Anda di halaman 1dari 16

Advanced Drug Delivery Reviews 57 (2005) 1779 1794 www.elsevier.

com/locate/addr

Optimization, evaluation, and characterization of molecularly imprinted polymers B


David A. Spivak
Department of Chemistry, Louisiana State University, 232 Choppin Hall, Baton Rouge, LA 70803, California, United States Received 25 January 2005; accepted 29 July 2005 Available online 2 November 2005

Abstract The underlying mechanisms for molecular recognition exhibited by the imprinting effect can be attributed to two processes. The pre-organization of complementary functional groups in the polymer by the template and the formation of a shape-selective cavity that is complementary to the template. However, measurements of binding and selectivity combine all effects contributing to molecular recognition in MIPs into one figure of merit. If the two molecules being compared are not enantiomers, then there are other factors which contribute to differential binding such as size or different partitioning effects due to differences in polarity, hydrophobicity, ionization state or shape and/or conformational effects. The best probe for the imprinting effect is therefore an enantiomeric pair. Therefore, the first section of this article discusses enantioselective optimization of polymerization, the second section will review methods employed for evaluation of MIPs and the last section will cover materials science methods used to characterize the physical properties of MIP materials. D 2005 Elsevier B.V. All rights reserved.
Keywords: Molecularly imprinted polymer; Optimization; Evaluation; Characterization; Pre-polymerization complex; Functional monomer; Crosslinking monomer

Contents 1. 2. Introduction . . . . . . . . . . . . . . . . . . Optimization of polymerization parameters . . 2.1. Optimization of functional monomer to 2.1.1. Covalent MIPs . . . . . . . . 2.1.2. Non-covalent MIPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . crosslinking monomer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (M/X) ratio. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1780 1781 1781 1781 1782

B This review is part of the Advanced Drug Delivery Reviews theme issue on bMolecularly imprinted polymers: Technology and applicationsQ, Vol. 57/12, 2005. E-mail address: david_spivak@chem.lsu.edu.

0169-409X/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.addr.2005.07.012

1780

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

2.2. Template concentration . . . . . . . . . . . . . . . . . . . . . . 2.3. Solvent effects . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Temperature effects. . . . . . . . . . . . . . . . . . . . . . . . 2.5. Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Evaluation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. General analysis of binding and selectivity using batch methods 3.2. Chromatographic evaluations . . . . . . . . . . . . . . . . . . . 4. Characterization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Surface area and porosity . . . . . . . . . . . . . . . . . . . . 4.2. Characterization of MIPs by spectroscopic analysis techniques . 4.3. Characterization of MIP swelling . . . . . . . . . . . . . . . . 5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

. . . . . . . . . . . . .

1782 1785 1786 1786 1787 1787 1790 1791 1791 1792 1792 1793 1793

1. Introduction Historically, the best indicator of the bimprinting effectQ by molecularly imprinted polymers (MIPs) has also been the best application of MIPs, namely the observation of increased binding for the template molecule versus other substrate molecules. The underlying mechanisms for molecular recognition exhibited by the imprinting effect have been attributed to two factors: (a) Pre-organization of complementary functional groups in the polymer by the template (b) Formation of a shape-selective cavity that is complementary to the template. Thus, the imprinting effect is essentially a threedimensional effect, i.e. it is the effective control of three-dimensional interactions by the template with surrounding functional monomers and the crosslinked matrix. However, measurements of binding and selectivity combine all effects contributing to molecular recognition in MIPs into one figure of merit. This figure of merit is usually in the form of a binding constant K, which is compared for different substrates; or as a selectivity factor a, which is the ratio of binding constants for two different substrates. If the two molecules being compared are not enantiomers, then there are other factors which contribute to differential binding (and hence separation) of these substrates such as size or different partitioning effects due to differences in: (i) polarity (ii) hydrophobicity

(iii) dipoles or ionization state (iv) shape and/or conformational effects. Thus, evaluation of the imprinting effect becomes quite complex when comparing several substrates or several different MIPs. The best probe for the imprinting effect, then, is the differential binding of a chiral template and its enantiomer, since all size and partitioning considerations remain the same, with the only difference being the topological orientation of the molecules components in three-dimensional space [1]. Because enantioselectivity is the best probe of the imprinting effect, it follows that enantioselectivity provides the clearest evidence of improvements in the imprinting effect when optimizing MIPs. This is particularly important for comparing one polymer to another, where differences in the polymer morphology and chemical composition contribute to differences in the partitioning effects between non-enantiomeric molecules. Therefore, the first section of this article discusses optimization of polymerization parameters focusing on studies that follow enantioselectivity as a figure of merit. The second section will review methods employed for evaluation of MIPs, including both physical methods of binding and selectivity characterization and the accompanying calculations for accurate assessment of the imprinting effect. The evaluation methods will also present appropriate methods for selectivity measurements in order to compare binding behavior of non-enantiomeric molecules when optimizing MIPs for separation of mixtures of different substrates. These methods allow determination of the imprinting effect when compar-

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

1781

ing non-enantiomeric substrates on two or more different MIP materials. The last section of this review will cover materials science methods used to characterize the physical properties of MIP materials. These are especially important when comparing the imprinting effects of two or more MIPs, in order to determine whether morphological or compositional effects are contributing to differences in performance between polymers.

2. Optimization of polymerization parameters Essentially two strategies for molecular imprinting have been established based on whether the template is associated with interactive monomers using covalent bonds or non-covalent interactions. Of the two strategies, the non-covalent approach has been used more extensively for three reasons: (i) Non-covalent methodology is easier because it does not require synthetic steps toward the prepolymer complex; interactions between monomers and template are easily obtained when all components are mixed in solution. (ii) Removal of the template is generally much easier, usually accomplished by continuous extraction. (iii) A greater variety of functionality can be introduced into the MIP binding site using noncovalent methods. Therefore, investigations into the behavior of MIPs have dealt more frequently with non-covalent systems. Consequently, most of the optimization studies discussed in this chapter cover non-covalent MIPs; however, studies on covalent MIP optimization will be presented in cases where information is available. 2.1. Optimization of functional monomer to crosslinking monomer (M / X) ratio 2.1.1. Covalent MIPs The requirements of covalent imprinting are different than those for non-covalent imprinting, particularly with respect to ratios of functional monomer (M), crosslinker (X), and template (T). Considering first covalently imprinted polymers, optimization has

been investigated by comparing selectivity versus ratios of crosslinking and non-crosslinking monomers. Using enantiomers of boronate ester templatebound functional monomers (Fig. 1), Wulff et al. have shown that for covalent molecular imprinting, enantioselectivity of MIPs increases with maximization of crosslinker [2]. In this study, three different crosslinkers were employed; ethyleneglycol dimethacrylate (EGDMA), butanediol dimethacrylate (BDMA), and 1,4-divinylbenzene (DVB). Polymers were prepared using this template and various ratios of crosslinker to a non-crosslinking comonomer (methylmethacrylate for EGDMA and BDMA, and styrene for DVB). Below 50% crosslinker, DVB and BDMA exhibit poor selectivity, with a slight increase for MIPs made using EGDMA. A dramatic increase was seen for enantioselectivity for MIPs employing EGDMA upon increasing crosslinking percent from 50% to 70%, after which enantioselectivity increases slowly to a maximum at 95% crosslinker and 0% comonomer (with the remaining 5% as the template-functional monomer complex). A similar response was seen for BDMA, which increases enantioselectivity almost linearly above 60% crosslinker and reaches a maximum at 95% crosslinker (Fig. 2). The maximum selectivity for BDMA is significantly lower than that found for EGDMA, attributed to the increased flexibility in the polymer afforded by the longer and more flexible four-carbon chain (versus two-carbon chain in EGDMA) between polymerizable groups. Polymers imprinted with DVB also see a rise in selectivity, however, it is slight in the range of 50% to 75%, after which enantioselective performance levels off. The critically lower values using DVB crosslinker are attributed to a relatively inflexible highly rigid matrix that exhibits kinetic hindrance to reversible rebinding of the template molecule. Overall, it can be concluded

O O O O O

Fig. 1. Covalent template-monomer complex of vinylboronic acid with h-mannopyrannoside.

1782

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794


Selectivity versus Percent Crosslinker
4
ethyleneglycol dimethacrylate butanediol dimethacrylate 1,4 divinylbenzene

3.5 Separation Factor 3 2.5 2 1.5 1 40

binding sites. Thus, optimization of X / M in noncovalent MIPs must be empirically derived; most reports, however, indicate that an optimum crosslinker percentage can be found in the range of 50% to 80% depending on the functional monomer used. 2.2. Template concentration The careful choice of functional monomer is of the utmost importance to provide complementary interactions with the template and substrates. The factors involved in the selection and optimization of the molecular structure of functional monomers are discussed elsewhere in this journal. In this section, the general effects of changing the template to functional monomer ratio will be discussed, providing the basis for optimization of the T / M ratio. Once the ratio of crosslinker to functional monomer has been determined, the template (T) concentration can then be optimized with respect to the functional monomer [4]. For covalent molecular imprinting, this is not necessary because the template dictates the number of functional monomers that can be covalently attached; furthermore, the functional monomers are attached in a stoichiometric manner. For non-covalent imprinting, the optimal T / M ratio is achieved empirically (similar to the optimization of the X / M ratio) by evaluating several polymers made with different formulations with increasing template. The underlying reason for this is thought to originate with the solution complex between functional monomers and template, which is governed by Le Chateliers principle. From the general mechanOptimization of Separation Factor vs. Percent Functional Monomer

50

60 70 80 Percentage Crosslinker

90

100

Fig. 2. Selectivity versus amount of crosslinker for covalently imprinted polymers (adapted from Ref. [2]).

from all cases studied that covalent molecular imprinting is optimized by maximizing the amount of crosslinking monomer in addition to the templatefunctional monomer complex. 2.1.2. Non-covalent MIPs The optimization of X / M for non-covalent molecular imprinting is complicated by a competing optimization of M / T which will be discussed in the next section. Seminal studies by Sellergren have investigated optimization of X / M in non-covalent imprinting systems while maintaining a fixed amount of template [3]. MIPs were formed complementary to the template l-phenylalanine anilide (l-phe-an) using the crosslinker EGDMA and the functional monomer methacrylic acid (MAA). Fig. 3 shows the separation factors (a) of MIPs made with different ratios of EGDMA / MAA. Initially, enantioselectivity for l-phe-an versus d-phe-an increases as the mole % of MAA increases. This increase starts to diminish above 20 mol% MAA, and enantioselectivity actually decreases above 30 mol% MAA. Loss of selectivity by imprinted polymers having more than 2030 mol% MAA has been postulated to arise for two reasons. First, an excess of MAA (or any functional monomer utilized) increases the amount of nonspecific binding which lowers the overall average selectivity of the MIP. Second, there is a minimum amount of crosslinker (i.e. EGDMA) necessary to form a rigid enough polymer network that will maintain the fidelity of the binding site. This limits the amount of non-crosslinking functional monomer MAA that can be used for formation of the MIP

5 Separation Factor 4 3 2 1 0 0 10 20 30 40 50 % Functional Monomer 60 70

Fig. 3. Determination of percent functional monomer for optimum X / M ratio (adapted from Ref. [3]).

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

1783

Template + Functional Monomer

Polymerize

Number of Binding Sites (N)

Pre-polymer Complex

Scheme 1. Relationship between solution complex and specific binding sites made in the MIP.

ism of formation of MIP binding sites, formation of each individual bbinding siteQ in the polymer is attributed to an individual template molecule that is surrounded by functional monomers in a pre-polymer complex (Scheme 1). It has been postulated, then, that each complex in the pre-polymer solution gives rise to each binding site (Scheme 2). Applying Le Chateliers principle to the complex formed prior to polymerization, increasing the concentration of components or binding affinity of the complex in the prepolymerization mixture would predict an increase in the pre-polymer complex. Correspondingly, there is an increase the number of final binding sites in the imprinted polymer, resulting in an increased binding or selectivity factor per gram of polymer. The pre-polymer complex can be increased by increasing either the amount of functional monomer (M) or the amount of template (T), or both. However, increasing the concentration of M results in an equal lowering of the crosslinker (X) concentration; in other words, the X / M ratio becomes smaller. As pointed out earlier, there is a limit to how small the (X / M) ratio can become before the MIP formulation falls below a critical amount of crosslinker needed to maintain the fidelity of the binding site. MIP optimization experiments in the literature have shown this ratio to have a lower limit of approximately 1.0, and optimum values around 4.0 [3].

On the other hand, the pre-polymer complex can also be increased by increasing the template concentration. This is an interesting prospect because the template can in theory be increased to very high concentrations without having any change on the composition of monomers in the final polymer. This is because the template is not covalently incorporated into the final polymer, and is removed at the end of the imprinting process. One strategy, then, to maximize the pre-polymer complex would be to increase the template concentration high enough to drive the complex equilibrium toward complex, while keeping the (X / M) ratio at the optimum 4.0. However, after some critical percent template (e.g. 20% template in Fig. 4), any excess amount of template would be anticipated to have no effect on binding site formation. In other words, once all of the functional monomer is complexed, any further addition of template will have no functional monomer to complex with, and thus is not expected to increase in the number of binding sites (assuming a binding site requires at least 1 functional monomer). However, the results shown in Fig. 4 do not occur, and will be explained first for templates with two interactive functional groups, and then the unexpected similarity for templates with only one interactive functional group. A seminal example of templates with two interactive functional groups has been reported using the template nicotine at var-

Complex II

Complex I high affinity sites low affinity sites

Complex III

low affinity sites

Scheme 2. Template/functional monomer complexes for templates capable of binding two functional monomers, that may account for nicotine binding data in Fig. 5.

1784
12 10 N (mole/g) 8 6 4 2 0 0 10 20

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

50 % template

100

Fig. 4. Theoretical behavior predicted for the graph of N versus percent template as template is increased for a model that has only one interaction between functional monomer and template.

ious concentrations while keeping the monomer amounts and the (X / M) ratio constant. Because nicotine has two amine groups, several interactions with this template and the functional monomer (methacrylic acid, MAA) are possible [5]. The data obtained from the analysis of these polymers showed that maximizing the amount of template did not optimize the MIP performance; instead, the peak performance was found at T / M ratios less than 1 (Fig. 5). From this data, a hypothetical model for 1-point binding, 2-point binding, and higher order complexes has been proposed to account for the data observed (Fig. 5). In this scheme, an initial increase in the T / M ratio maximizes the formation of the 2-point binding complex by Le Chateliers principle. However, as the T / M ratio increases further, the highly selective sites arising from 2-point binding become substituted by the less selective 1-point binding sites.
5 4 3 2 1 0 0 10 20 50 100 % Template (S-nicotine)

Enantioselectivity ()

Similar studies have carried out a similar investigation using simpler templates, such as cyclohexyl(phenylethyl)amine (CPA), for which only one solution complex is expected that has one-to-one stoichiometry of T / M [4]. Upon first consideration, MIPs made with CPA should more closely resemble the predicted result shown in Fig. 4 because the template is anticipated to position only one functional monomer for formation of the binding site according to the prepolymer stoichiometry (Scheme 3). Any further increase in template beyond the amount of available functional monomer should have no influence on the number of binding sites in the MIP. This is because once all of the functional monomer is complexed with template in the one-to-one complex established for solution complexes [6], addition of more template will not add nor otherwise interfere with the complexes made (Scheme 4). Thus the number of total binding sites would be anticipated to stay the same after this critical amount of template is reached, and remain unchanged with the addition of any more template. The results shown in Fig. 6, however, show results similar to the nicotine imprinted polymers. This would suggest a similar underlying mechanism for binding site formation, since all other materials properties remain unchanged (Fig. 6) [4]. Thus, multiple functional monomer interactions in the final polymer, regardless of the pre-polymer stoichiometry, appear to be responsible for the high-affinity binding sites seen for non-covalently imprinted polymers. Understanding the basis for optimization of noncovalent methods for molecular imprinting is important for two reasons: the methodology is far easier than covalent methods, and it produces higher affinity binding sites, versus covalent methods (this will be discussed further in the Evaluation section). The trends in binding and selectivity in non-covalently

Fig. 5. Graph of chromatographic selectivity versus mole percent template (relative to monomer concentration) for optimization of nicotine/methacrylic acid in nicotine imprinted polymers (adapted from Ref. [1]).

high affinity sites

Scheme 3. Template/functional monomer complex anticipated to result in theoretical binding behavior shown in Fig. 4.

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794


pre-polymerization post-polymerization

1785

+
low affinity sites

high affinity sites

Scheme 4. Model based on Fig. 6 for templates with only one interaction in solution but multiple interactions with the final polymer arising during polymerization.

imprinted polymers are explained best by a model that incorporates multiple functional monomers within the highest affinity binding sites. The increased number of binding interactions in the polymer binding site may account for greater fidelity of the site, and thus impart greater affinity and selectivity to the site. This would suggest that the number of functional groups in the polymer binding site is not determined directly by the solution phase pre-polymer complex; rather, it is determined during polymerization. Because of the difficulty of characterizing the binding site structures during and after polymerization, the actual events determining the final binding site structure are still unknown. Thus, the T / M ratios must still be empirically optimized by evaluation of a series of MIP formulations with various template concentrations at a fixed X / M ratio. As a general guideline, reports in the imprinting literature empirically comparing MIP performance versus T / M

ratio have often found best results for T / M ratios in the range of approximately 0.50.25, depending on the template used [3,4]. However, if more than one functional monomer is used, the process of optimization becomes exponentially more complex. The use of more than one functional monomer has been successful in a number of reported cases; optimized by changing the relative amounts of each functional monomer while keeping the template concentration constant [79]. 2.3. Solvent effects Solvent plays an important role in the formation of the porous structure of MIPs, which are a subset of a larger class known as macroporous polymers [10,11]. The morphological properties of porosity and surface area are determined by the type of solvent, referred to as bporogenQ, used in the polymerization. Porosity arises from phase separation of the porogen and the growing polymer during polymerization. Porogens with low solubility phase separate early and tend to form larger pores and materials with lower surface areas. Conversely, porogens with higher solubility phase separate later in the polymerization providing materials with smaller pore size distributions and greater surface area. It does not appear, however, that binding and selectivity in MIPs is dependent on a particular porosity. In fact, optimum results are often obtained when chloroform is used as porogen, which provides polymers with little or no porosity or surface area as determined by nitrogen adsorption BET measurements (discussed further in the Characterization section). However, the polymers made using chloroform are solvated to a high degree, and MIPs made without any porogen do not exhibit any selectivity

2.0

Enantioselectivity

1.6

1.2

20

40 60 % S-CNA

80

100

Fig. 6. Graph of chromatographic selectivity versus mole percent template (relative to monomer concentration) for optimization of SCPA [(S)-cyclohexyl(phenylethyl)amine] in S-CPA imprinted polymers (adapted from Ref. [35]).

1786

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

because substrate cannot access the polymer. Thus, diffusion of substrates through the MIP appears to be fast enough through the chloroform solvated polymers without requiring porosity. Another important role for solvent in the formation of MIPs is the effect it has on the complexation of functional monomers with the template before, during, and after polymerization. Before (and during) polymerization, the extent of the non-covalent prepolymer complex is affected by the polarity of the porogen solvent. Less polar solvents such as chloroform or benzene will increase complex formation by Le Chateliers principle, facilitating polar non-covalent interactions such as hydrogen bonding or bridging of ionic salts. On the other hand, more polar solvents tend dissociate the non-covalent interactions in the pre-polymer complex, especially protic solvents that afford a high degree of disruption to hydrogen bonds. This has been illustrated using a number of different porogens, shown in Table 1. This study was carried out by Sellergren and Shea, who demonstrated increased enantioselectivity by l-phe-an imprinted polymers as the porogen decreases in hydrogen bond capacity [12]. An important discovery in MIPs is that after polymerization, the rebinding performance is optimized when carried out in the same solvent used for imprinting [13,14]. The underlying cause for this effect has been postulated to arise from differences in solvation of the polymer structure in the binding site microenvironment. Different solvation properties of different solvents, such as chloroform and acetonitrile, may play a role in determining shape and distance parameters that are built into the forming polymer. In order to recreate and maintain these shape and distance
Table 1 Enantioselectivity of l-phe-an imprinted polymers as a function of hydrogen-bond capacity (adapted from Ref. [12]) Entry 1 2 3 4 5 6 Porogen Acetic acid Dimethylformamide Tetrahydrofuran Chloroform Acetonitrile Benzene Enantioseparation factor 1.9 2.0 4.1 4.5 5.8 6.8 Hydrogen-bond capability* S M M P P P

parameters, it is possible that optimum rebinding conditions require the same, or very similar, solvation conditions used for polymerization. 2.4. Temperature effects Several studies have shown that polymerization of MIPs at lower temperatures forms polymers with greater selectivity versus polymers made at elevated temperatures. To polymerize at colder temperatures, it is necessary to use photochemical polymerization. For example, Mosbach et al. presented a study on enantioselectivity of l-PheNHPh imprinted polymers, one polymer being thermally polymerized at 60 8C, the other photochemically polymerized at 0 8C [14,15]. The results of this study showed that better selectivity is obtained at the lower temperature versus the identical polymers thermally polymerized. The reason for this has again been postulated on the basis of Le Chateliers principle, which predicts that lower temperatures will drive the pre-polymer complex toward complex formation, thus increasing the number and, possibly, the quality of the binding sites formed. 2.5. Epilogue Because optimization of MIP formulations is largely empirical, high-throughput methods have recently been introduced for rapid screening of the different parameters [1618]. These methods make use of semi-automated procedures for parallel synthesis of small quantities of MIPs with different compositions. Rapid assays can be conducted to quickly determine candidates with improved binding or selectivity properties; these candidates can then be scaled up for more detailed analysis. Furthermore, chemometric tools for multi-variate analysis of screened libraries are also being developed to aid the optimization process [19]. These methods have been used to optimize X / M ratios in MIPs, as well as T / M ratios, type and amount of porogen, and initiation variables. In addition to these experimental methods, computational approaches are being developed for bvirtualQ high-throughput and combinatorial searches for functional group mixtures that can provide insight toward optimal MIP formulations. A last consideration is that all optimization methods

* P = poor hydrogen bonding solvent, M = moderate hydrogen bonding solvent, S = strong hydrogen bonding solvent.

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

1787

discussed here apply to traditional formation of MIPs in bulk polymer materials; different formats such as surface imprinting, small particle MIPs, and imprinting in thin films may have additional or different rules guiding their optimization. Finally, applications in the field of drug delivery may entertain alternative optimization variables such as mass transfer rates, control of swelling properties, and bioavailability.

3. Evaluation Initial methods for evaluating binding and selectivity used batch methods that resembled dialysis measurements for similar evaluations in biomolecules. Because of the heterogeneous nature of the MIP binding sites, evaluation methods have been refined that reflect better the binding behavior of imprinted materials. Difficulties with batch methods resulted in extensive use of chromatography for evaluation of MIPs, which afforded a single binding value (kV or a, discussed below) that did not need further adjustment or modeling. As solids, MIPs are naturally conducive to chromatographic methods, and the measurements sensitive and reproducible, making this the primary application of these materials. Both batch and chromatographic methods are discussed in this section, with the associated equations for obtaining binding and selectivity values from the collected data. It should be noted that comparisons of binding should be made using data acquired from the same technique, since batch and chromatographic methods do not appear to correspond quantitatively to each other. 3.1. General analysis of binding and selectivity using batch methods One of the best methods for evaluating the binding sites in MIPs is batch rebinding. Batch rebinding involves analysis of a heterogeneous mixture of MIP in a solution of substrate. After a designated amount of MIP is added to a solution of substrate (S), the amount of substrate remaining in solution after adsorption to the polymer is measured and referred to as C f (the concentration of free substrate). The solution can be separated from the polymer by filtra-

tion or the supernatant can be removed carefully in the presence of the polymer particles. The amount of substrate bound (S b) to the MIP is then calculated by subtraction of C f from the total substrate added (C t). Because the polymer is a solid, the amount of bound substrate is divided by the weight of polymer to give the amount of bound substrate per gram of polymer. At first glance, curvature in the binding isotherm is indicative of specific binding sites within the MIP, while a straight line suggests only the existence of non-specific binding (Fig. 7). In order to optimize conditions for obtaining binding isotherm measurements that are representative of the specific imprinted sites in the MIP, it is necessary to adjust two factors: the amount of polymer used and the solvent strength. Typical experiments with MIPs involve the use of organic solvents with low polarity to obtain high binding affinities; non-selective interactions are then reduced by the addition of a polar modifier until the greatest curvature in the binding isotherm is obtained. Because the focus of this review is drug delivery, the choice of solvent system is limited to aqueous systems that resemble fluids found in biological systems. Another way to increase the sensitivity to selective binding sites is to increase the amount of polymer in the heterogeneous solution. The more specific binding sites in the MIPs generally provide much higher affinities than the non-specific sites, which is then reflected as changes in the curvature of the binding isotherm. The addition of polymer to the substrate solution is limited, however, to amounts that will allow a range of substrate concentrations without depleting substrate absolutely.

Imprinted polymer Sb

Non-imprinted polymer

Cf

Fig. 7. Appearance of MIP binding isotherms for imprinted and non-imprinted polymers.

1788

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

Selectivity by a MIP arises from the differences in the free energy of adsorption of one substrate versus another. The free energy of binding the substrate by the MIP is determined by: DG RT lnKp 1

where K P is the partition coefficient of substrate on the MIP divided by the substrate in solution: Kp Sb =Cf : 2

of Shimizu et al. [2124], Sajonz et al. [25], and Kim [4] that MIPs are formed with a distribution of binding sites. This distribution is somewhat masked in the curvature exhibited by the binding isotherms (Fig. 8a); however, this becomes much clearer when the isotherm is presented in the form of a Scatchard plot (Fig. 8b) determined from Eq. (5). For a single Sb =Cf KN Sb Where: ! K = association constant (M 1) ! N = total number of binding sites class of binding site, the Scatchard plot provides a straight line from which the binding constant (K a) is determined from the slope. Scatchard plots of MIP isotherms are curved, however, which shows there is a distribution of different classes of sites in the imprinted polymers. Attempts have been made to fit two straight lines to these curved Scatchard plots (Fig. 8b), and the slopes of each interpreted to correspond to high-affinity and low-affinity binding constants; however this has been shown to be inaccurate [24]. Recent studies have shown that MIP isotherms (e.g. Fig. 8a) can be analyzed by fitting to the Freundlich equation (Eq. (6)) [2125]: Sb ACfv where: ! S b = moles of substrate bound per gram of polymer ! A = Freundlich constant ! m = Freundlich heterogeneity parameter (0 b m b 1).
250

The selectivity of one substrate versus another is quantified by the ratio of the two partition coefficients K P1 and K P2 (for substrates 1 and 2 respectively), which is referred to as the separation factor a: 3 a Kp2 =Kp1 : The corresponding difference in free energy of binding is written as: DG2 DG1 DDG RT lna: 4

When applying a to MIPs, the separation factor indicates how many times better substrate 2 binds to the polymers versus substrate 1. For measuring the imprinting effect, it is of particular interest to calculate a values from enantiomers of the same compound which have the same properties except for their three-dimensional topological geometry. Before using any of the equations presented above, it is important to discuss the heterogeneous distribution of binding sites found in MIPs and how to evaluate binding in a meaningful way. It was realized early on by Wulff [20] and more recently explained by the groups
3000

(a)

(b)

Sb (mole/g)

2000 Sb/ Cf 1000


Freundlich Eq. 6: Sb = AC f

200

150

100

20 40

60 80 100 120 140 Cf (mM)

400

800

1200

Cf (mM)

Fig. 8. Graph of binding isotherm (a) and corresponding Scatchard plot (b) for MIP elicited toward (S)-cyclohexyl(phenylethyl)amine (adapted from raw data used for Ref. [4]).

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

1789

Once the appropriate Freundlich equation is found with an acceptable fit to the overall binding isotherm data, the equation for the affinity distribution can be written incorporating the parameters of the Freundlich equation (Eq. (7)): N K i A sinpm v Ki : p 7

The graph of Eq. (7) (Fig. 9) is one way to represent the affinity distribution, and the area under this curve gives the total number of binding sites. By integrating Eq. (7), a new function is obtained which calculates the area under the curve and thus N, the total number of binding sites (Eq. (8)). N Z
lnKmax

the number of binding sites for each class with its corresponding association constant, shown as the numerator of Eq. (10). When this is divided by the number of binding sites N from Eq. (8), the numberaverage association constant (K n) is obtained. Z lnKmax N Ki Ki d lnK lnKmin 10 Kn Z lnKmax N Ki d lnK
lnKmin

N Ki d lnK

lnKmin

Asinpm m m Kmin Kmax : p 8

After substitution of Eq. (7) in the numerator and denominator, followed by integration we get the following solution for the number average association constant: 1m 1m Kmax Kmin m m Kn 11 m 1 m Kmin Kmax From Eq. (11), the number average association constant can be calculated using a binding isotherm that is modeled by the Freundlich equation (Eq. (6)). Once a number average binding constant is determined for each enantiomer, the a value shown in Eq. (3) can be determined. Using these calculations, the values found will be a combined average of the ensemble of sites tested. Furthermore, the values found are dependent on the ranges of substrate concentrations tested. Therefore, comparing a values alone may not be enough without consideration of how the binding site distribution will affect the a values through heterogeneity differences and the range of substrate concentrations that can be tested. Analysis of affinity distributions has uncovered some important aspects of MIPs. For example, a study comparing the affinity distributions of MIPs made using covalent versus non-covalent methods uncovered a distinct difference between the two. While non-covalent MIPs have an exponentially decaying distribution of binding strengths as higher affinities are approached, covalently formed MIPs appear to have a much narrower distribution around a moderate value. This might be expected from the better control over template/functional monomer stoichiometry of covalent pre-polymer complexes provide versus non-covalent complexes [24]. As a result, non-covalent bstoichiometric imprintingQ methods are currently being investigated in which

To find the number average binding constant (for a specified range of K min K max), the sum of all sites N i multiplied by the corresponding affinity constant, K i, is divided by the sum of N i, which is the total number of sites N (Eq. (9)). X Ni Ki .X Ni X Ni Ki =N : 9

From the continuous affinity distribution P the of graph of Eq. (7) (in terms of N versus ln K), N iK i can be substituted by the integration of the product of

2.5 2.0

K min

Kmax

N(K) (mole/g)

1.5 1.0 0.5 0.0 1 2 Area=N 3 4 5 6 Ln K (M-1) 7 8

Fig. 9. Graph of Eq. (7), the affinity distribution for a MIP elicited toward (S)-cyclohexyl(phenylethyl)amine based on the Freundlich model [4].

1790

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

strong non-covalent interactions are used for maintaining the stoichiometry of template/functional monomer pre-polymer complexes in an effort to reduce the variance in affinity distributions of MIPs [8,26]. A related finding, especially in the case of non-covalently imprinted polymers, is that the heterogeneity of the affinity distribution of binding sites is actually increased by the molecular imprinting process. This has a large impact on the optimization of MIPs because while the imprinting process increases the average affinity of the polymer for a particular template, applications such as chromatography will suffer greater peak-broadening and possibly reduced resolution. Obtaining a values is not limited to enantiomers, and any two compounds may be compared on the same polymer by computing their a value. In order to compare the imprinting effects of nonenantiomeric compounds the different partitioning effects encountered for each compound must be normalized. To do this, a modified partition coefficient can be used which is referred to as the imprinting factor (I) [27]. bIQ is obtained from the ratio of the partition coefficient of a substrate on an imprinted polymer, K Pmip, and the partition coefficient K Pnip, using the same substrate on a non-imprinted polymer with the same monomer formulation (Eq. (12)). I KPmip =Kpnip 12

3.2. Chromatographic evaluations Because they are synthesized as solids, MIPs can be used directly as chromatographic stationary phases which generally provide a quicker and easier method for analyzing binding properties of MIPs. For chromatography, the MIPs are sized to particles in the micron particle-size range (e.g. 2025 Am), and packed into stainless steel HPLC columns. Equations for evaluating chromatographic binding are proportional to the equations for batch rebinding above [28]. The fundamental equation relating batch rebinding to chromatography is: K kV=/: 14

The symbol / is the phase volume ratio (volume of stationary phase/volume of mobile phase), and k is the capacity factor that is found using the following equation: kV Rt Dt =Dt where: Rt Dt Retention time of the sample Retention time of an unretained sample (e.g. acetone). 15

To a first approximation, this normalization method removes binding due to non-specific interactions, leaving a value for the binding that can be attributed solely to the imprinting effect, namely the three-dimensional organization of monomers in the MIP. Taking the ratio of imprinting factors for two different substrates, I 1 for substrate one and I 2 for substrate two, we obtain the specific selectivity factor (S) [27]. S I1 =I2 13

The relationship between K and kV holds if both batch rebinding and chromatographic methods are under equilibrium conditions, allowing the equations derived for the batch rebinding method to be converted to the following equation for chromatographically derived data: a k 2 =k 1 V V I k mip =k nip V V S I1 =I2 16 17 18

In this way, the specific selectivity factor in Eq. (10) does not take into account partitioning effects between two molecules due to non-imprinted effects. For enantiomeric compounds, the value for K nip should be the same for both substrates, thus the specific selectivity factor simply reduces to a.

Because the amount of substrate that can be used for chromatographic analysis is limited to a very small range (in the low concentration range), the distribution of sites cannot be determined as in the case of batch rebinding. Therefore, comparison of chromatographic data is most meaningful if done under the same set of

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

1791

conditions, e.g. the same mobile phase and same temperature.

4.1. Surface area and porosity The morphology of MIPs, shown in Fig. 10, arises from nuclei that form around the initiator which grow to 1030 nm in diameter which then aggregate to form microspheres, that aggregate themselves into larger clusters that form the body of beads. The porosity and resulting surface area in MIPs is formed from irregular voids located between clusters of the microspheres (macropores, N 50 nm in diameter), or from the interstitial space of a given cluster of microspheres (mesopores, 250 nm in diameter), or even within the microspheres themselves (micropores, b2 nm in diameter). Typical values for surface area of the imprinted polymers are in the range of 100 to 400 m2/g. For pore size distribution there are both macropores and mesopores in the range 2 to 100 nm, and micropores of 0.6 to 2 nm in diameter. The most effective variables that control surface area and pore distribution are the percentage of crosslinking monomer, the type and amount of porogen, and the reaction temperature. Although binding and selectivity by MIPs in chromatographic or batch rebinding mode are not dependent on macroporosity, applications in drug delivery may rely on mass-transfer kinetics related to porosity. Surface area measurements in MIPs are primarily carried out using a nitrogen adsorption porosimeter using a BET (Brunauer, Emmett and Teller) analysis routine that is standard to all instruments. For pore size distributions in MIPs, the same nitrogen adsorption data can be analyzed using BJH (Barret, Joyner and Halenda) methods also available on porosimetry instruments. Results from this type of characterization are provided in Table 2 which compares the effect of different porogens on

4. Characterization The highly crosslinked network MIP materials formed during the molecular imprinting process are part of a class of materials known as macroporous polymers [10,11]. MIPs are solids, and therefore cannot be characterized by more commonly employed polymer characterization methods that would require polymer solutions; e.g. gel permeation chromatography, solution NMR techniques, and UV measurements directly on the polymers. Furthermore, because MIPs are amorphous, crystallographic or microscopy methods cannot be used to determine the structure of the MIP binding sites, although microscopy has aided the macroscopic understanding of MIP morphology [12]. Therefore, there are only a limited number of direct physical characterization methods for imprinted polymers. These include surface area and porosity measurements, IR spectroscopy, solid state NMR (13CPMAS spectroscopy), and swelling. The spectroscopy methods presented are best for investigating molecular level features of the MIP materials. The surface area, porosity, and swelling measurements characterize macroscopic features of MIPs; however, information provided by these on the binding site structure of MIPs is very limited. On the other hand, this data can be very useful for drug delivery applications such as controlling substrate release times, and swelling properties due to environmental factors.

Second family of mesopores (2-50 nm) Third family of macropores (50-1000 nm) 10-30 nm Nucleii

Large aggregates of microspheres (m)

Microsphere (100-200 nm)

First family of micropores (<2 nm)

Fig. 10. Model of morphology formation that provides the porous network in MIPs.

1792

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

Table 2 Surface area, pore volume, and average pore size in MIPs made with EGDMA / MAA monomers using l-phe-an as template (adapted from Ref. [12]) Entry Porogen Surface area (m2/g) 256 3.5 3.8 127 216 194 Pore volume (ml/g) 0.60 0.007 0.007 0.17 0.43 0.24 Average pore size (A) 94 91 71 52 78 52

1 2 3 4 5 6

Acetonitrile Chloroform Methylene chloride Dimethylformamide Benzene Tetrahydrofuran

surface area, pore volume, and average pore size in MIPs made with EGDMA/MAA monomers using lphe-an as template [12]. 4.2. Characterization of MIPs by spectroscopic analysis techniques FT-IR and solid state NMR methods are useful for the measurement of functional group incorporation, especially for the quantification of the degree of polymerization and reactivity for each type of polymerizable group on the monomers. For example, quantitative FT-IR can be used to measure the extent of unreacted double bonds using the CH out of plane bend at 900950 cm 1 and the CjC stretch at 1639 cm 1 [29]. A measure for the degree of polymerization is assessed from the number of unreacted double bonds, which are quantified by integration of the area under the peak corresponding to the wavelengths listed above. The integrated value is converted to number of double bonds using a calibration curve separately developed that correlates double bonds versus integration areas. Greater accuracy can be obtained if the ratio of peak areas for double bonds versus the area for the carbonyl group is used, eliminating any dependence on quantity of polymer used. Quantification of other functional groups can be done in similar fashion to measure the incorporation of different functional monomers. A more quantitative measure of overall unreacted double bonds in the different MIP materials can be obtained directly by 13C CP-MAS NMR (Fig. 11), without the need for calibration curves. All other functional groups of interest that are carbon-based, can also be quantified using this technique. As can

be seen in Fig. 11, resonances around 100130 ppm are due to double bonds of unreacted EGDMA; and peaks around 160180 ppm are attributable to the different CjO groups that can be found in EGDMA and MAA [30]. The percentage of unreacted double bonds are calculated using the ratio of the integrated intensity of the carbonyl resonance assigned to the unsaturated carbonyl, relative to the total integrated intensity of the saturated carbonyl: e.g. 167 versus 175 ppm for EGDMA. According to literature precedent [31] the relaxation times for CjO and CjC bonds used in this study should be directly comparable due to similar cross-polarization behavior [32]. Furthermore, studies have determined optimum cross-polarization times to be in the range of 15 ms, which is necessary to maximize signal intensity and small deviations from optimum contact time will not effect quantitative measurements of peak areas [33]. 4.3. Characterization of MIP swelling Swelling in MIPs has most often been measured using volumetric methods published by Sellergren and Shea [12]. There are some difficulties, however, due to buoyancy (i.e. the polymers float) especially for the polymers in chlorinated solvents; and general accuracy of volumetric methods. A more accurate technique that can be used measures changes in volume for a single bead [34]. In this case, the size of individual polymer particle can be observed under a microscope in the absence and presence of solvent. The particles are then photographed in swollen and unswollen states, and the

Fig. 11. Example of 13C CP-MAS NMR spectra for imprinted polymers formulated with a X / M ratio of 4 / 1, EGDMA / MAA (adapted from Ref. [30]).

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794

1793

ratios in area calculated to give the percent swelling. In many cases the particles have irregular shape or else there are wide ranges of different sizes between the particles; therefore, it is best to follow the same particle from the swollen state to the dry state.

5. Conclusion Molecular imprinting is a maturing science for tailoring specific binding properties into polymeric materials. Initial speculation on MIP binding applications was that they would serve as bsynthetic antibodiesQ; that is, they would provide specific binding for molecular targets in the same way antibodies do. However, MIPs have many differences versus antibodies; some of which are desired, such as greater stability to temperature, time, and compatibility with a wide range of solvents, additives and reagents. In other respects some differences are undesirable; for example, MIP binding sites are formed with a distribution of different binding affinities. Although antibodies are also originally formed with a distribution of affinities (i.e. a polyclonal response), single antibodies can be selected and amplified using monoclonal technology to amounts that can be used for analytical applications such as immunoassays, separations, and biosensors. This provides a uniform receptor molecule of high affinity that can be quantitatively analyzed and used to give one binding value that is uniform for any particular application without any concern regarding binding affinity distributions. MIPs on the other hand, have large numbers of different sites with a distribution of affinities that gives different average binding affinities depending on the number of sites accessed. All progress toward optimization of the MIP binding or selectivity by procedures in the bOptimizationQ section of this review have improved the bimprinting effectQ but have not eliminated problems associated with the distribution of sites. Methods presented in the Evaluation point out that comparison of MIPs is best under the same conditions, as in the chromatographic examples or other assay applications. If binding values are desired, it is important to understand the dependence of values on distribution heterogeneity and number of binding sites accessed. These aspects must be included in applications to separations, sen-

sors, or binding assays. Another difference is the single site format of monoclonal antibodies which affords easy quantification of effects by solution chemistry and immobilization strategies. MIPs are insoluble solids, and the binding sites cannot be manipulated by solution chemistries. The MIPs can be physically characterized at the macroscopic level by the techniques described in the third section; however, binding can still only be indirectly modeled at the molecular level using the Evaluation methods in the second section. Therefore, applications of antibodies which are molecular in nature, may not be directly replaceable by MIPs which are amorphous bulk materials in nature. It may be more fruitful instead to adjust applications or formats to exploit MIP capabilities. Drug delivery may be one of the best new applications of MIPs in this regard, wherein binding/release capability can be built into one material. For the binding aspect of these applications, it will be suitable to employ empirical (but reproducible) calibration of binding behavior. There are possibilities of linking binding phenomenon to release events directly through polymer conformation in a number of transduction pathways. This opens the door for the design and synthesis of a myriad new materials that cooperatively develop molecular imprinting methodologies. Thus, drug delivery is a key application that can take advantage of the materials characteristics of MIPs that are not available to other binding/receptor systems. References
[1] D.A. Spivak, J. Campbell, Systematic study of steric and spatial contributions to molecular recognition by non-covalent imprinted polymers, Analyst 126 (2001) 793 797. [2] G. Wulff, R. Kemmerer, J. Vietmeier, H.G. Poll, Chirality of vinyl polymers. The preparation of chiral cavities in synthetic polymers, Nouv. J. Chim. 6 (1982) 681 687. [3] B. Sellergren, Molecular imprinting by noncovalent interactions. Enantioselectivity and binding capacity of polymers prepared under conditions favoring the formation of template complexes, Makromol. Chem. 190 (1989) 2703 2711. [4] H. Kim, D.A. Spivak, New insight into modeling non-covalently imprinted polymers, J. Am. Chem. Soc. 125 (2003) 11269 11275. [5] H.S. Andersson, J.G. Karlsson, S.A. Piletsky, A.-C. KochSchmidt, K. Mosbach, I.A. Nicholls, Study of the nature of recognition in molecularly imprinted polymers: II. Influence of monomertemplate ratio and sample load on retention and selectivity, J. Chromatogr., A 848 (1999) 39 49.

1794

D.A. Spivak / Advanced Drug Delivery Reviews 57 (2005) 17791794 using the Freundlich isotherm-affinity distribution analysis, 76 (2004) 11231133. R.J. Umpleby II, S.C. Baxter, Y. Chen, R.N. Shah, K.D. Shimizu, Characterization of molecularly imprinted polymers with the LangmuirFreundlich isotherm, Anal. Chem. 73 (2001) 4584 4591. R.J. Umpleby II, S.C. Baxter, M. Bode, J.K. Berch, R.N. Shah, K.D. Shimizu, Application of the Freundlich adsorption isotherm in the characterization of molecularly imprinted polymers, Anal. Chim. Acta 435 (2001) 35 42. R.J. Umpleby II, M. Bode, K.D. Shimizu, Measurement of the continuous distribution of binding sites in molecularly imprinted polymers, Analyst 125 (2000) 1261 1265. P. Sajonz, M. Kele, G. Zhong, B. Sellergren, G. Guiochon, Study of the thermodynamics and mass transfer kinetics of two enantiomers on a polymeric imprinted stationary phase, J. Chromatogr., A 810 (1998) 1 17. M. Emgenbroich, G. Wulff, A new enzyme model for enantioselective esterases based on molecularly imprinted polymers, Chem.A Eur J. 9 (2003) 4106 4117. S.H. Cheong, S. McNiven, A. Rachkov, R. Levi, K. Yano, I. Karube, Testosterone receptor binding mimic constructed using molecular imprinting, Macromolecules 30 (1997) 1317 1322. M. Ringo, C. Evans, Liquid chromatography as a measurement tool for chiral interactions, Anal. Chem. 70 (1998) 315A 321A. K.J. Shea, D.Y. Sasaki, An analysis of small-molecule binding to functionalized synthetic polymers by 13CP/MAS NMR and FT-IR spectroscopy, J. Am. Chem. Soc. 113 (1991) 4109 4120. M. Sibrian-Vazquez, Ph.D. Thesis, Design, synthesis, and applications of bio-derived crosslinking monomers for molecular imprinting, Louisiana State University and Agricultural College, Baton Rouge, Louisiana, 2003. R.G. Earnshaw, C.A. Price, J.H. ODonnell, A.K. Whittaker, J. Appl. Polym. Sci. 32 (1986) 5337 5344. T. Hjertberg, T. Hargitai, P. Reinholdsson, Macromolecules 23 (1990) 3080 3087. J. Schaefer, E.O. Stejskal, R. Buchdahal, Macromolecules 10 (1977) 384 405. V.K. Sarin, S.B.H. Kent, R.B. Merrifield, Properties of swollen polymer networks. Solvation and swelling of peptide-containing resins in solid-phase peptide synthesis, J. Am. Chem. Soc. 102 (1980) 5463 5470. (a) H. Kim, D.A. Spivak, New insight into modeling noncovalently imprinted polymers, J. Am. Chem. Soc. 125 (2003) 11269 11275; (b) H. Kim, Ph.D. Thesis, "New insights into the binding site formation and the performance of molecularly imprinted polymers," Louisiana State University and Agricultural College, Baton Rouge, Louisiana, 2004.

[6] E.A. Yerger, G.M. Barrow, Acid-base reactions in nondissociating solvents-acetic acid and diethylamine in carbon tetrachloride and chloroform, J. Am. Chem. Soc. 77 (1955) 4474 4481. [7] O. Ramstrom, L.I. Andersson, K. Mosbach, Recognition sites incorporating both pyridinyl and carboxy functionalities prepared by molecular imprinting, J. Org. Chem. 58 (1993) 7562 7564. [8] C. Lubke, M. Lubke, M.J. Whitcombe, E.N. Vulfson, Imprinted polymers prepared with stoichiometric templatemonomer complexes: efficient binding of ampicillin from aqueous solutions, Macromolecules 33 (2000) 5098 5105. [9] S.A. Piletsky, H.S. Andersson, I.A. Nicholls, Combined hydrophobic and electrostatic interaction-based recognition in molecularly imprinted polymers, Macromolecules 32 (1999) 633 636. [10] A. Guyot, in: D.C. Sherrington, P. Hodge (Eds.), Synthesis and Separations Using Functional Polymers, John Wiley & Sons, New York, 1989, pp. 1 36. [11] L. Lloyd, J. Chromatogr. 544 (1991) 201. [12] B. Sellergren, K.J. Shea, Influence of polymer morphology on the ability of imprinted network polymers to resolve enantiomers, J. Chromatogr. 635 (1993) 31 49. [13] D.A. Spivak, K.J. Shea, Evaluation of binding and origins of specificity of 9-ethyladenine imprinted polymers, J. Am. Chem. Soc. 119 (1997) 4388 4393. [14] M. Kempe, K. Mosbach, Binding studies on substrate- and enantio-selective molecularly imprinted polymers, Anal. Lett. 24 (1991) 1137 1145. [15] D.J. OShannessy, B. Ekberg, K. Mosbach, Molecular imprinting of amino acid derivatives at low temperature (0 8C) using photolytic homolysis of azobisnitriles, Anal. Biochem. 177 (1989) 144 149. [16] F. Lanza, B. Sellergren, Molecularly imprinted polymers via high-throughput and combinatorial techniques, Macromol. Rapid Commun. 25 (2004) 59 68. [17] F. Lanza, B. Sellergren, Method for synthesis and screening of large groups of molecularly imprinted polymers, Anal. Chem. 71 (1999) 2092 2096. [18] T. Takeuchi, D. Fukuma, J. Matsui, Combinatorial molecular imprinting: an approach to synthetic polymer receptors, Anal. Chem. 71 (1999) 285 290. [19] M.P. Davies, V. De Biasi, D. Perrett, Approaches to the rational design of molecularly imprinted polymers, Anal. Chim. Acta 504 (2004) 7 14. [20] G. Wulff, R. Grobe-Einsler, W. Vesper, A. Sarhan, Enzymeanalogue built polymers: 5. On the specificity distribution of chiral cavities prepared in synthetic polymers, Makromol. Chem. 178 (1977) 2817 2825. [21] A.M. Rampey, R.J. Umpleby II, G.T. Rushton, J.C. Iseman, R.N. Shah, K.D. Shimizu, Characterization of the imprint effect and the influence of imprinting conditions on affinity, capacity, and heterogeneity in molecularly imprinted polymers

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31] [32] [33] [34]

[35]

Anda mungkin juga menyukai