Calixarene Monolayers as Quartz Crystal ... - ACS Publications

Anal. Chem. 1999, 71, 142-148

Calixarene Monolayers as Quartz Crystal Microbalance Sensing Elements in Aqueous Solution M. T. Cygan,† G. E. Collins,*,† T. D. Dunbar,‡ D. L. Allara,*,‡ C. G. Gibbs,§ and C. D. Gutsche§

Chemistry Division, Naval Research Laboratory, Code 6116, Washington, D.C. 20375-5342, 185 Materials Research Institute, The Pennsylvania State University, University Park, Pennsylvania 16802, and Texas Christian University, P.O. Box 32908, Fort Worth, Texas 76129

We have examined p-tert-butylcalix[4]arenetetrathiolate (BCAT) monolayers for their potential use as molecular recognition elements for in situ aqueous chemical sensors. Spectroscopic and wetting studies of BCAT monolayers on Au{111} reveal that the calixarene molecules exist in monolayers, preferentially oriented with their phenyl rings parallel to the surface normal axis. Using quartz crystal microbalance (QCM) sensors with goldcoated electrodes, the chemical specificity of monolayers and thin films to a variety of aromatic and aliphatic analytes in aqueous solution was examined. The response of BCAT sensors was compared to the responses of p-tertbutylcalix[4]arene (BCA)- and decanethiolate (DT)-coated QCM electrodes. BCAT is very selective for alkylbenzenes, much more so than either its spray-coated thin-film analogue, BCA, or the highly ordered DT monolayer. From these measurements, the factors behind molecular differentiation in each film are explored. Drawing upon these findings, the roles of cavitation and film order in molecular recognition for calixarene films are discussed. Gravimetric sensors such as quartz crystal microbalance (QCM) and surface acoustic wave (SAW) devices have shown promise as rapid, sensitive, and portable transducer elements for chemical sensors.1 The specificity of such devices is determined by the nature of the sensor-environment interface, the device being typically coated with a thin film chosen for its sensitivity to the analyte of interest. Metal oxide,2 polymer,3-5 biomolecular,6 and thin organic molecular7 films have been applied successfully to detect low levels of analytes in air and/or aqueous media. In * To whom correspondence should be addressed: (e-mail) gcollins@ ccf.nrl.navy.mil or [email protected]. † Naval Research Laboratory. ‡ The Pennsylvania State University. § Texas Christian University. (1) (a) Grate, J. W.; Martin, S. J.; White, R. M. Anal. Chem. 1993, 65, 940A; (b) Anal. Chem. 1993, 65, 987A. (c) McCallum, J. J. Analyst 1989, 114, 1173. (2) Semancik, S.; Cavicchi, R. E.; Kreider, K. G.; Suehle, J. S.; Chaparala, P. Sens. Actuators B 1996, 34, 209. (3) (a) Zellers, E.; Batterman, S. A.; Han, M.; Patrash, S. J. Anal. Chem. 1995, 67, 1092. (b) Zhou, X. C.; Zhong, L.; Li, S. F. Y.; Ng, S. C.; Chan, H. S. O. Sens. Actuators B 1997, 42, 59. (4) McGill, R. A.; Abraham, M. H.; Grate, J. W. CHEMTECH 1994, 24, 27. (5) Grate, J. W.; Abraham, M. H. NRL Memorandum Report 6692, 27 July 1990.

142 Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

vapor detection, work is currently progressing toward an electronic nose or smart sensor array systems.3-5,8 Progress in aqueous sensors has been retarded by complications due to the physical interaction of the sensor with the viscous medium. Quartz crystals are damped significantly (∼7 kHz for a 10-MHz crystal with one face in contact with an aqueous solution) through contact with a liquid solution and are sensitive to temperature changes which affect the viscosity and density of the liquid.9 Additionally, the surface-bound film may undergo nonanalyte-specific viscoelastic changes due to swelling, molecular rearrangement, or deformation.10 More precise control of the physical properties of the liquid and the structural properties of the film is required to minimize these physical effects which can produce spurious sensor response. Reducing the viscoelasticity of the film can be achieved by increasing the rigidity and decreasing the thickness of the film. A rigid sensing monolayer, strongly bound to the substrate and possessing strong intramonolayer forces that yield a stable structure, would meet these requirements. One such example is the alkanethiolate self-assembled monolayer (SAM), which is wellknown to form a highly ordered, stable overlayer on a metal surface.11 Additionally, the physicochemical nature of the SAMenvironment interface is adjustable, and a number of chemically and biochemically specific SAM-analyte systems have been created.12 Such interfacial specificity is an extremely important attribute for thin films to possess in order to offset the decrease (6) (a) Nakanishi, K.; Muguruma, H.; Karube, I. Anal. Chem. 1996, 68, 1695. (b) Bao, L.; Deng, L.; Nie, L.; Yao, S.; Wei, W. Anal. Chim. Acta 1996, 319, 97. (c) Ky¨ blinger, C.; Drost, S.; Aberl, F.; Wolf, H. Fresenius J. Anal. Chem. 1994, 349, 349. (7) (a) Dalcanale, E.; Hartmann, J. Sens. Actuators B 1995, 24-25, 39. (b) Dickert, F. L.; Baümler, U. P. A.; Stathopulos, H. Anal. Chem. 1997, 69, 1000. (8) Grate, J. W.; Rose-Pehrsson, S. L.; Venezky, D. L.; Klusty, M.; Wohltjen, H. Anal. Chem. 1993, 65, 1868. (9) Bruckenstein, S.; Shay, M. Electrochim. Acta 1985, 30, 1295. (10) (a) Martin, S. J.; Frye, G. C.; Senturia, S. D. Anal. Chem. 1994, 66, 2201. (b) Hauptmann, P.; Lucklum, R.; Hartmann, J.; Auge, J. Sens. Actuators A 1993, 37-38, 309. (11) (a) Nuzzo, R. G.; Allara, D. L. J. Am. Chem. Soc. 1983, 105, 4481. (b) Ulman, A. An Introduction to Ultrathin Organic Films; Academic Press: San Diego, CA, 1991. (12) (a) Wink, Th.; van Zuilen, S. J.; Bult, A.; van Bennekom, W. P. Analyst 1997, 122, 43R. (b) Mrksich, M.; Whitesides, G. M. Annu. Rev. Biophys. Biomol. Struct. 1996, 25, 55. 10.1021/ac980659b CCC: $18.00

© 1998 American Chemical Society Published on Web 11/21/1998

in film thickness that leads to fewer potential binding sites and, consequently, a decrease in signal intensity. As a result of their pervasive use in solvents, paints, and fuels, aromatic hydrocarbons are among the most commonly encountered environmentally hazardous chemicals encountered in groundwater and industrial waste sites throughout the United States.13 Efforts to develop sensors for in situ monitoring of aromatic hydrocarbon pollutant levels in aqueous solution have relied predominantly on plastic or silicone-clad fiber optic approaches.14 While the sensitivities are fairly good for these sensors, ranging from 0.4 to 18 ppm for aromatic hydrocarbons such as benzene and p-xylene, the detection mechanism is general and nonspecific, relying solely upon the hydrophobic enrichment of nonpolar organic species within the hydrophobic cladding material. We have been intrigued by the apparent molecular recognition properties of the calixarenes, cyclic molecules that are basketshaped in certain conformations and that act as versatile complexation agents in solution.15 For the analysis of volatile organic compounds, calixarene-based SAMs,16-18 and thin films19 have been employed previously as sensing elements. A class of alkanethiolate-modified calixarenes, referred to as resorcarenes, has recently been self-assembled through thiolate linkages to coinage metal surfaces to form sensing elements.16 These resorcarene SAMs were found to be highly sensitive to dichloromethane and underwent reversible and irreversible binding to a variety of organic compounds in the vapor phase. Initially, analyte molecules were thought to be bound or contained within the upper “cage” of the resorcarenes; however, recent studies propose that analyte molecules may, for some resorcarene films, instead be incorporated within the lower alkane portion of the resorcarene.20 To date, however, there have been no aqueous solution, chemical sensor studies of the calixarene-based SAMs capability for molecular recognition. To examine further the mechanism of molecular analyte binding to calixarene films, and to investigate new substrates for the detection of toxic organic pollutants in aqueous solution, we prepared monolayers of p-tert-butylcalix[4]arenetetrathiolate (BCAT, see Figure 1) on gold QCM electrodes. We investigated the sensor response of BCAT-coated electrodes to a number of dilute aqueous solutions of volatile organic compounds such as benzene, toluene, and xylene that are of current interest in wastewater analysis and remediation. The molecular sensitivity and selectivity of the highly (13) Sittig, M. Handbook of Toxic and Hazardous Chemicals and Carcinogens, 3rd ed.; Noyels Publications: Park Ridge, NJ, 1991; Vols. 1 and 2. (14) (a) Bu ¨rck, J.; Conzen, J.-P.; Beckhaus, B.; Ache, H.-J. Sens. Actuators B 1994, 18-19, 291. (b) Merschman, S. A.; Tilotta, D. C. Appl. Spectrosc. 1998, 52, 106. (c) Blair, D. S.; Bando, J. Environ. Sci. Technol. 1998, 32, 294. (15) (a) Gutsche, C. D. Calixarenes. In Monographs in Supramolecular Chemistry; Stoddart, J. F., Ed.; Royal Society of Chemistry; London, 1989. (b) Bo¨hmer, V., Vicens, J., Eds. Calixarenes: A Versatile Class of Macrocylic Compounds; Kluwer Academic Publishers: Dordrecht, 1991. (16) (a) Davis, F.; Stirling, C. J. M. Langmuir 1996, 12, 5365. (b) Huisman, B.-H.; Thoden van Velzen, E. U.; van Veggel, F. C. J. M.; Engbersen, J. F. J.; Reinhoudt, D. N. Tetrahedron Lett. 1995, 36, 3273. (17) Schierbaum, K.-D.; Weiss, T.; Thoden van Velzen, E. U.; Engbersen, J. F. J.; Reinhoudt, D. N.; Go ¨pel, W. Science 1994, 265, 1413. (18) Hill, W.; Wehling, B.; Gibbs, C. G.; Gutsche, C. D.; Klockow, D. Anal. Chem. 1995, 67, 3187. (19) (a) Hartmann, J.; Auge, J.; Lucklum, R.; Ry¨ sler, S.; Hauptmann, P.; Adler, B.; Dalcanale, E. Sens. Actuators. B 1996, 34, 305. (b) Go¨pel, W. Sens. Actuators. B 1995, 24-25, 17. (20) Greenblatt, J.; Kaushansky, N.; Liron, S.; Dalcanale, E. Electrochem. Soc. Proceed. 1997, 97-19, 141.

Figure 1. Pictorial representation of Au-bound BCAT molecules.

oriented BCAT monolayer coating are compared to that of its randomly deposited thin-film analogue, p-tert-butylcalix[4]arene (BCA), and that of the semicrystalline n-decanethiolate (DT) monolayer. Sensitivity, as used here, describes the magnitude of sensor response to a particular analyte (Hz/ppm); selectivity describes the ability of a sensor to discriminate between different analytes. From these studies, the molecular recognitive properties of each film are assessed. Through complementary spectroscopic studies of the molecular structure of the film coatings, the contributions of film cavitation, structure, and molecular orientation to the binding of solutes are explored. EXPERIMENTAL SECTION Materials. Monolayers for QCM analysis were prepared in a specially made glass cell wherein solution made contact with only one face of the quartz crystal. The 10 MHz quartz crystals with 0.25-cm-diameter Au electrodes were used as resonator substrates (Elchema, Inc.).21 The crystals were cleaned prior to monolayer deposition with hot “pirhana” solution, a 1:4 mixture of H2O2 and H2SO4. (Warning: solution reacts violently with many organic materials and must be used with extreme caution and should not be stored in sealed containers.) Monolayers for spectral analysis were prepared by first evaporating 10 nm of Cr onto 2-in.-diameter quartz disks as an adhesion layer, followed by 200 nm of Au (Aldrich, 99.99%).22 Solution depositions of BCAT onto the Au/Cr/quartz substrates were performed in Teflon containers. All monolayers were multiply rinsed in toluene, acetone, and ethanol. All solvents were analytical grade or better. BCAT used for analysis was prepared from BCA as described elsewhere.23 Monolayers of BCAT were prepared by immersion of Au substrates in 0.5 mM toluene for 4 days. BCA (Fluka, 97%) thin films were spray-coated using a Badger air brush (model 2003) at a low flow rate (30-90-min deposition times) from a 0.5 mM toluene solution. The BCA coatings, which ranged from 200 to 800 nm thick, were visible to the naked eye and exhibited a high degree of visual nonuniformity. BCA films were multiply rinsed in ethanol and water (18 MΩ, Millipore). To preserve the orientational disorder in these films, no annealing was performed, as annealing has been reported to induce order in a related calixarene thin film.24 Decanethiol (Aldrich, 96%) solutions were prepared in ethanol (1 mM), and monolayers were prepared through 4-day immersion of Au-coated quartz crystal in solution. Au-coated quartz crystals were cleaned immediately prior to use with hot pirhana solution. (21) Elchema, Potsdam, NY. (22) Laibinis, P. E.; Whitesides, G. M.; Allara, D. L.; Tao, Y.-T. Parikh, A. N.; Nuzzo, R. G. J. Am. Chem. Soc. 1991, 113, 7152. (23) Gibbs, C. G.; Sujeeth, P. K.; Rogers, J. S.; Stanley, G. G.; Krawiec, M.; Watson, W. H.; Gutsche, C. D. J. Org. Chem. 1995, 60, 8394. (24) Chaaˆbane, R. B.; Gamoudi, M.; Guillaud, G.; Jouve, C.; Gaillard, F.; Lamartine, R. Synth. Met. 1994, 66, 49.

Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

143

Instrumentation. Microgravimetric analysis was performed using an Elchema EQCN-400 nanobalance.21 Coated quartz resonators were placed in a Plexiglas flow-through cell (Universal Sensors, Inc.) and exposed to low- and submillimolar aqueous solutions of analyte.25 One side of the resonator was exposed to air to minimize damping of the quartz oscillation. A peristaltic pump was used to maintain a constant flow rate of 0.67 mL/min for optimal laminar flow.26 The flow-through cell was placed within a Faraday cage to minimize stray electromagnetic interference. Raw data were sampled by a Philips PM6674 universal frequency counter and interpreted by National Instruments LabView software. The mass-frequency response of the resonators was calibrated via underpotential deposition of a single monolayer of Pb and simultaneous monitoring of the resonant frequency. Deposition of a single monolayer of Pb, with deposition boundaries monitored by cyclic voltammetry, was found to result in 60-Hz shifts in resonator frequency. The mass, m, of the Pb monolayer can be approximated by m ) dAM, where d is the areal density of the Pb monolayer (in mol‚cm-2), A is the surface area of the Au resonator electrode (in cm2), and M is the molecular weight of Pb (in ng‚mol-1). Using 1.6 × 10-9 as the areal density of a closepacked Pb overlayer on Au[111], a mass of 64.9 ng/monolayer is obtained.27,28 This results in a frequency-mass relationship of 0.92 Hz‚ng-1. This result is close to that predicted by the Sauerbrey equation, f ) (f02/NFA)m, where f is the measured frequency (Hz), f0 is the fundamental frequency of the quartz resonator (Hz), N is the frequency constant of an AT-cut quartz crystal (1.67 × 106 Hz‚mm), F is the density of the quartz (2.648 g‚cm-3), and A is the surface area of the Au resonator electrode (0.196 cm2).27-29 This predicts a frequency response of 1.15 Hz‚ng-1, which is 125% of the measured value. This discrepancy between actual and theoretical values has been previously attributed to incomplete cleaning of the Au surface.27 Infrared external reflection spectra (IRS) were collected using a N2 (CO2- and H2O-free) purged, custom-modified Fourier transform infrared spectrometer as described in detail elsewhere.22,30 The spectral intensities are reported as reflectivities in absorption units, -log(R/R0), where R and R0 are the reflectivities of the BCAT-covered sample and a reference sample of a freshly UV-ozone cleaned gold-coated reference wafer, respectively.22 Optical function spectra, nˆ (ν˜) + ik(ν˜), where ν˜ is the wavenumber, of the bulk compounds were obtained from transmission spectra of KBr dispersions, as described elsewhere.30 These dispersions contained 1.86% BCAT by weight. Orientational analysis of the monolayer film structures was performed by spectral simulation procedures.30 Film thicknesses were determined by single-wavelength ellipsometry (632.8 nm and 70° angle of incidence) using a real refractive index of 1.66.22 This refractive index was determined using an atom fragment-based QSAR method.31 (The necessary density for this method was taken from the literature.32) Liquid (25) Model 1121, Universal Sensors, Metairie, LA. (26) Masterflex Console, Cole-Parmer Instrument Co., Chicago, IL. (27) Melroy, O.; Kanazawa, K.; Gordon, J. G., II; Buttry, D. Langmuir 1986, 2, 697. (28) Schultze, J. W.; Dickertmann, D. Surf. Sci. 1976, 54, 489. (29) Sauerbrey, G. Z. Phys. 1959, 155, 206. (30) Parikh, A. N.; Allara, D. L. J. Chem. Phys. 1992, 96, 927. (31) Ghose, A. K.; Crippen, G. M. J. Chem. Inf. Comput. Sci. 1987, 27, 21.

144


(water and hexadecane) contact angle measurements were performed in the static mode. Presentation of QCM Data. QCM data were originally obtained as the frequency difference (δf) between the working QCM and a reference QCM operated in air inside the same Faraday cage. The frequency change due to organic sample introduction was taken as the difference between the δf during water exposure and the δf at an equilibrium value after sample exposure. An attempt to compensate for thermal drift during each experiment was made by averaging the δf at pre- and postsample exposure. This compensation was only performed for samples that typically exhibited reversible interactions with the substrate. Since the frequency change of the QCM is dependent on the masses of the adsorbed molecules, it is not an appropriate parameter for comparison of different analytes. The massindependent parameter used in this paper is the number of adsorbed molecules per unit sensor surface area. This is denoted as surface coverage and represented in the units of adsorbed molecules per nanometer squared. BCAT results are additionally presented as an approximate percentage of filled sites on the surface. A “site” is defined here as a 0.22-nm2 area, selected to simulate the minimum surface area of a calixarene molecule positioned with the plane of its rings parallel to the surface normal. However, unknown variables such as the packing of molecules on the surface and the roughness of the substrate preclude a more accurate estimate, and the 0.22-nm2 figure must be considered an approximate lower limit. The percentage of filled sites is computed by dividing the frequency change by the calculated frequency change of a full monolayer (1 molecule/0.22 nm2) of adsorbed analyte. The calculated frequency change, using the measured frequency-to-mass relation of 0.92 Hz/ng, is 0.137 (Hz‚ mol)/g. As an example, toluene, with a molecular weight of 92.14, has a calculated monolayer frequency change of 12.6 Hz. RESULTS AND DISCUSSION Surface Characterization. The film thickness of BCAT on Au[111] was determined to be 8.2 ( 2 Å by single-wave ellipsometry, as described earlier. Wetting studies yielded a water contact angle of 74° and a hexadecane contact angle of 4°. These results portray the BCAT-liquid interface as hydrophobic and strongly oleophilic and are consistent with a monolayer-solution interface composed of disordered tert-butyl groups. Taken together, these studies confirm the formation of a single monolayer bound through thiol groups to the underlying gold substrate (see Figure 1). This has been shown, albeit indirectly, for BCAT selfassembled on silver, where surface-enhanced Raman spectroscopy found a loss of SH bonding during formation of the BCAT monolayer.18 Some spectroscopic analysis of the high-frequency IR region of p-tert-butylcalixarene, the parent compound of BCAT, has previously been done by Keller et al.33 Also, some characteristic low-frequency peaks of this tetramer have been listed by Gutsche.34 As mentioned above, we determined the intrinsic optical (32) Delaigue, X.; Harrowfield, J. M.; Hosseini, M. W.; Cian, A. D.; Fischer, J.; Kyritsakas, N. J. Chem. Soc., Chem. Commun. 1994, 13, 1579. (33) Keller, S. W.; Schuster, G. M.; Tobiason, F. L. Polym. Mater. Sci. Eng. 1987, 57, 906. (34) Gutsche, C. D. Structural Chemistry; Springer-Verlag: Berlin, 1984; Vol. 123, p 3.

Figure 2. Isotropic absorption coefficient spectra (k spectrum) of the low-frequency part of bulk polycrystalline phase of BCAT.

constants of BCAT through infrared spectra of dispersions of BCAT in KBr. A detailed analysis of the vibrational modes for a molecule as large as p-tert-butylcalix[4]arenetetrathiol is complicated and outside the scope of this report. The spectrum of bulk BCAT is shown in Figure 2 in the intrinsic form of k(ν˜), the imaginary part of the complex optical function. Appearing in the high-frequency part of the bulk-phase spectra of BCAT were several C-H stretching modes attributable to the methyl, methylene, and aromatic stretching modes. However, uncertainty with respect to the rotational motion of these alkyl groups precludes their use in orientation analysis, and the aromatic mode is not strong enough to detect in the monolayer spectra. A weak stretch in the bulkphase spectra appearing at 2560 cm-1 was assigned to the SH stretch. Although this peak is absent in the experimental reflection spectra, it is so weak that its absence cannot be taken as evidence of thiol to gold bonding. As a first approximation, the vibrational modes in BCAT can be viewed as being the same as those arising from an individual, tetrasubstituted phenyl ring. The substitution pattern of the individual ring can be described as 1,3,5-tri-“light”-2-“heavy” according to Varsańyi’s categorization methods.35 Many of the modes in this category are the same as those for 1,3-di-“heavy”2,5-di-“light”. For purposes of orientation analysis, we define the tilt and twist of the phenyl ring with respect to the surface normal according to Figure 3. The assignments of modes whose origin and transition dipole moment direction are fairly certain are listed in Table 1. For molecules with the substitution pattern of BCAT, mode ν8b typically occurs higher in frequency than mode ν8a.35 Thus, we can definitively assign the mode at 1593 cm-1 to ν8a and that at 1558 cm-1 to ν8b. Similarly, the peak at 1479 cm-1 is assigned to mode ν19a as it falls in the typical range of 1470-1490 cm-1. The two modes that appear at 869 and 879 cm-1 must be out-ofplane C-H bending modes ν11 and ν17a. Since transition dipole moment directions for both modes are in the Y-direction (out-ofplane), it is not necessary to differentiate between the two for purposes of orientational determinations. Simpson and Sutherland reported that 2,2-dimethylalkanes have a skeletal stretching mode (35) Varsańyi, G. Assignments for vibrational spectra of seven hundred benzene derivatives; John Wiley & Sons: New York, 1974.

Figure 3. Coordinate diagram defining the tilt and twist angles (θ, ψ) of the aromatic ring (sulfur to tert-butyl group) from the surface normal. Table 1. Band Assignments for Selected Vibrational Modes of BCAT mode

bulka

aryl C-H op (ν11 or ν17a) aryl C-H op (ν17a or ν11) C(CH3)3 skeletal CdC radial str (ν13) CdC tag str (ν14) CdC str (ν19a) CdC tang str (ν8b) CdC tang str (ν8a)

869 879 1199 1223 1279 1479 1558 1593

SAM

1202 (sh) 1217 1287 1592

Mb Y Y Z Z X Z X Z

a Vibrational mode positions are in wavenumbers (cm-1). b Transition dipole moment direction (see Figure 3).

with its transition moment parallel to the 2 and 3 carbons.36 We assign the 1199-cm-1 vibration to this mode. There are also several aromatic ring modes for which assignments are probable. Varsańyi suggested that phenols cause mode ν14 to shift to higher frequencies. Since BCAT is a thiol, we deduce that the shift will be to the higher end of the range of 1205-1280 cm-1 and that the mode at 1279 cm-1 arises from ν14. Mode ν13 probably occurs at 1223 cm-1. The most reliably defined value in the orientation of BCAT on the gold surface is the tilt. Figure 4 shows the experimental spectra of BCAT self-assembled on gold and the best-fit simulation based on an 8.2-Å-thick film. The best-fit tilt angle is 20 ( 10° from the surface normal (see Figure 5). The average twist of each ring has little effect on the spectral intensity even for the relatively intense modes (ν17a and ν11) and thus cannot be accurately determined.37 Sensor Response. To assess the molecular recognition properties of BCAT, monolayers were formed through spontaneous adsorption onto Au QCM electrodes and exposed to aqueous (36) Simpson, D. M.; Sutherland, G. B. B. M. Proc. R. Soc. 1949, A199, 169. (37) The intensities of out-of-plane modes ν17a and ν11 are maximized for a twist value of 90° for each particular tilt. Even at a twist of 20° and twist of 90°, the intensity of the out-of-plane modes is 6 × 10-5 absorbance unit, just barely above the signal-to-noise ratio at that part of the experimental spectra. For modes in the x-direction, the case is even worse, as modes ν8b and ν14 are relatively weak, consequently having intensities in the simulated spectra below the signal-to-noise level no matter what the twist is defined as for a tilt of 20°. Indeed, the tilt need only be decreased to 15° before no modes indicative of tilt or twist will show up based on our noise levels of ∼5 × 10-5 absorbance unit. Therefore, we refrain from assigning any value to the average phenyl ring twist.


145

Table 2. Response of DT-Coated and Uncoated Au Sensors surface coveragea

response, Hz analyte (5 mM)

DT

Au

DT

Au

tetrahydrofuran cyclohexanone dichloromethane chloroform benzene nitrobenzene m-dihydroxybenzene o-dihydroxybenzene phenol

1 4 1.3 2 b 5 1 2 2

0.5 2 0.5 b b 2 2 2 2.5

0.46 1.4 0.49 0.56 b 1.4 0.30 0.61 0.71

0.23 0.68 0.20 b b 0.54 0.61 0.61 0.89

a

Figure 4. Low-frequency region plot of experimental spectrum of BCAT self-assembled on a gold surface. The dotted lines are best-fit simulated spectra using the optical function spectra of the bulk compound. Note that though the simulation shown is for a tilt of 20° and twist of 45°, the modes indicative of twist for a phenyl ring tilted at 20° are so weak as to not be useful for making a definitive assignment of the twist.

Molecules per nanometer squared. b No response.

Table 3. Response of BCAT and BCA Sensors response,f Hz analyte (0.5 mM) tetrahydrofuran cyclohexanone nitrobenzene benzene p-xylene m-xylene o-xylene toluene m-dihydroxybenzene phenol 1,3,5-trihydroxybenzene

surface coveragea

% filled sitesb

BCAT BCA BCAT BCA BCAT BCA 1.5 1.8 2.5 1.0 15 12 12 10 2 1 d

3 c c 2 6 6 6 2 d d d

0.69 0.61 0.68 0.43 4.7 3.8 3.8 3.6 0.61 0.35 d

1.4 c c 0.85 1.9 1.9 1.9 0.72 d d d

15 13 15 9.4 104e 83 83 80 13 7.8 d

31 c c 19 41 41 41 16 d d d

a Molecules per nanometer squared. b Molecules of analyte per (interfacial) molecule in film. c Adsorption not reproducibly reversible. d No response. e Datum over 100% may reflect uncertainty in the BCAT surface packing (see text) or an additional amount of analyte entering film. f Noise in analyte response, 0.5 Hz.

Figure 5. Drawing of BCAT in its most likely orientation on a gold surface.

solutions of organic molecules at the ppm level. The interaction of the BCAT monolayer with the organic analytes was measured as a change in the resonant frequency of the quartz sensor. Two other films, as well as an uncoated Au sensor, were also studied to provide information on the mechanism of molecular recognition. BCA thin films (∼400 nm), acting as randomly oriented BCAT analogues, were chosen to study the influence of calixarene orientation on film-analyte recognition. DT, which forms a closepacked, crystalline overlayer, was chosen to represent an insensitive, nonselective sensor and was used as a background reference. Uncoated Au sensors were also used as background references and performed similarly to DT-coated sensors. The organic analytes chosen for study possessed a wide array of structural and functional groups. One additional fundamental requirement for selection was analyte aqueous solubility. As solubility may have a large impact on solute-film molecular partitioning, a wide solubility range was investigated to determine the extent of its effect on film selectivity. For BCA and BCAT trials, a lower solubility limit of 0.5 mM was discovered to be necessary to yield reliable sensor response for the entire group of analytes. Test solutions for the more unresponsive Au and DT sensors were prepared at 5 mM concentrations, which unfortunately exceeds the solubility limits of such molecules as the 146 Analytical Chemistry, Vol. 71, No. 1, January 1, 1999

xylenes (dimethylbenzenes), which were of interest due to their large response with BCAT films. The set of molecules studied here includes chlorinated methanes (CH2Cl2, CHCl3), aliphatic rings (THF, cyclohexanone), and a variety of substituted benzenes (xylenes, hydroxybenzenes, nitrobenzene). Table 2 lists the sensor responses of DT-coated and Au sensors to 5 mM aqueous analyte solutions. For both of these sensors, the average response was ∼2 Hz and yielded an average surface coverage of 1 molecule/2.0 nm2 sensor surface area. Neither sensor could detect benzene at the studied concentration. For the DT-coated sensor, two of the analytes, cyclohexanone and nitrobenzene, produced a markedly higher sensor response. The average surface concentration of these analytes was 1 molecule/ 0.7 nm2. In general, both sensors showed a similar pattern of response to the analytes. Table 3 shows the sensor response of BCAT-coated sensors to 0.5 mM aqueous analyte solutions. Analyte responses can be neatly separated into two categories, alkylbenzenes and nonalkylbenzene molecules. For non-alkylbenzene molecules, the average BCAT response was ∼1.6 Hz and yielded an average surface coverage of 1 molecule/1.9 nm2 sensor surface area. In contrast, for alkylbenzene analytes, the average response was ∼12 Hz and yielded an average surface coverage of 1 molecule/0.25 nm2 sensor surface area. This surface coverage is nearly that of

the BCAT molecules in the sensing monolayer, resulting in a nearly one-to-one correspondence between adsorbed alkylbenzene analyte and bound calixarene molecule. This near-unity correspondence is obtained by assuming a close-packed BCAT monolayer on a flat Au{111} plane. Less dense packing would result in an average of more than one analyte molecule per BCAT; increased surface roughness would decrease that ratio. Comparison of the average percentage of BCAT sites filled with alkylbenzene molecules (∼90%) to the average percentage of filled sites for other analytes (∼15%) highlights the magnitude of this response and the resulting specificity of this sensor. BCAT interacts much more strongly with alkylbenzenes than with other molecules that have similar size, shape, or aqueous solubility; this is a strong indication of molecular recognition. Table 3 also presents a direct comparison of BCA and BCAT responses to 0.5 mM analyte solutions. The percentage of analytefilled BCA sites at the film-solution interface is given, although this assumes analyte penetration only into the outer BCA layer as well as a surface packing identical to the BCAT film. BCA exhibited sensitivity to non-alkylbenzene analytes similar to that of BCAT. Results for BCA with nitrobenzene and cyclohexanone were omitted, as they did not interact entirely reversibly with the unannealed BCA film. BCAT shows a slight preference for p-xylene out of the three xylenes studied. Like BCAT, BCA shows selectivity toward the alkylbenzenes, although to a reduced degree, and with no particular selectivity within the xylenes. The ability of BCAT monolayers to increase the sensitivity of the QCM probe is apparent in its lower limit of detection as compared to uncoated or DT-coated probes. BCAT detection limits for the analytes studied were in the low-ppm range, with a detection limit of 3 ppm for p-xylene (most sensitive) and 39 ppm for benzene (least sensitive). For the DT-coated and uncoated sensors, benzene remained undetectable even at 390 ppm. The detection limits for the BCAT probe are in the same order of magnitude as those reported by Lucklum et al., who obtained a range of detection limits from 1 to 10 ppm for the benzene response of a variety of coated QCM sensors in aqueous solution.38 A study of the dependence of sensor response on analyte concentration over the 0.5-20 mM regime showed an approximately linear relationship to analyte concentration for each sensor. Specificity. As discussed above, each of the coatings displays a different response pattern to the analytes studied. In the case of BCAT, for instance, alkylbenzenes are highly preferred over other organic ring systems. In vapor-phase systems, factors such as analyte boiling point,20,39 electrostatic analyte-film interaction,5 and Lewis acid-base properties45,40 have been shown to influence the film selectivity of both polymers and monolayers. For the systems studied here, all of which are in solution rather than in vapor, analyte aqueous solubility might reasonably be expected to influence molecular film-solution partitioning. Figure 6 shows a comparison of the solubility and adsorption of analyte molecules into BCAT and DT films. Values for aqueous solubility are taken from literature.41 The sensor response of the (38) Lucklum, R.; Ry¨ sler, S.; Hartmann, J.; Hauptmann, P. Sens. Actuators. B 1996, 35-36, 103. (39) Patrash, S. J.; Zellers, E. T. Anal. Chem. 1993, 65, 2055. (40) Thomas, R. C.; Ricco, A. J.; DiRubio, C. R.; Yang, H. C.; Crooks, R. M. Electrochem. Soc. Proc. 1997, 97-19, 202.

Figure 6. Relationship of sensor response of DT-coated (triangle) and BCAT-coated (square) sensors to analyte aqueous solubility. Analyte concentrations are 0.5 mM for BCAT sensors and 5 mM for DT sensors. Analytes are (a) nitrobenzene, (b) cyclohexanone, (c) phenol, (d) tetrahydrofuran, (e) and 1,3-dihydroxybenzene.

DT films increases with decreasing analyte aqueous solubility. The response of the BCAT films, however, is indifferent to the analyte aqueous solubility. For DT films, solubility is a reasonably good predictor of adsorption and accounts for the heightened response of the film to nitrobenzene and cyclohexanone where no obvious chemically specific film-analyte interaction is present. This would also be consistent with the nonspecific binding expected from the methyl-like DT solution interface, noted recently in a SAW study by Ricco and co-workers.40 The quick, reversible sensor responses of bare Au and DT-coated Au, coupled with their adsorption-solubility behavior seen in Figure 6, indicates primarily nonspecific adsorption. Adsorption into these films has been observed to occur preferentially at defect sites such as kinks, terrace steps, and in the case of alkanethiolates, orientational domain boundaries.42,43 In contrast, the slower sensor response of BCAT films, their high preference for alkylbenzenes, and the lack of a direct solubility-adsorption relationship indicates that nonspecific binding is not the predominant adsorption mechanism defining their selectivity. Probable sources of the specific binding seen in BCAT films are electrostatic and/or topological interactions. As shown in Table 3, BCAT monolayers distinguish between molecules with similar structures such as toluene and phenol, and also m-xylene and m-dihydroxybenzene, providing evidence for the effect of electrostatic film-analyte interactions on binding. Topological considerations such as film cavitation, however, also appear to play an important role in film-analyte interaction. As shown in Table 3, BCAT exhibits a slightly greater response to p-xylene than to either o- or m-xylene, while the less oriented layer of BCA does not. BCAT presents preferentially oriented cavities toward the solution; BCA presents a less well defined surface, with fewer molecules offering direct access to the interior cavities. This orientational difference is reflected in the difference in the magnitude of the BCAT and BCA responses and may be a cause of discrimination between the xylenes. We must consider the possibility, however, that the analyte molecules actually prefer (41) Howard, P. H., Meylan, W. M., Eds. Handbook of Physical Properties of Organic Chemicals; Lewis Publishers: Boca Raton, FL, 1997. (42) Stranick, S. J.; Kamna, M. M.; Weiss, P. S. Science 1994, 266, 99. (43) Cygan, M. T.; Dunbar, T. D.; Arnold, J. J.; Bumm, L. A.; Shedlock, N. F.; Burgin, T. P.; Jones, L., II; Allara, D. L.; Tour, J. M.; Weiss, P. S. J. Am. Chem. Soc. 1998, 120, 2721.


147

the sites in between neighboring BCAT molecules of proper orientation rather than inside a particular molecular cage. As the degree of molecular packing in the BCAT monolayer is not wellknown, neither explanation can be ruled out from these experiments. However, Schierbaum and co-workers, in comparing surface and bulk layers of calixarenes, determined that accessibility to the interior calixarene cavities was an important factor in film-analyte activity.44 In addition, the well-documented ability of calixarenes to encapsulate these types of small organic molecules, including solid-state complexes of benzene, toluene, and xylene with tert-butylcalix[4]arene, would certainly suggest that in-cage binding is the more likely mechanism.15b CONCLUSIONS The ability of the surface-bound thiolcalixarenes to perform in-cage binding is of particular interest for the future design of monolayer-based sensors. Thiolcalixarene monolayers exhibit strong molecular selectivity despite a low degree of conformational mobility due to the formation of strong thiol-Au bonds. The nearunity alkylbenzene binding shown in this study demonstrates that the upper rim of calixarenes can play an important role in chemical sensor selectivity. This result is in contrast to studies of solvated calixarenes, wherein it has been determined that the lower rim, typically functionalized with hydroxyl or electron donating or -receiving groups, is primarily responsible for the formation of (44) Schierbaum, K.-D.; Gerlach, A.; Go¨pel, W.; Mu ¨ ller, W. M.; Vo¨gtle, F.; Dominik, A.; Roth, H. J. Fresenius J. Anal. Chem. 1994, 349, 372. (45) Arnand-Neu, F.; Barrett, G.; Harris, S. J.; Owens, M.; McKervey, M. A.; Schwing-Weill, M. J., Schwinte, P. Inorg. Chem. 1993, 32, 2644 and references therein. (46) Gutsche, C. D.; See, K. A. J. Org. Chem. 1992, 57, 4527 and references therein.

148


analyte complexes.45,46 Finally, the degree of analyte binding in the thiolcalixarene monolayer was shown to be strongly dependent upon the presence of properly oriented calixarene cavities, unlike the findings of a resorcarene binding study by Greenblatt et al., in which cavitation was found to have no discernible effect on sensor selectivity or sensitivity.20 The preference of BCAT coatings for alkylbenzenes may make them useful sensor elements in an array-based, or “chemical nose”type analysis system. While the detection limits of this system are currently too high for immediate application to most real-world systems, it is reasonable to expect that improvements to the QCM apparatus, or use of a similar technique such as flexural plate wave devices, might decrease the background noise and allow greater sensitivity. Additionally, DT monolayers show promise for providing a reliable, nonspecific sensor interface which could be used as an in situ “blank” or reference sensor in an array to compensate for physical changes in the analyte medium. Future plans include investigation of the electrochemical properties of calixarene monolayers and other structured thin films and the application of coated sensor arrays to multianalyte systems. ACKNOWLEDGMENT Support for this work by the Office of Naval Research is gratefully acknowledged. M.T.C. is a National Research Council Postdoctoral Fellow.

Received for review June 16, 1998. Accepted October 8, 1998. AC980659B