Anal. Chem. 2000, 72, 3696-3708
Conferring Selectivity to Chemical Sensors via Polymer Side-Chain Selection: Thermodynamics of Vapor Sorption by a Set of Polysiloxanes on Thickness-Shear Mode Resonators Andreas Hierlemann,*,†,‡ Antonio J. Ricco,*,†,§ Karl Bodenho 1 fer,| Andreas Dominik,| and | Wolfgang Go 1 pel
Microsensor Research & Development Department, Sandia National Laboratories, Albuquerque, New Mexico 87185-1425, and Institute of Physical and Theoretical Chemistry, University of Tu¨bingen, Auf der Morgenstelle 8, D-72076 Tu¨bingen, Germany
Entropy of mixing is shown to be the driving interaction for the endothermic physisorption process of organic vapor partitioning into seven systematically side-chainmodified (polar, acidic, basic, polarizable side groups and groups interacting via H-bridges) polysiloxanes on thickness-shear mode resonators. Each sensor was exposed to seven analytes, selected for their diversity of functional groups. This systematic investigation of sorption yields benchmarking data on physisorption selectivity: response data and modeling reveal a direct correlation of partition coefficients with interactions between specific polymer side chains and analyte functional groups. Partition coefficients were determined for every polymer/analyte pairing over the 273-343 K range at 10 K intervals; from partition coefficient temperature dependence, overall absorption enthalpies and entropies were calculated. By subtracting the enthalpy and entropy of condensation for a given pure analyte, its mixing entropy (primarily combinatorial) and mixing enthalpy (associated with intermolecular interactions) with each polymer matrix were determined. These two crucial thermodynamic parameters determine the chemical selectivity patterns of the polymers for the analytes. Simple molecular modeling based on the polymer contact surface share of the modified side group or the introduced functional group reveals a direct correlation between the partition coefficients and the side-group variation. Mass-sensitive transducers are commonly used for monitoring volatile organic compounds (VOCs) using polymer layers as chemically sensitive interfaces. Two devices are most often utilized: the thickness-shear mode resonator (TSMR) and the surface acoustic wave (SAW) device.1-4 As is the case for all †
Sandia National Laboratories. Present address: Physical Electronics Laboratory, IQE, ETH Hoenggerberg HPT-H4.2, CH-8093 Zurich, Switzerland. (Phone) ++41 1 633 3494 (fax) ++41 1 633 1054 (e-mail)
[email protected]. § Present address: ACLARA Biosciences, Inc., 1288 Pear Ave., Mountain View, CA 94043-1432. | University of Tu ¨ bingen. ‡
3696 Analytical Chemistry, Vol. 72, No. 16, August 15, 2000
chemical sensors, selectivity, speed of response, and reversibility are a consequence of the thermodynamics and kinetics of sensor material/analyte interactions. Consequently, it is necessary to compromise between high selectivity, typically associated with strong interactions, and perfect reversibility requiring weak interactions. To ensure reversibility, sensors coated with conventional organic polymers exhibiting only partial selectivity to most VOCs are commonly utilized. Analyte identification is then accomplished by using an array of different, partially selective sensors and applying mathematical methods of data evaluation.5-8 To obtain fundamental data on polymer/analyte interactions and derive physicochemical variables (sorption enthalpy and entropy) that are key to partial selectivity, a family of systematically side-chain-modified polysiloxanes and a set of carefully selected analytes were used in this study. Selecting only one type of polymer backbone, the polysiloxanes, and modifying the side chains offer several advantages.9 The influence of differences in the polymer backbone on the selectivity pattern is eliminated, and sorption behavior can be tailored by introducing specific side chains. Because acoustic wave device responses are affected by polymer mechanical properties, the close similarity of the temperature-dependent polymer modulus for coatings with an identical backbone is important, particularly with regard to the intentionally wide temperature range in the present study. A final benefit is (1) Ballantine, D. S.; White, R. M.; Martin, S. J.; Ricco, A. J.; Frye, G. C.; Zellers, E. T.; Wohltjen, H. Acoustic Wave Sensors: Theory, Design, and PhysicoChemical Applications; Academic Press: San Diego, 1997. (2) Bodenho ¨fer, K.; Hierlemann, A.; Noetzel, G.; Weimar, U.; Go ¨pel, W. Anal. Chem. 1996, 68, 2210. (3) Grate, J. W.; Martin, S. J.; White, R. M. Anal. Chem. 1993, 65, 940A948A, 987A-996A. (4) Nieuwenhuizen, M. S.; Venema, A. Sens. Mater. 1989, 5, 261-300. (5) Carey, W. P.; Beebe, K. R.; Kowalski, B. R. Anal. Chem. 1986, 58, 149153. (6) Lavine, B. K. Anal. Chem. 1998, 70, 209R-228R. (7) Osbourn, G. C.; Bartholomew, J. W.; Ricco, A. J.; Crooks, R. M. Acc. Chem. Res. 1998, 31, 297. (8) Hierlemann, A.; Schweizer-Berberich, M.; Weimar, U.; Kraus, G.; Pfau, A.; Go ¨pel, W. Pattern Recognition and Multicomponent Analysis. In Sensors Update; Baltes, H., Go ¨pel, W., Hesse, J., Eds.; VCH: Weinheim, Germany, 1996. (9) Hierlemann, A.; Kraus, G.; Weimar, U.; Go ¨pel, W. Sens. Actuators B 1995, 26, 126-134. 10.1021/ac991298i CCC: $19.00
© 2000 American Chemical Society Published on Web 07/08/2000
To frame the detailed discussion of absorption thermodynamics that follows, an overview of the responses of the polymers to the analytes is presented. The polar plots in Figure 2 display the partition coefficients of the selected analytes sorbed by six polysiloxanes. Partition coefficients are dimensionless thermodynamic equilibrium constants (“enrichment factors”) and are characteristic of a given organic volatile/polymer combination:
K ) cpoly/ca
Figure 1. Systematically side-chain-modified polysiloxanes: poly(dimethylsiloxane) (PDMS, dispersion only), poly(methyloctylsiloxane) (PMOS, dispersion only), poly[methyl(cyanopropyl)siloxane] (PMCPS, polar, weakly basic, dispersion), poly(methylphenylsiloxane) (PMPS, polarizability, dispersion), poly[methyl(aminopropyl)siloxane], 8% amino groups (PMAPS, basic, polar, dispersion), poly[methyl(isopropylcarboxylic acid)siloxane], 10% acid groups (PMiPCAS, acidic, polar, dispersion), and poly[methyl(2-carboxy(D-valinyl-tert-butylamide)propyl)siloxane], 10% valine groups (Chirasil-Val, CSVAL, slightly polar, H bonding).
the possibility to perform simple molecular modeling (contact surface calculations) with separation of the sorption effects of the side group from those of the backbone. Polysiloxanes have been shown to be very stable over extended times and temperatures, exhibiting extremely low polymer moduli and glass transition temperatures.9 To permit a systematic investigation of sorption effects, variation of the polymer side-chain functional groups was made according to the various fundamental physisorption mechanisms, which include (I) dispersion (induced dipole/induced dipole interactions), (II) polar interactions (dipole/dipole or dipole/ induced dipole), (III) polarizability (free electrons, π-electron systems), and (IV) Lewis acid/base interactions, including H bonding (Figure 1). It is difficult to realize only one type of interaction since most functional groups have more than one way to interact with a candidate analyte molecule: dispersion is always present; an acidic or basic group is in most cases also polar, and vice versa. The matrix of cross-correlations of these interactions is described in detail elsewhere, e.g., in ref 10. The representative set of seven candidate analytes covering most of the common solvent properties included n-octane, toluene, propan-1-ol, trichloromethane, tetrachloroethene, pyridine, and thiophene.
(1)
Here cpoly and ca denote the analyte concentration in the polymer and the gas phase. The partition coefficients can be calculated if a linear relation exists between VOC concentrations and the sensor signals (VOC concentration must be very low to approximate conditions of “infinite dilution”). TSMRs provide a convenient method to measure directly cpoly, provided all criteria necessary for them to act as mass-sensitive devices are satisfied (see below and refs 11 and 12). The size of the polar plot reflects the signal intensities; the plot shape is determined by the relative magnitude of the response along each axis (representing each polymer), thereby creating a characteristic “selectivity pattern” for a given analyte. In comparing polar plots associated with different solvents, one has to bear in mind that, in the absence of any specific chemical interactions, the partition coefficient is inversely proportional to the saturation vapor pressure (or proportional to boiling temperature and vaporization enthalpy) of the respective analyte. Thus, a highly volatile analyte, e.g., acetone, will always generate a much lower sensor response than a compound of low volatility, e.g., tetrachloroethene, at a comparable absolute concentration. (ppm by volume). This is not a consequence of a higher selectivity of a polymer to tetrachloroethene. The issue of selectivity and saturation vapor pressure was discussed in detail in ref 13. The different saturation vapor pressures influence the absolute size but not the relative shape of the polar plot. The shape of the plot is given by the ratios of the partition coefficients of the different polymers; its size then depends on the saturation vapor pressure, which is identical for all the axes (since it is the same analyte), and hence is just an extension factor. Examining, for example, the homologous series of the alkanes, the plot shape is identical for n-hexane and n-octane; in the case of n-octane, the size of the plot is larger since it has a lower saturation vapor pressure. Figure 2a shows polar plots for n-octane, toluene, and propan1-ol sorbed by six of the polysiloxanes. For n-octane, the nonpolar polymers in the upper part of the diagram show higher partition coefficients; the shape of the propan-1-ol plot is almost the converse. The polar plots for benzene (Figure 2b) and toluene are nearly centrosymmetric, a consequence of similar partitioning in nonpolar and polar polymers due to their polarizability. In the case of heteroaromatic compounds (Figure 2b), the partition coefficients of the polar polymers increase due to these molecules’ increased dipole moment; the basic pyridine additionally shows (10) Abraham, M. H.; Whiting, G. S.; Doherty, R. M.; Shuely, W. J. J. Chem. Soc., Perkin Trans. 2 1990, 1451-1460. (11) Lucklum, R.; Behling, C.; Hauptmann, P. Anal. Chem. 1999, 71, 24882496. (12) Martin, S. J.; Frye, G. C.; Senturia, S. D. Anal. Chem. 1994, 66, 2201. (13) Grate, J. W.; Patrash, S. J.; Abraham, M. H.; My Du, C. Anal. Chem. 1996, 68, 913.
Analytical Chemistry, Vol. 72, No. 16, August 15, 2000
3697
Figure 2. Summary of the effects of side-chain modification: polar plots of an array of six polysiloxanes. (a) Three different key analytes: n-octane, toluene, and propan-1-ol. (b) Increasingly polar aromatic compounds: benzene, thiophene, and pyridine. (c) Comparison of cyclic and aliphatic alkanes: cyclohexane and n-hexane. Table 1. Selected Partition Coefficients (K) Determined for Six Polysiloxanes at 303 K under Conditions of “Infinite” Dilution by TSMR Measurementsa polymer analyte
pa0 (Pa)
PDMS
PMiPCAS
PMAPS
CSVAL
PMPS
PMCPS
n-hexane n-heptane n-octane n-nonane cyclohexane benzene toluene p-xylene trichloromethane tetrachloromethane tetrachloroethene ethanol propan-1-ol propan-2-ol butan-1-ol ethyl acetate acetonitrile thiophene pyridine water
24600 7680 2430 770 16000 15700 4820 1530 31900 18700 3160 10400 3780 8010 1180 15800 15300 13200 3570 4180
350 800 2200 5300 420 400 1200 3100 260 450 1600 180 250 100 570 270
