Article pubs.acs.org/Langmuir
Surface Molecular Imprinting in Layer-by-Layer films on Silica Particles Jan Gauczinski,† Zhihua Liu,‡ Xi Zhang,‡ and Monika Schönhoff*,† †
Institute of Physical Chemistry, University of Muenster, Corrensstraße 28/30, 48149 Münster, Germany Key Laboratory of Organic Optoelectronics & Molecular Engineering, Department of Chemistry, Tsinghua University, Beijing 100084, P. R. China
‡
S Supporting Information *
ABSTRACT: An improvement to molecular imprinting in polymers, where bulk systems often suffer from slow dynamics of release and uptake, is the formation of thin films with imprinting sites that are more rapid to access by guest molecules. Based on our previous development of surface molecular imprinting layer-by-layer (LbL) films (SMILbL), the present paper presents selective imprinted sites in a surface film on dispersed silica particles, thus designing a SMILbL system with maximized active area and in addition allowing studies with bulk techniques. The multilayer is designed to include the template during the LbL buildup and to form a cross-linked network upon UV-irradiation for enhanced stability. A theophylline moiety is grafted to poly(acrylic acid) as the template, while a UV-sensitive diazo polycation cross-links the polymers after irradiation. Electrophoretic measurements prove the successful buildup of the multilayers by an alternating sign of the zeta potential. Template release is achieved by cleavage of the grafted template. The released amount of template is quantified in solution by 1H NMR spectra and is in good agreement with the prediction from surface coverage calculations. Rebinding studies of template to the now empty imprinted binding sites show a high affinity for a theophylline derivative with a rebound amount on the order of the original template content. In contrast to theophylline, caffeine with a very similar chemical structureonly differing in one functional groupshows very different binding properties due to a thiol moiety in the binding site. Thus, a particle system with very selective molecular imprinting sites is demonstrated.
■
INTRODUCTION Molecular imprinting is a well-known technique to achieve molecular recognition in a material by template assisted assembly. By removal of the template, binding sites are generated as cavities that represent the template in shape, size, and functionality. The most frequently used materials are molecularly imprinted polymers (MIPs), and these systems are promising to offer sites for specific separation, detection, or reactive sites for nanoreactors.1−3 The growing interest in this field in recent years leads to advanced applications including sensors for chemical or biochemical active components,4−6 catalysis,7−10 or novel drug design.11 The advantages of molecular imprinting lie in its easy preparation, thermal and chemical stability, and highly selective recognition capability. On the other hand, imprinting in a bulk polymer matrix has its limitation in rebinding capacity mostly due to deeply buried binding sites and long diffusion ways for the guest molecules. Finding a compromise between a highly rigid matrix which supports the shape of the binding site best, and a flexible matrix which hinders guest molecule penetration least, is a challenging task. A possible solution to these drawbacks is to create imprinting sites only on the surface of solid materials, e.g., with patterns of molecules in a monolayer on a glass substrate.12 © 2012 American Chemical Society
Other attempts are to build surface molecular recognition systems including construction of the molecular recognition moieties on silica particles by polymerization to bound monomers13−15 with subsequent dissolving of the core to obtain a thin hollow sphere.16 The challenge is to find structures with high stability, high affinity, and low diffusion time of guest molecules. A well-known and powerful way to modify a substrates surface is layer-by-layer (LbL) self-assembly of polyelectrolytes of alternating charge.17 These polyelectrolyte multilayers (PEM) can be prepared with a controlled architecture and tailored functionalities depending on the used building blocks18 and have shown their special properties in a variety of systems with electrostatic interactions17 and hydrogen bondings19,20 on planar surfaces as well as particle surfaces.21,22 These hydrated films, being only several tens of nanometers thick, are a promising material for short diffusion distances and therefore good accessibility of the binding sites, leading to a high rebinding capacity. A first demonstration for molecular imprinting in LbL films was done by Kunitake et al. Received: December 20, 2011 Revised: February 9, 2012 Published: February 10, 2012 4267
dx.doi.org/10.1021/la205027j | Langmuir 2012, 28, 4267−4273
Langmuir
Article
Scheme 1) was synthesized as reported elsewhere.29 It is termed PAAtheo, and a graft ratio of 18% was determined by 1H NMR
using the surface sol−gel method to introduce imprinted sites into inorganic−organic hybrid films.23 The hybrid film is rather rigid, and the selectivity is low. Kanekiyo et al. imprinted adenosine monophosphate into organic LbL films; however, without the concept of interchain cross-linking the stability is low.24 The photo-cross-linking of LbL films corresponds to the cross-linking by polymerization, which is a central step in all imprinted systems to achieve binding site stability and robustness of the film against external stimuli. The crosslinking of LbL films leads to structures similar to hydrogels; the latter were also investigated earlier with respect to their imprinting properties.25 Employing the advantages of the LbL technique, different types of molecules and functional moieties could be incorporated into a nanostructured thin film, with the aim of realizing their controlled loading and release. Stable LbL films for molecular imprinting were demonstrated recently by covalently cross-linking the LbL film after film formation, and the reversibility of uptake and release was demonstrated. The imprinted selectivity was monitored with UV−vis spectroscopy, quartz crystal microbalance, cyclic voltammetry,26,27 and permeation experiments28 on a system with an electrostatically bound porphyrin template. It was shown that this approach yields imprinted sites in robust multilayer structures and leads to stimuli responsive release and rebinding of the template upon swelling and shrinking of the LbL matrix.27 To diversify the types of interactions and enhance the recognition selectivity, another system, employing theophillyne derivatives imprinted into LbL films, was developed here; a combination of hydrogen-bonding and Coulomb interactions was employed for guest molecule binding.29 The selectivity could be shown by ATR-IR spectroscopy studies, but the system may need further improvement in terms of the active surface area and the corresponding total number of binding sites, as well as employing advanced sensitive spectroscopic methods to track the respective guest molecules. In the present paper, we accomplish the formation of crosslinked LbL films including template molecules on a nanoparticle surface and monitor the binding and release of the guest molecules in the imprinted sites employing NMR spectroscopy. It can be shown that we have obtained a surface molecular imprinted LbL film for the first time on silica nanoparticles. This allows us to use quantitative NMR spectroscopy, based on using a capillary with an external standard substance inside the NMR tube,30 to calculate the surface coverage of imprinted binding sites from the released amount of template. The rebinding selectivity could be determined for theool and caffeine by observation of the free amount of substance in contrast to the total amount added, as bound molecules disappear from the spectrum. With access to this method of investigations of the bulk material we were able to increase the surface area of the imprinted system. Very low concentrations could be detected for molecules that are difficult to observe by other spectroscopic methods. Thus, rebinding experiments can be performed for different guest molecules. We observed a stronger binding for the imprinted theool over caffeine with a very similar molecular structure. A surfacemolecular imprinted bulk system with very good selectivity has thus been created.
■
Scheme 1. Chemical Structure of DAR, PAAtheo, Theool, and Caffeine
spectroscopy, calculated from f = A/A0, where A is the integral of the peak at 8 ppm (template proton) and A0 is the area of the broad signal at 2.00−2.30 ppm (single polymer chain proton); therefore, the template is bound to 18% of PAA monomers. Poly(ethylene imine) (Mw = 50 000−60 000 g/mol, 50 wt % in water) (PEI) was obtained from Sigma-Aldrich. The photosensitive polycation diazo resin (solid, Mn = 2640) (DAR) was kindly provided by Prof. Yuping Dong (Beijing Institute of Technology). Poly(sodium p-styrenesulfonate) (Mw = 70 000 g/mol, solid) (PSS) was obtained from Acros Organics. PEI and DAR were used as received; PSS was dialyzed (pore size 25− 30 Å) against ultrapure water and freeze-dried before use. Tris(carboxyethyl)phosphin (TCEP) was obtained from Acros Organics. 7-(β-Hydroxyethyl)theophylline (theool) was obtained by Sigma-Aldrich. Caffeine (laboratory reagent grade, solid) was obtained by Fisher Scientific. Hexanal (pure) was obtained by Merck. Ultrapure water (Milli-Q, 18 MΩ cm) was used in all solutions and for all washing steps. HCl and NaOH were used as 0.1 M solutions from Riedel-de-Haen. D2O (99.9% isotope purity, “water peak” in spectra is HDO) and NaCl (p. a., solid) were obtained from Sigma-Aldrich. Silica particles (d = 403 nm, PDI = 0.02) were obtained as a 10 wt % aqueous dispersion from Micro Particles GmbH. Solutions. PAAtheo and DAR solutions were prepared with 1 mg/ mL concentration; the pH of both solutions was adjusted to 6.5. In addition, 0.5 M or 10 mM NaCl was added to PAAtheo or DAR solution, respectively. The PEI and PSS solutions had a monomer concentration of 10 mM, PSS being dissolved in 0.5 M NaCl, while PEI was dissolved in ultrapure water. Methods. UV−vis spectra of dispersions of coated particles were taken by a UV−vis spectrophotometer UV-2550 (Shimadzu). A 5 μL particle dispersion was added to 1 mL of water, and the spectra were recorded using a Suprasil quartz cuvette from Hellma. Zeta potential and particle size were determined by electrophoretic measurements and dynamic light scattering, respectively, employing a Zetasizer 3000 HSA (Malvern Instruments). Centrifugation was performed with a Sigma 6k15 centrifuge (Sigma Laboratory Centrifuges). NMR spectroscopy experiments were done on a 400 MHz Avance NMR spectrometer (Bruker) using 10 mm diameter sample tubes (Wilmad/ Lab Glass). A UV-irradiation lamp (NU-8 KL, 40 W, Konrad Bender Laborgeräte) provided UV emission lines at 254 and 366 nm. Particle Coating. Layer-by-layer deposition on particles was performed in analogy to the procedure developed earlier.22,31 Here, 5 mL of 2 wt % dispersion of silica particles was added dropwise under stirring to 40 mL polymer solution in a centrifuge tube. The deposition time was 30 min under continued stirring. Centrifugation with a force of 957g for 20 min led to complete sedimentation of the particles. The supernatant was replaced with water, the particles were redispersed, and the certifugation−washing procedure was repeated two more times. After each washing cycle the zeta-potential and the size were measured, and a UV−vis spectrum of the dispersion was
MATERIALS AND METHODS
Materials. The following chemicals were used: A Poly(acrylic acid)-template grafted polymer with theophylline side groups (see 4268
dx.doi.org/10.1021/la205027j | Langmuir 2012, 28, 4267−4273
Langmuir
Article
recorded as described above. The polymers were adsorbed in the following order: PEI/(PAAtheo/DAR)5/PSS, resulting in a total number of 12 layers. The dispersion was irradiated with UV-light in the tube under constant stirring until the absorbance of the DAR in the UV−vis spectra disappeared. After the cross-linking of the film, the particles were washed three times with D2O. Finally, for template release, 20 mL of 50 mM TCEP solution in D2O was added, and the dispersion was stirred for at least 10 h. Then, the particles were washed again three times with D2O. The amount of supernatant was adjusted to yield the desired volume of the final sample. 1 H NMR Spectroscopy. In order to provide an external standard in 1H NMR spectra for adsorption isotherm studies, a setup similar to the one suggested by Frise30 was employed: A melting point tube (dout = 1.35 mm, l = 8 cm) was filled with hexanal (pure) and flame-sealed. 3 mL of different theool or caffeine solutions (0.05−10 mM) in D2O were filled in a 10 mm NMR tube, and the standard was fixed in the center of the tube by a holder. Employing these solutions, calibration lines were recorded, defining the reference integral of the hexanal aldehyde proton at 10 ppm as 1 and employing it as a reference for the integral of the aromatic proton of the probe molecules. Adsorption measurements were then performed by adding 10 mM probe molecule solution stepwise (10−100 μL) to 2 mL particle dispersion, which contains ca. 20 mg of coated particles. Then, a spectrum is recorded of the suspended particle dispersion, following each addition step. Addition was stopped, when 10 μmol of probe molecules was added. The experiment was performed on separate particle samples for either target molecule. Adsorption experiments were also performed the other way round, i.e., by addition of particle dispersion to a target molecule solution. Here, the target molecule signal is decreasing due to target binding. In contrast to the first type of adsorption experiment the signal strength was much larger; thus it was sufficient to do quantitative 1H NMR detection of the free amount of target molecules by recording the spectrum without an external standard. In this case the integral of the target molecules was calculated using the HDO peak as a reference. As for all solutions 99.9% D2O is used and the samples were not exposed to air for longer times; it is assumed that the concentration of HDO does not change during the course of the experiments. Therefore, the error this method provides should be coming from the vicinity of other signal, thus very small. This method was used for adsorption of both target molecules separately by adding particles to the solution and recording the spectrum from the obtained suspension.
of only 10 mM was used for the DAR, since a higher salt concentration led to a strong charge screening after DAR adsorption, such that the particles coagulated. For the first layer of PEI, no salt was used, since PEI shows very good adsorption without salt, whereas the adsorption of PSS was performed with the addition of 0.5 M NaCl. The zeta potential after each adsorption and washing step represents the surface charge, which is clearly changing its sign after each layer, indicating successful adsorption and sufficient charge overcompensation (Figure 1). After each DAR layer a
Figure 1. Zeta potential (squares) and particle diameter (circles) upon adsorption of PEI/(PAAtheo/DAR)5/PSS in dependence of the number of polymer layers. Adsorption conditions are optimized as described in the text. Lines are guides to the eye.
potential of around +20 mV is obtained, which is sufficient for further adsorption steps, proving that problems with the charge overcompensation that occurred under different pH or salt conditions were overcome. PAAtheo leads to a strongly negative zeta potential of always less than −30 mV, showing the expected good adsorption of the chains with a high negative surface charge. As a last layer, PSS was adsorbed to ensure a stable dispersion after cross-linking. This is necessary, since particles terminated with DAR coagulated upon cross-linking. The reason is that DAR and PAAtheo lose their charge during irradiation: DAR loses charge completely; the charge of PAA is lost only for acid groups involved in the cross-linking reaction. PAAtheo is not favorable either for termination as it is a weak polyelectrolyte, which leads to coagulation at low pH values, where most charged groups are protonated. Being a strong polyelectrolyte, PSS makes the dispersion resistant to pH changes. The zeta potential after PSS adsorption is strongly negative (−49 mV), as expected. To control undesired coagulation of the particles, the size was measured as well (see Figure 1). Diameters between 380 and 600 nm were obtained after the layer formation steps, showing that no significant coagulation takes place. The exact value of the hydrodynamic radius RH = d/2 is not supposed to give more information about the layer thickness as the error is quite high due to several factors. First, there is the error of the zetasizer instrument; second, RH strongly depends on the extension of polyelectrolyte tails in solution. Conformations of adsorbed chain have a broad distribution and may vary between different adsorption steps. Thus, the size is only interpreted as a monitor of any potential coagulation. After the formation of each layer, a UV−vis spectrum is recorded from diluted dispersion. The adsorbance of the DAR at 380 nm26 should indicate the presence of DAR on the particle surface. This is a rather
■
RESULTS AND DISCUSSION Layer Formation on Silica Particles. LbL self-assembly of a theool imprinting system was demonstrated already on a planar substrate.29 Transferring the preparation method to silica particles is possible only with a careful adjustment of the adsorption conditions, since coagulation of partially coated particles, on the one hand, and desorption of instable layers, on the other hand, are critical processes when attempting LbL coating of colloids. Most important, the salt concentration and the pH strongly determine the outcome. The polyelectrolytes in the present work have specific properties which must be taken into account: DAR is a rather hydrophobic, short polymer with only about 12 repeat units, which is showing a tendency to surface charge compensation in contrast to the desired overcompensation. PAAtheo has a strongly hydrophobic side chain, leading to exponential growth under certain conditions (see Supporting Information), which is not desired, since exponentially growing films are less ordered, and complications in particle coating can be expected. The best conditions were found by employing pH 6.5 for all polymer solutions and all washing solutions. This ensures a controlled, linear growth of the film (see Supporting Information). The salt concentration was optimized for each polymer solution separately, resulting in 0.5 M for PAAtheo. A salt concentration 4269
dx.doi.org/10.1021/la205027j | Langmuir 2012, 28, 4267−4273
Langmuir
Article
leaving a thiol group in the binding site. As a way to observe and quantify the template release, 1H NMR spectroscopy was used because UV−vis spectroscopy is not applicable due to the overlap of the theophylline signal with the polymer absorption at around 200 nm. NMR spectroscopy allows to observe the single aromatic proton of the released template in the low-field region of the spectrum at 8 ppm, clearly separated from all other signals (see Figure 3). Exploiting this fact, a calibration
qualitative approach to ensure layer formation, as the light scattering of the particles superimposes the absorption band. Nevertheless, the absorbance band of DAR is clearly visible as a shoulder on the scattering curve of the particles (see Supporting Information). The growth of the absorbance band qualitatively demonstrates the growing surface coverage of DAR on the particles. In addition, a color change from a white silica dispersion to a yellow-brown dispersion caused by the yellow DAR color is clearly seen even by eye, while the supernatant was always clear and colorless after washing. Photo-Cross-Linking. A good stability of the film and the formation of stable imprinted sites in it are anticipated due to photoinduced cross-linking. Upon LbL formation, the diazonium moiety of DAR binds to the carboxylic acid from PAA by electrostatic interaction, leading to close vicinity of the functional groups. Cleaving the diazonium ion by UV irradiation leads to covalent ester bonds formed rapidly by the aryl cation and the acid. This cleavage can be observed by UV−vis spectroscopy in the aforementioned way; see the disappearance of the DAR absorbance over time (Figure 2). A
Figure 3. 1H NMR spectrum with signals for −CH= of theool at 8 ppm and aldehyde proton of hexanal at ∼10 ppm. Right inset: Calibration line for integral of theool proton relative to integral of external standard hexanal. Left inset: close-up of the 1H spectrum in the aromatic region showing hexanal and theool signals.
line is recorded for theool and caffeine solutions with the concentrations 0.05−10 mM with hexanal as an external standard. The external standard serves to enhance precision, when the probe molecule signal is normalized on the standard signal, since both are obtained simultaneously and with the same acquisition parameters. The signal intensity of the aromatic theophylline proton is measured relative to the aldehyde proton, the latter appearing nicely separated from other signals at ∼10 ppm. A linear fit of the theophylline signal for different theool concentrations (see Figure 3) is very close to the data points with a deviation R-square of >0.999. The calibration line for caffeine is identical and therefore not shown. The released template can be quantified in the release agent solution, since only the free guest molecules cause a liquid state signal at 8 ppm. The molecules incorporated in the adsorption layers have very slow dynamics, and their dipolar couplings lead to a broad signal not visible in the liquid state spectrum. Thus, from the spectrum in Figure 3 the line at 8 ppm is evaluated. The released amount obtained from 20 mg of particles is determined as 2 μmol of molecules. Taking the surface of the particles into account, this corresponds to a surface coverage of the template of Γtheo = 0.4 μg/cm2 (1.4 nmol/cm2). Comparing this result to the surface coverage from the buildup of the same layer system on a planar substrate, it is found in good agreement: From QCM it was estimated as Γtheo = 0.3 μg/cm2 (1.0 nmol/cm2), obtained under the assumption of a water content of the film of 40%. Though this estimate is rather crude, both surface coverages are in sufficient agreement to conclude that the layer buildup is very similar and the template is removed completely from the LbL film on particles. The amount released from particles also compares well with the results of Niu et al. for planar layers,29 where in the low concentration regime specific rebinding of about 1.3 nmol, corresponding to 1.7 nmol/cm2 (under the assumption of a total quartz surface area of 0.78 cm2) was concluded.
Figure 2. UV−vis spectra of a dispersion of SMILbL-coated particles during cross-linking. The inset shows the absorbance of the DAR band in dependence on irradiation time.
spectrum was recorded after every 5 min of irradiation, and the successive decrease of the absorbance at 380 nm can be observed (inset in Figure 2). On the other hand, a shoulder is appearing at around 300 nm, which can be attributed to the formation of the cleavage product. Complete decomposition of the diazonium ion is observed after 1 h; the zeta-potential of the particles is then −65 mV, in contrast to −49 mV before cross-linking. The outermost layer of PSS is responsible for the high surface charge even after cross-linking, which eliminates most of the charges in the film. The enhancement of the value of the zeta-potential may indicate that after removal of all positive charges in the film some negative charges remain. These might include charges in the binding sites of the template or excess charges not involved in the cross-linking. It turns out that the irradiation of the nontransparent dispersion has to be performed for a much longer time compared to the planar film on transparent substrates,26,27 and furthermore constant stirring has to be applied to ensure irradiation of all particles from all sides. The mechanism of cross-linking was discussed already on the molecular level for the case of planar SMILbL layers in a previous publication.27 Release. The release of the covalently bound template from the particles is achieved by treatment with TCEP, as the phosphin is very efficient in hydrolyzing the disulfide bond, 4270
dx.doi.org/10.1021/la205027j | Langmuir 2012, 28, 4267−4273
Langmuir
Article
Rebinding. The preparation of coated silica particles with an imprinted LbL film was successfully shown, and to complete this proof of concept, the rebinding selectivity of the chosen target molecules theool and caffeine was investigated. To study the rebinding selectivity two simple experiments were performed. In the first experiment, a target molecule solution with a concentration exceeding that resulting from the released amount (see previous section) is employed, where the particles are added stepwise. This way, an excess of target molecules is in solution, and the consecutive addition of particles leads to a decrease of the NMR signal of the free target molecule upon binding of a fraction of the molecules to the film. In a second binding study the whole particle batch of 20 mg is in the NMR tube, and a target molecule solution is added stepwise to observe an increase of the number of free molecules in solution, where only nonadsorbed target is detected. With both experiments, different regimes of the binding isotherm were monitored, since either an excess of small molecules or an excess of particles is employed. The results of the first experiment, employing addition of particle dispersion to guest solution, are given in Figure 4.
representing a part of the binding isotherm, where not the target molecules but the number of binding sites was in excess (Figure 5).
Figure 5. Amount of free target molecules, nfree, measured by 1H NMR, in dependence of total amount of target molecules added stepwise to a dispersion of imprinted particles (20 mg). The dashed straight line with slope of 1 represents the case of no binding.
With this experiment, in contrast to Figure 4 shown before, a better insight is given in the regime of a low bound amount, i.e., the low concentration part of the binding isotherm. Here, a slope of 1 (dashed line in Figure 5) would represent the case of no binding of the target molecules to the particles. The difference of the data points to the dashed line represents the number of target molecules, which are bound and thus not detected. The selectivity for theool is evident again because the deviation from 1 is clearly visible. For caffeine, on the other hand, the increase of the amount of free molecules is steeper; therefore, its binding is less strong than that of theool. For both targets, deviations from the case of no binding were most clearly observed in the low concentration region. Here, the slopes of the data differ most from 1, in particular for theool. At higher target molecule amounts, the slope for caffeine becomes close to 1, indicating that all newly added target molecules are free and detectable. Thus, for caffeine only a low number of binding sites is available and saturation is quickly reached. For theool the slope also increases with increasing target amount, but generally the difference to the dashed line and thus the uptake capacity of the particles is much larger. We note here that, though the data in Figure 5 appear to show changes of the slope in the range of n = 2−4 μmol, this could not be reproduced and has to be attributed to scattering of the data arising due to the challenging detection of very low concentrations. We attempted a further quantitative evaluation of the data in Figures 4 and 5 by converting them into an adsorption isotherm Γ(c); however, the quality of the data is not sufficient. Interpretation of Figures 4 and 5 thus is restricted to the clearly evident qualitative differences between caffeine and theool. A direct comparison between both types of experiments can only be drawn for the data point corresponding to 3 μmol target amount and 20 mg particles. For these amounts one reads a bound amount of caffeine of about 2 μmol from Figure 4, but about 0.8 μmol from Figure 5. This discrepancy might imply that the samples are not fully equilibrated, and the bound amount depends on the sample mixing history. Treating the particles with a large excess of caffeine at the beginning of the experiment (Figure 4) yields a stronger and irreversible binding
Figure 4. Amount of free target molecules, nfree, measured by NMR spectroscopy, in dependence on the added mass of dispersion. Data for either target molecule were determined in separate experiments starting with 3 μmol target molecules.
The binding of target molecules to the particles was clearly shown by an intensity loss of the aromatic signal and therefore a decrease of the amount of free target molecules, nfree, in the liquid phase upon particle addition. Theool is binding much stronger to the film than caffeine, as the signal loses 95% intensity after addition of the complete particle batch. The same amount of particles binds only 68% of the presented caffeine molecules, leading to the conclusion, that the selectivity of the imprinting site is responsible for the stronger binding of theool over caffeine. The strong uptake of theool into the film became even more evident by considering the fact that small amounts of particles do bind a significant amount of theool, but almost no caffeine: The slope of the curve in Figure 4 at low particle concentrations is particularly steep for theool. This is evidence for a large number of binding sites with a very high binding constant for theool, but a low binding constant for caffeine, i.e., specific binding sites. More insight into the binding properties of imprinted particles and the shape of the binding isotherm can be obtained by a second adsorption experiment, performed by adding theool to a particle dispersion. This experiment was 4271
dx.doi.org/10.1021/la205027j | Langmuir 2012, 28, 4267−4273
Langmuir
Article
as compared to a stepwise addition of target molecules to a large particle concentration. This implies that the binding might not be fully reversible under the conditions employed in the adsorption experiment. We will now discuss the rebinding experiments in terms of specific vs nonspecific binding of either target molecule to the particles. The difference in the adsorption curves is showing the imprinted selectivity for the hydroxy function of theool, which has a strong interaction with the thiol group of the binding sites. What contributes to unspecific binding might consist of adsorption due to hydrophobic interactions with the carbon backbone of the polymers. It was shown previously that nonimprinted polyelectrolyte multilayers from common polyions have a high affinity for small aromatic molecules, which adsorb into the layer, and the adsorption can be described by a Langmuir isotherm.32,33 The general capability of adsorbing small aromatic molecules into the film is the reason for the adsorption of caffeine. Considering caffeine as unspecifically bound, we have to discuss whether any unspecific binding of theool could yield the large amounts observed here. In detail, several interactions may be responsible for unspecific binding of theool or caffeine: A main binding interaction may come from charged carboxylic acid functions of PAA that were complexing DAR monomers before cross-linking and remained charged after cross-linking due to an incomplete reaction. These charged groups may interact with the hydroxyl group of theool and not with the more hydrophobic caffeine. On the other hand, not dissociated acid function could bind target molecules by hydrogen bonding as well, in which case theool and caffeine should not show different behavior. The hydroxyl function of theool even affects the polarity of the molecules, and it might be speculated whether different polarities or solubilities might control the binding strength to the polymer film. However, theool shows better solubility in water than caffeine; thus, it can be expected that unspecific binding of caffeine to any site is more pronounced. For the case of planar layers, binding experiments of theool and caffeine to not imprinted layers, where only nonspecific binding can occur, showed rather similar bound amounts for both targets.29 Altogether, we can conclude that the large difference in binding between both target molecules cannot be explained by a large difference of unspecific binding. Thus, it has to be attributed to additional, specific binding, which takes place in the case of theool, but not for caffeine. A comparison of both experiments further supports the specific binding of theool. In Figure 4, there is a remarkably strong uptake of theool at low particle concentrations. With the first step of particle addition a small amount of particles is capable of binding a major amount of the free target molecules. These particles are thus highly saturated with bound theool, which implies that an also possible binding reaction (see Scheme 2) is shifted to the bound sites by the excess of target molecules. As the next fraction of particles is added, the equilibriumas far as it is reversible, see discussion aboveis shifted more to the free site because of a lower ratio of free molecules to particles present in solution, and as more and more particles are added, the adsorption equilibrium moves further away from saturation. Nevertheless, in the whole concentration range in Figure 4 a huge difference between both targets is observed. This implies binding sites with a very large binding constant for theool, which are not occupied by caffeine. In the second experiment (Figure 5), upon addition of low concentrations of target molecules the binding sites with the
Scheme 2. Binding Processes for Both Theool and Caffeine in Imprinted and Unspecific Binding Sitesa
a
Different binding affinities control the bound amount in dependence on the concentration of free molecules.
highest affinity were occupied first, leaving less favored binding sites for further added molecules. At the end of the experiment two-thirds of the target molecules were still free in solution; the concentration was thus not sufficient to reach saturation of the particles. Also here, the difference of theool and caffeine data at low concentrations shows an equilibrium which is far more on bound side for theool, again proving the existence of additional, specific binding sites. For caffeine, on the other hand, the shape of the binding curve is different to theool in both experiments. The decrease of the amount of free molecules in Figure 4 decreases almost linearly with the amount of added particles. This implies that the binding constant is very low, and at all concentrations the system is far away from saturation. For caffeine no imprinted sites are offered as all binding from this target molecule is considered unspecific binding, although the affinity of different sites may vary. The binding of theool, on the other hand, is more sensitive to the excess of the free molecules. In Figure 5, the slope of the free caffeine in solution deviates just slightly from 1, showing again very low affinity of the film for caffeine. This underlines again the role of the imprint in creating the binding selectivity. The two experiments do not only display different regimes of binding in dependence of the ratio of target molecules to particle content but also represent different accessibility to the observed signal. In the first approach, a huge signal is observed from the start; therefore, no sophisticated observation method is needed. In the second experiment, only very small concentrations are available in the beginning, making very sensitive measurements necessary. Therefore, the external standard located inside a capillary is strongly needed to observe the signals. This supports the demand for samples with an increased surface area in these measurements, which is achieved by the development of SMILBL systems with a colloidal geometry, demonstrated in this work.
■
CONCLUSIONS The successful preparation of surface molecular imprinted LbL films on dispersed nanoparticles is presented in this work. This system offers a large active area and shows a significant selectivity toward the imprinted moiety in contrast to a very similar target structure. By quantitative NMR the selective adsorption could be studied in very low concentration regimes, exploiting the very well detectable aromatic proton of the target compounds. This allowed to approach the adsorption process from two sides, where small molecule and particles were present in excess, respectively. The desired selectivity is clearly visible in both regimes of the adsorption isotherm. 4272
dx.doi.org/10.1021/la205027j | Langmuir 2012, 28, 4267−4273
Langmuir
■
Article
Via a Template Approach. J. Am. Chem. Soc. 1986, 108 (5), 1089− 1091. (13) Titirici, M. M.; Sellergren, B. Thin molecularly imprinted polymer films via reversible addition-fragmentation chain transfer polymerization. Chem. Mater. 2006, 18 (7), 1773−1779. (14) Gao, D. M.; Zhang, Z. P.; Wu, M. H.; Xie, C. G.; Guan, G. J.; Wang, D. P. A surface functional monomer-directing strategy for highly dense imprinting of TNT at surface of silica nanoparticles. J. Am. Chem. Soc. 2007, 129 (25), 7859−7866. (15) Lu, C. H.; Zhou, W. H.; Han, B.; Yang, H. H.; Chen, X.; Wang, X. R. Surface-imprinted core-shell nanoparticles for sorbent assays. Anal. Chem. 2007, 79 (14), 5457−5461. (16) Xie, C. G.; Li, H. F.; Li, S. Q.; Gao, S. Surface molecular imprinting for chemiluminescence detection of the organophosphate pesticide chlorpyrifos. Microchim. Acta 2011, 174 (3−4), 311−320. (17) Decher, G.; Hong, J. D.; Schmitt, J. Buildup of Ultrathin Multilayer Films by a Self-Assembly Process 0.3. Consecutively Alternating Adsorption of Anionic and Cationic Polyelectrolytes on Charged Surfaces. Thin Solid Films 1992, 210 (1−2), 831−835. (18) Zhang, X.; Chen, H.; Zhang, H. Y. Layer-by-layer assembly: from conventional to unconventional methods. Chem. Commun. 2007, 14, 1395−1405. (19) Wang, L. Y.; Wang, Z. Q.; Zhang, X.; Shen, J. C. Macromol. Rapid Commun. 1997, 18, 509. (20) Stockton, W. B.; Rubner, M. F. Macromolecules 1997, 30, 2717. (21) Donath, E.; Sukhorukov, G. B.; Caruso, F.; Davis, S. A.; Möhwald, H. Novel hollow polymer shells by colloid-templated assembly of polyelectrolytes. Angew. Chem., Int. Ed. 1998, 37 (16), 2202−2205. (22) Sukhorukov, G. B.; Donath, E.; Davis, S.; Lichtenfeld, H.; Caruso, F.; Popov, V. I.; Möhwald, H. Stepwise polyelectrolyte assembly on particle surfaces: a novel approach to colloid design. Polym. Adv. Technol. 1998, 9 (10−11), 759−767. (23) Lee, S. W.; Ichinose, I.; Kunitake, T. Molecular imprinting of azobenzene carboxylic acid on a TiO2 ultrathin film by the surface solgel process. Langmuir 1998, 14 (10), 2857−2863. (24) Kanekiyo, Y.; Inoue, K.; Ono, Y.; Sano, M.; Shinkai, S.; Reinhoudt, D. N. ’Molecular-imprinting’ of AMP utilising the polyion complex formation process as detected by a QCM system. J. Chem. Soc., Perkin Trans. 2 1999, 12, 2719−2722. (25) Byrne, M. E.; Park, K.; Peppas, N. A. Molecular imprinting within hydrogels. Adv. Drug Delivery Rev. 2002, 54 (1), 149−161. (26) Shi, F.; Liu, Z.; Wu, G. L.; Zhang, M.; Chen, H.; Wang, Z. Q.; Zhang, X.; Willner, I. Surface imprinting in layer-by-layer nanostructured films. Adv. Funct. Mater. 2007, 17 (11), 1821−1827. (27) Gauczinski, J.; Liu, Z.; Zhang, X.; Schönhoff, M. Mechanism of Surface Molecular Imprinting in Polyelectrolyte Multilayers. Langmuir 2010, 26 (12), 10122−10128. (28) Liu, Z. H.; Yi, Y.; Gauczinski, J.; Xu, H. P.; Schönhoff, M.; Zhang, X. Surface Molecular Imprinted Layer-by-Layer Film Attached to a Porous Membrane for Selective Filtration. Langmuir 2011, 27 (19), 11806−11812. (29) Niu, H.; Liu, Z. H.; Fu, L.; Shi, F.; Ma, H. W.; Ozaki, Y.; Zhang, X. Surface-Imprinted Nanostructured Layer-by-Layer Film for Molecular Recognition of Theophylline Derivatives. Langmuir 2008, 24 (20), 11988−11994. (30) Frise, A. Nano-Segregated Soft Materials Observed by NMR Spectroscopy. PhD Thesis, Royal Institute of Technology, KTH, Stockholm, 2011. (31) Schwarz, B.; Schönhoff, M. A H-1 NMR relaxation study of hydration water in polyelectrolyte mono and multilayers adsorbed to colloidal particles. Colloids Surf., A 2002, 198, 293−304. (32) Choudhury, R. P.; Schönhoff, M. Pulsed field gradient NMR study of phenol binding and exchange in dispersions of hollow polyelectrolyte capsules. J. Chem. Phys. 2007, 127 (23). (33) Chakraborty, D.; Choudhury, R. P.; Schönhoff, M. Adsorption of Aromatic Alcohols into the Walls of Hollow Polyelectrolyte Capsules. Langmuir 2010, 26 (15), 12940−12947.
ASSOCIATED CONTENT
S Supporting Information *
Experimental details; Figures S-1 and S-2. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]; Ph: +49-2518323419; Fax: +49-2518329138. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This project is funded by the Deutsche Forschungsgemeinschaft (DFG) within the Chinese−German collaborative research center TR 61 “Multilevel Molecular Assemblies: Structure, Dynamics and Function”, project number B11. Additional funding was obtained by the Tsinghua University Initiative Scientific Research Program (2009-THZ02230). We thank Prof. Yuping Dong (Beijing Institute of Technology) for kindly providing the DAR azo resin polymer.
■
REFERENCES
(1) Wulff, G. Molecular Imprinting in Cross-Linked Materials with the Aid of Molecular Templates - a Way Towards Artificial Antibodies. Angew. Chem., Int. Ed. 1995, 34 (17), 1812−1832. (2) Haupt, K.; Mosbach, K. Molecularly imprinted polymers and their use in biomimetic sensors. Chem. Rev. 2000, 100 (7), 2495− 2504. (3) Alexander, C.; Andersson, H. S.; Andersson, L. I.; Ansell, R. J.; Kirsch, N.; Nicholls, I. A.; O’Mahony, J.; Whitcombe, M. J. Molecular imprinting science and technology: a survey of the literature for the years up to and including 2003. J. Mol. Recognit. 2006, 19 (2), 106− 180. (4) Kriz, D.; Ramstrom, O.; Svensson, A.; Mosbach, K. Introducing Biomimetic Sensors Based on Molecularly Imprinted Polymers as Recognition Elements. Anal. Chem. 1995, 67 (13), 2142−2144. (5) Kröger, S.; Turner, A. P. F.; Mosbach, K.; Haupt, K. Imprinted polymer based sensor system for herbicides using differential-pulse voltammetry on screen printed electrodes. Anal. Chem. 1999, 71 (17), 3698−3702. (6) Liang, C. D.; Peng, H.; Bao, X. Y.; Nie, L. H.; Yao, S. Z. Study of a molecular imprinting polymer coated BAW bio-mimic sensor and its application to the determination of caffeine in human serum and urine. Analyst 1999, 124 (12), 1781−1785. (7) Liu, J. Q.; Wulff, G. Molecularly imprinted polymers with strong carboxypeptidase A-like activity: Combination of an amidinium function with a zinc-ion binding site in transition-state imprinted cavities. Angew. Chem., Int. Ed. 2004, 43 (10), 1287−1290. (8) Polborn, K.; Severin, K. Molecular imprinting with an organometallic transition state analogue. Chem. Commun. 1999, 24, 2481−2482. (9) Robinson, D. K.; Mosbach, K. Molecular Imprinting of a Transition-State Analog Leads to a Polymer Exhibiting Esterolytic Activity. J. Chem. Soc., Chem. Commun. 1989, 14, 969−970. (10) Strikovsky, A. G.; Kasper, D.; Grun, M.; Green, B. S.; Hradil, J.; Wulff, G. Catalytic molecularly imprinted polymers using conventional bulk polymerization or suspension polymerization: Selective hydrolysis of diphenyl carbonate and diphenyl carbamate. J. Am. Chem. Soc. 2000, 122 (26), 6295−6296. (11) Allender, C. J.; Richardson, C.; Woodhouse, B.; Heard, C. M.; Brain, K. R. Pharmaceutical applications for molecularly imprinted polymers. Int. J. Pharm. 2000, 195 (1−2), 39−43. (12) Wulff, G.; Heide, B.; Helfmeier, G. Molecular Recognition through the Exact Placement of Functional-Groups on Rigid Matrices 4273
dx.doi.org/10.1021/la205027j | Langmuir 2012, 28, 4267−4273