Anal. Chem. 2003, 75, 2131-2139
Ion-Pairing Ability, Chemical Stability, and Selectivity Behavior of Halogenated Dodecacarborane Cation Exchangers in Neutral Carrier-Based Ion-Selective Electrodes Shane Peper, Yu Qin, Philip Almond, Michael McKee, Martin Telting-Diaz, Thomas Albrecht-Schmitt, and Eric Bakker*
Department of Chemistry, Auburn University, Auburn, Alabama 36849
Recently, it has been discovered that carba-closo-dodecaborates can be used as cation exchangers in neutral carrier-based ion-selective chemical sensors. Because of their inherent chemical stability and versatile functionalization chemistries, they offer many advantages that may potentially be exploited for ion analyses that require nontraditional sample conditions, including strongly acidic media. In this work, trimethylammonium salts of undecachlorinated (UCC), undecabrominated (UBC), hexabrominated (HBC), and undecaiodinated (UIC) carborane anions were prepared and evaluated for their potential use in solvent polymeric membrane-based sensors. Computational methods including Natural population analysis and electrostatic mapping were used to predict the ionexchanging ability of each lipophilic anion. In addition, the sandwich membrane technique was used to evaluate the ion-pairing ability of each carborane anion in situ (i.e., within bis(2-ethylhexyl) sebacate (DOS)- and 2-nitrophenyl octyl ether (o-NPOE)-plasticized ISE membranes). The results of the computational and potentiometric studies found that binding affinity of the anions followed the generalized trend HBC > UCC > UBC > UIC. PVC-DOS bulk optode thin films containing the chromoionophore ETH 5315 and a respective anion were used to determine the chemical stability/lipophilicity of the carboranes and tetrakis[3,5-bis(trifluoromethyl)phenyl] borate (TFPB) in acidic media (0.2 M HOAc) under flowing conditions. The studies found that in terms of stability/lipophilicity UIC > UBC > TFPB ≈ UCC . HBC. Electrodes containing a Pb2+-selective ionophore, tert-butylcalix[4]arene-tetrakis(N,N-dimethylthioacetamide)(lead IV), were used to evaluate the functionality of each cation exchanger. An evaluation of response characteristics such as slope and selectivity found that UIC and UBC were quite comparable to the behavior of TFPB. Interestingly, both UIC and UBC showed a marked selectivity improvement over cadmium, pot with log KPb,Cd values of -7.19 and -7.29, respectively, with TFPB giving a value of -5.89. Demonstrating excellent stability and suitable electrostatic properties, the carboranes, UIC in particular, are a very promising 10.1021/ac026056o CCC: $25.00 Published on Web 04/02/2003
© 2003 American Chemical Society
alternative to the tetraphenylborates and should find widespread application in the field of chemical sensors. Potentiometry (i.e., ion-selective electrodes (ISEs)) has become a routine analytical method for rapidly determining numerous analytes in a cost-effective manner.1 ISEs have been developed for ∼60 analytes, and they have found widespread application in the fields of clinical diagnostics and environmental monitoring.2 Probably the most common type of ISE that is used today is that based on a liquid polymer membrane. These electrodes are typically composed of a polymer matrix, usually poly(vinyl chloride) (PVC), a plasticizer, an ion-complexing agent (ionophore), and lipophilic ionic additives (i.e., either alkylammonium salts for anion sensing or tetraphenylborates for cation sensing). The idea of adding tetraphenylborate salts to ion-selective membranes was originally proposed by the group of Simon.3 They found that the presence of lipophilic anions in the sensing membrane drastically reduced the degree of interference encountered from lipophilic sample anions (Donnan exclusion), thus extending the measuring range of the electrodes to higher activities. In addition, for some neutral carriers, optimal selectivity for the primary ion was not realized in the absence of exogenous additives.4 There are several other advantages associated with the use of ionic additives, such as decreased membrane resistance,5 reduction in response time after an activity step,6 and significant change in selectivity for both neutral7 and charged carriers.8,9 * Corresponding author. E-mail:
[email protected]. (1) Wang, J. Electroanalytical Techniques in Clinical Chemistry and Laboratory Medicine; Wiley-VCH: New York, 1988. (2) Bakker, E.; Buhlmann, P.; Pretsch, E. Chem. Rev. 1997, 97, 3083. (3) Morf, W. E.; Ammann, D.; Simon, W. Chimia 1974, 28, 65. (4) Gehrig, P.; Morf, W. E.; Welti, M.; Pretsch, E.; Simon, W. Helv. Chim. Acta 1990, 73, 203. (5) Pretsch, E.; Wegmann, D.; Ammann, D.; Bezegh, A.; Dinten, O.; Laubli, M. W.; Morf, W. E.; Oesch, U.; Sugahara, K.; Weiss, H.; Simon, W. In Ion Measurements in Physiology and Medicine; Kessler, M., Ed.; SpringerVerlag: Berlin, 1985; p 11. (6) Meier, P. C.; Morf, W. E.; Laubli, M.; Simon, W. Anal. Chim. Acta 1984, 156, 1. (7) Eugster, R.; Gehrig, P. M.; Morf, W. E.; Spichiger, U. E.; Simon, W. Anal. Chem. 1991, 63, 2285. (8) Bakker, E.; Malinowska, E.; Schiller, R. D.; Meyerhoff, M. E. Talanta 1994, 41, 881. (9) Schaller, U.; Bakker, E.; Spichiger, U. E.; Pretsch, E. Anal. Chem. 1994, 66, 391.
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003 2131
For several decades, tetraphenylborates have been the only mobile lipophilic anions available for cation sensing. There are several inherent disadvantages, however, associated with their use such as susceptibility toward hydrolysis by acid, oxidants, and light.2,10,11 Even though substitution of electron-withdrawing groups onto the phenyl rings of the parent ion, tetraphenylborate (TPB), decreases acid hydrolysis, it still does occur albeit at a much slower rate.11 For optical sensing applications, a lack of photostability may result in a decrease of the operational lifetime of these sensors. With recent trends in ion sensing moving toward miniaturization, it becomes apparent that the lifetime of these sensors will ultimately be dictated by the lipophilicity of the active sensing components. Moreover, the sample matrix also plays a significant role in determining the required lipophilicity of each membrane constituent, with a higher lipophilicity being required for physiological samples (i.e., whole blood, serum, plasma, and undiluted urine).12,13 For microelectrodes, where in vivo measurements are commonplace, a loss of ionic sites results in a decreased signalto-noise ratio.14 This is of paramount importance for this type of sensor due to the small sample activities being measured, usually resulting in an emf change of only a few millivolts. For ion-selective optodes, several miniaturized sensing platforms have been reported, which include the immobilization of sensing films onto the end of optical fibers,15 self-referencing microspheres,16 and nanoscale intracellular probes.17 Loss of ion exchangers from optodes is crucial to their viability because they typically respond via coextraction or ion-exchange equilibria, whereby a decrease in the concentration of sites results in a decrease in sensor sensitivity. In an attempt to circumvent the effects of site leaching, some have tried to covalently anchor the ion exchanger to the polymer matrix.18-20 This typically involves cumbersome reactions and oftentimes a less than optimal sensor response is observed. Ion-exchanger leaching may also soon play a role in dictating the lower detection limit of ISEs. Now that zero-current ion fluxes are becoming better understood, many electrodes have been optimized with detection limits in the subnanomolar range.21 As transmembrane ion fluxes become reduced, other parameters will contribute to the detection limit, such as leaching of ionic sites. Although recent studies suggest that ion-exchanger leaching currently does not affect the detection limit, it has been suggested (10) Meisters, M.; Vandeberg, J. T.; Cassaretto, F. P.; Posvic, H.; Moore, C. E. Anal. Chim. Acta 1970, 49, 481. (11) Nishida, H.; Takada, N.; Yoshimura, M. Bull. Chem. Soc. Jpn. 1984, 57, 2600. (12) Dinten, O.; Spichiger, U. E.; Chaniotakis, N.; Gehrig, P.; Rusterholz, B.; Morf, W. E.; Simon, W. Anal. Chem. 1991, 63, 596. (13) Oesch, U.; Simon, W. Anal. Chem. 1980, 52, 692. (14) Armstrong, R. D.; Horvai, G. Electrochim. Acta 1990, 35, 1. (15) Kurihara, K.; Ohtsu, M.; Yoshida, T.; Abe, T.; Hisamoto, H.; Suzuki, K. Anal. Chem. 1999, 71, 3558. (16) Tsagkatakis, I.; Peper, S.; Retter, R.; Bell, M.; Bakker, E. Anal. Chem. 2001, 73, 6083. (17) Brasuel, M.; Kopelman, R.; Miller, T. J.; Tjalkens, R.; Philbert, M. A. Anal. Chem. 2001, 73, 2221. (18) Rosatzin, T.; Bakker, E.; Suzuki, K.; Simon, W. Anal. Chim. Acta 1993, 280, 197. (19) Kimura, K.; Sunagawa, T.; Yajima, S.; Miyake, S.; Yokoyama, M. Anal. Chem. 1998, 70, 4309. (20) Reinhoudt, D. N.; Engbersen, J. F. J.; Brzozka, Z.; Vlekkert, H. H. v. d.; Honig, G. W. N.; Holterman, H. A. J.; Verkerk, U. H. Anal. Chem. 1994, 66, 3618. (21) Bakker, E.; Pretsch, E. Trends Anal. Chem 2001, 20, 11.
2132 Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
Figure 1. Chemical structure of the carba-closo-dodecacarborane anions. Atom 1 is C, atoms 2-12 are B, atom 13 is H, and atoms 14-24 are Cl, Br, and I for UCC, UBC, and UIC, respectively. For HBC, atom 1 is C, atoms 2-12 are B, atoms 13 and 15-19 are H, and atoms 14 and 20-24 are Br.
that site leaching may play an important role soon.22 Thus, the development of more robust alternatives to the tetraphenylborates that exhibit improved stability and lipophilicity is warranted. In general, carboranes may be potential candidates for developing new, more versatile cation exchangers due to the manifold advantages afforded by their unique chemistry.23,24 Icosahedral carborane anions are exceptionally stable chemically, electrochemically, and photolytically. With a monovalent anionic charge delocalized throughout a stable cage, the carboranes are relatively noncoordinating. Because of their σ-aromaticity, they also offer an abundance of derivatives possessing tailored lipophilicities owing to their susceptibility to electrophilic substitution at both the carbon and boron vertexes.24,25 Halogenated carborane anions are typically preferred over their parent ion, CB11H12-, due to their weaker coordinating nature, their increased thermodynamic stability, and their decreased nucleophilicity.23,24 Recently, we reported the use of a closo-dodecacarborane anion, 1-HCB11Br11(UBC), as an excellent alternative to the best tetraphenylborate 3,5-bis(trifluoromethyl)phenylborate (TFPB).26 Both ISEs for Na+ and optodes for K+ showed behavior nearly identical to analogous sensors containing TFPB. Concurrently appearing in the literature was a metallacarborane anion containing a Co3+ bridge, which had less than optimal lipophilicity, but did however function as a suitable ion exchanger otherwise.27 In this work, several halogenated carborane anions (see Figure 1) are evaluated as suitable ion exchangers for ion-selective sensors. The fully unsubstituted carborane anion, CB11H12-, was (22) Telting-Diaz, M.; Bakker, E. Anal. Chem. 2001, 73, 5582. (23) Strauss, S. Chem. Rev. 1993, 93, 927. (24) Reed, C. Acc. Chem. Res. 1998, 31, 133. (25) Stasko, D.; Reed, C. J. Am. Chem. Soc. 2002, 124, 1148. (26) Peper, S.; Telting-Diaz, M.; Almond, P.; Albrecht-Schmitt, T.; Bakker, E. Anal. Chem. 2002, 74, 1327. (27) Krondak, M.; Volf, R.; Kral, V. Collect. Czech. Chem. Commun. 2001, 66, 1659.
EXPERIMENTAL SECTION Reagents. For membrane preparation, high molecular weight poly(vinyl chloride), bis(2-ethylhexyl sebacate) (DOS), 2-nitrophenyl octyl ether (NPOE), tert-butylcalix[4]arenetetrakis(N,Ndimethylthioacetamide) (lead ionophore IV), sodium tetrakis[3,5bis(trifluoromethyl)phenyl]borate (NaTFPB), and tetrahydrofuran (THF) were of Selectophore quality from Fluka (Milwaukee, WI). Cesium carborane (CsCB11H12) and silver 7,8,9,10,11,12 -hexabromocarborane (AgCB11Br6H6) were of the highest quality available from Strem Chemicals (Newburyport, MA). Chloride salts of sodium and calcium, and nitrate salts of cadmium and copper, were puriss quality from Fluka. Lead nitrate, sodium hydroxide, and sodium bisulfite were ACS grade from Fisher Scientific (Norcross, GA). All salt solutions were made with deionized Nanopure water (18 MΩ‚cm specific resistance). For syntheses, trifluoromethanesulfonic acid (triflic acid) (99%), bromine (99.8%), and iodine monochloride were purchased from Alfa Aesar (Ward Hill, MA). Iodine was obtained in the highest purity available from Mallinckrodt-Baker (Phillipsburg, NJ). Trimethylammonium chloride (98%), octadecanoic acid chloride, 4-aminoazobenzene, triethylamine (99.5%), and lithium aluminum hydride were acquired from Aldrich. All solvents were ACS grade from Aldrich and used as received. Syntheses. Synthesis of the undecahalogenated carborane trimethylammonium salts, [Me3NH][1-H-CB11X11] (where X ) Cl, Br, and I), was performed according to the procedure described by Xie et al.28 [Me3NH][1-H-CB11Cl11] (UCC). To summarize, a thick-walled Pyrex tube was charged with Cs[CB11H12] (0.08 g, 0.29 mmol), trifluoromethanesulfonic acid (1.0 mL, 11.3 mmol), and iodine monochloride (1.0 mL, 19.6 mmol). Caution: Triflic acid is extremely corrosive and should be handled with care. Proper protective gear should be worn when triflic acid is handled, and it should always be kept away from moisture and oxidizing agents. Tube loading was performed in a glovebox under an inert atmosphere. The tube contents were frozen with liquid nitrogen, sealed under vacuum, and placed in a furnace. The temperature was ramped at 0.5 °C/min to 200 °C and maintained for 48 h. After being cooled to room temperature, the resultant brown residue was treated with a 5% NaOH solution until the pH of the solution reached ∼7, causing the aqueous layer to clear. This was followed
by extraction with diethyl ether (3 × 20 mL). The dark brown ether portions were combined, concentrated to 30 mL, and subsequently treated with 10% NaHSO3 until the solution turned colorless. A concentrated solution of Me3NHCl was then slowly added to precipitate the trimethylammonium (TMA) salt of UCC. The precipitate was filtered, washed twice each with water (25 mL) and a mixture of CH2Cl2/hexanes (1:5, 25 mL), and then dried under vacuum to give [Me3NH][1-H-CB11Cl11] as a white solid (yield 150 mg, 89%). TMAUCC was characterized using ESI-MS and FT-IR. Elemental analysis calculated for C4H11NB11Cl11 (582.82): C, 8.24; H, 2.04; N, 2.40. Found: C, 8.50; H, 2.05; N, 2.32. [Me3NH][1-H-CB11Br11] (UBC). A Pyrex tube was charged with Cs[CB11H12] (0.11 g, 0.40 mmol), trifluoromethanesulfonic acid (1.0 mL, 11.3 mmol), and bromine (1.0 mL, 19.4 mmol). The temperature was ramped at 0.5 °C/min to 200 °C and maintained for 96 h. TMAUBC was obtained as a white solid (yield 350 mg, 81%). ESI-MS, 1H NMR (250 MHz, acetone-d6), and FT-IR were used to confirm product composition. Elemental analysis calculated for C4H11NB11Br11 (1071.00): C, 4.48; H, 1.11; N, 1.31. Found: C, 4.92; H, 1.15; N, 1.33. [Me3NH][1-H-CB11I11] (UIC). A Pyrex tube was charged Cs[CB11H12] (0.10 g, 0.69 mmol), trifluoromethanesulfonic acid (1.0 mL, 11.3 mmol), and I2 (2.0 g, 7.9 mmol). The temperature was ramped at 0.5 °C/min to 240 °C and maintained for 120 h. TMAUIC was obtained as a white-yellow solid (yield 980 mg, 90%). ESI-MS and FT-IR were used to confirm product composition. Elemental analysis is calculated for C4H11NB11I11 (1588.74): C, 3.02; H, 0.75; N, 0.88. Found: C, 4.14; H, 0.82; N, 0.88. TMAUIC was recrystallized at least three times prior to elemental analysis. Discrepancies in the calculated and experimental values for C and H are indicative of residual solvent present in the product. [Me3NH][1-H-CB11H5Br6] (HBC). Approximately 100 mg of Ag[1-H-CB11H5Br6] was dissolved in 2 mL of ethyl acetate. About 5 mL of an aqueous solution of saturated trimethylammonium chloride was added to the ethyl acetate solution, resulting in the precipitation of AgCl in the aqueous phase. Evaporation of ethyl acetate resulted in the recovery of TMAHBC. Due to the high lipophilicity of Ag+, it was necessary to exchange it for TMA+, making it possible to quantitatively exchange the cation during electrode conditioning. Elemental analysis is calculated for C4H16NB11Br6 (676.58): C, 7.10; H, 2.39; N, 2.07. Found: C, 9.93; H, 2.62; N, 2.04. The elemental analysis data again are slightly off for C and H; however, no reaction was done on the carborane anion, so the discrepancy is most likely due to residual solvent. The salt was recrystallized at least three times prior to elemental analysis. The procedure for the synthesis of ETH 5315 has previously been described29 and gave a yield of 163 mg (20%). The structure was confirmed by 1H NMR (250 MHz, CDCl3) and FAB-MS. Optode Leaching Experiments. Cocktails (240 mg total weight) containing 33 wt % PVC, 66 wt % DOS, and 10 mmol/kg each of ETH 5315 (1.1 mg), and ion exchanger, specifically, 2.2 mg of NaTFPB, 3.8 mg (UIC), of 2.6 mg UBC, 1.4 mg of UCC, or 1.6 mg of HBC, were dissolved in 1.8 mL of THF. Using 200-µL aliquots from each cocktail, two membranes of the same composi-
(28) Xie, Z.; Tsang, C.-W.; Sze, E. T.-P.; Yang, Q.; Chan, D. T. W.; Mak, T. C. W. Inorg. Chem. 1998, 37, 6444.
(29) Lerchi, M.; Bakker, E.; Rusterholz, B.; Simon, W. Anal. Chem. 1992, 64, 1534.
not included in this study because it has previously shown poor retention in ion-selective membranes.22,26 Ion coordination ability, chemical stability in acid and/or lipophilicity, and response characteristics (slope and selectivity) were used as criteria. The segmented sandwich membrane technique was used to determine relative ion-pairing trends of the carborane anions within plasticized ISE membranes. In addition, computational methods were used to calculate electrostatic contours for the carboranes in the presence of an approaching point charge, and the charge density distribution for each ion exchanger was determined using natural population analysis (NPA). Slope and selectivity of electrodes containing the ion exchangers and tert-butylcalix[4]arene-tetrakis(N,N-dimethylthioacetamide) were used to evaluate response characteristics.
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
2133
tion were cast onto quartz disks (35 mm in diameter) by means of a spin-coating device.2 After the films were air-dried for 1 h, they were placed in a flow-through cell, which was mounted into a Hewlett-Packard 8452A diode array UV-visible spectrophotometer and filled with 0.2 M HOAc. The solution was continuously replaced at a rate of 1.2 mL/min. Absorption spectra were recorded between 300 and 800 nm at 1-min intervals, with the exception of the films containing HBC, for which the absorbance was recorded every 15 s because of the rapid leaching behavior observed. Replicate experiments were performed to confirm the leaching behavior of each ion exchanger. Electrode Preparation and emf Measurements. For segmented sandwich membrane studies, cocktails (140 mg total weight) contained 10 mmol/kg ion exchanger i.e., TMA salts of carboranes or NaTFPB (0.9-1.1 wt %), plasticizer DOS or o-NPOE 66 wt %, and PVC 33 wt %. ISE membranes of ∼200-µm thickness were prepared by pouring the cocktails, dissolved in 1 mL of THF, into 22-mm glass rings affixed onto glass slides. After solvent evaporation, membranes containing carboranes were preconditioned in 10-2 M LiOH overnight in order to deprotonate the trimethylammonium cation and extract trimethylamine into the aqueous phase. Afterward, the membranes were conditioned overnight in either 10-1 M KCl or CaCl2 solutions. Disks 6 mm in diameter were cut from the parent membrane and assembled into Philips electrode bodies. The internal filling solutions used were 10-1 M KCl or CaCl2 depending on the composition of the conditioning solution. The measuring protocol that follows has previously been reported.30,31 To summarize, single membrane potentials were determined for each membrane containing either TFPB or a carborane anion. Then, the membranes were fused together with the TFPB-containing membrane in contact with the inner solution and the carborane-containing membrane in contact with the sample. The cell assembly used in this work was IFS (0.1 M MzClz) | TFPB | Carborane | sample (0.1 M MzClz), where Mz is either K+ or Ca2+. Measurements were done in triplicate for each carborane evaluated, and the means and standard deviations are reported. For determination of the pKa of ETH 5315, the segmented sandwich membrane method was employed as previously described.32 Membranes were prepared as mentioned above and contained PVC-DOS (1:2) and either 10 mmol/kg chromoionophore and 5 mmol/kg NaTFPB or 5 mmol/kg TFPB only. In contrast to the previous report, a symmetric cell was used, with both the IFS and the sample consisting of 0.01 M HCl. Measurements for pKa determinations were done in triplicate, and the mean and standard deviation are given. For neutral carrier-based ISEs containing Pb2+ ionophore IV and an ion exchanger, cocktails (140 mg total weight) consisted of 10 mmol/kg ionophore, 5 mmol/kg ion exchanger, 33 wt % PVC, and 66 wt % DOS and were dissolved in 1.5 mL of THF. Membranes were fashioned according to the protocol mentioned previously and were conditioned overnight in 10-2 M NaCl. The electrodes were conditioned in NaCl solutions so that unbiased selectivity coefficients could be determined. Disks 4 mm in diameter were cut from the parent membrane and glued to PVC (30) Mi, Y.; Bakker, E. Anal. Chem. 1999, 71, 5279. (31) Qin, Y.; Mi, Y.; Bakker, E. Anal. Chim. Acta 2000, 421, 207. (32) Mi, Y.; Bakker, E. Electrochem. Solid State Lett. 2000, 3, 159.
2134
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
tubing using a PVC/THF slurry according to the technique reported by Ceresa and Pretsch.33 The inner filling solution used was 10-2 M NaCl. Five electrodes were evaluated for each ion exchanger measured, and the mean values and standard deviations are given for response slopes and selectivity coefficients. Before exposure to Pb2+, electrode responses were measured toward Ca2+, Na+, Cd2+, and Cu2+ (from the most to least discriminated ion), according to previous recommendations.34-37 Selectivity coefficients were determined using the separate solution method (SSM) using the potentials from the highest sample activities lying within the Nernstian response range, as previously suggested.34 All emf measurements were made against a Ag/AgCl reference electrode (Metrohm 6.0729.100) with a 1 M LiOAc bridge electrolyte. The instrumentation used to acquire potentiometric data has been described earlier.37 Measurement values were corrected for liquid junction potentials using the Henderson formalism, and ion activities were calculated according to the Debye-Hu¨ckel approximation.38 Computational Details. The 1-HCB11X11-, X ) Cl, Br, I, and 1-HCB11H5Br6- were optimized at the AM1 level of theory39 within the Gaussian program.40 The electrostatic potentials were plotted onto the 0.001 au electron density contour at the STO-3G level using the Spartan program.41 The plots are color-coded with each color indicating the interaction energy (kcal/mol) for a unit positive charge (see legend). The NPA charges42-44 were computed at the B3LYP/3-21G(d)//AM1 level (wave function computed at the B3LYP/3-21G(d) level at AM1 geometries). RESULTS AND DISCUSSION There are several criteria that must be satisfied in order for a prospective ion exchanger to become routinely used in ionselective chemical sensors. Some of the more important characteristics that need to be exhibited include the presence of a low, delocalized charge (so that the electrostatic interactions are minimized), sufficient lipophilicity and chemical stability, and preservation of electrode response characteristics, such as sensitivity and selectivity. Charge delocalization allows for weaker interactions to occur at sites along the periphery of the anion. NPA is one computational (33) Ceresa, A.; Pretsch, E. Anal. Chim. Acta 1999, 395, 41. (34) Bakker, E.; Pretsch, E.; Buhlmann, P. Anal. Chem. 2000, 72, 1127. (35) Bakker, E.; Buhlmann, P.; Pretsch, E. Electroanalysis 1999, 11, 915. (36) Bakker, E. Anal. Chem. 1997, 69, 1061. (37) Bakker, E. J. Electrochem. Soc. 1996, 143, L83. (38) Meier, P. C. Anal. Chim. Acta 1982, 136, 363. (39) Dewar, M. J. S.; Zoebisch, E. G.; Healy, E. F. J. Am. Chem. Soc. 1985, 107, 3902. (40) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J., J. A.; Stratmann, R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A.; Gaussian, Inc.: Pittsburgh, PA, 1998. (41) Spartan, Wavefuntion, Inc.: Irvine, CA. (42) Reed, A. E.; Weinstock, R. B.; Weinhold, F. J. Chem. Phys. 1985, 83, 735. (43) Reed, A.; Curtiss, L. A.; Weinhold, F. Chem. Rev. 1988, 88, 899. (44) Weinhold, F. A. In Encyclopedia of Computational Chemistry; Schleyer, P. v. R., Ed.; Wiley Publishers: New York, 1998; Vol. 3, p 1792.
Table 1. Calculated Natural Population Analysis Charges for Several Carba-closo-dodecaborates UCa
UCC
UBC
UIC
HBC
atoqm
no.
charge
atom
no.
charge
atom
no.
charge
atom
no.
charge
atom
no.
charge
CH BH BH BH BH BH BH BH BH BH BH BH
1 2 3 4 5 6 7 8 9 10 11 12
-0.39 -0.11 0.03 0.03 0.03 0.03 0.03 -0.13 -0.13 -0.13 -0.13 -0.13
C B B B I3 B B B B B B B H Cl Cl Cl Cl Cl Cl Cl Cl Cl Cl Cl
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-0.8019 -0.0060 0.1332 0.1332 0.1332 0.1332 0.1332 -0.0254 -0.0254 0.0254 -0.0254 -0.0254 0.3589 -0.1053 -0.0945 -0.0945 -0.0945 -0.0945 -0.0945 -0.1025 -0.1025 -0.1025 -0.1025 -0.1025
C B B B B B B B B B B B H Br Br Bf Br Br Br Br Br Br Br Br
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-0.7922 -0.1030 0.0259 0.0259 0.0259 0.0259 0.0259 -0.1332 -0.1332 -0.1332 -0.1332 -0.1332 0.3541 0.0008 0.0124 0.0124 0.0124 0.0124 0.0124 0.0029 0.0029 0.0029 0.0029 0.0029
C B B B B B B B B B B B H I I I I I I I I I I I
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-0.7573 -0.2959 -0.1697 -0.1685 -0.1670 -0.1685 -0.1697 -0.3278 -0.3313 -0.3295 -0.3295 -0.3313 0.3596 0.1917 0.2038 0.2035 0.2059 0.2035 0.2038 0.1955 0.1954 0.1940 0.1940 0.1954
C B B B B B B B B B B B H Br Br Br Br Br Br H H H H H
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
-0.7717 -0.1111 0.0408 0.0408 0.0408 0.0408 0.0408 -0.1390 -0.1390 -0.1390 -0.1390 -0.1390 0.3276 -0.0170 -0.0267 -0.0267 -0.0267 -0.0267 -0.0267 0.0393 0.0393 0.0393 0.0393 0.0393
a
UC, unsubstituted carborane, CBl1H12-. Calculated at the B3LYP/6-31G(d) level.45
method that allows the charge density of a molecule to be partitioned among the various atomic nuclei. NPA was selected over Mulliken population analysis because Mulliken values are highly dependent on the basis set used.42-44 The NPA values calculated for the halogenated carboranes are found in Table 1 (for atomic numbering scheme, refer to Figure 1). There are primarily four distinct regions of charge density for each carborane, C1, B2, which is the boron atom opposite to C1 (also referred to as the antipodal position), B3-B7, which denotes the upper pentagonal belt, and B8-B12, which is the lower pentagonal belt. Atoms 13 and 14 are a proton and a halogen attached to the antipodal boron atom, respectively, while atoms 15-19 and 2024 refer to the substituents attached to each boron atom on the lower and upper pentagonal belts, respectively. Interestingly, the lower pentagonal belt possesses more negative charge than the upper belt (both boron atoms and substituents alike). This trend holds for all of the halogenated carboranes as well as for the unsubstituted parent ion (UC). It is also apparent that the substitution of halogens that are less electron withdrawing decreases the charge density on the periphery of the anion, thus shielding the charge located on the boron atoms. This would indicate a more weakly coordinating species, which is favorable in ISEs because it prevents the ion exchanger from interacting with extracted cations, which is sometimes seen with some of the tetraphenylborate anions (i.e., through π-interactions).23 It is also quite surprising that the overall charge magnitude obtained for each boron atom and its substituent is nearly identical for each of the carboranes. For example, if one sums the charge density for the antipodal boron (B2) and its substituent (R14) the overall charge is ∼0.1 for all of the halogenated carboranes as well as for the unsubstituted parent anion.45 This indicates that the amount (45) McKee, M. L. J. Am. Chem. Soc. 1997, 119, 4220.
of charge density for a given atom location on the cage (i.e., boron + substituent) is always the same; the distribution of charge, however, is different depending on the partitioning imparted by the substituents electron-withdrawing ability. Thus, it should be possible to create numerous anions with a variety of coordination abilities, lipophilicities, and solubilities. Furthermore, the susceptibility of carboranes to electrophilic substitution is an inherent advantage that may be exploited for the development of a wide variety of ion exchangers with tailored lipophilicities. Recently, a novel dodecacarborane anion has been described that contains a pentamethylated upper pentagonal belt and a hexahalogenated lower pentagonal belt.25 This initial report is an excellent example that demonstrates how tailored substitutions can be used to optimize the properties of the anion for specific applications. Another trend that is apparent for the undecahalogenated carboranes is the increased amount of charge density on the halogen (X14) bound to the antipodal boron relative to the other substituents of the lower belt. This phenomenon is known as the antipodal effect. It is this position that is the most susceptible to electrophilic attack. One would expect that this location would most likely be involved in weak electrostatic interactions within ISE membranes. The presence of weakly basic sites (i.e., halogens or hydrogen) is also advantageous because it reduces the coordination ability of the anion. Of the isostructural carboranes studied, it appears that UIC may be the best choice based on its superior charge delocalization. HBC is used as a comparative standard for the undecahalogenated carboranes because it is the most lipophilic carborane anion that is commercially available. As expected, the charge density at the antipodal position seems to be similar for HBC and UBC. However, for HBC, the antipodal effect is not observed. Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
2135
Table 2. Potentiometric Evaluation of Ion-Pairing Ability for Carborane Ion Exchangers Relative to the Binding Behavior of TFPBa ∆emf for K+
∆emf for Ca2+
anion
DOS
NPOE
DOS
NPOE
UIC UBC UCC HBC
13.4 ( 0.2 13.3 ( 0.2 18.1 ( 0.5 20.1 ( 0.9
11.6 ( 0.6 10.1 ( 0.7 19.3 ( 0.6 22.6 ( 1.1
22.9 ( 0.1 31.2 ( 0.4 34.6 ( 0.7 39.2 ( 0.7
47.9 ( 0.6 60.4 ( 0.7 72.3 ( 0.6 78.7 ( 0.6
a Mean and standard deviation of three measurements (in mV). Determined using the segmented sandwich membrane technique.
Figure 2. Electrostatic maps for HBC, UCC, and UBC. All anions were optimized at the AM1 level of theory, and the electrostatic potentials were plotted onto the 0.001 au electron density contour at the STO-3G level. The plots are color-coded with each color indicating the interaction energy (kcal/mol) for a unit positive charge (see legend). Negative regions are denoted in red, while positive regions are shown in blue.
Another computational tool that can be used to predict the coordination behavior of the carborane ion exchangers is the electrostatic contour or electrostatic map as it is sometimes called. The electrostatic map is a plot of the electrostatic potential (EP) at a fixed electron density. The electrostatic potential can be defined as the work done to bring unit positive charge from infinity to a point.46 Unlike electron density, the EP contains contributions from both the nuclei and the electrons. The electrostatic interaction energies can be visualized in the same manner as charge densities because the EP varies through space. The electrostatic maps calculated for UCC, UBC, and HBC are shown in Figure 2. Limitations of the Spartan program did not allow for the calculation of an electrostatic map for UIC. In Figure 2, the blue region corresponds to a more positive energy surface and the red region to a more negative energy surface. Often, electrostatic maps are used to predict where (46) Leach, A. R. Molecular Modelling: Principles and Applications, 2nd ed.; Prentice Hall: London, 2001.
2136 Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
electrophilic attack is most likely to occur, with the more negative (red) regions being more prone to substitution.46 This visual observation correlates well with the charge density data for the antipodal substituents in Table 1. Knowing the calculated interaction energies, it is possible to determine a trend for the strength of the ion-exchanger interactions. The interaction energies for UBC, UCC, and HBC, are 54, 64, and 60 kcal/mol, respectively. This means that the order of binding strength is UCC > HBC > UBC. These data validate the prediction that HBC would behave similarly to UCC. Moreover, had it been possible to generate an electrostatic contour for UIC, it is probable that it would appear after UBC in the series and that it should possess an interaction energy of less than 54 kcal/mol. The computational data, which give some prediction of the degree of electrostatic interactions, was compared to potentiometric data obtained using the segmented sandwich technique. This technique has previously been used to determine complex stability constants of ion-ionophore complexes,30,31 to determine chromoionophore pKa values in situ,32 to quantify the amount of endogenous ionic impurities present in plasticized PVC,47 and to determine the coextraction constants of dissolved electrolytes responsible for dictating upper detection limits.48 Here we have adapted it to determine ion-pairing interactions relative to the best tetraphenylborate, TFPB. For this study, single membrane potentials were determined for membranes containing either TFPB or carborane. Then, the membranes were fused together with the TFPB-containing membrane in contact with the inner solution and the carborane-containing membrane in contact with the sample. The potential generated from this configuration is dependent on the interaction of the carborane anion with the sample cations. By subtracting the average single membrane potentials from the average sandwiched membrane potentials, it is possible to determine a ∆emf value that indicates the degree to which a carborane anion has interacted with a sample cation relative to the binding behavior of TFPB. The emf values obtained for the carboranes in DOS and o-NPOE-plasticized membranes are found in Table 2. The data were statistically analyzed at the 95% confidence interval using a Student’s t-test in order to establish trends in binding strength. A positive charge sign indicates that all of the carboranes have a stronger interaction with the ions measured than that exhibited by TFPB. It is apparent that UIC forms the weakest ion pairs relative to the other halogenated carboranes in most cases. For binding studies with K+, the effect (47) Qin, Y.; Bakker, E. Anal. Chem. 2001, 73, 4262. (48) Qin, Y.; Bakker, E. Anal. Chem. 2002, 74, 3134.
of the plasticizer polarity was minimal on binding strength for most of the carboranes. Interestingly, however, in NPOE membranes UBC was slightly preferred over UIC. Conversely, for Ca2+ binding studies, stronger interactions occurred in the more polar o-NPOE plasticizer. It is also apparent from Table 2 that the carboranes interact more strongly with Ca2+ than with K+. This is expected due to the stronger Coulombic interactions exhibited by divalent ions as a result of their higher charge. It is noteworthy to mention that a Nernstian cationic slope was confirmed for each membrane segment in both plasticizers studied. The results of the binding studies show that UIC interacts minimally with cations relative to the other carborane anions. These results confirm the predicted interactions of the computational data found in Table 1 and Figure 2. The sandwich method also clarified the relative position of HBC, which binds slightly stronger than UCC in both polar and apolar membranes. For K+ binding in apolar DOS membranes, the binding trend observed was HBC > UCC > UBC ≈ UIC. Similarly, for Ca2+ binding, the trend was HBC > UCC > UBC > UIC. The affinity for K+ in polar NPOE membranes was found as HBC > UCC > UIC > UBC, while for Ca2+ binding the same trend observed in DOS membranes occurred. Further studies are needed to develop a theoretical explanation for the observed behavior. Other than charge delocalization, there are other characteristics that an ion exchanger must possess in order to be useful for ion-selective sensors. One parameter that affects the lifetime of sensing devices is the lipophilicity of the active components. Several important studies have been done over the years demonstrating the impact that component retention plays on sensor response and longevity.12,13 With today’s goals of miniaturized sensing platforms, the requirements become even more stringent. It has been reported for intracellular ion probes, consisting of nanoscale optodes, that the inherent lifetime of these sensors is ∼30 min.17 An inherent disadvantage of the tetraphenylborates is their susceptibility toward acid hydrolysis.49 It has been known for many years that cleavage of the boron-phenyl bond occurs in acidic media; however, due to a lack of alternative compounds, the tetraphenylborates have continually been used in developing sensors that require harsh acidic sample conditions, such as f-block elements50-53 and toxic heavy metals.33,54 It is known that carboranes are resistant to acid and may be easily functionalized with lipophilic groups. Facile solid-state synthetic routes are now available that produce single products in yields as high as 90%, thus allowing various derivatives to be synthesized.28 To compare the relative stability/lipophilicity of the halogenated carborane anions to TFPB, thin films, ∼2 µm in thickness, were fashioned using DOS-plasticized PVC that contained equimolar amounts of an ion exchanger and a chromoionophore with an acidic pKa value, ETH 5315 (pKa (DOS) ) 4.9 ( 0.03). (The pKa of the chromoiono(49) Bakker, E.; Pretsch, E. Anal. Chim. Acta 1995, 309, 7. (50) Amarchand, S.; Menon, S. K.; Agrawal, Y. K. Electroanalysis 1999, 12, 522. (51) Gupta, V. K.; Mangla, R.; Khurana, U.; Kumar, P. Electroanalysis 1999, 11, 573. (52) Hassan, S. S. M.; Ali, M. M.; Attawiya, A. M. Y. Talanta 2001, 54, 1153. (53) Senkyr, J.; Ammann, D.; Meier, P. C.; Morf, W. E.; Pretsch, E.; Simon, W. Anal. Chem. 1979, 51, 786. (54) Ceresa, A.; Bakker, E.; Hattendorf, B.; Gunther, D.; Pretsch, E. Anal. Chem. 2001, 73, 343.
Figure 3. Chemical stability and/or lipophilicity of TFPB and halogenated carborane anions in the presence of 0.2 M HOAc under flowing conditions. The ordinate values (1 - R) are the mole fraction of protonated chromoionophore.
phore was determined in situ using the segmented sandwich membrane technique. The value found here agrees nicely with the value of 5.2, which was previously obtained indirectly via optical and potentiometric experiments.55) The films were equilibrated in a flow cell that contained 0.2 M HOAc as previously described, and the acid was continuously replaced at a rate of 1.2 mL/min. Upon exposure to acetic acid, the cations present in the films (i.e., TMA+ for carboranes and Na+ for TFPB) are replaced by protons; therefore, the lipophilicity of the cation does not directly affect the rate of ion-exchanger leaching. The leaching behavior of TFPB and the carboranes appears in Figure 3. Because of the resilience of carboranes in acid, it is plausible that the primary mechanism affecting anion leaching is insufficient lipophilicity. It has been reported, however, that halogenated carboranes do not possess optimal solubility in organic solvents; therefore, leaching from the organic thin film may also reflect this potential shortcoming.25 It is noteworthy to mention that a visual inspection of the films prior to measurement resulted in no evidence of ion-exchanger crystallization or film heterogeneity. In contrast to the carboranes, TFPB leaching is predominantly caused by acid hydrolysis and lipophilicity limitations. From Figure 3, it appears that both UBC and UIC are substantially better than TFPB under flowing acidic conditions; however, due to the noticeable decrease, it is necessary for further developments to be made, either through the creation of new derivatives or through covalent modification of existing carboranes. The leaching behaviors of both UBC and TFPB are reasonably comparable to work previously reported, taking into consideration the experimental modifications (i.e., flowing mode vs static).26 Furthermore, it is evident that HBC is not a good choice due to its undesirable retention time. It should also be mentioned that evaluation of the (55) Bakker, E.; Lerchi, M.; Rosatzin, T.; Rusterholz, B.; Simon, W. Anal. Chim. Acta 1993, 278, 211.
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
2137
Table 3. Response Slopesa and Unbiased Selectivity Coefficientsb of PVC-DOS ISEs Containing Pb2+ Ionophore IV and a Respective Ion Exchanger TFPB ion Pb2+ Ca2+ Na+ Cd2+ Cu2+
slope
UIC log
pot KPb,J
slope
UBC log
pot KPb,J
slope
UCC log
pot KPb,J
slope
HBC log
pot KPb,J
slope
pot log KPb,J
30.58 ( 0.61 0 28.95c ( 0.89 0 29.48 ( 2.44 0 33.72 ( 1.37 0 27.11 ( 0.93 0 27.49 ( 0.73 -13.06 ( 0.07 20.11 ( 1.15 -13.31 ( 0.13 19.68 ( 2.92 -12.94 ( 0.27 11.31 ( 0.81 -12.08 ( 0.27 10.39 ( 3.38 -11.89 ( 0 59.72 ( 0.18 -7.27 ( 0.07 55.34 ( 0.42 -7.05 ( 0.06 53.65 ( 0.16 -7.56 ( 0.09 52.71 ( 0.56 -6.90 ( 0.21 51.04 ( 0.39 -6.31 ( 0. 31.22 ( 0.11 -5.89 ( 0.07 29.87 ( 0.21 -7.19 ( 0.04 29.10 ( 0.15 -7.29 ( 0.11 28.57 ( 1.18 -6.08 ( 0.22 26.12 ( 0.78 -6.30 ( 0.1 33.98 ( 0.44 -3.63 ( 0.11 35.14 ( 1.24 -3.61 ( 0.14 34.89 ( 0.76 -4.20 ( 0.06 41.64 ( 1.12 -4.47 ( 0.08 38.58 ( 0.97 -4.02 ( 0.
a Mean values of five electrodes; slope is in mV decade-1. b SSM; mean values calculated using the potentials from the highest sample activities lying within the Nemrnstian response range. c Calculated using experimental values from 10-2 to 10-4 M.
Figure 4. Response behavior of PVC-DOS ISE membranes containing Pb IV and either TFPB (left) or UIC (right). Solid lines denote Nernstian response slopes for each ion.
unsubstituted carborane anion, CB11H12-, was attempted; however, due to its poor lipophilicity, it was not sufficiently retained in the organic matrix and it was unable to fully protonate the chromoionophore; thus, it could not be used for comparison. It is known that hexahalogenated carboranes are much more stable in acid than tetraphenylborates;24 therefore, the loss of HBC from the film is primarily due to insufficient lipophilicity. Surprisingly, the behavior of UCC is very similar to that of TFPB. From the leaching data depicted in Figure 3, once again it appears that UIC is the most favorable carborane of those studied. To evaluate the functionality of the carboranes, ISEs were prepared containing a Pb2+-selective ionophore (lead IV). This ionophore is a calix[4]arene derivative and has been well studied.33,54,56 The electrodes were evaluated in terms of response behavior and selectivity, and the results appear in Table 3. Due to the hydrophilicity of the unsubstituted carborane electrode performance was not evaluated for this anion. It is noteworthy to mention that previous studies have shown that this anion readily leaches from ISE membranes, thus causing a deleterious effect on the lower detection limit.22 The values given in Table 3 are the mean of five electrodes, and the standard deviations are given. The separate solutions method was used in conjunction with the protocol for determining unbiased selectivity coefficients in order to evaluate any possible effects that the carboranes may have on ionophore-mediated selectivity.34 From the data it is apparent that (56) Sokalski, T.; Ceresa, A.; Fibbioli, M.; Zwickl, T.; Bakker, E.; Pretsch, E. Anal. Chem. 1999, 71, 1210.
2138
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
all of the carboranes give a Nernstian response toward lead. Moreover, both UBC and UIC appear to be quite comparable to TFPB, with only slight deviations from Nernstian behavior. Of interest, UIC and UBC both show a marked improvement in the discrimination of Cd2+ by nearly 1.3 orders of magnitude. Figure 4 compares the calibration curves for UIC and TFPB to illustrate this phenomenon. The selectivity values reported here for the system containing TFPB match quite well with previously published work.33 It should be noted that in some instances interfering ions exhibit sub-Nernstian responses for electrodes containing carborane ion exchangers; therefore, the selectivity coefficients reported in such cases represent an upper limit (see Table 3).34 In addition to differences in selectivity, UIC also showed promise for improving detection limits. Calibration curves for UIC and TFPB were extrapolated to the baseline potential of the electrodes in water prior to measurement of the primary ion; this allowed for a crude approximation of the detection limit in the absence of ion fluxes. It must be mentioned that this comparison was merely qualitative due to the super-Nernstian response of the electrodes for lead at lower sample activities. This was expected because of the constitution of the IFS (0.01 M NaCl). For TFPB, the estimated detection limit was -12.97 ( 0.32, while for UIC, it was estimated to be -16.36 ( 0.85 (data not shown). Interestingly, UIC was the only carborane that exhibited this behavior. This result suggests that UIC may be suitable for the realization of ion-selective electrodes with even lower detection limits than what are currently possible.
CONCLUSION Halogenated carboranes offer numerous advantages that may be exploited in ionophore-based sensing platforms. Of the carborane derivatives studied, UIC demonstrated the weakest interactions in both polar and apolar membrane solvents for K+ and Ca2+ as determined using the segmented sandwich membrane technique. Ion-pairing trends relative to TFPB were determined, with UIC exhibiting the weakest interactions. This result correlated nicely with the natural population analysis data and the electrostatic contours. Experiments interrogating the chemical stability/ lipophilicity of the carboranes showed that both UIC and UBC were retained longer than TFPB when in contact with 0.2 M HOAc under flowing conditions. The response and selectivity of Pb2+selective electrodes containing carboranes all demonstrated nearNernstian response slopes for the primary ion; however, the less lipophilic UCC and HBC saw sufficiently sub-Nernstian response slopes and slightly worse selectivity compared to TFPB. Surprisingly, UIC showed a marked improvement in selectivity and a
lower detection limit relative to TFPB. This exciting finding should spark a renewed interest in the development of ion exchangers for creating sensors with improved lifetimes and characteristics, such as selectivity. ACKNOWLEDGMENT The authors gratefully acknowledge funding from the National Institutes of Health (Grants GM59716 and DE14950) and Beckman Coulter, Inc. for this research. S.P. thanks the Electrochemical Society for the Colin Garfield Fink Summer Research Fellowship, and the American Chemical Society Division of Analytical Chemistry and Eli Lilly for an academic year research fellowship.
Received for review August 15, 2002. Accepted February 13, 2003. AC026056O
Analytical Chemistry, Vol. 75, No. 9, May 1, 2003
2139
