Metal ions as templates in molecular imprinting are interesting for two reasons: first, they are the smallest possible analytes in chemistry; second, they undergo directed interactions via coordinative bonds to a ligand. Therefore, we prepared imprinted polymers containing NVP as a functional monomer. First, experiments (results not shown here) indicated that the imprinting efficiency does not depend on the exact template, i.e., the respective anion only plays a marginal role. In fact, Cu(II) is able to form coordination complexes with NVP that are stabilized by the respective counter ion. Overall, the self-assembly procedures during imprinting thus indeed focus on the cation. Figure shows the QCM frequency pattern for copper MIP and the respective reference material towards 1
mM aqueous solutions of copper and some interfering ions. Notably, the polymer preferentially adsorbs cupric ions as the mass effect for Cu(II) is at least two times larger than that for Ni(II), Zn(II), and Co(II), respectively. In all cases, the sensors were exposed to solutions containing only one ion species. Furthermore, the reference material is insensitive to the nature of the metal that justifies the absence of any recognition source in its structure which in turn means that the preformed coordination complexes during MIP synthesis indeed lead to selective recognition. This also explains the selectivity between the bivalent ions despite the comparably small difference in ionic sizes as radii for Cu(II), Ni(II), Co(II), and Zn(II) ions are 87, 83, 79, and 88
pm, respectively [
19]. Na(I), on the other hand, is not only much larger in diameter (160
pm), but also does not form complexes. Therefore, it can be inferred that Cu(II) binds to NVP, and the structure and geometry of the complex are determined by the coordination sphere of the copper ion center. This also explains the substantial selectivity occurring despite the fact that the noble metal ions all have a rather similar globular geometry. Of course, further detailed studies are necessary for this system to further elucidate the recognition processes. Sensor responses on the non-imprinted material seem to be substantial. However, they are basically the same for all bivalent ions (and much smaller for Na(I)). This is a strong indication that the frequency shift in this case does not only depend on the mass change, but is also substantially determined by the change in conductivity between the deionized water used as the background matrix and the respective ionic solutions.
Not only coordinative bonds are highly directed, but also hydrogen bonds. They, of course, can also be applied to design materials with controlled binding networks for the selective detection of microorganisms. As an example for successful imprinting,
E. coli has to be mentioned. It is a common, gram-negative, rod-shaped bacterium that is found in the intestines of warm-blooded animals. Surface imprinting, e.g., into a polyurethane matrix, leads to the respective surface cavities, as can be seen in Figure . It shows the AFM image of the polyurethane, where template cavities of
E. coli (W strain) can be clearly identified, thus in principle, confirming the validity of the templating approach. Further insight into the material structure can be obtained from the sensor responses summarized in Figure . It shows the QCM data for both imprinted and non-imprinted materials towards 4
×
10
10 cells/ml of the W and B strains of
E coli, respectively. Clearly, the sensor responses reveal that the W strain is strongly preferred over the B strain in this case as its mass effect is four times larger. Additionally, bacteria of the W strain are usually slightly larger than those of the B strain. Therefore, cavities optimized for the larger species should also incorporate the smaller ones, meaning that also B strain bacteria should, in principle, be incorporated into cavities templated with the W strain specimen. The cell walls of
E. coli consist of liposacchrides (LPS) of characteristic sequences. Bacteria are classified on the basis of these LPS and different types of antigen attached with the lipids [
20]. In the case of MIP sensors, these surface sacchrides interact with the receptor material via the functional monomer, explaining the comparably high selectivity factor of 4.
Affinity interactions are further versatile means to achieve straightforward and robust chemical sensors [
16], whose response is indirectly proportional to the diameter of the nanoparticles applied. Such affinity can be determined by a range of properties; one of which is the Pearson hardness scale. There, thiols are categorized as ‘soft’ so an appropriate metal substrate being also soft should lead to optimized affinity. To test this claim and to demonstrate this concept, MoS
2 and Cu
2S nanoparticles have been synthesized for sensing thiol vapors. Figure a,b displays the respective AFM images showing that both compounds can be synthesized in the shape of nanoparticles having 100 and 150
nm in diameter, respectively. The size distribution in both cases is appreciably small. Figure summarizes the sensor data collected when MoS
2 and Cu
2S nanoparticles are exposed to different types of organic vapors. It can be seen clearly that both kinds of nanomaterials strongly prefer the thiol functionality over the other ones: the highest sensitivities are achieved for both 1-butanethiol and 1-octanethiol. For octane and other analytes, the responses are at least a factor of 2 (1-butanethiol) and 3 (1-octanethiol) lower, respectively. This selectivity pattern gives substantial evidence that the driving force behind recognition is the Pearson hardness. Furthermore, increasing the metal softness from Mo(IV) to Cu(I), the resulting frequency shifts towards both 1-butanethiol and 1-octanethiol become three times larger, thus strongly supporting the presence of such non-covalent interactions in this case. Additionally, Figure shows the sensor signals of the MoS
2 nanoparticles and polystyrene thin film, respectively, towards 10
ppm 1-octanethiol. The response of polystyrene is six times lower as compared to MoS
2 nanoparticles. PS does not contain any SH-functionality, so the sensor responses give is strong evidence in support of the aforementioned hard-soft interactions.
Finally, comparing artificial receptor materials with natural ones holds a key for assessing their recognition abilities for making biomimetic setups. One system, where this can be achieved in a rather straightforward way, is WGA lectin for which successful MIP has already been reported [
18]. WGA lectin is a gylcoprotein, and its natural receptor is an oligosaccharide with a glucosamine moiety. The immobilization of ligand on the QCM surface also leads to sensor layers for selective and reversible binding of WGA lectin. Figure summarizes the selectivities that can be achieved from both sensor layers, i.e., the ‘artificial receptor’ MIP and the ligand, when exposed to WGA and bovine serum albumin (BSA) solutions, respectively, with a concentration of 160
μg/l. BSA is a serum protein having a similar size with WGA [
21]. Evidently, the MIP approach leads to selectivity factors of about 3, whereas the ligand system prefers WGA by a factor of more than 7. Although this is still more than a factor of 2 which is better than the MIP system, it still shows that fully artificial recognition systems already approach natural ones in terms of selectivity in some cases.
1
