies using spectrophotometric mea-surements to investigate their enantiomeric recognition propertieshave been reported.31 One example uses near-infrared absorptionspectroscopy to determine the enantiomeric compositions ofpharmaceutical products using a novel CIL as both the solventand selector.29 Taking advantage of high sensitivity fluorescencedetection, the same group extended the technique to the enan-tiomeric determination of intrinsically fluorescent chiral analytesincluding propranolol, naproxen, and warfarin.30 In related work,we recently investigated the enantiomeric recognition of aminoacid-based CILs toward intrinsically fluorescent analytes.32 Yet,there is a continued need for a simple, universal technique tomeasure enantiomeric compositions at low concentrations. Highsensitivity fluorescence detection provides advantages; however,considering many analytes of interest display minimal intrinsicfluorescence emission, it would be advantageous to include aspart of the IL a fluorescent reporter.

A great amount of work reporting fluorescence in ILs hasappeared in the literature, including an investigation of dynamicStokes shift and excitation wavelength dependent fluorescenceof dipolar molecules in ILs.55,56 Additionally, a series of highlyphotoluminescent benzobis(imidazolium) salts were recentlydescribed by Bielawski and co-workers, and an application of thenew class of versatile fluorescent materials was demonstrated byconjugating a maleimide-functionalized derivative to bovine serumalbumin.57,58 It has also been reported that many ILs themselves,

(27) Kuhn, R. Electrophoresis 1999, 20, 2605–2613.(28) Hyun, M. H. Bull. Korean Chem. Soc. 2005, 26, 1153–1163.(29) Tran, C. D.; Oliveira, D.; Yu, S. F. Anal. Chem. 2006, 78, 1349–1356.(30) Tran, C. D.; Oliveira, D. Anal. Biochem. 2006, 356, 51–58.(31) Tran, C. D. Anal. Lett. 2007, 40, 2447–2464.(32) Bwambok, D. K.; Marwani, H. M.; Fernand, V. E.; Fakayode, S. O.; Lowry,

M.; Negulescu, I.; Strongin, R. M.; Warner, I. M. Chirality 2008, 20, 151–158.

(33) Anderson, J. L.; Ding, J.; Welton, T.; Armstrong, D. W. J. Am. Chem. Soc.2002, 124, 14247–14254.

(34) Ding, J.; Welton, T.; Armstrong, D. W. Anal. Chem. 2004, 76, 6819–6822.(35) Yuan, L. M.; Han, Y.; Zhou, Y.; Meng, X.; Li, Z. Y.; Zi, M.; Chang, Y. X.

Anal. Lett. 2006, 39, 1439–1449.(36) Baker, G. A.; Baker, S. N.; Pandey, S.; Bright, F. V. Analyst 2005, 130,

800–808.(37) Anderson, J. L.; Armstrong, D. W.; Wei, G. T. Anal. Chem. 2006, 78, 2892–

2902.(38) Soukup-Hein, R. J.; Warnke, M. M.; Armstrong, D. W. Ann. Rev. Anal. Chem.

2009, 2, 145–168.(39) Han, X.; Armstrong, D. W. Acc. Chem. Res. 2007, 40, 1079–1086.(40) Stalcup, A. M.; Cabovska, B. J. Liq. Chromatogr. Relat. Technol. 2004, 27,

1443–1459.(41) Shamsi, S. A.; Danielson, N. D. J. Sep. Sci. 2007, 30, 1729–1750.(42) Polyakova, Y.; Koo, Y. M.; Row, K. H. Rev. Anal. Chem. 2007, 26, 77–98.(43) Welton, T. Chem. Rev. 1999, 99, 2071–2083.(44) Smiglak, M.; Metlen, A.; Rogers, R. D. Acc. Chem. Res. 2007, 40, 1182–

1192.(45) Yu, S. F.; Lindeman, S.; Tran, C. D. J. Org. Chem. 2008, 73, 2576–2591.

(46) Bica, K.; Gaertner, P. Eur. J. Org. Chem. 2008, 3235–3250.(47) Patil, M. L.; Sasai, H. Chem. Rec. 2008, 8, 98–108.(48) Winkel, A.; Reddy, P. V. G.; Wilhelm, R. Synthesis 2008, 999–1016.(49) Ding, J.; Armstrong, D. W. Chirality 2005, 17, 281–292.(50) Baudequin, C.; Bregeon, D.; Levillain, J.; Guillen, F.; Plaquevent, J. C.;

Gaumont, A. C. Tetrahedron: Asymmetry 2005, 16, 3921–3945.(51) Rizvi, S. A. A.; Shamsi, S. A. Anal. Chem. 2006, 78, 7061–7069.(52) Fukumoto, K.; Yoshizawa, M.; Ohno, H. J. Am. Chem. Soc. 2005, 127,

2398–2399.(53) Ohno, H.; Fukumoto, K. Acc. Chem. Res. 2007, 40, 1122–1129.(54) Chen, X. W.; Li, X. H.; Hu, A. X.; Wang, F. R. Tetrahedron: Asymmetry 2008,

19, 1–14.(55) Mandal, P. K.; Sarkar, M.; Samanta, A. J. Phys. Chem. A 2004, 108, 9048–

9053.(56) Samanta, A. J. Phys. Chem. B 2006, 110, 13704–13716.(57) Boydston, A. J.; Pecinovsky, C. S.; Chao, S. T.; Bielawski, C. W. J. Am.

Chem. Soc. 2007, 129, 14550–14551.

5029Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

or possibly their impurities, display fluorescence when excited at280-430 nm. However, the origin of this fluorescence has notyet been explained.37 Universal fluorescence detection in themeasurement of enantiomeric compositions using a CIL as thesolvent, selector, and reporter would require a fluorescent CIL(i.e., a FCIL). Such a FCIL would make high sensitivity detectionpossible for nonfluorescent analytes. The intrinsic fluorescenceof aromatic amino acids provides this possibility.

Herein, we report the synthesis and characterization of anamino acid-based fluorescent chiral ionic liquid, L-PheC2NTf2, andinvestigate its enantiomeric recognition capabilities. The uniquespectral behavior of this FCIL makes it an ideal system toexploit the power of multidimensional fluorescence measure-ments. To the best of our knowledge, this is the first report ofa FCIL used as solvent and chiral selector, fluorescent reporterfor enantiomeric recognition of fluorescent, as well as nonfluo-rescent analytes with a wide range of structures.

EXPERIMENTAL SECTIONChemicals and Materials. L-Phenylalanine ethyl ester hy-

drochloride [L-PheC2Cl] (99%), bis(trifluoromethane) sulfonim-ide lithium salt (LiNTf2) (99.5%), 2-methoxy-2-(trifluoromethyl)phenylacetic acid (Mosher’s acid) (99%), methanol (ACScertified), Ludox AS-40 colloidal silica (40 wt % suspension inwater), tetramethyl silane (TMS), deuterated solvents (CD2Cl2and d6-DMSO), pure enantiomeric forms of fluorescent analytes(S-2,2,2-trifluoroanthrylethanol (S-TFAE) (98%), (R-TFAE) (98%)),and nonfluorescent analytes (D-serine (99%), L-serine (99%),D-glucose (99%), L-glucose (99%) D-mannose (99%), and L-mannose(99%)) were purchased from Sigma Aldrich Chemicals(Milwaukee, WI). All chemicals were used as received withoutfurther purification. Ultrapure water (18.2 MΩ · cm) was usedfor all purposes and obtained from an Elga model PURELABUltra water filtration system.

Solution Preparation for Fluorescence-Based Enantio-meric Recognition Screening. Fluorescent (TFAE) and non-fluorescent (serine, glucose, and mannose) analyte stock solutionswere prepared from ethanolic and aqueous stock solutions,respectively. Initially, 25 mL of 1 mM fluorescent R- and S-TFAEwas prepared in ethanol and diluted to 0.1 mM. Final concentra-tions were adjusted to ensure equal concentration based upon thefluorescence signal of these solutions acquired upon excitationat the absorbance maximum. These solutions were used forsample preparation as described below. The concentration ofnonfluorescent analyte stock solutions could not be equalized byfluorescence. In addition, the solutions displayed low absorbance.Therefore, a large volume (500 mL) of 1 mM aqueous solutionwas prepared to reduce potential errors and ensure equalconcentration and used in sample preparation. Required volumesof these analyte stock solutions were added to empty vials toprepare either 10 or 100 µM analyte solutions dissolved in FCIL.The aqueous or ethanolic phase was evaporated with a gentlestream of nitrogen followed by addition of a known volume ofFCIL (typically 0.5 mL as determined by mass based upon ameasured density of 1.49 g/mL) to the vials containing known

masses of analytes. Any errors in the final sample concentrationwere largely the result of errors in the final volume of FCIL.Samples were prepared in triplicate, and the precision of thesample preparation was on the order of 1% relative standarddeviation based upon the masses of added FCIL. These solutionswere subjected to ultrasonication for 20 min at room temperatureand stored overnight in the dark to ensure complete dissolutionand equilibration before fluorescence measurements.

Solution Preparation for Fluorescence-Based Determi-nation of Enantiomeric Composition. A series of mannose/FCIL solutions of varying known enantiomeric composition wereprepared by mixing known volumes (as determined by mass) ofthe pure enantiomeric mannose in FCIL solutions prepared above.The solutions were sonicated for 20 min and then allowed toequilibrate for 1 h before fluorescence measurements.

General Instrumental Methods. A Jasco-710 spectropola-rimeter was used to obtain the Circular Dichroism (CD) spectraof room temperature methanolic solutions (1 mM L-PheC2NTf2)in 1 cm quartz cells. The thermal decomposition temperature(Tdec) was determined with a thermal analysis instrument, 2950TGA HR V6.1A (module TGA 1000 °C), at a heating rate undernitrogen of 5 °C min-1 from 25 to 300 °C. The NMR spectrawere recorded on a Bruker-250 MHz instrument with tetram-ethyl silane (TMS) as an internal standard. For 1H and 13CNMR, d6-DMSO was used, whereas CD2Cl2 with 30% d6-DMSOwas used for 19F NMR measurements. Electrospray ionization(ESI) mass spectrometry (MS) analysis was performed on anAgilent 6210 electrospray time-of-flight mass spectrometer. Thesample was dissolved in methanol before analysis.

Absorption measurements were acquired at room temperatureusing a Shimadzu, model UV-3101PC, UV-Vis-NIR spectropho-tometer with samples either in 4 mm square quartz cells orbetween two quartz plates in the case of the neat L-PheC2NTf2

film. Steady-state and frequency-domain lifetime fluorescencemeasurements were recorded at room temperature on a SpexFluorolog-3 spectrofluorimeter (model FL3-22TAU3; HORIBAJobin Yvon, Edison, NJ). Excitation was provided by a 450 Wxenon arc lamp. Dual monochromators were used to select theexcitation and emission wavelengths with both bandpasses setto 3 nm. The integration time was set to 0.1 s per point, and950 V was applied to a Hamamatsu R928 PMT. All fluorescencespectra were ratio corrected through division by the signal froma reference detector which monitored the lamp output. Excita-tion-Emission Matrix spectra (EEMs) of neat FCIL werecollected in 4 mm square quartz cells with excitations from250 to 400 nm and emission from 250 to 500 nm using 2.5 nmand 5 nm step sizes for emission and excitation, respectively,while EEMs for enantiomeric recognition studies were col-lected in 3 mm square quartz cells over the same region. Dataprocessing and analysis was performed in Microsoft Excel andMATLAB (The MathWorks, Inc., Natick, MA). Multivariateregression analysis (PLS1) was also performed using Unscram-bler (CAMO, Inc., Corvallis, OR, version 9.8) chemometricsoftware.

The fluorescence quantum yield was measured in a 1 cm path-length quartz cell with right-angle collection geometry followingthe comparative method reported by Williams et al.59 A quantumyield of 0.27 for anthracene in ethanol was used as the reference.

(58) Boydston, A. J.; Vu, P. D.; Dykhno, O. L.; Chang, V.; Wyatt, A. R.; Stockett,A. S.; Ritschdbrff, E. T.; Shear, J. B.; Bielawski, C. W. J. Am. Chem. Soc.2008, 130, 3143–3156.

5030 Analytical Chemistry, Vol. 82, No. 12, June 15, 2010

Fluorescence decay times were measured using a variablefrequency phase-modulation technique. Full-sized 1 cm quartz cellswere used with 330 nm excitation. The emission was collectedthrough a 370 nm long pass filter. Thirty-one logarithmicallyspaced frequencies were collected over a frequency range of10-136.9 MHz using five averages and a 10 s integration time ateach frequency. Frequency-domain measurements were collectedversus Ludox, a scatter reference solution, which showed a lifetimeof 0 ns. Phase and modulation decay profiles were analyzed usingthe Globals software package developed at the Laboratory forFluorescence Dynamics (University of Illinois at Urbana-Cham-paign). Constant phase and modulation errors of 0.5° and 0.005were used in analyses for consistency and ease of day-to-day datainterpretation.

Synthesis of L-Phenylalanine Ethyl Ester Bis(trifluo-romethane) Sulfonimide (L-PheC2NTf2). The procedure wassimilar to that used in the preparation of alanine-based CILsreported previously.32 Aqueous solutions of 5 g (21.7 mmol) ofL-phenylalanine ethyl ester chloride (L-PheC2Cl) and 6.25 g (21.7mmol) of bis(trifluoromethane) sulfonimide lithium salt(LiNTf2) were mixed together to a combined volume of 800mL. The mixture was stirred for 16 h at room temperature.The ionic liquid was obtained after removal of water using arotary evaporator in vacuo at 35 °C and purified by washingwith a 10 mL volume of water 5 times. Subsequently, the ionicliquid was freeze-dried for 12 h using a lyophilizer to removeany trace amounts of water remaining after rotary evaporationaffording the ionic liquid product in good yield (83% yield). Theinitial batch was used for characterization and initial studies.Multiple batches were then synthesized with minimal batch-to-batch variation as determined by their spectral properties.All subsequent batches were pooled for use in enantiomericrecognition studies. The pooled FCIL was sealed tightly andstored in the dark when not in use.

The decomposition temperature (Tdec) was found to be 270°C using thermal gravimetric analysis (see Supporting Informa-tion, Figure S-1). Characterization by 1H NMR and 13C NMRproduced the expected peaks, consistent with the chemicalstructure of the ionic liquid. 1H NMR (250 MHz, d6-DMSO) δ(ppm) 8.3 (s, 3H), 7.34-7.21 (m, 5H), 4.28 (t, J ) 7.5, 1H),4.12 (q, 2H), 3.02-3.16 (ddd, J ) 6.4, 7.5, 13.2 Hz, 2H), 1.10 (t,J ) 7.2 Hz, 3H). 13C NMR δ (ppm) 169.91, 135.35, 130.28,129.44, 128.16, 62.57, 59.50, 54.02, 14.53. These results wereconfirmed using elemental analysis and mass spectrometry.Calculated for C13H16N2O6S2F6: C, 32.91; H, 3.40; N, 5.90. Found:C, 32.00; H, 3.79; N, 5.31. MS (ESI+) m/z calculated for[C11H16NO2]+: 194.2100, found 194.1271. MS (ESI-) m/zcalculated for [N(SO2CF3)2]-: 280.1390, found 279.9343.

RESULTS AND DISCUSSION

Confirmation of Chiral Character and Stability uponExtended Heating. One goal of the current study was todemonstrate chiral discrimination by use of L-PheC2NTf2 whenused simultaneously as the solvent and selector. Therefore, thechiral integrity of the FCIL product was important as some CILshave previously been reported to undergo racemization during

synthesis, although its precursor was chiral.60 Circular dichro-ism measurements of L-PheC2NTf2 product ionic liquid (1 mMin methanol) and its chiral starting material, L-PheC2Cl (1 mMin methanol), were carried out to confirm the retention of chiralconfiguration following the ion exchange. The CD spectral bandof the L-PheC2Cl precursor is positive (data not shown) andthe same shape and direction as that of the L-PheC2NTf2

product, thus confirming the retention of chiral configuration.It is also important to confirm the retention of chiral configu-ration upon heating the FCIL for an extended period of time.This was investigated through a comparison of CD spectra of1 mM methanolic solutions prepared from neat L-PheC2NTf2

at room temperature to that of a corresponding FCIL solutionin which neat L-PheC2NTf2 was heated in the oven for morethan 18 h at 115 °C. As shown in Figure 1, this FCIL does notracemize after prolonged heating.

Thermal analysis by TGA revealed that L-PheC2NTf2 is stableto thermal decomposition up to 270 °C (Supporting Information,Figure S-1A). It has been observed that upon prolonged heatingsome ILs decompose at temperatures lower than their measureddecomposition temperatures.61,62 This observation prompted usto heat L-PheC2NTf2 for 30 min at a constant temperature of 225°C. A minimal loss of mass was observed (Supporting Information,Figure S-1B). Results obtained from these studies suggest thatthis FCIL is relatively stable when heated for a long period oftime and may be potentially suitable as a solvent medium for high-temperature reactions and as a chiral stationary phase in gaschromatography.

Enantiomeric Recognition Screening Using 19F NMR. Theenantiomeric recognition ability of L-PheC2NTf2 was investigatedusing the diastereomeric interaction toward classical racemicMosher’s sodium salt. A binary solvent system consisting ofCD2Cl2 with 30% d6-DMSO was used to dissolve the mixture of

(59) Williams, A. T. R.; Winfield, S. A.; Miller, J. N. Analyst 1983, 108, 1067–1071.

(60) Jodry, J. J.; Mikami, K. Tetrahedron Lett. 2004, 45, 4429–4431.(61) Baranyai, K. J.; Deacon, G. B.; MacFarlane, D. R.; Pringle, J. M.; Scott,

J. L. Aust. J. Chem. 2004, 57, 145–147.(62) Kosmulski, M.; Gustafsson, J.; Rosenholm, J. B. Thermochim. Acta 2004,

412, 47–53.

Figure 1. Circular dichroism spectra of 10-3 M L-PheC2NTf2 inmethanol at room temperature (broken line) and corresponding 10-3

M L-PheC2NTf2 in methanol prepared from neat L-PheC2NTf2 previ-ously heated at 115 °C for 18 h (continuous line). The structure ofL-phenylalanine ethyl ester bis(trifluoromethane) sulfonimide (L-PheC2-NTf2) is included as an inset.


racemic Mosher’s sodium salt and L-PheC2NTf2. The 19F NMRspectrum of Mosher’s salt in the absence of L-PheC2NTf2

displayed only a single band at 69.803 ppm (SupportingInformation, Figure S-2A). Addition of L-PheC2NTf2, 142.30 mg(0.3 mmol), to the racemic Mosher’s sodium salt, 8.38 mg (0.03mmol), resulted in a splitting as well as a shift to 70.832 and70.876 ppm (Supporting Information, Figure S-2B). The differ-ence in chemical shifts (∆δ) was calculated as 11 Hz. This is anindication of the strong diastereomeric interaction between theracemic Mosher’s sodium salt and the FCIL, suggesting thatL-PheC2NTf2 has good chiral discrimination ability leading usto further investigate its enantiomeric recognition propertiesusing steady-state fluorescence.

Spectroscopic Properties of L-PheC2NTf2. The absorptionand fluorescence properties of L-PheC2NTf2 were studied as neatFCIL and using ethanol as a solvent as needed. These dataare presented in Table 1. Neat L-PheC2NTf2 was observed tobe colorless. A thin film of neat L-PheC2NTf2 displayed amaximum absorbance at 257 nm and a spectral shape similarto the characteristic phenylalanine amino acid absorptionspectrum in water. However, unlike phenylalanine and L-phenylalanine ethyl ester hydrochloride (a precursor),L-PheC2NTf2 displayed a weakly absorbing tail in its absorptionspectrum extending to wavelengths beyond 400 nm. Theabsorption spectra of neat L-PheC2NTf2 in a 4 mm path-lengthcell (Figure 2) displayed an absorbance of ∼0.1 near 350 nmdecreasing to ∼0.05 at 400 nm. Absorbance was greater than 3 atwavelengths below ∼270 nm. It is also interesting to note thatsubtle differences appeared in the spectral region between 280and 320 nm as L-PheC2NTf2 was diluted in ethanol (Figure 2).

The high absorbance of the neat L-PheC2NTf2 necessitateduse of dilute ethanolic solutions in determination of its quantumyield excited at its maximum absorbance. A quantum yield of0.06 was calculated for L-PheC2NTf2 dissolved in ethanol uponexcitation at 257 nm. For comparison, this value is slightlygreater than the value of 0.022 reported for phenylalanineamino acid in water.63 It should be noted that a low quantumyield does not hinder the use of this FCIL for enantiomericrecognition studies. Its use as solvent (without dilution)resulted in sufficiently high concentration (molarity ) 3.14 mol

L-1, calculated based on its measured density) to obtain morethan adequate absorbance and fluorescence signal.

The absorption spectrum of a thin film (∼29 µm) of L-Phe-C2NTf2 between two quartz plates was qualitatively similar tothat of a 1 mM solution of L-PheC2Cl (a precursor) in ethanol(Figure 2). The molar absorptivity of the L-PheC2NTf2 thin filmwas estimated to be 95 L mol-1 cm-1 at 257 nm, which is lowerthan that reported for phenylalanine amino acid dissolved inH2O (195 L mol-1 cm-1 at 257.5 nm).64 Decreased absorbancewas also observed for 0.01 M L-PheC2Cl (a precursor) in H2Oas compared to 0.01 M phenylalanine in H2O (factor of 1.77 atthe maximum absorbance) which accounts for the majority ofthe observed decrease in absorbance for the phenylalanine-based FCIL as compared to phenylalanine. The seemingly lowmolar absorptivity of L-PheC2NTf2 does not pose a problem forour applications. In fact, high absorbance values and theresulting inner filter effects presented greater difficulties.Therefore, front face collection geometry was used for fluo-rescence measurements in an effort to minimize inner filtereffects. It was not generally practical to use excitations below∼280 nm where absorption of the incident excitation light wasobserved to reduce the observed emission intensity.

(63) Chen, R. F. J. Res. Natl. Bur. Stand. Sect. C-Eng. Instrum. 1972, A 76,593–606.

(64) Fasman, G. D. Handbook of Biochemistry and Molecular Biology, Proteins,I, 3rd ed.; CRC Press: Boca Raton, Fl, 1976.

Table 1. Photophysical Characteristics of L-PheC2NTf2

fluorescence properties

absorption propertiesa emission propertiesb lifetime (ns)c

FCIL λmax (nm) ε (L mol-1 cm-1) λexc (nm) peak(s): λem (nm) τ1 τ2 τ3

L-PheC2NTf2d 257 ∼95

260 ∼280, 320280 320, ∼400 0.68 3.57 12.72300 ∼340, ∼400 (0.25) (0.61) (0.14)320 ∼450

a Properties of neat ionic liquid. A thin film (∼29 µm) of neat L-PheC2NTf2 between two quartz plates displayed a spectral shape similar tophenylalanine. Neat ionic liquid in a 4 mm path-length cell displayed an absorbance greater than 3 at wavelengths below ∼270 nm, ∼0.1 near 350nm, and ∼0.05 at 400 nm. b Properties of neat ionic liquid. Front face collection geometry was used in an effort to minimize inner filter effects.c Properties of neat ionic liquid. Frequency-domain measurements were collected versus Ludox, a scatter reference solution, which showed alifetime of 0 ns. Full-sized 1 cm quartz cells were used with 330 nm excitation. The emission was collected through a 370 nm long pass filter.Fractional contributions of each lifetime are shown in parentheses. d The high absorbance of the neat L-PheC2NTf2 necessitated using dilute ethanolicsolutions in the determination of its quantum yield. A quantum yield of 0.06 was calculated upon excitation at 257 nm in a 1 cm path-length quartzcell with right-angle collection geometry.

Figure 2. Absorption spectra of neat L-PheC2NTf2, various dilutionsin ethanol (EtOH), and 1 mM L-PheC2Cl ionic liquid precursor in EtOHmeasured in a 4 mm thick cell. The figure also includes a 29 µMthick layer of neat L-PheC2NTf2.


The fluorescence Excitation-Emission Matrix (EEM) ofL-PheC2NTf2 revealed complex fluorescence behavior withmultiple excitation and emission bands for neat FCIL. The fullEEM contour plot can be found in Figure 3A with a subset ofemission spectra collected with various excitation wavelengthsshown in Figure 3B. Although the emission intensity was impactedby absorption of the incident light, excitation at 260 nm (near the257 nm maximum absorbance of L-PheC2NTf2) revealed broadoverlapping emissions with indications of peaks near thecharacteristic 280 nm emission of phenylalanine, as well as near320 nm. The latter peak was previously identified as aphenylalanine excimer by Leroy et al.65 Longworth had alsoobserved similar excimer emission for L-polyphenylalanine inTHF.66 Increasing the excitation wavelength to 280 nm reducedthe impact of inner filter effects as the absorbance decreased.A prominent excimer peak at 320 nm resulting from 280 nmexcitation displayed the greatest intensity of all emissionsresulting from any excitation. A lower intensity emission peaknear 400 nm was also apparent upon 280 nm excitation andbecame more readily observable as the excitation wavelengthincreased to 300 nm. At excitation of 320 nm, the peak near400 nm appeared as the only emission peak. Excitations greaterthan 320 nm resulted in a red shift of this long-wavelengthemission. Concentration-based studies conducted by Leroy etal.65 clearly demonstrated excimer emission at 320 nm for aphenylalanine concentration of 0.1 M in water. However, nolonger wavelength emitting species were reported at thisconcentration. It is not surprising that the extremely highphenylalanine concentration (3.14 mol L-1) occurring in neatL-PheC2NTf2 results in much more complex spectral behavior

than low concentration solutions. Explaining these complexitiesis beyond the scope of this present work; however, it shouldbe noted that imidazolium ILs have also been reported todisplay unconventional fluorescence behavior. Fluorescenceemission shifted toward the red with increasing excitationwavelength. This observation was attributed to the presenceof energetically different associated forms of the constituentions and a slow rate of excited state relaxation processes inthe viscous IL.55 Interestingly, the long-wavelength emissionwas found to increase (with a corresponding decrease in shortwavelength phenylalanine and excimer emissions) as this FCILaged (Supporting Information, Figure S-3). This was not theresult of degradation. Other properties of L-PheC2NTf2 were notobserved to change with time. This phenomenon is of interestand is currently under investigation. As will be discussed below,it is this spectral region that displays significant enantiomericrecognition ability toward some of the nonfluorescent analytes.One possible explanation for this observation is that there is agradual reorganization of the constituent ions over time.

The complexity observed in the steady-state EEMs was alsodisplayed in time-resolved measurements. Instrumental limitationsand the high absorbance of neat L-PheC2NTf2 complicated themeasurements and required excitation at a longer wavelengththan the 257 nm maximum absorbance of L-PheC2NTf2. In aneffort to minimize inner filter effects, measurements wereperformed upon 330 nm excitation with emission collectedthrough a 370 nm long pass filter. Leroy et al. reported thelifetime of phenylalanine amino acid (8 × 10-4 M in H2O at 20°C) to be 6.8 ± 0.2 ns and observed a slight increase in decaytime at increased concentrations.65 In this work, three lifetimecomponents including both significantly longer (∼12 ns) andshorter (<1 ns) lifetimes were required to adequately modelthe decay of L-PheC2NTf2. Fitting parameters are reported inTable 1 with data, residuals, and 2 surfaces presented in theSupporting Information (Figures S-4 and S-5). The more complexdecay observed for the phenylalanine-based FCIL is expectedconsidering the complexity of the spectral behavior describedabove. It is interesting to note that complex spectral behavior wasalso observed in a phenylalanine-based fluorescent chiral molec-ular micelle, previously reported by our group.20,21 Long-wavelength emission and a triple exponential decay similar to thatobtained for L-PheC2NTf2 were observed.

Enantiomeric Recognition Using Steady-State Fluores-cence Spectroscopy. In these studies, L-PheC2NTf2 servedsimultaneously as a solvent and chiral selector, as well as afluorescent reporter in the case of nonfluorescent chiralanalytes. A variety of fluorescent and nonfluorescent analyteswith a wide range of chemical structures were investigated. Itis interesting to note that L-PheC2NTf2 could dissolve structur-ally diverse analytes (including TFAE, serine, glucose, andmannose) and induce spectral differences between enantio-merically pure forms of each pair of analytes. The ability ofILs to solvate a wide range of both relatively polar and nonpolaranalytes may eliminate the need for mixed solvent systems,thus simplifying the measurement and increasing the universalnature of the method.

The spectral properties of neat L-PheC2NTf2 were complex,with multiple emissions upon various excitations. Therefore,

(65) Leroy, E.; Lami, H.; Laustria, G. Photochem. Photobiol. 1971, 13, 411–421.(66) Longwort, J. W. Biopolymers 1966, 4, 1131–1148.

Figure 3. Fluorescence of neat L-PheC2NTf2 (A) Excitation-EmissionMatrix (EEM). (B) Subset of emission spectra collected with variousexcitation wavelengths extracted from the EEM at various excitationwavelengths (260-360 nm in 20 nm intervals).


the best single excitation wavelength or emission region for agiven analyte could not be predicted. The choice of the wrongexcitation wavelength could at worst lead to the incorrectconclusion that the system displays no selectivity or, at best,lower the degree of selectivity and increase the errors in theanalysis. The complex emissions made this an ideal system totake advantage of the multidimensional nature of fluorescence.In our previous methods employing fluorescence,19-21,32 we didnot take full advantage of its multidimensional properties. A singleexcitation wavelength was generally chosen near the maximumabsorbance of the fluorescent analyte or fluorescent chiral selectorused to collect an emission spectrum. Then, a single wavelength(or a relatively small spectral window) in the emission wasanalyzed. While a certain degree of added selectivity was providedthrough use of two wavelengths (both excitation and emission),little thought was put into the choice of excitation wavelength.Judicious selection of excitation and/or use of multiple excitationscan potentially improve the results (vide infra).

In addition to the complex emission from the FCIL, it wasessential to measure the EEM over a broad range of excitationsand emissions because many dipolar fluorescent molecules exhibitrelatively strong excitation-wavelength-dependent fluorescencebehavior in ILs.55 The excitation-wavelength-dependent shift ofthe fluorescence maximum has been reported to be between 10and 35 nm.55 The dynamic stokes shift and excitation wavelength-dependent fluorescence of dipolar molecules in room temperatureILs have been discussed at length.56

In the case of chiral analytes possessing intrinsic fluorescence,spectral alterations induced upon the analyte’s emission weregenerally much greater than those induced upon the FCIL’semission by the presence of the analyte. However, the moreinteresting application of the technique resulted from its abilityto allow more sensitive and selective fluorescence detection inthe enantiomeric recognition of nonfluorescent analytes throughtheir effects on the emission of the FCIL. It is this aspect thatwill be emphasized in this paper after first discussing the behaviorof a single fluorescent analyte (TFAE).

Enantiomeric Recognition of a Fluorescent Analyte. Chiraldiscrimination of TFAE was investigated by measuring the EEM

of three independently prepared solutions of 10 µM R- or S-TFAEdissolved in FCIL. The average R- and S-EEMs were calculated.For simplicity, emission spectra of each enantiomeric solutioncollected at representative excitation wavelengths (20 nm intervalsbetween 260 and 360 nm) were extracted and presented in Figure4. The entire EEM of R-TFAE in L-PheC2NTf2 is shown in FigureS-6A (Supporting Information). Comparison of emission spectrafor neat FCIL with spectra of each enantiomer dissolved inL-PheC2NTf2 for a given wavelength is informative. Emissionspectra of neat FCIL and each enantiomer dissolved inL-PheC2NTf2 upon excitations at 280 nm and 360 nm arepresented in the Supporting Information (Figure S-6B). Thesewavelengths were chosen because excitation at 280 nm is nearthe excitation maximum generating maximum phenylalanineexcimer emission from L-PheC2NTf2, and excitation at 360 nmis near the excitation maximum of TFAE in L-PheC2NTf2. Themaximum emissions of TFAE enantiomers (the structuredemission between 375 and 500 nm) were greater than themaximum FCIL emission. R-TFAE emission was greater thanS-TFAE emission. Interestingly, the phenylalanine excimeremission of L-PheC2NTf2 was quenched by the presence of bothenantiomeric forms of TFAE.

Differences between the emissions in the presence of eachenantiomer (R-EEM minus S-EEM) were calculated on a wave-length-by-wavelength basis for the entire EEM (including propa-gated errors). A representative subset of differences resulting froma range of excitation wavelengths is presented as an inset in Figure4. The greatest differences resulted where TFAE emissiondominated. The entire difference EEM is shown in Figure S-7A(Supporting Information). The greatest difference was at emissionof ∼420 nm upon excitation at 365 nm. However, although thedifferences in the FCIL excimer emission were small (spectralregion between 275 and 350 nm upon 260 and 280 nm excitations),they were greater than the error in the measurements.

It is sometimes useful to express the spectral differences aspercentages (%∆) because the largest magnitude differences maynot lead to the greatest percent difference %∆ or lowest relativeerrors in the spectral differences. In this case, the FCIL excimeremission region displayed ∼2-3% difference in the presence of

Figure 4. Emission spectra of 10 µM pure enantiomeric solutions of TFAE in L-PheC2NTf2 presented at various excitation wavelengths (260-360nm in 20 nm intervals). Solid lines/filled symbols represent the S-enantiomer. Broken lines/open symbols represent the R-enantiomer. Inset:difference spectra (average of triplicate measurements; R-TFAE minus S-TFAE) for the same representative excitation wavelengths. Representativeerror bars for emission region 475-500 nm upon 360 nm excitation are included.


the enantiomeric forms. Although the largest difference occurredin the emission of TFAE upon excitation at 365 nm, the greatest%∆ occurred in the same emission region but upon excitation at275 nm. A contour map of the entire surface for %∆ is providedin Figure S-7B (Supporting Information). It shows structure,indicating that the spectral changes induced are more complexthan simple wavelength-independent quenching or enhancement.It should be clear that optimization is not a trivial task and thatcareful thought and experimentation must be invested in choiceof the proper excitation and emission region when a quantitativeanalysis result is desired.

Enantiomeric Recognition of Nonfluorescent Analytes.Many analytes of interest have limited or virtually no intrinsicfluorescence emission. Therefore, it is advantageous to measurealterations in the emission of a fluorescent selector, L-PheC2NTf2,caused by diastereomeric interactions with chiral analytes. Thevarious nonfluorescent chiral analytes used for this studyinclude serine, glucose, and mannose.

Chiral discrimination was investigated by measuring the EEMof three independently prepared solutions of each of the three100 µM D- or L-nonfluorescent analytes dissolved in FCIL. Thepresence of the nonfluorescent analytes induced spectral alter-ations to L-PheC2NTf2. Difference spectra (D-form minus L-form)of representative excitation wavelengths are presented inFigure 5. Clearly, the differences are wavelength dependent anddifferent for each analyte. The smallest differences were betweenD- and L-serine, while the largest differences were between D- andL-mannose. Interestingly, the phenylalanine excimer emissiondisplayed the greatest enantiodiscrimination for serine andglucose, while the long-wavelength FCIL emission displayed thegreatest enantiodiscrimination for mannose. See Figure S-8 (Sup-porting Information) for the EEMs of D-enantiomeric solutions inFCIL and representative emission spectra of neat FCIL and inthe presence of each enantiomer excited at 280 and 360 nm. Theseexcitation wavelengths were chosen to show phenylalanine exci-mer and wavelength-dependent long-wavelength emission. Thepresence of serine quenched both the eximer and long-wavelengthemissions of FCIL emission with small differences in the magni-tude of quenching resulting from the two enantiomeric forms ofserine. In contrast, the presence of aldohexose stereoisomersglucose and mannose resulted in completely different behavior,leading to enantioselective enhancement of FCIL emissions.

Although the spectral changes induced upon L-PheC2NTf2 bydifferent enantiomeric forms of nonfluorescent analytes weresmaller than those observed for different enantiomers of thefluorescent analyte, TFAE, the presence of 100 µM nonfluo-rescent analytes induced more than adequate alterations to theemission of the FCIL solvent to differentiate enantiomeric formsof these analytes. Maximum %∆’s for the nonfluorescentanalytes were all greater than 5% and as high as ∼15% in thepresence of mannose. The entire contour plots of both differ-ence spectra and %∆ are provided for reference in FiguresS-9-S-11 (Supporting Information).

Potential for Chemoselective and Enantioselective Fluo-rescent Recognition of Monosaccharides. Interestingly, thisFCIL displayed enantioselectivity toward both glucose and man-nose, and this selectivity occurred in different spectral regions,thus providing for the possibility of chemo- and enantioselectivity.

Chiral discrimination of monosaccharides has been challengingwith few examples reported in the literature. James et al. havedescribed the chiral discrimination of D- and L-monosaccharidesusing a chiral diboronic acid which incorporated a fluorescentnaphthyl moiety.67 Binding of each enantiomer of various monosac-charides altered the fluorescence intensity of the sensor todiffering degrees, thus allowing them to be distinguished. Otherwork in this area includes a recent example of related boronicacid-based enantio- and chemoselective fluorescent chemosensors

Figure 5. Difference spectra (average of triplicate measurements;D-form minus L-form) for 100 µM pure enantiomeric solutions ofnonfluorescent analytes, (A) serine, (B) glucose, and (C) mannosein L-PheC2NTf2, presented at various excitation wavelengths (260-360nm in 20 nm intervals).


for sugar alcohols.68 In recent years, this area of research hasfocused largely on boronic acid-based molecular receptors sincetheir “rediscovery” in the mid 1990s.69 Boronic acid-based sensorshave been reviewed recently.70 While there has been significantprogress using these boronic acid-based sensors, the developmentof boronic acid-based molecular receptors has been hampered bysynthetic difficulties. Another potential difficulty is the cleavageof boronic acids in relatively mild conditions. The FCIL describedherein is not hampered by these difficulties.

The FCIL, L-PheC2NTf2, described herein displays the great-est selectivity for glucose in the short-wavelength excitationof the phenylalanine excimer emission region (Figure 5B).However, it shows a great enantioselectivity toward mannose inthe long-wavelength emission region (Figure 5C). Differencesbetween D- and L-mannose were approximately 4 times greaterthan difference between D- and L-glucose in the long-wavelengthregion, while the short-wavelength excitation of the phenylalanineexcimer emission did not show enantioselectivity toward mannose(Figure 5C). Increased sensitivity for mannose is a benefit as mostbiological samples contain excess mannose as compared toglucose.

Enantiomeric Composition Analysis of Mannose. As a firststep toward the multicomponent chemo- and enantioselectivedetermination of glucose and mannose, we herein report the useof the technique to measure the enantiomeric composition of amannose solution. First, the EEM of this FCIL in the presence ofvarying enantiomeric compositions of 100 µM mannose wasmeasured over the entire range from 100% L-mannose to 100%D-mannose. The intensity as a function of D-mannose mole fractionof a representative data point (emission at 450 nm upon excitationat 340 nm) is shown in Figure 6A. Error bars for the solutionsmeasured in triplicate are included. Interestingly, the greatestdifference was between 100 µM solutions of racemic mannoseand 100% D-mannose. The emission of L-PheC2NTf2 showed highsensitivity to the presence of small amounts of L-mannose inthe mole fraction region between 81.25 and 100% D-mannose.This FCIL was also sensitive to the presence of small amountsof D-mannose to a lesser degree (mole fraction region between0 and 50% D-mannose). It is likely that the other form of thisFCIL, D-PheC2NTf2, would show the opposite behavior and bemore sensitive to small amounts of D-mannose. Unlike typicalmeasurements which use only a single excitation and emission,this work measures the entire EEM to allow differentiation ofcompositions resulting in the same intensity for a single datapoint (for example, points near 25% D-mannose and 81.25%D-mannose). The subtle spectral differences over variousregions of the EEM allow these compositions to be distin-guished. Work is ongoing to further investigate this interestingsystem.

A regression model of each point in the EEM was preparedusing a mole fraction region between 81.25% and 100% D-mannose.The mixture with a composition of 96.875% D-mannose (ee % )93.75%) was excluded from the regression and used as anunknown. The composition of this point was then determined for

each point in the EEM, and these compositions are shown inFigure 6B. Compositions near the true value (96.875% D-mannose)are presented as white, with high values shown in red and lowvalues shown in blue. It is clear that the choice of excitation and

(67) James, T. D.; Sandanayake, K.; Shinkai, S. Nature 1995, 374, 345–347.(68) Liang, X. F.; James, T. D.; Zhao, J. Z. Tetrahedron 2008, 64, 1309–1315.(69) James, T. D.; Sandanayake, K.; Shinkai, S. Angew. Chem., Int. Ed. Engl.

1996, 35, 1910–1922.(70) Wang, W.; Gao, X. M.; Wang, B. H. Curr. Org. Chem. 2002, 6, 1285–1317.

Figure 6. Determination of enantiomeric composition. (A) Intensityas a function of D-mannose mole fraction in L-PheC2NTf2 (ex 340 nm,em 450 nm). This point is represented by a black dot in parts B andC. Part A includes a regression line for the linear region most sensitiveto the presence of small amounts of L-mannose. The empty diamondat 96.875% D-mannose is designated as an unknown. (B) CalculatedD-mannose mole fraction as a function of excitation and emissionwavelength. Actual D-mannose mole fraction is 96.875%. Regionsclosest to the true value are represented as white in the contour map.A 10 data point by 10 data point data matrix (ex 310-355 nm; em445-467.5 nm) giving the most reasonable results is indicated by agray box. (C) EEM of L-PheC2NTf2 with regions of interest defined.See text for further discussion. Black dot and gray square asdescribed above. Horizontal dashed line is an emission spectrumexcited at 300 nm. Vertical dotted line is an excitation spectrummonitored at 400 nm. Top inset is D-mannose mole fraction calculatedbased on excitation spectra. Right inset is D-mannose mole fractioncalculated based on emission spectra.


emission wavelengths has a large effect on the quality of therecovered answer. The long-wavelength emission region betweenabout 425 and 500 nm upon excitations between about 300 and350 nm resulted in the best predictions. The representative datapoint shown in Figure 6A and shown as a black dot in Figure 6Bfalls within this region. The predicted value based upon this pointwas 96.9393, relative error ) -0.066% (ee % ) 93.8786%, relativeerror ) -0.137%). A 10 × 10 block of data points (ex 310-355nm; em 445-467.5 nm) was chosen for further analysis and isshown as a gray box in Figure 6.

The predictions need not be based on a single data point.Additionally, the composition could be determined by using (1)any entire emission spectrum, (2) any entire excitation spectrum,(3) any region of EEM, or (4) the entire EEM. The valuescalculated based on excitation and emission spectra are shownas insets in Figure 6C. As examples, the horizontal dashed line isan emission spectrum excited at 300 nm. The predicted valuebased on this emission spectrum was 95.7529, relative error )1.158% (ee % ) 91.5058, relative error ) 2.394%). The verticaldotted line is an excitation spectrum monitored at 400 nm. Thepredicted value based on this excitation spectrum was 94.9937,relative error ) 1.942% (ee % ) 89.9874, relative error ) 4.013%).It was found that both the excitation and emission spectra resultedin good prediction (<1-2% error). However, it is arbitrary andsomewhat difficult to choose the best excitation and emissionwavelengths. Therefore, the best strategy chooses a large regionof the EEM (the long-wavelength fluorescence highlighted in grayin Figure 6). The prediction based upon this region was 96.8369,relative error ) 0.039% (ee % ) 93.6738, relative error ) 0.081%).For comparison, the prediction based upon the entire EEM was95.8847, relative error ) 1.022% (ee % ) 91.7694, relative error )2.113%). Although further work is needed to demonstrate that theimproved accuracy will translate to even higher ee % in this andother systems, these results compare very favorably with othertechniques based upon regression modeling of spectral data andto other methods of enantiomeric excess analysis such as arecently reported high-throughput screening method with errorson the order of 15%.71,72

CONCLUSIONS

In summary, we have synthesized, characterized, and inves-tigated the enantiomeric recognition properties of a fluorescentamino acid-based chiral ionic liquid, L-phenylalanine ethyl esterbis(trifluoromethane) sulfonimide (L-PheC2NTf2), which wasderived from commercially available starting materials. Thiscompound was obtained as a liquid at room temperature, wasstable to thermal decomposition up to 270 °C, and retained itschiral configuration upon prolonged heating. Importantly,L-PheC2NTf2 was shown to be superior to previously reportednonfluorescent chiral ionic liquids whose applications in enan-tiomeric recognition were limited to less sensitive modes ofdetection or to the use of fluorescence in only a small subsetof intrinsically fluorescent analytes. To the best of our knowl-edge, this is the first report on the use of an FCIL as a solvent,chiral selector, and fluorescent reporter in enantiomericrecognition of both fluorescent and nonfluorescent analytes.Wavelength- and analyte-dependent changes in L-PheC2NTf2

fluorescence emission were observed and provide the potentialfor simultaneous chemo- and enantioselectivity toward multipleanalytes, including the monosaccharides glucose and mannose.Importantly, use of excitation-emission matrices combinedwith judicial selection of excitation and emission wavelengthregions was shown to improve the prediction of enantiomericcomposition.

ACKNOWLEDGMENTI.M. Warner acknowledges the National Science Foundation

(NSF) and Phillip W. West Endowment for support of this work.The authors also acknowledge Ioan Negulescu for help withthermal gravimetric analysis measurements.

SUPPORTING INFORMATION AVAILABLEFigures S-1-S-11. This material is available free of charge via

the Internet at http://pubs.acs.org.

Received for review December 7, 2009. Accepted May 1,2010.

AC9027774

(71) Leung, D.; Folmer-Andersen, J. F.; Lynch, V. M.; Anslyn, E. V. J. Am. Chem.Soc. 2008, 130, 12318–12327.

(72) Leung, D.; Anslyn, E. V. J. Am. Chem. Soc. 2008, 130, 12328–12333.


amino acid-based fluorescent chiral ionic liquid for enantiomeric recognition

Documents